A Novel Protein-Tyrosine Phosphatase Related to the Homotypically Adhering κ and μ Receptors

Here we describe a novel member of the receptor-like protein-tyrosine phosphatases (PTPs) termed PTP λ, which is homologous to the homotypically adherent PTPs κ and μ. Murine PTP λ contains MAM, IgG, fibronectin type III, and dual phosphatase domains. As has been demonstrated for PTPs κ and μ, PTP λ mediates homotypic adhesion in vitro, and PTP λ is associated with β catenin in kidney epithelial cells. The extracellular domain of PTP λ is proteolytically processed in cell culture as well as in vivo Northern blot analysis reveals that PTP λ is expressed throughout embryonic development and is predominately found in adult brain, lung, and kidney. In situ hybridization to 15.5-day old rat embryos reveals that PTP λ is expressed in a variety of embryonic neuronal sites as well as in the esophagus, lung bronchiolar epithelium, kidney glomerular epithelium, olfactory epithelium, and various cartilagenous sites. Analysis of neonatal brain demonstrates expression in cells of the hippocampus, cortex, and the substantia nigra. Finally, immunohistochemical analysis reveals expression of this PTP on specific neurons of the spinal cord as well as on isolated cortical neurons.

Tyrosine phosphorylation is induced by a plethora of receptor-like molecules as well as by a wide range of intracellular enzymes. The effects of tyrosine phosphorylation are numerous, and they modulate a range of developmental as well as other cellular operations. Of course, the importance of tyrosine phosphorylation is underlined by the need for mechanisms that carefully regulate the levels of these events. Thus, proteintyrosine kinases represent positive mediators of tyrosine phosphorylation, while protein-tyrosine phosphatases (PTPs) 1 induce the removal of phosphate from tyrosine. The balance of the levels of tyrosine phosphate is thus mediated by the relative activities of these two types of enzymes. It is therefore clear that the mechanisms which regulate cellular function via tyrosine phosphorylation require specific proteins that mediate both the up-regulation as well as the down-regulation of the levels of this modified amino acid.
PTPs represent a growing family of enzymes that are found in both receptor as well as non-receptor forms (1)(2)(3). The re-ceptor PTPs are a highly diverse group that are unified by the inclusion of a hydrophobic domain that disposes them to the plasma membrane of the cell. Recently, the receptor PTPs have been subdivided into eight types based upon their domain content (4). These subtypes all contain one or two PTP domains on their cytoplasmic sides, with a variety of extracellular motifs including heavily O-glycosylated mucin-like domains (for example, CD45), chondroitin sulfate domains (for example, PTP ␥), and short, highly glycosylated segments (for example, PTP ␣). The largest family of PTPs is the family that contains various motifs related to those found in adhesion molecules. These motifs include immunoglobulin-like (IgG) domains and fibronectin type III (FNIII) regions similar to those found in cell adhesion molecules such as ICAM, N-CAM, and Ng-CAM (5). In addition, a subset of these adhesion-like PTPs, including the PTPs and , contain a third domain termed the MAM, for meprin/A5/PTP , motif (6). The MAM motif has been previously shown to be involved with cell-cell recognition in neurons (7)(8)(9). Interestingly, recent data suggest that three of these adhesion-like PTPs appear to be involved with neuronal pathfinding during Drosophila development (10,11).
PTPs and are the receptors that are most well characterized as homotypic adhesion molecules (4,12,13). Thus, a diversity of assays, including cell-as well as molecule-based, have demonstrated that the extracellular domain of these enzymes can bind with high specificity in a homophilic manner (14 -16). Interestingly, mixing experiments have revealed that these related PTPs will not bind to each other in a heterophilic manner, suggesting that the extracellular region is meant to recognize other cells specifically expressing identical receptors, a situation highly reminiscent of the cadherin homotypic adhesion system (17). While the extracellular domains required for this homotypic binding remain controversial, it appears likely that both the MAM motif as well as the IgG region are involved with homophilic interactions (18,19). While these data suggest that these homophilic adhesion enzymes are involved with the recognition of other cells expressing similar types of receptors, other data have suggested that this recognition event may play a role in the attachment of such cells to each other. Thus, Tonks and coworkers (20) have recently demonstrated that the receptor PTP specifically associates with the catenin/cadherin complex of homotypic cell adhesion molecules. They also demonstrated that treatment of cells with the PTP inhibitor pervanadate resulted in the up-regulation of tyrosine phosphorylation of cadherins and catenins, a result which suggested a role for a PTP, potentially PTP , in the maintenance of the cadherin/catenin complex in an underphosphorylated state. However, another recent report casts somewhat of a shadow on these findings (21). Interestingly, two other recent reports suggest that PTPs and LAR also appear to interact with the catenin/cadherin complex in a specific manner (22,23). Importantly, previous work suggested that the level of tyrosine phos-phorylation of this complex was correlated with the adhesive capacity of the cadherins (24), a result which is consistent with the hypothesis that the adhesion between cells mediated by the cadherins might be regulated by their tyrosine phosphorylation levels as determined by homotypic interactions between receptor PTPs such as and .
The finding that PTPs and mediated homotypic adhesion, together with the somewhat restricted tissue distribution of these PTPs (12,13), has suggested that additional members of this family of adhesive enzymes might exist. Here we report the cloning and characterization of the third member of this receptor PTP family, termed murine PTP . The receptor reported here contains structural motifs that are very similar to those found in PTP and and is capable of mediating homotypic adherence and associating with ␤ catenin. In addition, this receptor PTP reveals a tissue distribution, particularly with respect to its expression on neurons in the central nervous system, which is divergent from that previously described for the other members of this family. These data are consistent with a burgeoning family of MAM-containing receptor PTPs potentially involved with cell adhesion in various tissues.

RNA Isolation and Polymerase Chain Reaction-Messenger
RNA was isolated from the non-adherent Lin lo CD34 hi fraction of fetal yolk sac hematopoietic cells (25) (Micro-FastTrack, InVitrogen). Poly(A) ϩ RNA was reverse transcribed with random hexamers (Promega) and Molony murine leukemia virus reverse transcriptase (SuperScript II, Life Technologies, Inc.). One quarter of this cDNA was amplified by PCR using degenerate mixed oligonucleotides primers. Sense and antisense primers corresponding to the amino acid sequences (H/D)- Isolation of cDNA Clones-Adapter-linked double-stranded cDNA was prepared from A ϩ RNA of day 10 mouse embryos (Marathon ready cDNA synthesis kit, Clontech) using either random hexamer or oligo(dT) primer. Full-length cDNA was isolated by 5Ј or 3Ј rapid amplification of cDNA ends (RACE) of the marathon ready cDNAs. A cDNA library derived from adult mouse lung was screened following standard protocols using cDNA fragments isolated by RACE as probes. The DNA sequences of the RACE products as well as the clones were determined using standard sequencing protocols.
Bacterial Expression of GST-PTP Fusion Protein-cDNA sequences encoding amino acids 43-741 containing the extracellular region of the PTP were obtained by PCR using primers containing SalI or NotI. The PCR fragment was treated with SalI and NotI restriction enzymes and cloned into the pGEX-4T-1 plasmid (Pharmacia Biotech Inc.). Fusion protein was affinity purified using glutathione-Sepharose columns (Pharmacia). Polyclonal anti-serum against the extracellular region was generated by repeatedly immunizing rabbits with the purified GST-fusion protein. Affinity-purified extracellular domain antibody was obtained by binding serum IgG to the receptor GST-fusion protein immobilized on glutathione-Sepharose, and eluting the bound antibody with low pH and high salt.
Expression and Purification of the MAM-IgG-FnIII Immunoglobulin Fusion Protein-To express the extracellular portion of PTP , including MAM, IgG-like, and FnIII domains fused with human IgG hinge, CH2, and CH3 domains, fragments were subcloned into the pRK5 vector (26). The fusion proteins were expressed under the control of a cytomegalovirus promoter with a herpes simplex virus glycoprotein D signal sequence and monoclonal antibody 5B6 binding epitope. Transfection was done in 293 cells using the calcium phosphate precipitation method with 40 150-mm plates. One day after transfection, the cells were transferred to serum-free conditional media for an additional 5-6 days. Enzyme-linked immunosorbent assay in 96-well plates to test the expression levels was performed on the 3rd or 4th day. To purify protein, the supernatant (about 1 liter) was spun to remove cell debris, filtered, and concentrated ϳ10-fold. The sample was then added to a protein G column three times, and the column was washed with at least 10 bed volumes of PBS. Washing was monitored by A 280 . Finally, the fusion protein was eluted with 0.1 M glycine, pH 2.8, and neutralized with 0.1 volume of 1 M Tris, pH 8.0. The eluted protein was dialyzed against PBS overnight.
In Vitro Homotypic Adhesion Assay-Fluorescently-labeled green and red spheres (ϳ2 m, Duke Scientific) were incubated in 300 g/ml solutions of either the PTP -IgG or FLT 4-IgG (27) fusion proteins in PBS for 30 min at room temperature. The beads were blocked with 10 mg/ml bovine serum albumin in PBS for an additional 30 min at room temperature and subsequently washed three times with PBS, after which they were suspended in their original volume of PBS and used immediately. Different combinations of protein-coated beads were incubated in microtitre wells at room temperature for 30 min on a rotary shaker. 10-l aliquots were removed and examined under the fluorescence microscope at a wavelength of 359 nM. At this wavelength, the green and red beads appear to be blue and orange, respectively, and both can therefore be visualized simultaneously.
Analysis of Proteolytic Cleavage-Various truncation mutants of PTP , which contained the herpes simplex virus glycoprotein D signal sequence and the monoclonal antibody 5B6 binding epitope, were produced in the PRK 5 vector using the polymerase chain reaction. Constructs were transfected into 293 cells and incubated for up to 4 days. The supernatant was collected and centrifuged to remove cell debris. Then it was incubated with 5B6 at 1 g/ml for 2 h at 4°C and protein G-Sepharose (Pharmacia) was added for another hour. The immune complexes were washed 5 times with 0.2% deoxycholic acid, 0.2% Tween 20 in PBS, boiled for 5 min with SDS sample buffer, and analyzed on 4 -20% SDS-PAGE. The gel was transferred to a ProBlott membrane (Applied Biosystems) with electroblotting buffer (10 mM CAPS, pH 11, and 10% MeOH). Immunoblots were probed with the 5B6 antibody and developed using chemiluminescence (ECL, Amersham Corp.). Alternatively, if the cell lysates were used, the transfected cells were washed once with PBS and lysed in E1A lysis buffer (0.25 M NaCl, 5 mM EDTA, 0.1% Nonidet P-40, 50 mM HEPES, pH 7.6, and 1 l/ml aprotinin) at 4°C. After centrifugation, the incubation of the lysate with antibodies and protein G-Sepharose, SDS-PAGE separation of the immune complex, and immunoblot detection were as described above. Neonatal brains were dounce-homogenized in the same lysis buffer and immunoprecipitated with affinity purified anti-extracellular domain antibody, transferred to membranes, and probed with the same antibody.
Coprecipitation with ␤ Catenin-Lysates of human embryonic kidney (293) cells, which had been previously transfected with either the fulllength PTP or a truncated version missing the dual phosphatase domains but containing the juxtamembrane region, were immunoprecipitated with an anti-␤ catenin monoclonal antibody directed against the C terminus of the protein (Transduction Laboratories) using the immunoprecipitation conditions described above. The anti-␤ catenin immunoprecipitates were run on SDS gels, immunoblotted, and probed with the monoclonal antibody directed against the glycoprotein D 5B6 epitope on the N terminus of PTP . The blots were exposed, stripped, and reprobed with the anti-␤ catenin monoclonal antibody used for immunoprecipitation. Neither the 5B6 nor the anti-␤ catenin monoclonal antibodies revealed cross reactivity with ␤ catenin or PTP , respectively, in Western blot experiments.
Northern Analysis-A 2.5-kilobase cDNA fragment encoding the cytoplasmic region of PTP was used to probe the murine multi-tissue Northern blot (Clontech) In Situ Hybridization-Rat E15.5 embryos and postnatal day 1 brains were immersion fixed overnight at 4°C in 4% paraformaldehyde, then cryoprotected overnight in 15% sucrose. Adult rat brains were fresh frozen with powdered dry ice. All tissues were sectioned at 16 m and processed for in situ hybridization for PTP using [ 33 P]UTP labeled RNA probes. Sense and antisense probes were synthesized from a 2.5-kilobase DNA fragment of PTP using SP6 or T7 polymerase, respectively.
Immunohistochemistry-The cerebral cortex from postnatal day 1 (P1) rat pups was dissected out and minced into small pieces in Hanks' balanced salt solution. The tissue was dissociated in papain (Worthington) for 30 min. Triturated cells were resuspended in Neurobasal medium with B27 supplement (Life Technologies, Inc.) and plated onto chambered microscope slides (Nunc) coated with poly-D-lysine. The cells were maintained in an incubator for 1 day to allow the neurons to extend neurites and then fixed in 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, for at least 15 min. Brains from embryonic day 14 (E14) rat pups were immersion fixed in 4% paraformaldehyde, cryopro-tected in 30% sucrose, frozen in OCT mounting medium, and sectioned at 20 m in the coronal plane through the spinal cord. For immunostaining, cortical cells or spinal cord sections were rinsed for 30 min in PBS, blocked with goat serum, and then incubated with affinity purified antibody against the external portion of PTP (15 ng/ml) overnight at 4°C followed by a Texas Red-conjugated goat anti-rabbit secondary antibody (1:200, Vector). Immunostaining reagents were diluted in 1 mg/ml bovine serum albumin in PBS. All stained tissues were imaged on a Zeiss Optiphot microscope with an Optronics 3 CCD color video or a Xillix digital 10-bit camera. Staining with pre-immune serum under identical conditions did not reveal any detectable signal (data not shown).

Isolation and Characterization of the cDNA Encoding PTP
-In order to isolate novel PTPs expressed in murine primitive hematopoietic cells, we undertook the cloning of PCR fragments produced by priming with sequences directed against conserved protein motifs found in PTPs from a number of different genes and species (28). Analysis of 70 different PCRderived subclones revealed an array of previously described PTPs as well as two novel PTPs. One of these novel PTPs, termed PTP HSCF, is a member of the PTP PEST family of enzymes, and it has been previously described (29). The second novel PCR fragment was homologous to PTPs and , both related receptor-type PTPs that mediate homophilic adhesion (4). In order to further characterize the cDNA encoding this novel PTP, a combined cloning approach that utilized RACE as well as cloning from phage cDNA libraries was performed. The composite cDNA and derived protein sequences determined from these various clones is shown in Fig. 1. The ATG start codon utilized for translation of this large open reading frame was embedded within a consensus Kozak sequence, and there are several translational stop codons upstream of this initiator codon. As can be seen from this figure, the protein derived from this cDNA is a large receptor-like molecule of 1,436 amino acids and a molecular mass of approximately 161,176 Da. Closer perusal of the sequence reveals 13 consensus N-linked glycosylation sites in the extracellular region of the receptor, consistent with a potentially significant degree of glycosylation. Fig. 2 illustrates that the novel, hematopoietically derived PTP-related protein reported here shows a high degree of homology to both PTP (ϳ60%) and PTP (ϳ53%) throughout their entire lengths (12)(13). Thus, the novel PTP contains MAM, IgG, four fibronectin type III, and two cytoplasmically localized phosphatase domains ( Fig. 2) (4,12,13). These homologies with the novel PTP are somewhat less than the homology between PTP and (ϳ62%), suggesting that the novel PTP reported here is more distantly related to these two PTPs rather than they are to each other. These data suggest that this novel PTP is the third member of the homotypically interacting PTP family containing PTPs and , and we have therefore named the novel receptor PTP .
As can be seen from Fig. 3, the relative sequence homologies in each of the domains of these three receptors suggests that they are indeed closely related. Interestingly, previous data suggested that both the MAM and IgG domains mediated specific homotypic adhesion between PTPs and (18,19), and it is clear from the sequence comparisons between these three related proteins that these two domains are homologous. However, the fact that there are a large number of sequence changes between these two motifs is also consistent with the supposition that they can mediate specific homotypic interactions. The overall sequence homologies between the three proteins is also relatively high in the FnIII domains although the homology in the first of these domains is significantly higher than in the others. Previous work has also demonstrated that a juxtamembrane site between the transmembrane domain and the first phosphatase domain is distantly homologous to a similar region in the cadherins (20), and this site shows a high degree of homology between these three receptors, a point that will be discussed further. A significant degree of sequence homology is also found between the first PTPase domains of these three receptors, with a somewhat lower level of homology between the second PTPase domains of these proteins. This latter result may be important since it has been reported that the first phosphatase domain is the critical enzymatic motif of the dual phosphatase regions in the receptor PTPs (30,31). The homology between these PTPase domains includes many of the residues previously found to be important for substrate recognition and tyrosine dephosphorylation in the PTP 1B (32) although not all of these residues are identically conserved, and immunoprecipitation studies have demonstrated that PTP has weak, but detectable, phosphatase activity against two different tyrosine-phosphorylated peptide substrates as well as against the aritificial substrate p-nitrophenyl phosphate (data not shown).
Homotypic Adhesion Mediated by PTP -Previous data have demonstrated that the PTP -related receptor PTPs and induce homotypic adhesion and that these adhesive events appear to be mediated by the MAM and IgG domains. In order to examine the ability of PTP to mediate such homotypic adhesion, a soluble construct containing the MAM, IgG, and first FnIII domain attached to the hinge, CH2, and CH3 regions of human IgG was produced (-IgG chimeric protein). This chimeric protein was purified to ϳ90% by passage over protein A-Sepharose (data not shown) and was used in in vitro binding experiments. As a control, the extracellular domain of the tyrosine kinase receptor FLT 4, in the form of an IgG chimera, was used (FLT 4-IgG) (27). As can be seen in Fig. 4, fluorescent beads containing the FLT 4-IgG were unable to aggregate, while beads containing the -IgG protein formed large mixed aggregates. As a further control, the aggregation of a mixture of red -IgG and green FLT 4-IgG beads was examined. Fig. 4 shows that only the red beads aggregated in this experiment, with the green beads remaining largely unaggregated. This in vitro binding assay thus demonstrates that a form of PTP can mediate homotypic aggregation and is consistent with previous data suggesting that the MAM and IgG domains are critical for this adhesive event (18,19).
Proteolytic Cleavage of PTP in Vitro and in Vivo-A number of previous studies have suggested that various receptor PTPs, including and , appear to have a regulated, specific proteolytic cleavage that gives rise to a secreted region derived from the extracellular domain (12,33). We examined this possibility for PTP in two different ways. In vivo proteolytic cleavage was examined by probing extracts of neonatal mouse brain with an affinity purified polyclonal antibody directed against a GST fusion of the extracellular domain of PTP . This source of tissue was chosen because of the high transcript levels in Northern blots of adult brain as well as the in situ hybridization results (both described below). As can be seen in Fig. 5, immunoprecipitation of brain extracts with this antibody and subsequent Western blotting revealed three specific bands in neonatal brain. The highest molecular mass band (ϳ190 kDa) appears to correspond to a glycosylated full-length form of this PTP. A second band at ϳ115 kDa and a smaller band at ϳ70 kDa are observed and are likely to be proteolytic fragments of the full-length receptor since they specifically react with the affinity purified antibody directed against the extracellular domain. These data thus demonstrate that PTP appears to be proteolytically processed in brain extracts to two lower molecular weight forms, and additionally demonstrates the specificity of the affinity purified antibody directed against this phosphatase.
In order to examine this processing event in greater detail, we produced a number of different truncated constructs to examine the cleavage of the protein in vitro. All of the constructs contained a herpes simplex virus glycoprotein D signal sequence and N-terminal monoclonal antibody (5B6) epitope to aid in the analysis. As can be seen in Fig. 6, expression of a full-length form of the receptor in 293 cells results in a ϳ190-kDa protein as well as a ϳ115-kDa proteolytically clipped fragment, very similar to those seen in neonatal brain extracts. Examination of cell supernatants reveals that the smaller ϳ115-kDa fragment appears to be shed from the cells. This cleavage occurs in the absence of the PTP domains since the truncation mutant lacking the dual phosphatase domains also shows a ϳ115-kDa fragment in both the cell lysates as well as in the supernatants. Interestingly, this cleavage appears to occur when the MAM domain is truncated from the receptor (appearing as a smaller band in the cell lysates transfected with construct number 3 in Fig. 6) although this form of the protein does not appear to be secreted, possibly due to incorrect folding of this form of the protein and intracellular degradation. Finally, removal of both the MAM and IgG domains results in ϳ64and ϳ50-kDa fragments in the cell lysates, with the ϳ50-kDa fragment being very efficiently secreted into the media. The molecular mass of the secreted ϳ50-kDa fragment is consistent with the proteolytic cleavage event occurring within the fourth FnIII repeat (12; also see "Discussion" below). Finally, the construct containing only the FnIII repeats also shows a smaller ϳ25-kDa fragment in the medium, suggesting that the ϳ50-kDa fragment may be further cleaved after secretion to give rise to this smaller fragment although such a smaller fragment was not readily apparent in the cells transfected with either construct 1 or 2 (Fig. 6). These data strongly suggest that the extracellular domain of PTP is efficiently and specifically processed in 293 cells as it is in neonatal brain, and further demonstrates that the processing site appears to be contained within the fourth FnIII domain, a result similar to that found previously for PTP (12).
PTP Interacts with ␤ Catenin in Human Kidney Epithelial Cells-Previous data have suggested that PTPs and interact specifically with the cadherin/catenin complex (20,22). In addition, recent data have also suggested that PTP LAR interacts with this homotypic adhesion complex in neuronal PC12 cells (23). However, other investigators have cast some doubts about the interactions between the catenin/cadherin complex and PTP , predominantly due to antibody cross-reactivity (21). In order to examine if PTP interacts with this adhesion complex, a coprecipitation experiment was performed. Because transcript analysis suggests that PTP is expressed in vivo in embryonic kidney epithelial cells (see below), a human 293 embryonic kidney cell line was transfected with either the full-length construct or a construct lacking the phosphatase domains but containing the juxtamembrane domain homologous to cadherin cytoplasmic domains (constructs 1 and 2 in Fig. 6). Lysates from these transfected cells were then immunoprecipitated with a monoclonal antibody directed against ␤ catenin, and the immunocomplex was electroblotted and probed with antibodies against the glycoprotein D tag contained within the N terminus of PTP . As can be seen in Fig.  7, immunoprecipitation with the ␤ catenin monoclonal antibody resulted in the specific coprecipitation of the full-length PTP but not of the cleaved extracellular domain. In addition, this figure illustrates that only the intracellular juxtamembrane domain is required for an association with ␤ catenin, consistent with previous data suggesting that this domain of PTP is required for catenin association (22). The relatively low percentage (ϳ10%) of PTP in the immunoprecipitated ␤ catenin complex can be explained in a number of ways. For example, the system analyzed the coprecipitation of a murine form of the PTP complexed with human ␤ catenin, and the wash conditions utilized may have been overly stringent. While a recent report suggests that some degree of antibody crossreactivity may have been responsible for the apparent interactions between PTP and ␤ catenin or E-cadherin (21), the use of highly specific monoclonal antibodies for both the ␤ catenin immunoprecipitation and PTP blotting steps of this experiment eliminates any such potential artifacts. Thus, we have not observed any indication that the anti-␤ catenin antibody recognizes PTP on Western blots, nor have we found that the anti-gD antibody used to detect PTP reacts with immunoprecipitated ␤ catenin (data not shown). These data thus support previous studies and suggest that PTP , like PTPs , , and LAR, interacts with the cadherin/catenin adhesion complex.
Tissue Expression of the PTP Transcript-As can be seen in Fig. 8, Northern blot analysis of fetal as well as adult tissues demonstrates that PTP is expressed in a diversity of organs outside of the hematopoietic progenitor cells from which it was originally cloned. Thus, the expression of this PTP is detected throughout embryonic development beginning in the very early embryo at day 7. Interestingly, analysis of adult organs reveals that this receptor is expressed specifically in only a subset of tissues. Thus, there appears to be a very high level of expression of the enzyme in brain, lung, and kidney, a much decreased level in heart, skeletal muscle, and testis, and a lack of obvious expression at this exposure in spleen and liver. The high level expression in lung and brain together with the lack of expression in liver are in contrast to PTP , a PTP which is expressed at high levels in liver but is almost undetectable in lung and brain (12). Thus, in spite of the fact that PTP was originally isolated from hematopoietic stem cells, there is no obvious expression in two sites which contain hematopoietic cells, the spleen, and the liver. The lack of signal in the spleen, an organ which contains mostly mature hematopoietic cells, suggests, therefore, that this receptor may be expressed specifically in earlier hematopoietic progenitor cells and not in the predominately mature cells found in this organ as was previously found for PTP HSCF (34). Interestingly, there appears to also be an alternatively spliced transcript in the lung, which is not detected in the other two organs that express this receptor at high levels or in the embryos although the nature of this alternatively spliced transcript has not been determined. In summary, these data suggest that PTP is specifically expressed in a subset of adult tissues, some of which are divergent from PTP .
In Situ Hybridization Analysis-We performed in situ mRNA analysis of the rat E15.5 embryo, and P1 and adult rat brain to determine potential sites of PTP production. Extensive PTP expression was observed in developing skeletal, epithelial, and neuronal structures throughout the E15.5 embryo (Fig. 9). Systemic expression was observed in various developing skeletal elements such as vertebral perichondrium, intervertebral discs, teeth, mandible, and maxilla (Fig. 9, A and  B). Expression within urogenital structures included the genital tubercle (panels A and B), urethra, and urogenital sinus (not shown). Other positive areas of PTP expression included the anal canal (not shown), skin, olfactory and oral epithelium, esophagus (panels A and B), pituitary (panels A-C), dura mater (panels A, B, and D), kidney (panels A and B), and lung (panels A and B). Higher magnification reveals expression restricted to developing glomeruli in the cortical region of the kidney (panels F and G), and bronchiolar epithelium of the lung (panels H and I). Within the E15.5 embryonic nervous system, high levels of expression were observed in the developing cerebral cortex (panels A and B), floor of the midbrain, choroid plexus primordium, gigantocellular reticular nucleus of the brain stem (panels A-C), dura mater, and spinal cord (panels A, B, and D). High magnification of the spinal cord reveals highest expression of PTP in the ventrolateral motor column (panel D).
In P1 and adult brain, expression of PTP was localized to regions derived from embryonic anlage that also contained high levels of expression. For instance, expression in the embryonic midbrain preceded the high levels of PTP expression in the P1 and adult substantia nigra (Fig. 10, C and E, respectively). Expression in the embryonic forebrain (Fig. 9A) preceded expression observed in the inner layers (cortical layers 5 and 6) of the P1 and adult cortex (Fig. 10, A, B, D, and E  respectively). Expression in the choroid plexus primordia of the embryo begets high levels of expression in the P1 brain (Fig.  10A) and low levels of expression in the adult brain (Fig. 10D).
In general, PTP expression in the adult brain appears to be down-regulated relative to the P1 brain (Fig. 10). However, other areas of prominent expression in both P1 and adult brain include piriform cortex and endopiriform nucleus (Fig. 10, A   FIG. 2-continued and D, respectively), amygdaloid nuclei, subiculum, and CA1, CA2, and to a lesser extent, CA3 of the hippocampal formation (Fig. 10, B and E, respectively). The P1 brain also exhibits strong expression throughout the septal area, basal ganglia, thalamus, and midbrain (panels A, B, and C). Weak expression is observed in the adult superior colliculus as well as scattered expression throughout the thalamus (panel E).
PTP Is Expressed on Neurons of the Central Nervous System and Cortex-The Western blot, Northern blot, and in situ analyses suggested that PTP is expressed in the nervous system although the resolution of these techniques was insufficient to determine if the receptor was expressed on neurons. In order to examine this possibility, immunostaining of the spinal cord and isolated cortical neurons of the E14 rat was done. As Fig. 11A shows, a subset of spinal neurons, including ventral and dorsal root neurons extending from the spinal cord as well as those in ventral and ventrolateral funiculi within the spinal cord, were specifically stained with the antibody. Examination of isolated cortical neurons for reactivity with the antibody revealed that both the soma as well as the axons and dendrites extending from the neuron stained with the antibody. In addition, growth cones were also positive for this receptor (Fig. 11, B and C). Together, these data suggest, in agreement with the in situ hybridization data, that spinal cord and cortical neurons express PTP , consistent with a function for this receptor in the formation of the nervous system.

DISCUSSION
The relative levels of tyrosine phosphorylation of a diversity of proteins are critical for the regulation of a number of activities during embryonic differentiation and throughout the life of the organism. The absolute levels of this modification are mediated through the balance of the enzymatic activities of tyrosine kinases with those of tyrosine phosphatases. In both cases, these large families of proteins perform their roles through conserved enzymatic domains that are coupled to a plethora of specificity-determining motifs. These various motifs are found in the context of both membrane traversing, receptorlike molecules as well as intracellular forms of the enzymes. These similarities in overall structure of the kinases and phosphatases suggest that they mediate their relative specific activities through the use of these various domains. A subset of receptor phosphatases also contain a diversity of domains, including immunoglobulin-and fibronectin-like, which are associated with cell adhesion and ligand binding activities in other protein families. Among the most interesting of these types of adhesion-associated PTPs are the and receptors which are involved with homotypic types of interactions. Earlier predictions, based upon the likely function of these receptors in mediating cell adhesion as well as their limited tissue distribution, suggested that there might be other and -like receptor PTPs with different tissue dispositions (4). We report here the isolation of the third member of this family of homophilically interacting receptor PTPs which may be associated with the construction of epithelial and neural structures during development and in the adult.
The strongest data suggesting that the novel PTP described here, termed PTP , is homologous to the and receptors lies in the high degree of sequence conservation among these three proteins. Analysis of these three receptors clearly revealed that the novel PTP had a significant degree of sequence homology throughout its entire length. This homology included the four major types of domains contained in this family including the MAM, the IgG, the FnIII, and the dual phosphatase (PTP) domains (12,13). Because previous data have suggested that both the MAM as well as the IgG domain appear to be involved with homotypic adhesion (18,19), it is likely that these motifs are used for a similar function in PTP , a hypothesis that is consistent with a role for this receptor in cell adhesion (see below). However, the degree of sequence homology of these domains between the newly reported receptor and the and receptors is quite divergent, suggesting that the novel receptor may also specifically mediate a homophilic interaction only to itself and not to these domains in the other family members (19). As will be discussed below, these results, together with the tissue localization of this receptor, suggest that it may be involved with the formation of very specific edifices during development. While it is difficult to currently interpret the significance of the conservation of the FnIII domains, which may act as spacer domains to extend the functionally critical MAM and IgG domains from the cell surface, the conservation of the dual PTP domains lends itself to some comment. Thus, the higher degree of conservation of the first domain as compared with the second substantiates previous work suggesting that the N-terminal PTP motif is the enzymatically active one, while the C-terminal domain may be involved with the regulation of enzyme activity (30). Indeed, recent structural data of the phosphatase domains of PTP ␣ reveal that the second domain may regulate the enzymatic activity of the first domain (31). In summary, the data reported here are consistent with PTP being the third member of the homotypically interacting receptor PTP family.
As with other members of this family of PTPs, PTP can mediate homotypic adhesion. The in vitro binding studies using the IgG chimeric protein containing the MAM, IgG, and first FnIII domains of PTP are consistent with previous data suggesting that the first two of these domains appear to be critical for homotypic adhesion (14,19). While this paper was under review, Ullrich and colleagues also reported a novel receptor PTP, termed PCP-2, which appears to be the human homologue of PTP (35). As predicted by the homotypic adhesion studies reported here, PCP-2 is concentrated at the intracellular contact points of cells expressing this receptor. Together, these data support the hypothesis that PTP is the third member of this family of proteins to mediate homotypic adhesion and further suggest that this adhesion is induced by, at most, the first three N-terminal domains of the protein. The proteolytic cleavage studies reported here also are consistent with previous data on this family of homotypically adherent PTPs (12,33), and have implications as far as the adhesion modulating functions of this receptor. The proteolytic mapping studies suggested that a specific cleavage site within the fourth fibronectin domain was the point where proteolysis occurred, and examination of this domain in PTP revealed a furin-like cleavage sequence ( 636 RLRR 639 ) which is highly conserved in all three members of this receptor PTP family (12,22). Interestingly, this site appears to be missing in the PCP-2 phosphatase in spite of its otherwise high degree of homology with PTP , and PCP-2 is apparently not cleaved in transient transfection assays (35), which is in sharp contrast to the results reported here. In addition, analysis of PTP expression in neonatal brain extracts revealed that this proteolytic cleavage event also appears to occur at a significant level in vivo at a potentially similar site to that seen in vitro although a further processed form of the extracellular domain was also observed in brain extracts. Because the cleaved form of the extracellular region of PTP contains all of the motifs necessary to mediate homotypic adhesion, it is likely that this soluble protein can bind to, and potentially inhibit, the homotypic adhesion mediated by the cell surface form of PTP . Thus, it is possible that this specifically cleaved form of the protein acts to regulate the cell-associated form of the enzyme.
Previous data have suggested a role for this category of receptor PTPs in cadherin/catenin regulation, and other investigators have pointed to an intracellular juxtamembrane site with significant homology to a similarly localized region in the cadherins (4,20). Analysis of ␤ catenin immunocomplexes clearly revealed the association of PTP with this adhesion system. We have also found a very high degree of sequence conservation in the juxtamembrane region with the other members of this phosphatase family, consistent with a potential role for this domain in catenin/cadherin interactions. In agreement with this supposition, we have demonstrated that a PTP construct containing only this juxtamembrane region intracellularly can associate with ␤ catenin, consistent with previous data demonstrating that a homologous region of PTP alone can mediate catenin association in vitro (22). These data thus support the proposal from Tonks and coworkers (4,20), suggesting a role for this family of PTPs in the regulation of cadherin-mediated adhesion. An interesting possibility along these lines, therefore, is that the lack of contact inhibition observed in cells treated with the phosphatase inhibitor vanadate might have been due to the inhibition of the dephosphorylation of the catenin/cadherin complex by one or more members of this receptor PTP family (36). Together with the in situ data demonstrating expression of PTP on epithelial sheets in lung and kidney particularly, this possibility suggests that the cadherin-mediated adhesion of these cells may in part be regulated by the activity of PTP during development and in the adult. In addition, the possibility that such phosphatase-catenin interactions in neuronal cells exist has recently been suggested by work from Reichardt and coworkers (23). The demonstration here that PTP is expressed in neurons of the central nervous system, together with the suggestion that it mediates homotypic adhesion and associates with the catenincadherin complex, is consistent with the possibility that this novel PTP may be involved with specific adhesion during neural development as well (see below).
The in situ hybridization analysis of the expression of PTP in the developing embryo and adult suggest some potentially important hypotheses. The expression of this receptor in a diversity of developing skeletal areas, as well as in epithelial sites which line various organ systems with a layer of these cells, coupled with the proposed role for PTP (20) (4) and potentially PTP (22) in the control of cadherin-mediated adhesion, suggests that the novel PTP might be involved in a similar type of adhesion control in the developing embryo. For example, the development of epithelial layers in the lung bronchioles and kidney glomeruli requires that a sheet of epithelial cells that is 1-cell thick be constructed. Thus, as the cells grow and migrate during embryogenesis, they would require a mechanism where they sensed the location of other epithelial cells that were in contact with them so that this cellular contiguity initiated an adhesive response that inhibited further epithelial movement via the enhancement of cell adhesion. The formation of 1-cell thick epithelial structures in these embryonic organs could be mediated by such a sensing mechanism using the homotypic adherence of PTP in association with the catenincadherin complex. The expression of this receptor PTP in bone forming chondrocytes would also be expected to perform a similar type of sensing and adhesion function to assemble these structures although this type of anatomy, which is more complex than the thin walled epithelial-like morphology described above, would be expected to involve more elaborate types of sensing and adhesive mechanisms. Finally, because many common types of tumors of the lung and other organs involve   7. Coprecipitation of PTP with ␤ catenin. Human embryonic kidney cells (293 cells) were transfected with either full-length or PTP domain-deleted constructs expressing PTP as described in Fig. 6  (constructs 1 and 2, respectively). Transfected cells were then immunoprecipitated with a monoclonal antibody directed against the C terminus of ␤ catenin, and the immunoprecipitates were run on SDS gels and subsequently transferred to membranes. In addition, lysates of the transfected cells were also run on these gels and transferred. Blots were then probed with either the anti-gD monoclonal antibody 5B6, specific for the N terminus of PTP (upper panel) or the anti-␤ catenin monoclonal antibody (lower panel). The two closed arrows illustrate the full-length and truncated forms of PTP , while the open arrow illustrates the proteolytically cleaved form of the receptor (see Fig. 6). Note that the cleaved ϳ115-kDa form of PTP does not appear to be coprecipitated with the anti-␤ catenin monoclonal. The blot on the right illustrates PTP expression in RNA isolated from: heart (a), brain (b), spleen (c), lung (d), liver (e), skeletal muscle ( f ), kidney (g), and testis (h).  cells, it is possible that disruptions in the proposed function of this type of adhesion sensing mechanism might be involved with the disorganized morphology and often high rate of metastasis of these tumors (17,24). Together, these hypotheses suggest a critical role for PTP in the formation of various epithelial-like structures in the embryo.
Recent data from the Drosophila system also suggest interesting possibilities for the function of PTP in the developing nervous system (10,11). In these reports, three different Drosophila receptor PTPs, termed DPTP69D, DPTP99A, and DLAR, which all contain IgG and fibronectin type III adhesion domains similar to those found in PTP , were shown to be critically involved with neuronal pathfinding in the developing nervous systems. Thus, mutations in either of these receptors resulted in a loss of the ability of certain neural subsets to become reoriented during their formation in the embryo. These data, together with the recent observations on the association of mammalian LAR with the catenin-cadherin complex, are consistent with a role for receptor PTPs in specific neural pathfinding and adhesion. Because PTP , a homotypically adhering PTP associated with the catenin complex, is expressed in a number of developing neural sites, it is possible that it plays a similar role in the pathfinding of nerves in mammals. Thus, the expression of this PTP in the developing midbrain, forebrain, and other neural sites would dispose it to function as a mediator of pathfinding in these maturing sys-FIG. 10. PTP mRNA expression in P1 and adult rat brain. Emulsion autoradiographs of coronal sections of P1 rat brain (A-C) and adult rat brain (D and E) hybridized with a [ 33 P]UTP labeled PTP antisense probe are shown. Coronal sections of the P1 brain are at the level of the septum (A), hippocampus (B), and substantia nigra (C). For the adult animal, coronal brain sections are at the level of the septum (D) and the hippocampus and substantia nigra (E). Hybridization using a PTP sense strand control probe is shown in an adult coronal section at the level of the substantia nigra (F). Bar, 1.0 mm for A-C; bar, 1.0 mm for D-F. tems. Interestingly, the expression of this receptor in these embryonic anlage was confirmed by expression in the adult sites which arise from these embryonic structures. However, the expression in the adult appeared to be somewhat reduced as compared with that observed in the embryo, and it was far more organized. These data suggest that this enzyme might be utilized during adult neuronal formation although the apparent decrease in adult expression suggests a potentially more critical role during embryogenesis. The observation that PTP is expressed on neurites of isolated cortical neurons as well as on the growth-cone like structures at the tips of these processes is also consistent with a potential role for this receptor in neuronal pathfinding in the mammalian nervous system. Interestingly, we have observed that PTP is specifically expressed on layer 5 and 6 cortical neurons only during their differentiation and pathfinding phases, again consistent with a role for this phosphatase in aspects of neural architecture. 2 Finally, while the clear observation of the loss of pathfinding in Drosophila will be difficult to recapitulate in the mouse due to the relatively high complexity of the mammalian nervous system, it will nevertheless be potentially of great interest to examine the formation of the nervous system in animals which have been made null for the expression of this receptor.
In summary, the data reported in this paper demonstrate the existence of a third member of the family of receptor PTPs that appear to be involved with homotypic adhesion and, potentially, cadherin/catenin-mediated organ formation. The role that this novel receptor might play in the formation of epithelial sheets and neuronal structures remains to be determined.
However, the existence of three of these types of receptors further suggests that this growing family may be involved with the specific formation of various types of complex structures during development, as well as in the adult.