Determination of sub-cellular localization of membrane proteins
Kevin C Miranda1, Rajith Aturaliya2, Melissa Davis, Zheng Yuan, Cameron Flegg and Rohan Teasdale
1k.miranda@imb.uq.edu.au, University of Queensland; 2r.aturaliya@imb.uq.edu.au, University of Queensland
The RIKEN Representative Protein Set (RPS) contains 17,209 readily identifiable full-length proteins (Nature 420, 563-573 (2002)). These we subdivided into five groups depending on the predicted combinations of signal peptides (SP) and transmembrane domains (TM). Three of these groups, secreted proteins (SP +ve, TM -ve), type I transmembrane proteins (SP +ve, TM +ve) and type II transmembrane proteins (SP -ve, TM +ve) were used for this study. Functional analysis of these subgroups was performed using InterPro and SCOP domain predictions. This analysis revealed domains unique and excluded from groups, for example in the secreted protein class 693 InterPro domains were present, of these 241 were unique to this class of protein. Further analysis of the type II group of membrane proteins was performed aiming to predict sub-cellular localization. In specific proteins targeted to and endocytosed from the basolateral membrane of polarized cells and golgi targeted proteins. This was done by searching for experimentally proven localization signals such as, tyrosine and dileucine based motifs. A profile of motifs was used to search the type II dataset, hits were scored according to strict criteria and are presented here. These predictions were tested experimentally for sub-cellular localization in a polarized epithelial cell line, preliminary results are shown.