Characterization of the interaction between CD45 and CD45-AP.

CD45, a leukocyte-specific transmembrane protein tyrosine phosphatase, is required for critical signal transduction pathways in immune responses. To elucidate the molecular interactions of CD45 with other proteins involved in CD45-mediated signal transduction pathways, we have recently cloned a 30-kDa phosphorylated protein, CD45-AP, which specifically associates with CD45. Binding analysis employing several deleted or chimeric forms of CD45-AP and CD45 demonstrated that the potential transmembrane segment of CD45-AP bound to the transmembrane portion of CD45. CD45-AP was found in particulate fractions of lymphocytes along with CD45, indicating that it is likely to be a transmembrane protein. In addition, CD45-AP was resistant to proteolysis by tosylphenylalanyl chloromethyl ketone-treated trypsin applied to intact cells. This is consistent with the most likely membrane orientation of CD45-AP predicted from the amino acid sequence, that is, only a short amino-terminal segment of CD45-AP is extracellular. We propose that CD45-AP interacts with CD45 at the plasma membrane and that the bulk of CD45-AP located in the cytoplasm act as an adapter which directs the interaction between CD45 and other molecules involved in CD45-mediated signal transduction pathways.

CD45 plays a critical role in signal transduction pathways essential for immune responses (1). The intracellular domain of CD45 exhibits protein tyrosine phosphatase activity (2) and differential usage of extracellular NH 2 -terminal exons results in several isotypes of CD45 which are variably expressed in a cell lineage-and developmental stage-specific manner (1,3). It has been proposed that CD45 activates src family protein tyrosine kinases associated with the T cell receptor signaling complex, p56 lck (4,5) and p59 fyn (6,7), by dephosphorylating their down-regulatory tyrosyl residues. Thus, CD45 appears to be required at the onset of the T cell receptor-mediated signal transduction pathway. CD45 appears to play a similar role in B lymphocytes in combination with their src protein tyrosine kinases (8). Furthermore, CD45 is required for signal transduction pathways of monocytes (9), natural killer cells (10), and mast cells (11). Physical associations between CD45 and various membrane proteins involved in T cell activation, such as CD2 (12), CD4 (13,14), CD8 (13,14), Thy-1 (15), T cell receptor (14), CD26 (16), and LFA-1 (14), as well as components of the B cell antigen receptor complex (17), have been documented under certain circumstances. It is not clear how specifically these proteins interact with CD45 and what their binding sites are. A cytoskeletal protein, fodrin, has been reported to interact with the cytoplasmic portion of CD45 and stimulate its protein tyrosine phosphatase activity (18). In addition, CD45 has been reported to be one of surface glycoproteins to which CD22, the B lineage-specific cell surface glycoprotein and a sialic acidbinding lectin, binds (19).
We have recently reported the molecular cloning of a novel 30-kDa phosphorylated protein, CD45-AP, which is specifically associated with CD45 (20). CD45-AP shares no homology with previously known sequences and appears to be expressed specifically in leukocytes. The association of CD45-AP with CD45 has been observed in various types of lymphocytes (21), indicating that the association involves the isotypically invariant portion of CD45. No tyrosine phosphorylation of CD45-AP has been detected either in vivo or in vitro, and the predicted sequence contains no consensus tyrosine phosphorylation sites (20), indicating that it is not a substrate for protein tyrosine kinases. In addition, the predicted sequence of CD45-AP does not contain conserved sequences of GTP-binding proteins. Given the importance and complexity of CD45-mediated signal transduction, it is essential to determine how this novel protein, CD45-AP, interacts with CD45. In the present study, we characterize the interaction between CD45 and CD45-AP and summarize our findings in a model.

EXPERIMENTAL PROCEDURES
Rabbit Antisera-Antisera against CD45-AP were raised by immunizing rabbits with CD45-AP (20) isolated from YAC-1 cells, a murine T cell line (22), as well as with a recombinant glutathione S-transferase (GST) 1 -CD45-AP fusion protein (described below). Rabbit antiserum to the cytoplasmic domain of CD45 (23) was a generous gift from Dr. H. Ostergaard. Rabbit antiserum which recognizes a common segment present in the extracellular portion of all CD45 isoforms (24) was a generous gift from Dr. J. Marth.
Construction of Deleted Forms of CD45-AP cDNA-CD45-AP cDNA was subcloned into pGEM-3Z plasmid (Promega) downstream from the T7 RNA polymerase promoter using the EcoRI site, and two deleted forms, "N" and "C", of the cDNA were derived from it. The N form contains a termination codon at amino acid position 75 and the region 3Ј to the new termination codon is eliminated. The C form starts from amino acid position 47, and the region 5Ј to the new initiation codon is eliminated. For construction of the pN1 form, the sequence from nucleotide positions 67 to 158 was amplified from the full-length CD45-AP cDNA by the polymerase chain reaction (PCR) and was cloned in pGEM-3Z. Therefore, the pN1 form contains a segment which corresponds to amino acid positions 1-24. Likewise, the pN2 form was constructed by amplifying the sequence from nucleotide positions 67 to 241 which corresponds to amino acid positions 1-52. A schematic drawing of each construct is shown in Fig. 1.
Binding Assay of in Vitro Translated CD45-AP-In vitro transcription and translation were carried out with the TNT TM T7-coupled reticulocyte lysate system lacking cysteine (Promega) with the addition of Tran 35 S-label TM (1076 Ci/mmol, ICN Biochemicals). CD45 immunocomplex was formed by incubating a YAC-1 cell lysate in 0.8% polyoxyethylene 10 oleyl ether (BRIJ 96) containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM MgCl 2 , 0.1 mM EGTA, 2.5 mM thioglycolic acid, and 1 mM phenylmethylsulfonyl fluoride (PMSF) with the anti-CD45 monoclonal antibody (mAb) (25)-conjugated beads (26) followed by extensive washing as described (20). LFA-1 and transferrin receptor immunocomplexes were formed in the same manner by employing anti-LFA 1 (27) and anti-transferrin receptor (28) mAbs, respectively. In vitro translated products and immunocomplexes were incubated at 37°C for 1 h for binding, followed by extensive washing in lysis buffer containing 0.4% BRIJ 96 in order to remove unbound material prior to SDS-PAGE analysis.
Construction and Preparation of Recombinant CD45-AP-The fulllength CD45-AP cDNA or its deleted forms, N and C, described above were subcloned into pGEX-1T expression vector (Pharmacia Biotech Inc.) downstream from GST using the EcoRI site. The pN1 and pN2 forms described above were subcloned likewise using the BamHI and EcoRI sites. The BL21(DE3) strain of Escherichia coli was transformed by each of these constructs, and plasmids obtained from the transformants were sequenced in the region of the fusion sites in order to confirm that the ligation occurred in frame. Recombinant proteins fused to the carboxyl terminus of GST and a control GST protein were produced in BL21(DE3) by isopropylthio-␤-D-galactoside induction with various efficiencies. Proteins were purified from equal volumes of bacterial culture by affinity chromatography using equal amounts of glutathione-Sepharose 4B (Pharmacia Biotech Inc.) for the binding assay. The amount of GST fusion proteins obtained, as assessed from Coomassie Blue-stained SDS-PAGE gels, varied among different recombinants 2 and was in the decreasing order of GST control Ͼ C Ͼ N ϭ pN1 ϭ pN2 Ͼ F.
Construction and Preparation of Recombinant Cytoplasmic CD45-Recombinant cytoplasmic CD45 was constructed, expressed, and purified as described before (29). Briefly, cDNA encoding amino acid residues 564-1268 of the cytoplasmic domain of mouse CD45 (numbered according to Ref. 30) was subcloned into the pET-three-dimensional vector (31) with an additional six histidine residues and a factor Xa cleavable site at the amino terminus. This construct was expressed in BL21(DE3) E. coli and purified using nickel affinity chromatography and ion-exchange chromatography.
L Cell Transfectants Expressing Truncated or Chimeric Recombinant CD45-A truncated version lacking the majority of the cytoplasmic domain (amino acid positions 574-1268) was constructed from the mouse CD45RABC cDNA (30) in pBluescript (Stratagene) by site-directed mutagenesis (32). The truncated and full-length forms of CD45RABC cDNA were subcloned into the eukaryotic expression vector, pH␤apr-1-neo (33) and were transfected into L cell fibroblasts using calcium phosphate precipitation. Stable, G418-resistant clones (L106A6 and L12B5) expressing surface CD45 were selected for further study. A stable G418-resistant L cell clone that expresses transfected CD44.TM2 on the surface was a generous gift from Dr. J. Lesley (34). CD44.TM2 is a CD44 mutant consisting of the extracellular and intracellular domains of CD44 and the transmembrane segment (ALIIFLVFLIIVTSI-ALLVVLY) of CD45 (34). Control L cells (L40B7) and their transfectants were cultured in the media described above for YAC-1 cells but containing supplements of 0.35-0.5 mg/ml of active G418 for the transfectants.
Binding Assay of Recombinant CD45-AP-CD45 was purified from YAC-1 cells by forming CD45 immunocomplex as described above and eluting in 50 mM diethanolamine containing 0.4% BRIJ 96, 150 mM NaCl, 2.5 mM thioglycolic acid, and 0.2 mM PMSF (21). HEPES-NaOH, pH 7.4, and EDTA were added to the eluate to final concentrations of 25 and 5 mM, respectively, and the pH of the eluate was adjusted to 7.4. Recombinant GST fusion proteins bound to equal amounts of glutathi-one-Sepharose 4B beads in 0.4% BRIJ 96, 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM MgCl 2 , 0.1 mM EGTA, 2.5 mM thioglycolic acid, 1 mM PMSF, 2 g/ml chymostatin, 2 g/ml leupeptin, and 10 g/ml trypsin inhibitor were incubated with various cell lysates, CD45 purified from YAC-1 cells, or purified recombinant cytoplasmic CD45 at 37°C for 1 h for binding. After extensive washing, material bound to the beads was analyzed by SDS-PAGE followed by autoradiography or Western blotting. Western blotting was carried out with a combination of rabbit antisera and horseradish peroxidase-conjugated Protein A or a combination of rat mAb and horseradish peroxidase-conjugated anti-rat Ig antibody using the ECL Western blotting system (Amersham Corp.).
Subcellular Fractionation of Lymphocytes and Localization of CD45 and CD45-AP-YAC-1 cells were suspended at 4 ϫ 10 7 cells/ml in cold hypotonic buffer consisting of 25 mM HEPES-NaOH, pH 7.4, 5 mM KCl, 1 mM MgCl 2 , 2.5 mM thioglycolic acid, and 1 mM PMSF and were disrupted with a tight-fitting Dounce homogenizer. The homogenate was centrifuged at 200 ϫ g for 10 min at 4°C, and the 200 ϫ g pellet (nuclear fraction) was resuspended in the hypotonic buffer supplemented with 0.8% BRIJ 96 and 150 mM NaCl. Insoluble material was removed by a brief centrifugation in a microcentrifuge and the final volume of the fraction was adjusted to be equal to the volume of the original cell suspension. The 200 ϫ g supernatant was centrifuged at 100,000 ϫ g for 1 h at 4°C, and 0.8% BRIJ 96 and 150 mM NaCl were added to the 100,000 ϫ g supernatant (cytoplasmic fraction). The 100,000 ϫ g pellet (microsomal fraction) was resuspended in gradient buffer consisting of 25 mM HEPES-NaOH, pH 7.4, 5 mM KCl, 5 mM NaCl, 0.1 mM EDTA, 2.5 mM thioglycolic acid, and 1 mM PMSF and was overlaid on a discontinuous sucrose gradient of 20, 35, 40, and 50% sucrose in gradient buffer. After centrifugation at 200,000 ϫ g for 2 h at 4°C (40,000 rpm in an SW40 swinging bucket rotor), material at the interfaces was collected into three fractions, i.e. M1 (the pooled interfaces of 0%/20% and 20%/35%), M2 (the 35%/40% interface), and M3 (the 40%/50% interface) (35). Each fraction was diluted in gradient buffer without sucrose and was centrifuged at 100,000 ϫ g for 40 min at 4°C. Each pellet was then resuspended in the same manner as described above for the nuclear fraction. A portion of each fraction was subjected to SDS-PAGE for Western blotting with antiserum against CD45-AP and horseradish peroxidase-conjugated Protein A. Another portion of the same fraction was immunoprecipitated by anti-CD45 mAb (25), and the immunoprecipitates were analyzed by SDS-PAGE.
TPCK-Trypsin Treatment of Lysates and Cells-YAC-1 cell lysates (10 ϫ 10 6 cell/ml lysis buffer) prepared in Hank's balanced salt solution containing 0.8% BRIJ 96, 25 mM HEPES-NaOH, pH 7.4, 1 mM EDTA, and 2.5 mM thioglycolic acid or intact cells (10 ϫ 10 6 cells/ml) suspended in Hank's balanced salt solution containing 25 mM HEPES-NaOH, pH 7.4, and 1 mM EDTA were treated without or with TPCK-treated trypsin (10 g/ml, specific activity 12,200 units/mg of protein) at 37°C for 1 h. After the incubation, PMSF and trypsin inhibitor were added to the lysate samples to final concentrations of 2 mM and 20 g/ml, respectively. The cell samples were washed once in the presence of 10% fetal calf serum and 20 g/ml trypsin inhibitor followed by lysis in 0.8% BRIJ 96 containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM MgCl 2 , 0.1 mM EGTA, 2.5 mM thioglycolic acid, 1 mM PMSF, and 20 g/ml trypsin inhibitor. A portion of each sample derived from an equal number of cells was subjected to SDS-PAGE for Western blotting with antiserum against either CD45-AP or the extracellular domain of CD45 followed by horseradish peroxidase-conjugated Protein A (24). Intact cells radiolabeled by incorporation of 14 C-labeled amino acid mixture were treated without or with TPCK-treated trypsin as described above, and immunoprecipitates obtained with mAb against LFA-1 (27) or the transferrin receptor (28) were analyzed by SDS-PAGE.

Binding of Various Deleted Forms of in Vitro Translated
CD45-AP to CD45-In a previous report, we have shown that in vitro translated CD45-AP bound specifically to CD45 (20). In the present study, the segment of CD45-AP that is responsible for the interaction with CD45 was determined by constructing deleted forms of CD45-AP cDNA and examining the ability of their in vitro translated products to bind to CD45. The in vitro translated full-length (F) or the deleted forms (N or C) of CD45-AP cDNA (Fig. 1) were either analyzed directly in SDS-PAGE or were examined for their ability to bind specifically to CD45 (Fig. 2). As expected, the in vitro translated full-length CD45-AP exhibited specific binding to the CD45 immunocomplex but not to the immunocomplexes of LFA-1 and the trans-ferrin receptor. The in vitro translated products of the N and C forms migrated at 14 and 22 kDa, respectively. The N form exhibited specific binding to CD45 whereas the C form failed to bind to CD45. These results indicate that interaction between CD45-AP and CD45 requires only a small NH 2 -terminal segment of CD45-AP, and the rest of CD45-AP, including most of the consensus phosphorylation sites, is not directly involved in the interaction.
Binding of Various Deleted Forms of Recombinant CD45-AP to CD45-In order to further define the segment of CD45-AP required for binding to CD45, two additional deleted forms of CD45-AP cDNA, designated pN1 and pN2, were prepared by PCR (Fig. 1). The full-length and the various deleted forms of CD45-AP cDNA were then ligated to pGEX-1T expression vector downstream from GST, and recombinant GST fusion proteins were produced in E. coli. The recombinant proteins bound to equal amounts of glutathione-Sepharose 4B were incubated with lysates of YAC-1 cells metabolically labeled by 14 C-labeled amino acid incorporation. Material bound to the beads was analyzed by SDS-PAGE (Fig. 3). As expected, a prominent band of 180 kDa which comigrates with CD45 immunoprecipitated from YAC-1 cells bound to the beads containing the GST fusion protein of full-length CD45-AP. The 180-kDa protein also bound to the GST fusion proteins of the N and pN2 forms but not to the GST fusion proteins of the C and pN1 forms. These results demonstrate that the potential transmembrane segment of CD45-AP is required for binding to CD45.
Binding of Recombinant Cytoplasmic Portion of CD45 to CD45-AP-Next we sought to determine what portion of CD45 interacts with CD45-AP. First, a recombinant cytoplasmic CD45 which lacks the entire extracellular domain and the transmembrane segment was produced in E. coli with six histidine residues fused to its amino terminus and purified as described previously (29). GST fusion proteins of the full-length (F) or the deleted forms (N or C) of CD45-AP bound to glutathione Sepharose 4B beads were incubated with either CD45 isolated from YAC-1 cells or the purified recombinant cytoplasmic CD45 protein. Material bound to the recombinant CD45-AP beads was subjected to SDS-PAGE and analyzed by Western blotting with antiserum against the cytoplasmic portion of CD45 (23) (Fig. 4). CD45 isolated from YAC-1 cells (YAC-1 CD45) bound to the full-length and the N form of CD45-AP, but not to the C form, in agreement with the results described above (Fig. 4A). The recombinant cytoplasmic form of CD45 (Cyto-CD45), however, failed to exhibit any significant amount of binding to any of the recombinant CD45-AP (Fig.  4B). The minute amount of the cytoplasmic CD45 bound to the N and C forms probably reflects nonspecific binding, since a similar amount of binding was also detected with control glutathione beads that did not contain any protein. 2 Binding of Recombinant Noncytoplasmic Portion of CD45 to CD45-AP-In order to examine whether the noncytoplasmic portion of CD45 binds to CD45-AP, recombinant forms of CD45 were expressed in L cells. When an anti-CD45 immunoprecipitate was obtained from L106A6 cells which were transfected with the full-length CD45RABC and was analyzed by Western blotting, a 220-kDa protein was detected, as expected, by antiserum which recognizes a common segment in the extracellular domain of all CD45 isoforms (24) (Fig. 5A). A minor band of 200 kDa observed in this sample probably represents a degradation product of the full-length CD45RABC. 3 When a similar analysis was carried out for L12B5 cells transfected with the noncytoplasmic form of CD45 which consists of the entire extracellular domain of CD45RABC and the transmembrane segment but lacks most of the cytoplasmic portion, 160-, 125-, and 95-kDa proteins were detected. The 160-and 125-kDa proteins appear to be derived from a single precursor protein, 3 and they comigrate as a 115-kDa protein after N-glycanase treatment. 2 The 95-kDa protein may represent an incompletely translated product or a degradation product of the recombinant noncytoplasmic CD45.
For binding analysis, GST fusion proteins of the full-length 3 A. Maiti and P. Johnson, unpublished results. (F) or the deleted forms (N or C) of CD45-AP bound to glutathione-Sepharose 4B beads were incubated with a cell lysate of untransfected control L cells (L40B7), L106A6, L12B5, or YAC-1 cells. Material bound to the recombinant CD45-AP beads was subjected to SDS-PAGE and was analyzed by Western blotting with antiserum against the extracellular portion of CD45 (Fig. 5B). The full-length as well as the noncytoplasmic form of CD45RABC bound to the full-length and N forms of CD45-AP. This indicates that CD45-AP interacts with the noncytoplasmic portion of CD45.
Binding of the Transmembrane Segment of CD45 to CD45-AP-As described above, the potential transmembrane segment of CD45-AP is required for binding to CD45 (Figs. 2 and  3). Therefore, if CD45-AP is indeed a transmembrane protein, it is most likely that the transmembrane segment of CD45 is directly involved in the interaction with CD45-AP. In order to examine whether the transmembrane portion of CD45 alone is capable of binding to CD45-AP, we utilized an L cell transfectant expressing recombinant chimeric CD44 (CD44.TM2) in which the transmembrane segment of CD45 replaces that of CD44 (34). Immunoprecipitates were obtained from untransfected control L cells (L40B7) and the CD44.TM2 transfectant with an anti-CD44 mAb, KM201 (36), and were analyzed by Western blotting with the same antibody (Fig. 6A). L cells express endogenous CD44 (37), and this was detected as a broad 85-kDa band in the control L cells. A similar band in the transfectant represents both endogenous CD44 and the transfected chimera, since they cannot be discriminated by size.
For binding analysis, GST-fusion proteins of the full-length (F) or the deleted forms (N or C) of CD45-AP bound to glutathione-Sepharose 4B beads were incubated with a cell lysate of the control L cells or the L cells transfected with CD44.TM2. Material bound to the recombinant CD45-AP beads was subjected to SDS-PAGE and was analyzed by Western blotting with the anti-CD44 mAb (36) (Fig. 6B). In the control L cells, endogenous CD44 did not bind to any form of CD45-AP. In the L cells transfected with CD44.TM2, on the other hand, CD44 bound to the full-length and N forms of CD45-AP but not to the C form. This indicates that the transmembrane portion of CD45 is sufficient for specific interaction with CD45-AP.
Subcellular Localization of CD45-AP-The CD45-AP cDNA predicts a stretch of hydrophobic amino acids near the NH 2 terminus that qualifies as a potential transmembrane segment (38). It was of great interest, therefore, to examine in which subcellular fraction CD45-AP is found and whether the CD45-AP localization pattern is similar to that of CD45. Subcellular fractions of YAC-1 cells derived from equal numbers of cells were subjected to SDS-PAGE and Western blotting with antiserum against CD45-AP (Fig. 7A), while another portion of the same fraction was immunoprecipitated by anti-CD45 mAb and the immunoprecipitates were analyzed by SDS-PAGE (Fig.  7B). CD45-AP was found in particulate fractions and not at all in the cytoplasmic fraction, indicating that it is likely to be a membrane protein. CD45-AP was found largely in the microsomal fraction and to a lesser extent in the nuclear fraction which would contain nondisrupted cells and cytoskeletal components besides nuclei. When the microsomal fraction was further separated into three subfractions, CD45-AP was found in all three fractions. The two high density fractions, M2 and M3, contained an equal amount of CD45-AP and the low density fraction, M1, contained somewhat less. The distribution of CD45 among all subcellular fractions was quite similar to that Membrane Orientation of CD45-AP-The results described above indicate that CD45-AP is likely to be a membrane protein and that it interacts with CD45 at the membrane. It is critical to determine the membrane orientation of CD45-AP in order to understand the role of CD45-AP in CD45-mediated signal transduction. YAC-1 cell lysates or intact cells were treated without or with TPCK-treated trypsin and a portion of each sample derived from equal numbers of cells was subjected to SDS-PAGE. The gel was then analyzed by Western blotting with antiserum against CD45-AP (Fig. 8A) or against a peptide segment present in the extracellular domain of CD45 (24) (Fig.  8B). TPCK-trypsin treatment of the lysates digested CD45-AP to a completely undetectable level and cleaved CD45 to a smaller fragment of about 120 kDa.
On the other hand, TPCK-trypsin treatment of intact cells did not alter the amount of CD45-AP detected at all, indicating that no arginine or lysine residue of CD45-AP is located extracellularly or is accessible to TPCK-treated trypsin. The 28-kDa band seen both in the presence and absence of TPCK-treated trypsin (Fig. 8A) was detected in control samples lacking cells as well 2 and represented a component present in the fetal calf serum used for washing cells at the end of the incubation. In contrast to CD45-AP, the amount of CD45 detected was significantly reduced after TPCK-trypsin treatment of intact cells due to degradation of the extracellular domain. The 120-kDa fragment is not seen with TPCK-trypsin treatment of intact cells indicating that it probably results from digestion of the cytoplasmic domain of CD45.
In order to confirm that the proteolytic conditions employed above for intact cells were appropriate for degrading other transmembrane proteins as well, a similar experiment was carried out using cells that were radiolabeled by incorporation of 14 C-labeled amino acid mixture. The labeled intact cells were treated without or with TPCK-treated trypsin as described above, and immunoprecipitates of LFA-1 (27) or the transferrin receptor (28) were analyzed by SDS-PAGE (Fig. 8C). The TPCK-trypsin treatment dramatically reduced the amounts of LFA-1 (the 180-kDa ␣ chain and the 95-kDa ␤ chain) and the mature form of the transferrin receptors (95 kDa) but not the nascent intracellular form (90 kDa) demonstrating that extracellular proteins were degraded under the proteolytic conditions employed. DISCUSSION Binding analysis employing several different deleted forms of CD45-AP demonstrated that the potential transmembrane segment of CD45-AP is required for binding to CD45 and that the rest of CD45-AP, including most of the consensus phosphorylation sites, is not directly involved in the interaction (Figs. 2  and 3). Interestingly, binding analysis employing recombinant cytoplasmic or noncytoplasmic forms of CD45, and various recombinant forms of CD45-AP showed that the noncytoplasmic portion of CD45 bound to CD45-AP (Figs. 4 and 5). Furthermore, binding analysis employing a recombinant chimeric transmembrane protein which consists of the extracellular and intracellular domains of CD44 and the transmembrane segment of CD45 demonstrated that the transmembrane segment of CD45 is sufficient for specific interaction with CD45-AP (Fig.  6). These data indicate that CD45 and CD45-AP interact at the plasma membrane through their respective transmembrane domains.
Consistent with this notion are the following observations: (i) CD45-AP can associate with various isotypes of CD45 (21), indicating that the interaction involves the isotypically invariant portion of CD45, (ii) the association between CD45-AP and CD45 is disrupted by some nonionic detergents (20), suggesting that the interaction between the two proteins involves hydrophobic segments, and (iii) CD45-AP is found in particulate fractions along with CD45 by subcellular localization studies (Fig. 7), indicating that CD45-AP is likely to be a membrane protein. The recombinant noncytoplasmic and transmembrane forms of CD45 used for binding studies were produced in L cell fibroblasts and the recombinant forms of CD45-AP were purified from E. coli. Therefore, these binding studies establish that the physical association between CD45 and CD45-AP does not require the presence of other leukocyte-specific proteins, such as p56 lck and p59 fyn(T) .
When the CD45-AP cDNA was in vitro transcribed and translated with the rabbit reticulocyte lysate system in the absence of microsomal fraction, the product migrated as 30 kDa in SDS-PAGE (Fig. 2). Moreover, the in vitro translated product comigrated with the mature form of CD45-AP obtained from YAC-1 cells (20). In the absence of microsomal fraction, NH 2 -terminal signal peptides of nascent proteins cannot be cleaved (39). Therefore, it seems unlikely that the predicted sequence of the CD45-AP cDNA contains an NH 2 -terminal signal peptide for protein translocation because, in that case, the mature form of CD45-AP would be smaller than the in vitro translated product. Instead, CD45-AP is likely to be a singlepass membrane protein with an internal signal peptide that remains as a membrane spanning segment. All three charged amino acid residues in the NH 2 -terminal side of the potential transmembrane segment of CD45-AP are acidic amino acids. In contrast, the 42-residue peptide segment immediately on the carboxyl side of the potential transmembrane segment contains nine charged residues which are all basic. This charge distribution of the amino acids in the vicinity of the potential signal/ transmembrane segment of CD45-AP would most likely cause translocation of the amino-terminal end into the endoplasmic reticulum lumen (40,41), as seen with the ␤-adrenergic receptor (42) and glycophorin C (43). As a result, only a short segment at the NH 2 terminus of CD45-AP would be located extracellularly, and the bulk of the protein would be intracellular. This is consistent with the results obtained by treating cell lysates and intact cells with TPCK-treated trypsin (Fig. 8A). CD45-AP was completely resistant to proteolysis by TPCKtreated trypsin applied to intact cells but was susceptible when the enzyme was applied to cell lysates. CD45-AP contains 10 arginine and one lysine residues, the TPCK-treated trypsin target sites, and all of them are located in the carboxyl-terminal side of the potential transmembrane segment (20). A proposed model of the physical orientation of CD45-AP is depicted in Fig. 9. Recently, a cDNA clone of a CD45-AP human homologue has been reported (44). The predicted amino acid sequence of the human homologue has a high percentage of sequence identity with murine CD45-AP. Interestingly, the human homologue differs from murine CD45-AP in two important aspects. (i) The human homologue appears to have an NH 2 -terminal signal peptide that is cleaved to produce the mature form, and (ii) the human homologue has a potential tyrosine phosphorylation site. It is of interest to know (i) whether the two highly conserved proteins have different mechanisms of protein translocation, and (ii) why is the tyrosine phosphorylation lacking in murine CD45-AP if the human homologue plays a role as a physiological substrate for CD45 as proposed (44). It is possible that a family of CD45-AP proteins of different structures exist. In fact, we have obtained and analyzed a human genomic clone of CD45-AP that also predicts a protein of very high homology to the murine CD45-AP, but differs in part from the reported sequence of the human homologue. 4 The homology between the human CD45-AP characterized in our laboratory and the murine CD45-AP is higher than that between the recently reported human homologue (44) and the murine CD45-AP.
The present study predicts that the major portion of CD45-AP including most of the potential phosphorylation sites is located intracellularly (Fig. 9). The cytoplasmic portion of CD45-AP may act as an adapter which directs the interaction between CD45 and other molecules involved in CD45-mediated signal transduction pathways. Proteins that interact with the intracellular portion of CD45-AP may be a substrate or a regulator of CD45 protein tyrosine phosphatase, and such proteins are currently under investigation. Distinct differences in the specific protein tyrosine phosphatase activity of CD45 have been detected among various populations of CD45 separated by sucrose gradient ultracentrifugation with a nondisruptive detergent (21). It is possible that CD45-AP affects the protein tyrosine phosphatase activity of CD45 directly or indirectly since a population of CD45 without associated CD45-AP appeared to have a higher specific activity. This possibility can now be more closely examined by using various forms of recombinant CD45-AP. YAC-1 cell lysates or intact YAC-1 cells were treated without or with TPCK-treated trypsin, and a portion of each sample derived from equal numbers of cells was subjected to SDS-PAGE in an 8 -15% acrylamide gradient under reducing conditions. They were then analyzed by Western blotting with antiserum against either CD45-AP (panel A) or the extracellular domain of CD45 (panel B). Panel C, YAC-1 cells were radiolabeled by incorporation of 14 C-labeled amino acid mixture and were treated without or with TPCK-treated trypsin. Immunoprecipitates were obtained from these cells with anti-LFA 1 or anti-transferrin receptor mAb and were analyzed by SDS-PAGE in an 8 -15% acrylamide gradient under reducing conditions. The positions of molecular mass markers (expressed in kilodaltons) are shown.