Vav family proteins constitute disparate branching points for distinct BCR signaling pathways

Antigen recognition by B‐cell antigen receptors (BCRs) activates distinct intracellular signaling pathways that control the differentiation fate of activated B lymphocytes. BCR‐proximal signaling enzymes comprise protein tyrosine kinases, phosphatases, and plasma membrane lipid‐modifying enzymes, whose function is furthermore coordinated by catalytically inert adaptor proteins. Here, we show that an additional class of enzymatic activity provided by guanine‐nucleotide exchange factors (GEFs) of the Vav family controls BCR‐proximal Ca2+ mobilization, cytoskeletal actin reorganization, and activation of the PI3 kinase/Akt pathway. Whereas Vav1 and Vav3 supported all of those signaling processes to different extents in a human B‐cell model system, Vav2 facilitated Actin remodeling, and activation of Akt but did not promote Ca2+ signaling. On BCR activation, Vav1 was directly recruited to the phosphorylated BCR and to the central adaptor protein SLP65 via its Src homology 2 domain. Pharmacological inhibition or genetic inactivation of the substrates of Vav GEFs, small G proteins of the Rho/Rac family, impaired BCR‐induced Ca2+ mobilization, probably because phospholipase Cγ2 requires activated Rac proteins for optimal activity. Our findings show that Vav family members are key relays of the BCR signalosome that differentially control distinct signaling pathways both in a catalysis‐dependent and ‐independent manner.


Introduction
Engagement of BCR by cognate antigen initiates a multistep process of cellular activation and differentiation that eventually can lead to formation of antibody-secreting plasma cells [1,2]. Antigen-ligated BCRs activate intracellular signaling pathways by means of ITAMs in the cytoplasmic domains of their signaltransducing transmembrane proteins Igα and Igβ (CD79A and position 204 (Y204) of Igα [5,6]. Tyrosine-phosphorylated SLP65 serves as a docking platform for both Bruton´s tyrosine kinase (Btk) and its immediate substrate phospholipase Cɣ2 (PLCɣ2) [7,8]. This process of PLCɣ2 activation can furthermore be supported by accessory proteins such as Grb2 and CIN85 [9,10]. Activated PLCɣ2 hydrolyzes the plasma membrane phospholipid phosphatidyl-inositol-4,5-bisphosphate (PIP2), thus, generating the second messenger molecules diacylglycerol (DAG) and inositol-1,4,5-trisphosphate (IP3). Plasma membrane-bound DAG recruits C1 domain-containing signaling proteins like protein kinase C and Ras guanine nucleotide release proteins, whereas the soluble IP3 induces opening of IP3-gated Ca 2+ channels in the membrane of the endoplasmic reticulum called IP3 receptors (IP3R), allowing Ca 2+ ions to enter the cytosol [5]. This first wave of Ca 2+ influx into the cytosol is linked to a second wave controlled by store-operated Ca 2+ channels in the plasma membrane [11]. Both the intensity and duration of BCR-induced Ca 2+ mobilization are tuned by a variety of activating or inhibiting cell surface coreceptors and by intracellular enzymes and adaptor proteins [12,13]. The mobilization of the second messenger Ca 2+ and the activation of BCR signaling in general were furthermore reported to be under control of the cortical actin cytoskeleton as treatment of certain B cells with actin filament destabilizing compounds can cause a "BCR-like" Ca 2+ signal [14,15]. However, this situation may be restricted to B cells that coexpress mIgD-BCRs, CD19, and the chemokine receptor CXCR4 together on their cell surface [16]. Whatever the case may be, the integrated activity of cellular Ca 2+ -tuning signaling molecules not only shapes the kinetics of the Ca 2+ flux, but also that of Ca 2+ -and DAG-sensitive signaling cascades, such as the NF-AT, NF-κB, or MAP kinase pathways, in B cells. Eventually, the combined activities of these pathways adjust the cellular activation and differentiation processes and, thus, control the strength and quality of a humoral immune response [12,17].
Although the molecular composition of the BCR signalosome as well as the order of events that lead to Ca 2+ mobilization have been studied in considerable detail, the picture remains incomplete. Genetic mouse models showed that members of the Vav family of guanine nucleotide exchange factors (GEFs), which control downstream activation of small G proteins of the Rho/Rac family [18], are moreover essential for BCR-(and TCR-) induced Ca 2+ mobilization. Whereas solitary deficiency for Vav1 had little impact on BCR-induced Ca 2+ signaling, deficiency for Vav1 and Vav2 or all three Vav family members severely impaired Ca 2+ signaling in mouse B cells [19][20][21]. Similarly, a Vav3-deficient variant of the chicken B-cell line DT40 showed reduced BCR-induced Ca 2+ signaling and impaired activation of PI3 kinase [22]. Yet, the mechanistic basis for these observations remains unknown. Specifically, it remains unclear if the catalytic GEF activity of Vav proteins is required for BCR-proximal Ca 2+ signaling or if Vav proteins fulfil some kind of adaptor function [23]. Moreover, Vav was shown to associate with tyrosine-phosphorylated CD19 and to be required for CD19-mediated amplification of BCR-induced Ca 2+ signaling [24]. However, it remains unclear how Vav is employed by the BCR in the absence of CD19 stimulation.
Using a human B-cell model system that was made deficient for Vav1 by genome editing and used for subsequent expression of distinct Vav isoform variants, we show that Vav1 and Vav3 coordinate BCR-induced Ca 2+ mobilization via their catalytic activity and by additional means that probably involve adaptor functions. We furthermore show that the GEF activity of Vav is essential for BCR-induced actin reorganization, a process that is supported by all three members of the Vav family, which however, is uncoupled from Ca 2+ mobilization. Furthermore, Vav proteins mediate activation of the PI3 kinase/Akt pathway in antigen-activated B cells, which only partially depends on their GEF activity. Mechanistically, Vav1 is directly recruited to the activated BCR and its signalosome by binding to phosphorylated Igα tyrosine residues and to phospho-SLP65. Also, the substrates of Vav family GEFs, small G proteins of the Rho/Rac family, were required for optimal Ca 2+ signaling in human B cells. Collectively, our data show that BCR-proximal signaling is critically and differentially coordinated by GEF of the Vav family.

Expression of Vav1 in a human B-cell model system controls BCR-induced Ca 2+ mobilization
To investigate the functional principles of Vav family proteins in BCR-induced signaling reactions, we generated a cellular genetic model system by inactivating the VAV1 gene in the human B-cell line DG75 using TALEN-mediated genome editing (for details see Supporting information Fig. S1 and reference [25]). Several Vav1deficient DG75 subclones showed severely compromised BCRinduced Ca 2+ mobilization and, thus, mirrored the phenotype of primary mouse B-cells deficient for either Vav1 and Vav2 or all three Vav isoforms (Fig. 1A). Re-expression of citrine-tagged Vav1 restored the Ca 2+ mobilization defect, demonstrating that impaired BCR signaling in Vav1-deficient cells was indeed caused by the lack of Vav1 ( Fig. 1B and C). However, the Ca 2+ kinetics of cells expressing ectopic Vav1 was enhanced in comparison to parental DG75 cells. Comparison of endogenously and ectopically expressed Vav1 showed that the latter was present in larger amounts (Supporting information Fig. S2A). Consistently, ectopic overexpression of Vav1 in parental DG75 cells enhanced BCR-induced Ca 2+ mobilization (Supporting information Fig. S2B and C), indicating that Vav1 regulates the intensity of BCR-proximal signaling in a dose-dependent manner. We then tested if the impaired BCR-induced Ca 2+ signal in Vav1-deficient DG75 cells was due to impaired phosphorylation of components of the B cell Ca 2+ initiation complex. Thus, we affinity-purified PLCγ2 from lysates of BCR-stimulated B cells, and tested tyrosinephosphorylation of PLCγ2 and associated proteins by immunoblotting. However, neither tyrosine phosphorylation of PLCγ2 nor that of copurified SLP65 or Igα/β was reduced in cells lacking Vav1 (Fig. 1D). Furthermore, analysis of SLP65 tyrosine phosphorylation by intracellular flow cytometry did not reveal any differences between Vav1-deficient and -proficient cells (Fig. 1E)     hydrogen bonds with Rac1 [29]. Replacement of this glutamate with alanine (E201A) was reported to eliminate Vav1´s GEF activity [29], which we confirmed in in vitro GEF assays toward the small G proteins Rac1, RhoA, and Cdc42 (Supporting information Fig. S3). When expressed in Vav1-deficient DG75 B cells, the E201A variant of Vav1 only poorly supported BCR-induced Ca 2+ signaling ( Fig. 2B and C). In addition to E201, the glutamine at position 331, the asparagine at position 371, and the glutamate at position 378 of Vav1 make contacts with Rac1 [29,30]. The corresponding alanine replacements (Q331A, N371A, and E378A) also resulted in markedly reduced BCR-induced Ca 2+ kinetics (Supporting information Fig. S4A-D). Yet another way to inactivate the catalytic activity of Vav1 was used by Saveliev et al., which involved the replacement of the two adjacent amino acids leucine 334 and lysine 335 in murine Vav1 with alanine residues (LK/AA) [31]. The phenotype of LK/AA-mutant murine Vav1, which was indeed inactive toward Rac1, RhoA, and Cdc42 (Supporting information Fig. S4E) closely resembled that of E201A-mutant human Vav1 (Supporting information Fig. S4F). Importantly, all of the analyzed DH domain mutants were expressed in amounts comparable to WT Vav1 ( Fig. 2C and Supporting information Fig. S4B, D, & G). The zinc finger (ZF) of Vav1 makes intramolecular contacts with the DH and PH domain and, thus, supports the catalytic activity of Vav [29]. We inactivated the ZF by replacing either glutamine 542 or tyrosine 544 with alanine (Q542A, Y544A), which abrogates Vav1´s GEF activity (Supporting information Fig. S3) [32]. Similar to the E201A variant, both ZF-mutant variants of Vav1 could only inefficiently restore BCR-induced Ca 2+ mobilization ( Fig. 2D and E). The GEF activity of Vav proteins is furthermore autoinhibited by a peptide loop, termed acidic region (AR), located between the N-terminal CH and the catalytic DH domain (see Fig. 2A). Deletion of the AR enhances the catalytic activity of Vav family proteins [33], which could also be observed C 2020 The Authors. European Journal of Immunology published by Wiley-VCH GmbH www.eji-journal.eu in our in vitro GEF assays toward Rac1 and RhoA (Supporting information Fig. S3). Expression of a Vav1 variant lacking the AR (Vav1-AR) resulted in a markedly enhanced Ca 2+ profile in BCR-activated DG75 cells (Fig. 2F), even though this variant was less well expressed than WT Vav1 (Fig. 2G). Together, these experiments reveal a critical function of the catalytic GEF activity of Vav1 for optimal BCR-induced Ca 2+ signaling, and thus, show that Vav1 does not just act as an adaptor protein in that process.

Vav1 is directly recruited to the phosphorylated BCR
Having established the importance of the catalytic activity of Vav1 for BCR-induced Ca 2+ signaling, we tested the role of its C-terminal adapter region for that process. We first inactivated the Vav1 SH2 domain by replacing the critical arginine residue in the SH2 domain binding pocket with alanine (R696A), which completely suppressed Vav1´s function in the BCR signaling cascade ( Fig. 3A and B). To identify the Vav1 SH2 domain interaction partners within the BCR signalosome, we performed affinity purification experiments using a recombinant Vav1 SH2 domain and lysates of BCR-activated B cells. As controls, we used the R696A-mutant SH2 domain and the tandem SH2 domains of the PTK Syk. Immunoblotting with anti-phosphotyrosine antibodies showed that the WT, yet not the R696A-mutant Vav1 SH2 domain interacts with a distinct set of phosphoproteins from activated B cells (Fig. 3C). The higher molecular weight proteins were identified as SLP65, Syk, and HS1 in separate experiments (Supporting information Fig. S5 and data not shown). Strikingly, the Vav1 SH2 domain purified a diffusely running set of phosphoproteins that was also purified by the tandem SH2 domains of Syk, which are well known to bind to the Igα/β heterodimer of the BCR. We identified these proteins as Igα/β by direct immunoblotting (Supporting information Fig.  S5A). To further characterize the interaction between Vav1 and the BCR, we used a set of phosphorylated peptides encompassing the known tyrosine-phosphorylation motifs of Igα and Igβ, respectively, for affinity purification experiments from B cell lysates (Supporting information Fig. S5B). Indeed, Vav1 interacted with phosphopeptides from Igα in which either the ITAM tyrosines were phosphorylated (Fig. 3D, Igα-pepI 1 and 1+2) or in which the non-ITAM tyrosine residue Y204 was phosphorylated (Igα-pepII 2+3 and 3). However, Vav1 was not purified by the phosphorylated Igβ peptide, indicating that the SH2 domain of Vav1 prefers the phosphotyrosine binding motifs of Igα over those of Igβ. To verify the integrity of the used phosphopeptides, we tested for the presence of the known BCR-interacting proteins Syk and SLP65, which were bound by either doubly-phosphorylated ITAM motifs (Syk) or the Igα non-ITAM Y204 motif (SLP65) as expected (Fig. 3D, middle and lower panel, respectively). Finally, binding of a recombinant Vav1 SH2 domain to a phosphopeptide from Igα proved that complex formation between Vav and phospho-Igα is due to a direct interaction of the two partners (Fig. 3E).

Complex formation between Vav1 and the BCR facilitates Ca 2+ signaling
To investigate whether the newly discovered interaction between Vav1 and Igα tyrosine docking sites in the BCR is of functional relevance, we generated chimeric variants of Vav1 in which we replaced its endogenous SH2 domain with either those of Syk or that of SLP65 (see Supporting information Fig. S6A). The SH2 domains of Syk and SLP65 specifically bind to either ITAM or non-ITAM tyrosine motifs of the BCR (see Fig. 3D) and, hence, target the chimeric Vav1 variants directly to the activated BCR (Supporting information Fig. 6B). Indeed, both Vav1-SH2 domain chimeras restored BCR-induced Ca 2+ mobilization albeit to different extents ( Fig. 4A and Supporting information Fig. S6C). These differences are most likely due to diverging affinities and numbers of available binding sites (one non-ITAM tyrosine (Y204) motif for the SLP65 SH2 domain versus two ITAM motifs for the tandem SH2 domains of Syk per BCR complex, see Supporting information Fig. S6D). Note that also the Vav1 SH2 domain has several potential docking sites per BCR complex and furthermore can interact with SLP65 and Syk (see Fig. 2), which might explain why it works better than the SH2 domain of SLP65. To exclude that the SLP65 SH2 domain in the context of the chimeric Vav1 protein uses an unknown alternative binding site besides Igα Y204, we additionally expressed a chimeric transmembrane protein consisting of the extracellular and transmembrane parts of CD8 and the intracellular domain of Igα in cells expressing either WT Vav1 or the Vav1-SLP65 SH2 domain chimera (Supporting information Fig. S6D). Stimulation of this CD8/Igα chimera resulted in a BCR-like Ca 2+ signal in cells expressing WT Vav1 and a slightly reduced signal in cells expressing the Vav1-SLP65 SH2 domain chimera, while inactivation of the SLP65 SH2 domain (Vav1-SLP65-SH2-RL) disabled Ca 2+ signaling (Fig. 4B). Likewise, inactivation of Y204 in the CD8/Igα chimeric receptor (CD8/IgαY204F) selectively abolished Ca 2+ mobilization in cells expressing the Vav1-SLP65 SH2 domain chimera, but not in cells expressing WT Vav1 (Fig. 4C). These results show that the SH2 domains of Syk and SLP65 targeted the Vav1 chimeras to the phosphorylated BCR, and that this route of Vav1 plasma membrane recruitment is functionally relevant for BCR-induced Ca 2+ mobilization.

Binding of Vav to phospho-SLP65 represents an alternative membrane recruitment pathway
Previous studies had shown that Vav1 can associate via its SH2 domain with phosphorylated SLP65 [7,34], but a functional role of this protein-protein interaction was never described. Hence, we tested whether this interaction is of functional relevance using a similar approach as before. This time we replaced the Vav1 SH2 domain with those of the Tec family PTKs Btk or Itk (Supporting information Fig. S6E), both of which interact with one specific tyrosine-phosphorylation site, Y96, in human SLP65 [7,8] (Supporting information Fig. S6F). Noteworthy, the SH2 domain of Itk has a superior affinity compared to that of Btk (Supporting  information Fig. 6G) as previously reported [8]. Both SH2 domain chimeras could support BCR-induced Ca 2+ mobilization to some extent (Fig. 4D), showing that the recruitment of Vav1 to phospho-SLP65, that is, to the immediate microenvironment of activated BCRs, is sufficient to allow Vav proteins to exert their function in BCR-proximal signaling reactions.   human peripheral blood B cells and in our DG75 B cell model system (Fig. 5A). All three isoforms of Vav were readily detected in lysates of primary human B cells. By contrast, DG75 B cells expressed considerable amounts of Vav1 and Vav2, but hardly any Vav3. When expressed in Vav1-deficient DG75 B cells, all three Vav family members differed strongly in their ability to promote BCR-induced Ca 2+ signaling. Whereas expression of Vav3 in DG75 cells caused a much enhanced and strongly prolonged Ca 2+ signal as compared to expression of Vav1, expression of Vav2 did not restore BCR-induced Ca 2+ mobilization at all (Fig. 5B). Notably, Vav3 was a very strong stimulator of Ca 2+ mobilization despite being expressed at reduced amounts compared to Vav1 and Vav2 (Fig. 5C). To test if the robust Ca 2+ signal caused by Vav3 required its GEF activity, we inactivated the catalytic DH domain in Vav3 by replacing the glutamate at position 199 with alanine (E199A, Supporting information Fig. S7). Similar to the Vav1 E201A variant, E199A-mutant Vav3 gave rise to a strongly reduced Ca 2+ signal (Fig. 5D), even though its expression level was comparable to that of WT Vav3 (Fig. 5E). We furthermore investigated why Vav2 failed to support BCRinduced Ca 2+ signaling. We hypothesized that its inhibitory AR might interfere with proper activation of this isoform in B cells. To test this notion, we deleted the AR in Vav2 (Vav2-AR) and tested its functionality as before (Fig. 5F). Even though Vav2-AR was less well expressed than WT Vav2 (Fig. 5G), it fully restored Ca 2+ signaling, indicating that Vav2 can mediate Ca 2+ mobilization in B cells when its catalytic activity is liberated.

Vav proteins stimulate BCR-induced activation of the PI3 kinase/Akt pathway
A previous report suggested a role for Vav3 in the activation of PI3 kinase in the chicken B-cell line DT40, yet the mechanistic basis for this observation remained unclear [22]. Hence, we used our human B-cell model system to address that issue. As readout for PI3 kinase activity, we measured phosphorylation of the serine/threonine kinase Akt (PKB) by intracellular flow cytometry. Indeed, BCR-induced phosphorylation of Akt was reduced in Vav1-deficient B cells to about 50% of the level observed in parental DG75 cells (Fig. 6A and Supporting information Fig. S8). This defect was restored by re-expression of WT Vav1, and also by expression of the catalytically inactive Vav1 E201A variant, albeit to a somewhat reduced extent (Fig. 6A), indicating that Vav-mediated PI3 kinase activation is partially uncoupled from its GEF activity. Like Vav1, also Vav3 fully supported activation of Akt (Fig. 6B).
Since the magnitude of BCR-induced Ca 2+ mobilization roughly correlated with the efficiency of Akt phosphorylation in our experiments, we tested whether Ca 2+ mobilization, is required for PI3 kinase activation in B cells. For this purpose, we used DG75 B cells deficient for both PLCγ isoforms, PLCγ1 and PLCγ2 (PLCγdko) that are completely deficient for BCR-induced Ca 2+ signaling [25]. More precisely, we tested in this experiment if any of the PLCγ-derived second messengers controls PI3 kinase signaling. Figure 6C shows that BCR-induced Akt phosphorylation did not require expression of PLCγ. Hence, the Ca 2+ -promoting activity of Vav1 and Vav3 is not directly responsible for PI3 kinase signaling in BCR-activated cells.
We furthermore tested the ability of Vav2 to promote activation of PI3 kinase. Surprisingly, BCR-induced phosphorylation of Akt could be restored by expression of Vav2 (Fig. 6D) although this isoform failed to support Ca 2+ signaling (see Fig. 5B). This result indicates that insufficient activation of PI3 kinase in the Vav1-deficient cells is not responsible for defective Ca 2+ signaling. Notably, the level of Akt phosphorylation in cells expressing Vav2 resembled that of cells expressing catalytically inactive Vav1. Thus, we tested if deletion of the AR in Vav2 would further enhance activation of Akt and found that the AR-variant of Vav2 indeed had a somewhat increased ability to promote phosphorylation of Akt (Fig. 6D). In summary, these results demonstrate that distinct members of the Vav GEF family differentially control BCR-induced Ca 2+ mobilization and activation of the PI3 kinase/Akt pathway.

Vav proteins are a branching point for BCR-induced Actin reorganization and Ca 2+ mobilization
Reorganization of the actin cytoskeleton has been implicated to be involved in the initiation of BCR-induced Ca 2+ mobilization [14,15]. Since Vav proteins are activators of Rho/Rac family small G proteins, which control Actin reorganization in many cell types, we tested whether they control actin remodeling downstream of the BCR (Supporting information Fig. S9). Cells expressing WT Vav1 showed an increased Phalloidin staining intensity that peaked after 30 to 60 s following BCR activation and reached a plateau between the 3 and 7 min stimulation time points (Fig. 7A, blue curve). By contrast, BCR-induced formation of F-Actin was severely compromised in cells lacking Vav1 (Fig. 7A, black curve). Catalytically inactive Vav1-E201A was impaired in its ability to support BCR-induced Actin remodeling (Fig. 7B), indicating that the GEF activity of Vav is required for that process. In line with our previous observations regarding Ca 2+ mobilization, Vav3 was even more effective than Vav1 in actin remodeling despite being less well expressed (Figs. 5C and 7C). Surprisingly, also Vav2 was a very effective mediator of BCR-induced actin reorganization (Fig. 6D). This result indicated that BCR-induced reorganization of the actin cytoskeleton and the mobilization of Ca 2+ ions are separable events. To test this notion in more detail, we titrated the Actin polymerization-inhibiting drug Latrunculin A (LatA) to determine the minimal effective concentration that was needed to prevent formation of F-Actin in BCR-activated DG75 cells (Supporting information Fig. S10). We then pretreated Vav1expressing B cells for 5 min (as we did in the titration experiments shown in Supporting information Fig. S10) with concentrations of either 0.5 or 1 μM LatA to block actin reorganization, and then analyzed BCR-induced Ca 2+ mobilization. Evidently, inhibition of actin reorganization had no detectable effect on BCRinduced Ca 2+ signaling (Fig. 7E). However, since administration of LatA to B cells was previously reported to instantly evoke a BCR-like Ca 2+ signal [15], we repeated the experiment and this time monitored Ca 2+ flux during the 5 min incubation time with LatA, followed by BCR stimulation as before. Under these conditions, LatA caused a detectable, yet very subtle release of Ca 2+ , which was strongly enhanced by BCR stimulation (Fig. 7F)  BCR-induced Ca 2+ mobilization and actin reorganization, yet they appear to represent branching points at which these two processes diverge.

Vav family substrates promote BCR-induced Ca 2+ signaling
Since our experiments revealed that the catalytic GEF activity of Vav proteins regulates BCR-proximal Ca 2+ mobilization, we tested whether the substrates of Vav, small G proteins of the Rho/Rac family, are involved in that process as well. Recently, two effective pharmacological inhibitors of Rho/Rac proteins were described, called Rhosin and EHop-016, respectively [35,36]. Rhosin is reported to be specific for RhoA whereas EHop-016 seems to preferentially target Rac proteins. Both inhibitors strongly impaired Ca 2+ signaling in DG75 B cells, resulting in Ca 2+ kinetics that resembled that of Vav1-deficient cells (Fig. 8A). Treatment of primary human B cells with EHop-016 caused a similar reduction in BCR-induced Ca 2+ signaling as in DG75 B cells (Fig. 8B & Supporting information Fig. S11), indicating that Rac and Rho family proteins are indeed involved in Ca 2+ signaling in antigenactivated B cells. To verify the inhibitor experiments and to more precisely identify the small G protein(s) involved in BCR-induced Ca 2+ mobilization, we furthermore generated genetic model systems. Since inactivation of Rac2 in the mouse was previously reported to affect Ca 2+ mobilization in B cells [37], we first deleted Rac2 in DG75 B cells using CRISPR/Cas9-directed genome editing (Supporting information Fig. S12A-D). Similar to mouse B cells, deficiency for Rac2 had a moderate effect on BCR-induced Ca 2+ mobilization in DG75 cells (Fig. 8C). As controls, we ectopically expressed Rac2 in the Rac2-ko cells and in parental DG75  Fig. S12E), which rescued the Ca 2+ mobilization defect of Rac2-deficient cells (Fig. 8D) and resulted in an elevated Ca 2+ signal in parental DG75 B cells, respectively (Supporting information Fig. S12F). Nevertheless, since Rac2deficiency did not fully recapitulate the Ca 2+ mobilization defect of DG75 cells expressing catalytically inactive Vav1 or Vav3, we furthermore inactivated Rac1 by CRISPR/Cas9-mediated genome editing (Supporting information Fig. S13A-C). Deficiency for Rac1 resulted in a noticeable reduction of BCR-induced Ca 2+ mobiliza-tion (Fig. 8E). Again, ectopic re-expression of Rac1 reverted the Ca 2+ phenotype of the Rac1-ko cells (Supporting information Fig.  13D and Fig. 8F). We also tried to generated Rac1/2 doubledeficient DG75B cells and to inactivate RhoA by CRISPR/Cas9 mutagenesis, however, neither Rac1/2 double-deficient cells nor RhoA-deficient clones could be obtained, probably because DG75 cells require these proteins for survival and/or proliferation. Hence, we could not test the role of RhoA in BCR-proximal Ca 2+ signaling in a genetic model system. Nevertheless, the available C 2020 The Authors. European Journal of Immunology published by Wiley-VCH GmbH www.eji-journal.eu genetic and pharmacological data indicate that probably both, Rac and RhoA, are involved in BCR-proximal second messenger generation. Rac1 and Rac2 have previously been implicated in the activation of PLCγ2 in BCR-activated chicken B cells [38]. This mechanism seems to involve a tyrosine-phosphorylationindependent protein-protein interaction between Rac and PLCγ2 [39], which can be abrogated by replacing phenylalanine 897 of PLCγ2 with glutamine (F897Q) [38]. To test whether the missing interaction between Rac and PLCγ2 could explain the Ca 2+ phenotype of DG75 cells lacking Rac1 or Rac2, we tested the performance of F897Q-mutant PLCγ2 (tagged with EGFP) in PLCγ2-deficient DG75 B cells. As controls, we analyzed cells expressing WT PLCγ2 and cells lacking Rac2. Indeed, cells expressing F897Q-mutant PLCγ2 showed a reduced Ca 2+ kinetics that was even a little lower than that of Rac2-deficient cells (Fig. 8G), which may indicate that Rac2 and Rac1 are redundant in their ability to activate PLCγ2. Importantly, WT and F897Q-mutant PLCγ2 were expressed in equal amounts in the transfected cells (Fig. 8H).

Discussion
Here, we have shown that Vav proteins regulate critical BCR signaling processes including mobilization of the key second messenger Ca 2+ , activation of the PI3 kinase/Akt pathway, and actin cytoskeleton remodeling in the human Burkitt lymphoma cell line DG75. The involvement of Vav isoforms in B-cell Ca 2+ signaling was observed earlier in gene-targeted mice lacking more than one Vav family member. However, these previous results lacked a mechanistic explanation and furthermore were somewhat inconsistent. Whereas some groups reported that BCRinduced Ca 2+ signaling is normal in cells lacking only Vav1 but defective in B cells lacking Vav1 and Vav2 [19,20] others found essentially normal Ca 2+ mobilization in B cells from the very same Vav1/2 double-deficient mice [15]. In addition, another group showed defective Ca 2+ signaling in mouse B cells lacking all three Vav family members [21]. Our human Vav1-deficient B cells expressed only low amounts of Vav2 and hardly any Vav3. They thus represent a suitable model system to study distinct Vav isoforms. Whereas Vav1 and Vav3 were able to promote B cell Ca 2+ signaling, Vav2 failed to support this process. This was not due to a general functional defect of this isoform, since Vav2 efficiently promoted BCR-induced activation of PI3 kinase/Akt as well as cytoskeletal reorganization. Interestingly, also in Jurkat T cells, Vav2 failed to support TCR-induced Ca 2+ mobilization [40]. By contrast, murine and chicken Vav2 were previously shown to facilitate BCR-induced Ca 2+ mobilization in a murine and in a chicken B cell line, respectively [22,41]. This discrepancy to our results may be due to species-related differences of either the used Vav2 orthologues or the respective cellular context. Indeed, species-dependent differences were recently observed with regard to activation of the Erk MAP kinase pathway in antigen-activated B cells [25]. Whatever the exact reason may be, the inability of human Vav2 to facilitate BCR-induced Ca 2+ signaling could be relieved by deletion of its inhibitory AR, indicating that this domain effectively and selectively blocks the Ca 2+ -promoting activity of Vav2 [42]. Furthermore, our mutational analyses of the DH domain and the ZF region in Vav1 and Vav3 clearly showed that the catalytic activity of Vav is mandatory for proper BCR-induced Ca 2+ signaling. However, the Ca 2+ kinetics of DH domain-mutant Vav1 and Vav3 were not as strongly reduced as that of Vav1-ko cells, indicating that additional Vav-specific mechanisms are required for an optimal Ca 2+ response, which may involve adaptor functions of either the C-terminal SH3 domains and/or the N-terminal CH domain. Future experiments will be required to reveal the function of these domains in BCR-proximal signaling. The requirement of Vav3 for BCR-induced activation of PI3 kinase/Akt has previously been observed in the chicken B cell line DT40 [22]. This observation led to the conclusion that impaired PI3 kinase activation is responsible for the Ca 2+ mobilization defect in Vav3 deficient DT40 B cells [22]. However, our results challenge this interpretation, since expression of human Vav2 in DG75 cells restored PI3 kinase signaling, yet completely failed to support Ca 2+ mobilization. Hence, besides facilitating the activation of PI3 kinase there must be an alternative and/or additional mechanism by which Vav proteins promote Ca 2+ signaling in B cells. We tested whether this alternative mechanism involves reorganization of the Actin cytoskeleton, since previous studies had inferred that the actin membrane skeleton controls BCR activation [14,15]. Indeed, BCR-induced formation of F-Actin was strongly reduced in Vav1-deficient DG75 B cells and -like the Ca 2+ mobilization kinetics -could only partially be restored by catalytically inactive E201A-mutant Vav1. Similar to the activation of PI3 kinase, Actin remodeling was very efficiently restored by Vav2, which however, failed to promote Ca 2+ mobilization. Furthermore, pharmacological inhibition of Actin reorganization by Latrunculin A treatment had hardly any detectable effect on Ca 2+ signaling in B cells. These findings combined make it unlikely that the Actin cytoskeleton is a key control element of BCR-proximal Ca 2+ signaling. Consistently, a recent report showed that Latrunculin A-induced Ca 2+ signaling is specific for IgD-containing BCRs and furthermore requires coexpression of CD19 and the chemokine receptor CXCR4 [16]. Taken together, Vav proteins are apparently involved in the activation of several distinct signaling processes that are initiated by BCR engagement, yet these distinct signaling reactions appear to be -at least partially -uncoupled from each other. Thus, Vav proteins seem to serve as branching points in the BCR signaling cascade (Supporting information Fig. S14).
It is important to note that we have used soluble reagents for BCR stimulation in our experiments. However, B cells can as well detect cell surface-bound antigens, in which case the Actin cytoskeleton plays a central role when B cells spread over membranes and use mechanical force to actively take up antigen [43][44][45][46]. Yet the factors that couple BCR signaling to reorganization of the actin cytoskeleton during recognition of membrane-bound antigen are not well characterized. Our data suggest that Vav proteins may be key mediators in that process.
Since the catalytic activity of Vav was required for proper Ca 2+ signaling in B cells, we tested the involvement of small G proteins of the Rho/Rac family using two approaches: pharmacological inhibition and genetic inactivation. The RhoA inhibitor Rhosin (which may also target additional closely related Rho family members) strongly impaired BCR-induced Ca 2+ mobilization. Pharmacological inhibition of Rac proteins using the inhibitor EHop-016 had a similar effect. EHop-016 probably inhibits several Rac isoforms, leaving unclear precisely which one is employed in BCR-proximal signaling. However, since pharmacological inhibitors always bear the risk of causing nonspecific side effects, we furthermore performed genetic experiments. Genetic inactivation of either Rac1 or Rac2 caused a reduction of BCR-induced Ca 2+ mobilization, which however was less pronounced than that caused by EHop-016. This indicates that both Rac isoforms have partially redundant functions. Unfortunately, this could not be tested in DG75 cells since Rac1/2 double-deficient cells (like RhoA deficient cells) could not be obtained. More recently, even Cdc42 has been implicated to regulate BCR-proximal signaling [47]. However, in our in vitro GEF assays Vav was not a very robust activator of Cdc42. While Rho/Rac proteins have been widely described as regulators of the actin cytoskeleton [48], their exact functions in BCR-proximal Ca 2+ signaling and PI3 kinase activation remains to be clarified in future studies. Notably, RhoA, Rac, and Vav proteins have all been implicated in the activation of phosphatidylinositol 4-phosphate 5-kinases, which generate phosphatidyl-inositol-4,5-bisphosphate (PIP2), the substrate for PLCγ [24,49,50]. This scenario could explain why PLCγ phosphorylation was normal in Vav1-deficient cells, since it would affect the output of PLCγ but not its activation. In addition, Rac proteins were reported to directly activate PLCγ2 via a protein-protein interaction, which is prevented in the F897Q-mutant variant of PLCγ2 [38,39]. Indeed, the Ca 2+ kinetics of cells expressing F897Q-mutant PLCγ2 closely mirrored those of Rac2-deficient cells, indicating that this mechanism may -at least in partbe responsible for the diminished Ca 2+ kinetics of Rac-deficient DG75 and mouse B cells [37] and, thus, probably also for that of cells expressing catalytically inactive Vav1 or Vav3. Importantly, the requirement of GTP-bound Rac proteins for optimal phospholipase activation applies to PLCγ2 but not to PLCγ1 [51], which may explain why expression of the catalytically inactive LK-AA-mutant Vav1 in T cells did not result in diminished TCRinduced Ca 2+ signaling, since T cells express PLCγ1 but not PLCγ2 [31].
We furthermore showed that in B-cells Vav1 needs to be recruited via its SH2 domain to the BCR signalosome. The association between Vav1 and SLP65 (BLNK) was already reported together with the initial identification and characterization of SLP65 [7,34]. However, the functional significance of this interaction was never tested. Our data demonstrate that the recruitment of Vav1 to phosphorylated SLP65 is indeed of functional importance in BCR-proximal signaling. However, Vav1 also associates directly with the activated BCR and this interaction is of functional relevance as well. In conclusion, Vav proteins probably do not need to be precisely localized at a defined docking site within the BCR signalosome like, for example, Btk, which specifically interacts with one particular tyrosine phosphorylation motif in SLP65 [7,8], but instead can opt between multiple phosphorylated tyrosine residues, depending on their availability. In conclusion, the differential functional capabilities of Vav family members within the BCR signaling cascade may indicate specialized functions of individual Vav proteins in different B-cell subpopulations and/or different stages of B-cell development and furthermore may allow for pharmacological inhibition of these molecules in the treatment of B cell-related diseases.

Cells and genome editing
The human Burkitt lymphoma B-cell lines DG75 and Ramos were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). All cells were maintained in RPMI1640 + Glutamaxx (Biochrom, Berlin, Germany) supplemented with 10 % heat-inactivated FCS and antibiotics. The DG75EB variant of the cells, which expresses the murine cationic amino acid transporter 1 (SLC7A1) to make them susceptible to infection with MMLV-based retrovirus particles as well as the PLCγ2-deficient and the PLCɣ1/2 double-deficient sublines of DG75 cells were described before [25,52]. DG75 cells deficient for Vav1 were generated by TALEN-mediated genome editing. TALEN cassettes were designed using the TAL Effector Nucleotide Targeter 2.0 (https://tale-nt.cac.cornell.edu/) and generated using a modified version of the "Golden Gate" TALEN assembly method [53]. The plasmid kit used for generation of TALENs was a gift from Daniel Voytas and Adam Bogdanove (Addgene kit # 1000000024). Modifications included shortening of the linker regions Nand C-terminal of the DNA binding modules and optimization of the FokI nuclease by introducing "Sharkey" mutations that cause a higher activity without increasing off-target effects [25,54]. The TALEN constructs (designated left and right TALEN, respectively) were cloned into expression vectors pmax-IE and pmax-IR, containing an IRES-EGFP (IE) and an IRES-tagRFP (IR) cassette, respectively. DG75 cells were transiently transfected with both plasmids using the Amaxa Nucleofector TM II device (Lonza, Basel, Switzerland) in combination with the Lonza Human B cell Nucleofector TM Kit (program T-015). Two days after electroporation, EGFP/RFP doublepositive cells were sorted, expanded, subcloned, and screened for deficiency of the target protein (see Supporting information Fig.  S1 for further details). Vav1-deficient cells were retrovirally transduced to express SLC7A1. The resultant DG75EB Vav1 −/− cells were used in all experiments. Single guide RNAs were designed with the Zhang-lab online software (crispr.mit.edu) and were cloned as synthetic oligos into the pSpCas9(BB)-2A-GFP vector (Addgene #48138) [55], followed by Amaxa nucleofection, cell sorting for GFP-positive cells, subcloning, and screening by immunoblotting.

Expression vectors and retroviral gene transfer
The cDNAs encoding for human Vav and Rac family members were purchased from Dharmacon (Lafayette, CO, USA). The cDNA encoding rat PLCγ2 was provided by Dr. Tomohiro Kurosaki, Osaka University, Japan, and was equipped with an N-terminal EGFP tag. Chimeric variants of Vav1 were generated using overlap PCR. Chimeric CD8/Igα was described previously [56]. For sitedirected mutagenesis, cDNAs were cloned into the plasmid pCR2.1 (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). The integrity of all constructs was confirmed by Sanger DNA sequencing (SeqLab-Microsynth, Göttingen, Germany). For expression, all Vav constructs were C-terminally fused to citrine and cloned into pMSCVpuro (Clontech, Takara Bio Inc, Kyoto, Japan). Rac isoforms were expressed using the MIGRII vector containing an IRES-EGFP cassette. Retroviral transductions were done with the packaging cell line Plat-E. Transduced cells were selected in the presence of 2 μg/mL puromycin for 5-7 days followed by expression analysis by flow cytometry and western blotting.

Biochemical assays, antibodies, and reagents
For BCR stimulation, cells were incubated with 20 μg/mL goat antihuman IgM F(ab') 2 fragments (Jackson ImmunoResearch, West Grove, PA, USA) at 37°C in RPMI without FBS for the indicated times. Preparation of cellular lysates in 1% NP40-containing lysis buffer for affinity purifications and western blot analyses was done as described [6]. Briefly, the cells were lysed with lysis buffer composed of 50 mM Tris-HCl (pH 7.8), 137 mM NaCl, 0.5 mM EDTA, 1 mM sodium orthovanadate, 10% v/v glycerol, 1% v/v NP40, and a protease inhibitor cocktail containing AEBSF, Aprotinin, Bestatin, E-64, Leupeptin and EDTA (Sigma Aldrich, #P2714). Lysis was performed on ice for 10 min and the cell debris was pelleted at 20 000 × g at 4°C for 10 min. The supernatant containing the cell lysate was mixed with reducing SDS-PAGE sample buffer, boiled at 95°C for 5 min and analyzed by SDS-PAGE and immunoblotting. For affinity purification of proteins from B-cell lysates using biotinylated peptides or GST fusion proteins, cell lysates were prepared as described above from 3 × 10 7 Ramos cells and incubated with 2 μM peptides or ß20 μg GST fusion proteins bound to glutathione-sepharose (GE Healthcare). Biotinylated peptides and associated binding partners were purified using streptavidin-sepharose beads (GE Healthcare, Chicago, IL, USA) under gentle rotation at 4°C for 2 h. For immunopurification of PLCγ2, 2 μg of rabbit anti-PLCγ2 antibodies (#Q-20, Santa Cruz Biotechnology, Santa Cruz, CA, USA) were added to the lysate of 5 × 10 7 cells followed by incubation on a rotator for 4 h at 4°C. Subsequently, 25 μL protein A/G PLUS agarose beads (Santa Cruz Biotechnology) were added, followed by additional incubation for 1 h on a rotator at 4°C. All affinity-purification samples were repeatedly washed in lysis buffer, before they were subjected to SDS-PAGE and immunoblotting. Antibodies for immunopurification and immunoblotting: the monoclonal antibodies against β-Actin (clone 13E5), SLP65/BLNK (clone D8R3G), Vav1 (clone D45G3), phospho-p38 MAPK (Thr180/Tyr182, clone 3D7), phosphotyrosine (clone 100) and the polyclonal anti-p38 MAP kinase antibody, were from Cell Signaling Technology (Danvers, MA, USA). The monoclonal antibodies to Vav2 (clone EP1067Y), CD79A (Igα, clone EP3618), and to CD79B (Igβ, clone EPR6861) were from Abcam (Cambridge, UK). The polyclonal anti-PLCɣ2 antibody (Q-20), anti-Rac2 (C-11), and the monoclonal anti-Syk antibody (clone 4D10) were from Santa Cruz Biotechnology. Rabbit polyclonal anti-Vav3 was from Merck Millipore (Burlington, MA, USA) and the polyclonal anti-GST antibody was from Molecular Probes (Eugene, OR, USA). Monoclonal anti-Rac1 (ARC03) was from Cytoskeleton Inc. (Denver, CO, USA). Biotinylated (phospho-) peptides encompassing the intracellular regions of Igα or Igβ were synthesized by CASLO ApS (Lyngby, Denmark) or by Eurogentec (Liège, Belgium). Immunoblot images were processed using Photoshop CS4 and Corel Draw software. Quantification of band intensities was done with LabImage 1D software (Kapelan Bio-Imaging, Leipzig, Germany). The inhibitors EHop-016, Rhosin, and Latrunculin A were from Calbiochem (Merck Millipore).

Measurement of intracellular free Ca 2+
The day before the measurement, ß1 × 10 6 cells were seeded in a 10-cm culture dish. The next day, cells were loaded under gentle mixing for 30 min at 30°C with 1 μM Indo-1-AM (Invitrogen, Thermo Fisher Scientific) in RPMI containing 10% fetal bovine serum and 0.015% Pluronic-F-127 (Invitrogen). Subsequently, cells were washed twice and resuspended in Kreb's-Ringer solution composed of 10 mM HEPES (pH 7.0), 140 mM NaCl, 10 mM glucose, 4 mM KCl, 1 mM MgCl 2 , and 1 mM CaCl 2 . After 30 min of resting prior to the measurement, the fluorescence ratio of Ca 2+ -bound Indo-1 (405 nm) to Ca 2+ unbound Indo-1 (530 nm) was monitored on an LSR II cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). The basal Indo-1-AM ratio was monitored for 25 s, followed by stimulation with either 20 μg/mL goat anti-human IgM F(ab ) 2 (Jackson ImmunoResearch) or 10 μg/mL anti-human CD8 (clone MEM-31, a gift of Dr. Vaclav Horejsi, Prague, Czech Republic). Data were analyzed using FlowJo (FlowJo LLC, Ashland, OR, USA), Microsoft Excel, and GraphPad Prism. A forward scatter/side scatter gate as shown in Supporting information Fig.  S11 was applied in each analysis to exclude cellular debris.

GEF activity assay
The constructs of Vav family members used for the guanine nucleotide exchange assay were expressed with an N-terminal His 10 -tag in E.coli BL21-CodonPlus cells. Proteins were purified with HisPur Ni-NTA Superflow Agarose (Thermo Fisher Scientific). Guanine nucleotide exchange assays were carried out using Bodipy-GDP. A total of 400 nM of small G proteins (purchased from Cytoskeleton Inc., Denver, CO, USA) were loaded with 300 nM Bodipy-GDP (Invitrogen, Thermo Fisher Scientific) in 20 mM Tris (pH7.5), 50 mM NaCl, 4 mM EDTA, 100 μg/mL BSA and 500 μM DTT for 20 min at 22°C. Loading was locked by addition of MgCl 2 to a final concentration of 5 mM. To enable the exchange of fluorescent GDP for nonfluorescent GTP, 500 nM GTP and 800 nM of purified Vav protein fragments were added. The fluorescence of Bodipy-GDP-bound small G proteins was measured every 30 s for 20 min at 20°C using a Cytation3-Imager (BioTek Instruments, Inc, Winooski, VT, USA). The used Vav protein fragments were as follows: human Vav1: amino acids 170-575 carrying a Y174D substitution, human Vav1-AR: 189-575, human Vav3: 169-573 (Y173D), murine Vav1: 170-575 (Y174D). Additional mutations are specified in the respective figure legends.

Isolation of primary human B cells
Primary human B cells from peripheral blood of healthy donors were isolated by magnetic cell sorting using the B-cell isolation kit II (Myltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer´s instructions. The purity of the isolated B cells was tested by surface staining of CD19 (anti-CD19-PE clone LT19, Myltenyi Biotec) and IgM (anti-human IgM-AF647, Southern-Biotech, Birmingham, AL, USA), and was typically around 90% or higher. Experiments involving human participants were approved by the ethical review committee of the University Medical Center Göttingen and were performed in accordance with relevant guidelines and regulations. An informed consent was obtained from all participants.