Analysis of Tyrosine Phosphorylation-dependent Interactions between Stimulatory Effector Proteins and the B Cell Co-receptor CD22*

The B cell-restricted transmembrane glycoprotein CD22 is rapidly phosphorylated on tyrosine in response to cross-linking of the B cell antigen receptor, thereby generating phosphotyrosine motifs in the cytoplasmic domain which recruit intracellular effector proteins that contain Src homology 2 domains. By virtue of its interaction with these effector proteins CD22 modulates signal transduction through the B cell antigen receptor. To define further the molecular mechanism by which CD22 mediates its co-receptor function, phosphopeptide mapping experiments were conducted to determine which of the six tyrosine residues in the cytoplasmic domain are involved in recruitment of the stimulatory effector proteins phospholipase Cγ (PLCγ), phosphoinositide 3-kinase (PI3K), Grb2, and Syk. The results obtained indicate that the protein tyrosine kinase Syk interacts with multiple CD22-derived phosphopeptides in both immunoprecipitation and reverse Far Western assays. In contrast, the Grb2·Sos complex was observed to bind exclusively to the fourth phosphotyrosine motif (Y828ENV) from CD22 and does so via a direct interaction based on Far Western and reverse Far Western blotting. Although both PLCγ and PI3K were observed to bind to multiple phosphopeptides in precipitation experiments, subsequent studies using reverse Far Western blot analysis demonstrated that only the carboxyl-terminal phosphopeptide of CD22 (Y863VTL) binds directly to either one. This finding suggests that PLCγ and PI3K may be recruited to CD22 either through a direct interaction with Tyr863 or indirectly through an association with one or more intermediate proteins.

CD22 is a B cell-restricted transmembrane glycoprotein that is also an I-type lectin, which specifically recognizes ␣2-6linked sialic acid residues (1). CD22 expression is tightly linked with that of membrane IgM and IgD on mature B cells (2,3), and studies have demonstrated that it physically associates with the B cell antigen receptor (BCR) 1 complex, albeit at a low stoichiometry (4,5). Moreover, CD22 is phosphorylated rapidly on tyrosine residues in response to BCR cross-linking, and this promotes the recruitment of several effector proteins that contain Src homology 2 (SH2) domains (1). By virtue of its ability to recruit intracellular effector proteins, CD22 functions as a co-receptor that is able to modulate B cell activation in response to BCR cross-linking (6 -9).
CD22 contains three immunoreceptor tyrosine-based inhibitory motifs in its cytoplasmic domain, similar to other inhibitory co-receptors (10). Studies have shown that tyrosine phosphorylation of CD22 promotes the recruitment of the protein tyrosine phosphatase SHP-1, which has been shown to regulate signal transduction negatively via growth factor and cytokine receptors (6,11,12). The SH2 domain-mediated binding of SHP-1 to CD22 results in potentiation of its catalytic activity (6). Thus, it was hypothesized that recruitment of SHP-1 to the CD22⅐BCR complex is responsible for attenuation of signal transduction. Independent ligation of CD22 using immobilized anti-CD22 mAb was observed to potentiate B cell proliferation in response to anti-Ig and interleukin-4 and actually decreased the threshold of stimulus required by more than 10-fold (6). This finding was interpreted as providing evidence that sequestration of CD22 away from the BCR leads to enhanced signal transduction. Furthermore, co-cross-linking experiments in which CD22 is co-localized in the membrane with the BCR complex leads to suppression of mitogen-activated protein kinase activation (13). Additional proof that CD22 negatively regulates signal transduction via the BCR has been provided by a series of studies examining the B cell compartment in CD22-deficient mice, revealing that the loss of CD22 expression causes B cells to become hyperresponsive to acute stimulation through the BCR (14 -17).
Although the studies described above indicate that CD22 functions as an inhibitory co-receptor, existing evidence indicates that CD22 may transduce stimulatory signals under certain circumstances. It has been shown that BCR cross-linking promotes the recruitment of multiple stimulatory effector proteins to CD22 including Syk, phospholipase C␥ (PLC␥), phosphoinositide 3-kinase (PI3K), and Lyn (18 -21). Additionally, it has been shown that human tonsilar B cells lacking CD22 are unresponsive to stimulation through the BCR based on their inability to mobilize Ca 2ϩ or to proliferate (22). These findings suggest that expression of CD22 and its association with stimulatory effector proteins could be involved in potentiating signal transduction through the BCR (18 -20). Experiments have also demonstrated that ligation of CD22 alone can indeed deliver stimulatory signals to the B cell independent of BCR cross-linking. Incubation of tonsilar B cells in the presence of an anti-CD22 mAb that blocks binding of CD22 to ligand induces proliferation and antibody production in the presence of interleukin-2 (21). Moreover, the anti-CD22-blocking mAb induced tyrosine phosphorylation of CD22 and recruitment of stimulatory effector proteins (21). These data suggest that engagement of CD22 by CD22 ligand(s) may induce a stimulatory signal independent of BCR cross-linking.
Delineation of the specific molecular mechanisms underlying the function of co-receptors like CD22 will yield important insight regarding the factors that regulate the balance between tolerance and immunity. Because CD22 exhibits the ability to recruit both inhibitory (i.e. SHP-1) and stimulatory effector (i.e. Syk, PLC␥, and PI3K) proteins, it is essential to define further the nature of these interactions to develop an understanding of the molecular mechanism(s) underlying the function of CD22. Toward this goal, phosphopeptides were synthesized based on the cytoplasmic domain of murine CD22 and were used to identify the binding sites for several stimulatory effector proteins. The results obtained indicate that the three distal tyrosine residues in the cytoplasmic domain of CD22 are involved in the recruitment of stimulatory effector proteins and that all of the effector proteins studied exhibit direct binding either to native phosphorylated CD22 or CD22-derived phosphopeptides. Finally, the current study has identified a novel interaction between CD22 and the Grb2⅐Sos complex.

EXPERIMENTAL PROCEDURES
Biological Reagents-Biotinylated phosphopeptides derived from CD22 (see Fig. 1) were purchased from Quality Controlled Biochemicals (Hopkinton, MA). Additionally, six control peptides were also purchased from Quality Controlled Biochemicals in which the tyrosine residues were changed to phenylalanine.
Antibodies used in these studies included mouse anti-bovine PLC␥1 mixed monoclonal antibody (IgG isotype, Upstate Biotechnology, Lake Placid, NY), rabbit anti-rat PI3K whole antiserum (Upstate Biotechnology), polyclonal rabbit anti-mouse Syk antibody (Upstate Biotechnology), polyclonal rabbit anti-Grb2 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), and mouse anti-Sos1 monoclonal antibody (IgG3 isotype, Transduction Laboratories, Lexington, KY). The anti-phosphotyrosine (Tyr(P)) mouse monoclonal antibody 4G10 conjugated to horseradish peroxidase (HRP) was purchased from Upstate Biotechnology. Additional antibody reagents used in Western blotting included goat anti-mouse Ig conjugated to HRP and goat anti-rabbit Ig conjugated to HRP (BIOSOURCE, Camarillo, CA). Streptavidin conjugated to HRP was also purchased from BIOSOURCE. Dr. K. Mark Coggeshall (Ohio State University, Columbus) provided the polyclonal rabbit anti-glutathione S-transferase (GST) antibody used in these studies.
GST Fusion Proteins-GST fusion proteins were produced as described below. The plasmids encoding GST-Grb2-SH2 and GST-Grb2-SH2/3 fusion proteins as well as the GST-Syk-SH2 fusion protein have been described previously (19). The plasmid encoding the GST-PLC␥ fusion protein containing the NH 2 -and COOH-terminal SH2 domains of PLC␥ was provided by Dr. K. Mark Coggeshall (Ohio State University). The plasmid encoding the GST-PI3K fusion protein containing the NH 2 -and COOH-terminal SH2 domains of the p85 subunit was obtained from Dr. Gordon Mills (M. D. Anderson Cancer Center, Houston, TX). Bacteria containing the GST fusion protein constructs were grown in LB medium containing 100 g/ml ampicillin and 20 mM glucose. Bacterial growth was monitored by A 600 nm , and once a value of 0.6 -0.8 was reached GST expression was induced by the addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration ranging from 0.2 to 3 mM. The bacterial culture was induced overnight at room temperature on a shaker (100 rpm). The next day the bulk culture was harvested by centrifugation for 15 min at 5,000 rpm (Sorvall GS-3 rotor) and the supernatant drained. The bacterial pellet was vortexed prior to resuspension with TBS (Tris-buffered saline) with complete protease inhibitors (Roche Molecular Biochemicals). Next, the resuspended bacteria were sonicated five times for 1 min with 2-min incubations on ice between each pulse. 10% of the resulting volume of the bacterial lysate was mixed with 10% Triton X-100 Surfact-Amps (Pierce) and the lysate rotated for 1 h at 4°C. Afterward, the lysate was centrifuged at 12,000 rpm for 15 min. The GST fusion proteins were recovered from the supernatant solution by the addition of 1-2 ml of a 50% (v/v) suspension of glutathione-Sepharose resin (Sigma) equilibrated in 1% Triton X-100 and TBS buffer and rotated overnight at 4°C. The next day the glutathione beads were washed, and the GST fusion protein yield was analyzed by 10% SDS-PAGE and Coomassie staining.
Immunoprecipitation and Immunoblotting-K46 cells were used at a concentration of 2 ϫ 10 7 cells/sample. After harvesting the cells, the cells were stimulated with goat anti-mouse IgG/M/A (FabЈ) 2 (2 g/ml) for various time points or stimulated with pervanadate if needed. Stimulation of the cells was terminated by diluting the cell suspensions with ice-cold PBS. The cells were washed three times and lysed in 0.5 ml of lysis buffer (150 mM NaCl, 10 mM NaF, 10 mM Tris, pH 7.3, 2 mM EDTA, 1 mM Na 3 VO 4 , 1% Nonidet P-40). The lysates were precleared by centrifugation and incubation with streptavidin-conjugated beads (Sigma). Subsequently, biotinylated phosphopeptides or control peptides were added at a final concentration of 10 M. Peptide-effector protein complexes were recovered using streptavidin-conjugated beads. The precipitates were then washed in 0.2% Nonidet P-40 lysis buffer and resuspended in SDS-PAGE reducing sample buffer. The samples were boiled, and the precipitated proteins were separated by 10% SDS-PAGE after which the proteins were transferred electrophoretically to nitrocellulose membranes. The membranes were blocked with 10% milk and probed with antibodies. The recovery of specific effector proteins was visualized using enhanced chemiluminescence (ECL). If needed, the nitrocellulose membranes were stripped by incubating the membranes in stripping buffer (10 mM Tris, 150 mM NaCl, pH 2.3) for 1 h at 60°C. Then the membranes were washed with TBST (150 mM NaCl, 10 mM Tris, pH 8.0, 0.005% Tween 20) and reprobed.
Far Western-The Far Western assay employed was adapted from previous protocols (23)(24)(25)(26). K46 cells (2 ϫ 10 7 cells/sample) were stimulated with F(abЈ) 2 fragments of polyclonal rabbit anti-mouse IgG/M/A (2 g/ml), pervanadate, or they were left untreated. Diluting cell suspensions with ice-cold PBS terminated stimulation of the cells. Afterward, cells were washed three times and lysed in 1% Nonidet P-40 lysis buffer. The lysates were precleared by centrifugation and the addition of Sepharose 4B beads conjugated to the irrelevant mAb RG7. After preclearing, the lysates were incubated with anti-CD22 mAb (CY34) coupled to Sepharose 4B beads overnight at 4°C. The beads were washed, and the immunoprecipitates were resolved by SDS-PAGE, after which the proteins were transferred to nitrocellulose membranes. Immobilized proteins were denatured for 30 min by incubating the membranes with 6 M urea in HEPES balanced buffer (HBB, 25 mM HEPES-KOH, pH 7.5, 25 mM NaCl, 5 mM MgCl 2 , 1 mM dithiothreitol). Subsequently, the proteins were renatured by progressive dilution of urea in HBB and incubated overnight at 4°C in 0.1 mM urea and HBB with shaking. The membranes were blocked with 5% milk in HBB and later incubated overnight in hybridization buffer (20 mM HEPES-KOH, pH 7.5, 75 mM MgCl 2, 0.1 mM EDTA, 1 mM dithiothreitol, 0.1% Nonidet P-40, 1% milk) containing GST fusion proteins at a concentration of 0.1-4 g/ml). Next, the membranes were washed three times for 5 min each in HBB and incubated for 1 h at room temperature with polyclonal rabbit anti-GST antibody. Again, the membrane was washed briefly and incubated with a tertiary antibody (goat anti-rabbit Ig HRP) and the blots developed by chemiluminescence.
Reverse Far Western-The reverse Far Western assay was adapted from previous work (27). GST fusion proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. After blocking with 10% milk and TBST the membranes were incubated with the appropriate biotinylated CD22 phosphopeptide or control peptide at a final concentration of 100 nM. After an overnight incubation at 4°C, the membranes were washed in TBST three times for 5 min and incubated with streptavidin-HRP. The interaction between GST fusion proteins and phosphopeptides was visualized using ECL.

RESULTS
Analysis of Tyr(P)-dependent Recruitment of Effector Proteins to CD22-The cytoplasmic domain of CD22 contains six tyrosine residues that are conserved between mouse and man (1). Based on previous studies, it is apparent that one or more tyrosine residues are phosphorylated in response to BCR crosslinking, thereby generating Tyr(P) motifs that are able to recruit SH2 domain-containing effector proteins. To explore the potential role of CD22 Tyr(P) motifs in the recruitment of effector proteins, we synthesized 12 peptides (10 amino acids each) based on the sequence of murine CD22 (Fig. 1). Six peptides were synthesized in which the tyrosine residue of each was phosphorylated (phosphopeptides 1-6). Six complementary peptides were generated in which the tyrosine residues were replaced with phenylalanine (Y 3 F peptides 1-6) to prevent phosphorylation when added to cell lysates. These peptides were then incubated with lysates from unstimulated K46 B lymphoma cells to determine whether they were able to interact with SH2 domain-containing effector proteins. Based on Western blot analysis several stimulatory effector proteins were observed to associate with CD22 phosphopeptides, including PLC␥, PI3K, Grb2, and Syk (Fig. 2). The ability of PLC␥, PI3K, and Syk to bind to phosphopeptides derived from CD22 is in agreement with previous work from other laboratories (6, 18 -20). However, the recruitment of Grb2 is a novel observation. The CD22 phosphopeptides have also been used to immunoprecipitate SHP-1, SHP-2, and SHIP from B cell lysates (27) 2 ; however, these interactions were not examined further in the current study.
With the exception of Grb2, which bound specifically to phosphopeptide motif 4 containing Tyr 828 , the other effector proteins were observed to interact with multiple phosphopeptides, although with varying affinities. Syk bound to phosphopeptide motifs 2, 3, 5, and 6, exhibiting the greatest affinity for the motifs containing Tyr 783 and Tyr 863 . PLC␥ and PI3K exhibited a binding preference for CD22 phosphopeptide motif 6 containing Tyr 863 , although both PI3K and PLC␥ also associated to a lesser extent with motif 4. Finally, PLC␥ was observed to interact weakly with motif 2. Subsequent experiments revealed that in all cases, the interaction of PLC␥, PI3K, Syk, and Grb2 with peptide was dependent on the presence of Tyr(P) ( Fig. 3; Syk data not shown). Although these studies indicate that multiple effector proteins can bind to CD22 in a Tyr(P)-dependent manner, it was not possible to determine whether the interactions resulted from the direct binding of effector proteins to CD22-derived phosphopeptides. Additionally, it was not possible to determine if the interactions between effector proteins and specific motifs derived from CD22 are physiologically relevant because studies have not been performed to identify the specific tyrosines that are phosphorylated in the cytoplasmic domain of CD22.
Grb2 Binds Directly to CD22 via Tyr(P) Motif 4 -The association of Grb2 with CD22 phosphopeptide motif 4 represents a novel interaction between CD22 and a stimulatory effector protein. Therefore, additional studies were performed to determine whether Grb2 co-precipitates with native CD22 from activated B cells. K46 cells were incubated in medium alone or in the presence of anti-Ig antibody or pervanadate to stimulate tyrosine phosphorylation of CD22 after which the cells were lysed, and CD22 was immunoprecipitated. Experiments demonstrated that Grb2 is inducibly recruited to CD22 in response to stimulation of B cells with F(abЈ) 2 fragments of polyclonal anti-Ig or pervanadate (Fig. 4). As expected, binding of Grb2 correlates with tyrosine phosphorylation of CD22 as demonstrated by stripping and reprobing the membrane with anti-Tyr(P) mAb. These results clearly demonstrate the recruitment of Grb2 to CD22 in response to stimulation of the B cell through the BCR complex.
Next, Far Western and reverse Far Western assays were performed to determine if Grb2 binds directly to CD22 or whether it is recruited via an intermediate linker protein. For the Far Western assay, K46 cells were incubated in medium alone or were stimulated either with F(abЈ) 2 fragments of polyclonal anti-Ig or pervanadate. CD22 was immunoprecipitated from cell lysates, resolved by SDS-PAGE, and then transferred to nitrocellulose. CD22 derived from untreated or activated B cell was probed with GST fusion proteins containing the SH2 domain of Grb2. Binding of fusion proteins to CD22 was de-2 J. Yohannan and L. B. Justement, unpublished observation. The motifs are designated 1-6 starting with the motif that is proximal to the transmembrane region of CD22. Two sets of 10 amino acid peptides were synthesized based on these motifs. In one set, each of the tyrosines was phosphorylated (phosphopeptides); in the other, each of the tyrosine residues was changed to phenylalanine (Y 3 F mutants).
FIG. 2. Analysis of effector protein binding to phosphopeptides representing the Tyr(P) motifs in the cytoplasmic domain of CD22. K46 cells (2 ϫ 10 7 /sample) were washed in PBS, pelleted, and lysed in buffer containing 1% Nonidet P-40 for 1 h on ice. Lysates were precleared by centrifugation, and biotinylated phosphopeptides were added at a final concentration of 10 M. Peptide-effector protein complexes were recovered by the addition of streptavidin-conjugated beads. The beads were washed in lysis buffer containing 0.2% Nonidet P-40 and were resuspended in SDS-PAGE reducing sample buffer. The samples were boiled, and precipitated proteins were separated by SDS-PAGE on 10% acrylamide gels. The resolved proteins were transferred to nitrocellulose membranes that were subsequently blocked in 10% milk before incubation with antibodies specific for PLC␥, PI3K, Syk, or Grb2. Precipitation of specific effector proteins was visualized by the addition of secondary antibody coupled to HRP, after which the blots were developed using ECL detection reagents.
tected by incubating the filters with anti-GST antibodies. The results revealed that the SH2 domain of Grb2 but not the fusion partner GST alone was able to interact with CD22 (Fig.   5). The interaction was only observed under conditions that led to tyrosine phosphorylation of CD22, as detected by anti-Tyr(P) immunoblotting of the membranes after they had been stripped. This observation is consistent with the hypothesis that Grb2 binds directly via its SH2 domain to phosphorylated Tyr 828 in the cytoplasmic tail of CD22.
To examine further this possibility reverse Far Western blotting was used to determine whether the CD22 phosphopeptide corresponding to motif 4 (Tyr 828 ) was able to interact with the recombinant SH2 or SH2/SH3 domain of Grb2. GST alone, GST-Grb2-SH2, or GST-Grb2-SH2/3 fusion proteins were resolved by SDS-PAGE and were transferred to nitrocellulose after which the membranes were probed with biotinylated phosphopeptides representing the six motifs from CD22. Peptide binding was detected using streptavidin coupled to HRP and chemiluminescence. Only phosphopeptide motif 4 bound to the Grb2 fusion protein (data not shown). The specificity of peptide binding in the reverse Far Western assay is demonstrated in Fig. 6. As depicted, phosphopeptide 4 bound to the Grb2 SH2 domain but not to GST alone. In contrast, the Tyr 3 Phe mutant peptide did not bind to Grb2 even though the GST-Grb2 fusion proteins were present at equal levels on the membrane (Fig. 6, lower panels). Together, these data show that the SH2 domain of Grb2 is capable of specific and exclusive binding to CD22 at Tyr 828 .
Grb2 is an adapter protein that stimulates Ras activation by engaging and promoting membrane translocation of Sos, a Ras-specific guanine nucleotide exchange enzyme, to the membrane (28 -30). To test the possibility that Sos is co-localized with Grb2 in CD22 phosphopeptide precipitates, lysates from unstimulated K46 cells were incubated with phosphorylated or Tyr 3 Phe peptides corresponding to the motif containing FIG. 3. Effector proteins bind specifically to tyrosine-phosphorylated but not Tyr 3 Phe mutant peptides from CD22. K46 cells (2 ϫ 10 7 /sample) were washed, pelleted, and lysed in buffer containing 1% Nonidet P-40 for 1 h on ice. Lysates were centrifuged and precleared after which either biotinylated phosphopeptides or Tyr 3 Phe mutant peptides were added at a final concentration of 10 M. Peptide-effector protein complexes were recovered using streptavidin-conjugated beads. The beads were washed in lysis buffer containing 0.2% Nonidet P-40, SDS-PAGE reducing sample buffer was added, and the proteins were resolved by SDS-PAGE on 10% gels. Proteins were transferred to nitrocellulose membranes, and the membranes were blocked in 10% milk. After probing the membranes with antibodies specific for PLC-␥, PI3K, or Grb2, the presence of specific effector proteins was visualized using secondary antibody coupled to HRP and ECL.
FIG. 4. Grb2 is recruited to native CD22 in an activation-dependent manner. K46 cells (2 ϫ 10 7 /sample) were stimulated with F(abЈ) 2 fragments of polyclonal goat anti-mouse Ig antibody (␣Ig, 3 g/ml) or with pervanadate (PV) for the length of time indicated. Stimulation of cells was terminated by the addition of ice-cold PBS (Ͼ10 volumes). The cells were washed two times in PBS, pelleted, and lysed in buffer containing 1% Nonidet P-40. Subsequently, the lysates were centrifuged at 13,000 ϫ g, precleared with Sepharose 4B beads coupled to RG7 mAb, and then incubated with anti-CD22 mAb (CY34) coupled to Sepharose 4B beads. The beads were washed and boiled with SDS-PAGE sample buffer after which the immunocomplex proteins were resolved by SDS-PAGE on 12% acrylamide gels. The proteins were transferred to nitrocellulose, and the membranes were probed with polyclonal rabbit anti-Grb2 antibody. The association between CD22 and Grb2 was visualized using secondary goat anti-rabbit Ig coupled to HRP and ECL (lower panels). Next, the membranes were stripped and reprobed with anti-Tyr(P) mAb coupled to HRP. Tyrosine phosphorylation of CD22 was visualized with ECL (upper panels). NT, no treatment.
FIG. 5. The SH2 domain of Grb2 mediates direct binding to CD22 in a phosphorylation-dependent manner. K46 cells (2 ϫ 10 7 /sample) were incubated in medium alone (NT) or were stimulated with F(abЈ) 2 fragments of polyclonal goat anti-mouse Ig (aIg, 3 g/ml) or pervanadate (PV) for 10 min at 37°C. Stimulation of cells was terminated by the addition of 10 volumes of ice-cold PBS. The cells were washed and then lysed in buffer containing 1% Nonidet P-40. Lysates were incubated with anti-CD22 mAb (CY34) conjugated to Sepharose 4B beads. The beads were washed in buffer containing 0.2% Nonidet P-40, and the immunocomplex proteins were resolved by SDS-PAGE on 12% acrylamide gels. Resolved proteins were transferred to nitrocellulose membranes, and duplicate membranes were probed with anti-Tyr(P) mAb coupled to HRP. Phosphorylation of CD22 was visualized using ECL (upper panels). Subsequently, a Far Western assay was performed. CD22 on the membranes was denatured and then renatured as described under "Experimental Procedures." The membranes were blocked by incubating them for 1 h in buffer containing 5% milk. They were then incubated overnight in hybridization buffer containing either GST alone (bottom left) or a GST-Grb2-SH2 fusion protein (bottom right). The fusion proteins were added at a final concentration of 4 g/ml. The membranes were washed and probed with polyclonal rabbit anti-GST antibody. Binding of GST fusion proteins to CD22 was detected by the addition of a secondary goat anti-rabbit Ig antibody coupled to HRP followed by development of the blots with ECL.
Tyr 828 , or as a negative control, Tyr 783 . Immunocomplex proteins precipitated by these CD22 peptides were resolved by SDS-PAGE and transferred to nitrocellulose. The membranes were probed with antibodies directed against Grb2 or Sos. The results depicted in Fig. 7 demonstrate, as shown previously, that Grb2 specifically associates with the phosphopeptide containing Tyr 828 but not Tyr 783 . The studies also demonstrated that binding of Grb2 to CD22 phosphopeptide 4 recruits Sos into the complex (Fig. 7). Finally, experiments were performed demonstrating the recruitment of Sos to native phosphorylated CD22 isolated from K46 cells that had been stimulated with anti-Ig antibody (data not shown). Thus, tyrosine phosphorylation of native CD22 promotes recruitment of Grb2, which binds to Tyr 828 via its SH2 domain, and this in turn leads to localization of the Ras-activating enzyme Sos.
PI3K Binds Directly to CD22-It has been demonstrated previously in human B cells that PI3K physically associates with CD22 (18,21). Whereas human CD22 contains an ideal motif (Y(PO 4 )XXM, 31) that would be predicted to bind to the SH2 domains of PI3K, murine CD22 does not (1). Therefore, we performed experiments to extend the peptide binding data (Fig.  1) and to confirm that PI3K is recruited to CD22 in a tyrosine phosphorylation-dependent manner in murine B cells. Fig. 8 depicts a representative experiment in which K46 B cells were incubated in medium alone or were stimulated with F(abЈ) 2 fragments of polyclonal anti-Ig or pervanadate. CD22 was immunoprecipitated, resolved by SDS-PAGE, and transferred to nitrocellulose. The membrane was probed with anti-Tyr(P) mAb, stripped, and then reprobed with mAb directed against the p85 subunit of PI3K. As can be seen, PI3K is indeed recruited to CD22 in response to tyrosine phosphorylation even though murine CD22 lacks the consensus YXXM motif (Fig. 8).
Because murine CD22 does not contain the YXXM motif, FIG. 6. Phosphopeptide motif 4 containing Tyr 828 from the CD22 cytoplasmic domain binds directly to GST-Grb2 fusion proteins. GST alone, GST-Grb2-SH2, and GST-Grb2-SH2/3 fusion proteins were resolved on 12% acrylamide gels using SDS-PAGE. The separated proteins were transferred to a nitrocellulose membrane that was cut into duplicate strips. These were then blocked in 10% milk and probed overnight at 4°C with either biotinylated CD22 phosphopeptide motif 4 or the corresponding Tyr 3 Phe mutant peptide at a final concentration of 100 nM. After the overnight incubation, the membranes were washed and probed with streptavidin coupled to HRP. Binding of CD22 peptides to GST fusion proteins was detected using ECL. Equivalent loading of fusion proteins was confirmed by stripping the blots and probing with rabbit anti-GST polyclonal antibody followed by a secondary goat anti-rabbit Ig antibody coupled to HRP. GST fusion proteins were visualized by ECL. FIG. 7. Binding of Grb2 to CD22 phosphopeptide motif 4 recruits Sos. K46 cells (2 ϫ 10 7 /sample) were lysed in buffer containing 1% Nonidet P-40 for 1 h on ice. The lysates were precleared and incubated with biotinylated phosphopeptides representing motifs 1 and 4 as well as the corresponding Tyr 3 Phe mutant peptides. Peptideeffector protein complexes were recovered by the addition of streptavidin-conjugated agarose beads. The beads were washed extensively, boiled in SDS-PAGE sample buffer, and the proteins resolved by SDS-PAGE on 10% acrylamide gels. Whole cell lysate and immunoprecipitation controls with anti-Sos and anti-Grb2 polyclonal antibodies were included. The resolved proteins were transferred to nitrocellulose, and the membrane was probed with rabbit anti-Grb2 polyclonal antibody followed by goat anti-rabbit Ig antibody coupled to HRP. Precipitation of Grb2 was visualized using ECL. Subsequently, the membrane was stripped and reprobed with mouse anti-Sos antibody followed by rabbit anti-mouse polyclonal antibody coupled to HRP. Sos was visualized using ECL.
FIG. 8. PI3K binds to native CD22 in an activation-dependent manner. K46 cells (2 ϫ 10 7 /sample) were stimulated with F(abЈ) 2 fragments of polyclonal goat anti-mouse Ig antibody (␣Ig, 3 g/ml) or with pervanadate (PV) for the length of time indicated. Stimulation of cells was terminated by the addition of ice-cold PBS (Ͼ10 volumes). The cells were washed two times in PBS, pelleted, and lysed in buffer containing 1% Nonidet P-40. Subsequently the lysates were centrifuged at 13,000 ϫ g, precleared with Sepharose 4B beads coupled to RG7 mAb, and then incubated with anti-CD22 mAb (CY34) coupled to Sepharose 4B beads. The beads were washed and boiled with SDS-PAGE sample buffer after which the immunocomplex proteins were resolved by SDS-PAGE on 12% acrylamide gels. The proteins were transferred to nitrocellulose, and the membranes were probed with polyclonal rabbit anti-PI3K antibody. The association between CD22 and PI3K was visualized using secondary goat anti-rabbit Ig coupled to HRP and ECL (lower panel). Next, the membranes were stripped and reprobed with anti-Tyr(P) mAb coupled to HRP. Tyrosine phosphorylation of CD22 was visualized with ECL (upper panel). NT, no treatment. subsequent experiments were performed to determine if PI3K binds directly to CD22. A Far Western assay was performed in which K46 cells were incubated in medium alone or were stimulated with anti-Ig antibody or pervanadate. CD22 was isolated from cell lysates, resolved by SDS-PAGE, and transferred to nitrocellulose. Duplicate membranes were probed with GST alone or GST fused to the dual SH2 domains of the p85 subunit of PI3K (GST-PI3K) as described under "Experimental Procedures." The results depicted in Fig. 9 demonstrate that GST-PI3K binds to tyrosine phosphorylated CD22, whereas the GST-PI3K fusion protein did not interact with nonphosphorylated CD22. These results provide evidence that PI3K can indeed bind directly to CD22 in a Tyr(P)-dependent manner.
CD22 Tyr(P) Motif 6 Mediates Binding of Multiple Stimulatory Effector Proteins-Because PI3K was observed to bind directly to native CD22 in a tyrosine phosphorylation-dependent manner, it was of interest to identify the specific Tyr(P) motif involved. Based on phosphopeptide immunoprecipitation experiments, motifs 4 and 6 were potential binding sites for PI3K. The reverse Far Western was used to identify which of these phosphopeptides are able to bind directly to a GST fusion protein with PI3K. The results of experiments with PI3K fusion proteins demonstrated that phosphopeptide 6, but not 4, is able to bind to GST-PI3K (Table I).
Previous studies have demonstrated Tyr(P)-dependent recruitment of PLC␥ to CD22 (18,20). It has also been shown that PLC␥ binds directly to CD22 using the Far Western assay (20). We extended these observations by demonstrating that PLC␥ binds to a specific phosphopeptide motif using the re-verse Far Western assay. Although PLC␥ can be precipitated from cell lysates using phosphopeptide motifs 2, 4, and 6, only phosphopeptide motif 6 containing Tyr 863 bound to a GST-PLC␥ fusion protein in the reverse Far Western assay (Table I). The selective ability of phosphopeptide 6 to bind directly to GST-PI3K or GST-PLC␥ suggests that one or more intermediate proteins might target these effector proteins to other phosphopeptides (i.e. motifs 2 and 4) via an indirect binding process. Thus, it appears as though PI3K and PLC␥ could be recruited to CD22 via both direct and indirect binding mechanisms.
Previous studies have demonstrated binding of the protein tyrosine kinase Syk to CD22 in a Tyr(P)-dependent manner (20). Moreover, these same studies have shown that the interaction between Syk and CD22 results from the direct interaction of these proteins with one another based on Far Western blotting. Therefore, experiments were performed to determine which phosphopeptides are able to bind directly to a GST-Syk dual SH2 domain fusion protein using the reverse Far Western assay. The results depicted demonstrate that phosphopeptides 2, 5, and 6 exhibit strong binding to GST-Syk, whereas weak binding of phosphopeptide motif 3 was observed in selected experiments (Table I). Thus, optimal recruitment of Syk may require that both SH2 domains be bound to Tyr(P) motifs in the cytoplasmic domain of CD22. This is in contrast to PI3K and PLC␥, which also possess dual catalytic domains but apparently interact directly with CD22 via a single tyrosine residue (Tyr 863 ).

DISCUSSION
The results from several studies indicate that CD22 negatively regulates signaling through the BCR via recruitment and activation of the protein tyrosine phosphatase SHP-1 (6,(13)(14)(15)(16)(17). Although there has been a great deal of interest in the finding that CD22 recruits SHP-1, it is evident that tyrosine phosphorylation of CD22 also mediates the recruitment of several stimulatory effector proteins. The current study, and work from other laboratories, clearly establishes that tyrosine phosphorylation of CD22 leads to the recruitment of PLC␥, PI3K, Syk, and Grb2 (18 -20). The functional significance underlying the ability of CD22 to recruit both inhibitory as well as stimulatory effector proteins is not well understood. However, it is possible that recruitment of stimulatory effector proteins may facilitate their dephosphorylation by SHP-1. This would presumably involve multiple CD22 molecules because motif 6 (Tyr 863 ), for example, appears to be involved in mediating binding of SHP-1 as well as Syk, PLC␥, and PI3K, effectively precluding recruitment of multiple SH2-containing proteins to a single CD22 molecule. Thus, it is possible that SHP-1 bound to one CD22 molecule could act on PLC␥, Syk, or PI3K bound to another CD22 molecule that is localized in the same region of the plasma membrane.
Alternatively, it is interesting to note that CD22 has the ability to recruit a wide range of stimulatory effector proteins that, in aggregate, would theoretically be sufficient to promote signal transduction via both Ca 2ϩ -and mitogen-activated protein kinase-dependent pathways. Recent studies have demonstrated that treatment of B cells with anti-CD22 mAbs promotes tyrosine phosphorylation of CD22 and the recruitment of effector proteins (21). Moreover, it has been shown that anti-CD22 mAbs, which recognize the ligand binding site on CD22, have the ability to induce B cell proliferation and antibody secretion (21). These findings indicate that under certain circumstances CD22 may function as a stimulatory receptor independent of the BCR.
The novel finding that Grb2 interacts with CD22 was characterized further in the present study by demonstrating that GST-Grb2 binds directly to the fourth phosphopeptide motif of FIG. 9. The SH2 domains of the p85 subunit of PI3K mediate direct binding to CD22 in a phosphorylation-dependent manner. K46 cells (2 ϫ 10 7 /sample) were incubated in medium alone (NT) or were stimulated with F(abЈ) 2 fragments of polyclonal goat anti-mouse Ig (␣Ig, 3 g/ml) or pervanadate (PV) for 10 min at 37°C. Stimulation of cells was terminated by the addition of 10 volumes of ice-cold PBS. The cells were washed and then lysed in buffer containing 1% Nonidet P-40. Lysates were incubated with anti-CD22 mAb (CY34) conjugated to Sepharose 4B beads. The beads were washed in buffer containing 0.2% Nonidet P-40, and the immunocomplex proteins were resolved by SDS-PAGE on 12% acrylamide gels. Resolved proteins were transferred to nitrocellulose membranes, and duplicate membranes were probed with anti-Tyr(P) mAb coupled to HRP. Phosphorylation of CD22 was visualized using ECL. Subsequently, a Far Western assay was performed as described under "Experimental Procedures." The membranes were incubated overnight in hybridization buffer containing GST-PI3K fusion protein. The fusion proteins were added at a final concentration of 4 g/ml. The membranes were washed and probed with polyclonal rabbit anti-GST antibody. Binding of the GST-PI3K fusion protein to CD22 was detected by the addition of a secondary goat anti-rabbit Ig antibody coupled to HRP followed by development of the blots with ECL.
CD22 (Tyr 828 ) based on the reverse Far Western assay. This finding is in agreement with previous predictions that the SH2 domain of Grb2 selectively binds to Tyr(P) motifs exhibiting the sequence YXN(V/P) (31). Additional experiments indicate that binding of Grb2 to Tyr 828 of CD22 mediates Sos recruitment to the complex, presumably through the association of Grb2 SH3 domains with proline-rich sequences in Sos. The recruitment of Sos to CD22 suggests that CD22 may be able to affect Ras activation positively or negatively. If CD22 is physically localized in the membrane with the BCR complex, then recruitment of Sos could potentiate BCR-dependent Ras activation. In contrast, if CD22 is physically excluded from the BCR activation complex, then it could actively compete with the BCR for recruitment of the Grb2⅐Sos complex, thereby attenuating Ras activation.
Experiments also demonstrated that the sixth phosphopeptide motif from CD22 (Tyr 863 ) is able to interact directly with a GST-PLC␥ fusion protein based on both Far Western (20) and reverse Far Western assays. The observation that the aminoand carboxyl-terminal SH2 domains of PLC␥ bind preferentially to peptides containing Y(V/I)X(P/V) and Y(V/L)X(V/L), respectively (31), supports the conclusion that the sixth tyrosine motif in the cytoplasmic domain of CD22 does indeed play a role in the recruitment of PLC␥. Although it is formally possible that motif 2 (Y 783 AIL) or 4 (Y 828 ENV) could play a role in recruitment of PLC␥ as well, because they were observed to precipitate PLC␥ from cell lysates, it is likely that these motifs would recruit PLC␥ through an indirect interaction involving one or more intermediate proteins. Whether PLC␥ is indeed recruited to CD22 primarily via Tyr 863 , as opposed to Tyr 783 or Tyr 828 , remains to be established definitively as it is not known which of the six tyrosine residues in the cytoplasmic domain of CD22 are phosphorylated in response to cross-linking of the BCR.
Results from the current study demonstrate that Tyr(P) motifs 4 and 6 are able to precipitate PI3K from B cell lysates. This finding is in agreement with a previous report demonstrating that phosphopeptides containing the human equivalent of either motif 4 or 6 were able to bind to PI3K when added to human B cell lysates (18). The results from this study further indicate that PI3K binds directly to tyrosine-phosphorylated CD22 based on the Far Western assay. Phosphopeptide libraries have been used previously to characterize the optimal motif to which the tandem SH2 domains of PI3K will bind. The results of these analyses reveal that both the amino-and carboxyl-terminal SH2 domains of PI3K exhibit specificity for the sequence YXXM (31). Whereas human CD22 does indeed contain such a tyrosine motif (YNPM), the analogous tyrosine motif in mouse CD22 is Y 773 NPAM (1). Thus there is no consensus motif in murine CD22 to which PI3K would be predicted to bind based on previous analyses using degenerate phosphopeptide libraries (31). It is interesting to note that precipitation experiments using a phosphopeptide based on the hu-man sequence for CD22 containing the YNPM motif fail to precipitate PI3K from cell lysates (18). Similarly, an association between phosphopeptide motif 1 from mouse CD22 (Y 773 NPAM) and PI3K was not observed in the current studies. Nevertheless, the sixth phosphopeptide motif from CD22 (Y 863 VTL) was observed to bind directly to a GST fusion protein of PI3K. Although previous studies have indicated that the SH2 domains of the p85 subunit of PI3K exhibit selectivity for the sequence YXXM, the amino-terminal SH2 domain exhibits additional specificity for a hydrophobic residue at the ϩ1 position in relationship to tyrosine (31). Motif 6 in the cytoplasmic domain of CD22 (Y 863 VTL) does have a hydrophobic residue in the ϩ1 position, which may be important for binding of PI3K. As is the case with PLC␥, it is not possible to establish definitively if PI3K is primarily recruited to CD22 via a direct interaction with motif 6 or indirectly through binding to motif 4 via an intermediate protein(s).
Like the stimulatory effector proteins discussed previously, it is clear that CD22 and Syk can interact directly with one another based on the results from Far Western blotting experiments (20). However, in contrast to Grb2, PLC␥, or PI3K, Syk interacts with multiple CD22 phosphopeptides. Indeed, both peptide precipitation experiments and reverse Far Western blotting indicate that Syk can bind to multiple Tyr(P) motifs in the cytoplasmic tail of CD22, including motifs 2, 5, and 6. Binding of Syk to the immunoreceptor tyrosine-based activation motifs of CD79a and CD79b involves both of its SH2 domains and requires that both tyrosine residues within the immunoreceptor tyrosine-based activation motif be phosphorylated (32,33). Thus, optimal recruitment of Syk to CD22 may occur only when both of its SH2 domains are bound to phosphorylated tyrosine residues in the cytoplasmic domain.
It is interesting to note that binding inhibition and phosphopeptide precipitation experiments indicate that the inhibitory protein tyrosine phosphatase SHP-1 can associate with motifs 2, 5 and 6 from CD22 (6). 2 Additional studies using reverse Far Western blotting suggest that phosphopeptides 5 and 6 can interact directly with a GST:SHP-1 fusion protein. 2 These results have been corroborated by studies in which each of the tyrosine residues in the cytoplasmic tail of CD22 has been mutated to phenylalanine. Mutation of Tyr 843 or Tyr 863 dramatically decreases the binding of SHP-1 to CD22, suggesting that motifs 5 and 6 are important for recruitment of this protein tyrosine phosphatase (34). 2 Findings from the current study suggest that PLC␥, PI3K, and perhaps Syk may bind directly to the same motif (i.e. motif 6) in the cytoplasmic tail of CD22. If this is indeed true, then it is likely that these effector proteins would compete for available binding sites on CD22. Furthermore, it is possible that binding of positive effector proteins to CD22 would decrease binding of SHP-1 and vice versa. It follows then that the extent to which a specific effector protein or class of effector proteins actually interacts with CD22 could have a significant effect on the functional role that TABLE I Analysis of CD22 phosphopeptide binding to stimulatory effector proteins using the reverse Far Western assay Blots were probed with CD22 phosphoproteins at a final concentration of 100 nM as described under "Experimental Procedures." Control peptides containing Tyr 3 Phe mutations were also used to probe duplicate membranes at a final concentration of 100 nM. Control peptides were not observed to bind to any of the GST fusion proteins tested (data not shown). Tyr 773  Tyr 783  Tyr 817  Tyr 828  Tyr 843  Tyr 863 GST alone

Phosphopeptide binding b
a Equivalent loading and transfer of GST fusion proteins were confirmed based on Western blotting with anti-GST antibody. b Ϫ ϭ no interaction; ϩ ϭ weak interaction; ϩϩ ϭ moderate interaction; ϩϩϩ ϭ strong interaction.
CD22 plays in regulating B cell activation. The association between CD22 and specific effector proteins could be determined by several factors including the relative level of effector protein expression in the cell, the subcellular localization of effector proteins relative to CD22, the relative affinity of effector protein SH2 domains for specific tyrosines in the cytoplasmic tail of CD22, and whether other proteins in the cell compete with CD22 for binding to a particular effector protein.
To determine the precise mechanism(s) responsible for recruitment of the individual effector proteins described herein, additional studies will be required using cells that express CD22 in which each of the tyrosine motifs in the cytoplasmic domain has been selectively altered to affect the binding of specific effector proteins. By selectively inhibiting the binding of individual effector proteins, it should be possible to determine whether the inhibitory function of CD22 is mediated solely by the interaction with SHP-1. Conversely, it should be possible to determine if recruitment of stimulatory effector proteins but not SHP-1 enables CD22 to function primarily as a stimulatory co-receptor.