Association of 75/80-kDa phosphoproteins and the tyrosine kinases Lyn, Fyn, and Lck with the B cell molecule CD20. Evidence against involvement of the cytoplasmic regions of CD20.

CD20, a non-glycosylated cell-surface protein expressed exclusively on B lymphocytes, is one of a family of 4-pass transmembrane molecules that also includes the beta chain of the high affinity receptor for IgE. The precise function of CD20 is unknown, although in vitro effects of CD20-specific antibodies on resting B cells indicate that it is able to transduce an extracellular signal affecting the G0/G1 cell cycle transition. Previous studies have demonstrated that CD20-initiated intracellular signals involve tyrosine kinase activation and that CD20 is tightly associated with both serine and tyrosine kinases. Here, analysis of CD20-associated molecules has revealed that CD20 is associated with the Src family tyrosine kinases p56/53lyn, p56lck, and p59fyn and with 75/80-kDa proteins phosphorylated in vivo on tyrosine residues. Mutagenesis of CD20 was performed to define regions of CD20 involved in intermolecular interactions. Mutants were analyzed in the human T lymphoblastoid cell line Molt-4, in which ectopically expressed wild-type CD20 associated with p59fyn, p56lck, and 75/80-kDa phosphoproteins. Deletion of major portions of the cytoplasmic regions of CD20 did not abolish its association with either p75/80 or tyrosine kinases. The interaction between CD20 and the Src-related kinases is therefore likely to be independent of CD20 cytoplasmic domains and may occur indirectly. The interaction may be mediated by the p75/80 phosphoproteins, which were found to be tightly associated with the Src family kinases isolated from the CD20 complex.

Growth and differentiation of B lymphocytes occurs primarily in the follicles and germinal centers of the spleen and lymph nodes and depends not only on contact with specific antigen but also upon co-stimulatory signals derived from accessory cell surface molecules. Accessory signals may be mediated by soluble factors such as IL-4 or by intercellular receptor-ligand interactions, such as the interaction between CD40 on B cells and its ligand on the surface of activated T cells (1). Like CD40, CD20 is a B cell surface protein with the capacity to initiate intracellular signals and modify cell growth and differentiation (2)(3)(4)(5)(6)(7)(8). CD20 is a non-glycosylated 33-kDa protein expressed on all mature B cells (9 -13). The predicted amino acid sequence of CD20 suggests a structure that has four transmembrane-spanning regions with both amino and carboxyl termini located on the cytoplasmic side of the plasma membrane (14 -16). CD20 is unrelated to a large family of proteins with similar overall structure (CD9, CD37, CD69, TAPA-1, and others) but does possess limited sequence homology, particularly within the transmembrane regions, to a recently identified molecule expressed in hemopoietic cells, HTm4, and to the ␤ chain of the high affinity receptor for IgE (Fc⑀RI) (17,18). The cytoplasmic region of Fc⑀RI␤ contains an immunoreceptor tyrosine-based activation motif (ITAM) 1 and is thought to be involved in relaying signals for the IgE binding ␣ chain (19). HTm4 and CD20 do not contain ITAMs and are not presently known to be part of receptor complexes.
Recent studies have shown that treatment of resting B cells with anti-CD20 mAb induces the accumulation of c-Myc mRNA via a protein-tyrosine kinase (PTK)-dependent signaling pathway (20). Furthermore, PTK activity was found to be associated with CD20, and antibodies against CD20 induced tyrosine phosphorylation of phospholipase C ␥ and other unidentified substrates (20). Tyrosine kinase activation is a critical early component of signal transduction pathways initiated by the engagement of several cell surface receptors, including the B cell antigen receptor (BCR), CD19, and CD40 (1). These receptors lack intrinsic kinase activity but are functionally, and in some cases physically, associated with non-receptor tyrosine kinases. Several tyrosine kinases expressed in B cells have been implicated in signal transduction events. Specifically, p72 syk and the Src family members p53/56 lyn , p59 fyn , p55 blk and p56 lck have been found in association with the BCR complex, where they are thought to be involved in antigen-stimulated signal transduction (21)(22)(23). Lyn and Lck have also been reported to associate with the CD19 complex (24), and Lyn is activated by anti-CD40 treatment of a B cell line (25). An unrelated tyrosine kinase, p77 btk , is required for normal B cell development (26), but its precise function is not yet known. Tyrosine kinases can associate with the cytoplasmic regions of cell-surface molecules by several mechanisms. In this study, we sought to identify the tyrosine kinases associated with CD20 and to localize the site of interaction. We found specific association of Lyn, Fyn, and Lck tyrosine kinases with CD20, whereas no association with Blk, Btk, or Syk was observed. CD20 also coprecipitated a ϳ75/80-kDa doublet that was phosphorylated in vivo on tyrosine residues. When transfected into the human T lymphoblastoid line Molt-4, CD20 associated with Lck and Fyn, the only Src family kinases detected in these cells, and also with ϳ75-80-kDa proteins, suggesting that p75/80 or a closely related molecule is also expressed in T cells. Deletion of 49 of the 51 amino acids, which constitute the amino-terminal cytoplasmic tail, and 45 of the ϳ85 amino acids of the carboxyl tail did not affect the ability of CD20 to associate with either p75/80 or PTK in Molt-4 cells. Deletion of an additional 37 residues at the carboxyl terminus or deletion of 8 of the 13 amino acids in the internal loop compromised the ability to detect CD20 at the cell surface. Despite the low level of surface CD20 expression caused by these deletions, CD20associated kinase activity was detected. These results demonstrate that the majority of the cytoplasmic residues in CD20 are not required for its association with Src family kinases and suggest that the association is mediated by an additional molecule. p75/80 was coprecipitated by antibodies against Fyn, Lyn and, to a lesser extent, Lck, from CD20 complexes following their elution in 1% SDS, suggesting that p75/80 is tightly associated with these tyrosine kinases and may mediate their interaction with CD20.

MATERIALS AND METHODS
Reagents and Cells-Monoclonal antibodies 2H7 (IgG2b, anti-CD20), G28.1 (IgG1, anti-CD37), and LB-2 (IgG2b, anti-CD54) have been described (9,11,27). These antibodies were purified from ascites fluid, dialyzed, and filtered prior to use. B1 mAb was purchased from Coulter Corp. (Hialeah, FL). 4G10 anti-phosphotyrosine mAb and antiserum against the p85 subunit of PI-3 kinase were purchased from Upstate Biotechnology Inc. Rabbit antisera to Src family tyrosine kinases have been described (22), as has antiserum to p72 syk (28). Antisera to p77 btk were generated in rabbits against peptides corresponding to amino acids 608 -625 and 176 -194 of human Btk. The peptides were conjugated to keyhole limpet hemocyanin (Calbiochem). Antisera from rabbits immunized separately with each peptide conjugate were pooled. Rabbit anti-mouse IgG was from Jackson Immunoresearch. The Raji and Molt-4 human lymphoblastoid cell lines were originally obtained from ATCC and were maintained in RPMI 1640, 10% fetal calf serum.
Immunoprecipitation and Western Immunoblotting-Precipitation of CD20 complexes, unless otherwise noted, used 10 7 cells. 5 g of 2H7 mAb was added for 30 min on ice, then pelleted and lysed in 1% Nonidet P-40 detergent buffer containing 0.25% deoxycholate, 50 mM Tris, pH 7.5, 12.5 g/ml leupeptin, 0.5% aprotinin, 1 mM phenylmethylsulfonyl fluoride, 5 mM EDTA, and 2.5 mM sodium vanadate. After 15 min on ice, the samples were centrifuged at 14000 ϫ g for 15 min to remove insoluble material. Lysates were transferred to clean tubes and mixed with 25 l of protein A-Sepharose (Repligen Corp., Cambridge, MA) for 1-2 h. Precipitation of protein-tyrosine kinases was performed as above except that 2 l of antiserum was added to cell lysates and mixed overnight at 4°C before adding protein A-Sepharose. Beads were washed three times in lysis buffer and then once in phosphate-buffered saline before addition of 30 l of 2 ϫ SDS sample buffer. Samples were boiled for 5 min, separated on 8.5% polyacrylamide gels, and transferred to Immobilon (Millipore). The filters were blocked in 5% bovine serum albumin, incubated with antiserum, and then followed by washing, detection with protein A-horseradish peroxidase (Bio-Rad), and development by enhanced chemiluminescence (ECL) (Amersham). Prestained molecular weight markers (Bio-Rad) were run on each gel. Bands were visualized by autoradiography using Kodak X-OMAT film (Eastman Kodak Co.).
In Vitro Phosphorylation-Immune complexes precipitated from cell lysates with protein A-Sepharose beads were washed three times in lysis buffer, twice in kinase buffer (20 mM Pipes, pH 7.2, 5 mM MnCl 2 , and 5 mM MgCl 2 ), and resuspended in 25 l of kinase buffer containing 5 Ci of [␥-32 P]ATP (specific activity, 3000 Ci/mmol, Amersham). Reactions were allowed to proceed at room temperature for 15 min and then were stopped by the addition of 3 l of 0.5 M EDTA. Samples were washed with lysis buffer, boiled for 5 min in SDS sample buffer, and run on 8.5% SDS-PAGE gels. For reprecipitation experiments, 125 l of 1% SDS in 10 mM Tris, pH 7.5, was added to samples following the in vitro kinase reaction. Samples were heated to 95°C for 10 min, diluted to 0.1% SDS with lysis buffer, and cleared with 20 l of protein A-Sepharose (Repligen Corp.) for 1 h. Antisera (2 l) or mAb (2 g) was mixed with the samples for 16 h on ice, and then 20 l of protein A-Sepharose was added for 1 h. Samples were then washed with lysis buffer, boiled for 5 min in SDS sample buffer, and run on 8.5% SDS-PAGE gels. Bands were visualized by autoradiography using Kodak X-OMAT film. Unless otherwise noted, exposures were at Ϫ70°C with intensifying screens.
Mutagenesis-Deletions of the putative cytoplasmic tails and the internal loop region of CD20 was accomplished by polymerase chain reaction using the primer pairs listed in Table I and the CD20.6 cDNA template obtained from Dr. Ivan Stamenkovic (Charleston, MA). CD20.6 cDNA was subcloned into pBluescript (Stratagene). To delete the amino-terminal, an internal unique EcoRI site at position 451 of the CD20 cDNA was used in combination with XhoI in the multiple cloning site of pBluescript to remove the 5Ј-end of CD20.6 cDNA. This region was then replaced with the XhoI-EcoRI-digested polymerase chain reaction product derived from primer pairs shown in Table I. Similarly, to delete the carboxyl tail, the EcoRI site was used in combination with NotI in pBluescript to remove the 3Ј-end of CD20.6 cDNA. This region was then replaced with the EcoRI-NotI-or EcoRI-AflII-digested polymerase chain reaction product derived from primer pairs shown (Table I). Deletion of 8 of the 13 amino acids constituting the putative internal loop was accomplished by replacing the XhoI-EcoRI fragment of CD20.6 cDNA with the digested polymerase chain reaction product from primer pairs shown. In this case the 3Ј-primer included sequence spanning the loop region but with bases encoding amino acids 107 through 114 deleted. The sequence of each construct was confirmed by dideoxy-DNA sequencing (Sequenase version 2.0, U. S. Biochemical Corp.). All constructs were subcloned into the BCMGSneo mammalian expression vector (29).
Transfections-4 ϫ 10 6 Molt-4 cells were mixed with 20 g of DNA and electroporated at 250 V and 500 microfarads across a 0.4-cm electrode chamber using the Bio-Rad Gene Pulser. Geneticin (Life Technologies, Inc.) was added at 0.8 mg/ml after 2 days, and individual clones were screened for CD20 expression after 3-4 weeks.
Immunofluorescence-Cells (ϳ10 5 ) were suspended in 100 l of phosphate-buffered saline, 2% horse serum, 0.01% sodium azide with 0.1 g of 2H7, 1F5, or isotype-matched control mAb left on ice for 20 min, washed once, and resuspended for 20 min on ice with 100 l of 1 ⁄100 dilution of goat anti-mouse IgG conjugated to fluorescein isothiocyanate (Southern Biotechnology Associates, Inc., Birmingham, AL). After washing, cells were resuspended in phosphate-buffered saline, 0.01% azide and analyzed on a FACScan cytometer (Becton Dickinson).

TABLE I
Polymerase chain reaction primers used in the preparation of CD20 deletion mutants All sequences are written in the 5Ј to 3Ј direction. Sequences in italics are restriction enzyme sites as described in the text.

Designation
Primer Sequence of primer   N⌬50  5Ј  CAGTCTCGAGAGCAAAATGAAGACTTTGGGGGCTGTC  3Ј  CAATGAATTCATTATCATTTTTCC  C⌬252  5Ј  ATAATGAATTCATTGAGCCTCTTT  3Ј  GTCAGTCTTAAGTTATGTTTCAGTTAGCCCAA  C⌬222  5Ј  ATAATGAATTCATTGAGCCTCTTT  3Ј  CAGTGCGGCCGCTTATCTGGAGCACGTTCTTTT  C⌬215  5Ј  ATAATGAATTCATTGAGCCTCTTT  3Ј  CAGTCTTAAGTTATTCATTCTCAACGATGCC  IL⌬106-115  5Ј  AGTCTCGAGAGCAAAATGACAACA  3Ј CAATGAATTCATTATCATTTTTCCTTTCTCCGTTGCTGC RESULTS CD20 Is Constitutively Associated with Tyrosine Kinases-CD20 was precipitated from lysates of the Raji human B cell line using mAb 2H7 and then subjected to an in vitro phosphorylation reaction in the presence of 32 P-labeled ␥-ATP and enolase, an exogenous substrate. Phosphorylated species of 50 -60 kDa and 75-80 kDa were detected (Fig. 1A), shown previously by phosphoamino acid analysis to include phosphotyrosine (20). In addition, enolase was phosphorylated but not in control samples lacking CD20 mAb (Fig. 1A, lane 1), establishing that the kinase activity observed in CD20 precipitates was due to specifically precipitated kinase(s). Since anti-CD20 mAb 2H7 does not precipitate CD20 when added directly to cell lysates, the antibody was bound to whole cells prior to lysis. To investigate the possibility that the association of kinases with CD20 was induced by binding of 2H7 antibody to CD20 on intact cells, two types of experiments were performed. First, the effect of binding 2H7 mAb to the cells at physiologic temperatures versus ice-cold temperatures was compared: No difference in CD20-associated kinase activity was observed when antibody was bound at 37°C versus 0°C (Fig. 1B). Second, the B1 mAb, which is capable of binding to CD20 in detergent lysates as well as on intact cells (12,13), was used to determine if kinase activity could be precipitated with CD20 after cell lysis. B1 does not precipitate as much kinase activity from cells as does 2H7 for reasons that are not clear at present; however, B1 coprecipitated kinase activity from both cell lysates and whole cells (Fig. 1C), suggesting that kinase activity was associated with CD20 prior to the addition of antibody. Additional proteins were observed in the kinase assay following CD20 precipitation from cell lysates when compared to precipitation from the cell surface (Fig. 1C).
CD20 Is Associated with 75-80-and 50 -60-kDa Proteins Tyrosine Phosphorylated in Vivo-To determine if CD20 is associated with proteins that are tyrosine phosphorylated in vivo, CD20 complexes were isolated from Raji cells, and the constituent proteins were separated by SDS-PAGE, transferred to membranes, and probed with anti-phosphotyrosine antibody. At least two bands in the range of 50 -60 kDa were distinct from the heavy chain of the 2H7 mAb (Fig. 2, compare  lanes 3 and 4). Proteins that migrated as a broad smear of 75-80 kDa ( Fig. 2A, lane 3) could be seen as two distinct bands in some experiments (Fig. 2B, lane 4). When two bands were visible, the lower one was always more heavily tyrosine phosphorylated as determined by reactivity with anti-phosphotyrosine. For comparison, Fig. 2B also shows immunoprecipitation of the p85 subunit of PI-3 kinase. Immunoblotting of the filter shown in Fig. 2B with an antiserum against the p85 subunit of PI-3 kinase demonstrated that the lowest of the major bands in lane 2 represented p85 (data not shown). No reactivity of anti-p85 with CD20-associated proteins was observed.
p72 syk and p77 btk Are Not Detected in CD20 Immunoprecipitates-Since most known tyrosine kinases can autophosphorylate, the CD20-associated PTK would most likely be represented by one or more of the bands observed following kinase assays performed on CD20 immune complexes. Candidate PTKs fall into 3 groups: p72 syk /ZAP70; p77 btk/itk , and the Src family (50 -60 kDa) (31). Of these, ZAP70 and Itk were considered to be unlikely candidates on the basis of their low to undetectable expression in B lymphocytes (31). To test for the presence of Syk in CD20 complexes, filters containing mAb 2H7-or anti-Syk-precipitated proteins were probed with anti-Syk antiserum (Fig. 3A), then stripped and reprobed with anti-phosphotyrosine (Fig. 3B). The results showed that although Syk is present in the B cells examined, CD20 immune precipitates contained no detectable Syk, and CD20-associated p75/80 migrated slightly slower than Syk (Fig. 3B). The second candidate kinase in the 75-80-kDa size range is Btk. Btk was expressed FIG. 1. Constitutive association of kinase activity with CD20. In vitro kinase assays were performed on CD20 complexes immunoprecipitated from Raji cells in the presence of 5 g of acid-denatured enolase (A) (arrow). Cells were incubated at 37°C for 10 min with medium alone (lane 1) or with CD20 mAb 2H7 (lane 2) prior to lysis. B, cells were incubated either at 37°C for 10 min (lanes 1-3) or at 0°C for 30 min (lanes 4 -6), with medium alone (lanes 1 and 4), with CD20 mAb 2H7 (lanes 3 and 6), or with isotype matched control mAb LB2 (CT, lanes 2 and 5). C, 2H7 or B1 mAb was added to whole cells at 0°C for 30 min prior to lysis (lanes 1 and 2), or B1 was added to cell lysates (lane 3) for 2 h. in Raji cells, as shown in Fig. 3, C and D, but anti-Btk did not detect p75/80 in immunoblots of 2H7 precipitates (Fig. 3D). Furthermore, Btk was not recovered from in vitro phosphorylated CD20 immune complexes (Fig. 3E). Thus, p75/80 did not appear to be either Syk or Btk.
CD20 Is Associated with Src Family Tyrosine Kinases-Several members of the Src family of tyrosine kinases are expressed in B cells and in B cell lines (32). To determine which of these kinases were expressed in the Raji cell line, cell lysates were screened for the presence of Blk, Fyn, Hck, Lck, Lyn, and Yes by precipitation of each PTK using its specific antibody, followed by an in vitro kinase assay. The results showed that Raji cells express Blk, Lck, Lyn, and Fyn kinases (Fig. 4A). To determine which, if any, of these PTK were associated with CD20, 2H7 immunoprecipitates were first phosphorylated in an in vitro kinase reaction and then released from the Sepharose beads by boiling in 1% SDS and subjected to reprecipitation with antibodies against Blk, Lck, Lyn, and Fyn. The results demonstrated the presence of Lyn, Lck, and Fyn, but not blk, in the CD20 complex (Fig. 4B). Interestingly, p75/80 was also reprecipitated by Lyn and Fyn antisera and to a lesser extent by anti-Lck, despite prior disruption of the CD20 complex in 1% SDS. The presence of Lyn accounted for most of the PTK activity in the CD20 complex. The proportion of total cellular Lyn that coprecipitated with CD20 was estimated to be 3-5%, as determined by comparing the cpm associated with Lyn following reprecipitation from the CD20 complex, with that following direct precipitation of Lyn. This estimate is likely to be low due to partial loss of CD20-associated components during isolation of the complex and because the efficiency of reprecipitation of Lyn may be less than 100%. In control experiments, only 18 -24% of Lyn precipitated by anti-Lyn was reprecipitated by the same antibody. Thus, the proportion of total cellular Lyn that is associated with CD20 could be 12-20% or more.
CD20 Is Also Associated with p75/80 and Src Family PTK When Expressed by Transfection in the Human T Cell Line   FIG. 4. Association of p59 fyn , p56 lck , and p56/53 lyn with CD20. A, Raji cell lysates were mixed with antisera against Src family kinases as indicated, and in vitro kinase assays were performed on recovered immune complexes. Exposure, 1 h at room temperature. B, CD20 was precipitated from Raji cells with mAb 2H7 and released from the Sepharose beads in 1% SDS. Samples were cleared with protein A-Sepharose, and antibodies were added as indicated. Exposure, 24 h. Lane 7 shows a 1-h exposure of lane 5. NRS, nonimmune rabbit serum. Molt-4-To try and determine the site(s) of interaction between CD20 and associated kinases, we first needed to identify a cell line in which CD20, expressed ectopically by transfection, associated with Src-related tyrosine kinases. Attempts to express human CD20 in the murine B cell lines WEHI 231 and BAL 17 resulted in very low levels of expression of the protein at the cell surface, as determined by the ability of anti-human CD20 antibodies to bind to the transfected cells. However, in the human T lymphoblastoid cell line Molt-4, transfected CD20 was expressed at a level that was similar to that found in the B lymphoblastoid cell line, Raji. CD20 was precipitated from the cell surface of independently derived transfected Molt-4 clones and subjected to in vitro kinase assays; [␥-32 P]ATP was incorporated into 50 -60-and 75-80-kDa proteins (Fig. 5A). Phosphoamino acid analyses demonstrated that all of the radioactive phosphate was coupled to tyrosine (data not shown). Molt-4 cells do not express p56/53 lyn but do express p59 fyn and p56 lck (data not shown), as is typical for T cells (31,32). Both p59 fyn and p56 lck were recovered from the in vitro phosphorylated CD20 complexes (Fig. 5B), with Lck accounting for most of the PTK activity. A minor proportion of p75/80 coprecipitated with p59 fyn and p56 lck as previously observed in Raji cells. In some experiments, for example in Fig. 5B, lane 2, additional CD20-associated proteins were observed between 25 and 40 kDa and were reprecipitated by anti-phosphotyrosine.
Cytoplasmic Domain Deletions Do Not Prevent CD20 Association with PTK-Src family kinases associate with the cytoplasmic regions of other transmembrane proteins by various means, including interactions involving Cys residues in the case of CD4 and CD8, with an acidic region in IL-2R␤, and via ITAMs and proline-rich regions recognized by SH2 and SH3 domains, respectively (33,34). Hence, it was of interest to determine which cytoplasmic region of CD20 may be responsible for the observed association with Src family kinases. Each of the three cytoplasmic domains of CD20 was deleted in turn, and the deleted forms of CD20 were expressed by transfection in Molt-4 T cells. The amino-terminal tail was deleted, leaving three amino acids to anchor the first TM domain (N⌬50 , Table  II). N⌬50 was expressed well on the surface of Molt-4 cells (Fig.  6). The internal loop region consists of 13 amino acids; 8 central residues were deleted, leaving 5 residues to anchor the 2nd and 3rd TM domains (IL⌬106 -115, Table II). This deletion impaired the expression of CD20 but was detectable at approxi-mately 100-fold lower level than transfected wild-type CD20 (Fig. 6). The precise length of the carboxyl cytoplasmic tail is unclear. Kyte-Doolittle hydropathy plots predict that the first cytoplasmic amino acid is Gln 204 or Glu 205 (16); however, Glu 205 is followed by a stretch of seven hydrophobic residues. The beginning of the cytoplasmic carboxyl-terminal region is therefore not clearly defined and may be 85 or 93 residues in length. The carboxyl tail was deleted at three sites. First, a deletion was made immediately upstream of acidic and prolinerich regions (residues 259 -276 and 280 -283, see Table II) considered to be potential sites for kinase association. Second, C⌬222 deleted 75 residues, leaving either 10 or ϳ18 amino acids in the cytoplasm, depending on the precise location of TM4. C⌬222, like the internal loop deletion IL⌬106 -115, was poorly expressed at ϳ100-fold lower level than wild-type CD20 (Fig. 6). A third carboxyl tail deletion was made terminating at Glu 215 . However, the expression of C⌬215 was undetectable at the cell surface (Fig. 6). As deletions at both sites (222 and 215) resulted in poor or no expression of CD20, the sequence between Gln 204 and Thr 252 may be important for either transport of CD20 to the cell surface or maintenance of the overall structure. The term "expression" as used here refers to expression of the epitope detected by mAb 2H7. In the case of the C⌬215 deletion, it is possible that only the 2H7 epitope was lost; however, a second mAb 1F5, thought to recognize a different, albeit overlapping, epitope (35), was also used to detect expression, and the results were always equivalent to that observed with 2H7.
In vitro kinase assays were performed on CD20 immune complexes precipitated from Molt-4 transfectants expressing cytoplasmic domain deletion mutants of CD20. The results, shown in Fig. 7, demonstrate that tyrosine kinase activity was still associated with CD20 in all cases. Both Lck and Fyn were detected by reprecipitation from the in vitro phosphorylated CD20 complexes isolated from transfectants expressing the N⌬50 and C⌬252 deletions (data not shown). The internal loop and C⌬222 deletions were expressed at too low a level to allow reprecipitation experiments to be performed. Thus, it is possible that one or both of these deletions could have affected the association of one, but not both, of the PTKs. p75/80 was clearly detected in association with each of the CD20 cytoplasmic domain deletion mutants. DISCUSSION CD20-associated tyrosine kinases are shown in this report to include p53/56 lyn , p59 fyn , and p56 lck , all members of the Src family. p56 blk and the non-Src family kinases p77 btk and p72 syk were not detected. The BCR, like CD20, is associated with multiple tyrosine kinases, including Lyn, Fyn, Blk, Lck, and Syk (21)(22)(23). It is not clear whether each associated Src family kinase has distinct or overlapping functions, but association of PTK with the BCR is required for signal transduction upon antigen recognition. The high affinity receptor for IgE, expressed on mast cells and basophils, is associated with p56 lyn in the rat basophilic leukemia cell line RBL-2H3 (37). This association was specific since RBL-2H3 cells also express p60 src , but p60 src did not associate with Fc⑀RI. In the murine mast cell line, PT-18, Fc⑀RI is associated with another Src family kinase, p62 yes (37). Thus, Fc⑀RI, which transmits intracellular signals upon IgE binding, associates with some but not all of the Src family tyrosine kinases, as does CD20 and the BCR.
In some B cell lines, i.e. SKW6, T51, and a hairy cell leukemia line, HCLL 7876, CD20 fails to coprecipitate kinase activity (20), despite the fact that they all express Lyn. SKW6 and T51 also express Fyn, but none of these cell lines express Lck (data not shown). Interestingly, Lck was expressed in all five human B cell lines in which CD20 was observed to associate with PTK activity. This correlation raises the possibility that the expression of Lck is required for the association of Lyn and Fyn to CD20. Alternatively, the PTK may associate with CD20 through an intermediate protein, which is absent in these three cell lines. The addition of recombinant Lck to CD20 immunoprecipitates isolated from T51, SKW6, and HCLL did not reveal the presence of p75/80, although an exogenous substrate enolase was phosphorylated (data not shown). This suggests that CD20 precipitates isolated from these lines may lack p75/80 as well as p56 lck , consistent with a model in which PTKs associate with CD20 indirectly by binding to a co-associated molecule, p75/80. Src family tyrosine kinases can bind to other proteins via Src-homology 2 (SH2) domains, which recognize phosphorylated tyrosine residues within specific amino acid sequences. Src-related PTKs also associate with other proteins by mechanisms that are independent of tyrosine phosphorylation, such as via SH3 domains or their unique amino-terminal sequences (33). Since the cytoplasmic domains of CD20 contain no tyrosine residues, the PTK must either bind to CD20 indirectly or via tyrosine-independent means. The association between Fc⑀RI and Lyn has been shown to occur through the carboxyl tail of the Fc⑀RI ␤ chain, a CD20-related molecule (38). The carboxyl tail of Fc⑀RI ␤ includes an ITAM motif ((YXXL)X 6 -7 (YXXL)), although it is unclear whether this motif is involved in the binding of Lyn. The cytoplasmic regions of CD20 do not include this motif; however the carboxyl tail is highly charged with 16 acidic residues among the 40 COOH-terminal amino acids and also includes a potential SH3-binding site (PEPP). Deletion of these acidic and proline-rich regions, however, did not affect the ability of CD20 to bind to either PTK or p75/80 in CD20-transfected Molt-4 T cells. Further deletions of the carboxyl tail resulted in low or no surface expression of CD20.
Nevertheless, associated kinase activity was detected, albeit weakly, after deletion of 75 COOH-terminal residues. This deletion left potentially 10 -18 amino acids protruding into the cytoplasm, depending on the location of the fourth TM domain. Thus, although CD20 associations with PTK and p75/80 were independent of major portions of all three cytoplasmic regions of CD20, it is possible that the region of the carboxyl tail immediately proximal to the plasma membrane may be involved in intermolecular associations. Since deletion of this region results in the loss of CD20 expression, it is not possible to determine if it is also involved in association to PTKs.
The identity of p75/80 is not known at the present time. Neither is it known whether the proteins represented by the two bands migrating at ϳ75/80 kDa are different proteins or variants of the same protein. Recently, it was reported that an TABLE II Amino acid sequence of the cytoplasmic domains of CD20 deletion mutants Designations include the numbers of amino acid residues retained in the constructs. All sequences are written with the amino group to the left. Slashes indicate the predicted position of the inner leaflet of the plasma membrane. The possible alternative location for the transition point between the fourth transmembrane domain and the COOH-terminal cytoplasmic region is indicated with a second slash. Capital letters indicate amino acids retained in the constructs. Lower case letters indicate deleted amino acids in the internal loop deletion and in C⌬252. The underlined sequences in the deleted portion of the COOH-terminal tail indicate acidic and proline-rich regions.  85-kDa protein was phosphorylated in in vitro kinase assays performed on immunoprecipitated glycosyl phosphatidylinositol-linked proteins, Thy-1 and CD48, molecules that have been shown to associate with Src family kinases Lck and Fyn in T cells (39). Interestingly, this 85-kDa protein was also reprecipitated by anti-Fyn and anti-Lck antibodies, as was the case for CD20-associated p75/80. The tyrosine kinase Lyn has been found to associate with the p85 subunit of PI-3 kinase and with HS1, a ϳ75-kDa protein of unknown function (40,41). Antisera against p85 and HS1, however, did not react with CD20-associated p75/80 in immunoblots or in reprecipitation experiments (data not shown).
CD20-associated kinase activity was detectable using the B1 antibody to precipitate CD20 from cell lysates. This indicates that at least some tyrosine kinase activity is associated with CD20 in intact cells prior to the addition of antibody. Regulation of the tyrosine kinase activity may occur by extracellular interactions, by relocalization of the kinases, or by modifying access to substrates. A possible alternative means of regulating CD20 function may be related to the fact that CD20 is also associated in B cells with serine/threonine kinase (PSK) activity, as well as with tyrosine kinases (20). In Molt-4 cells, transfected CD20 did not coprecipitate detectable PSK activity for reasons that are not clear at present. The CD20-associated PSK has not been identified, but there is evidence that CD20 may be a substrate for PKC (42,43). Interestingly, Fc⑀RI␤ has been recently shown to be associated with the ␦ isoenzyme of protein kinase C (44). CD20 is phosphorylated in vivo on serine/ threonine residues, and this phosphorylation is increased by activation signals initiated by phorbol 12-myristate 13-acetate, anti-CD40, or anti-BCR (12,13,30,42,43,45). Phosphorylation of CD20 may affect the activity or localization of the associated tyrosine kinases. Alternatively, CD20-associated PSK may act on the associated tyrosine kinases directly; serine phosphorylation of Lck in an activated T cell line has been reported to down-regulate its activity (46).
In conclusion, CD20 is shown here to be a component of a multimolecular complex that includes the tyrosine kinases Lyn, Fyn, and Lck, and a 75/80-kDa doublet of proteins phosphorylated on tyrosine residues in vivo. Mutational analysis of the cytoplasmic domains of CD20 suggested that the PTKs do not directly associate with the cytoplasmic regions of CD20 and therefore more likely associate indirectly via an additional protein, perhaps p75/80. As p75/80, or a closely related molecule, appears to be expressed in both B and T cells, it may function as an adaptor protein linking Src-like kinases to members of this 4-pass TM family in lymphocytes. In this regard, as HTm4 mRNA is expressed in Molt-4 T cells (18), it will be important to determine if HTm4 protein is also associated with p75/80 and Src family kinases. Both CD20 and HTm4, like Fc⑀RI␤, may be signal-transducing components of complexes that include receptor subunits. Whether this receptor subunit is p75/80 or whether p75/80s are adaptor molecules functioning to link CD20 to downstream activation signals remains to be determined. Characterization of the p75/80 components of the complex is in progress to further elucidate the role of CD20 in B cell growth and differentiation.