When planning a drug discovery project we need to examine the information available for the target in question. Based on it, we can design a strategy and a plan for the project. There are 5 possible basic scenarios:
• The structure of the protein is available: A known structure will allow the use of structure-based drug design (SBDD) tools, including an efficient use of in silico tools (computer-aided drug design, CADD). The useful computational chemistry tools at the first stages of the project include virtual library design, in silico screening, docking and scoring.
The second factor to take into consideration is the availability of ligands known to bind to the protein target. The most preferable is of course an atomic X-ray structure of the complex of the ligand with the protein. Such structure will clearly define the ligand binding site, the types of interactions present there (hydrogen bond interactions, the presence of a hydrophobic pocket, the protein residues involved in these interactions, etc.). This information will allow a reliable construction of a structure-based pharmacophore model (a pharmacophore, originally introduced in 1909 by Ehrlich, is a set of structural features that are recognized at the binding site and are responsible for the biological activity of the compound, for details see, Yang, Drug Discov Today). A pharmacophore model helps in the preparation of virtual screening, but also in lead optimization.
The blue quadrant (the best place to be at!) in the image below corresponds to a situation when a protein and protein-ligand complex structures are known.
For virtual screening we use our proprietary virtual library, which contains millions of purchasable compounds. It has already been pre-filtered to remove e.g. reactive groups, too-lipophilic compounds, too large compounds, etc. The libraries used may also be designed to contain, for example, drug-like molecules of a specific class, natural compounds, or only small-fragment molecules. Following a virtual screening campaign, hits are clustered into different structural classes to aid the selection of compounds for purchase. The purchased compounds may then be tested in biochemical assays.
Compound activity assays are performed using assay capabilities of our close partner Red Glead. In our experience, virtual screening substantially increases hit rates. To filter out false positives, we also assess the binding of the identified compounds using biophysical screening methods, like Nuclear Magnetic Resonance (NMR), thermal-shift assay and of course X-ray crystallography. This step is particularly important in the case of fragment screening, since the frequency of false positives tends to be high for fragments, which are screened at relatively high concentrations.
A follow-up in silico similarity search may also be performed after testing the initial hits in order to identify similars of the active molecules. This offers the fastest possible establishment of a SAR model (structure-activity relationships).
The second option is of course to screen the target against a library of real compounds. We use our proprietary fragment library (fragment-based drug design, FBDD), which includes a collection of 1300 fragments. The library is designed to be general purpose, (not target-directed) covering a diverse chemical space. For screening we use the proprietary WAC™ technology or various biophysical techniques.
In case there is no experimental structure of the target, SARomics Biostructures offers gene-to-drug services. In this case the protein is cloned, expressed, purified and crystallized and its X-ray crystallographic structure is determined at the company.
The three-dimensional structure of the protein-ligand complex helps to reveal the details of protein-ligand interactions, enabling rational expansion and linking of identified hits, further optimization of the chemistry and physical-chemistry characteristics of the compounds. Multiple protein-ligand complexes also help in detailed mapping of the binding site and in improving the pharmacophore model, which in turn may be used in the design of new focused compound libraries and new screens. Optimization of ligand structure should result in improving its binding affinity and other properties, a process which normally take several cycles. The result may be a series of compounds, which after further optimization and tests will result in a lead molecule.
As part of our custom protein crystallography services, SARomics Biostructures also offers protein-ligand co-crystallization with customer's compounds.
SARomics Biostructures Integrated drug discovery services are usually adapted to the needs of each project and the requirements of our the customer. Please contact us to discuss your project.
Outline of the possible strategies
As noted in the text on the right, the design and planning of a new project depends on the information at hand. This is illustrated in the image on this page. On the image along the imaginary X-axis we have two options - protein structure known and protein structure unknown. The Y-axis corresponds to the two options of known and unknown ligand structure. Each quadrant is a combination of these two possibilities: for example the blue quadrant discussed on the right, corresponds to a best-case scenario where both protein and ligand structures are known. The worst-case scenario shown in magenta corresponds to a case of unknown protein and ligand structures. The yellow and green quadrants correspond to intermediate cases. Each quadrant describes the initial strategy of the project. For example, in the yellow quadrant we will need to start with a ligand-based approach and build a QSAR model (Quantitative structure–activity relationship), which can be used later for virtual screening. Meanwhile, it would of course be beneficial to try to determine the three-dimensional structure of a protein. Structural information will always accelerate the drug discovery project!