Vav Links the T Cell Antigen Receptor to the Actin Cytoskeleton and T Cell Activation Independently of Intrinsic Guanine Nucleotide Exchange Activity

Background T cell receptor (TCR) engagement leads to formation of signaling microclusters and induction of rapid and dynamic changes in the actin cytoskeleton, although the exact mechanism by which the TCR initiates actin polymerization is incompletely understood. The Vav family of guanine nucleotide exchange factors (GEF) has been implicated in generation of TCR signals and immune synapse formation, however, it is currently not known if Vav's GEF activity is required in T cell activation by the TCR in general, and in actin polymerization downstream of the TCR in particular. Methodology/Principal Findings Here, we report that Vav1 assembles into signaling microclusters at TCR contact sites and is critical for TCR-initiated actin polymerization. Surprisingly, Vav1 functions in TCR signaling and Ca++ mobilization via a mechanism that does not appear to strictly depend on the intrinsic GEF activity. Conclusions/Significance We propose here a model in which Vav functions primarily as a tyrosine phosphorylated linker-protein for TCR activation of T cells. Our results indicate that, contrary to expectations based on previously published studies including from our own laboratory, pharmacological inhibition of Vav1's intrinsic GEF activity may not be an effective strategy for T cell-directed immunosuppressive therapy.


Introduction
In developing and mature T cells, the T cell receptor (TCR) activates Src family kinases that phosphorylate immunoreceptor tyrosine-based activation motifs (ITAMs) in CD3 and TCRf proteins, providing docking sites for Syk/ZAP-70 family kinases. Subsequently, the recruitment of the adaptors LAT, GADS, and SLP-76, and enzymes such as Tec family kinases, phosphoinositol-3 kinase (PI3K), and phospholipase Cc1 (PLCc1), leads to the generation of the secondary signaling intermediates, 1,4,5-inositol triphosphate (IP 3 ) and diacylglycerol (DAG), activating intracellular Ca ++ and mitogen-activated protein kinases (MAPK) (reviewed in [1,2]). Together, these events promote the transcription of genes involved in T cell proliferation and differentiation. The engagement of the TCR also leads to rapid and dynamic changes in the T cell actin cytoskeleton that can be visualized by imaging F-actin. In a model of TCR stimulation on a planar surface, F-actin is induced at TCR-surface contact sites, but then spreads circumferentially to the cell periphery driving plasma membrane extensions such as filopodia and lamellipodia [3]. In addition, recent live cell imaging studies using total internal reflection fluorescence microscopy (TIRFM) in combination with stimulatory antibodies or planar bilayers containing peptide:MHC complexes revealed the formation of microclusters of signaling proteins including TCRf, CD3, ZAP-70, SLP-76 and Vav, suggesting that these structures could be the sites of signal generation [4,5,6,7,8,9]. Nevertheless, while the importance of the actin cytoskeleton in lymphocytes has been appreciated for over 30 years, the exact mechanism(s) by which the TCR initiates actin polymerization remains incompletely understood [10].
Several models have been proposed for TCR-initiated actin polymerization (reviewed in [10,11,12,13]). While most studies point to the involvement of WASp/WAVE proteins as the downstream effectors, important differences exist in the proposed mechanisms regarding how the TCR is linked to actin assembly. For example, one model suggests that CD3 chains directly recruit an Nck-WASp complex via Nck SH3 binding to proline-rich sequences in CD3 [14], providing an explanation of how F-actin induction could occur at the TCR independently of ITAM phosphorylation. However, the preponderance of evidence indicates that tyrosine phosphorylation and the recruitment of ZAP-70, SLP-76, and LAT are required for TCR initiation of Factin assembly, and recent studies suggest that microclusters of these signaling proteins (also termed proto-synapses) can recruit WASp to sites of TCR contacts [6,15,16].
In this context, WASp/WAVE-mediated nucleation of actin filaments, through their interaction with the Arp2/3 complex, can be induced by Nck binding independently of Rho GTPases [17,18]. Alternatively, WASp/WAVE activation can be mediated by Rho GTPases, such as Rac1 and Cdc42, which are activated by guanine nucleotide exchange factors (GEF), including Vav, aPIX, bPIX, and DOCK2 [19,20,21,22,23,24]. Vav has been implicated in T cell cytoskeletal regulation based on its Dbl-homology (DH) domain, tyrosine phosphorylation, and recruitment to T cell-APC contacts (reviewed in [25]), although recent studies indicated the importance of Vav in integrin activation and T cell-APC conjugate formation, rather than in F-actin assembly [26,27]. Thus, while Vav1 also regulates ERM [28] and MTOC polarization [26], no conclusive evidence exists, to date, in support of an essential role for Vav proteins in the TCR initiation of actin polymerization. In this regard, because studies of T cells lacking all three Vav proteins revealed redundancy of Vav1 with other Vavs [29], direct examination of TCR-induced actin polymerization in Vav1/2/3-deficient (Vav NULL ) T cells should conclusively establish whether or not the Vav family is essential in this process. While Vav is considered a Rho GEF, it is unknown if the intrinsic GEF activity is indeed required for Vav function downstream of the TCR. In this context, disruption of TCR-induced Ca ++ and MAPK signaling in T cells lacking all Vav proteins (Vav NULL ) suggests that Vav may function downstream of the TCR as a critical linker rather than exclusively as a Rho GEF [29]. Consistent with such a view, GEF-inactivated Vav has been shown to augment NFAT-dependent transcriptional activation in Jurkat T cells [30]. In addition, Vav contains several tyrosine residues that may be involved in direct binding of SH2 domaincontaining proteins [9,31,32]. Thus, it is possible that Vav mediates TCR signals independently of its intrinsic GEF activity, however this remains to be tested in T cells lacking all endogenous Vav proteins.
In this report, we address these unresolved issues. Using live-cell imaging, we show that Vav forms signaling microclusters at TCR contact sites, similar to other TCR linker proteins, and demonstrate that the Vav family is critical for TCR initiation of actin polymerization. Surprisingly, the intrinsic GEF activity is dispensable for Vav function in TCR signaling and mobilization of intracellular Ca ++ fluxes. Here, we propose a model for Vav as a critical linker in TCR-induced activation of T cells.

Results
Vav proteins are essential for the initiation of actin polymerization at the TCR In view of the functional redundancy of Vav proteins, we decided to examine if the Vav family is required in TCR-initiated actin polymerization using Vav NULL T cells lacking all 3 Vav proteins [29]. To this end, we first analyzed WT T cells by confocal imaging of F-actin structures at the plane of cell contact with the stimulatory coverslip, visible by DIC microscopy, and then in increments along the Z-axis (Fig. 1A) [3,4,6]. Initially, the cell-contact sites appeared round and did not show significant Factin content beyond a small ring along the circumference of the cell contact. Subsequently, within 2-5 minutes, WT T cells showed dramatic F-actin accumulation throughout the region of coverslip contact and formed filopodia and lamellipodia stretching beyond the circumference of the F-actin ring (Fig. 1A, and data not shown). This process continued for approximately 10 minutes, at which time the cell perimeter (Fig. 1B) and F-actin content (Fig. 1C) reached their maximum. We next analyzed Vav1deficient (Vav1 2/2 ) T cells and found that cell spreading and the induction of F-actin structures were delayed relative to WT (Fig.  S1), indicating that Vav1 regulates but is not essential for TCRinduced actin polymerization in this system. In sharp contrast to WT or Vav1 2/2 T cells, F-actin production and cell spreading of Vav NULL T cells was virtually blocked (Fig. 1A,B,C), resembling non-stimulated cells at all of the time points studied (Fig. 1A). These results show that the Vav family is critical for the initiation of TCR-induced actin polymerization and T cell spreading. Thus, together with the involvement of Vav1 in signaling microclusters [9], these data indicate that Vav may function as a critical linker for TCR-initiated actin polymerization, raising the question of whether or not the intrinsic GEF activity is necessary for its function in this process.

GEF-inactive Vav1 participates in signaling microclusters and restores TCR function in J.Vav cells and Vav1deficient T lymphocytes
Live-cell imaging studies of T cell-planar surface contacts revealed microclusters of signaling proteins that included ZAP-70, LAT, SLP-76, Nck, Grb2, and WASp, which have been implicated in the initiation of T cell activation and actin polymerization at the sites of TCR contacts [4,5,6,7,33]. Since Vav1 has been implicated in T cell cytoskeleton regulation, we decided to examine its dynamic redistribution in live T cells. To this end, we generated Vav1-deficient Jurkat cells [34] that express Vav1-GFP (J.Vav1 WT ) at the level of endogenous Vav1 in the WT parental Jurkat line [9] (Fig. 2). Such cells were analyzed using stimulatory coverslips and real-time total internal reflection fluorescence microscopy (TIRFM), allowing visualization of Vav1-GFP in the direct vicinity (100-200 nm) of plasma membrane-coverslip contacts. Consistent with our recent report, Vav1-GFP quickly assembled (within 5-10 seconds of initial contact) into microclusters at the cell-coverslip interface ( Fig. 2A) [9]. Notably, kymographic analyses of microcluster fluorescence intensity over time, indicate that Vav1-GFP microclusters are stable ( Fig. 2B-D), and Vav1 showed little, if any, lateral diffusion as indicated by laser-bleaching (data not shown). Control experiments using J.Vav cells expressing GFP-only (GFP), or J.Vav1 WT cells incubated on coverslips with irrelevant antibody or poly-L-lysine showed no significant microcluster formation ( Fig.  S2 and data not shown). To extend these initial observations, we used confocal imaging and found that TCR-induced Vav1-GFP microclusters colocalized with SLP-76 microclusters ( Fig. 2F and Fig. S3). Thus, given that the redistribution pattern of Vav was reminiscent of other signaling molecules implicated in microcluster formation [4], and that Vav colocalized with SLP-76, these data suggest that Vav could be involved at the sites of initial TCRinduced actin polymerization, which is consistent with our finding that Vav is required for generation of F-actin and cell spreading (Fig. 1).
To determine if the intrinsic GEF activity of Vav1 is required for its function in TCR signaling, we first generated J.Vav cells expressing Vav1 protein with a previously characterized GEF loss-of-function mutation L278Q (corresponding to L213Q in onco-Vav), fused to GFP (J.Vav1 GEF2 ) ( [9,22,35,36,37] and Fig.  S4). We first examined such J.Vav1 GEF2 cells by TIRFM, as in experiments described in Fig. 2, and found that, similar to Vav1 WT , Vav1 GEF2 generated stable microclusters at the T cellstimulatory coverslip interface ( Fig. 3A-D). Moreover, similar to Vav1 WT , Vav1 GEF2 microclusters colocalized with TCRinduced SLP-76 microclusters (Fig. S3). In addition, tyrosine phosphorylation and SLP-76 binding of Vav1 GEF2 in response to TCR stimulation showed no discernible differences from Vav1 WT (Fig. 3E). Thus, neither the pattern of Vav1 redistribution, nor its tyrosine phosphorylation and SLP-76 binding, appear to be affected by the loss of intrinsic GEF activity ( Fig. 3A-E).
To determine if Vav GEF activity is required for TCR induction of NFAT and NFkB, we used J.Vav1 WT and J.Vav1 GEF2 cells transfected with NFAT or NFkB luciferase reporter-gene constructs and analyzed luciferase activity upon stimulation with anti-CD3 antibodies (Fig. 3F). As expected, such treatment led to a strong induction of both NFAT-and NFkBdependent luciferase activity in J.Vav1 WT T cells. Notably, J.Vav1 GEF2 cells showed no statistically significant differences in activity in this assay as compared to J.Vav1 WT (Fig. 3F) and responded similarly to PMA and Ionomycin (Fig. S5). These experiments suggest that, even in the absence of endogenous Vav1, a GEF-inactive Vav1 is capable of rescuing TCR-induced NFAT-and NFkB-dependent transcriptional activation. These observations are consistent with previous studies showing GEFindependent effects of Vav in this pathway [30]. Strikingly, however, the same GEF-inactivating mutation completely abolished the ability of Vav to activate NADPH-oxidase in myeloid cells ([9,36,37], and our unpublished observations). Thus, it appears that in contrast to the TCR signaling pathway, in myeloid cells Vav GEF activity is critical for its function in regulating the NADPH oxidase complex.
Since signaling properties of Jurkat T cells differ in some aspects from those of primary T cells, for example due to PTEN deficiency, we decided to examine the requirement for Vav1 GEF activity in primary T lymphocytes. In this regard, while anti-CD3or superantigen SEE-induced proliferation of Vav1 2/2 T lymphocytes was diminished, as expected based on previously published studies [38,39,40], expression of retrovirally-encoded Vav1 GEF2 protein in primary Vav1 2/2 T lymphocytes restored their proliferative capacity, as compared to Vav1 WT T cells ( Fig. 3G). In addition, TCR-mediated Ca ++ signaling, which is defective in Vav1 2/2 T cells, was restored in Vav1 GEF2 cells (data not shown).
Taken together, these results suggest that the intrinsic GEF activity is dispensable for Vav1 function in J.Vav cells and in Vav1 2/2 T cells. However, because neither J.Vav cells nor Vav1 2/2 T lymphocytes show appreciable defects in TCRinduced actin polymerization ( Fig. S1 and data not shown), we reasoned that the requirement for GEF activity must be conclusively addressed in T cells in the Vav NULL background.

Expression of Vav1 GEF2 restores T cell development in Vav NULL mice
To address the requirement of Vav GEF activity, without the complicating issue of compensatory effects of endogenous Vav proteins, we decided to generate T cells that express Vav1 GEF2 in  the absence of any other Vav protein. In this regard, we first examined if Vav1 GEF2 protein could, by itself, support Vav NULL T cell development. To this end, we developed a Vav NULLhematopoietic stem cell complementation (HSCC) approach and, as a validation of this approach, showed that Vav1 WT -GFP expression rescued Vav NULL T cell development (Fig. 4A,B). Thus, while Vav NULL mice showed severely reduced populations of both developing and mature T cells [29], Vav1 WT chimera mice developed populations of thymocytes and peripheral T lymphocytes similar to WT mice (Fig. 4A,B), although the total number of thymocytes generated in such RAG-chimera is typically somewhat lower, as compared to WT (Fig. 4A and data not shown). Thus, having established that the introduction of Vav1 WT rescues development of Vav NULL T cells, we next examined the effects of Vav1 GEF2 in this same assay. Strikingly, both numbers and percentages of thymocytes and peripheral T cell subsets in Vav1 WT and Vav1 GEF2 mice were similar (Fig. 4A,B). Importantly, the levels of expression of Vav1 WT and Vav1 GEF2 proteins were virtually equal to that of endogenous Vav1 (Fig. 4C). Also, similar to Vav1 WT , a majority of Vav1 GEF2 thymocytes and peripheral T cells were GFP + (Fig. 4A,B), and these GFP+ cells contained the mutated Vav1 GEF2 , as confirmed by direct sequencing of genomic DNA from purified peripheral T cells (data not shown). Together, these results show that GEF-inactive Vav1 is capable of restoring development of T cells lacking all endogenous Vav family proteins. We conclude from these experiments that Vav GEF activity is not essential in T cell development.

Expression of Vav1 GEF2 rescues Vav NULL T cell proliferation and cytokine production
Although the Vav family is necessary for T cell proliferative responses [29,38,39,40], the requirement for Vav GEF activity is not known. To address this issue, Vav1 WT and Vav1 GEF2 T cells generated by Vav NULL -HSCC were stimulated with anti-CD3 antibodies, in the presence or absence of anti-CD28 antibodies, and proliferation was measured by 3 H-thymidine incorporation (Fig. 5A). While, as we have previously shown, Vav NULL T cells showed essentially no proliferation in this assay [29], surprisingly, Vav1 GEF2 T cells showed a robust response that was similar to Vav1 WT at all concentrations of stimulatory antibodies tested ( Fig. 5A and data not shown). As an alternative measure of T cell proliferation, we used CFSE dye-dilution assays, which also showed comparable proliferative responses of Vav1 GEF2 and Vav1 WT T cells (Fig. 5B). Moreover, analyses of T cell proliferation induced by superantigen SEE-pulsed APCs showed similar results (Fig. 5C,D), indicating that the intrinsic Vav GEF activity is not required for T cell proliferation.
Since Vav1-deficiency has been shown to impair generation of effector T cells and cytokine production with deficient IL-4 expression and enhanced Th1 development [41], we examined if Vav1 GEF activity may be essential in this process. To this end, purified naïve CD4 + CD62L hi T cells from Vav1 WT or Vav1 GEF2 mice were stimulated under Th1 or Th2 polarizing conditions and then assayed for IFNc or IL-4 production. Results of these experiments showed similar cytokine production profiles of Vav1 WT and Vav1 GEF2 T cells (Fig. 5E). Taken together, these experiments indicate that while Vav proteins are essential for the induction of T cell proliferative responses, the intrinsic GEF activity appears dispensable for Vav function in T cells. Of note, while previous reports indicated involvement of Vav GEF activity in CD28 signaling (reviewed in [42]), our results suggest that there may also exist a GEF-independent mechanism for Vav-mediated CD28 co-stimulation.
Expression of Vav1 GEF2 rescues defects in TCR signaling, actin cytoskeleton remodeling, Rac1 activation, and MTOC polarization Our previous studies showed defects in TCR-induced Ca ++ and Ras/MAPK signaling in Vav NULL T cells [29], however it is not known if the intrinsic Vav GEF activity is required in these processes. To address this issue, we examined Ca ++ mobilization in response to TCR stimulation in Vav1 WT and Vav1 GEF2 T cells and found that both types of cells showed a similar response (Fig. 5F). Similarly, activation of Erk-1/2 appeared normal in both Vav1 WT and Vav1 GEF2 T cells (Fig. 5G). These results indicate that although the activation of Ca ++ and Erk signaling downstream of the TCR requires Vav [29], it does not depend on the intrinsic Vav GEF activity. In this context, in accord with reports of a defect in TCR activation of Rac1 in Vav1 2/2 T cells, [43,44], we also found defective TCR-induced Rac1 activation in Vav NULL T cells and a modest reduction in Rac1 activation in J.Vav cells (Fig. 6). Given the disruption of Ca ++ and MAPK signaling in Vav NULL T cells, we reasoned that defective Rac activation in these cells likely results from the loss of Vav linker function. Consistent with this view, the induction of activated Rac1 in TCRstimulated Vav1 WT and Vav1 GEF-T cells was similar, as was that in J.Vav WT compared to J.Vav GEF2 (Fig. 6), indicating that Vav GEF activity is dispensable for TCR induction of Rac1.
To examine if the GEF activity of Vav is essential for TCRinduced actin polymerization, Vav1 WT or Vav1 GEF2 T cells were incubated on stimulatory coverslips, and F-actin structures were visualized as in Fig. 1. While Vav NULL T cells completely failed to spread and form lamellipodia or filopodia following TCR stimulation (Fig. 1, Table 1), both Vav1 WT and Vav1 GEF2 T cells showed robust actin polymerization and spreading, virtually indistinguishable from that of WT T cells (Fig. 7A, Table 1). These data indicate that while Vav proteins are indispensable for TCR-induced F-actin remodeling (Fig. 1), the intrinsic GEF activity does not appear to be required in this process.
Since a recent report implicated Vav in TCR-induced MTOC polarization [26], we examined the requirement for Vav GEF activity in this process. These experiments showed that while MTOC polarization in Vav NULL T cells was reduced essentially to background levels, as compared to WT (Fig. 7B, Table 2), MTOC polarization in Vav1 WT and Vav1 GEF2 cells was similar to WT (Fig. 7C, Table 2). Collectively, these data indicate that while T cells require Vav proteins for TCR signaling and cytoskeletal regulation, Vav GEF activity appears dispensable. Therefore, Vav appears to mediate TCR signals as a critical linker protein rather than as a bona fide Rho GEF.

Discussion
Stimulation of T cells with anti-CD3 antibodies immobilized on a planar surface permits analyses of the initial formation and the stability of TCR-induced signaling microclusters, or protosynapses, in live cells [3,4,5,6,7,33]. In this regard, following contact of a T cell with a stimulatory surface, ZAP-70, SLP-76, LAT, GADS, and Grb2 are quickly incorporated into signaling microclusters [4,5,33]. Here, we report that Vav1 rapidly assembles into TCR-induced microclusters, and remains stable and lacks lateral motion. A recent adaptation to visualizing the dynamic redistribution of TCR-induced microclusters involves stimulation of T cells with cognate TCR ligands embedded in fluid lipid bilayers instead of immobilized anti-CD3. Although this approach allows engaged TCRs to diffuse freely throughout the T cell membrane and to coalesce at the cSMAC within the immunological synapse [7,8], unlike immobilized anti-CD3 stimulation, data generated using either approach indicates that signaling microclusters form at early time points following TCR stimulation and are relevant sites of TCR signaling initiation and maintenance.
Vav1 may interact with the TCR/CD3-complex in several different ways, including via direct interaction with TCRf chains [45] or by binding to ZAP-70 or SLP-76 [46,47]. In this regard, together with the observation that Vav1 is rapidly recruited to signaling microclusters at the initial sites of actin polymerization, similar to other essential linkers such as LAT, these results indicate that Vav itself may function as a linker in TCR-induced actin polymerization, independently of its other potential role as a Rho GTPase activator. Thus, given that Vav recruitment to ZAP-70, SLP-76, or LAT is dependent upon tyrosine phosphorylation [47,48], our results support a model in which TCR-induced actin polymerization is initiated in the context of phosphorylated ITAMs. Consistent with this view, our analyses indicate that Vav colocalizes with other linkers, such as SLP-76, in TCRinduced microclusters that rapidly form at TCR contacts.
Although Vav1-deficient T lymphocytes and J.Vav cells show defects in TCR signaling, surprisingly little or no evidence exists in support of the requirement for Vav1 for TCR-induced actin polymerization. In this regard, two main issues appear to have precluded significant inroads. First, the functional redundancy of Vav proteins, all of which are expressed in T cells, produces compensatory effects in cells lacking individual family members. Second, in studies with T cell-APC conjugates or with other systems involving an immune synapse, Vav-dependent signals emanating from integrins and/or costimulatory molecules are difficult to discriminate from the TCR-specific signals that may depend upon Vav. Therefore, in this report we examined the requirement for the entire Vav family in actin reorganization using Vav NULL T cells and anti-CD3 stimulation on a planar surface and find a virtually complete disruption in actin polymerization, which is the first such direct demonstration. Strikingly, these defects are rescued by expression of GEF-inactive Vav. In this regard, several potential scenarios could explain the lack of requirement for the intrinsic Vav GEF activity. For example, a previously described SLP-76-Nck-WASp complex may control actin reorganization independent of Rho-protein involvement [6,15,16]. Alternatively, a recently described Dynamin2 function in TCR-induced actin polymerization could contribute Vavdependent, but GEF activity-independent, effects [49].
While the requirement for Vav SH2, SH3, CH, and PH domains for Vav function in TCR signaling is well established [43,47,50,51], the requirement for the GEF activity remains controversial [30,34,50,52,53]. In this regard, the truncated form of Vav1 (with constitutive GEF activity) does not enhance NFATdependent transcription [30,47], suggesting that Vav GEF activity is not sufficient to propagate signals leading to NFAT. However, while several reports indicated that GEF activity of Vav1 may be required in TCR-induced NFAT-and NFkB-mediated transcriptional activation [34,50,52,53], another study showed that Vav1 GEF activity is not required in enhancing NFAT-dependent transcription [30]. Several potential explanations exist for these apparent discrepancies. For example, the effects of Vav1 may vary depending on relative levels of protein expression, as transient expression of Vav1 in Jurkat cells was shown to potently stimulate NFAT-dependent signaling, even in the absence of TCR engagement [30,47]. In this context, overexpression of GEFinactive Vav1 could conceivably result in dominant negative effects on gene transcription [53]. Alternatively, ectopic expression of GEF-inactive Vav1 could exert positive effects on downstream signaling pathways, for example via a mechanism involving transcomplementation of the missing GEF activity by the activity of endogenous (GEF-sufficient) Vav1 protein [30]. Here, we show that GEF-inactive Vav1, expressed stably at endogenous levels in J.Vav cells, rescues TCR-induced NFAT-and NFkB-dependent transcription. In this regard, we used a previously characterized L278Q loss-of-function mutant [22,35] and verified the loss of catalytic activity by comprehensive analyses of GDP/GTP exchange in vitro and by in vivo assays for F-actin induction by the N-terminally truncated Vav (Fig. S4 and [9,36]). However, these experiments do not rule out the possibility that the GEF activity of other Vav proteins, Vav2 and/or Vav3, both of which are expressed in J.Vav cells, may contribute compensatory effects  to Vav1 GEF2 -mediated signals. To address this issue, we used Vav NULL T cells that lack all three endogenous Vav proteins. We note, however, that other non-Vav GEFs could also be responsible for contributing compensatory activity, such as bPIX, which is activated in response to TCR stimulation in J.Vav cells [54]. While mice lacking individual Vav family proteins show partial to no defects in T lymphocytes, Vav NULL mice show a severe block in T cell development [29,38,39,40,55]. We reasoned, given our earlier observation that Vav1 +/2 /Vav2 2/2 /Vav3 2/2 mice (which express only Vav1 but not Vav2 or Vav3) show no discernible defects in T-lineage cells [29], that reintroduction of Vav1 alone should be sufficient to rescue Vav NULL T cell development and function. Indeed, we found that the expression of Vav1 WT is capable of completely restoring development and activation of Vav NULL T cells. The levels of expression of recombinant Vav1 in these ''rescued'' T cells closely approximate that of endogenous Vav1 in WT T cells (Fig. 4C), a finding that is notable because in T cells generated by the Vav NULL -HSCC assay, Vav1 expression is not controlled by the endogenous promoter elements but rather by retroviral-based LTRs. Thus, these data indicate that one or more mechanisms may regulate Vav1 expression in T cells, or possibly, this could be due to a developmental advantage of T cell progenitors that express a certain level of Vav1. While at present we do not completely understand how the levels of Vav1 expression may be regulated in T cells, expression of either WT or GEF-inactive Vav1, at levels indistinguishable from endogenous Vav1, can support T cell development.
Consistent with recent studies implicating Vav1 in control of microtubular reorganization [26], Vav NULL T cells show disrupted MTOC polarization (Fig. 7, Table 2). While this function of Vav could, conceivably, require the GEF activity for activation of GTPases such as Cdc42 that can modulate MTOC polarization [56], analyses of Vav1 GEF2 T cells suggest that Vav effects on MTOC polarization are Vav GEF-independent. However, while Vav appears to be essential for both TCR-mediated regulation of MTOC polarization and actin polymerization, any GEF activity(ies) required in these processes could be controlled by other effectors such as aPIX, bPIX, DOCK2, or DOCK180, or the RhoA effectors p160ROCK and p190RhoGEF [19,20,21,23,24,57]. However, recent studies clearly show that WASp/WAVE-mediated actin polymerization can be induced by the Arp2/3 complex independently of Rho GTPases, for example via binding of Nck [3,6]. In this context, the rescue of TCR-induced F-actin defects in Vav NULL T cells by GEFinactive Vav expression indicates that the intrinsic Vav GEF activity is not essential for actin polymerization downstream of the TCR. These data suggest that Vav functions as a TCRproximal linker critical for cytoskeletal reorganization that could be Rho GTPase-independent. Interestingly, similar to Vav, the Rac-GEF kalirin induces lamellipodia formation independently of its intrinsic GEF activity [58] suggesting that regulation of actin dynamics by some GEFs may not require the catalytic activity of the DH domain.
Alternatively, however, Rac activation downstream of the TCR may be mediated by other Rho-GEFs, such as aPIX, bPIX, or DOCK2. Indeed, T cells deficient in DOCK2 show defective Rac1 activation by the TCR, but unlike Vav NULL T cells, show no defects in Ca ++ or MAPK signaling [24], indicating distinct mechanisms for regulation of Rac and Ras GTPases downstream of the TCR. In this regard, the function of Vav appears to be as a TCR linker required for both Rac and Ras signaling. Thus, taken together, the reduction of specific catalytic activity of the Vav GEF-mutant used in our study to essentially undetectable levels (less than 1% of wild type), combined with no evidence for any local increases in the concentration of the mutant protein, as judged by TIRFM analyses of activated T cells, and no evidence of any titratable differences in the ability of Vav1 GEF2 T cells to respond to TCR stimulation, provide compelling evidence that defects in TCR signaling (including Ca ++ , MAPK, and Rac1 activation), actin polymerization, MTOC polarization and proliferation of Vav NULL T cells are due to the loss of adaptor/ linker, rather than GEF, function of Vav. Consistent with effective reduction of the GEF activity of the Vav1 GEF2 constructs, combining the GEF-killing mutation with the GEF-activating mutation (Y174F) completely abolished the effects of the latter [9]. Of note, we obtained similar results using another GEF-inactive form of Vav (Vav1 E201A/K335A ) [59,60,61] (data not shown). Thus, the preponderance of evidence presented in our report indicates to us that a scenario in which any residual GEF activity would account for the rescue of T cell function by the Vav1 L278Q mutant is unlikely. Moreover, recent reports demonstrate that expression of the same Vav1 GEF2 mutant (L278Q) in Vavdeficient myeloid cells does not rescue LPS-or FccR-triggered oxidative burst [36,37], indicating that in these cells the intrinsic GEF activity of Vav is essential for its function, in contrast to TCR-induced signaling.
We also note that because mice congenitally lacking Vav1 show primarily T-lineage specific defects [62], one could reason that the intrinsic GEF-activity of Vav1 could be an attractive potential target for pharmacological inhibition in the context of T celldirected immunosuppressive therapies. However, our data presented here suggest that the inhibition of the Vav1 enzymatic activity as a GEF would likely not be an effective strategy for suppressing T cell activation and proliferative expansion.
While Vav proteins also contain a PH domain, implicated in PIP 2 and PIP 3 binding and the regulation of Vav plasma membrane interactions as well as GEF activity [25], a recent study showed that a mutation rendering Vav1 PH domain incapable of binding to phosphatidylinositol metabolites leads to TCR signaling defects [43]. Thus, given the results of our studies presented in this report, it is possible that the PH domain could contribute to Vav function in TCR signaling independently of its effects on GEF activity.
Vav has been among the first phosphotyrosine-proteins identified in TCR signaling pathways [63,64] and indeed, tyrosine phosphorylation distinguishes the Vav family from a plethora of other Dbl proteins. While Vav tyrosine phosphorylation has mainly been considered in the context of the regulation of the intrinsic GEF activity [31,65], our data presented in this report suggest that tyrosine phosphorylation of Vav could also contribute to its function as a TCR linker for activated T cells. In this regard, we propose that Vav mediates TCR signals in a GEF-independent manner.
TIRFM Imaging. Imaging of dynamic Vav1-GFP microcluster assembly and movement was performed using TIRF microscopy as described in [9,66]. Image recording and processing were performed using AQUACOSMOS software (Hamamatsu Photonics, Japan) and image analyses were performed using Metamorph Software (Molecular Devices Corp., Sunnyvale, CA). Kymographic analysis was performed as in [9]. See Supplemental Methods S1 for more extensive descriptions.

Actin Polymerization and MTOC Polarization
T cells were purified from LN cell suspensions by removal of B cells with anti-Ig-coated Dynabeads (Invitrogen, Carlsbad, CA) using standard procedures. T cells were resuspended in plain DMEM and incubated on anti-CD3e-coated coverslips (clone 145-2C11, 1 mg/mL, BD Biosciences) for the indicated time points. Actin polymerization was visualized by staining of F-actin with Alexa-Fluor-488-phalloidin (Molecular Probes, Eugene, OR). MTOC polarization was performed as previously described [67]. MTOCs were visualized by staining with fluorescein (FITC)-antia-tubulin (Sigma, St. Louis, MO). Confocal and differential interference contrast (DIC) images were taken using Zeiss LSM510 confocal system and analyzed by ImageJ software and LSM Image Browser software.

Luciferase Assays
Cells were transfected with 5 mg luciferase plasmid containing NFATx3 binding sites from the IL-2 promoter, or NFkBx2 binding sites from the IFNb promoter. Sixteen hours following transfection, cells were either left unstimulated or stimulated with anti-CD3+anti-IgG2a for 6 hours. Luciferase assays were then performed according to manufacturer's instructions (Promega, Madison, WI).

Mice, Cell Suspensions, Antibodies, and Flow Cytometry
Germline Vav1 2/2 and Vav NULL mice have been previously described [29,55] and were maintained in the SPF facility of Washington University School of Medicine according to institutional protocols. Cell suspensions were prepared, counted, and stained with antibodies following standard procedures. The following antibody conjugates were used (BD Biosciences): phycoerythrin (PE) and allophyocyanin (APC)-H129.19 (anti-CD4) and cytochrome C (CyC)-53-6.7 (anti-CD8a). All samples were analyzed on a FACSCalibur flow cytometer (Becton Dickinson) with FlowJo software.

T Cell Polarization and Analysis of Cytokine Production
Naïve CD4 + CD62L + LN T cells FACS sorted from fresh LN were activated and polarized to Th1 or Th2 as previously described in [68]. For ELISA, resting cells were stimulated with anti-CD3 antibodies for 24 hrs. Cytokine concentrations were measured in culture supernatants using Cytometric Bead Array (BD Biosciences) according to manufacturer's instructions.
Rac assay. Purified LN T cells were starved for 30 mins in media lacking serum. Cells were treated with 1 mg/mL anti-CD3 antibodies for 2 mins and Rac assay performed using EZ-Detect Rac1 Activation Kit (Pierce, Rockford, IL) according to manufacture's instructions.
Purification of GST-Rac1 and MBP-Vav1, and guanine nucleotide exchange assays Bacterially expressed GST-Rac1 was purified as previously described in [69]. MBP-Vav1 fusion proteins were expressed in E. coli strain BL21(DE) followed by purification using amylose resin according to the manufacture's protocol (NEB, Beverley, MA), with the exception that the column was washed with 20 mM Tris, pH 7.4, 200 mM NaCl after binding the protein to the resin. The MBP-fusion proteins were eluted with the same buffer containing 10 mM maltose. Exchange assays were performed essentially as described in [70,71].

Statistical Analysis
Data are expressed throughout as mean+standard deviation. Data sets derived from the indicated genotypes were compared using the two-tailed unpaired Student's t-test. Differences were considered statistically significant when p,0.05.