The ras superfamily of GTP-binding proteins: guidelines on nomenclature.

Over the past several years, a large superfamily of monomeric GTP-binding proteins structurally related to the ras oncogene proteins has been described (1-4). The number of members of this superfamily identified has increased dramatically in recent years, and the number of descriptive terms or names used to denote these proteins has increased almost as dramatically as well. During the recent FASEB Summer Research Conference on “Low Molecular Weight GTP Binding Proteins:’ an informal session on nomenclature was held in an effort to establish guidelines for the literature in this rapidly expanding field, as well as in the naming of new genes and proteins. The need for such a session clearly results from the abundance of sequence information on new genes and proteins and the relative paucity of information regarding protein function, a common theme in the biological sciences today. The discussion focused on three main questions: 1) what should we call the superfamily of 20to 29-kDa GTP-binding proteins, 2) what proteins should be considered members of this superfamily, and 3) how many subfamilies currently comprise this superfamily? Although no unanimous decisions were reached, several consensus agreements were reached and are summarized below. It was generally agreed that the best name for the large number (currently containing more than 50 members) of structurally related small GTP-binding proteins is the “ras superfamily” It was also recognized that in discussing or writing about this superfamily it is still necessary to use more descriptive phrases, and low molecular weight GTP-binding proteins or small GTP-binding proteins are suggested. However, the term G-protein should be reserved for the family of heterotrimeric signal-transducing proteins with GTP-binding (a) subunits in the 39to 52-kDa range. The term ras-related, although less accurate or informative, was also considered to be acceptable. Entry into the ras superfamily is dependent on three criteria: 1) protein size in the 20to 29-kDa range; 3) the presence of consensus motifs for GTP-binding proteins (actual demonstration of specific GTP binding preferred); and 3) occurrence of sequence motifs that further characterize subfamilies (see below and Table 1). With regard to the naming of new proteins, everyone is encouraged to use existing nomenclature for highly related proteins. For example, a new protein sharing 85-95% identity to the racl protein should be designated raciB, whereas one showing 35-85% identity should be designated the next available number (e.g., rac3). A combination of computer alignments and visual inspection should be sufficient to determine if such a relationship exists. The subdivision of the ras superfamily into subfamilies is again made on structural grounds rather than on functional ones (1-4). It was agreed that the ras superfamily can be subdivided into five subfamilies. Three closely related major subfamiies are the ras and related proteins, the rho and related proteins, and the rab and related proteins. A more distantly related fourth major subfamily is composed of arf and related proteins. Finally, the 1C4 and ran proteins have been tentatively designated as a fifth subfamily of ran and related proteins. The principal determinant of this division is based on sequence alignments (see Fig. I in ref 1). Alignment of a member of one subfamily with a member of another subfamily reveals about 30% sequence identity, whereas this value is generally at least 40% for two sequences within the same subfamily and often quite higher. One useful hallmark of the rho proteins is the presence of a 12 amino acid insertion between residues 122 and 123 of an aligned ras protein (1). Other useful indicators for subfamily assignment occur at the NH2 and COOH termini where subfamilies tend to have well-conserved motifs. For example, all currently known arf proteins have a glycine at position 2, are N-myristoylated, and lack a cysteine near the COOH terminus (3). Three subfamilies contain sites for COOH terminal, but not NH2-terminal, processing but still fall into distinct patterns. Members of the ras and rho subfamiies end exclusively with a CAAX motif, whereas most rab subfamily members end with either CC or CXC motifs. Finally, the ran subfamily lacks any apparent NH2or COOHterminal lipid modification signal sequences. A related issue, the terminology of the known regulatory factors associated with the various members of the ras superfamily, was also addressed. A clear consensus was obtained on the use of the terms GTPase-activating protein (GAP), GDP dissociation inhibitor (GDI), and GDP/GTP dissociation stimulator (GDS) for these proteins.