Computational Molecular Biology 2016, Vol.6, No.1, 1-20
12
faster and more accurate. Computer programs can calculate the binding energies and interaction energies of a
protein-ligand, protein-protein, nucleic acid-protein and nucleic acid-ligand interaction which helps in finding the
lead compound. Crystal structures of proteins, nucleic-acid and ligands are easily accessible from the databases.
Computational methods can also be used to characterize and optimize the existing ligand libraries Apart from
screening techniques, computational tools could also be used to identify the peptides residing in the
protein-protein interface, which could be targeted to inhibit protein-protein interactions, hence discovering
potential drug targets (Fletcher and Hamilton, 2007). Peptide drugs are advantageous in being selective or specific
in inhibition, they also exhibit less cellular toxicity. The major problem or drawback of peptide inhibitors is their
membrane impermeability and poor proteolytic stability (Fosgerau and Hoffmann, 2015; Otvos and Wade, 2014).
Researchers have found solution to this problem by developing drugs with specificity and efficacy of peptide and
efficient cellular uptake of small molecules. These kinds of peptides which are attached to a non-natural
compound are called stapled peptides. This concept was laid by Blackwell and Grubs. Stapling of peptides
provides stabilization of α-helical structure that makes peptides resistant to proteolysis and increases cell
permeability (Verdine and Hilinski, 2012). The most promising examples of this new technique is the
hydrocarbon-stapled α-helical peptides (Walensky and Bird, 2014). Recent research has developed stapled peptide
against MDM2 E3 ligase which inhibits MDM2-p53 interaction (Chang et al., 2013; Madden et al., 2011). The
process of drug discovery involves four steps:
1)
Prediction of the target: A drug target is selected based on the prior knowledge of its involvement in the
progression of the disease. Also, an in-depth knowledge of its structural detail along with the critical amino acid
residues involved in its interaction with other compounds and also its stabilization can help to confer drug
specificity (Hughes et al., 2011). There are many tools and servers (Table 2) which can predict the binding site of
proteins. Crystal structure of the target could be fetched from various databases, e.g., protein database (PDB)
(Berman, 2008). If crystal structure of the protein is not available one could go for computer aided homology
modeling which provides a predictive model of the target (Vyas et al., 2012).
Table 2 Computer based prediction tools for finding binding pocket and identifying critical residues in target protein.
Critical site/ residue identification
Computer Programmes
Pocket Prediction
CastP, PocketFinder, PocketPicker, Pass
Binding Site Prediction
ConSurf, Crescendo
Ligand Binding Site Prediction
3DLigandsite, LIGSITE, FINDSITE, MetaPocket, Q-siteFinder
2)
Lead generation through virtual screening: Lipinski et al., analyzed described the ideal properties of a drug
after analysing more than 2000 molecules. He laid out five rules which makes a candidate molecule fit as a ligand.
The five rules are:
1) less than 500 Dalton molecular mass 2) lipophilicity (LogP) less than 5 3) less than 5
H-bond donor 4) less than 10 H-bond acceptors 5) molar refractivity of 40-130. Any molecule has to pass the
Lipinski’s five rules to be considered as a potential drug candidate (Leeson, 2012). Screening for lead can be
carried out by two approaches: knowledge based design and screening or random screening. Zinc database is a
library of small molecules with drug potential in their 3-D conformations. It has about 727 842 molecules ready to
be used for virtual screening (Irwin et al., 2012). Knowledge based design requires a thorough knowledge of the
chemical structure of the ligand or small molecule. This helps in narrowing of the lead compound to be screened.
On the other side random screening does not require any prior knowledge of the compounds and involves
screening hits from a large collection of compounds(Hoelder et al., 2012). Virtual screening is a computer
program which uses structure based docking approach to screen chemicals from the library which bind to a
specific target. Two different methods could be followed for screening. First, ligand based approach in which
ligand is screened from the library on the basis of prior knowledge of its structural similarity to an already existing
ligand. Second, structure based approach in which screening is done on the basis of fit of the ligand in the active
