Protein-tyrosine phosphatases.

Phosphate ester formation and hydrolysis are likely the most crucial chemical reactions carried out by living organisms. Polymerization of the genetic material involves these reactions as does activation of substrates in many biosynthetic pathways. Phosphate ester chemistry is central to the properties of biological membranes and often mediates interactions between macromolecules. Formation of phosphate monoesters on proteins, phosphorylation, and hydrolysis of these protein-associated phosphate monoesters, dephosphorylation, are two of the most widely studied cellular reactions involving phosphate esters. These reactions regulate numerous biological processes (1) and are catalyzed by protein kinases and protein phosphatases, respectively. In eukaryotic organisms, protein kinases and phosphatases are typically categorized based upon substrate specificity. For the protein kinases, this classification system defines two groups: the serinekhreonine kinases and the tyrosine kinases. In spite of differences in substrate specificity, all of the eukaryotic protein kinases share amino acid sequence identities as well as similarities in three-dimensional structure (2, 3). Phosphatases represent a diverse family of biological catalysts. For purposes of introduction, we will focus on three groups: the nonspecific phosphatases, the phosphoprotein (serine/threonine) phosphatases (PPases),’ and the protein-tyrosine phosphatases (PTPases). Unlike the protein kinases, each of these groups lacks sequence and structural similarity with the other two groups. The nonspecific phosphatases include enzymes such as the alkaline phosphatases and the acid phosphatases. These enzymes are often capable of hydrolyzing both proteinaceous and non-proteinaceous phosphate monoesters. The nonspecific phosphatases typically function in degradative or catabolic pathways and are b lieved to serve as “phosphate scavengers.” These diverse enzymes do not share amino acid sequence similarities with one another. The catalytic mechanisms of the nonspecific phosphatases have been widely studied (4-6). In brief, phosphate ester hydrolysis by alkaline phosphatases proceeds through a phosphoserine intermediate (7) while phosphate ester hydrolysis by at least some acid phosphatases proceeds through a phosphohistidine intermediate (8). In contrast with the nonspecific phosphatases, most of the PPases share amino acid sequence similarity with one another (9). The PPases are distinct in amino acid sequence from both the nonspecific phosphatases and the PTPases (10). The PPase catalytic mechanism does not appear to involve a phosphoenzyme intermediate (11). Recently, a sequence motif has been described that is found in all type 1,2A, and 2B PPases and a variety of otherwise unrelated phosphoesterases (12) (Fig. 1). These latter phosphoesterases include enzymes involved in sphingomyelin phosphate hydrolysis and the debranching of RNA. The “phosphoesterase signature motif’ includes a number of amino acid residues that have a demonstrated role in metal ion binding, substrate binding, and/or catalysis (12). The widespread distribution of this motif may suggest that enzymes which display the motif employ a common catalytic mechanism.

Phosphate ester formation and hydrolysis are likely the most crucial chemical reactions carried out by living organisms. Polymerization of the genetic material involves these reactions as does activation of substrates in many biosynthetic pathways. Phosphate ester chemistry is central to the properties of biological membranes and often mediates interactions between macromolecules. Formation of phosphate monoesters on proteins, phosphorylation, and hydrolysis of these protein-associated phosphate monoesters, dephosphorylation, are two of the most widely studied cellular reactions involving phosphate esters. These reactions regulate numerous biological processes (1) and are catalyzed by protein kinases and protein phosphatases, respectively. In eukaryotic organisms, protein kinases and phosphatases are typically categorized based upon substrate specificity. For the protein kinases, this classification system defines two groups: the serinekhreonine kinases and the tyrosine kinases. In spite of differences in substrate specificity, all of the eukaryotic protein kinases share amino acid sequence identities as well as similarities in three-dimensional structure (2,3). Phosphatases represent a diverse family of biological catalysts. For purposes of introduction, we will focus on three groups: the nonspecific phosphatases, the phosphoprotein (serine/threonine) phosphatases (PPases),' and the protein-tyrosine phosphatases (PTPases). Unlike the protein kinases, each of these groups lacks sequence and structural similarity with the other two groups.
The nonspecific phosphatases include enzymes such as the alkaline phosphatases and the acid phosphatases. These enzymes are often capable of hydrolyzing both proteinaceous and non-proteinaceous phosphate monoesters. The nonspecific phosphatases typically function in degradative or catabolic pathways and are believed to serve as "phosphate scavengers." These diverse enzymes do not share amino acid sequence similarities with one another. The catalytic mechanisms of the nonspecific phosphatases have been widely studied (4-6). In brief, phosphate ester hydrolysis by alkaline phosphatases proceeds through a phosphoserine intermediate (7) while phosphate ester hydrolysis by at least some acid phosphatases proceeds through a phosphohistidine intermediate (8).
In contrast with the nonspecific phosphatases, most of the PPases share amino acid sequence similarity with one another (9). The PPases are distinct in amino acid sequence from both the nonspecific phosphatases and the PTPases (10). The PPase catalytic mechanism does not appear to involve a phosphoenzyme intermediate (11). Recently, a sequence motif has been described that is found in all type 1,2A, and 2B PPases and a variety of otherwise unrelated phosphoesterases (12) (Fig. 1). These latter phosphoesterases include enzymes involved in sphingomyelin phosphate hydrolysis and the debranching of RNA. The "phosphoesterase signature motif' includes a number of amino acid residues that have a demonstrated role in metal ion binding, substrate binding, and/or catalysis (12). The widespread distribution of this motif may suggest that enzymes which display the motif employ a common catalytic mechanism.
* This minireview will be reprinted in the Minireview Compendium, which will be available in December, 1994. This work was supported in part by grants from the National Institutes of Health (to J. E. D PTPases in their ability to utilize phosphoserine and phosphothreo-nine2 as substrates in addition to phosphotyrosine. Without exception, all PTPases contain an active site signature motif, (W)HCXAGXGR(S/T)G, which harbors the catalytic cysteinyl residue involved in formation of a phosphoenzyme reaction intermediate (14,15).
The PTPases regulate cell growth and proliferation, the cell cycle, and cytoskeletal integrity in response to a variety of external stimuli (13). The relative level of tyrosine phosphorylation is balanced by the antagonistic actions of the protein-tyrosine kinases and the PTPases. The PTPases can function as tumor suppressors in some systems (16) and have a demonstrable role in cell cycle regulation (17,18) as well as T-cell activation (19). PTPases also seem to play key roles in ontogeny (20, 21). Quite surprisingly, the PTPases play a role in bacterial pathogenesis. The Yersinia genus, which is responsible for the bubonic plague or the Black Death, encodes a PTPase that is essential for virulence and pathogenesis (14,22). The structures of the Yersinia PTPase catalytic domain (23) and another intracellular PTPase, PTPlB (24), have recently been reported.
While the nonspecific phosphatases and the PPases also play critical roles in all biological systems and constitute important subjects for study, a full accounting of these enzymes would be beyond the scope of this review. Instead, we will focus on the protein-tyrosine phosphatases. A number of other excellent reviews have appeared in recent years, which discuss these enzymes (10,13,(25)(26)(27)(28)(29)(30)(31)(32)(33). The purpose of the present review is t o introduce the reader to current developments within the field of PTPase structure, function, and biology. Although the first reported purification of a PTPase, PTPlB (341, appeared only 6 years ago, progress in this field has been rapid. It is not possible in the format of a minireview to cite all of the findings that have taken place. We apologize to those whose work has been omitted.

Structure and Catalytic Properties of the PTPases
The Yersinia PTPase and PTPlB are derived from very different sources, a bacterium and a mammal, respectively. In addition, the two enzymes share only 20% sequence identity. Nonetheless, it is clear from structural elucidation that these molecules share close similarities in secondary and tertiary structure (23,24). The catalytic domains of the Yersinia PTPase (residues 163-468) and of PTPlB (residues 1-321) consist of a twisted, mixed p-sheet structure flanked by a-helices. Structural features within the molecular crevice that makes up the active site of each PTPase are remarkably similar. The PTPase signature motif of the Yersinia PTPase is located on a loop structure C-terminal to a p sheet (residues 402-406) and on the first turn of the a helix following the loop (residues 407-410). Residues 403-410 form the binding loop for the phosthreonine, or phosphotyrosine as phosphatase substrates, we are r e t ! % $ ? : these amino acid residues found within the primary structure of a protein or peptide, not to the free amino acids. phate present on the substrate (the P-loop), with the nucleophilic cysteinyl residue, Cys-403, centered within the loop (23) (Fig. 2). A similar structure is present in the PTPlB active site signature motif (24). There are several elements of the PTPase catalytic mechanism that are at least partially explained by structural analysis. One of these elements is activation of the catalytic cysteinyl residue. Proteinaceous sulfhydryls typically have a dissociation constant (PR,) of approximately 8.5. Cys-403 of Yersinia PTPase has an apparent PK, of 4.7, suggesting active site stabilization of the corresponding thiolate anion at neutral pH (35). The thiolate anion of Cys-403 in Yersinia PTPase seems to be stabilized by an intricate and extensive network of hydrogen bonds. The Cys-403 sulfur atom is less than 4 A away from 5 amide nitrogens located within the P-loop structure (23). Additionally, these P-loop amide nitrogens are all coupled electrostatically through their carbonyl oxygens to hydrogen bonds radiating away from Cys-403. The net result is a series of microdipoles, all of which have their 8' ends oriented toward the Cys-403 thiol. The combined effects of the previously discussed hydrogen bonds and the P-loop microdipoles would appear to contribute to lowering the apparent PK, of Cys-403. The PTPlB structure shows a similar network of hydrogen bonds (24). Barford and colleagues (24) have also noted that PTPase signature sequence residue Arg-221 of PTPlB is within 3.0 A of the thiol of the catalytic cysteine. The positioning of this signature sequence arginine residue may also play a role in stabilizing the reacting thiolate ion prior to nucleophilic attack.
Another key aspect of PTPase catalysis clarified by the structural studies is the role of a specific general acid and general base in the catalytic mechanism. The anion tungstate has aided these studies, When the Yersinia PTPase was crystallized in the presence of sodium tungstate, a competitive inhibitor, the structure revealed that tungstate was bound in the active site of the enzyme. The structure also revealed that the carboxylate of Glu-290 forms a bidentate salt bridge with the guanidinium group of Arg-409 (23). A similar structure is also evident between the corresponding glutamic acid and arginine residue in the PTPlB structure (24).  in the Yersiniu PTPase is an invariant glutamic acid residue (the corresponding residue is Glu-215 in PTPlB) and has been proposed to function as a general base (36). The interactions between Yersinia PTPase residues Glu-290 and Arg-409 also orient the Arg-409 guanidinium group so that it can form a bidentate salt bridge with two oxygens of tungstate (23) (Fig. 2). The salt bridges An invariant acidic residue (Asp-356 in the Yersinia PTPase) has been proposed to function as a general acid in the catalytic mechanism (36). Interestingly, in the unliganded Yersinia PTPase structure, Asp-356 is greater than 10 A from the phosphate binding site.
In the tungstate-complexed structure, the loop that contains the Asp-356 residue has moved like a "flap" to "cover" the active site.
The C-cu of Asp-356 itself moves 6 A toward the active site upon tungstate binding, positioning its carboxylate 3.5-3.8 A away from the only unliganded oxygen of the bound tungstate (23) (Fig. 2). This tungstate oxygen is directly opposite the thiol of Cys-403. It is most likely that this tungstate oxygen corresponds to the scissile monoester bond in a phosphotyrosine substrate and is therefore ideally positioned to accept a proton during catalysis from Asp-356.
More surprising than the structural similarities between PTPlB and the Yersinia PTPase are the structural similarities between these two molecules and low molecular weight phosphatases (37,38). The low molecular weight phosphatases (LMWPs) fall within our nonspecific phosphatase designation and are capable of hydrolyzing phosphotyrosine. While PTPlB and the Yersinia PTPase share 20% sequence identity, the LMWps share no discernible sequence identity with any of the PTPases. Nonetheless, the architecture of the PTPase active site and the LMWP active site shares a number of common features. The LMWP active site is comprised of a P-loop that contains an invariant, nucleophilic cysteinyl residue and an invariant arginyl residue (37). The nucleophilic cysteine is activated by P-loop amide dipoles. In addition, the LMWp catalytic structure contains an aspartic acid residue on a loop adjacent to the P-loop, which is positioned correctly to function as a general acid in the catalytic mechanism (37). In light of the overall lack of sequence identity, these striking structural similarities between the PTPases and the LMwps may well represent an example of convergent evolution.

Receptor-like and Intracellular PTPaee Biology
A subset of the receptor-like PTPases has extracellular domains that share sequence similarities with the cell adhesion molecules (CAMS). One potential implication of this architecture is that these PTPases may participate in cell-cell interactions and perhaps im-

m a s e s
pact upon cellular signal transduction as a result. Two of the most carefully studied of these CAM-like PTPases are PTPp (Fig. 3) and P T P K , mammalian PTPases that both display a single extracellular immunoglobulin-like domain and four fibronectin type I11 repeats (39,40). In addition, these molecules possess a motif at their amino termini that may be important for formation of disulfide bridges (41). Expression of PTPp in the insect cell line SF9 leads to cell aggregation (42,43). Deletion studies revealed that the extracellular domain of the PTPase was the mediator of the cell-cell interactions. The extracellular domain of PTPp was also found to be capable of self-association ("homophilic" interactions) in the absence of the transmembrane domain of the PTPase. The purified extracellular domain conjugated to chromatography beads also caused bead aggregation. The closely related P T P K confers the same properties upon cells in which it is expressed (44). Interestingly, in spite of the similarities in the extracellular architecture of PTPp and P T P K , when cells expressing P T P K were mixed with cells expressing -, only homophilic adherence occurs (44). To date, changes in phosphatase activity caused by this sort of homophilic binding have not been detected (42,43,45). In this regard, it should be noted that cellular proliferation is often associated with tyrosine phosphorylation. Cell-cell contact is often associated with cell quiescence and a decrease in tyrosine phosphorylation (2). Further elucidation of the role of CAM-like PTPases in these latter processes is clearly of importance.
The murine enzyme PTPp defines another subclass of receptorlike PTPases. The extracellular domain of this subclass of PTPases is characterized by an amino-terminal carbonic anhydrase-like domain, fibronectin type I11 repeats, and an additional cysteine-free domain (46) (Fig. 3). The extracellular domain of PTPp has recently been shown to be a chondroitin sulfate proteoglycan (47). Further, the extracellular domain of PTPp has been demonstrated to interact specifically with the extracellular matrix protein tenascin. As Barnea and colleagues (47) point out, tenascin has been developmentally associated with the migration pathways of neural crest cells and with outgrowing peripheral nerves. In addition, tenascin seems to define boundaries between regions of the developing brain. The identification of tenascin as the ligand for PTPp may have profound implications in the effort to define factors that regulate development of the mammalian central nervous system. The role of tenascin in regulating PTPp activity has not been elucidated, but it clearly represents an important focus for future research. A good deal of work has focused on defining the substrate specificity determinants for individual PTPases (15,4843). Synthetic, tyrosine-phosphorylated peptides based upon tyrosine kinase receptor autophosphorylation sites have been employed in much of this work. Interestingly, k,JK,,, values for a given enzyme and multiple substrates do not often M e r by more than 2 orders of magnitude. This may suggest that PTPase substrate specificity is governed by factors other than substrate primary amino acid sequence alone. One additional factor in regulating PTPase substrate specificity is localization of the catalyst. This is of particular importance for the intracellular class of PTPases. A review of this topic has recently appeared (13). Structural motifs within the intracellular PTPases serve to localize these enzymes to a variety of places within the cell. These include: 1) membranes, 2) the nucleus, 3) cytoskeletal elements, and 4) tyrosine-phosphorylated signaling proteins. This latter type of targeting has received a great deal of attention recently with the discovery of PTPases that possess Src homology 2 (SH2) domains. The SH2 domain represents a conserved motif of approximately 100 amino acids, which binds to phosphotyrosine residues (54).
Two SH2 domain-containing PTPases have been studied in some detail, SH-PTP1 and SH-PTP2 (Fig. 3). SH-PTP1 is predominantly expressed in hemopoietic cells and is also referred to as PTPlC, HCP, and SHP. SH-PTP2 has a wider tissue distribution and is also referred to as SYP, PTPlD, PTP2C, and SH-PTP3. Both SH-PTP1 and SH-PTP2 possess dual, amino-terminal SH2 domains. These molecules have been shown to interact with autophosphorylated platelet-derived growth factor receptor (PDGFR), epidermal growth factor receptor, and phosphorylated insulin receptor substrate 1 (55-58) via their SH2 domains.
The intracellular location of SH-PTP1 changes in response to extracellular stimulus. Within 1 min of platelet stimulation by thrombin, SH-PTP1 becomes associated with the cytoskeleton (59). This mimics the translocation of the tyrosine kinase pp6OPs" in response to the same stimulus and implicates SH-PTP1 in the complex series of cytoskeletal rearrangements that accompany platelet activation. SH-PTP2, on the other hand, may be involved in tyrosine kinase receptor signal transduction. SH-F'l"2 binds to and becomes phosphorylated by activated PDGFR. Phosphorylated SH-PTP2, in turn, can be bound by the Ras pathway adapter molecule GRB2 (59). The role of SH2-PTP2 phosphatase activity in this signaling cascade remains to be elucidated.
Removal of the SH2 domains from SH-PTP2 results in a 12-45fold increase in phosphatase activity (60)(61)(62). Exogenous SH2 domains reverse this effect (62). These studies illuminate the complex interactions that are likely to occur between the SH2 domains and the catalytic center of the PTPase. The fact that SH2-containing PTPases can be tyrosine phosphorylated further adds to the regulatory complexity. In the case of SH-PTP2, phosphorylation by activated PDGFRp actually enhances phosphatase activity (63).
The importance of SH2-containing PTPases is further underscored by recent developmental studies with Drosophila and mice. The Drosophila gene corkscrew encodes an SH2-containing PTPase, which is required for signal transduction by the tyrosine kinase receptor torso (21). Proper functioning of the torso signal transduction pathway is obligatory for normal formation of anterior and posterior structures during embryogenesis. Homozygously carried defects in the murine homolog of SH-PTP1, Hcph, lead to the motheuten phenotype (20). The motheaten mouse has a profoundly altered immune response, and the animals only survive a few days or weeks beyond birth.
The dual specificity PTPases are unique among the PTPases in their ability to hydrolyze not only phosphotyrosyl but phosphoseryl and threonyl monoesters as well. Dual specificity phosphatase cDNA clones have been isolated from a number of organisms including yeast (63,641, pox viruses (65), and mammals (VHR (66); Pac-1 (67); CLlOO (68)(69)(70)(71)). The time course of CLlOO and Pac-1 mRNA expression during mitotic stimulation is characteristic of immediate-early genes (67,721. Recent work has begun to identify the substrates for this interesting class of molecules. Mitogen-activated protein kinase (MAPK) plays a critical role in a variety of signal transduction pathways. MAPK activity peaks and declines following mitogen stimulation with kinetics reminiscent of immediate early gene mRNA accumulation. MAPK activation is dependent upon phosphorylation of a neighboring tyrosine Minireview: Protein. and threonine residue. Evidence suggests that multiple isoforms of MAPK are expressed within a given cell type, each of which serves a particular cellular function (73). A variety of dual specificity PTPases exhibit activity toward activated MAPK isoforms both i n vitro3 and in vivo (70,74,75). In each instance, dephosphorylation of a MAPK isoform by a dual specificity PTPase leads to loss of MAPK activity. It is interesting that multiple dual specificity PTPases can be found within a given cell type, as can multiple MAPK isoforms. Future experiments should reveal a preference, if any, of particular dual specificity PTPases for particular MAPK isoforms. In this same vein, it is important to recognize that new classes of MAPK-related kinases have recently been reported (76,77). These Jun kinases and stress-activated protein kinases are phosphorylated on nearly juxtaposed threonine and tyrosine residues, like MAF'K. Furthermore, these phosphorylation events are important for kinase activity. It is not beyond the realm of possibility that these molecules will be prove to be substrates for dual specificity PTPases as well. tory at the University of Michigan for their continued contribution and Heidi