Mapping Analogous Heterogens onto Residue Interactions (Mahori)

Summary: A web service for identification of ligand & residue interactions. It offers the link between pharmaceutically important fragments to all biomolecular binding partners in the PDB.

“ Crystal structures serve as a template for many facets of drug discovery (Breitenlechner et al., 2005). A survey on market value of small-molecule drugs has shown that two-thirds of the sales resulted from analogue design (Wermuth, 2006). Understanding of i  nteractions made by chemical analogues presented in the Protein Data Bank may suggest synthetic strategies for lead optimisation. Nevertheless, gathering binding characteristics of a particular analogue is time-consuming and there is no publicly available resource that facilitates this type of understanding. Our web-based tool, Mahori, is acronymed from its function for Mapping Analogous Heterogens onto Residue Interactions. It aims to provide visualisation and classification of molecular interactions made by the user-query atoms obtained from the heterogen section (HET) of the PDB file (wwPDB, 2007).

There exists a few web resources that allow for querying PDB structures and their superpositions based on the ligand structure; these include Relibase (Bergner et al., 2001) and IsoStar (Bruno et al., 1997). Selection of amino acid residues that interact with a queried ligand can also be achieved by FireDB (Lopez et al., 2007) and classification of ligand-protein interactions based on the residue contact can be obtained from MSDsite (Golovin et al., 2005). However, these websites do not allow for comparing molecular interactions of multiple structures at the level of ligand substructure.

The rationale to allow for user-defined substructure comparison is based on the idea of pharmacophore, which shares a similar number and position of interactions such as hydrogen bonds. Selection of the analogous part of the molecule could give an indication of the significant interactions that the substructure can contribute to the binding. Analysing all the contacts made by the whole ligand is also available from our website. However, the user might be overwhelmed by the amount of weak residue interactions that may not greatly contribute to the ligand binding. Therefore, selecting either atoms that comprise the same analogue across several structures or non-carbon atoms that have higher tendency to make significant interactions is recommended.

Mahori supports a robust querying against the whole PDB by its underlying PDB-ligand database called Credo developed by Adrian Schreyer. This database stores all the interactions the ligand make using the method adapted from interaction types assignment described in the approach for optimising fragment and scaffold docking (Marcou and Rognan, 2007). Every atom of the heterogen and its contacting neighbour atoms have their pre-defined types and the distances are calculated for every interacting pair. By prioritising the types, the distance and the angle in the case of Pi-Pi interaction, the interaction types can be assigned for all atom pairs.

The user can make a query by providing a chemical structure, a SMILES string, or a PDB ligand three-letter code, which will bring up a substructure selection panel. Interacting residues and the type of interactions made by the user-defined substructure can be retrieved and displayed in the same page. 

An example to illustrate the interaction interpretation is the general kinase inhibitor Staurosporine, which can be queried from Mahori using the three-letter code STU. The number of hydrogen bonds that residues around N4 of Staurosporine made corresponds well with the trend of binding affinitites. Kinase structures that have two residues making hydrogen bonds to N4 of staurosporine (i.e. PDB code 1AQ1, 1NVR, 1STC, 1OKY, 1YHS), or make hydrogen bonds with both N4 and O6 (i.e. PDB code 1SM2, 1QPD), all have binding affinities below 50 nM. The majority (6/7) of structures have only one residue contribute in the hydrogen bonds to N4 have binding affinities between than 50-820 nM (i.e. PDB code 1BYG, 2DQ7, 2CLQ, 1NXK, 1U59, 2BUJ). Structures that do not make any hydrogen bond with N4 at all have binding affinities 9000 nM (PDB code 1E8Z). Therefore, if one wishes to modify Staurosporine to achieve better affinity for kinase in the latter group, the best strategy is to identify a residue around the N4 region which can make another hydrogen bond with the modified staurosporine structure.

The comparisons of the amount, the type, and the position of interactions can be achieved from one resulting page on the computer screen. This kind of information might be useful for rational drug design and medicinal chemistry education. ”

References:

Bergner, A., Gunther, J., Hendlich, M., Klebe, G. and Verdonk, M. (2001) Use of Relibase for retrieving complex three-dimensional interaction patterns including crystallographic packing effects. Biopolymers, 61, 99-110.

Breitenlechner, C.B., Bossemeyer, D. and Engh, R.A. (2005) Crystallography for protein kinase drug design: PKA and SRC case studies. Biochimica et Biophysica Acta (BBA) - Proteins & Proteomics

Inhibitors of Protein Kinases (4th International Conference, Inhibitors of Protein Kinases) and Associated Workshop: Modelling of Specific Molecular Recognition Processes (Warsaw, Poland, June 25-29, 2005), 1754, 38-49.

Bruno, I.J., Cole, J.C., Lommerse, J.P.M., Rowland, R.S., Taylor, R. and Verdonk, M.L. (1997) IsoStar: A library of information about nonbonded interactions. Journal of Computer-Aided Molecular Design, 11, 525-537.

Golovin, A., Dimitropoulos, D., Oldfield, T., Rachedi, A. and Henrick, K. (2005) MSDsite: a database search and retrieval system for the analysis and viewing of bound ligands and active sites. Proteins, 58, 190-199.

Lopez, G., Valencia, A. and Tress, M. (2007) FireDB--a database of functionally important residues from proteins of known structure

10.1093/nar/gkl897. Nucl. Acids Res., 35, D219-223.

Marcou, G. and Rognan, D. (2007) Optimizing Fragment and Scaffold Docking by Use of Molecular Interaction Fingerprints. J. Chem. Inf. Model., 47, 195-207.

Wermuth, C.G. (2006) Similarity in drugs: reflections on analogue design. Drug Discovery Today, 11, 348-354.

wwPDB. (2007) Heterogen Section. Protein Data Bank Contents Guide: Atomic Coordinate Entry Format Description Version 3.0.1, Vol. 2007.