Abstract
Antifreeze proteins (AFPs) adsorb to surfaces of growing ice crystals, thereby arresting their growth. The prevailing hypothesis explains the nature of adsorption in terms of a match between the hydrophilic side chains on the AFP's ice-binding surface (IBS) and the water molecules on the ice surface. The number and spatial arrangement of hydrogen bonds thus formed have been proposed to account, respectively, for the binding affinity and specificity. The crystal structure of a type III AFP from ocean pout (isoform HPLC-3) has been determined to 2.0-A resolution. The structure reveals an internal dyad motif formed by two 19-residue, loop-shaped elements. Based on of the flatness observed on the type I alpha-helical AFP's IBS, an automated algorithm was developed to analyze the surface planarity of the globular type III AFP and was used to identify the IBS on this protein. The surface with the highest flatness score is formed by one loop of the dyad motif and is identical to the IBS deduced from earlier mutagenesis studies. Interestingly, 67% of this surface contains nonpolar solvent-accessible surface area. The success of our approach to identifying the IBS on an AFP, without considering the presence of polar side chains, indicates that flatness is the first approximation of an IBS. We further propose that the specificity of interactions between an IBS and a particular ice-crystallographic plane arises from surface complementarity.