Tyrosine Phosphorylation of Vav Stimulates IL-6 Production in Mast Cells by a Rac/c-Jun N-Terminal Kinase-Dependent Pathway

The proto-oncogene vav is primarily expressed in cells of hemopoietic origin (1). The role of Vav in hemopoietic cell development and signaling is unclear, although its presence is required for normal T and B cell development and thymic selection (2, 3, 4). In contrast, normal development of cells of erythroid, myeloid, and mast cell lineages occurs in the absence of Vav expression (3). Several studies demonstrated that Vav is a guanine nucleotide exchange factor (GEF)³ whose activity is directed toward Rac1 (5, 6). The activation of Rac1 by Vav requires the tyrosine phosphorylation of Vav (5, 7). Furthermore, activation of Rac/Cdc42/Rho can lead to the activation of c-Jun N terminal (JNK)/stress-activated protein kinase, an event that may be important in the transcription of specific genes (8, 9, 10).

Tyrosine phosphorylation of Vav in response to engagement of the high affinity receptor for IgE (FcεRI) was reported (11). We previously demonstrated that a fraction of Vav associates with the FcεRI γ-chain immunoreceptor tyrosine-based activation motif (12). A more recent study using a chimeric Tac-FcεRI γ chain construct demonstrated a link between the FcεRI γ-chain and Vav-dependent activation of JNK1 (7). Other studies demonstrated the interaction of the Vav SH2 domain with phosphorylated tyrosines between the C-terminal SH2 domain and the catalytic domain of Syk (13). However, little is known about what FcεRI-dependent mast cell response(s) may be regulated by Vav-dependent activation of JNK1. Early studies in T cells demonstrated that Vav overexpression resulted in increased NF-AT activity and induced the response of an IL-2 reporter construct in T cells (14). Later studies showed that Vav and the adapter molecule SLP-76 were synergistically coupled in IL-2 production (15). Thus, in the present study we explored whether Vav could also modulate cytokine production in mast cells.

What signaling pathways regulate cytokine production in mast cells is not clearly understood. An apparent species or developmental difference may exist, which has led to reports of JNK activity as important to TNF-α production in MC/9 murine mast cells (16), while ERK2 regulates this cytokine in RBL cells (17). In addition the activation of JNK in mast cells can be mediated by PKC-dependent and -independent pathways, with the latter being initiated by two distinct effectors, namely SEK1 and MKK7 (18). In the present study we took advantage of the observation that overexpression of Vav in the RBL cell line resulted in its tyrosine phosphorylation in the absence of additional stimuli to analyze its role in the absence of the activation of multiple signaling pathways by FcεRI engagement. We also tested the effects of FcεRI engagement on Vav-initiated signals. Finally, we examined whether the Vav-mediated induction of IL-6 observed in RBL cells could also be observed in nonimmortalized bone marrow-derived mast cells (BMMC).

The enzymes used in these studies were purchased from either New England Biolabs (Beverly, MA) or Life Technologies (Gaithersburg, MD). Deoxyadenosine 5′-α-[³⁵S]thiotrisphosphate, deoxyadenosine 5′-α-[³²P]trisphosphate, l-[³⁵S]cysteine, and enhanced chemiluminescence reagents were purchased from Amersham (Arlington Heights, IL). The PRISM Ready Reaction DyeDeoxy Terminator Cycle Sequencing Kit was purchased from Applied Biosystems (Foster City, CA). Piceatannol was purchased from Calbiochem (San Diego, CA). All other chemicals were purchased from Sigma (St. Louis, MO). Anti-dinitrophenyl-specific murine IgE was obtained and purified as previously described (19, 20), and dinitrophenylated (DNP)-human serum albumin (HSA) was purchased from Sigma. A rabbit polyclonal Ab and a mouse mAb to Vav were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and Upstate Biotechnology (Lake Placid, NY), respectively. Goat and rabbit polyclonal Abs to JNK1 were also purchased from Santa Cruz Biotechnology. A mouse mAb to phosphorylated JNK was purchased from New England Biolabs (Beverly, MA). Rabbit IgG was purchased from Jackson ImmunoResearch Laboratories (West Point, PA). Rac1 N17 and Rac1 V12 were gifts from Marc Symons (Onyx Pharmaceuticals, Richmond, CA). Ras N17 and Ras V12 were provided by Deborah Morrison (National Cancer Institute, National Institutes of Health). The nonactivatable JNK1 (APF) and the wild-type JNK1 were described previously (21). The previously described NF-AT (7x) (22) luciferase reporter was provided by Dr. K.-I. Arai (Institute of Medical Sciences, University of Tokyo, Tokyo, Japan). Catalytically inactive Btk (K430R) was a gift from T. Kawakami (La Jolla Institute of Allergy and Immunology, La Jolla, CA).

A cDNA plasmid library of RBL-2H3 cells was custom prepared by Clontech (Palo Alto, CA) according to a modified Gubler and Hoffman procedure (23). This cDNA library was generated from methylmercuric hydroxide-denatured mRNA, which was 5′-stretched, and oligo(dT) and random primed. The library generated was directly cloned into the BstXI site of the plasmid vector pCDM8. The probe specific for Vav was generated by PCR from a cDNA library of RBL-2H3 made by RT-PCR of the mRNA using oligo(dT) priming. Primers used in the amplification reaction were of the nucleotide sequence of Vav conserved between human and mouse. The 5′- primer was 5′-GTGAGAAGTTCGGCCTCAAGC-3′, and the 3′-primer was 5′-ACCGAGGTGAAGAACAACAGAGC-3′, which generated a fragment of 1160 bp starting at bp 261 and ending at bp 1419 of the mouse sequence with 98% identity. A 600-bp fragment of the 5′ end of the PCR-derived rat Vav fragment was used as a probe for screening the cDNA library. Five independent clones were isolated, the longest of which was 4.1 kb. All clones were partially sequenced and were identical for the regions sequenced in the open reading frame (ORF). One clone was sequenced to completion and was found to be missing 28 bp at the 5′ end of the ORF compared with the mouse sequence. Analysis of the remaining clones revealed the absence of an ATG that would result in an ORF for a 95-kDa product of Vav. To obtain the 5′ sequence, a rapid amplification of 5′-cDNA ends was performed on an RT-PCR single-stranded cDNA library using a 3′ clone-specific oligonucleotide and a 5′ random priming oligonucleotide. The product obtained was subcloned into the sequenced clone, generating a clone of 2858 bp. The nucleotide sequence was determined and showed a complete ORF and a 5′-untranslated sequence of 56 bp.⁴ The translation of the ORF (with 94% identity to murine Vav) resulted in a protein of 843 aa with a net charge of −9, a predicted isoelectric point of 6.51, and a predicted molecular mass of ∼98 kDa. Expression of the rat Vav cDNA encoding the ORF in CHO cells resulted in expression of a protein with an apparent molecular mass of 105 kDa.

Sequencing was performed using the PRISM Ready Reaction DyeDeoxy Terminator Cycle Sequencing Kit and the 373A DNA Sequencing System from Perkin-Elmer/Applied Biosystems (Foster City, CA) as described by the manufacturer. Plus and minus strands were sequenced with a minimum of two sequence reactions per primer. Analysis of the acquired data was performed using the GCG Sequence Analysis Package and Sequencer (Genetics Computer Group, Madison, WI).

The rat basophilic leukemia cell line (RBL-2H3) was cultured essentially as previously described (24). The hamster kidney (BHK) and Chinese hamster ovary (CHO) cell lines was cultured as described by American Type Culture Collection (Manassas, VA). Transfection of plasmids carrying the complete cDNA of Vav, the mutant cDNA of Vav, NF-AT-luc, and the IL-6 promoter-luc were in CHO or RBL-2H3. Stable transfectants of Vav were previously described (25), using the εMTH vector (26). For transient protein expression we used a viral expression system based on the Semliki Forest virus (SFV) and described briefly below and in detail previously (27).

Vav, Rac1 N17, Rac1 V12, Ras N17, Ras V12, JNK1(APF), and JNK1 were subcloned in the pSFV1 or pSFV1-green fluorescent protein (GFP) vectors, which were modified as previously described (27). The production of recombinant SFV infectious particles and the determination of titer were performed according to the procedure previously described (28). For transient protein expression by SFV infection, adherent cells (RBL cells) were plated in six-well plates or in 100- × 20-mm tissue culture dishes for RT-PCR or in-gel kinase assays, respectively. Adherent cells were washed, infected by addition of activated virus in serum-free medium in the presence of 7.5% polyethylene glycol, and incubated at 37°C for 30 min as previously described (27). Nonadherent cells (BMMC) were recovered by centrifugation of exponential growth cultures at 1000 rpm for 10 min, washed once in PBS, and resuspended in serum-free medium containing activated virus followed by rapid addition of polyethylene glycol. Following incubation of the nonadherent cells with activated virus as described above, the cells were washed once in serum-free medium to remove polyethylene glycol and excess virus, and growth medium was added. Cells were allowed to incubate for 4 h at 37°C, and expression of the recombinant proteins was determined using a FACScan cytometer (when tagged with GFP) or by Western blot.

Transfected cells grown in culture dishes and sensitized with IgE were rid of unbound IgE by two washes with growth medium. Subsequently, growth medium containing DNP-HSA (100 ng/ml) was added, and cells were further incubated for 8 min. After stimulation, DNP-caproic acid (25 μM/ml) in ice-cold PBS (pH 7.4) was added to stop further receptor engagement. Cells were then harvested with a cell lifter (Costar, Cambridge, MA) and lysed with 1% Nonidet P-40 lysis buffer as previously described (12). Soluble lysates was equally divided in three for immunoprecipitations with goat anti-JNK1, rabbit anti-p38 MAPK, and rabbit anti-ERK2 Abs as previously described (12). For detection of phosphorylated JNK1 the recovered proteins were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and probed with Ab specific to phosphorylated JNK1.

In-gel kinase assays were performed essentially as described by Kameshita et al. (29) with a minor modification to the kinase buffer. Briefly, GST-ATF2 (300 μg/ml) was used as a specific substrate for JNK1 or p38 MAPK, and MBP (0.5 mg/ml) was used as a substrate for ERK2. After denaturation of the gel with 6 M guanidine HCl and a 16-h renaturation with a buffer containing 50 mM Tris-HCl (pH 8.0), 5 mM 2-ME, and 0.04% Tween 40 at 4°C, the gel was equilibrated with 20 ml of kinase buffer containing 40 mM HEPES (pH 7.5), 2 mM DTT, 0.1 mM EDTA, and 5 mM MgCl₂ for 30 min at 30°C. The kinase reaction was conducted for 1 h at 30°C with 20 ml of fresh kinase buffer containing, in addition to the above, 10 μM cold ATP and 50 μCi of [γ-³²P]ATP. The gels were then washed extensively with washing buffer (5% TCA plus 1% sodium pyrophosphate) to remove unincorporated [γ-³²P]ATP before drying and exposure to x-ray film.

A semiquantitative RT-PCR (25) was used to determine the expression of cytokines in cells expressing different constructs. Briefly, transfected cells were washed twice in PBS and solubilized with Tri-Reagent (Molecular Research Center, Cincinnati, OH), and total RNA was then isolated as described in the provided protocol. First-strand cDNA synthesis was performed using reverse transcriptase, 2.5 μg of total RNA, and poly(dT) primer. The cDNA generated was then amplified by PCR using conditions where amplification remained linear and primer sets previously described (25).

Luciferase reporter assays were previously described (22, 30) with minor modification. Briefly, 36 h posttransfected cells were IgE sensitized and serum deprived overnight before engagement of FcεRI. Luciferase activity was measured with the dual luciferase detection kit from Promega (Madison, WI). The response of 2 × 10⁶ cells was assayed in each sample and was normalized to either the response observed for a control luciferase vector or determined protein concentrations. For IL-6 secretion assays, IgE-sensitized and GFP-, Vav-GFP-, or Vav (DH-)-GFP-expressing BMMC cells (1.0 × 10⁷ cells) were grown for 4 h post-SFV infection, and expressing cells were recovered by FACSorter. The observed efficiency of viral infection for BMMC ranged from 15 to 40%. The highest infection efficiency was observed for cultures 4–5 wk of age. The sorted cells were recovered, washed once in growth medium, resuspended in the same containing DNP-HSA (100 ng/ml), and further incubated for 1 h at 37°C. The IL-6 secreted in the medium was measured by an ELISA kit purchased from Endogen (Woburn, MA) as previously described (31).

Overexpression of the Vav cDNA in RBL-2H3 stable transfectants gave 4- to 8-fold greater levels of Vav in these cells (data not shown). Analysis of seven experiments of the tyrosine phosphorylation profile of total proteins in the Vav-stable transfectants revealed no significant change compared with control cells, except that Vav was heavily tyrosine phosphorylated (Fig. 1^,A). The level of tyrosine phosphorylation of Vav was similar to that observed by FcεRI engagement of the vector control transfectant (Fig. 1^,B). The ratio of phosphorylated to unphosphorylated Vav was the same in control and Vav-transfected cells, with ∼5% of the total Vav phosphorylated in either transfectant. SFV transient overexpression of Vav showed a similar increase in the amount of tyrosine-phosphorylated Vav, and the levels of Vav expression were up to 10-fold greater than the levels of endogenous Vav (27). To determine whether the basal activity of Syk might lead to Vav phosphorylation, we immunoprecipitated Vav from 2.5 × 10⁷ vector- or Vav-transfected cells either untreated or treated with genistein or piceatannol. The phosphorylation of Vav was inhibited by the Syk-selective inhibitor, piceatannol (32), but not by the Src-selective inhibitor, genistein (Fig. 1 C), thus demonstrating a basal Syk activity sufficient for Vav phosphorylation. This was further confirmed in other studies where Vav phosphorylation was not seen in Syk-deficient cells (33).⁵ In addition, we explored whether Btk activity could contribute to Vav phosphorylation in control and Vav-overexpressing cells. Expression of a catalytically inactive Btk (K430R) had no effect on Vav phosphorylation in the absence or the presence of FcεRI engagement (data not shown). Because we previously demonstrated that low levels of Syk activity are found in resting RBL-2H3 cells (34), we conclude that the increased fraction of phosphorylated Vav results from the presence of increased substrate for Syk.

The tyrosine phosphorylation of Vav has been demonstrated to be critical for its Rac1-directed GEF activity (5). Thus, the observation of constitutive phosphorylation of Vav by overexpression in the RBL cells in the absence of FcεRI engagement provided a suitable system to investigate the consequences of Vav activation in mast cell responses. Because Vav has been described to promote IL-2 production in T cells (14, 15) we investigated, by RT-PCR, the profile of cytokine mRNA. We found increased IL-2 and IL-6 mRNA levels in Vav-overexpressing cells, but little (GM-CSF) or no effect on IL-3, IL-4, TNF-α, and TGF-β mRNA levels (Fig. 2). In most experiments the increased levels of IL-2 and IL-6 mRNA were similar to the levels observed in response to FcεRI engagement, although a slight enhancement was observed upon FcεRI engagement (Fig. 2). Because the levels of IL-2 mRNA in RBL cells were not easily detectable, further efforts were focused on investigating the molecular mechanisms regulating the IL-6 response in these cells.

Tyrosine-phosphorylated Vav was shown to activate JNK1 in response to an external stimulus (7). Furthermore, JNK activity has been implicated in mast cell cytokine production (16). Therefore, we analyzed whether the constitutive phosphorylation of Vav induced by its overexpression in RBL cells could directly result in increased JNK1 activity or whether FcεRI engagement is a prerequisite. Fig. 3,A shows that transient Vav overexpression led to activation of JNK1 with little effect on p38 MAPK and ERK2 activities. In the control transfectant (where only GFP is expressed) no stimulation of JNK1 activity was observed. Engagement of the FcεRI in the control (GFP) transfectant resulted in a greater than 4-fold stimulation of JNK1 activity, while in Vav-transfected cells <2-fold enhancement of the JNK1 activity was observed, suggesting that almost complete activation of the kinase is achieved by Vav overexpression alone (Fig. 3,A). Engagement of the FcεRI potently activated both p38 MAPK and ERK2 activities in the GFP control and Vav-transfected cells (Fig. 3 A).

Because products of the PI-3 kinase pathway have been implicated in the activation of Vav (35) we explored whether the constitutive activation of JNK1 by Vav was dependent on PI-3 kinase activity. We treated GFP control and Vav-transfected cells with 100 nM wortmannin, a concentration that inhibits PI 3-kinase activity in these cells (36) and has been demonstrated to inhibit JNK activation (16, 37, 38). We analyzed JNK1 phosphorylation before and after FcεRI engagement, since we found, by kinetic analysis, that the level of JNK1 phosphorylation was a direct reflection of its activity (J. Song and J. Rivera, unpublished observation). As shown in Fig. 3 B, wortmannin inhibited (∼40–50%) the activation of JNK1 in both the control and Vav-transfected cells. Treatment of the transfected cells with 10–300 nM wortmannin gave similar results, although inhibition ranged from 25 to 50%. These results are consistent with prior studies demonstrating a 60% inhibition of JNK activation in RBL cells after treatment with 100 nM wortmannin (38). Wortmannin had little effect on the basal JNK1 activity of the GFP control transfectant before FcεRI engagement. However, the extent of inhibition for GFP control and Vav-transfected cells was the same following FcεRI engagement. Therefore, basal and stimulated PI-3 kinase activities are required for the complete activation of JNK1 by Vav, but Vav overexpression does not overcome the inhibitory effect of wortmannin, demonstrating that PI-3 kinase is upstream of Vav.

Because Vav overexpression activated JNK1 we investigated the role of Rac and Ras in this constitutive activation by coexpression of the inactive forms of Rac1 (N17) and Ras (N17) with Vav. Coexpression of Ras N17 with Vav or with the control GFP-transfected cells had no effect on JNK1 activity (Fig. 4, lanes 1 and 2 vs 3 and 4). In fact, a slight enhancement of JNK1 activity was observed in several experiments, including the representative experiment shown. In contrast, coexpression of Rac1 N17 with Vav or with the GFP control led to the marked inhibition of JNK1 activity, with almost complete inhibition of Vav-induced JNK1 activity in some experiments (Fig. 4, lanes 1 and 2 vs 5 and 6). This suggested the possibility that the active form of Rac1 (Rac1 V12) would mimic the Vav-mediated induction of JNK1. Expression of Rac1 V12 with GFP resulted in the activation of JNK1, and expression of Rac1 V12 in the presence of Vav enhanced the Vav-induced JNK activity (Fig. 4, lanes 1 and 2 vs 7 and 8). These results demonstrated that Vav-mediated activation of JNK1 is Rac1-dependent.

To assess whether the Vav-dependent activation of JNK1 was a direct result of Vav activation of Rac1, we tested whether a mutant Vav could activate JNK1 in the RBL-2H3 cells. The Dbl homology (DH) domain of Vav is a guanine nucleotide exchange domain with specificity to Rac1. Deletion of this domain should inhibit JNK1 activation, if Rac1 activity is required for JNK1 activation (8) in these cells. As shown in Fig. 5,A, the DH-Vav did not show a dominant-negative phenotype, as it did not inhibit JNK1 activation below the level of the GFP control (lanes 1 and 3 vs 7 and 9). However, overexpression of DH-Vav did not result in constitutive activation of JNK1 (lane 7) to the levels that result from overexpression of wild-type Vav (lane 4). This suggested that Vav guanine nucleotide exchange activity is important for JNK1 activation. Thus, one might expect that coexpression of wild-type Vav with the inactive JNK1 (APF), expressing a dominant-negative phenotype, should inhibit the Vav-dependent activation of JNK1 as demonstrated in Fig. 5,A (lane 5). Furthermore, expression of wild-type JNK enhanced the JNK1 activity of all transfectants (Fig. 5 A, lanes 3, 6, and 9). These results showed that the guanine nucleotide exchange activity of Vav mediates the activation of JNK1 in these cells.

To determine whether the constitutive cytokine response observed in Vav-overexpressing cells was a direct result of JNK1 activation we studied the IL-6 mRNA response in cells expressing the DH-Vav and the mutant JNK1 (APF). Because PCR amplification and the rapid plateauing of the IL-6 mRNA response could mask any apparent differences, we chose to work in the early phase of the IL-6 mRNA response where the rate increased exponentially and also used limited cycles of PCR. As shown in Fig. 5,B, overexpression of the DH-Vav led to minimal increases in the IL-6 mRNA (lane 7) above that seen in the GFP control transfectant (lane 1). Thus, as noted above, the inability of the DH-Vav to effectively activate JNK1 also led to a poor IL-6 mRNA response. In contrast, coexpression of wild-type JNK with either the control GFP or DH-Vav gave a 2-fold IL-6 mRNA response (Fig. 5,B, lane 3 vs 9). Coexpression of wild-type Vav with wild-type JNK caused a 3-fold enhancement of the IL-6 mRNA response (Fig. 5,B, lane 6). Collectively, the findings demonstrated a link of JNK1 activity to IL-6 mRNA accumulation. Thus, as might be expected, coexpression of GFP, wild-type Vav, and DH-Vav with the mutant JNK1 (APF) effectively inhibited the Vav-induced IL-6 mRNA accumulation (Fig. 5 B, lanes 2, 5, and 8). These results demonstrated the requirement of JNK1 activity in the Vav induction of IL-6.

Vav-induced NF-AT activity was reported in T cells and correlated with the induction of IL-2 in these cells (14). NF-AT activity in response to FcεRI engagement of mast cells has also been demonstrated (39, 40). We explored whether Vav overexpression induced the constitutive activation of NF-AT using an NF-AT-luciferase reporter construct (22). We used a suboptimal dose of Ag (1 ng) to assess whether any additional FcεRI-mediated effect of Vav on NF-AT activity might be mediated by Vav. As shown in Fig. 6, we observed a significant basal NF-AT activity in the control transfected cells. Serum deprivation had no significant effect on the basal NF-AT activity in these cells and had only a minor effect on the stimulated response (data not shown). FcεRI engagement caused a 2-fold enhancement of NF-AT activity in the control transfectant (Fig. 6, Vector/+Agn). Vav overexpression caused a constitutive activation of NF-AT activity that ranged from 4- to 8-fold enhancement of the basal activity (Fig. 6, Vav/-Agn). In addition, FcεRI engagement caused a small, but consistent, increase in the Vav-induced NF-AT response (Fig. 6, Vav/+Agn). A comparison of the FcεRI-stimulated control (vector/+Agn) and Vav-transfected (Vav/+Agn) cell NF-AT response showed a 2- to 4-fold increase in NF-AT activity. Collectively, our results demonstrated the Vav-mediated constitutive activation of NF-AT activity (Fig. 6) and suggest that Vav may contribute to NF-AT activation by more than one signaling pathway, because FcεRI stimulation enhanced Vav-mediated NF-AT activation.

To determine whether Vav enhanced IL-6 production in a nonimmortalized mast cell, we studied the FcεRI-dependent secretion of IL-6 in BMMC-transfected with wild-type and DH-Vav. For reasons that are unclear, RBL cells did not secrete detectable levels of IL-6 (above the background of the available assay), although the protein was present in these cells (data not shown). Because transient transfection or viral infection of BMMC is inefficient, cells were sorted using the GFP tag of the expressed protein to isolate the expressing cell population. Before stimulation both GFP- and Vav-transfected cells secreted very low levels of IL-6 in the medium, with the latter secreting 30–40% more IL-6 under resting conditions. Kinetic analysis of IL-6 secretion from nontransfected BMMC showed that maximum secretion occurred at 4 h post-FcεRI engagement. Thus, we chose to work at 1 h post-FcεRI engagement because the IL-6 levels were easily detected, and secretion was at a maximal rate. Using these conditions we found that the overexpression of Vav resulted in an FcεRI-stimulated increase of as much as 90% in secreted IL-6 compared with the overexpression of GFP alone (Fig. 7). The GFP control transfectants secreted 550–750 pg/ml/10⁶ cells, while Vav transfectants secreted 1.0–1.4 ng/ml/10⁶ cells. Furthermore, the increase in secreted IL-6 required the GEF activity of Vav, as DH-Vav failed to enhance the levels of secreted IL-6 (Fig. 7). Thus, we conclude that Vav induction of IL-6 mRNA results in increased protein production that is secreted upon FcεRI engagement.

We used a simple model for the study of Vav function in mast cells, based on the finding that overexpression of Vav in RBL cells leads to its phosphorylation and activation in the absence of FcεRI engagement. This model demonstrated a direct link between Vav activation and IL-6 induction in mast cells (see Fig. 8). Of particular importance is the finding that Vav phosphorylation is both necessary and sufficient for increased IL-6 mRNA in the absence of ERK2 and p38 MAPK activation. In addition Ras V12, which enhanced ERK2 activity in RBL cells (data not shown), failed to induce an IL-6 response, while Rac1 V12, which activated JNK1, caused an increase in IL-6 transcripts. This demonstrates that Vav has the potential to induce IL-6 expression in a Rac1/JNK1-dependent and Ras/ERK2-independent manner. Nevertheless, following FcεRI engagement other signaling pathways appear to synergize to activate JNK1, because only a partial inhibition of JNK1 activity by Rac1 N17 or JNK1 (APF) was observed under these conditions. It is possible that this synergy is mediated by the Ras pathway, because we previously found that components of the Ras pathway coimmunoprecipitated with Vav (12), and in kinetic studies we presently found that Vav overexpression enhances the rate of ERK2 activation after FcεRI engagement by Agn (data not shown). This enhancement was most pronounced when a suboptimal concentration (1 ng) of Agn was used. These results suggest that Vav GEF activity may not directly activate the Ras pathway, as previously suggested (41), but at suboptimal concentrations of Agn Vav may increase the rate of ERK2 activation by facilitating the formation of a Vav-containing plasma membrane macromolecular complex (see Footnote 5) that contributes to Ras and Rac signaling (12, 42, 43).

Vav expression primarily affected the message levels of IL-2 and IL-6 without a significant effect on IL-3, IL-4, TNF-α, or TGF-β and only a slight effect on GM-CSF. This specificity is rather remarkable, because the Vav-dependent increase in NF-AT activity that we found in RBL cells should also affect other cytokine promoters (44). To the best of our knowledge an NF-AT binding site in the IL-6 promoter has not been described, although it is present in the IL-2 (22) and IL-4 (44) promoters. Vav induction of NF-κB activity has been reported (45), and an NF-κB binding site is present in the IL-6 promoter (30). In preliminary experiments we could not detect induction of NF-κB-luciferase reporter activity in response to Vav overexpression in RBL cells. However, since a previous study demonstrated an NF-κB-like activity (not comprised of the p50 or Rel proteins) in RBL cells that is important for TNF-α gene expression (46), it is possible that our reporter construct may not be activated by this NF-κB-like activity, or the promoter insert did not contain the appropriate binding site. Nevertheless, because we observed no significant Vav-mediated induction of IL-4, GM-CSF, or TNF-α (genes known to be regulated by NF-AT), this would suggest that the increased NF-AT, c-Jun (47), or NF-κB/NF-κB-like (45, 46) activities cannot solely explain the specific accumulation of IL-2 and IL-6. Thus, one possible explanation would be that other unidentified transcription factors are induced by Vav that provide the required complex for specific gene activation. The recent findings (48) of two distinct NF-ATc isoforms (α and β) in mast cells whose individual expression is either constitutive (α) or inducible (β; by FcεRI engagement) would support the aforementioned hypothesis if Vav activity is also capable of inducing early response genes.

Recent studies by Y. Kawakami et al. (49) showed that mast cells from btk-null mice failed to activate JNK to the levels of normal mast cells. Interestingly, these mice were also defective in the production of cytokines, and the cytokine responses could be reconstituted by transfection with Btk (31). Given that a similar phenotype was observed with overexpression of Vav, DH-Vav, and downstream effectors, this suggested a possible connection between Btk and Vav. However, we did not find evidence for a direct link of Btk activity to Vav function under overexpression conditions that led to activated JNK1. In fact, when FcεRI was engaged on RBL cells expressing the inactive Btk (K430R), only a minimal effect on Vav phosphorylation was observed. Neverthe-less, Btk and Vav may communicate. For example, Btk and Vav activities are regulated by binding, to their pleckstrin homology domains, of components of the phosphoinositide pathway (35, 50, 51, 52, 53). Our finding that wortmannin, at concentrations that effectively inhibit PI-3 kinase (36), partially inhibited Vav-induced JNK1 activity is consistent with prior results (38) and suggests that Vav activity is at least partly responsible for JNK1 activation. A more complete JNK1 inhibition by the same concentration of wortmannin was observed in prior studies on MC/9 cells, suggesting that the primary pathway of JNK1 activation in these cells is PI-3 kinase dependent (16, 37). It is possible that the sustained activation of Vav may be dependent on Btk activation and its regulation of other effectors, such as phospholipase Cγ (52), because in SLP-76 (an adapter molecule for Vav)-deficient T cells, phosphorylation of phospholipase Cγ and the calcium response is inhibited (54), and the latter is also seen in Vav-null T cells (55, 56). Alternatively, Btk and Vav may be part of a molecular signaling complex, where, in the absence of one protein, the function of the others may be affected (54, 57, 58). Regardless, the present study is most consistent with prior studies supporting the critical role for Syk activity in Vav phosphorylation and activation (33, 59). In addition, we recently found that the presence of Syk is critical not only to Vav phosphorylation but also to Vav compartmentation and function (see Footnote 5).

The overexpression of the inactive JNK1 (APF), which competes with wild-type JNK1 (TPY) as a nonphosphorylatable substrate, also demonstrated the inhibition of Vav-induced JNK1 activity and of the IL-6 mRNA response, thus establishing the link of (pY)Vav→Rac1→JNK1→IL-6 (Fig. 8). Given the specificity observed for IL-2 and IL-6 mRNA responses, our findings underscore published studies that suggest that no single signaling pathway is capable of inducing a general mast cell cytokine response (16, 17, 25, 31, 48). It is clear, however, that Vav-dependent activation of JNK1 is sufficient for the induction of an IL-6 response in RBL and BMMC. Thus, one might speculate that because Vav-dependent induction of IL-6 may be relatively direct, IL-6 production could occur in circumstances where multiple and converging signaling pathways might not be active (47, 60, 61, 62).

Recent studies clearly demonstrate that Vav regulates cytoskeletal reorganization by the TCR (55, 56). In these studies, vav^−/− T cells were found to be defective in TCR-mediated actin cap formation and in IL-2 production. Surprisingly, these studies also demonstrated that JNK/SAPK activity is normal in vav^−/− T cells, a finding that would not be predicted from previous (7, 8, 9, 10) and present studies. These differences may be explained by the presence of redundant pathways or the up-regulation of compensatory mechanisms, such as the possible increased expression or activation of another GEF activity such as Vav2 (63), which in the absence of Vav may target an alternate pathway leading to JNK activation (18). Regardless, the present study demonstrates the potential of Vav to activate JNK1 activity via the activation of Rac1 (Fig. 8). We also found that IL-6 mRNA accumulation required the GEF activity of Vav, could be induced by Rac, and required activation of JNK1. Finally, because Vav, but not DH-Vav, expression in BMMC promoted increased FcεRI-dependent IL-6 secretion, we conclude that Vav GEF activity contributes to the production of IL-6 in mast cells. Collectively, these findings support the idea that activation of a particular signaling pathway can have specific consequences. However, the cross-talk of signaling pathways in response to receptor engagement is likely to modulate the kinetics, extent, and profile of cellular responses.

We thank Drs. K.-I. Arai, T. Kawakami, D. Morrison, and M. Symons for providing reagents. We also acknowledge the contributions of Dr. V. Parravicini and Ms. C. Friedman in preparing some of the viral plasmids used in this study.