A MODEL OF CDC25 PHOSPHATASE CATALYTIC DOMAIN AND CDK-INTERACTION SURFACE-BASED ON THE PRESENCE OF A RHODANESE HOMOLOGY DOMAIN

Mammalian Cdc25 phosphatase is responsible for the dephosphorylation o f Cdc2 and other cyclin-dependent kinases at Thr14, and Tyr15, thus ac tivating the kinase and allowing cell cycle progression. The catalytic domain of this dual-specificity phosphatase has recently been mapped to the 180 most C-terminal amino acids. Apart from a CX3R motif, which is present at the active site of all known tyrosine phosphatases, Cdc 25 does not share any obvious sequence similarity with any of those en zymes. Until very recently, the Cdc25 family was the only subfamily of tyrosine phosphates for which no three-dimensional structural data we re available. Using the generalized profile technique, a sensitive met hod for sequence database searches, we found an extended and highly si gnificant sequence similarity between the Cdc25 catalytic domain and s imilarly sized regions in other proteins: the non-catalytic domain of two distinct families of MAP-kinase phosphates, the non-catalytic doma in of several ubiquitin protein hydrolases, the N and C-terminal domai n of rhodanese, and a large and heterogeneous groups of stress-respons e proteins from all phyla. The relationship of Cdc25 to the structural ly well-characterized rhodanese spans the entire catalytic domain and served as template for a structural model for human Cdc25a, which is f undamentally different from previously suggested models for Cdc25 cata lytic domain organization. The surface positioning of subfamily-specif ic conserved residues allows us to predict the sites of interaction wi th Cdk2, a physiological target of Cdc25a. Based on the results of thi s analysis, we also predict that the budding yeast arsenate resistance protein Acr2 and the ORF Ygr203w encode protein phosphatases with cat alytic properties similar to that of the Cdc25 family. Recent determin ation of the crystal structure of the Cdc25a catalytic domain supports the validity of the model and demonstrates the power of the generaliz ed sequence profile technique in homology-based modeling of the three- dimensional structure of a protein having a weak but significant seque nce similarity with a structurally characterized protein. (C) 1998 Aca demic Press.