Common drug targets
The majority of available drugs have protein molecules as their targets. Although nucleic acids may also be considered, their use as drug targets in drug discovery and structure based drug design has been limited due to various effects like toxicity, difficulty in achieving high specificity, etc. In a survey of currently available marketed drugs, it was shown that out of the 21 000 registered in the USA drugs, only 1357 were unique. Of these 1204 were small-molecule drugs, while 166 were biological agents. The total number of protein drug targets, which included both human proteins and proteins from pathogenic organisms, was found to be 324. Of these 207 were human protein drug targets (Overington, Al-Lazikani & Hopkins, 2006). Around 70% of the targets belonged to 10 protein families, while almost 50% of the drugs were shown to exert their activity via 4 families: G-protein coupled receptors, nuclear receptors, ligand-gated and voltage-gated ion channels. Looking at the structural side, the authors used the CATH and SCOP databases, which contain classification of protein domains, and identified 130 “druggable” domains. This is a nice example of how structural bioinformatics resources are used in drug discovery and structure-based drug design.
Starting a structure-based drug discovery project - some general considerations
Most modern drug discovery projects start with protein target identification and verification to obtain a “verified drug target”. For structure-based drug design the three-dimensional structure of the protein in question needs to be determined by one of the available high resolution experimental methods: Protein crystallography or NMR. When identifying a drug target, we first need to answer some general questions:
Does the target protein belong to a biochemical pathway, which can be bypassed by the cell, if inhibited? Obviously, if the pathway can be bypassed, inhibiting it will not make much difference.
If our aim is to inhibit a protein, which belongs to a pathogen, an obvious question would be: Are there any related proteins in the human host, which may be affected by the drug?
If the protein is not so well studied one could also ask if it is actually drugable, in the sense that it has a small-molecule binding site for which a binding compound can be designed. Of course, in the best case scenario there will be some known inhibitory compounds which can be co-crystallized with the protein. This will help in mapping available interactions within the active site, which in turn will help in the next step when new compounds will be designed.
If there is no three-dimensional structure available for the protein target one could try to find a structure of a homologous protein, which may subsequently be used for homology modeling.
There are of course just few examples, many more questions, often specific to each target, need to be answered before starting a structure-based drug design project.
Actually, many of the questions related to drug discovery and structure based drug design may be answered using the tools of structural bioinformatics. For example, a search of sequence databases followed by sequence alignment and analysis may easily answer questions related to the specificity of a particular target in a given organism. In the following page we will discuss in more details the questions related to finding a useful compound (hit generation) and the use of structural information in optimization of newly identified compounds.