Cell, vol. Ss. 1013-1016, September 22, 1989, Copyright0 1989by Cell Press Pmtein-Tyrosine Phosphatases: The Other Side of the Coin Minireview Tony Hunter Molecular Biology and Virology Laboratory The Salk Institute San Diego, California 92138 For many years protein phosphatases have been the poor relation in the protein phosphorylation field, but that era appears to be at an end. Although the regulation of protein function by phosphorylation necessarily requires protein phosphatases in addition to protein kinases, the recent discovery that certain cell cycle genes and a transcrip- tional regulatory gene encode protein-serine phospha- tases indicates that protein phosphatases do not simply constitutively reverse the effects of protein kinases, but rather themselves play central and specific roles in cellu- lar physiology (Cyert and Thorner, 1989). To add to the excitement over protein-serine phospha- tases, there have been equally important recent advances in our understanding of protein-tyrosine phosphatases (PTPases). Prior to last year little was known about the structure of PTPases. In contrast, protein-serine phospha- tases have been extensively studied (for review, see Co- hen, 1989, and Cohen and Cohen, 1989). Two of these, PP-2A and PP-PB, do have weak activity toward phos- photyrosine as well as phosphoserine, but because all the F’TPases characterized to date have absolute specificity for phosphotyrosine, the PTPases clearly constitute a sep- arate class of enzyme (for review, see Tonks et al., 1989). Recent interest in PTPases was sparked by Tonks, Fischer, and colleagues, who purified to homogeneity a 35 kd solu- ble PTPase from placenta, PTPase 1B (Tanks et al., 1988a, 1988b), and obtained a nearly complete protein sequence (Charbonneau et al., 1988). Surprisingly, this sequence showed no relationship to any of the protein-serine phos- phatase catalytic subunits. However, it had an unexpected similarity to both copies of the 300 amino acid imperfect tandem repeat in the cytoplasmic domain of the leukocyte cell surface protein CD45 (also known as LCA, Ly-5, or T200; for review, see Thomas, 1989). This immediately suggested that CD45 might have PTPase activity. Rapid verification of this idea came with the demonstration that immunoaffinity-purified CD45 has PTPase activity (Tonks et al., 1988c). That CD45 has intrinsic PTPase activity has recently been confirmed by two groups who have expressed the cytoplasmic domain of CD45 in nonmammalian sys- tems. They find that the cytoplasmic domain of CD45 can dephosphorylate phosphotyrosyl-angiotensin when this domain is expressed either as a secreted protein in a baculovirus system (Ostergaard et al., 1989) or as a solu- ble protein in E. coli (Streuli et al., 1989). The structure of CD45, which consists of a large exter- nal N-terminal domain, a single membrane spanning re- gion, and a large C-terminal cytoplasmic region, is in many ways analogous to that of the growth factor receptor protein-tyrosine kinases (PTKs; see figure). This similarity suggests that the activity of the CD45 cytoplasmic domain might be regulated by an extracellular ligand in a manner similar to the regulation of the growth factor receptor PTK activity. No ligand is known for CD45, but the cell type- specific polymorphism in the structure of the CD45 exter- nal domain strongly implies that this domain has an im- portant function (see below). By analogy with the large family of receptor F’TKs, one might predict that there will be a family of “receptor PTPases” regulated by extracellu- lar ligands. There is already evidence that this is the case. Using a CD45 cDNA probe for low stringency screening, Saito and colleagues were able to isolate from a placental cDNA library another putative receptor PTPase, termed LAR (Streuli et al., 1988). In its cytoplasmic domain the predicted structure of the LAR protein is similar to that of CD45, including the tandem repeat, but the external do- mains of the two proteins are completely unrelated (see figure). The large externai domain of LAR shows similarity to both the lg-like and the fibronectin type Ill-like repeats of the neural adhesion molecules, N-CAM and fasciclin II. This similarity suggests that LAR may have a function in cell adhesion or cell-cell interaction. N-CAM molecules are capable of homotypic interaction. LAR has not yet been tested for this property, but one could speculate that LAR might be involved in cell-cell signaling through the interaction of two LAR molecules on different cells. Unlike CD45, whose expression is restricted to leuko- cytes, LAR is expressed in many tissues and cell types. Interestingly, T cells express both CD45 and LAR (Streuli et al., 1988). In this sense the receptor F’TPases may prove to be like the receptor PTKs, with a single cell displaying a repertoire of receptor PTPases available for selective triggering by a variety of external stimuli. In addition to a multiplicity of receptor PTPases, it is al- ready clear that there is more than one soluble nonrecep- tor PTPase. Over the past five years several groups have reported the existence of multiple PTPase activities re- solved by column chromatography of soluble cell or tissue extracts (for review, see Lau et al., 1989). For example, In- gebritsen and co-workers recently identified seven major peaks of PTPase activity in soluble brain extracts (Jones et al., 1989). Although this heterogeneity could be due to a single catalytic subunit associated with different regula- tory subunits, there is likely to be more than one non- receptor PTPase catalytic subunit among the brain PTP- ases, based on their differential sensitivities to polyanion inhibitors; one of these may well be PTPase 1B. Structural evidence for multiple PTPases comes from the peptide mapping studies of Tonks et al. (1988b), which concluded that placental PTPase 1A is distinct from that of PTPase 1B. In addition, Cool et al. (1989) have recently isolated a human T cell cDNA clone using oligonucleotides based on the PTPase 1B sequence. The predicted product of the T cell cDNA clone, which is about 10 kd larger than PTP- ase 1B due to a C-terminal extension, is only -85% identi- cal in sequence, and it is evidently the catalytic subunit of a soluble PTPase distinct from PTPase 1B. Analysis of monkey tissue RNAs indicates that expression of this PTPase is not restricted to T cells. The larger size of the “T cell” F’TPase could mean that the purified placental