Sequence and structure-based prediction of eukaryotic protein phosphorylation sites¹

Abstract

Protein phosphorylation at serine, threonine or tyrosine residues affects a multitude of cellular signaling processes. How is specificity in substrate recognition and phosphorylation by protein kinases achieved? Here, we present an artificial neural network method that predicts phosphorylation sites in independent sequences with a sensitivity in the range from 69 % to 96 %. As an example, we predict novel phosphorylation sites in the p300/CBP protein that may regulate interaction with transcription factors and histone acetyltransferase activity. In addition, serine and threonine residues in p300/CBP that can be modified by O-linked glycosylation with N-acetylglucosamine are identified. Glycosylation may prevent phosphorylation at these sites, a mechanism named yin-yang regulation.

The prediction server is available on the Internet at http://www.cbs.dtu.dk/services/NetPhos/ or via e-mail to [email protected]

Introduction

Protein kinases catalyze phosphorylation events that are essential for the regulation of cellular processes like metabolism, proliferation, differentiation, and apoptosis Kolibaba and Druker 1997, Hunter 1998, Johnson et al 1996, Johnson et al 1998, Pinna and Ruzzene 1996, Graves et al 1997. This very large family of enzymes share homologous catalytic domains and the mechanism of substrate recognition may be similar despite large variation in sequence. Crystallization studies indicate that a region, between seven and 12 residues in size, surrounding the acceptor residue contacts the kinase active site (Songyang et al., 1994).

The specificity of protein kinases is dominated by acidic, basic, or hydrophobic residues adjacent to the phosphorylated residue, but the large variation makes it difficult manually to inspect protein sequences and predict the location of biologically active sites. This prompted us to investigate if the fuzzy sequence patterns can be recognized using artificial neural networks techniques. Neural networks are capable of classifying even highly complex and non-linear biological sequence patterns, where correlations between positions are important. The network recognizes the patterns seen during training, and retains the ability to generalize and recognize similar, but non-identical patterns. Artificial neural networks have been extensively used in biological sequence analysis Wu 1997, Baldi and Brunak 1998. Since determinants of phosphorylation sites probably are no longer than about ten residues, most local sequence alignment tools, such as BLAST and FASTA, will not be useful for detecting phosphorylation sites due to a large number of irrelevant hits in the protein databases, even to non-phosphorylated proteins.

The related proteins p300 and CBP (CREB (cAMP-response-element-binding)-binding protein) integrate molecular signals at the level of gene transcription and chromatin modification. p300 and CBP interact with transcription factors CREB, Jun and Fos, viral oncoproteins E1a and SV40 large T antigen, and kinases pp90^RSK and cyclin E-complexed cyclin-dependent kinase (CDK)-2 Shikama et al 1997, Ait-Si-Ali et al 1998. These interactions may possibly be regulated by reversible phosphorylation of p300/CBP. We demonstrate that regions of p300/CBP, which have been shown to interact with other molecules, contain probable phosphorylation sites. In addition, we describe sites that possibly are regulated by both phosphorylation and glycosylation by N-acetylglucosamine (GlcNac), a regulatory mechanism described as a yin-yang dynamic phosphorylation/glycosylation Hart et al 1995, Hart 1997.