Structural motif screening reveals a novel, conserved carbohydrate-binding surface in the pathogenesis-related protein PR-5d

The proteins identified in the screen were significantly associated with carbohydrate-related functions according to gene ontology (GO) enrichment analysis, and predicted motifs were found frequently within novel folds and glycan-binding sites not included in the training set. In addition to numerous binding sites predicted in structural genomics proteins of unknown function, one novel prediction was a surface motif (W34/W36/W192) in the tobacco pathogenesis-related protein, PR-5d. Phylogenetic analysis revealed that the surface motif is exclusive to a subfamily of PR-5 proteins from the Solanaceae family of plants, and is absent completely in more distant homologs. To confirm PR-5d's insoluble-polysaccharide binding activity, a cellulose-pulldown assay of tobacco proteins was performed and PR-5d was identified in the cellulose-binding fraction by mass spectrometry.Based on the combined results, we propose that the putative binding site in PR-5d may be an evolutionary adaptation of Solanaceae plants including potato, tomato, and tobacco, towards defense against cellulose-containing pathogens such as species of the deadly oomycete genus, Phytophthora. More generally, the results demonstrate that coplanar aromatic clusters on protein surfaces are a structural signature of glycan-binding proteins, and can be used to computationally predict novel glycan-binding proteins from 3 D structure.Carbohydrate-binding proteins (CBPs) are highly diverse in terms of their sequences, structures, binding sites, and evolutionary histories [1]. Sequence-based classifications (e.g., as used in the CAZy database [2]) are an attempt to organize this diversity, and do so by grouping CBPs into evolutionarily related families and subfamilies. Many of these families have a common function and mechanism, while in others functions have diversified [2]. Prediction of novel CBPs with unique binding sites and mechanisms that are unrelated to known cases is a more difficult task, as there is no si