A Powerful Computational Approach to Identify Surface of Protein-Protein Interaction

Researchers from UCSD and the Scripps Research Institute have developed a robust and cost effective computational approach to identify protein-protein interaction surfaces, including those of transient interaction partners based on multi-species sequence data without the need of structure information.

Background: The interactions between proteins are fundamental to all cellular processes.  Information about these interactions improves the understanding of diseases and can provide the basis for new therapeutic approaches.

Conventional computational methods for the identification of interaction surface between proteins have drawbacks of not being able to differentiate indirect from direct interacting residues.  Experimental approaches such as surface scanning mutagenesis and co-crystal structure generation are labor intensive, expensive and often serendipitous.  Furthermore, transient interaction partners are difficult to obtain by co-crystal structure generation.

Technology: Using sequence information alone, a covariance method is first applied to a large set of multi-species homologous protein sequences with known interaction partners to identify the correlated residues.  A global inference of message passing process is then used to infer direct coupling between pairs of residue positions.  Using this approach direct contact residues can be identified and the protein complexes assembled without the requirement of costly experimental input. 

This powerful computational tool has been validated using the bacteria two-component signal system of sensor kinase (SK) and response regulator (RR). Direct contact amino acid residues both for SK/RR hetero-dimer interactions and the RR/RR homo-dimer interactions were successfully and correctly identified, and confirmed with the previously known surface residues information derived from co-crystal structures.

Potential Applications are:

  • illuminating mechanism of protein-protein interaction,
  • predicting interacting protein partners,
  • identifying potential therapeutic targets for infectious diseases, cancers, neurological disorders and many others.

References: Martin Weigt, Robert A. White, Hendrik Szurmant, James A. Hoch, Terence Hwa  Identification of direct residue contacts in protein-protein interaction by message passing

Proc Natl Acad Sci U S A. 2008 Dec 30. [Epub ahead of print].

New technique is quantum leap forward in understanding proteins. http://www.physorg.com/news149276015.html

Keywords: bioinformatics, therapeutic target prediction, protein-protein interaction, amino acid residues, message passing

Case Number: 2009-193

Inquiries To: invent@ucsd.edu