Academic Sciences International Journal of Pharmacy and Pharmaceutical Sciences

ISSN- 0975-1491 Vol 5, Suppl 1, 2013

Review Article

MOLECULAR MODELLING: A NEW SCAFFOLD FOR DRUG DESIGN

SONIKA REDHU1, ANIL JINDAL2

1JCDM College of Pharmacy, Sirsa 125055, Haryana, India, 2L. R. Institute of Pharmacy, Rajgarh Road, Solan 173212, India.
Email: sonika285@yahoo.co.in
Received: 30 Oct 2012, Revised and Accepted: 05 Dec 2012
ABSTRACT
Molecular modeling is becoming ever more important in the fields of protein engineering and drug design. X-ray diffraction and nuclear magnetic
resonance techniques are improving rapidly, but nevertheless it is unlikely that the three-dimensional structures of all molecules of interest will be
elucidated in the foreseeable future. Another line of work is drug design. Drug design is a creative act of the same magnitude as composing,
sculpting or writing. The results can touch the lives of millions, but the creator is rarely one scientist and the rewards are distributed differently in
the arts than in the sciences. The aim of this review is to give an outline of studies in the field of medicinal chemistry in which molecular modeling
has helped in the discovery process of new drugs. The emphasis will be on lead generation and optimization.
Keywords: Molecular modeling, Drug design, Molecular software, Computational chemistry

INTRODUCTION Computational Science typically unifies three distinct elements: (i)

modeling, algorithms and simulations (e.g. numerical and
Molecular modeling encompasses all theoretical methods and nonnumerical, discrete and continuous); (ii) software developed to
computational techniques used to model or mimic the behavior of solve natural science, social science, engineering, and medical
molecules. The techniques are used in the fields of computational problems; (iii) computer and information science that develops and
chemistry, drug design, computational biology and materials science optimizes advanced hardware systems, software, networking and
for studying molecular systems ranging from small chemical data management components.
systems to large biological molecules and material assemblies. The
simplest calculations can be performed by hand, but inevitably DRUG DESIGN
computers are required to perform molecular modeling of any
reasonably sized system. The common feature of molecular Drug design is a synthesis of scientific knowledge, experience,
modeling techniques is the atomistic level description of the intuition, and aesthetics. However, unlike the arts, this beauty has
molecular systems. Most molecular modeling studies involve three limited distribution; the general public is severely under informed
stages. In the first stage a model is selected to describe the intra- and about the creative process whereby molecules are designed and
inter- molecular interactions in the system. The two most common created. Indeed, like artists, scientists are hard-pressed to enunciate
models that are used in molecular modeling are quantum mechanics their intuitive insights. Three crucial components have converged to
and molecular mechanics. These models enable the energy of any accentuate structure-based drug design as a product of the end of
arrangement of the atoms and molecules in the system to be this century. Because of expanded databases (especially from
calculated, and allow the modeler to determine how the energy of crystallography and NMR), computer graphics displays, and linkage
the system varies as the positions of the atoms and molecules by facile web tools (Meyer & Funkhouser, 1998), these resources put
change. The second stage of a molecular modeling study is the precise molecular structures before our eyes and at our fingertips.
calculation itself, such as an energy minimization, a molecular The structural databases, which have been growing exponentially
dynamics or Monte Carlo simulation, or a conformational search. for three decades, now offer a number of interesting targets for
Finally, the calculation must be analyzed, not only to calculate structure-based drug design. While graphics hardware and
properties but also to check that it has been performed properly. prototypical software for drug design have been available since the
Computational chemistry/molecular modeling is the science (or art) 1960s (Meyer, 1980), only during the past decade have they
of representing molecular structures numerically and simulating matured to the extent that a synthetic chemist can now master
their behavior with the equations of quantum and classical physics. molecular modeling without having to become a computer scientist;
Computational chemistry programs allow scientists to generate and and not very many computer scientists have become synthetic
present molecular data including geometries (bond lengths, bond chemists
angles and torsion angles), energies (heat of formation, activation
energy, etc.), electronic properties (moments, charges, ionization MODELLING CHALLENGES
potential and electron affinity), spectroscopic properties The current explosion in gene databases and “structural genomics”
(vibrational modes, chemical shifts) and bulk properties (volumes,
eventually will provide essential sequence information, but the lack
surface areas, diffusion, viscosity, etc.). As with all models however,
of knowledge of the laws of protein folding generally prohibits the
the chemist's intuition and training is necessary to interpret the
inference of structure from sequence. Targeted studies, which seek
results appropriately. One of the earliest and still one of the largest
to understand the basic chemistry and physiology of a disease are
uses of computers is to solve complex problems in the natural
more promising. For example, human immunodeficiency virus (HIV)
sciences and engineering disciplines and more specifically to obtain
proteinase inhibitor development surely must be one of the most
solutions of mathematical models that describe chemical or physical
phenomena (or processes). The techniques used to obtain such spectacular successes in the brief history of structure-based drug
solutions are part of the general area called Scientific Computing, design. A novel approach scans a pathological vector using the tools
and the use of these techniques to obtain insights into scientific or of molecular biology of the many relevant proteins produced a few
engineering problems is called Computational Science. can be isolated, crystallized and structurally elucidated. For
Computational Science is a rapidly-emerging transdisciplinary field example, Pyrobaculum aerophilum is reported to be a cofactor for
at the intersection of the natural sciences, computer science, and the Rev and Rex Transactivator proteins of HIV-1 and T-cell
mathematics because much scientific investigation now involves leukemia virus I. It has been studied (Peat et al., 1998) at 1.75 Å
computing as well as theory and experiment: resolutions and identified as a target for chemical interdiction. The
structures of normal and pathological molecules can be compared
Computational Science = mathematics+ computer science + field of and compounds designed to inhibit pathogenic enzymes or
application. receptors selectively. So with the structure of even one target
 Redhu et al.
Int J Pharm Pharm Sci, Vol 5, Suppl 1, 5-8

protein and the knowledge of function of its receptor or active site it 3. Yasara is a molecular-graphics, modeling and simulation
is now possible to use computer tools to build and dock a ligand or package for Linux and Windows. Yasara is powered by PVL
inhibitor (“new leads”) prior to investing time and resources for (Portable Vector Language), a new development framework.
synthesis and testing. Conversely, large-scale screening may detect PVL allows you to visualize even the largest proteins and
“new leads” that then must be modeled in order to explore enables true interactive real-time simulations with highly
subsequent synthetic analogs. In either case, molecular modeling is accurate force fields on standard PCs.
essential for understanding and exploring the structure-function 4. RasMol is a molecular graphics program developed at the
relationship. Attractive and repulsive forces can be summed and the University of Edinburgh. The software is intended for the
fit quantified. Ideally, one seeks a correlated listing of experimental visualization of proteins, nucleic acids and small molecules.
and computational values to give assurance that novel compounds The program has the ability to read in PDB as well as several
can be evaluated before being synthesized. other formats. Coloring schemes including atom type,
temperature factor and hydrophobicity.
COMPUTER GRAPHICS SOFTWARE 5. MacroModel is a computer program for molecular modelling of
Software for molecular modeling is steadily improving both in organic compounds and biopolymers. It features various force
software standards and the underlying science. Commercial fields coupled with energy minimization algorithms for the
interests play a part in this process. There are indications that this prediction of geometry and relative conformational energies of
improvement will continue. The first molecular modeling facilities molecules. MacroModel also has the ability to perform
were relatively simple program which took a file of protein atomic molecular dynamics simulations to model systems at finite
positions from disk storage or tape displayed them and allowed a temperatures using stochastic dynamics and mixed Monte
user to alter the positions with I/O devices. Periodically during the Carlo algorithms.
session the modeler would write the new coordinate file back onto 6. SYBYL-X provides capabilities for crucial small molecular
the disk. If a new representation was used this involved a different modeling and simulation, includng structure-activity
program which could require the coordinate data in a different relationship modeling, pharmacophore hypothesis generation,
format or which used a different style of interaction. Today molecular alignment, conformational searching, homology
researchers realize the benefits of using a true computer system modeling, sequence alignment, and other key tasks required to
where all modules (replacing program) have the same internal data understand and model the static and dynamic 3D structural
format and the modeler has one mode of interaction. There are four properties of proteins and other biological macromolecules.
parts of a molecular modeling system: 7. SOMA2 environment is a web server based system offering a
framework for integrating molecular modelling applications
 Molecular database including molecular data exchange. SOMA2 allows users to
 Graphics system combine software available in the computing system into
 Application programs unique workflows which are automatically executed.
 System skeleton 8. Amber is a suite of programs for molecular simulation and
analysis of proteins, nucleic acids, lipids, carbohydrates.
The molecular database holds the positions of the atoms Amber" refers to two things: a set of molecular mechanical
(coordinates) and other information such as the protein sequence, force fields for the simulation of biomolecules (which are in the
amino acid properties, and atom properties. The Winchester public domain, and are used in a variety of simulation
Graphics System is based upon a relational database which lacks the programs); and a package of molecular simulation programs
efficiency of the UCSF implementation but is more flexible. which includes source code and demos.
Graphics systems make the picture from the data. Having done this, the 9. VMD is designed for modeling, visualization, and analysis of
graphics system looks after the user interaction with the device. This biological systems such as proteins, nucleic acids, lipid bilayer
involves polling the I/O devices, carrying out analog to digital conversion assemblies etc. VMD provides a wide variety of methods for
of the device signals, and loading the resulting transformation matrices rendering and coloring a molecule: simple points and lines,
into the special registers of the graphics hardware. These are applied CPK spheres and cylinders, licorice bonds, backbone tubes and
and a new picture is calculated. This is carried out about 30 times a ribbons, cartoon drawings, and others. VMD can be used to
second to obtain smooth rotation of the molecules. There are two parts animate and analyze the trajectory of a molecular dynamics
to the graphics subsystem. There is one part that is dependent upon the (MD) simulation.
graphics hardware used and the second which depends upon the 10. MOE internal representation of organic chemical structures
structure of the database and the system skeleton. and flexible architecture provide a solid foundation for
molecular modeling and computational chemistry.
Application programs are very important in protein modeling. In 11. The PUPIL (Program for User Packages Interfacing and
particular, little modeling can be carried out effectively without Linking) system is designed to facilitate communication and
using a molecular mechanics energy program. This program can be control among the different software packages (Calculation
used for finding the minimum energy conformation, modeling the Units, “CU”s) used in a multi-scale simulation. All simulation
potential field around the protein, and simulating molecular data were packed and unpacked appropriately to pass through
dynamics. The need for energy minimization which can keep up with the PUPIL application. All stubs were blocked properly (to
the graphics system has caused groups to look at new algorithms, synchronize with the information on which they are waiting),
the use of parallel processors, and specialized hardware. The system and released at the right moment (that is, when the framework
skeleton handles the passing of data between the subsystems and had put the information into the stub data structures).
takes care of input/output. 12. SIMLYS is a tool to aid in the analysis of molecular dynamics,
Monte Carlo and other stimulations. Its purpose is twofold it is
MOLECULAR MODELING SOFTWARE a system performing the actual analysis and it serves as a shell
1. Abalone is designed for macromolecular simulations (proteins, to integrate new analysis functions. SIMLYS allows one to
DNA). It supports both explicit and implicit solvent models. In analyse the results from various simulations, as for example
contrast to Ascalaph, tailored to the simulation of small from proteins or polymers, by using the trajectories. The
molecules, Abalone is focused on molecular dynamics modeling program is separated into modules performing the
of biopolymers. It supports such effective methods as the input/output, building the interface to the user, preparing the
Replica Exchange and hybrid Monte Carlo. coordinates and performing the calculations.
2. Ascalaph is general purpose molecular modeling software that PROTEIN DATA
performs quantum mechanics calculations for initial molecular
model development, molecular mechanics and dynamics To access crystal structures rapidly, all files containing protein atom
simulations in the gas or in condensed phase. It can interact coordinates are stored on a local network Furthermore, the in-house
with external molecular modeling packages (MDynaMix, protein structural data is preprocessed-- in the .pdb files we
NWChem, CP2K, PC GAMESS/Firefly and Delphi). explicitly define ligands in active sites with HETATM records and all

6
 Redhu et al.
Int J Pharm Pharm Sci, Vol 5, Suppl 1, 5-8

other atoms with ATOM records. This enables the active sites of conformation without van der Waals contacts. The random
proteins to be identified immediately. Extra information about the increment must be in user-defined limits. The energy associated to
structures, necessary for certain program macros, is coded into the the new conformation N+1 is computed (EN+1). Subsequently, we
CGI programs. This information includes tables of conserved active compare both energies EN and EN+1. If the new energy is lower than
site residues (used for the superimposition of related proteins), the the previous one, the program accepts the new conformation N+1 as
names of structures that do not contain waters, and the names of the starting conformation for the next perturbation. If EN+ 1 > EN,a
dimeric proteins. Boltzman distribution is used for a decision. A random number
between 0 and 1 is generated. Then the following Boltzman
MOLECULAR DOCKING distributed term is computed: exp[-(EN+1 -EN/RT] where R is the gas
Molecular docking can suggest a favorable configuration for two constant and T the absolute temperature. If the exponent is lower
molecules forming a complex system. Molecular docking has been than the random number, the conformation N+1 is taken as the new
applied to studies of protein–ligand interactions, and structural reference conformation. N will be re-used as the starting
information from the theoretically modeled complex may help us conformation for the next perturbation. The purpose of the Monte-
clarify the mechanism of molecular recognition. Such models can Carlo method is to provide an efficient sampling of the
even suggest modifications to lead structures to improve biological conformational space. In order to increase the speed of the
activity. We have developed different scoring functions for two algorithm, it was decided to treat bond length and angles as fixed. As
stages of molecular docking. In the first step, the dot surface is a consequence, only van der Waals and torsional terms need to be
generated using the program written by Connolly. The coordinates computed to evaluate energetic variations from the conformation N
of the probe molecule and the target molecule surfaces are randomly to N + 1. A simplified force field including only previous terms has
rotated and translated. An initial solution is randomly generated also been implemented in the Monte-Carlo algorithm.
containing six variables: three translational degrees of freedom, and
APPLICATIONS
three rotational degrees of freedom. The three rotational variables
are described by three Euler angles. The position of the target  Molecular modeling has been used to predict the ability of the
molecule is fixed and the six variables define the orientation of the enzymes to discriminate between the enantiomers of a series
probe molecule. The initial solution is evaluated using surface of substrates, and has also been applied for guidance on
complementarity. The evaluation score is composed of two parts: redesigning CRL by site-directed mutagenesis in order to test
the matching score and penalty score of atomic overlapping. the importance of a couple of amino acids selected on the basis
of the computational results.
Fitness = Scorematch - const X Scoreoverlap
 Molecular modeling techniques have been applied to allow the
Where Scorematch is the matching score and Scoreoverlap is the penalty identification of a possible binding model for a class of
score. Const is a coefficient balancing the contributions of the two suramin-related anti-HIV-1 inhibitors (Suradistas) on the HIV-
parts. Const is mainly determined by the dot density, an important 1 cellular receptor (CD4) surface, without detailed structural
parameter of MS program, which is defined as the average number information about the binding site for the ligand on CD4.
of dots per square angstrom area of both probe and target  In a third application, finally, in the absence of any
molecules. experimental structure (NMR or X-ray) of the macromolecular
biological target (Candida albicans lanosterol 14α-
Molecular Docking Studies of Two Unbound Complexes demethylase) a model has been deduced from the ligands that
It has been shown that for bound complexes, surface bind to it by means of pseudo receptor modeling.
complementarity usually is sufficient to obtain correct binding  Computer modeling has the potential to provide new insights
conformations. However, the ultimate goal of molecular docking is to into reaction pathways, to predict properties of catalysts that
predict protein–protein and protein–peptide interaction without have not yet been synthesized, and to bring information for a
requiring a complexed crystal structure. Compared with docking of given system from many different experimental techniques. e.g.
these complexes with crystal structures, calculations for complexes kinetic studies of reaction rates, thermodynamic information
without crystal structures are more difficult. During the formation of on adsorption and spectroscopic data on molecular-level
a complex, some molecules will undergo conformational changes, so structure.
the docking procedure must be sufficiently flexible to manage
REFERENCES
conformational changes, yet specific enough to identify the correct
solution. In some cases, especially when the binding regions 1. http://en.wikipedia.org/wiki/Molecular_modelling
between proteins and/or peptides are unknown, complete 2. Andrew R,Leach. Molecular Modelling:Principles and
conformational searches are not tractable. Even using rigid-body Applications
approximations, it is very difficult to determine the global minimum 3. Allen B. Richon: An Introduction to Molecular Modeling
using conventional minimization algorithms. To test our hybrid Molecular Solutions, Inc. 4411 Connecticut Avenue NW, STE
minimization algorithm and docking procedure, two uncomplexed 514 Washington, DC 20008-8677.
systems were studied. One study employed an uncomplexed trypsin 4. Rama Rao Nadendla. Molecular Modeling: A Powerful Tool for
inhibitor (4PTI in PDB) and an uncomplexed trypsin (3PTN in PDB). Drug Design and Molecular Docking
The other example used an uncomplexed serine proteinase (2PKA) 5. T Perun, C L Propst. Computer Aided Drug Design, Marcel
and an uncomplexed bovine pancreatic trypsin inhibitor (2BPI). The Dekker Inc. New York.1989;2-4.
PDB codes of these two cases are 2PTC and 2KAI. All 6. N Claude Cohen, Guide Book on Molecular Modelling in Drug
crystallographic water molecules were eliminated from the Design Academic Press. 1995;56.
structures. Some missing hydrogen atoms were added to the 7. Boyd DB, Lipkowitz KB, J.Chem.Educ. 1982:29(4):269.
complexes using the molecular design software. The crystal 8. Marshall GR, Maylor CB. Comprehensive Medicinal Chemistry
structures of uncomplexed trypsin (3PTN) and an uncomplexed Pergamon.New York.1990:431.
trypsin inhibitor (4PTI) have been solved separately in different 9. Venkataraghavan R, J. Med. Chem.1982;29:.2149.
systole forms. A comparison of their structures with the 10. http: www. netsci.org/Science/ Compchem/feature 01.html.
corresponding components of a complex has indicated that 11. Aithal KS, Singh UV, Satyanarayan K, Udupa N.Indian
relatively large conformational changes have occurred, especially in Drugs.1998;60(2):68.
the trypsin inhibitor. 12. Daniluk A.Visual modeling for scientific software architecture
THE MONTE-CARLO SIMULATION design. A practical approach Computer Physics
Communications 2012;183:213–230
The principle of a Monte-Carlo algorithm is the following: Firstly 13. Denning PJ. Computer science: the discipline in: Encyclopedia
consider a conformation (number N) of energy E N. Then modify one of Computer Science. Nature Publishing Group, London, 2000.
(or more) torsion angle(s) by adding a random increment to the 14. Eykhoff P, System Identification: Parameter and State
current torsion value. This process is repeated until reaching a Estimation, Wiley. New York.1974.

7
 Redhu et al.
Int J Pharm Pharm Sci, Vol 5, Suppl 1, 5-8

15. Vriend G. WHAT IF: A molecular modeling and drug design 29. Hou T, Xu X. A new molecular simulation software package:
program BIOSON Research Institute BIOSON Research Peking University Drug Design System (PKUDDS) for structure-
Institute, Laboratory of Chemical Physics, Department of based drug design. Journal of Molecular Graphics and
Chemistry, University of Groningen, Groningen. The Modelling 2001;19(5):455–465.
Netherlands J. Mol. Graphics. 1990;8. 30. Wang JM, Hou TJ. Automated docking of peptides and proteins
16. Bernstein FC. J. Mol. Biol. 1977;112:535-542 by genetic algorithm. Chemometr. Intell. Lab. 1999;45:281–
17. Cherfils J, Vaney MC, Morize I, Surcouf E, Colloc N, Momon JP. J. 286.
Mol. Graphics 1988;3:155-160 31. Hou TJ, Wang JM, Xu XJ. Automated docking of peptides and
18. Vinter JG, Gardner M. Molecular Modeling and Drug Design. proteins by using a genetic algorithm combined with a tabu
London: Macmillan; 1994. search. Protein Eng. 1999;12:639–647.
19. Kollman P: Advances and continuing challenges in achieving 32. Connolly, M.L. Solvent-accessible surfaces of proteins and
realistic and predictive simulations of the properties of organic nucleic acids. Science 1983, 221, 709–713.
and biological molecules. Accounts Chem Res 1996;29:461- 33. Ooms F. Molecular Modeling and Computer Aided Drug Design.
469. Examples of their Applications in Medicinal Chemistry Current
20. Jorgensen WL, Kaminski G: Performance of the AMBER94, Medicinal Chemistry. 2000;7:41-158
MMFF94 and OPLS-AA force fields for modeling organic liquids. 34. Torras J, He Y, Cao C, Muralidharan K, Deumens E, Cheng HP,
J Phys Chem 1996;100:1801-8013. TrickeySB. PUPIL:A systematic approach to software
21. Case DA, Cheatham TE, Darden T, Gohlke, Luo R, Merz KM, integration in multi-scale simulations Computer Physics
Onufriev A, Simmerling C, Wang B, Woods B. The Amber Communications 2007;177:265–279
biomolecular simulation programs. J. Computat. Chem. 35. Aqvist J, Warshel A, Chem. Rev. 2001;93:597–653.
2005;26:1668-1688. 36. Zhu T, Li J, Yip S, Bartlett RJ, Trickey SB, Leeuw DNH. Molecular
22. Ponder JW, Case DA. Force fields for protein simulations. Adv. Simulation 2003;29:671.
Prot. Chem. 2003;66:27-85. 37. Kruger P, Luke M, Szameit A. SIMLYS - a software package for
23. Case DA, Cheatham TE, Darden T, Gohlke H, Luo R, Onufriev A, trajectory analysis of molecular dynamics simulations.
Simmerling C, Wang B, Woods RJ. The Amber biomolecular Computer Physics Communications.1991;62:371-380.
simulation programs. J Comput Chem. 2005 Dec;26(16):1668-88. 38. Kinnunen T, Nyrönen T, Lehtovuori P. SOMA2 – Open source
24. Taylor R, Mullier GW, Sexton GJ. Automation of conformational framework for molecular modelling workflows CSC – Finnish
analysis and other molecular modeling calculations J. Mol. IT Center for Science
Graphics, 1992;10. 39. Lehtovuori P, Nyrönen T. J. Chem. Inf. Model. 2006;46(2):620.
25. Cohen NC, Blaney JM, Humblet C, Gund P, Barry DC. J. Med. 40. Taylor RN, Smith R: The World Wide Web as a graphical user
Chem. 1990;33:883-894 interface to program macros for molecular graphics, molecular
26. Odell BJ. Comput.Aid. Molec. Design. 1988;2:191-216. modeling, and structure-base drug design. Journal of Molecular
27. SYBYL Users Manual,Tripos Associates, St. Louis, MO 63117, USA Graphics 1996;14:291-296.
28. Broughton HB. Molecular modeling Current Opinion in 41. Meyer FE, Swanson MS, Williams AJ. Molecular modelling and
Chemical Biology. 1997;1:392-398 drug design. Pharmacology & Therapeutics 2000;85:113–121.