The Crystal Structure of Domain 1 of Receptor Protein-tyrosine Phosphatase μ*

Receptor-like protein-tyrosine phosphatases (RPTPs) play important roles in regulating intracellular processes. We have been investigating the regulation and function of RPTPμ, a receptor-like PTP related to the Ig superfamily of cell adhesion molecules. Recently, the crystal structure of a dimer of the membrane proximal domain of RPTPα (RPTPα D1) was described (Bilwes, A. M., den Hertog, J., Hunter, T., and Noel J. P. (1996)Nature 382, 555–559). Within this crystal structure, the catalytic site of each subunit of the dimer is sterically blocked by the insertion of the N-terminal helix-turn-helix segment of the dyad-related monomer. It was proposed that dimerization would lead to inhibition of catalytic activity and may provide a paradigm for the regulation of the RPTP family. We have determined the crystal structure, to 2.3 Å resolution, of RPTPμ D1, which shares 46% sequence identity with that of RPTPα D1. Although the tertiary structures of RPTPα D1 and RPTPμ D1 are very similar, with a root mean square deviation between equivalent Cα atoms of 1.1 Å, the quaternary structures of these two proteins are different. Neither the catalytic site nor the N-terminal helix-turn-helix segment of RPTPμ D1 participates in protein-protein interactions. The catalytic site of RPTPμ D1 is unhindered and adopts an open conformation similar to that of the cytosolic PTP, PTP1B (Barford, D., Flint, A. J., and Tonks, N. K. (1994) Science 263, 1397–1404). We propose that dimerization-induced modulation of RPTP activity may not be a general feature of this family of enzymes.

Numerous cellular events are regulated by reversible phosphorylation of tyrosine residues, including growth, differentiation, the cell cycle, cell-cell adhesion, and cell-matrix contacts (1). Phosphotyrosine levels are controlled by the coordinated and competing actions of protein-tyrosine kinases and proteintyrosine phosphatases (PTPs). 1 The PTPs comprise a diverse family of transmembrane receptor-like and nontransmembrane, cytosolic enzymes (1). Receptor-like PTPs are likely to play crucial roles in transducing transmembrane signals and evidence is accumulating to link these enzymes to the control of phenomena mediated by cell adhesion (1). By interacting with ligands through their extracellular domains, the intracellular catalytic activity of RPTPs may be regulated, hence controlling the levels of cellular phosphotyrosine. The receptor-like PTPs RPTP, RPTP, and RPTP, which have extracellular segments sharing structural similarities with Ig superfamily cell adhesion molecules, have been shown to mediate cell aggregation via homophilic interactions (2)(3)(4). For RPTP, the homophilic binding site(s) reside on the immunoglobulin domain, whereas the intracellular segment consists of a juxtamembrane, cadherin-like domain, and two PTP domains (5). A role for RPTP in regulating cell junctions and cytoskeletal organization is suggested by the finding that the RPTP is associated with cadherin-catenin complexes at adherens junctions (6). Similar results have been demonstrated for RPTP (7) and RPTP (4).
The crystal structure of PTP1B in a complex with a phosphotyrosyl peptide substrate (8), together with the results of a number of kinetic studies (9), have revealed important insights into the catalytic mechanism of members of the PTP family. The signature motif, (I/V)HCXAGXXR(S/T)G, which defines this family of enzymes, contains the catalytically essential Cys and Arg residues. It forms a rigid cradle structure that coordinates binding of the phosphate moiety of the substrate and positions the cysteine residue to act as a nucleophile. This motif lies at the base of a cleft on the surface of the enzyme, which is defined at one end by the invariant residue Tyr 46 and at the other by Phe 182 , which is in a loop containing the invariant general acid Asp 181 within the highly conserved motif WPD. It is the depth of the cleft, equivalent to the length of a phosphotyrosine side chain, that is a major determinant of the specificity of these enzymes for Tyr(P) substrates, since the side chains of Ser(P) and Thr(P) are too short to permit access of the phosphate to the catalytic cysteine. Binding of substrate induces a profound conformational change in the enzyme such that on engagement of Tyr(P) the WPD loop flips down onto the Tyr(P) side chain (8). This creates a predominantly hydrophobic recognition pocket that surrounds and interacts with the phenyl ring of the Tyr(P) side chain and allows Asp 181 to act as a catalytic acid. Catalysis proceeds via nucleophilic attack by the invariant Cys (Cys 215 ) on the substrate phosphorous atom and protonation of the tyrosyl-leaving group of the substrate by Asp 181 resulting in formation of a cysteinyl-phosphate covalent intermediate. The reaction is then completed by hydrolysis of the covalent intermediate by a water molecule, possibly activated by Asp 181 (10).
Recently, a model for the regulation of RPTP activity was proposed by Bilwes et al. (11) on the basis of their crystal structure of domain 1 of RPTP␣. In contrast to the cytosolic * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The atomic coordinates and structure factors (PDB ID code 1rpm (coordinates); PDB ID code r1rpmsf (for structure factor)) have been deposited in the Protein Data Bank, Brookhaven National Laboratory, Upton, NY.
§ PTPs, PTP1B and Yersinia PTP, which are monomeric (12,13), the crystal structure of domain 1 (D1) of RPTP␣ revealed a homodimeric organization that would result in an inhibited PTP (8,11). The dimer results from a structural segment at the N terminus of the RPTP␣ D1 catalytic domain, consisting of a turn connecting helices ␣1Ј and ␣2Ј that wedges into the dyadrelated catalytic site. The insertion of residues of the helixturn-helix segment of one subunit into the catalytic site of the dyad-related subunit would be predicted to inhibit catalytic activity. This occurs first, as a result of sterically blocking substrate access to the catalytic site, and second, by constraining the WPD loop in an open conformation that is unable to adopt the closed conformation necessary for catalysis (8). Distinct features of tertiary structure of RPTP␣ D1 that are absent from PTP1B are thought to facilitate this dimeric interaction. These include first, a two-residue insertion in the loop connecting ␣1Ј and ␣2Ј, that inserts into the dyad-related catalytic site, and second, a ␤-strand (␤x), immediately N-terminal to ␣1Ј that forms a distinctive two-strand ␤-sheet with ␤y. The ␤sheet between ␤x and ␤y, together with conserved residues within ␣1Ј and ␣2Ј, contribute to defining the conformation of the helix-turn-helix segment of RPTP␣ D1 (11).
Interestingly, a sequence analysis of the PTP family suggested, on the basis of residue conservations, that the tertiary structural features of RPTP␣ D1 would be shared with D1s of other RPTPs but not cytosolic PTPs, such as PTP1B, or D2 of RPTPs (11). The crystal structure of RPTP␣ D1 lead Bilwes et al. (11) to propose a dimerization model for the regulation of RPTP␣ and, as a result of the sequence similarity with other RPTP D1s, it was suggested that this model could provide a paradigm for the regulation of RPTPs generally (11). The model proposes that ligand binding to the extracellular segment would modulate dimerization and hence the catalytic activity of the PTP domain.
We have been pursuing an investigation of the function of the Ig superfamily cell adhesion molecule related-phosphatase RPTP. At present little is known about the regulation of RPTP activity. Although it associates through its intracellular segment with the cadherin-catenin complex (6), the effects on PTP activity remain unclear. We have initiated a structural analysis of RPTP to provide further insights into the mechanisms for control of its enzymatic activity. In this regard, the possibility that the structure of RPTP␣ is indicative of a general mechanism for ligand-induced inhibition of RPTP activity is exciting. Here, we describe the crystal structure of domain 1 of human RPTP and show that although it is present in the crystal as a dimer, unlike RPTP␣ D1, the active site of RPTP is in an open, uninhibited conformation. Therefore our data indicate that regulation of RPTP activity by dimerization is not a general feature of these enzymes.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-Residues 874 -1168 of human RPTP D1 were cloned into a modified version of the pET28a vector (Novagen Inc.) 2 that incorporates an N-terminal hexa-His tag immediately preceding Gly 874 . The protein was overexpressed in Escherichia coli (strain B834) grown at 18°C for 16 h. Protein purification was achieved with the following steps. Bacterial cells were lysed in 25 mM Tris⅐HCl (pH 8.0), 500 mM NaCl, 20 mM imidazole, 2 mM benzamidine, 0.01 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml aprotinin, 1 g/ml pepstatin, 1 g/ml DNase, 5 g/ml lysozyme and the cleared lysate loaded onto a nickel-nitrilotriacetic acid-agarose column Structure Determination-X-ray crystallographic data were collected to 2.3-Å resolution and processed using DENZO and SCALEPACK (14). The structure was determined by molecular replacement using the CCP4 integrated version of AMoRe (15, 16) using a polyalanine model of RPTP␣ (11), including side chains of residues identical with those of RPTP, as a search model. Crystallographic refinement was performed using X-PLOR (17), applying strict noncrystallographic symmetry (NCS) restraints. Calculated phases using the model were improved by several cycles of combined solvent flattening (solvent content 0.50) and NCS averaging using the PHASES package (18). Iterative rounds of simulated annealing refinement, combined with NCS averaging and solvent flattening, were applied to improve the quality of the maps and the protein model. During the final stages of refinement, the NCS restraints were removed. Analysis of the coordinates and electron density maps was performed with O (19). The final model consists of residues 879 -1156 of RPTP D1 that are visible within the electron density and 340 water molecules (Table I). The structure of RPTP␣ D1 consists of residues 877-1156 (RPTP numbering (20)).

RESULTS AND DISCUSSION
In the crystal structure of RPTP D1, the subunits associate to form a dimer with 2-fold symmetry (Fig. 1). It is most likely that the dimerization of RPTP D1 is a consequence of crystallization, since size exclusion chromatography studies revealed that the protein was monomeric with an apparent molecular mass of ϳ28 kDa in concentrated solutions of ϳ7.5 mg/ml (Fig. 2). This is in contrast to RPTP␣ D1, which forms monomers, dimers, and higher oligomers in solution at protein concentrations of 0.1-5.0 mg/ml (11). The oligomeric state of RPTP D1 is entirely different from that of RPTP␣ D1, since the dimer interface of RPTP D1 does not include the Nterminal helix-turn-helix segment or the catalytic site (Fig. 1). Crystal packing analysis also shows no obstruction of the catalytic site of RPTP D1. The interactions between the two 2 D. Barford   dimeric subunits of RPTP are hydrophobic in nature and mediated by residues Thr 1025 and Ile 1027 from one subunit and residue Ile 1050 and the aliphatic moeity of Glu 1052 from the other. These residues are not conserved within other RPTPs (Fig. 3). Within RPTP D1, the catalytic site is unhindered and adopts an open conformation similar to that of the apo form of PTP1B (12). Residues of RPTP, equivalent to those of the RPTP␣ helix-turn-helix that inserts into the dyad related catalytic site, are exposed to solvent (Fig. 1).
In contrast to the differences in quaternary structure between RPTP␣ and RPTP, the tertiary structures of domain 1 of these two RPTPs are very similar with a root mean square deviation between equivalent C␣ atoms of 1.1 Å after superimposition (Fig. 4). The major differences in tertiary structure are present within more mobile surface loops. Of particular note, the tertiary arrangements of the N-terminal helix-turn-helix segment and N-terminal ␤-strand, ␤x, and associated ␤-sheet with ␤y, of RPTP are essentially identical to that of the equivalent structural elements of RPTP␣. These features distinguish RPTP D1s from the catalytic domains and N-terminal flanking sequences of cytosolic PTPs and are important in promoting the dimeric organization of RPTP␣ D1. Interestingly, Asp 227 and Asp 228 of RPTP␣ D1, which are present within the helix-turn-helix segment that inserts into the dyadrelated catalytic site, superimpose less well with equivalent residues of RPTP D1 (Ala 895 and Glu 896 , respectively). The C␣ atoms of these residues differ by 1.8 Å after global superimposition (Fig. 4), although this region of the polypeptide chain is well ordered.
Among RPTPs, conservation of amino acids within the Nterminal helix-turn-helix that contribute to the dimer interface of RPTP␣ D1 is low (Fig. 3). Of the five residues implicated in this dimeric interface, only Glu 234 is conserved between RPTP Bottom, residues from ␤8 through to the WPD loop of the catalytic site. Residues of RPTP D1 and RPTP␣ D1 that form interactions at their respective dimer interfaces are indicated with vertical arrows (top) and stars (bottom), respectively. The residues that form the dimer interface of RPTP D1 are poorly conserved throughout the family, whereas residues of the RPTP␣ D1 interface are poorly conserved within the helix-turn-helix segment, but well conserved within the catalytic site. Invariant residues are in white type on a black background, and highly conserved residues are boxed. D1 and RPTP␣ D1. However, catalytic site residues that form the dimer interface of RPTP␣ D1 are either invariant or highly conserved between different RPTPs (Fig. 3). Thus, it is unlikely that invariant catalytic site residues would have the potential to interact with all of the structurally diverse N-terminal helixturn-helix segments within the RPTP family. Moreover, if this were possible, it would imply that RPTP heterodimers may form as the catalytic sites of any given RPTP would not distinguish between the helix-turn-helix segments of other RPTPs.
The structure of D1 of RPTP␣ determined by Bilwes et al. (11) indicated the existence of a dimer in which a helix-turnhelix segment on the N-terminal side of one catalytic domain occupied the catalytic site of the opposing PTP domain in the dimer. This observation led the authors to propose the exciting possibility that ligand-induced dimerization of RPTPs in general may lead to inhibition of activity. They used this structural model to interpret previous observations regarding the signaling properties of an EGF receptor-CD45 chimera (21). Ablation of expression of CD45, the prototypic RPTP, from T or B cells disrupts normal signaling responses to engagement of antigen receptor. However, in some T cell lines, signaling can be restored by expression of membrane-targeted constructs containing the catalytic domain of CD45. In the instance of the chimeric molecule in which the extracellular and transmembrane segments of CD45 were replaced with those of the EGF receptor, signaling was restored only in the absence of EGF. It was proposed that in the presence of growth factor, dimerization of the chimera is induced with resulting inhibition of PTP activity. However, a direct effect of ligand binding on the catalytic activity of the EGFR-CD45 chimera has not been reported. Our results suggest caution in extrapolating beyond the structure of RPTP␣ to other RPTPs. We show that the N-terminal helixturn-helix of RPTP D1 does not cause formation of an inhibited dimeric state, similar to that of RPTP␣ D1 (11), and therefore the dimeric structure of RPTP␣ D1 may not represent the situation for all members of the RPTP family. Dimerization may be a feature of only a subfamily of RPTPs, such as RPTP␣ and its close relatives, for example CD45, which share sequence similarity in those residues participating at the N-terminal helix-turn-helix dimer interface (Fig. 3). Our results do not exclude the possibility that in the context of the plasma membrane the full-length form of RPTP may form an inhibited dimer. However, the fact we do not observe dimers of RPTP D1 in solution, unlike that for RPTP␣ D1, and that within the RPTP D1 crystal, which would mimic the restricted diffusion of RPTP in the plasma membrane, inhibition of catalytic sites is not observed, suggests that the Nterminal helix-turn-helix of RPTP D1 does not play a role in regulating the activity of this phosphatase.