Prediction of phosphoinositide binding sites

Phosphoinositides (PIs) are a class of phospholipids that are involved in cellular signaling, organelle identity and membrane trafficking. We have applied structural bioinformatics techniques in an attempt to predict the affinity and specificity of protein structures for phosphoinositide compounds. The ability to predict whether a protein is able to recognize phosphoinositide compounds would focus both practical and theoretical research when investigating pathways believed to involve phosphoinositide compounds. We have investigated several known PI binding domains that have structures represented in the PDB to investigate features shared for the purpose of binding PI. We have shown that in the majority of cases the Pleckstrin Homology (PH) domain appears to have a well defined binding groove despite its very low sequence homology (<30%). Steps have also been made to attempt to explain sequence features that account for the varying affinity of PH domains. We have also begun to investigate the amino acid composition of all known PI binding domains with the eventual aim of designing an algorithm capable of predicting the PI binding properties, from protein structure.