Vav and Rac Activation in B Cell Antigen Receptor Endocytosis Involves Vav Recruitment to the Adapter Protein LAB

The signal transduction events supporting B cell antigen receptor (BCR) endocytosis are not well understood. We have identified a pathway supporting BCR internalization that begins with tyrosine phosphorylation of the adapter protein LAB. Phosphorylated LAB recruits a complex of Grb2-dynamin and the guanine nucleotide exchange factor Vav. Vav is required for activation of the small GTPases Rac1 and Rac2. All these proteins contribute to (and dynamin, Vav, and Rac1/2 are required for) BCR endocytosis and presentation of antigen to T cells. This is the first description of a sequential signal transduction pathway from BCR to internalization and antigen presentation.

All receptors for antigens stimulate a complex series of intracellular biochemical events to stimulate biological changes in the state of the host cell. Antigen receptors have common structural features that include a low molecular weight polypeptide bearing an immunoreceptor tyrosine-based activation motif. The tyrosines of the immunoreceptor tyrosinebased activation motif are phosphorylated to recruit a variety of Src homology 2 (SH2) 2 -containing lipid, tyrosine, and serine/ threonine kinases, many adapter proteins, and several enzymes. Together, the enzymes generate second messengers that include increased cytoplasmic Ca 2ϩ , inositol trisphosphate, diacylglycerol, and phosphoinositide lipids to cause cell activation (reviewed in Refs. [1][2][3]. BCR signal transduction causes B cells to increase expression of MHC class II, CD80, and CD86 (4). However, the expression of these proteins by themselves is insufficient to promote B-T interaction and the production of high affinity, class-switched antibodies and long lived memory. Rather, T-B interaction is driven by antigenic peptides derived from BCR-endocytosed antigens (5). Many studies have identified signal transduction pathways that support transcription factor activation leading to up-regulation of surface proteins that support B-T interaction. Very few studies have shown a connection between those sig-naling events and BCR internalization following antigen binding.
Other receptors in the immune system that are similar to BCR include the T cell antigen receptor and a variety of immunoglobulin receptors, including IgG receptors Fc␥RI, -II, and -III and IgE receptors Fc⑀RI and -II. The IgG receptors generally do not endocytose their targets upon engagement; instead, these receptors phagocytose large particles having IgG bound to them. The high affinity FceRI is endocytosed in a lipid raft using the GTPase dynamin (6). The low affinity Fc⑀RII uses clathrin and dynamin for endocytosis (7). The T cell antigen receptor is endocytosed after antigen binding along with a protein complex that contains the tyrosine kinase ZAP-70 and two tyrosine-phosphorylated adapter proteins LAT (linker of activated T cells) and SLP-76 (8). T cell antigen receptor internalization is associated with immunoreceptor tyrosine-based activation motif phosphorylation and recruitment of ZAP-70, phosphorylation of ZAP-70 at Tyr-292, and subsequent recruitment of the adapter protein Cbl to phosphorylated ZAP-70 (9). T cell antigen receptor internalization also involves WASp (Wiskott Aldrich syndrome protein), which links actin reorganization and the small GTPase Cdc42 (10). WASp recruits Intersectin 2, a Dbl homology-containing protein that potentially activates Cdc42 to stimulate WASp-catalyzed actin reorganization.
We recently reported that receptor-triggered BCR endocytosis required Vav1 and/or Vav3 isoforms such that BCR endocytosis was completely blocked in B cells of Vav1,3 Ϫ/Ϫ mice. Other BCR-triggered functions in Vav1,3-deficient B cells like up-regulation of MHC class II, CD80, and CD86 and signal transduction events like the influx of Ca 2ϩ , the activation of MAPK modules, and of Akt were equivalent to those of wildtype B cells. Thus, Vav1 and/or -3 are uniquely and absolutely required for BCR endocytosis, whereas Vav2 can substitute for all other BCR-triggered functions, including B cell development and maturation.
Vav contains a Dbl homology domain, a pleckstrin homology (PH) domain, and a C-terminal protein interaction domain that includes an Src homology 2 (SH2) domain flanked by two SH3 domains (11). Vav is activated by tyrosine phosphorylation at Tyr-174 (12), which allows the catalytic Dbl homology domain to act on small GTPases (13). Vav phosphorylation is achieved by Vav recruitment to the site of tyrosine kinases associated with the receptors. Vav is recruited to the scaffolding protein LAT in T cells (14) and to phosphatidylinositol 3-kinase (PI3K)generated lipids in many cytokine receptors (15)(16)(17)(18). Both CD19 tyrosine phosphorylation (19) and PI3K-generated 3-phosphoinositide lipids (20) are induced upon BCR stimulation. It is not clear which of these mechanisms are operative in supporting Vav activation during BCR endocytosis.
The Vav family of proteins (Vav1, -2, and -3) acts as guanine nucleotide exchange factors for many small GTPases, including Rac1/2, RhoA/G, and possibly Cdc42 (11). Vav1 and -3 isoforms prefer Rac1/2 and RhoA/G but do not recognize Cdc42 (21). Vav2 is capable of causing GTP loading of Cdc42 and Rac1/2 but has less activity toward RhoA/G (22,23). It is not clear which of these small GTPases is the molecular target of Vav in BCR endocytosis.
Here, we have tested the role of these small GTPases using RNA interference to selectively down-regulate their expression and show that Rac1 and Rac2 are required for BCR endocytosis, whereas other Vav targets are not. We also explored the mechanism by which Vav is activated in B cells. We found that PI3K is not required for BCR endocytosis, Vav phosphorylation, or Rac-GTP loading, whereas the protein interaction domain of Vav is involved in these processes. Vav bound to the B cell adapter protein LAB (linker of activated B cells) in a complex that also contains the GTPase dynamin and Grb2. LAB-deficient B cells show an incomplete defect in BCR endocytosis, Vav tyrosine phosphorylation, and Rac activation. We conclude that Vav is activated by recruitment to LAB and another unidentified adapter protein(s), which serves to stimulate Rac-GTP loading to support BCR endocytosis.

MATERIALS AND METHODS
Mice-Animals were obtained from The Jackson Laboratories (Bar Harbor, ME), Duke University Medical Center, or bred at our facility. Wild-type C57BL/6 mice were used as controls for LAB Ϫ/Ϫ . All experiments have been approved by the IACUC.
Cell Purification and Stimulations, Lysates-Splenic B cells were prepared from the spleens of 5-8-week-old BALB/c mice as described earlier (26). In brief, erythrocytes were removed by osmotic lysis, and T cells were removed by incubation with anti-Thy-1.2 followed by guinea pig complement. The resulting population was 85-95% pure B cells, based on staining with antibodies to CD19 or B220. For immunoprecipitations, 10 ϫ 10 6 cells were washed and resuspended in 100 l of phosphatebuffered saline and stimulated at 37°C with 15-20 g/ml F(abЈ) 2 rabbit anti-mouse IgG or IgM. In the indicated experiments, LY294002 was added to temperature-equilibrated cells for 10 min at 37°C prior to cell stimulation. Cells were then lysed in TN-1 buffer (50 mM Tris-HCl, pH 8.0, 125 mM NaCl, 10 mM EDTA, 1% Triton, 10 mM NaF, 3 mM Na 3 VO 4 , 10 mM Na 4 P 2 O 7 , 10 g/ml aprotinin, 10 g/ml leupeptin) and centrifuged at 16,000 ϫ g for 10 min at 4°C. The cleared lysate was incubated overnight on a rocker at 4°C with 2-3 g of immunoprecipitating antibody/ml. Protein A/G-Sepharose was added for 1 h, and immune complexes were washed five times in lysis buffer and eluted with SDS sample buffer. The proteins were separated by SDS-PAGE. The proteins were electrophoretically transferred to nitrocellulose membranes, probed with appropriate antibodies, and visualized by an enhanced chemiluminescence. The bands were quantitated by measuring individual chemiluminescence by using a Lumi-Imager F1 workstation (Roche Applied Science).
BCR Internalization Assay-B cells were incubated with 15 g/ml unlabeled anti-BCR (primary cells) or 5 g/ml PC-keyhole limpet hemocyanin (M12g4Rd) for 30 min at 4°C and washed to remove unbound antibody or antigen. The cells were then incubated at 37°C for an additional 30 min (for anti-BCR) or 4 h (for antigen). The receptors remaining on the surface were detected by a fluorescently labeled anti-BCR (M12g4Rd) or anti-rabbit Ig (primary cells). Mean fluorescence intensity (MFI) of the samples was determined using a FACS Scan. As controls, samples were either stained with secondary antibody alone or the samples were maintained at 4°C to prevent BCR internalization. Percent internalized BCR was obtained by the MFI of the sample and the control in the following calculation: percent internalization ϭ ((MFI no internalization, 4°C sample control) Ϫ (MFI sample) Ϭ (MFI no internalization, 4°C sample control)) ϫ 100.
Transfections-For transfections, plasmids containing ⌬6AVav, dominant-negative Rac1 and Rac2 genes (15), and the other Vav constructs were cloned into the expression vector pLEGFP-C1, which co-expresses green fluorescent protein (GFP) along with the gene of interest. The GST-Grb2 fusion protein has been described elsewhere (27). M12g4Rd cells were transfected by electroporation as follows: 1 ϫ 10 7 cells were electroporated with the indicated plasmid DNA in a 4-mm gap cuvette containing warm RPMI 1640 medium, using a Bio-Rad GenePulser at 340 V and 960 microfarads. The cells were resuspended in 8 -10 ml of warm growth medium and cultured for 36 -72 h before analysis. Following incubation, cells were sorted for equal expression of GFP using MoFlo. For siRNA targeting of LAB, Grb2, or small GTPases, the M12g4Rd cells were electroporated as above with commercial (Santa Cruz Biotechnology) siRNA or scrambled siRNA and kept at 37°C as above.
Rac Assay-The protocol is based on a Rac-GTP pulldown method using GST-p21-activated kinase as described (28). Briefly, cells were stimulated and lysed as above, and the lysates were incubated with GST-PAK pre-bound to glutathione-agarose beads. After washing, the beads were incubated with Laemmli SDS-PAGE sample buffer, and the eluted proteins were subjected to SDS-PAGE. After the transfer of proteins to a nitrocellulose membrane, Western blotting was performed using a pan monoclonal antibody to Rac isoforms. The amount of GTP-loaded Rac was quantitated by measuring the mass of GTP-Rac in comparison with the mass of total Rac. The value was then normalized to the GTP-Rac/total Rac ratio of unstimulated cells.
Antigen Presentation Assay-M12g4Rd B cells were cultured in RPMI 1640 medium in with PC-Ova (5 g/ml) to generate class II MHC loaded with Ova peptide on their surface. The cells were washed and co-cultured at 100,000 cells per well with the T cell hybridomas (DO.11.10; (25)) at 100,000/well, and the co-culture was maintained at 37°C for 24 h. Following incubation, the plates were frozen, thawed, and centrifuged at 2,000 ϫ g, and supernatants were harvested. The supernatant was then assayed for IL-2. In the indicated experiments, B cells were cocultured with DO.11.10 in the presence of 5 g/ml of an Ova peptide ISQAVHAAHAEINEAGR.
Statistics-All experiments were done in triplicate and repeated multiple times, as indicated in the figure legends. The data are reported as the average and S.E. of the triplicate samples from within each experiment.

RESULTS
BCR Is Endocytosed upon Treatment with Antigen or Anti-BCR Antibody-We used two models for BCR endocytosis: primary splenic B cells or a B cell line (M12g4Rd (24)) that has been transfected with cDNA encoding an antigen-specific (PC) BCR. Each model has advantages and disadvantages. Splenic B cells can be derived from various knock-out mice to test the contribution of a known signaling protein to BCR endocytosis. Likewise, splenic B cells form a distinct BCR cap in 2 min following anti-BCR treatment and hence are conducive to fluorescence co-capping studies. However, primary B cells cannot be transfected with dominant-negative constructs or RNA interference reagents and cannot be used for antigen presentation studies because of their polyclonal antigen specificities. The M12g4Rd cell line is easily transfected and can be used for antigen presentation studies with PC-labeled proteins (24), but it does not form a BCR cap.
For endocytosis measurements, B cells are stimulated with unlabeled antigen or goat anti-BCR antibodies at 4 or 37°C. At various times, the B cells are stained with PE-labeled anti-mouse BCR antibodies (for antigen treatments) or PE-donkey anti-goat (for anti-BCR treatments) and analyzed by flow cytometry. A typical result is shown in Fig. 1A. BCR expression is high in cells held at 4°C (Fig. 1A, solid gray) and lower in cells incubated at 37°C with anti-BCR antibodies (dotted line). The solid line in Fig. 1A marks the PE-anti-goat isotype control, showing only background staining. The percent internalization is determined as the ratio of the MFI at 37 and 4°C, as described in detail previously (29). Using this measurement, we find that both models readily endocytose their BCR upon antigen stimulation ( Fig. 1B) or anti-BCR stimulation of the M12 cell line or primary B cells (Fig. 1C).
To ensure the receptor is contained within the B cell, we stimulated primary splenic B cells with PE-labeled anti-BCR for 30 min at 4°C (Fig. 1D) or at 37°C (Fig. 1E). The B cells were fixed and stained with antibodies to the surface protein, CD45, and visualized by confocal fluorescence microscopy. The images show the red-stained BCR within the cells that are also stained with anti-CD45 to mark the plasma membrane. Our earlier report shows that the receptor is not shed from the sur-FIGURE 1. BCR endocytosis assay. B cells were incubated with goat antibodies to BCR, held at 4°C or incubated at 37°C for various times, and then stained with PE-labeled anti-goat Ig or isotype control. A shows a typical result of BCR expression level of cells kept at 4°C (solid gray) or incubated for 30 min at 37°C (dotted line); the staining level of the isotype control is shown with a solid line. % BCR internalization is determined as the ratio of MFI of the 37°C sample to the 4°C sample, minus the MFI or the isotype, as described (29). B shows the kinetics of BCR internalization of the M12g4Rd cell line using antigen ( face, is not endocytosed at 4°C, requires actin reorganization, and requires expression of both dynamin and either or both of the Vav1 or -3 isoforms (29). Thus, our assay accurately measures BCR endocytosis with either model.
Rac1 and -2 Are the Molecular Targets of Vav in BCR Endocytosis-To identify which small GTPase acts distal to Vav, we used siRNAs that individually target the small GTPases known to be Vav targets. The siRNAs were transfected into M12g4Rd, and the transfectants were compared with untransfected control cells in the BCR endocytosis assay described in Fig. 1. The controls are shown in Fig. 2A, showing the isotype staining control (gray peak), the sample kept at 4°C showing maximal BCR fluorescence (blue peak), and the sample incubated for 30 min at 37°C (red peak). The decrease in surface BCR fluorescence indicates BCR endocytosis, and we use the ratio of their MFI as described above.
Upon transfection of siRNA targeting small GTPases, we found (see Fig. 2B) that siRNA targeting either Rac1 or Rac2 completely blocked BCR endocytosis such that B cells expressing these RNA interference reagents have as much surface BCR as did the B cells held at 4°C. In contrast, siRNA targeting Rac3, Cdc42, RhoA, or RhoG or scrambled siRNA to Rac1 or -2 (data not shown) had no effect, and endocytosis was as complete as the untransfected control cells shown in Fig. 2B (red line). We confirmed ( Fig. 2, C and D) that protein expression of the siRNA target was indeed down-regulated by Western blots of whole cell lysates of the transfectants using antibodies to the individual small GTPases. Rac3 showed low-to-absent expression levels in B cells, so it was not quantitated. The data show indeed that transfection of B cells with all the siRNAs (ϩ), but not the scrambled (Scr) control siRNA, efficiently knocked down expression of the targeted small GTPase to less than 20% of control (scrambled siRNA) levels.
As an additional test for Rac contributions to BCR endocytosis, we transfected M12g4Rd with N17 mutants of Rac1 or Rac2 having high affinity for GDP and hence unable to bind GTP (30). The transfection was done as a dose response, titrating the amount of dominantnegative Rac mutant and measuring the level of BCR endocytosis. We found (Fig. 2E) that both dominantnegative mutants of Rac were able to block BCR endocytosis in this model, with the Rac2 mutant consistently acting slightly more efficient than Rac1.
Rac1, Rac2, and Rac3 are over 90% identical and differ primarily in the region near the C-terminal CAAX (where AA is aliphatic amino acid) prenylation site (31). Because the siRNA reagents are proprietary commercial ones and do not list sequences targeted, we tested whether the siRNAs down-regulate the protein expression of their specific target alone or cross-react with related small GTPases. We therefore transfected M12g4Rd with scrambled siRNA (control) or siRNA targeting Rac1, Rac2, Rac3 or Cdc42 as an additional control, and we measured the expression level of Rac1, Rac2, Rac3, and the less related small GTPase, Cdc42 (32), and used actin as a loading control. We found (Fig. 2F) that the siRNAs were indeed specific and only knocked down protein expression of their advertised targets. The siRNA targeting Rac3 showed a slight decrease in Rac2 expression (an off-target effect), but this decrease in Rac2 level was not sufficient to alter BCR endocytosis (Fig. 2B, purple line). Together, these findings show that, of all the known small GTPases targeted by Vav isoforms, only Rac is required for BCR endocytosis and surprisingly that both Rac1 and Rac2 isoforms are separately required.
PI3K Is Not Required for BCR Internalization and Does Not Affect Vav or Rac Activation-In colony-stimulating factor-1stimulated macrophages (15) and a variety of other cytokine receptors, PI3K is necessary for Vav membrane recruitment and tyrosine phosphorylation. To test if PI3K acts upstream of Vav tyrosine phosphorylation and Rac activation in mouse B cells, we used a chemical inhibitor of PI3K (LY294002). The level of Vav tyrosine phosphorylation is a marker of Vav activation (13) because phosphorylation relieves auto-inhibition allowing Vav to activate its small GTPase target. We consistently found that treatment of B cells with LY294002 increased rather than inhibited Vav tyrosine phosphorylation (Fig. 3A) and had no effect on Rac activation (Fig. 3B). Likewise, LY294002 did not alter the rate or extent of BCR internalization induced by anti-BCR (Fig. 3C), which requires Vav activity (29). Fig. 3C, inset, shows the level of phosphorylated Akt in stimulated or unstimulated B cells, treated or not with LY294002, to establish that PI3K activity was indeed blocked by the drug. The data indicate that PI3K is not needed for BCR-triggered Vav phosphorylation, Rac activation, or BCR endocytosis.
As an additional test for the contribution of PI3K to Vav activation, we overexpressed various Vav mutants shown in Fig.  4A in the M12g4Rd B cell line. The ⌬6AVav construct has a mutation within the catalytic domain and was used previously (15) to disrupt Vav/Rac-dependent functions. We earlier showed that expression of the ⌬6AVav mutant in M12g4Rd B cells completely blocked BCR endocytosis (29). For the experimental conditions, we overexpressed the isolated PH domain of Vav (which blocks PI3K-triggered Vav recruitment by consum-ing available 3-phosphoinositides (33)) and the isolated protein interaction domains present in the C terminus of Vav (CT-Vav). Expression of either the PH and C-terminal Vav domains interferes with Vav function by virtue of overexpression, above that of endogenous Vav. Accordingly, the overexpressed proteins adsorb mediators (phosphoinositides for the PH domain) or protein-binding sites (phosphotyrosine or proline-rich sequences for the SH2/3 domains).
To establish overexpression of the PH domain, we measured the BCR-induced phosphorylation of Akt, a PI3K-dependent event in B cells (34), and we found that Vav PH domain overexpression completely blocked this event (Fig. 4B). Thus, the PH domain was overexpressed sufficiently to block PI3K function. We have established overexpression of ⌬6AVav in our previous report (29). It is difficult to establish overexpression (expression greater than endogenous Vav expression) with the CT-Vav mutant because there is no C terminus-specific Vav antibody that recognizes these constructs as well as endogenous Vav. We attempted probes using phosphopeptides con-  taining a pYESP motif, a sequence found in both ZAP-70 (35) and SLP-76 (36) that directly bind the SH2 domain of Vav in far Western blots. Unfortunately, the peptide failed to bind endogenous Vav or any of the proteins derived from the transfected cDNAs. Nevertheless, and consistent with the result using LY294002 (Fig. 3), we found that overexpression of the Vav PH domain (to block PI3K function) did not alter BCR endocytosis (Fig. 4C). In contrast, overexpression of the C-terminal protein interaction domain of Vav or the ⌬6AVav mutant completely blocked BCR endocytosis. Likewise, overexpression of the isolated C terminus but not the PH domain of Vav blocked BCRtriggered Vav phosphorylation (Fig. 4D) and formation of Rac-GTP (Fig. 4E). Thus, although we cannot formally establish overexpression of the C-terminal construct, the fact that we can disrupt Vav function by transfection of the Vav protein interaction domain suggests that Vav protein interactions and not recruitment to 3-phosphoinositides are key in Vav activation for BCR endocytosis.

Linker of Activation in B Cells (LAB/NTAL) Is Required for Optimal Vav Phosphorylation and BCR Internalization-Pre-
vious studies indicated that LAB, a scaffolding protein with multiple tyrosine phosphorylation sites, is important in BCR internalization (37). We confirmed the defect in BCR internalization of B cells from LAB Ϫ/Ϫ mice (Fig. 5A) using our endocytosis measurement and applying anti-BCR antibodies. We found that B cells from LAB Ϫ/Ϫ animals showed an approximate 50% decrease in the extent of BCR internalization compared with wild-type B cells.
The data in Fig. 4 uncovered an important role for the protein interaction C-terminal domain of Vav in regulating its phos-phorylation and the formation of Rac-GTP. If LAB is involved in recruitment of Vav as LAT is involved in T cells (14,38), the LAB-deficient B cells would show a defect in Vav phosphorylation. We tested this possibility by probing Vav immunoprecipitates from resting or stimulated B cells of wild-type or LAB Ϫ/Ϫ mice with antibodies to phosphorylated Vav (pVav). We found (Fig. 5B) that Vav from LAB-deficient B cells showed an approximate 50% decrease in phosphotyrosine level relative to that of wild-type B cells. The decreased Vav tyrosine phosphorylation was associated with a corresponding decrease in Rac-GTP loading, as shown in Fig. 5C. These findings suggest that LAB at least in part regulates Vav activation, possibly by recruitment of the C-terminal protein interaction domain to permit Vav tyrosine phosphorylation. Other adapter protein(s) in B cells likely can substitute for LAB in LAB-deficient B cells. The data are remarkably similar to findings in LAT-deficient Jurkat T cells that likewise show suboptimal Vav tyrosine phosphorylation (38).
LAB Forms a Complex with Dynamin and Vav-Our earlier work established an important role for the GTPase dynamin, in addition to Vav (29). Both dynamin and Vav co-capped with BCR upon BCR stimulation. Likewise, in data not shown, we observed that LAB and BCR co-cap upon BCR stimulation, as was shown previously (37). Given these previous findings, it may be that both Vav and dynamin physically interact with the LAB scaffolding protein. To test this possibility, we immunoprecipitated dynamin or Vav from resting or BCR-stimulated B cells and probed the immunoprecipitates with antibodies to LAB. We found (Fig. 6A) that LAB associated with both proteins but only in BCR-stimulated samples, indicating that LAB can bind both dynamin and Vav after B cell activation and suggesting the involvement of LAB tyrosine phosphorylation. The finding raises the possibility that LAB can form individual complexes; some that contain only Vav and others that contain only dynamin. Alternatively, LAB might form a single complex that contains both Vav and dynamin. Because LAB contains 10 phosphorylatable tyrosines (39,40), it is likely that LAB recruits other interacting partners.
We applied an immunodepletion approach to distinguish these possibilities. Here, the strategy is to remove all Vav from the lysate and ask if there are remaining complexes of LABdynamin. Similarly, we removed all dynamin from lysates and asked if there is remaining LAB-Vav complexes. We exhaustively immunodepleted Vav or dynamin by three sequential immunoprecipitations from B cell lysates prepared from resting (Ϫ) or BCR-stimulated (ϩ) splenic B cells. As a control, we depleted samples with normal rabbit immunoglobulin. The material obtained from the first and third immunoprecipitation was tested by Western blot (Fig. 6B for Vav immunodepletion and Fig. 6D for dynamin immunodepletion) and showed the presence of the target protein in the first but not the third immunoprecipitate. Thus, we successfully depleted the target protein from the lysate. The supernatant was then subjected to immunoprecipitation with antibodies to dynamin or Vav. We found antibodies to dynamin failed to co-precipitate LAB (Fig.  6C), indicating the LAB-dynamin complexes were removed in the immunoprecipitation with anti-Vav. Likewise, we found no LAB-Vav complexes in the samples immunodepleted with anti- FIGURE 5. LAB is required for optimal BCR internalization. A, primary B cells were prepared from wild-type or LAB Ϫ/Ϫ mice splenocytes with anti-Thy-1.2 and guinea pig complement. BCR internalization was measured for the indicated times. The data are representative of five separate experiments and are the average and S.E. for one representative experiment. B, splenic B cells of wild-type or LAB Ϫ/Ϫ mice were left unstimulated (Ϫ) or stimulated with anti-BCR for 2 min (ϩ). Lysates were subjected to immunoprecipitation with antibodies to Vav and probed with antibodies to phosphotyrosine (labeled pVav) or to Vav (total Vav). The ratio of pVav to total Vav for four experiments is shown below the figure. C, Rac activation using GST-PAK was as described in Fig. 3. Wild-type or LAB Ϫ/Ϫ splenic B cells were left unstimulated (Ϫ) or stimulated with anti-BCR for 5 min (ϩ). The ratio of Rac-GTP to total Rac and normalized to the unstimulated sample is shown below the blot for four experiments.
bodies to dynamin (Fig. 6E). In contrast, when the samples were depleted three times with NRIg, we found complexes of LABdynamin (Fig. 6C) and LAB-Vav (Fig. 6E), indicating that the associated proteins survive the immunodepletion protocol. The data suggest that there are only single complexes of LAB bound to both Vav and dynamin and not individual complexes of LAB-Vav and LAB-dynamin. We cannot exclude single complexes of LAB-Vav or LAB-dynamin that are themselves associated.
Grb2 Adapter Is Needed for BCR Internalization and Bridges Dynamin and LAB-There is no immediately obvious way by which LAB or phosphorylated LAB can associate with dynamin. Unlike Vav, dynamin lacks SH2 and SH3 domains, but dynamin contains a proline-rich region that is known to associate with SH3 domains of various proteins, including that of Grb2 (41). LAB contains five tyrosine residues in a YXNX sequence that upon phosphorylation are able to engage the SH2 domain of Grb2 (39,40). Thus, a distinct possibility is that Grb2 recruits dynamin to tyrosine-phosphorylated LAB. We tested this possibility by immunoprecipitation of Grb2 and probing with antibodies to proteins within the complex: dynamin, Vav, and LAB. We found (Fig. 7A) that Grb2 immunoprecipitates did indeed contain all three proteins from activated but not unstimulated splenic B cells, consistent with the notion that dynamin, Vav, and LAB are present within a single complex.
We hypothesize that Grb2 and dynamin are associated with tyrosine-phosphorylated LAB because the SH2 domain of Grb2 is bound to one or more of the five Grb2-binding sites on LAB. To test this hypothesis, we immunoprecipitated the LAB-dynamin complex overnight in the presence or absence of the recombinant SH2 domain of Grb2 as a GST fusion protein as described previously (27). The recombinant SH2 domain of Grb2 maintains its preference for the pYXNX sequences as does the full-length protein, as we previously found (42). The expectation in this experiment was that the recombinant SH2 domain, being in excess, would compete with the endogenous Grb2 for binding to LAB and hence disrupt dynamin binding. Indeed, probing the dynamin immunoprecipitates for LAB (Fig.  7B) showed that LAB failed to co-immunoprecipitate when the recombinant SH2 domain of Grb2 was present, but not when the GST fusion partner alone was present. These findings are consistent with the hypothesis that a dimer of Grb2-dynamin associates with phosphorylated LAB through the Grb2 SH2 domain. In contrast, including the recombinant SH2 domain of Grb2 did not affect the association of Vav to LAB (Fig. 7B), indicating these proteins do not associate through the Grb2 SH2 domain.
Grb2 Is Required for Optimal B-T Interaction-The data above suggest Grb2 is an important protein for dynamin recruitment to permit BCR internalization, antigen processing, and presentation. To more rigorously test a possible connection between Grb2 and antigen-triggered BCR internalization, we applied siRNA targeting Grb2 or a scrambled siRNA to M12g4Rd B cells and tested their ability to internalize their BCR. We found that siRNA targeting Grb2 reduced BCR internalization induced by anti-BCR to about half the level of untreated B cells or B cells exposed to scram-  bled siRNA (Fig. 8A). Transfection with siRNA targeting Grb2 reduced Grb2 protein expression about 50%, consistent with the reduction in BCR endocytosis (Fig. 7B). The data are consistent with a critical role for Grb2 in recruiting dynamin to LAB to support BCR endocytosis.
Once internalized, the BCR and its bound antigen traffic to lysosomes, where the protein antigen is digested into peptides. After lysosome digestion, the peptides are loaded onto MHC class II for presentation to T cells. Thus, conditions that block BCR endocytosis should likewise block presentation of antigenic peptides to T cells. To test this possibility, we applied an antigen presentation system using PC-Ova incubated with M12g4Rd B cells and allowed the Ova-derived peptides to be presented to the Ova-specific T cell line, DO.11.10, as described previously (29). In this model, the responder DO.11.10 T cells are used as an indicator for the presence of antigen (PC-Ova) that was endocytosed via BCR and processed into peptides. The T cells in the presence of the M12g4Rd B cells will also respond by IL-2 production upon addition of the cognate Ova peptide corresponding to residues 323-339 (ISQAVHAAHAEINEAGR), because the peptide will displace existing peptides bound to B cell MHC class II (43).
For these studies, M12g4Rd was pulsed with 5 g/ml PC-Ova for 4 h to allow endocytosis. The B cells were washed and cocultured with DO.11.10 T cells for an additional 24 h to stimulate IL-2 production. The amount of IL-2 present in the supernatant was measured by enzyme-linked immunosorbent assay, and the data are shown in Fig. 8C. Addition of the cognate Ova peptide corresponding to residues 323-339 to a culture containing M12g4Rd and DO.11.10 T cells stimulates the maximum amount of IL-2 in this system (about 80 pg/ml) and served as the positive control. The absence of peptide or protein anti-gen PC-Ova (no peptide) and the absence of M12g4Rd B cells (no B cells) represent the negative controls. M12g4Rd B cells expressing the same siRNA targeting Grb2 (Grb2 siRNA) but not the scrambled siRNA lowers the DO.11.10 about 50%, consistent with the extent of the Grb2 knockdown shown in Fig. 8B and the effect on BCR endocytosis shown in Fig. 8A (black  bars). However, addition of the exogenous 323-339-residue Ova peptide to the B cells expressing Grb2 siRNA completely reconstituted IL-2 production (Fig. 8A gray bars). This peptide control indicates there is no defect in the presentation process itself by B cells expressing the Grb2 siRNA and focuses the defect on the BCR endocytic process. Thus, Grb2 is an important part of the signal transduction pathway leading to BCR internalization and B-T interaction.

DISCUSSION
Mature B cells bind soluble protein antigens through a unique and clonally restricted BCR. Antigen binding stimulates a complex signal transduction biochemical cascade that supports two distinct biological events, First, BCR signaling induces transcription factors to cause expression of new genes, particularly those involved in antigen presentation to T cells. Second, BCR signaling stimulates BCR endocytosis that ultimately leads to antigen-dependent B-T interaction to drive the humoral immune response. There have been many studies linking BCR signaling to the activation of transcription factors. Conversely, how BCR and bound antigen is internalized to initiate humoral immunity is relatively unstudied.
We earlier reported (29) that BCR internalization requires the protein dynamin and was completely blocked in B cells lacking Vav1 and -3 isoforms. Remarkably, Vav1,3-deficient B cells were fully competent in other BCR-triggered signaling events, including the influx of extracellular Ca 2ϩ , the activation of MAPK modules (p38, c-Jun N-terminal kinase (JNK), and extracellular signal-regulated kinase (ERK)), and the phosphorylation of Akt. Similarly, anti-BCR-stimulated B cells lacking Vav1 and -3 could up-regulate those proteins involved in B-T interaction as follows: MHC class II, CD80, and CD86 (29). These findings suggest that the signal transduction events and proteins that support transcription factor activation and those supporting BCR endocytosis are, at least in some cases, completely distinct.
Here, we have extended those previous findings to identify a molecular target for Vav and a mechanism by which Vav 1/3 proteins are activated. First, we showed that RNA interferencemediated depletion of Rac1 or -2 protein but not other small GTPases known to be targets of Vav (Rac3, Cdc42, RhoA, and RhoG) block BCR endocytosis. Second, we showed that Vav is recruited through its protein interaction C-terminal domain to the adapter protein LAB to form a complex that also contains Grb2-dynamin. The association of dynamin to LAB required the participation of Grb2. We provided evidence that the SH2 domain of Grb2 binds to phosphorylated LAB, and others have shown that the SH3 domain of Grb2 binds to the proline-rich sequence of dynamin (41). We also show that the four proteins, LAB, Grb2, dynamin, and Vav, are all present in a complex. Finally, we show here that Grb2 and earlier that dynamin and Vav (29) are all important for the generation of cognate MHC class II-peptide complexes after native antigen internalization, because their reduction or elimination by RNA interference techniques prevents antigen presentation. These data and the previous report (29) are summarized in a model shown in Fig. 9.
It was surprising that both Rac1 and Rac2 are separately required for BCR endocytosis (Fig. 2). These Rac isoforms are difficult to functionally distinguish because the amino acid sequences are highly similar. The main sequence differences are in the C termini; Rac1 has six basic amino acids preceding the CAAX site (KKRKRKCLLL) involved in prenylation and in binding to effectors (44). In Rac2, three of the six basic residues are neutral amino acids (RQQKRPCSLL). Isoform-specific knock-out animals have been generated, but Rac1 Ϫ/Ϫ mice are embryonic-lethal (45), whereas Rac2 Ϫ/Ϫ mice show defects in B cell development and generation of antibody-secreting cells (46), consistent with this report. Conditional Rac1,2 Ϫ/Ϫ mice lack all mature peripheral B cells (47) and thus the functional requirements for Rac proteins and potential isoform-specific functions (BCR endocytosis, formation of class II peptide, or humoral immune responses) in mature B cells cannot be addressed in these models. This study establishes that both isoforms are individually required for BCR endocytosis and likely have unique downstream effectors that have not been identified.
There are no data showing any functional consequences toward BCR endocytosis for the formation of the LAB-Grb2dynamin-Vav protein complex (Figs. 6 and 7). However, an important function is likely because the LAB-deficient B cells show a defect in Vav phosphorylation and Rac1/2 activation (Fig. 5) and because both Vav and Rac are obligatory for BCR endocytosis (29).
In the LAB Ϫ/Ϫ mice, BCR internalization was decreased as shown earlier (37). Vav and Rac activation were similarly and partially decreased, showing the important scaffolding role LAB plays in these BCR-triggered events. However, these processes were reduced but not absent in LAB-deficient B cells, whereas Rac activation and BCR endocytosis are completely absent in B cells of Vav1,3 Ϫ/Ϫ mice (29). BCR-stimulated Rac activation can only occur through Vav1 or -3 proteins (29), which accounts for the defect in BCR endocytosis in Vav1,3deficient B cells. The partial effect of LAB deficiency on BCR endocytosis and Vav-Rac activation suggests that Vav activation might have an alternative mechanism besides Vav protein recruitment to LAB or that another scaffolding protein can substitute for LAB. Bam32 is also important for BCR internalization and acts upstream of Rac (48). However, Bam32 requires PI3K for activation (48), and our data suggest BCR internalization does not appear to have such a requirement. It is possible that Bam32 can substitute for LAB in BCR internalization under the unique conditions when LAB is absent. Alternatively, and more likely, there may be an additional scaffolding protein(s) that is at least partially redundant with LAB to support Vav-Rac activation that is essential for BCR endocytosis.
Given the partial effect on BCR endocytosis of the LAB-deficient B cells, the efficiency of humoral immunity in LAB-deficient mice might be decreased but not lost. The efficiency issue has not been addressed, but an earlier report showed that LAB Ϫ/Ϫ and wild-type mice were not different in IgG responses following a single intraperitoneal injection of 75 g of trinitrophenyl-Ova and a single measurement of serum IgG levels after 14 days (49).
The interaction of LAB with dynamin and Vav was complex (Figs. 6 and 7). The interaction required that the B cells be stimulated and required Grb2 to recruit dynamin but not Vav. The need for B cell stimulation likely reflects a need for LAB tyrosine phosphorylation to create a docking site for SH2 domain-containing proteins like Grb2 and Vav. Five of the 10 tyrosines in LAB (tyrosines 95, 118, 136, 193, and 233) are present in motifs that are able to bind Grb2 (39,40). Phosphorylated tyrosine residues 191 (pYVNV) and 226 (pYENL) of LAT, the T cell homolog of LAB, directly bind to the SH2 domain of Vav (14). LAB Tyr-136 is present as YENV (similar to the Vav-binding site YENL of LAT), and Tyr-233 is present as YVNG (similar to the Vav-binding site YVNV of LAT). Studies on Tyr-to-Phe mutants of LAB indicated that Tyr-136, Tyr-193, and Tyr-233 are the primary phosphorylation sites (50). Thus, it is possible that the SH2 domain of Vav directly binds LAB at one or another of these phosphorylated tyrosine residues.
Dynamin appeared to associate with LAB via the adapter protein Grb2 (Fig. 7), although we have no data showing that the interaction is important for any of the processes investigated here. A role for Grb2 in LAB recruitment of dynamin was likely, because Grb2 is known to interact with dynamin (41), and LAB contains numerous Grb2-binding sites (39,40). Indeed, we found that the recombinant SH2 domain of Grb2 was able to block dynamin association to LAB. However, the recombinant Grb2 SH2 domain did not alter LAB-Vav interac- FIGURE 9. Model for signaling events that support BCR endocytosis and B-T interaction. BCR clustering by antigen stimulates tyrosine phosphorylation of LAB, which then recruits Grb2-dynamin and Vav1,3. An unknown protein appears to be partially redundant with LAB, indicated by ? and based on the partial BCR endocytosis effects in LAB-deficient B cells shown in Fig. 4. Grb2 binds to LAB via its SH2 and to dynamin via its SH3 domains. It is not known how Vav interacts with LAB, but the Vav SH2 domain binds to the LAB homolog LAT in T cells (14). Vav binding to LAB and/or to the unknown protein promotes Vav tyrosine phosphorylation and activation. Active Vav serves to stimulate GTP loading on Rac1,2 to promote F-actin reorganization and BCR endocytosis. The endocytosed BCR ultimately traffics to compartments allowing antigen hydrolysis and peptide loading on MHC class II. It is not known what proteins act downstream of Rac in the process, but it may include WAVE and the Arp2/3 complex. tion despite the fact that Grb2 is known to associate with Vav (51), and Vav co-precipitated with Grb2 in stimulated B cells (Fig. 6A). It may be that Vav is able to directly bind LAB by its own SH2 domain, as discussed above. The hypothesis that the Vav SH2 domain directly binds to phosphorylated LAB is consistent with the need for the protein interaction C-terminal region of Vav for its own activation (Fig. 4).
It is interesting that we failed to find separate complexes of LAB-Vav or LAB-Grb2-dynamin (Fig. 6), and the reasons for this are not understood. It is possible that separate complexes are present but also associate through LAB-LAB or another protein interaction. We have observed that cells expressing short hairpin RNA targeting dynamin showed a decrease in Vav phosphorylation (data not shown) after BCR stimulation. This latter finding suggests that dynamin recruits a kinase needed for Vav phosphorylation when Vav is present in the LAB protein interaction complex.
We have considered proteins that might act downstream of Rac to reorganize F-actin during BCR endocytosis. WASp is a member of a family of proteins that regulate the actin cytoskeleton organization (52) and include WASp, N-WASp, and WAVE1-3. WASp and N-WASp contain a Cdc42/Rac interaction binding motif that directly binds activated Cdc42 (52). WAVE1-3 lack the Cdc42/Rac interaction binding motif but act downstream of Rac in a variety of situations (53,54). The members of the WASp family bind actin monomers and Actinrelated protein 2 and 3 (Arp2/3 (55)). The Arp2/3 dimer serves as an actin nucleation site by binding to pre-existing actin filaments (56). Arp2/3 thus enhances actin polymerization to produce a branched filament structure. Repeated branching by Arp2/3 and actin monomer binding produces an actin network, and the process is greatly accelerated in the presence of active Rac (57,58). A recent report showed that WASp was recruited to the BCR cap and that WASp became phosphorylated by Btk upon BCR stimulation (59). Furthermore, BCR-triggered actin reorganization was dependent on Btk, raising the possibility that WASp acts downstream of Btk in actin reorganization. However, this hypothesis contrasts with findings in WASp Ϫ/Ϫ mice, where WASp appeared not to function in BCR capping or in the humoral immune response to vaccination (60). BCR endocytosis was not measured in either report. There is no information on the role of N-WASp or WAVE in BCR internalization. We are actively pursuing the possible involvement of WAVE and Arp2/3 in BCR endocytosis and in B-T interaction.