Discrimination of mammalian GPI-anchored proteins by hydropathy and amino acid propensities.

The glycosylphosphatidylinositol (GPI) attachment is a most important post-translational modification of proteins that plays essential roles in promoting the biochemical activities of eukaryotic cells. Described here is an analysis of the amino acid properties of mammalian GPI-anchored proteins (GPI-APs) and the development of an innovative method of detecting them. GPI-APs are characterized by two high-hydropathy regions: the signal peptide, located inside the Endoplasmic Reticulum (ER), and the GPI attachment signal, a sequence adjacent to the GPI-anchoring site (the ω-site). Especially in sequence analysis of known GPI-APs, there were some distinct aspects of the amino acid propensities around the ω-sites. Therefore, a method of detecting GPI-APs was developed based on hydropathy profiles and a position-specific scoring matrix (PSSM) calculated by position-specific amino acid propensities. First, sequences of GPI-APs and negative controls, determined by screening based on hydropathy and residue volume profiles, were aligned based on residue volume profiles in the C-terminal region, and the position-specific amino acid propensities of each group were calculated according to their alignment positions. Then, a PSSM was devised using the amino acid propensities of GPI-APs and negative controls, and discrimination scores were estimated for each dataset. Based on these scores at a threshold was fixed for each dataset. GPI-APs were detected with 81.1% sensitivity and a 0.818 success rate in an optimized calculation region determined by adjusting the window size of this region using a 5-fold dataset. The results indicate that a PSSM around the ω-site can effectively discriminate GPI-APs.