WO1994009037A9 - A NEW CLASS OF RPTPases: THEIR STRUCTURAL DOMAINS AND LIGANDS - Google Patents

Abstract

A new class of receptor protein tyrosine phophatase molecule, the family of ligands that binds this new class of receptor, and uses of such receptors and ligands is described.

Description

A NEW CLASS OF RPTPases: THEIR STRUCTURAL DOMAINS AND LIGANDS

1. INTRODUCTION 5 The present invention relates to a new class of receptor protein tyrosine phosphatase molecule, the family of ligands that binds this new class of receptor, and uses of such receptors and ligands. Specifically, the members of this new class of

10 receptor protein tyrosine phosphatase molecule are proteoglycans and/or possess an extracellular carbonic anhydrase structural domain. The characterization of one member of this new class, RPTP3, is described in the working examples presented herein. The ligands

15 which bind the receptor protein tyrosine phosphatases of the invention are members of the cell adhesion molecule (CAM) family of extracellular molecules.

BACKGROUND OF THE INVENTION

20

2.1 PROTEIN PHOSPHORYLATION AND SIGNAL TRANSDUCTION

Cells rely, to a great extent, on extracellular molecules as a means by which to receive stimuli from

25 their immediate environment. These extracellular signals are essential for the correct regulation of such diverse cellular processes as differentiation, contractility, secretion, cell division, contact inhibition, and metabolism. The extracellular

30 molecules, which can include, for example, hormones,

-^•" growth factors, or neurotransmitters, act as ligands

* that bind specific cell surface receptors. The

* binding of these ligands to their receptors triggers a cascade of reactions that brings about both the ampli-

₃₅ fication of the original stimulus and the coordinate regulation of the separate cellular processes mentioned above.

A central feature of this process, referred to as signal transduction (for recent reviews, see Posada, J. and Cooper, J.A. , 1992, Mol. Biol. Cell 2:583-592; Hardie, D.G. , 1990, Symp. Soc. Exp. Biol. 4_4:241-255) , is the reversible phosphorylation of certain proteins. The phosphorylation or dephosphorylation of amino acid residues triggers conformational changes in regulated proteins that alter their biological properties.

Proteins are phosphorylated by protein kinases and are dephosphorylated by protein phosphatases. Protein kinases and phosphatases are classified according to the amino acid residues they act on, with one class being serine-threonine kinases and phosphatases

(reviewed in Scott, J.D. and Soderling, T.R. , 1992, 2_.:289-295), which act on serine and threonine residues, and the other class being the tyrosine kinases and phosphatases (reviewed in Fischer, E.H. et al. , 1991 Science 253:401-406; Schlessinger, J. and Ullrich, A., 1992, Neuron 9_.:383-391; Ullrich, A. and Schlessinger, J. , 1990, Cell _.6_1:203-212) , which act on tyrosine residues. The protein kinases and phospha¬ tases may be further defined as being receptors, i.e., the enzymes are an integral part of a transmembrane, ligand-binding molecule, or as non-receptors, meaning they respond to an extracellular molecule indirectly by being acted upon by a ligand-bound receptor. Phosphorylation is a dynamic process involving competing phosphorylation and dephosphorylation reactions, and the level of phosphorylation at any given instant reflects the relative activities, at that instant, of the protein kinases and phosphatases that catalyze these reactions. While the majority of protein phosphorylation occurs at serine and threonine amino acid residues, phosphorylation at tyrosine residues also occurs, and has begun to attract a great deal of interest since the discovery that many oncogene products and growth factor receptors possess intrinsic protein tyrosine kinase activity. The importance of protein tyrosine phosphorylation in growth factor signal transduction, cell cycle progression and neoplastic transformation is now well established (Cantley, L.C. et al. , 1991, Cell 64:281-302; Hunter T. , 1991, Cell 61:249-270; Nurse, 1990, Nature 344:503-508; Schlessinger, J. and Ullrich, A., 1992, Neuron 9 :383-391; Ullrich, A. and Schlessinger, J., 1990, Cell 61:203-212). Subversion of normal growth control pathways leading to oncogenesis has been shown to be caused by activation or overexpression of tyrosine kinases which constitute a large group of dominant oncogenic proteins (reviewed in Hunter, T. , 1991, Cell 64:249-270) . In addition, since the initial purification, seguencing and cloning of a protein tyrosine phosphatase (Thomas, M.L. et al. , 1985, Cell 41:83) , additional potential protein tyrosine phosphatases have been identified at a rapid pace. (See, for example, Kaplan, R. et al. , 1990, Proc. Natl. Acad. Sci. USA 82:7000-7004; Krueger, N.X. et al.. 1990, EMBO J. _.9:3241-3252; Sap, J. et al.. 1990, Proc. Natl. Acad. Sci. USA 87_.:6112-6116) . Because the number of different protein tyrosine phosphatases that have been identified is increasing steadily, speculation has arisen that the protein tyrosine phosphatase family may be as large as the protein tyrosine kinase family (Hunter, T. , 1989, Cell 58_.: 1013-1016) . With this increase in the reported cloning of protein tyrosine phosphatase genes, the role that the regulation of dephosphorylation may have in the control of cellular processes has also begun to receive more attention.

2.2 PROTEIN TYROSINE PHOSPHATASES As mentioned above, protein tyrosine phosphatases (PTPases) can be classified into two subgroups, the non-receptor and receptor classes. The non-receptor class is composed of low molecular weight, cytosolic, soluble proteins. All known non-receptor PTPases contain a single conserved catalytic phosphatase domain of approximately 230 amino acid residues. (See, for example, Charbonneau et al. , 1989, Proc. Natl. Acad. Sci. USA 86:5252-5256; Cool et al.. 1989, Proc. Natl. Acad. Sci. USA JLj5:5257-5261; Guan et al.. 1990, Proc. Natl. Acad. Sci. USA 87:1501-1502;

Lombroso et al. , 1991, Proc. Natl. Acad. Sci. USA 88_.:7242-7246. ) Sequence analysis reveals that about 40 of the amino acids of the catalytic domain are highly conserved, and a very highly conserved segment of 11 amino acid residues with the consensus sequence [I/V]HCXAGXXR[S/T]G, is now recognized to be a hallmark of the protein tyrosine phosphatase catalytic domain.

The receptor class is made up of high molecular weight, receptor-linked PTPases, termed RPTPases. Structurally resembling growth factor receptors, RPTPases consist of an extracellular, putative ligand- binding domain, a single transmembrane segment, and an intracellular catalytic domain (reviewed in Fischer et al. , 1991, Science 253:401-406) . The intracellular segments of almost all RPTPases are very similar. These intracellular segments consist of two catalytic phosphatase domains of the type described above, separated by an approximately 58 amino acid residue segment. This two domain motif is usually located approximately 78 to 95 amino acid residues from the transmembrane segment and is followed by a relatively short carboxy-terminal amino acid sequence. The only known exception is the isoform HPTP/S (Krueger, N.X. et al. , 1990, EMBO J. 9_:3241), which contains only one catalytic phosphatase domain.

While the intracellular RPTPase segments are remarkably highly conserved, the RPTPase extracellular domains are highly divergent. For example, certain RPTPases possess a heavily glycosylated external domain and a conserved cysteine-rich region (Thomas, M.L. et al.. 1985, Cell 4_1:83; Thomas, M.L. et al. , 1987, Proc. Natl. Acad. Sci. USA 8_.4:5360; Ralph, S.J. et al.. 1987, EMBO J. 6 :1251-1257) while others contain immunoglobulin G-like (Ig) domains linked to fibronectin type III repeats (Streuli, M. et al. _f 1989, Proc. Natl. Acad. Sci. USA 8j5:8698; Streuli, M. et al.. 1988, J. Exp. Med. 168:1523) . Still other RPTPases contains only multiple fibronectin type III repeats (Krueger, N.X. et al.. 1990, EMBO J. 9_.:3241), while certain RPTPases have smaller external domains that contain several potential glycosylation sites (Jirik, F.R. et al.. 1990, FEBS Lett. 221:239). The ligands that regulate RPTPs have not been identified. It has been speculated that circulating extracellular factors are unlikely to bind to those receptors containing Ig and/or fibronectin Type III repeats and that interaction with other surface antigens, perhaps on other cells, is more likely to be the case with these receptors.

Because enhanced tyrosine phosphorylation has been shown to be responsible for causing cellular transformation, underexpression, or inactivation, of protein tyrosine phosphatases may also potentially result in oncogenesis. For this reason, tyrosine- specific phosphatase genes are candidate recessive oncogenes or tumor suppressor genes. In support of this theory, the human RPTPase, RPTP7, has been shown to map to a chromosomal region, 3pl4-21, which is frequently deleted in renal cell and lung carcinomas (LaForgia, S. et al. , 1991, Proc. Natl. Acad. Sci. USA 88_.:5036-5040) . Recent studies, however, indicate that protein tyrosine phosphatase action need not only be suppressive. It has been shown that members of the src family of non-receptor tyrosine kinases contain inhibitory tyrosine phosphorylation sites in the carboxy terminal tails (reviewed by Hunter, T. , 1987, Cell 9_:1-14). When these sites are phosphorylated, the molecules' tyrosine kinase activity is inhibited (Nada, S. et al.. 1991, Nature 3_.51:69-72). It has further been demonstrated that, in T cells, the dephosphorylation of such inhibitory sites by a protein tyrosine phosphatase (CD45) leads to enhanced tyrosine phosphorylation (Ledbetter, J.A. et al., 1989, Proc. Natl. Acad. Sci. USA 86:8628-8632) , indicating, therefore, that phosphatases may function as activating and well aε inhibitory signaling enzymes. Also, dephosphorylation of a tyrosine residue has been suggested to be an obligatory step in the mitotic activation of the maturation-promoting factor kinase (Morla, A.O. et al. , 1989, Cell 58:193- 203) . Taken together, the above observations suggest that PTPases may play an important role in cellular control mechanisms, as effectors in mechanisms of transmembrane signaling, as cell-cycle regulators, and as potential oncogenes and anti-oncogenes. 3. SUMMARY OF THE INVENTION The present invention relates to a new class of receptor protein tyrosine phosphatase molecule, to the family of ligands that binds this new class of receptor, and to the uses of such receptors and ligands. Specifically, the members of this new class of receptor protein tyrosine phosphatase molecule are proteoglycans and/or possess an extracellular carbonic anhydrase structural domain. The characterization of one such receptor molecule, RPTP , is described in the working examples presented herein.

The ligands which bind the receptor protein tyrosine phosphatases of the invention are members of the cell adhesion molecule (CAM) family of extracel- lular molecules. The discovery that CAMS bind receptor protein tyrosine phosphatases represents the first identification of a natural ligand for this type of receptor. Binding of two CAMs, namely N-CAM and Ng-CAM, to the receptor protein tyrosine phosphatases of the invention is demonstrated in the working examples presented herein. The receptors and the receptor-binding ligands of the invention may be used to develop compounds and strategies for modulating cellular processes under the control of the receptor protein tyrosine phosphatases. Such processes include, but are not limited to, normal cellular functions such as differentiation, metabolism, cell cycle control, and neuronal function; cellular behavior such as motility and contact inhibition, in addition to abnormal or potentially deleterious processes such as virus-receptor interactions, inflammation, cellular transformation to a cancerous state, and the development of Type 2, insulin Independent, diabetes mellitus. Compounds that may interfere with ligand binding are described and methods for identifying other potential ligands, such as CAM-type ligands, growth factors, or extracellular matrix components, are discussed.

4. BRIEF DESCRIPTION OF THE FIGURES

FIG. 1. The amino acid sequence of RPTP/3. The protein sequence of RPTP/3 containing 2308 amino acids is indicated. The hydrophobic signal peptide is underlined, the transmembrane peptide is underlined and the transmembrane peptide is designated in bold- type. The 21 potential N-glycosylation sites are indicated by the arrows. The CAH-related domain and the two phosphatase domains, DI and DII, are indicated by the boxes.

FIG. 2. Chromosomal localization of human RPTPβ. A. Presence of the RPTP/3 gene in a panel of 17 rodent-human hybrids. A completely stippled box indicates that the hybrid named in the left column contains the chromosome indicated in the upper row; lower-right stippling indicates presence of the long arm (or part of the long arm, indicated by a smaller fraction of stippling) of the chromosome shown above the column; upper left stippling indicates presence of the short arm (or partial short arm) of the chromosome listed above the column; an open box indicates absence of the chromosome above the column; the column for chromosome 7 is boldly outlined and stippled to highlight correlation of presence of this chromosome with the presence of the RPTP/3 gene. The pattern of retention of the RPTP/3 sequences in the hybrids is shown to the right where presence of the gene in the hybrids is indicated by a stippled box with a plus sign and absence of the gene is indicated by an open box enclosing a minus sign. B. RPTP/3 maps to 7q31- q33. Chromosome in situ hybridization of a 1.8 kb RPTP/3 cDNA to normal human metaphases confirmed local¬ ization of the gene to 7q and revealed a peak of grains centered over region 7q31.3 - 7q32 as illustrated on the right to the chromosome sketch. Each dot representing an autoradiographic grain.

FIG. 3. Analysis of the expression of RPTP/S in various murine tissues and cell lines. A. Poly A+ RNA (1 μg per sample) from the various murine tissues indicated were loaded onto a 1.0% agarose/2.2M formaldehyde gel and probed with the per amplified murine DNA fragment, pBSMBDII (described in Materials and Methods, Section 6.1.4). B. The blot in A. was stripped of probe and rehybridized with a ³²P labeled rat actin probe. C. 20 μg of total cellular RNA (lanes 1-5) and 1 μg of Poly A+ RNA (lane 6) isolated from the various glioblastoma and neuroblas- toma cell lines indicated were loaded onto on RNA gel and probed with a DNA fragment isolated from the human brain stem cDNA clone that begins with sequences just 5' of the transmembrane region and extends and includes all of the sequences in phosphatase domain I.

FIG. 4. Northern blots to identify alternative splicing of RPTP/3 transcripts. A. A schematic diagram of the protein encoded by the full length RPTP0 cDNA compared to the putative protein encoded by the two independently isolated cDNA clones that carry an identical deletion of 258 bp in the extracellular region of the protein. The position of the deletion is indicated by the dotted line with the number of the amino acid that remains at both the 5 ' and 3 ' end of the deletion indicted. The location of the two probes used in Northern analysis (probes 1 and 2) are indicated. TM, transmembrane peptide; DI, phosphatase domain I and DII, phosphatase domain II. B. Poly A+ RNA (1 μg) isolated from the Lan 5 neuroblastoma cell line was separated on a RNA formaldehyde gel and probed with human probe 1 (PI) that contains 1.3 kb of sequences derived from the extreme 5' end of the cDNA clone and human probe 2 (P2) that contains 1.6 kb of sequences derived from the portion of the full length cDNA clone that is deleted in clones BS-dl4 and Cau- dll.

FIG. 5. In situ hybridization analysis of RPTP/3 in developing and adult mouse brain. A. A sagittal section through an embryonic day 20 (E20) mouse shows that RPTPβ is preferentially expressed in the developing central nervous system. The highest level of expression is seen in the ventricular zone (VZ) . B. A sagittal section through the adult mouse brain shows discrete bands of expression in the Purkinje cell of the cerebellum, the dentate gyrus (OG) , and the anterior horn of the lateral ventricle (AH) .

FIG. 6. Identification of endogenous RPTP/3 protein expression in Lan 5 cells. Immunoprecipi- tation of RPTPS with normal rabbit serum (NRS, lane 1) and immune RPTPβ antiserum (αPTPβ, lanes 2 and 3) from lysates of ³⁵S methionine-labeled Lan 5 cells that had been labeled in the absence (lanes 1 and 2) or presence of tunicamycin (lane 3) . Apparent molecular weight is approximately 300 kD in the absence, and 250 kD is the presence, of tunicamycin. Immunoprecipita- tion of the EGF receptor with RK2 antibody (λaEGFR, lanes 4 and 5) from lysates of ⁵S methionine-labeled Lan 5 cells labeled in the absence (lane 4) and presence (lane 5) of tunicamycin. FIG. 7. Identification of a CAH-related domain in the extracellular region of RPTP/3. A. The align¬ ment of the amino acid sequence of the CAH-related domain of RPTP/3 with the corresponding domain of RPTP7 and the six different isόforms of CAH (Deutsch, H.F., 1987, Int. J. Biochem. JL9_:101-113) . The amino acid sequences that are boxed in black are those that are identical in all six isoforms of CAH. The sequences that are boxed in the gray hatches are those that are identical between the CAH-related domains of RPTP/S and RPTPγ. B. The percent similarity, taking into account conservative substitutions of amino acids utilizing the program, between the CAH-related domains of RPTP/3 and RPTPγ and the six isoforms of CAH is indicted in this grid.

FIG. 8 Polyacrylamide gel of an immunoprecipita- tion, using ³⁵S-NaS0₄-labeled cell lysates from 293 cells transfected with RPTPjβ DNA (Lane 1) or from control, 293 cells transfected with vector alone (Lane

2) . Antiserum used was directed against RPTPjδ, as described in Section 6.1.5.

FIG. 9 Polyacrylamide gel of an immunoprecipita- tion, using ³⁵S-Met-labeled cell lysates from 293 cellε transfected with RPTPβ DNA (Lane 1) or from control, 293 cells transfected with vector alone (Lane 2) . Antiserum used was directed against RPTP , as described in Section 6.1.5.

FIG. 10 Polyacrylamide gel of an immunoprecipi- tation, using ³⁵S-Met-labeled cell lysates from 293 cellε transfected with RPTPβ DNA (Lanes 3 and 4) or from control, 293 cells transfected with vector alone (Lane 1 and 2) . Lanes 2 and 4 represent lysates that have been chondroitinase ABC-treated, while 1 and 3 are untreated lysates. Antiserum used was directed against RPTP/3, as described in Section 6.1.5.

FIG. 11. Effects of the proteoglycan 3F8 on aggregation of Ng-CAM-Covaspheres. Green-fluorescing Ng-CAM-Covaspheres after incubation for 2 hours at 25° (A) in the presence of 10 μg/ml of BSA. (B) 30 μg/ l 3F8 proteoglycan. Covaspheres were visualized using a Nikon Diaphot microscope equipped for fluorescence and were photographed using a N2000 camera.

FIG. 12. Inhibition of NG-CAM-Covasphere aggre¬ gation by 3F8. The appearance of superthreshold aggregates of Covaspheres coated with Ng-CAM was measured after 2 hours using a Coulter counter. 6 μl samples of Ng-Covaspheres were mixed in a final volume of 60 μl PBS in the presence of variouε concentrations of native 3F8 proteoglycan (PG, solid lines) and chondroitinase-treated 3F8 proteoglycan core (Chase, dashed lines) from 7-day brain. Data represent means (N=3) ± the standard error of the % of the control levels of superthreshold aggregates detected. The mean level of superthreshold particles in control samples was 32,582 ± 788.

FIG. 13. Inhibition of N-CAM-Covasphere aggre¬ gation by chondroitinaεe-treated 3F8 (circleε) . The appearance of superthreshold aggregates of Covaspheres coated with N-CAM was measured after 2 hours.

Chondroitinaεe-treated 3F8 used. Data represents a mean (N=3) of the % control levels of superaggregates detected. The mean level of superthreshold particles in control samples was 19,993 +/- 2,190. FIG. 14. Comparison of the amino acid sequences of the carbonic anhydrase domains contained in rat 3F8 and human RPTPS proteinε. Top sequence represents the RPTP3 sequence, bottom line the 3F8 sequence.

5. DETAILED DESCRIPTION OF THE INVENTION This invention involves a new class of receptor protein tyroεine phoεphatase molecule whose members are proteoglycans and/or possess an extracellular carbonic anhydrase structural domain. In addition, a family of extracellular molecules, the cell adhesion molecules (CAMs) , that bind to, and act as ligands for, this new class of receptor is also described. The discovery that CAMs bind receptor protein tyrosine phosphatases represents the first identification of a natural ligand for this type of receptor. Binding of two CAMs, namely N-CAM and Ng-CAM, to the receptor protein tyrosine phosphatases of the invention is demonstrated in the working examples preεented herein. A number of uses for the receptors and the receptor- binding ligands of the invention are alεo encompaεsed in the invention. Briefly, the receptor and the receptor-binding ligands may be used to develop compounds and strategies for modulating cellular processes under the control of the receptor protein tyrosine phosphatases. Such processeε include, but are not limited to, normal cellular functionε εuch aε differentiation, metaboliεm, cell cycle control, and neuronal function; cellular behaviorε such as motility, contact inhibition, and signal transduction; in addition to abnormal or potentially deleterious processes such as virus-receptor interactions, inflam¬ mation, cellular transformation to a cancerous state, and the development of Type 2, insulin independent diabetes mellitus. Compounds that may interfere with ligand binding are described and methods for identifying other potential ligands, such as CAM-type ligands, growth factors, or extracellular matrix components, are discussed. Finally, working examples are presented in which one member of the new RPTPase family of molecules, RPTPβ, is characterized, and two members of the CAM family, N-CAM and Ng-CAM, are shown to bind to the new RPTPase class of receptor.

5.1 RPTPases

The RPTPases of the invention that are proteoglycans may be modified with macromolecules composed of glycosa inoglycan (GAG) chains (glycans) covalently bound to the RPTPase protein core. GAG components may consist of such units as hexosamine (D- glucosamine (GlcN) or D-galactosamine (GalN) ) , and either hexuronic acid (HexA; D-glucuronic acid (GlcA) or L-iduronic acid (IdoA)) or galactose units (as in keratin sulfate) that are arranged in alternating, unbranched sequence, and carry sulfate substituents in various poεitions. The glycan backbones of the RPTPase molecules may include, but are not limited to, a basic structure composed of (HexA-GalN)_n, (HexA- GlcN)_n, or (Gal-GlcN)_n disaccharide units. While these structures connote the basic structure of the RPTPase modifications, such modifications may also contain marked heterogeneity within as well as between the individual polysaccharide chainε. Such heterogeneity iε an expected byproduct of the mechanism of GAG biosynthesis, and may include, but iε not limited to differences in εulfate substitutions along the chain and epimerization of one unit to another (GlcA to IdoA, for example) . At least one glycan chain must be attached to the protein core of each proteoglycan RPTPase. Glycan chainε may, but are not required to, be attached to the protein core at the serine (Ser) amino acid residue of the sequence, Ser-Gly-X-Gly, where Gly iε a glycine amino acid residue and X is any amino acid residue. The members of the RPTPase class of the invention may include an extracellular stretch of amino acids that shares similarity with the known carbonic anhydrase isoformε (Deutεch, H.F., 1987, Int. J. Biochem. 19_.:101-113) . Such sequences need not have carbonic anhydrase enzymatic activity. One or more complete or partial carbonic anhydrase motifs may be present on a single RPTPase molecule. Within the CAH region of similarity there may exist amino acid substitutions, as well as short amino acid deletions, and/or short amino acid additions that diverge from the known CAH isoforms. Such divergent sequences are acceptable as long as the overall amino acid sequence similarity to CAH remains at least about 25% and/or the tertiary structure or the domain remains similar to that of CAH.

Presented in the working example in Section 6 is the characterization of one member, RPTP/3, of this new class of RPTPase molecule. In this example, it is shown that RPTP3 not only contains a CAH-like domain but is also a proteoglycan.

5.2 RPTPase LIGANDS The molecules that act as the preferred ligands for the receptors of the invention are cell adhesion molecules (CAMs) . Such moleculeε include, but are not limited to, any member of the classes of Ca²⁺-indepen¬ dent CAMs, cadherins, which are Ca²⁺-dependent CAMs, and integrinε, which are Ca²⁺- or Mg²⁺-dependent CAMs. Ca²⁺-independent CAMs include such molecules as the N- CAM family, Ng-CAM, LI, Jl, Fasciclin III, and MAG molecules. The cadherins include such molecules as N- cadherin, E-cadherin, P-cadherin, L-CAM, B-cadherin, and T-cadherin. As is demonstrated in the working example presented in Section 7, two members of the CAM family of molecules, N-CAM and Ng-CAM, bind members of the RPTPase clasε of molecule described in this invention.

It had previously been εpeculated that receptor phosphatases themselves may function as cell adhesion molecules because some of them contain motifs such as IgG-like or fibronectin Type III repeats typical of CAMs. In addition, because CAMs are known to undergo homotypic ("like" molecule) binding, it had been proposed that PTPases with IgG and fibronectin motifs may also under go homotypic interactions. It is of note, however, that IgG-like and fibronectin motifs are found in many surface receptors and proteins which do not undergo homotypic interations. Contrary to thiε proposal, though the working Example of Section 6 described herein demonstrateε that CAMs act as ligands for the RPTPaεe molecules of this invention, which contain no IgG-like and fibronectin Type III motifε. Thuε, even in the absence of peptide domain similarities, a ligand/receptor interaction does, in fact, occur between the RPTPase clasε of molecule diεcloεed in thiε invention and CAMε, where no interaction haε previouεly been predicted to occur.

The ligands of the invention may be transmembrane proteins, glycosylphosphatidylinositol-linked membrane proteinε, or εecreted proteins. The moleculeε that constitute the ligands of this invention may contain one or more peptide domains, including, but not limited to, one or more Ig (immunoglobulin) domains (Williams, A.F., 1987, Immunol. Today 8_:298-303), one or more fibronectin type III domains (Hynes, R.O., 1990, Fibronectins, Springer-Verlag, New York) , and/or one or more ectodomains (Takeichi, M. , 1991, Science 251:1451-1455) . Ig domains may share characteristics with both immunoglobulin constant and variable regions. Such characteristics may include pairs of cysteine residueε, spaced approximately 60 amino acids apart, that form disulfide bonds with each other. Molecules may exhibit one or amino acid repeats of the sequence DRE, DXNDN, DXD, DVNE, DXE, and/or DPD. If the molecules are transmembrane proteins, such sequences should be present in the extracellular portion of the molecule.

Because the RPTPase moleculeε of thiε invention may be proteoglycans, several other non-CAM-like ligands may exist. For example, such extracellular matrix molecules as vitronectin, fibronectin, and laminin have been known to bind to the GAGs of certain proteoglycans. Also, growth factors, such as fibro- blast growth factors, and Schwann cell growth factor, have also been demonstrated to have affinity for proteoglycan GAG chains. Therefore, molecules including, but not limited to extracellular matrix molecules and growth factors are potential ligands for the RPTPase class of molecule presented in this invention.

5.3 USES AND ADMINISTRATION OF RPTPaεe LIGANDS Depending on the individual molecule, εome RPTPase molecules may become activated upon ligand binding, and others may become inactivated (the activity referred to here being the RPTPases' phospha¬ tase activity) . Ligand binding to RPTPase molecules may affect a variety of cellular processes. Such processes include, but are not limited to, normal cellular functions such as differentiation, metabo- lism, cell cycle control, and neuronal function; cellular behavior, such aε motility and contact inhibition; in addition to abnormal or potentially deleterious processes such as virus-receptor interac- tions, inflammation, cellular transformation to a cancerous state, and the development of Type 2, insulin independent diabetes mellitus. RPTPase/CAM binding may exert an effect on the above-mentioned processes within the RPTPase-exhibiting cell. In addition, because CAMs are often cell surface proteins, RPTPase/CAM binding may elicit an effect on the CAM-exhibiting cell. Alternatively, RPTPases may contribute to the control of such cellular processes by exerting an effect directly on the CAM ligand itself, via, for example, a CAM phosphorylation/ dephosphorylation reaction. The receptors and the receptor-binding ligands of the invention may be used as drugs that can modulate the cellular processes under the control of the RPTPaseε. In addition, methodε are presented below for the identification of compounds that affect RPTPase activity, and such compounds may also be uεed as drugs that can modulate one or more of the cellular processeε mentioned above. The receptors or their ligands may be used directly to modulate processes such as those mentioned above. For example, soluble RPTPases may be adminis¬ tered, uεing techniques well known to those skilled in the art, that could act to compete with endogenous transmembrane receptor molecules for available ligandε, thuε reducing or inhibiting ligand binding to endogenouε RPTPases. The effect of such a procedure would be to activate, reduce or block the signal normally transduced into the cell (either the RPTPase- exhibiting cell, or the CAM-exhibiting cell) via ligand binding to transmembrane RPTPase. The RPTPases used here may include the entire molecule or, alternatively, only the RPTPase extracellular domain, or a part of the RPTPase extracellular domain thereof. In addition, ligands may be administered, again, using techniques well known to those in the art. Such administration would lead to a greater than normal number of transmembrane RPTPases being bound by ligand, potentially causing an amplification of the ligand-bound state within cells exhibiting RPTPases. Alternatively, the administered ligand may be compoεed of a modified form of said ligand such that receptor binding may still occur, but the normal result of such binding (receptor activation or inactivation, as the case may be) does not occur. A ligand with such a design would act in much the same way that administra¬ tion of soluble RPTPase would, in that both procedures would have the final effect of reducing the number of functionally bound RPTPase transmembrane molecules, therefore lowering or blocking the normal extracel- lular signal being transduced into the RPTPase- exhibiting cell via normal ligand binding to transmembrane RPTPase. The effect on a CAM ligand- exhibiting cell would also be the same in that an overall lower number of endogenous CAM ligands would be bound, therefore lowering or blocking the effect of RPTPase binding on such CAM-exhibiting cells.

Depending on the specific conditionε being treated, agentε may be formulated and adminiεtered systemically or locally. Techniques for formulation and adminiεtration may be found in "Remington's

Pharmaceutical Sciences," Mack Publishing Co., Easton, PA, latest edition. Suitable routes may include oral, rectal, transmucosal, or intestinal administration; parenteral delivery, including intramuscular, subcu- taneous, intramedullary injections, as well as intrathecal, direct intraventricular, intravenous, intraperitoneal, intranasal, or intraocular injec¬ tions, just to name a few. For injection, the agents of the invention may be formulated in aqueous solu- tions, preferably in physiologically compatible bufferε such as Hanks's solution. Ringer's solution, or physiological saline buffer. For such transmucosal administration, penetrants appropriate to the barrier to be permeated are used in the formulation. Such penetrants are generally known in the art.

RPTPases and/or their ligands may also be used to screen for additional molecules that can act to modu¬ late the activity of cellular processes such as those described above. For example, compounds that bind to RPTPase moleculeε may be identified. One method that may be pursued in the isolation of such RPTPase- binding moleculeε would include the attachment of RPTPase molecules to a solid matrix, such as agarose or plaεtic beadε, microtiter wells, or petri dishes, and the subsequent incubation of attached RPTPase molecules in the presence of a potential RPTPase- binding compound or compounds. After incubation, unbound compounds are washed away, and the RPTP-bound compoundε are recovered. In this procedure, large numbers of types of molecules may be simultaneouεly εcreened for RPTPase-binding activity. Bound molecules could be eluted from the RPTPase molecules by, for example, competing them away from the RPTPase molecules with the addition of excess ligand. The effect of a compound on the phosphatase activity of RPTPaεe moleculeε can also be determined. Such a compound may, for example, be one isolated using a procedure such as the binding technique described above. One method that may be utilized for determining the effects of a compound on RPTPase phosphatase activity would involve exposing such a compound to a preparation of cultured cellε that express the RPTPase of the invention, and subsequently measuring the phosphatase activity of the culture. The compound of interest may be introduced to the cells, for example, by addition of the compound to the tissue culture medium. The phosphatase activity of the cells within the tissue culture preparation may be determined by measuring the level of cellular phospho- tyrosine within the culture, using method that are well known in the art (Honegger et al. , 1987, Cell .51:199-209; Margolis et al.. 1989, Cell 57:1101-1107) . To properly determine the effects of addition of the compound, the cellular phosphotyrosine levels of the same type of tissue culture preparation that has not been exposed to this compound must also be measured, and the two levels must then be compared. For example, RPTPases may be incorporated into apparatuses including but not limited to affinity columns such that large numbers of molecules may be screened quickly by being applied to said apparatuses. Those molecules with an affinity for RPTPases will be bound. Such binding will also bring about a partial purifica¬ tion of the molecules of interest. In order to continue the purification process, the bound molecules εhould be eluted off the above deεcribed apparatuses, for example by competing them away from the RPTPases with excess ligand, and the process should be repeated until the molecule of interest is purified to the extent neceεsary.

6. EXAMPLE: CHARACTERIZATION OF THE RECEPTOR PROTEIN TYROSINE PHOSPHATASE RPTP3

The subsections below describe the characteriza¬ tion of a human receptor protein tyrosine phosphatase molecule, RPTPβ. It is shown that this RPTP/3 contains an extracellular carbonic anhydrase domain and is a proteoglycan. In addition, it is shown that RPTP/3 mRNA and protein are predominantly expressed in brain tissue. 5

6.1 MATERIALS AND METHODS

6.1.1 ISOLATION OF cDNA CLONES AND DNA SEQUENCE ANALYSIS

10 A cDNA clone containing a portion of the coding sequences for RPTP/3 was isolated after screening a λgtll human infant brain stem cDNA library under conditions of reduced stringency with a nick trans¬ lated LCA probe that included both phosphatase domains _ (Kaplan, R. et al. , 1990, Proc. Natl. Acad. Sci. USA 2:7000-7004). Since the 5' end of thiε gene was not present in the original clone, the library was rescreened with a DNA fragment that was generated from the 5'end of the original clone. The probe was

₂₀ labeled with ³²P dCTP utilizing the random prime method (USB) and hybridization was performed under moderately stringent conditions at 42°C in a buffer containing 50% formamide, 5XSSC, 20mM Tris-CL pH 7.6, IX Denhardt's solution, 0.1% SDS and 100 μg/ml of sheared

₂₅ and denatured salmon sperm DNA. After hybridization, phage filters were washed three times for 20 min at 50°C in a buffer containing O.lXSSC/0.1% SDS and then were processed for autoradiography. The brainstem library was rescreened a total of three timeε in order

₃₀ to iεolate overlapping cDNA clones that contained the entire coding εequence for RPTP . cDNA inserts from positive recombinant plaque- purified were subcloned into the plasmid vector, Blue Script (Stratagene) , and sequenced by the dideoxy

_₅ chain termination method using the Sequenase Version 2.0 Kit (USB) . 6.1.2 CHROMOSOMAL LOCALIZATION Isolation, propagation and characterization of parental and somatic cell hybrids used in this study have been described (Durst, M. et al. , 1987, Proc. Natl. Acad. Sci USA 84_.:1070-1074) . Presence of specific human chromosomes or regions of chromosomes has been confirmed by DNA hybridization using probeε for geneε assigned to specific chromosome regions. Fig. 2A depicts diagrammatically the chromosomes or partial chromosomes retained in most of the hybrids used.

Chromosomal in situ hybridization was performed as described previously (Cannizzano, L.A. et al.. 1991, Cancer Res. 5_1:3818-3820) . Slides containing etaphase chromosomes from normal male (46 XY) peripheral blood lymphocytes were aged at 4°C for 7-10 days and pretreated with ribonuclease A (Sigma) for 1 hour at 37°C. The chromosomal DNA was denatured in a hybridization mixture containing 50% formamide, 2X SSC and 10% dextran sulfate (pH 7.0). Hybridization was carried out at 37°C overnight. After rinsing at 39°C in three changes of 50% formamide and 2X SSC, then five changes of 2X SSC, slides were dehydrated, air dried, subjected to autoradiography, and banded with Wright's - Giemsa stain solution mixed with 1-3 parts of pH 9.2 borate buffer (Cannizzano, L.A. et al. , 1991, Cancer Res. 51:3818-3820) .

6.1.3 ISOLATION OF MOUSE SEQUENCES HOMOLOGOUS TO HUMAN RPTPβ

Two oligonucleotideε in conserved phosphatase domain II were εyntheεized according to the nucleotide sequence of human RPTP/3. These oligos, in conjunction with phage DNA from a mouse brain cDNA library that was purchased from Clonetech (Palo Alto, CA) , were used in the polymerase chain reaction with Taq polymeraεe (Perkin-Elmer) to amplify homologouε mouse RPTP3 sequences. The amplified product was purified and cloned into the Blue Script plasmid vector (Stratagene, La Jolla, CA) . Ho ology was confirmed by DNA sequence analysis as described above. This subcloned fragment will be referred to as pBSMBDII.

6.1.4 NORTHERN ANALYSIS Total cellular RNA was prepared with the Strategene RNA isolation kit. Poly A⁺ RNA was further selected utilizing oligo dT cellulose chromatography (Stratagene) . For Northern analysis, the RNA was separated on a 1.0% agarose/2.2 M formaldehyde gel and transferred to a Nytran membrane (Schleicher and Schuell) by capillary action. The membrane was prehy- bridized and hybridized in 0.5 M sodium phosphate pH 7.2, 7% SDS, lmM EDTA, 100 μg/ml salmon sperm DNA and then washed in 40mM sodium phosphate ph 7.2, 1% SDS, 1 mM EDTA at 65°C. For the blot containing RNA isolated from various mouse tissues, a ³²P-labeled probe was made utilizing pBSMBDII as template in the random prime labeling reaction (USB) . The human glioblastoma and neuroblastoma RNA blots were probed with labeled restriction fragmentε iεolated from different partε of the human RPTPβ cDNA clones.

6.1.5 ANTIBODIES A peptide derived from the carboxy-terminal 15 amino acids of human RPTP/3 was synthesized and coupled to Keyhole limpet hemocyanin according to previously published procedures (Harlow, E. and Lane, D. , 1988, in Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 77-88) . This was used as immunogen to inoculate two rabbits to produce polyclonal antisera against RPTP/3. Anti-EGF receptor immunoprecipitates were performed with RK2 antibody which recognizes the glycosylated and nonglycosylated forms of the EGF receptor (Kris, R.M. et al. , 1985, Cell 4_0: 619-625).

6.1.6 CELL LABELING AND IMMUNOPRECIPITATION The human neuroblastoma cell line, Lan 5, waε maintained in Dulbecco's modified Eagles medium (DMEM) contain 10% fetal bovine serum (FBS) . Cell treatment with tunicamycin involved incubating the cultures with 10 μg/ml tunicamycin (Sigma) for 1 hour prior to ³⁵S methionine labeling. Treated and untreated cells were washed twice with methionine free DMEM and then labeled for 4 hours with 0.15 mCi/ml ³²S methionine (purchased from New England Nuclear) in DMEM minus methionine containing 1% dialyzed FBS. During the labeling period, 10 μG/ml tunicamycin was added to the medium of the treated cells. Cells were then washed with ice cold phosphate-buffered saline and solubilized in a lysis buffer containing Hepes (N-2- hydroxyethylpiperazine-N*-2-ethane-sulfonic acid) pH 7.5, 150 mM NaCl, 1.0% Triton X-100, 10% glycerol, 1.5 mM MgCl₂, 1 mM ethylene glycol-bis (B-aminoethyl ether)-N, N, N', N'-tetracetic acid (EGTA) , 10 μg leupeptin per ml, 1 mM phenylmethylsulfunyl fluoride, and 10 μg aprotinin per ml. For ³⁵S-NaS0₄ labeling, cells were washed twice with phosphate-buffered saline (PBS) , at which time labeling mix was added (NEX-041H medium plus 10% calf serum, gentamicin, and 200 μCi/ l of ³⁵S-NaS0₄) . Cells were labeled for 20 hours, washed twice with PBS, and solubilized as described above. Cell lysateε were clarified and then immunoprecipitated. Lysates from ³⁵S-Methionine- labeled cultures were immunoprecipitated with normal rabbit antiserum, anti-RPTPβ antiserum or RK2 antiserum for 2 hours at 4°C. Lysates from ³⁵S-NaS0₄- labeled cultures were immunoprecipitated without preclearing, with anti-RPTPβ antiserum for 2 hours at 4°C. The immunocomplexes were then precipitated with Protein A Sepharose (Sigma) for 45 min at 4°C and washed 10 times with RIPA buffer (20mM Tris-Cl ph 7.6, 300 mM NaCl, 2mM EDTA, 1.0% Triton X-100, 1.0% sodium deoxycholate and 0.1% SDS). The immunoprecipitated material was analyzed on SDS-polyacrylamide gels (7.5% for ³⁵S-Methionine, 5% for ³⁵S-NaS0₄) and then fluorographed.

6.1.7 IN SITU HYBRIDIZATION ANALYSIS Fresh frozen tissue was cut on a cryostat into 20 μm thick sections and thaw mounted onto gelatin coated slides. The sections were fixed in 4% paraformaldehyde in 0.1 M sodium phosphate (pH 7.4) for 30 min and rinsed 3X for 5 min each in 0.1 M sodium phosphate and IX for 10 min in 2X SSC. The probe used in the hybridization analysis was a 49 base oligonucleotide complementary to a portion of the RPTPβ mRNA in phosphatase Domain II. The oligonucleotide was labeled with [α-³⁵S] dATP (NEN Dupont) using terminal deoxynucleotidyltransferaεe (Boerhinger Mannheim) and purified uεing Sephadex G25 quick spin columns (Boerhinger Mannheim) . The specific activity of the labeled probes was between 5 X 10⁸ - 1 x 10⁹ cpm/μg. Prehybridizationε and hybridizationε were carried out in a buffer containing 50% deionized formamide, 4X SSC, IX Denhardt'ε, 500 μg/ml denatured εalmon εperm DNA, 250 μg/ml yeaεt tRNA and 10% dextran εulfate. The tissue waε incubated for 12 h at 45°C in hybridization solution containing the labeled probe (1 x 10⁶ cpm/section) and 10 mM dithiothreitol (DTT) . Controls for specificity were performed on adjacent sections by diluting the labeled oligonucleotides with a 30 fold concentration of the appropriate unlabeled oligonucleotide and by hybridization with a sense probe. After hybridization the sections were washed in 2 changes of 2X SSC at room temperature for 1 h, IX SSC at 55°C for 30 min. , 0.5X SSC at 55°C for 30 min, 0.5X SSC at room temperature for 15 min and dehydrated in 60%, 80%, and 100% ethanol. After air drying, the sections were exposed to X-ray film for 5-10d.

6.2 RESULTS

6.2.1 THE PRIMARY AMINO ACID SEQUENCE OF RPTPβ

Four overlapping cDNA clones were isolated from the human brain stem library that contained the entire coding sequence for RPTPS. The primary amino acid sequence deduced from the nucleotide sequence of the cDNA cloneε is shown in FIG. 1. RPTP/3 belongε to the high molecular weight, tranεmembrane class of tyrosine phosphatases and is encoded by 2308 amino acids. The protein contains a signal peptide (underlined in FIG.

1) followed by a long extracellular domain of 1636 amino acids containing 21 potential N-glycoεylation sites (indicated by the arrows) . A hydrophobic, transmembrance peptide (bold sequences) joins the extracellular portion of the protein to two tande ly repeated and conserved phosphatase domains (designated

DI and DII) .

6.2.2 CHROMOSOMAL LOCALIZATION OF HUMAN RPTP0 The chromosomal localization of the human RPTP/3 gene was initially determined utilizing a panel of rodent-human hybrids carrying defined human chromo¬ somes or chromosome regions. DNAs from seventeen rodent-human hybrids, carrying overlapping subsets of human chromosome regions representing the entire human genome (see FIG. 2A) were cleaved with Eco RI, electrophoresed, transferred to filters and hybridized to the radiolabelled human RPTPβ probe. Results are summarized in FIG. 2A in which it can be seen that the presence of human chromosome 7. A more precise localization of the RPTP/3 gene was determined by chromosomal in situ hybridization to metaphase chromosomes of normal human lymphocytes. This tech¬ nique places the RPTP/3 gene at 7q31-33 with the most likely position at 7q31.3-7q32, which is diagrammatically shown to the right of the chromosome 7 sketch in Fig. 2B.

6.2.3 NORTHERN BLOT ANALYSIS Northern hybridization analysiε of variouε murine tissue RNAs was performed to determine the tissue specific expression of RPTP/3. The probe used in this analysis was a portion of the murine homolog of RPTPjβ that was amplified in the polymeraεe chain reaction (see Materials and Methods) and contains 405 nucleo- tides encoding 135 amino acids of Domain II. Baεed on a nucleotide sequence comparison to the equivalent region of the human cDNA clone, the murine and human clones are 88% identical at the nucleotide level in this region of Domain II of RPTP/3. The results of thiε Northern analysiε (FIG. 3A) indicate the presence of two major transcripts of 8.8 and 6.4 kb, respecti- vely. A minor transcript of greater than 9.4 kb is sometimes observed and might represent cross-reaction to a highly related phosphatase. Both the transcripts are restricted to brain tisεue, and could not be detected in other tissue. The absence of expresεion of RPTPβ in the majority of tiεεues is not due to degraded RNA since the presence of intact actin trans¬ cripts was observed utilizing the same blot (FIG. 3B) . Because the expres ion of RPTP/3 was restricted to brain tissue, expression of this phosphatase in different human glioblaεtoma cell lineε and a human neuroblastoma cell line, Lan 5, was examined. A human RPTP/3 probe hybridized to three major transcripts of 8.8, 7.5 and 6.4 kb respectively (FIG. 3C) . These transcripts were only detected in RNA isolated from the Lan 5 neuroblastoma cell line and were absent in the RNA isolated from the four glioblastoma cell lineε even though similar amounts of total cellular RNA were loaded as determined from ethidium bromide staining of the 28S and 18S ribosomal RNAs. The 8.8 and 6.4 kb transcripts were identical in size to the two transcripts observed in RNA isolated from mouse brain tissue (FIG. 3A) . The presence of three transcripts in Lan 5 RNA could be due to cross hybridization with other highly related phosphatases since the probe used in this analysis was derived from sequences in the conserved phosphatase Domain I or be due to alteratively spliced RPTPβ transcripts. In order to obtain further insight into the nature of the three RPTP3 transcripts, a similar Northern analysis was performed on RNA isolated from Lan 5 cells with probes isolated from the 5' portion of the human cDNA clones. The probes utilized were derived from sequences in the extracellular domain that are unique for RPTP/3. This analyεiε showed an identical pattern of hybridization as what is observed in FIG. 3C. These results suggest that the three distinct transcripts are products of alternatively spliced mRNAs. 6.2.4 IDENTIFICATION OF A VARIANT FORM OF RPTPβ The overlapping human cDNA clones collectively contain greater than 8.1 kb of coding and noncoding sequences and appear to represent the largest tran- script that is 8.8 kb in length. In screening the human brain stem library and another human caudate library (Stratagene) , two independent cDNA clones, called Bsdll and Caudll, were isolated that each con¬ tained an identical deletion of 2581 nucleotides from the extracellular domain of RPTPβ. This deletion did not introduce a stop codon or interrupt the open reading frame of RPTPβ and joined amino acid 754 to amino acid 1615 as shown in FIG. 4A. The deleted clones maintained the same extreme 5' and 3 ' ends of the RPTP/3 gene in addition to the sequences encoding the transmembrane peptide and the two phosphatase domains. A transcript corresponding to the deleted clone would be approximately 2.6 kb smaller than the transcript corresponding to the undeleted, full-length clone. As shown in FIG. 3C, there is a transcript of 6.4 kb that is approximately 2.4 kb εmaller than the largest transcript which is 8.8 kb in length. In order to determine whether the 6.4 kb transcript repreεents the deleted form, Northern blot hybridization analysis was performed utilizing RNA iεolated from the Lan 5 cell line. Duplicate blots were made from this RNA and hybridized with two distinct probes. One probe (probe 1) waε derived from sequences in the 5' end of RPTP3 that are present in the deleted and full length cDNA clones. The other probe (probe 2) encompasses the sequences that are no longer present in the deleted cDNA clones. The location of probes A and B in the full length RPTPβ cDNA is shown in FIG. 4A. Comparison of the Northern analysis with the two probes revealed that probe l hybridized to the three distinct transcript (FIG. 4B, PI) whereas probe 2 hybridized to the 7.5 and 8.8 kb transcripts but failed to hybridize to the 6.4 kb transcript (FIG. 4B, P2) . These results provide preliminary evidence that the 6.4 kb transcript represents a deleted and alternatively spliced form of RPTP/3. The nature of the 7.5 kb transcript remains to be determined.

6.2.5 IN SITU HYBRIDIZATION ANALYSIS

In order to obtain further insight into the expression of RPTP/3, an in situ hybridization analysis was performed to look for the expreεsion of RPTP/3 in the developing mouse embryo. Studies were also performed to determine whether RPTP3 gene expression iε diffuse or restricted to specific regions of the adult brain. The results of thiε analysis confirm that RPTP/5 is preferentially expressed in the central nervous system. In the day 20 mouse embryo (E20) , high level of expression was observed in the ventricular zone of the brain (FIG 5A.) and in the spinal cord. A similar pattern of expression, with variable levels of intensity, has been seen from embryonic day 13 to postnatal day 7. The level of expression is much lower in the adult brain, and is discretely localized to the Purkinje cell layer of the cerebellum, the dentate gyrus, and the anterior horn of lateral ventricle (FIG. 5B) . The addition of a 30 fold excesε of unlabeled oligonucleotide completely blocked the labeling in all of theεe areaε indicating that this probe is hybridizing to mRNA in a sequence specific manner. Results from the Northern blot and in situ hybridization analyses demonstrate that RPTP/3 has a restricted tissue specificity to specific regions in the central nervous system and therefore may play an important role in the development of the nervous system.

6.2.6 ENDOGENOUS RPTPβ PROTEIN EXPRESSION Since RPTP3 transcripts were identified in the Lan 5 neuroblastoma cell line, these cells were subse¬ quently used to detect endogenous protein expression. Cell lysates prepared from cultures labeled with ³⁵S- methionine for 4 hours were immunoprecipitated with normal rabbit serum or anti-RPTP/3 antiserum (FIG. 6) . A protein with apparent weight of approximately 300 kd was recognized by the immune but not by the normal rabbit serum (lanes 1 and 2) . Since there are 21 potential N-glycosylation sites, it was necessary to determine whether the protein immunoprecipitated by the anti-RPTP/3 antiserum represented the core protein or a glycosylated form of the protein. In order to do this, tunicamycin was added to the cells during the ³⁵S-methionine labeling period. The effects of tunicamycin treatment on RPTP3 mobility was compared to the cell line were drug'ε ability to inhibit the glycosylation of the EGF receptor, which is also expressed in this cell line. Untreated cell lysates and lysates prepared from cells treated with tunicamycin were immunoprecipitated with an antibody (RK2) that recognizes the 170 kd glycosylated form and the 135 kd nonglycosylated form of the EGF receptor. (Kris, R.M. et al.. 1985, Cell 4J):619-625; and FIG. 6, lanes 4 and 5) . The protein immunoprecipitated with anti RPTPβ antiserum from Lan 5 cellε, that had been metabolically labeled in the preεence of tunicamycin, migrated faster than the immunoprecipitated product isolated from the untreated cells (compare lanes 3 and 2) . The molecular weight of the protein detected in FIG. 6, lane 3, is approximately 250 kD a value consistent with that of the core protein whose predicted molecular weight as deduced from the amino acid sequence is approximately 254 kd.

6.2.7 THE PRESENCE OF CARBONIC ANHYDRASE

RELATED SEQUENCES IN THE EXTRACELLULAR DOMAIN OF RPTPβ

A εtretch of 283 amino acids in the extracellular domain of a related transmembrane-type phosphatase, RPTP7 that shares homology with different isoforms of the enzyme carbonic anhydrase (CAH) has recently been identified. A similar stretch of CAH-related amino acids in RPTP3 has now been identified at the extreme amino terminus of the protein (designated as CAH-like in FIG. 1) . Alignment of the CAH-related domains of RPTP/3 and RPTP7 with the six known isoforms of CAH is shown in FIG. 7A. FIG. 7B shows the percent similarity, taking into account conservative amino acid substitutions between the CAH-related domain of RPTP/3 and the corresponding domain of RPTP7 and the six CAH enzymes. The CAH-related domain of RPTP/3 ranges anywhere from 45-50% in amino acid sequence similarity to the six different isoforms of CAH. The highest degree of similarity (58%) exists between the CAH-related sequences of RPTP/3 and RPTP7. Therefore it appears that RPTPaεes, β and 7, may represent a new subgroup of tyrosine phosphatases that will be charac¬ terized by the presence of CAH-related sequences in their extracellular domains.

6.2.8 RPTPβ IS A PROTEOGLYCAN In this εeries of experiments, it is demonstrated that RPTP/3 exhibits the characteristicε of a proteoglycan. Specifically, it is shown that the RPTP/3 protein is covalently modified with high molecular weight, sulfate-containing moieties, and that such moieties are sensitive to chondroitinase ABC treatment.

6.2.8.1 ³⁵SULFATE LABELING In order to begin to address what post-transla- tional modifications RPTP0 undergoes, 293 cells transfected with RPTP/S DNA as well as control, 293 cells transfected with vector alone were ³⁵S-NaS0₄ labeled. Immunoprecipitations of the lyεateε, using an anti-RPTPjβ antiserum were then performed. The gel illustrated in FIG. 8 shows the results of one such immunoprecipitation. As can be seen in lane 3, a significant amount of labeled material that reacts with the RPTP_/8 antiserum does not enter the running portion of the gel. This is easily contrasted to the non-transfected control lysate in lane 4, in which no such material is detectable.

6.2.8.2 ³⁵S-METHIONINE LABELING in continuing to investigate the posttranεla- tional modifications that RPTP3 undergoes, 293 cellε tranεfected with RPTP3 DNA aε well as control 293 cells transfected with vector alone were ³⁵S-methionine labeled. Immunoprecipitationε of the lysates, using an anti-RPTPS antiserum were then performed. The gel illustrated in FIG. 9 εhows the results of one such immunoprecipitation. Lane 4 contains a large amount of labeled material that reacted with the anti-RPTPβ antiserum which does not enter the running portion of the gel and a significant amount that does not even enter the stacking portion of the gel. By contrast, lane 5, which contains the control lysate, exhibits no such material. 6.2.8.3 CHONDROITINASE TREATMENT 293 cellε transfected with RPTP0 DNA as well as control 293 cells transfected with vector alone were ³⁵S-methionine labeled. Lysates were immunoprecipitated using an anti-RPTPβ antiserum and then chondroitinase ABC treated for 1 hour. The gel illustrated in FIG. 10 shows the resultε of one such immuno¬ precipitation. Lane 3 and 4 contain non-treated and treated RPTP3-transfected lysates, respectively. As can be seen, the bulk of the labeled material that had not entered the gel in the non-treated sample is absent in the treated sample, and in its place, a labeled band of a lower molecular weight has appeared. In lanes 1 and 2 are non-treated and treated control lysates, respectively. No such shift in high molecular weight labeled material is detectable here.

EXAMPLE: THE CELL ADHESION MOLECULES, N-CAM AND Ng-CAM ARE LIGANDS OF THE RECEPTOR PROTEIN TYROSINE PHOSPHATASE, RPTPβ

The experiments described below demonstrate that a receptor protein tyrosine phoεphatase, the human RPTPβ molecule, binds the cell adhesion molecules N- CAM and Ng-CAM. Section 7.2.1 demonstrate that the rat proteoglycan, 3F8, binds these two CAM molecules. Section 7.2.2, demonstrateε that the carbonic anhydraεe domainε of rat 3F8 and human RPTP/3 are nearly identical, leading to the conclusion that RPTP3 is the human counterpart of the rat proteoglycan 3F8. When taken together, these two pieces of information, indicate that RPTPβ alεo bindε the two CAM molecules, N-CAM and Ng-CAM. 7.1 MATERIALS AND METHODS 7.1.1 PROTEINS AND ANTIBODIES Ng-CAM and N-CAM were purified from 14-d embryonic chicken brains by immunoaffinity chromato- graphy using specific monoclonal antibodies (Grumet, M. and Edelman, G.M. , 1988, J. Cell Biol. 106:487- 503) . Analysis of the proteins on SDS/PAGE showed that Ng-CAM consisted of a major component of 135 kDa and lesεer amounts of the 200 kDa and 80 kDa species as described (Grumet, M. and Edelman, G.M. , 1988, J. Cell Biol. 106:487-503) and N-CAM ran as hetero- disperse species above 12 kDa as described previously (Hoffman, S. et al. , 1982, J. Biol. Chem. 257:7720- 7729) . Chondroitin sulfate proteoglycans were extracted with PBS from the brains of 7-day or 2- to 2-month old Sprague-Dawley rats, and purified by ion exchange chromatography and gel filtration (Kiang, W.-L. et al.. 1981, J. Biol. Chem. 256:10529-10537) . 3F8 proteoglycan waε then isolated by immunoaffinity chromatography, using monoclonal antibodies coupled to CNBr-activated Sepharose 4B (Rauch, U. et al.. 1991, J. Biol. Chem. 266:14785-14801) . Analysiε of the proteins on SDS-PAGE following chondroitinase-treat- ment showed that the core glycoprotein obtained by chondroitinase treatment of the 3F8 proteoglycan from either early poεtnatal or adult brain migrated on SDS- PAGE aε a single bad at 400 kDa (Rauch, U. et al.. 1991, J. Biol. Chem. 266:14785-14801) . For studies of the core proteins, proteoglycans were digested for 45-60 min at 37°C with protease-free chondroitinase ABC (Seikagaku America Inc., Rockville, MD) in 100 mM Tris-HCl buffer (pH 8.0 at 37°C) containing 30 mM sodium acetate. A ratio of 1.5 mM chondroitinase/μg proteoglycan protein was used for the 3F8 proteoglycan. Completeness of digestion was confirmed by SDS-PAGE, which demonstrated that the large native proteoglycan which did not enter the separating gel was converted to discrete core glycoprotein bands after enzyme treatment (Rauch, U. et al. , 1991, J. Biol. Chem. 266:14785-14801) .

Polyclonal rabbit antibodies raised against chicken Ng-CAM were prepared as previously described (Grumet, M. et al.. 1984, Proc. Natl. Acad. Sci USA 81:267-271).

7.1.2 COVASPHERE AGGREGATION Proteins (50μg) were covalently coupled to 200μl of 0.5-μm Covaspheres (Duke Scientific Corp., Palo Alto, CA) , washed twice in PBS containing 1 mg/ml

BSA/10 mM NaN₃, and resuspended in 200μl of buffer as previously described (Grumet, M. and Edelman, G.M., 1988, J. Cell Biol. 106:487-503) . Quantitative measurements indicated that under these conditions approximately 20% of the Ng-CAM protein was bound to the Covaspheres. For Covasphere aggregation experi¬ ments, prior aggregates in the bead preparationε were firεt diεsociated by sonication for 10-20 sec, and 6 μl aliquots (containing about 0.3 μg of protein) were mixed with 54 μg of PBS containing the indicated amounts of proteins. After a 30 min incubation on ice, the samples were resonicated- and aggregation was monitored at 25°C. The appearance of superthreshold aggregates of Covaεphereε waε measured using a Coulter Counter (Model ZBI) fitted with a 100 μm aperture set at amplication = 0.17, aperture current = 0.33, threshold 10-100; these settings allowed detection of particles >^~4 μm as described previously (Grumet, M. and Edelman, G.M. , 1988, J. Cell Biol. 106:487-503) . Superthreshold particles were measured in samples of 20 μl that were diluted into 20 ml of PBS. To test the sensitivity of proteoglycans to proteolysis, solutions containing 0.1 mg/ml proteoglycan were treated with 10 μg/ml of trypsin for 1 h at 37°C and the reaction was terminated by addition of 20 μg/ml of soybean trypsin inhibitor.

7.1.3 DNA SEQUENCING Sequencing was performed according to standard dideoxy techniques.

7.2 RESULTS 7.2.1 THE RAT PROTEOGLYCAN. 3F8, BINDS CAMS In previous studies, it was found that Ng-CAM (Grumet, M. and Edelman, G.M. , 1988, J. Cell Biol.

106:487-503) and N-CAM (Hoffman, S. and Edelman, G.M. , 1983, Proc. Natl. Acad. Sci. USA 80:5762-5766) individually mediated homophilic binding when recons¬ tituted into liposo es or when covalently bound to the surface of beads (Covaspheres) . The rate of aggrega¬ tion of Ng-CAM-Covaspheres was meaεured using a Coulter Counter to detect aggregates larger than a given size and was shown to be highly dependent on the concentration of Covaspheres in suspension. It was previously determined that at a concentration of -85 cm² of bead surface area/ml, the appearance of aggregates began to reach a plateau level after -1 h of incubation (Grumet, M. and Edelman, G.M. , 1988, J. Cell Biol. 106:487-503) . Therefore, to explore potential interaction between proteoglycans and CAMs, the effects of various proteins and proteoglycans on the rate of aggregation of Ng-CAM-Covaεphereε were tested. Whereas control proteins including BSA and fibronectin did not inhibit aggregation of the Ng-CAM- coated beads, one distinct chondroitin sulfate proteoglycan, 3F8, inhibited aggregation (FIG. 11). In contrast, aggrecan, a rat chondrosarcoma chondroitin sulfate proteoglycan (Doege, K.M. et al. 1987, J. Biol. Chem. 2J52:17757-17767) did not inhibit the aggregation, indicating that the effects were not simply related to the presence of chondroitin sulfate. This conclusion was further supported by the finding that the core glycoproteins obtained by chondroitinase treatment of the proteoglycans were equally effective in inhibiting the aggregation of Ng-CAM-Covasphereε (FIG. 12) . In contrast, most of the inhibitory activity was eliminated after treating the proteo¬ glycans with trypsin (see Materials and Methods, Section 7.1.2). These results demonstrate that the effects of the 3F8 proteoglycan on Ng-CAM binding are not mediated by the glycosa inoglycan chains, but that a protein domain (or possibly a cluster of oligosac- charides present on the proteoglycan core protein) is involved. Based on these results, all further experi- ments were performed using chondroitinase-treated proteoglycans.

The 3F8 proteoglycan inhibited aggregation of Ng- CAM-Covasphereε at 30 μg/ml (FIG. 11) . It is unlikely that the proteoglycans inhibited Covasphere aggregation by a trivial mechanism such aε proteolysis of Ng-CAM because it waε found that incubation of the 3F8 proteoglycan with Ng-CAM for 1 h at 37°C had no effect of the molecular εizes of the Ng-CAM components when resolved by SDS-PAGE. To compare the effects of different proteoglycanε we meaεured the appearance of εuperthreshold aggre¬ gates of Covaspheres using a Coulter Counter to detect aggregates larger than a given size. The aggregation of Ng-CAM-Covaspheres was inhibited in a concentra- tion-dependent manner by the 3F8 proteoglycan (FIG. 12) .

To determine whether the proteoglycans could affect other GkMs, a similar series of experiments waε performed using N-CAM coated beads. The aggregation of N-CAM-Covaspheres was inhibited in a concentration- dependent manner in the presence of the 3F8 proteo¬ glycan (FIG. 13) . These resultε are εimilar to those obtained using Ng-CAM-Covaspheres, (compare to FIGS. 11 and 12) .

The inhibitory effect of 3F8 proteoglycan on the aggregation of Ng-CAM- and N-CAM-coated beads were maximal at approximately 10 μg/ml. In a typical assay (60 μl) at this concentration of proteoglycans, the amount of proteoglycan in solution was 0.6 μg and the amount of Ng-CAM on the Covaspheres was approximately 0.3 μg (see Materials and Methods, Section 7.1.2), suggeεting that the brain proteoglycan can perturb homophilic Ng-CAM binding at approximately εtoichio- metric levelε with Ng-CAM.

7.2.2 RPTP0 REPRESENTS THE HUMAN COUNTERPART OF THE RAT PROTEOGLYCAN, 3F8

Comparison of the sequence of human RPTP3 with a partial sequence of a proteoglycan designated 3F8 cloned from a rat brain stem library (R. Margolis, personal communication) , reveals that these two proteins contain carbonic anhydrase-like domainε and are 91.9% identical at the amino acid level (FIG. 14) . The maximum amino acid εequence identity between the different memberε of the carbonic anhydraεe family of enzymeε is 64%. This sequence information indicates that the two proteins, the murine proteoglycan 3F8 and the human proteoglycan RPTP/3, are counterparts of each other. It iε apparent that many modifications and varia¬ tions of this invention as set forth here may be made without departing from the spirit and scope thereof. The specific embodiments described hereinabove are given by way of example only and the invention iε limited only by the terms of the appended claims.

Claims

WHAT IS CLAIMED IS:

1. An isolated proteoglycan tyrosine phosphatase receptor protein.

2. The receptor protein of Claim 1 f rther comprising an extracellular carbonic anhydrase structural domain.

3. The receptor protein of Claim 2 in which the receptor protein comprises RPTPjS.

4. An isolated tyrosine phosphatase receptor protein having an extracellular carbonic anhydrase structural domain.

5. The receptor protein of Claims 1, 2, 3, or 4 which noncovalently binds a ligand to form a receptor- ligand complex.

6. The receptor-ligand complex of Claim 5 in which the ligand is a cell adhesion molecule.

7. The receptor-ligand complex of Claim 6 in which the cell adhesion molecule is N-CAM.

8. The receptor-ligand complex of Claim 6 in which the cell adhesion molecule is Ng-CAM.

9. The receptor-ligand complex of Claim 5 in which the ligand is a growth factor.

10. The receptor-ligand complex of Claim 5 in which the ligand is an extracellular matrix component.

11. A method for identifying a ligand that binds to the receptor protein of Claims 1, 2, 3, or 4 comprising exposing at least one molecule to the receptor protein for a time sufficient to allow formation of a receptor-ligand complex, removing non- co plexed molecules, and detecting the presence of the molecule bound to the receptor protein.

12. A method for isolating molecules that bind to the receptor protein of Claims 1, 2, 3, or 4 comprising exposing a mixture of moleculeε to the receptor protein, removing non-complexed molecules, and eluting the molecules bound to the receptor protein to thereby obtain isolated molecules, capable of binding the receptor protein.

13. A method for identifying compoundε that modulate the enzymatic activity of the receptor protein of Claimε 1, 2, 3, or 4 comprising exposing cells exhibiting the receptor protein to known ligands for a time sufficient to allow formation of receptor- ligand complexes, measuring phosphotyroεine levels within the cells to obtain an enzyme activity level, and comparing the measured enzyme activity level to the enzyme activity level in cells not exposed to ligand,

14. A method of modulating the endogenous enzymatic activity of a tyroεine phosphatase receptor in a mammal in which the receptor is a proteoglycan or has an extracellular carbonic anhydrase domain, comprising administering to the mammal an effective amount of a ligand to the receptor protein to modulate the enzymatic activity.

15. A method of modulating the endogenous enzymatic activity of a tyrosine phosphatase receptor in a mammal in which the receptor is a proteoglycan or has an extracellular carbonic anhydrase domain, comprising administering to the mammal an effective soluble amount of the receptor protein or the extracellular domain of the receptor protein to modulate the enzymatic activity.

16. The method of Claim 14 or 15 in which the enzymatic activity of the receptor protein is increased.

17. The method of Claim 14 or 15 in which the enzymatic activity of the receptor protein is decreased.

18. The method of Claim 14 or 15 in which the modulation of receptor protein enzymatic activity regulates cellular functions compriεing differentiation, metaboliεm, cell cycle control, or neuronal function.

19. The method of Claim 14 or 15 in which the modulation of receptor enzymatic activity regulateε cellular behavior comprising contractility or contact inhibition.

20. The method of Claim 14 or 15 in which the modulation of receptor enzymatic activity regulates abnormal or deleterious processes comprising virus- receptor interactions, inflammation, cellular transformation to a cancerous state, or the development of Type 2, insulin independent diabetes mellituε.

21. A method of modulating cell adhesion molecule function in a mammal comprising administering to the mammal an effective amount of a ligand to the receptor protein of Claim 5 to modulate cell adhesion molecule function.

22. A method of modulating cell adhesion molecule function in a mammal comprising administering to the mammal an effective soluble amount of the receptor protein or the extracellular domain of the receptor protein of Clai ε l, 2, 3 or 4 to modulate cell adhesion molecule function.