Drug repositioning: in silico platform to study Drug-Protein relationships

The aim is to create a Bioscience and Biotechnology's platform with tools that uses existing drugs to help advance basic biological research.

Drug repositioning, where an existing drug is found to be effective against a different disease, has been growing in importance in the past years. The best example is Pfizer's Viagra for erectile dysfunction, which originally targeted hypertension. Repositioning of drugs is useful as the drug's safety is already approved and it can shorten the development time by many years. It is relatively economical and can bring treatment to patients faster. However, it is not easy to reposition drugs. Most of the discovery relies on side-effects observed during research. A in silico platform that can facilitate such discovery would be highly desirable.

The well-established 2D Tanimoto Index is used to measure the molecular similarity between two molecules. Past studies showed that when the Tanimoto between two molecules is greater than or equal to 0.85, they are likely to exhibit a similar bioactivity towards the same protein. Likewise, similar proteins are likely to have similar binding sites and biochemical function when their protein sequence identity is high, > 70%. The protein sequence similarity is also a well-known index, often measured by Blast or Fasta. Based on these concepts, similar chemical compounds bind to a protein family. Indeed, this has been validated in many families, e.g. protein kinases, where some compounds have selectivity issues as the ATP binding site is conserved. If two compounds known to bind to different proteins are highly similar, then the chances are high that each might show activity towards the known target of the other, leading to the possibility of drug repositioning.

We would like to establish an in silico platform to show the dynamic relationship between ligands and protein using data in the Protein Data Bank (PDB). Currently, the number of macromolecular structures publicly available from the PDB is over 110K entries, with about 60K entries have protein-ligand interaction information. The platform will allow queries such as: which ligands are similar, what proteins and protein families they bind to? Are these proteins similar or not? A platform that investigates the complex relationship between ligand and protein will answer many questions, which can assist with drug repositioning, off-target side effects, structure activity relationships, allosteric binding.

Using the Tanimoto alone can only find molecules that are structurally similar. Cresset's 3D Field Similarity methodology can compare molecules based on their electrostatic and shape properties rather than just chemical structure, allowing the identification of pairs of molecules with similar binding properties even though the structures might be considerably different. This technique has shown its value in drug repositioning. The addition of field similarity on the ligands to the platform will greatly improve its effectiveness. Recent developments in the Cresset field technology allow it to be applied to protein binding sites, thus enabling a shape- and electrostatic-based comparison of both ligands and proteins.

The platform will allow users to query a ligand/drug in the PDB and return information about the members of any protein families that bind ligands similar, 2D or 3D to the query molecule. Such a platform may reveal distantly related proteins that nevertheless exhibit similar binding sites and might bind the drug.