The Scaffolding Adapter Gab2, via Shp-2, Regulates Kit-evoked Mast Cell Proliferation by Activating the Rac/JNK Pathway*

The scaffolding adapter Gab2 mediates cell signaling and responses evoked by various extracellular stimuli including several growth factors. Kit, the receptor for stem cell factor (SCF), plays a critical role in the proliferation and differentiation of a variety of cell types, including mast cells. Kit, via Tyr567 and Tyr719, activates Src family kinases (SFK) and PI3K respectively, which converge on the activation of a Rac/JNK pathway required for mast cell proliferation. However, how Kit Tyr567 signals to Rac/JNK is not well understood. By analyzing Gab2–/– mast cells, we find that Gab2 is required for SCF-evoked proliferation, activation of Rac/JNK, and Ras. Upon Kit activation in wild-type mast cells, Gab2 becomes tyrosyl-phosphorylated and associates with Kit and Shp-2. Tyr567, an SFK binding site in Kit, and SFK activity were required for Gab2 tyrosyl phosphorylation and association with Shp-2. By re-expressing Gab2 or a Gab2 mutant that cannot bind Shp-2 in Gab2–/– mast cells or acutely by deleting Shp-2 in mast cells, we found that Gab2 requires Shp-2 for SCF-evoked Rac/JNK, Ras activation, and mast cell proliferation. Lastly, by analyzing mast cells from mice with compound Gab2 and Kit Y719F mutations (i.e., Gab2–/–: KitY719F/Y719F mice), we find that Gab2, acting in a parallel pathway to PI3K from Kit Tyr719, regulates mast cell proliferation and development in specific tissues. Our data show that Gab2 via Shp-2 is critical for transmitting signals from Kit Tyr567 to activate the Rac/JNK pathway controlling mast cell proliferation, which likely contributes to mast cell development in specific tissues.

Mast cells are the major effector cells for IgE-dependent allergic responses, and also are important for resistance to parasitic and bacterial infections (1). The progenitors of mast cells originate in the bone marrow (BM), 5 circulate in peripheral blood, and differentiate in various tissues (1). Stem cell factor (SCF) and its receptor Kit are essential for the mast cell development in vivo, as illustrated by the phenotype of mice with null mutations in the Kit (W) or SCF (Sl) loci. Such mice lack tissue mast cells, and also have defects in melanogenesis, gametogenesis, and hematopoiesis (2).
Kit belongs to a subfamily of receptor-tyrosine kinases (RTK) that include the receptors for colony-stimulating factor and platelet-derived growth factor. Ex vivo studies of mast cells or heterologous cells expressing Kit wild type (WT) or various mutants indicate that Kit activates multiple downstream pathways, which together are important for the proliferation, survival, migration, and adhesion of mast cells (3). Upon SCF binding, Kit becomes dimerized and activated. Activated Kit transphosphorylates tyrosine (Y) residues in the receptor cytoplasmic tail, creating docking sites for various SH2 domaincontaining signaling molecules. Among the various tyrosine residues in the cytoplasmic tail of Kit, Tyr 719 , and Tyr 567 are critical for transmitting Kit signals (4,5). Recent studies of Kit "knock-in" mice indicate that Tyr 719 and Tyr 567 contribute to mast cell development. However, Tyr 567 is more important than Tyr 719 for the mast cell development in specific tissue such as peritoneal cavity (6). Tyr 719 , a site for recruiting the p85 regulatory subunit of PI3K (7), is required for SCF-evoked PI3K activation (8,9). Tyr 567 is reported to recruit various signaling molecules including Src family kinases (SFKs) (4, 10), Shp-2 (11), and Shc (12). Tyr 567 is important for SCF-evoked SFK (4) (13,14) and Ras activation (14,15). However, the role of Shp-2 recruitment to Tyr 567 is less clear. Importantly, activation of SFK from Tyr 567 and activation of PI3K from Tyr 719 both contribute to proliferation and survival of mast cells in vitro. SFK and PI3K can both activate the Rac/JNK pathway that promotes SCF-evoked mast cell proliferation (4). It is not well understood what signaling pathway activated by SFK and PI3K promotes SCF-evoked mast cell survival. One report suggested that Rac2 promotes survival by activating Akt and suppressing the expression of pro-apoptotic protein Bad (16). PI3K can activate Rac through Rac GEF (17). PI3K via activation of Rac also mediates SCF-evoked mast cell migration (18). However, it is not clear what molecule mediates the activation of Rac by SFK from Kit Tyr 567 .
Gab2 (Grb2-associated binder-2) is a member of Gab/Dos subfamily of scaffolding adapters that also include mammalian Gab1 and Gab3, Drosophila DOS (daughter of sevenless), and Caenorhabditis elegans Soc-1 (19). Like other Gab/DOS family members, Gab2 contains an N-terminal pleckstrin homology (PH) domain, several proline-rich motifs (PXXP), and multiple tyrosine phosphorylation sites (19). Gab2 PH domain preferentially binds PI 3,4,5-P 3 (PIP3) (20). Two of the proline-rich motifs in Gab2 are Grb2-SH3 domain binding sites (21), and are important for coupling Gab2 to upstream receptors through the Shc⅐Grb2 complex (22). Gab2 plays an important role in transmitting signals downstream of receptors for several cytokines and growth factors as well as multichain immune receptors. Upon receptor activation, Gab2 becomes tyrosyl-phosphorylated and recruits SH2 domain containing signal relay molecules, including the tyrosine phosphatase Shp-2 and p85, the regulatory subunit of Class 1A PI3K. Gab2 association with Shp-2 is important for cytokine induced immediate early gene expression (23) and growth factor-induced Erk activation (24 -27). Genetic and biochemical evidence indicate that Shp-2 via Gab/Dos is required for activation of the Ras/Erk pathway (28). However, the critical Shp-2 substrate in controlling Ras/Erk activation is still not clear. Gab2 association with p85 is critical for cytokine, Fc-receptor, and growth factor-evoked PI3K activation (19,29).
The in vivo functions of Gab2 have been elucidated through the analysis of Gab2 knock-out (Ϫ/Ϫ) mice. Gab2 is essential for allergic responses and is reportedly important for RANKmediated osteoclastogenesis (30,31). In addition, we and others have found that Gab2 Ϫ/Ϫ mice have selective loss of mast cells in certain tissues such as peritoneal cavity and stomach (30,32). Because Kit is essential for mast development in vivo, these data strongly suggest that Gab2 mediates Kit regulated mast cell development. Consistent with this idea, SCF-evoked proliferation is reduced in Gab2 Ϫ/Ϫ mast cells in vitro (32). However, the mechanism by which Gab2 participates in Kit-evoked mast cell proliferation is still not clear. In this study, we provide evidence that the scaffolding adapter Gab2 via Shp-2 mediates signal from Kit Tyr 567 in an SFK-dependent manner to activate the Rac/JNK pathway that is important for the Kit-regulated proliferation. The KitY567-Gab2 pathway is likely to be critical for mast cell development in specific tissues in vivo.
Plasmid Constructs-The HA-Gab2 WT and Gab2-⌬Shp-2 fragments, released by restriction enzyme digestion from pBluescript (23), were cloned into the retroviral vector pMXspuro (a gift from T. Kitomura, Tokyo University). The GST-RBD (Rac Binding Domain of PAK1) plasmid was a kind gift from Dr. C. Carpenter (Beth Israel Deaconess Medical Center, BIDMC, Boston).
Immunoprecipitation and Western Blots-Cells were lysed in 1% Nonidet P-40 lysis buffer as described previously (22). Total cell lysates or immunoprecipitates were resolved by SDS-PAGE, immunoblotted with the indicated primary antibodies,

Gab2/Shp-2 Activates the Rac/JNK Pathway in Kit Signaling
followed by horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG (Amersham Biosciences), and developed by enhanced chemiluminesence (ECL) (Amersham Biosciences). The intensities of bands in Western blots were quantified by densitometry analysis using NIH Image 1.63F software.
Proliferation and Apoptosis Assays-BMMC proliferation assays were performed as described (8 For quantifying apoptosis, cells were stained with FITC-Annexin V (Promega) according to the manufacturer's instructions, and analyzed by flow cytometry.
Retroviral Infection-of BMMC PMXs-puro retroviral plasmids were transfected into the ecotrophic packaging line Plate-E (36) using FuGENE (Roche Applied Science). Viruscontaining culture supernatants were collected 2-days later. BM cells cultured in IL-3-containing IMDM medium for 12 days were spin-infected (2500 rpm, 90 min) with pMXs-puro virus supernatants in the presence of 4 g/ml polybrene, and then incubated at 37°C for 18 -24 h. Infected cells were selected in the presence of 1 g/ml of puromycin for 10 -12 days, and then cultured in the absence of puromycin for ϳ4 weeks.
Analysis of Tissue Mast Cells-Back skins and ears from WT, Kit Y719F/Y719F, Gab2 Ϫ/Ϫ , KitY719F/Y719FϩGab2 Ϫ/Ϫ mice (2-5-month old) were dissected, fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned (5 m). Mast cells were visualized by staining tissue sections with toluidine blue (Histology core facility at Beth Israel Deaconess Medical Center), and counted under a microscope. Data were presented as the total number of mast cells per 1-cm tissue with similar thickness. Hypodermis of the back skin consists of the fat cells and smooth muscle cells underneath the dermis.
Statistical Analysis-Paired data were evaluated by twotailed Student's t test. Comparisons of multiple groups were performed using two-way ANONA. (30,32) reported that mast cell numbers are decreased in certain tissues in Gab2 Ϫ/Ϫ mice. Most strikingly, there was Ͼ95% decrease of mast cells in the peritoneal cavity and stomach of Gab2 Ϫ/Ϫ mice. These data suggested that Gab2 plays an important role in the growth and development of mast cells in vivo, which is controlled by Kitinitiated signals. To begin to address this question, we cultured BMMC from Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ mice (see "Experimental Procedures"). Flow cytometric analysis showed that Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ BMMC have similar cell surface expression of Kit (Fig. 1A). Consistent with a previous report (32) and the role of Gab2 in IL-3 signaling (22), IL-3 evoked proliferation was lower in Gab2 Ϫ/Ϫ BMMC compared with Gab2 ϩ/ϩ BMMC (data not shown). Importantly, we found that SCF-evoked cell proliferation also was impaired in Gab2 Ϫ/Ϫ BMMC (Fig. 1B), as reported (32). We also analyzed cell death by apoptosis by Annexin-V staining, and found that loss of Gab2 had little effect on apoptosis of BMMC in the presence of SCF (Fig. 1C), suggesting that Gab2 regulates SCF-evoked cell cycle progression in BMMC.

Gab2 Is Required for Optimal Kit-initiated BMMC Proliferation-Previously, we and others
Kit Tyr 567 Signaling to Gab2 Depends on SFK Activity-To begin to address how Gab2 mediates SCF-evoked proliferation in mast cells, we first asked whether Gab2 forms a complex with Kit upon SCF stimulation. Gab2 was immunoprecipitated from starved and SCF-stimulated WT BMMC lysates, and subjected to immunoblotting with several antibodies ( Fig. 2A). Consistent with a previous report (32), Gab2 became tyrosyl-phosphorylated in response to SCF stimulation. We also found that Gab2 associated with Kit, Shp-2, and p85 ( Fig. 2A).
To examine how Kit signals to Gab2, we analyzed Gab2 tyrosyl phosphorylation in BaF3 cells (Kit-negative) reconstituted with WT Kit and the Kit mutants Y567F, Y569F, Y702F, and Y719F. These BaF3 cells express similar levels of WT and mutant Kit, as described previously (13). Upon SCF stimulation, Gab2 was robustly tyrosyl-phosphorylated (Fig. 2B) and associated with tyrosyl-phosphorylated Kit (data not shown) in Kit WT cells (Fig. 2B), as well as in cells expressing Kit Y569F, Y702F, and Y719F. In marked contrast, Gab2 tyrosyl phosphorylation ( Fig. 2B) and its association with tyrosyl phosphorylated Kit (data not shown) were inhibited in Kit Y567F cells (Fig.  2B) although residual Gab2 tyrosyl phosphorylation could be seen in Kit Y567F cells after longer exposure (data not shown). Loss of SCF-evoked Gab2 tyrosyl phosphorylation is not because of the defective kinase activity of the Kit 567F mutant since we found that SCF-evoked Kit 567F autophosphorylation and proliferation of Baf3 Kit 567F cells were not impaired compared with Kit WT (13). This result indicates that Tyr 567 is required for SCF-evoked Gab2 tyrosyl phosphorylation.
Kit Tyr 567 is primarily thought to recruit and activate SFK, although reports suggested that it also recruits Shp-2 (11) and Shc (12). Therefore, it is possible that Gab2 is recruited to Kit via Shp-2 or Shc. To distinguish these possibilities, SCF-evoked Shp-2 and Shc tyrosyl phosphorylation were examined in Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ BMMC. Although Shp-2 became tyrosyl phosphorylated in Gab2 ϩ/ϩ BMMC, its phosphorylation was lost in Gab2 Ϫ/Ϫ BMMC, indicating that Gab2 is required for SCF-evoked Shp-2 tyrosyl phosphorylation (Fig.  2C). This result strongly suggests that Shp-2 recruitment to Kit is actually dependent on Gab2, rather than the converse.
Our previous data show that Shc via Grb2 recruits Gab2 to ␤c of IL-3R/GM-CSFR (22). Therefore, our data support a model that Shc/Grb2 recruits Gab2 to Kit Tyr 567 . To test this hypothesis, we examined the biochemical behavior of Gab2 ⌬Grb2 mutant in MC/9 mast cell line. Previously, we established pool of MC/9 cells expressing HA-Gab2 WT and Gab2 ⌬Grb2 by retroviral transduction (20). Gab2 ⌬Grb2 is a Gab2 mutant that cannot bind Grb2 constitutively because its two Grb2 SH3 binding sites are mutated (21). Upon SCF stimulation, we found that HA-Gab2 WT became strongly tyrosyl phosphorylated and associated with Shc and Kit. In contrast, HA-Gab2 ⌬Grb2 was barely phosphorylated and lost its association with Shc and Kit (Fig. 2F). This data strongly support that the Grb2/Shc complex is required for Gab2 association with the activated Kit (Fig. 8).
To ask whether SFK activity is required for SCF-evoked Gab2 tyrosyl phosphorylation, we pretreated WT BMMC with the selective SFK inhibitor, SU6656, (40) before stimulating the cells with SCF. Notably, SU6656 strongly inhibited SCF-evoked Gab2 tyrosyl phosphorylation and its association with Shp-2 ( Fig. 2G). Interestingly, Gab2-associated Shc tyrosyl phosphorylation is not affected by SU6656 treatment (Fig. 2G), suggesting that SFK activity is not required for Shc tyrosyl phosphorylation. Collectively, these data suggest that Kit Tyr 567 via Shc is required for Gab2 recruitment, and Gab2 tyrosyl phosphorylation by SFK.
Kit-evoked Full Activation of Rac and JNK Requires Gab2-To investigate which Kit-activated downstream signaling pathway(s) is(are) regulated by Gab2, we first examined the activation of Erk and Akt in Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ BMMC by performing immunoblotting using phosphospecific antibodies. Upon stimulation with SCF with doses that induce BMMC proliferation (Fig. 1B), Gab2 Ϫ/Ϫ BMMC showed a 50% decrease in Erk phosphorylation at later time point (15 min) compared with Gab2 ϩ/ϩ BMMC. In contrast, Akt phosphorylation was similar in Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ BMMC (Fig. 3A). We and others (26,27) have shown that Gab2-mediated Erk activation is important for proliferation of the mammary epithelial cells. To ask whether decreased Erk activation is responsible for the impaired SCF-evoked proliferation of Gab2 Ϫ/Ϫ BMMC, we examined the effects of the Mek inhibitor UO126. We found that pretreatment of BMMC with UO126, which inhibited Erk activation, did not inhibit SCF-evoked thymidine incorporation in Gab2 ϩ/ϩ BMMC (Fig. 3B), indicating that Gab2-activated Erk is not required for SCF-evoked mast cell proliferation.
These data suggest that Gab2, via its association with Shp-2, is required for SCF-evoked activation of the Rac/JNK pathway and mast cell proliferation. However, the Shp-2 binding sites in Gab2 may also bind other SH2-containing signaling molecules such as SOCS proteins (42). To test whether Shp-2 itself is required for SCF-evoked Rac activation and mast cell proliferation, we acutely inhibited Shp2 expression in BMMC. To . Kit Tyr 567 signals to Gab2 depending on SFK activity. A, Gab2 forms a complex with Kit upon SCF stimulation. WT BMMC were starved, stimulated with 50 ng/ml SCF for 5 min, lysed, and immunoprecipitated with anti-Gab2 antiserum. Gab2 immune complexes were immunoblotted with antibodies against phosphotyrosine (pTyr), Kit, p85, Shp-2, and Gab2. B, Tyr 567 is required for Kit-evoked Gab2 tyrosyl phosphorylation. BaF3 cells expressing WT Kit or the indicated Kit mutants, were starved (0), and stimulated with 100 ng/ml SCF for 3 or 10 min. Nonidet P-40 lysates (10 7 cells) were immunoprecipitated with antibodies against Gab2, immunoblotted with anti-pTyr, and reprobed with antibodies against Gab2. C and D, SCF-evoked Shp-2 and Shc tyrosyl phosphorylation in Gab2 Ϫ/Ϫ BMMC. Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ BMMC were starved and then stimulated with 50 ng/ml SCF for the indicated times. Shp-2 (C) and Shc (D) were immunoprecipitated with Shp-2 and Shc antibodies, respectively, immunoblotted with anti-pTyr, and reprobed with Shp-2 and Shc antibodies, respectively. E, SCF-evoked Shc tyrosyl phosphorylation is inhibited in Kit Y567F BaF3 cells. BaF3 cells expressing Kit WT or Kit Y567F were starved and stimulated with SCF as in A. Shc tyrosyl phosphorylation was analyzed as in D. Data shown in A-D represent one of the three independent experiments with similar results. F, Grb2 SH3-binding sites in Gab2 are required for SCF-evoked Gab2 association with Shc and Kit. MC/9 cells expressing HA-tagged Gab2 WT and ⌬Grb2 mutant were starved and stimulated with 50 ng/ml of SCF for 5 min. Nonidet P-40 lysates (10 7 cells) were immunoprecipitated with HA antibody, immunoblotted with anti-pTyr, and reprobed with antibodies against HA, Shc, and Kit respectively. G, Src family kinase (SFK) activity is required for SCF-evoked Gab2 tyrosyl phosphorylation. BMMC were starved, pretreated with the Src family kinase inhibitor SU6656 (5 M) for 1 h before stimulation with 25 ng/ml SCF. Gab2 was immunoprecipitated with Gab2 antibodies, immunoblotted with anti-pTyr, and reprobed with antibodies against Gab2 and Shp-2, respectively. Data shown in F and G represent one of the two independent experiments with similar results. achieve this, we derived BMMC from mice bearing a homozygous "floxed" (fl/fl) allele of Shp-2 and the ERCre TM transgene (ERCre TM :fl/flShp2 mice). Addition of the estrogen analog, Tamoxifen, into the culture medium results in the acute activation of the ERCre TM fusion protein and excision of the floxed Shp-2 allele via Cre-loxp-mediated recombination.
Immunoblot analysis showed that the maximal reduction in Shp-2 protein level (ϳ75% compared with control) occurred between day 3-4 after Tamoxifen addition to fl/fl Shp-2 ERCre TM BMMC (Fig. 5A). Shp-2 deletion resulted in reduced (ϳ50%) SCF-evoked cell proliferation, as revealed by reduced thymidine incorporation (Fig. 5B). SCF-evoked Rac activation and JNK phosphorylation also were decreased following Shp-2 deletion (Fig. 5C). In contrast, SCF-evoked Erk and Akt activation were minimally affected in the presence of Tamoxifen (Fig.  5D). As a negative control, Tamoxifen treatment had no effects on the proliferation of WT BMMC (data not shown).
Because previous biochemical and genetic evidence indicates that Shp-2 acts upstream of Ras in RTK signaling, we also examined the effect of Shp2 deletion on SCF-evoked Ras activation. Ras was activated robustly in control BMMC. In contrast, SCFevoked Ras activation was reduced in Shp-2-deleted cells (Fig.  5E). Similarly, SCF-evoked Ras activation also was reduced (to a similar extent) in Gab2 Ϫ/Ϫ , compared with Gab2 ϩ/ϩ , BMMC (Fig. 5F). Thus, our result suggests that Shp-2 via interaction with Gab2 contributes to SCF-evoked Ras activation in mast FIGURE 3. Gab2 is required for full activation of Rac and JNK in response to SCF stimulation. A, SCF-evoked phosphorylation of Erk and Akt in Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ BMMC. BMMC were starved, stimulated with 50 ng/ml SCF for the indicated times. Lysates were immunoblotted with antibodies against phospho-Akt (Ser 473 ) and phospho-Erk1/2 followed by reprobing with Akt and Erk2 antibodies. Densitometry was used to quantify bands in Western blots. Numbers under the phospho-Erk2 blot show the ratios (arbitrary units) of phospho-Erk/total Erk2. B, pretreatment with UO126 has no effect on SCF-evoked BMMC proliferation. [ 3 H]Thymidine incorporation was measured as in Fig. 1A, except that WT BMMC were pretreated with Me 2 SO alone or 8 M UO126 for 30 min. The phospho-Erk immunoblot (bottom panel) shows that UO126 inhibits SCF-evoked Erk activation. Data shown in A and B represent one of the three independent experiments with similar results. C, SCF-evoked Rac activation is reduced in Gab2 Ϫ/Ϫ BMMC. Gab2 ϩ/ϩ and Gab2 Ϫ/Ϫ BMMC were starved and stimulated with 50 ng/ml SCF for indicated times. Activated Rac1 (Rac1-GTP) was assayed by incubating the lysates with GST-RBD, followed by immunoblotting with Rac1 antibodies, according to "Experimental Procedures." TCL (5%) of each sample was loaded as a control for the total Rac1 in BMMC (left panel ). Activated Rac1 was calculated by quantifying the Rac1-GTP level normalized for total Rac1, and presented as relative to signal for Gab2 ϩ/ϩ cells at 3 min (right panel). Results represent mean Ϯ S.E. from three independent experiments, *, p Ͻ 0.02. D, SCF-evoked JNK activation is reduced in Gab2 Ϫ/Ϫ BMMC. BMMC were starved and stimulated as indicated. JNK1 was immunoprecipitated, subjected to in vitro kinase (IVK ) assay using GST-c-Jun as substrate, and reprobed with antibodies against JNK1. JNK1 activity was quantified by analyzing phosphorylated c-Jun level normalized for JNK1 loading, and presented as relative to signal for Gab2 ϩ/ϩ cells at 7 min. Results represent mean Ϯ S.E. from three independent experiments, *, p Ͻ 0.01. E, SCF-evoked Rac activation is partially inhibited by pretreatment with PI3K inhibitor LY294002. WT BMMC were starved, pretreated with vehicle alone or 10 and 25 M LY294002 for 30 min, and stimulated with 50 ng/ml of SCF with the indicated times. Activated Rac1 in lysates was analyzed as in C. Numbers under the Rac1 blot show the ratios (arbitrary units) of Rac1-GTP/total Rac1. Total cell lysates were also immunoblotted with antibodies against phospho-Akt (473) and reprobed with anti-Akt antibodies. Note that LY294002 at 25 M concentration completely inhibits SCF-evoked Akt phosphorylation. Data shown in D and E represent one of the two independent experiments with similar results.
We also examined SCF-evoked PI3K activation in these cells by assessing the activation of Akt, a downstream effector of PI3K, using phospho-Akt antibodies. We found that SCFevoked Akt phosphorylation was dramatically reduced by ϳ90% in Kit Y719F/Y719F BMMC compared with WT BMMC (Fig. 6C), consistent with previous reports that Kit Tyr 719 is the major route for SCF-evoked PI3K and Akt activation (8,9). Also, as we observed in Fig. 3B, no change in Akt phosphorylation could be detected in Gab2 Ϫ/Ϫ compared with WT BMMC (Fig. 4D). However, the residual level of Akt phosphorylation in Kit Y719F/Y719F BMMC was eliminated in Gab2 Ϫ/Ϫ Kit Y719F/Y719F BMMC. This result indicates that whereas Gab2 is only responsible for a minor fraction (Ͻ10%) of SCF-evoked Akt (or PI3K) activation in BMMC, it nonetheless contributes to PI3K activation if the main route for PI3K activation through Kit719F is compromised.
The Gab2 Pathway Plays a Distinct Role in Mast Cell Development in Different Tissues-We also analyzed mast cells in various tissues in the aforementioned mice by toluidine blue staining (Fig. 7). Although there was no decrease in mast cell numbers in the dermis of the back skins of either Gab2 Ϫ/Ϫ mice or Kit Y719F/Y719F mice, mast cell numbers were decreased (ϳ50%) ( p Ͻ 0.05) significantly in compound Kit Y719F/ Y719FϩGab2 Ϫ/Ϫ mice compared with WT mice (Fig. 7A). Similarly, although there was an ϳ30% ( p Ͻ 0.05) or ϳ40% decrease ( p Ͻ 0.01) in mast cell numbers in the ears of Kit Y719F/Y719F and Gab2Ϫ/Ϫ mice, respectively, Gab2 deficiency combined with the Kit Y719F/Y719F mutation resulted in further decrease in mast cell numbers (ϳ80%) ( p Ͻ 0.001) (Fig. 7C). These results indicate that Gab2 and Kit Tyr 719 both contribute equally to mast cell development in the dermis and ears. In the hypodermis, Gab2 deficiency alone resulted in an ϳ50% decrease ( p Ͻ 0.01) in mast cell numbers (Fig. 7B), whereas the Kit Y719F/Y19F mutation alone did not affect mast cell numbers. Strikingly, Gab2 deficiency combined with the Kit Y719F/Y719F mutation almost eliminated (ϳ95% reduction) ( p Ͻ 0.001) mast cells in this location (Fig. 7B). In agreement with previous reports (32) (6), we also found that Gab2 deficiency led to Ͼ95% decrease in peritoneal mast cells, whereas Kit Y719F/Y719F mutation resulted in an ϳ75% decrease in these cells (data not shown). Furthermore, in double mutant mice, no mast cells were detected at this location (data not shown). These data indicate that the Gab2 pathway is more critical for mast cell development in specific tissues such as peritoneal cavity and hypodermis of back skins. However, both the Gab2 and the Kit Tyr 719 pathways contribute to mast cell development in other tissues such as ears and the dermis of the back skins. . Gab2, via Shp-2, is required for SCF-evoked Rac/JNK activation and BMMC proliferation. Gab2 Ϫ/Ϫ BMMC were transduced with pMXs-puro virus alone (Vector), or pMXs expressing Gab2 WT, or Gab2 ⌬Shp-2. A, Shp-2 association is required for Gab2-mediated SCF-evoked Rac and JNK activation. Gab2 Ϫ/Ϫ BMMC reconstituted with vector, Gab2 WT, or Gab2 ⌬Shp-2 were starved and stimulated with SCF, as indicated. Rac activation was assayed as in Fig. 3C. JNK activation was assessed by immunoblotting lysates with antibodies against phospho-JNK. Western with equal amount of lysates was probed with antibodies against JNK1 and Gab2, respectively. Numbers under the blots show the ratios (arbitrary units) of Rac1-GTP/ total Rac1 and phosphorylated JNK1/JNK1 respectively. B, Shp-2 association is required for Gab2-mediated, SCF-evoked proliferation. SCF-evoked [ 3 H]thymidine incorporation into Gab2 Ϫ/Ϫ BMMC reconstituted with vector, Gab2 WT, or ⌬Shp-2 mutant was assayed as in Fig. 1A, *, p Ͻ 0.001 by ANOVA. Data shown here represent one of the two independent reconstitution experiments with similar results. SEPTEMBER 29, 2006 • VOLUME 281 • NUMBER 39  Fig. 1A. *, p Ͻ 0.001. C, Shp-2 is required for SCF-evoked Rac and JNK activation. BMMC were starved, stimulated with 50 ng/ml SCF for the indicated times, and lysed. Activated Rac1-GTP was assayed as in Fig. 3C, calculated by quantifying the Rac1-GTP level normalized for total Rac1, and presented as relative to signal for Tamoxifen (Ϫ) cells at 15 min. Results represent mean Ϯ S.E. from three independent experiments (right panel, *, p Ͻ 0.01). Lysates were immunoblotted with anti-phospho-JNK, and reprobed with antibodies against JNK2 (bottom panel). Numbers under the phospho-JNK blot show the ratios (arbitrary units) of phosphorylated JNK1/JNK1. D, loss of Shp-2 has minimal effects on SCF-evoked Akt and Erk activation. Lysates, prepared as in Fig. 5C, were immunoblotted with antibodies against phospho-Akt (Ser 473 ) and phospho-Erk, and reprobed with antibodies against Akt1/2 and Erk1, respectively. E, SCF-evoked Ras activation is reduced in Shp-2-deleted BMMC. BMMC were starved, stimulated with 50 ng/ml SCF for the indicated times, and Ras activation was assessed as described under "Experimental Procedures." Lysates were also immunoblotted with anti-Ras antibodies as a loading control. Activated Ras was calculated by quantifying the Ras-GTP level normalized for total Ras, and presented as relative to signal for Tamoxifen (Ϫ) cells at 2 min (bottom panel ). Results represent mean Ϯ S.E. from three independent experiments, *, p Ͻ 0.01. Data shown in A, B, and D represent one of the three independent experiments with similar results. F, SCF-evoked Ras activation is reduced in Gab2 Ϫ/Ϫ BMMC. Ras activation was measured as in E and G, expression of Gab2-⌬Shp-2 mutant in Gab2 Ϫ/Ϫ BMMC failed to rescue Ras activation. Gab2 Ϫ/Ϫ BMMC reconstituted with Gab2 WT and Gab2-⌬Shp-2 as in Fig. 4. SCF-evoked Ras activation were measured as Fig. 5E. Numbers under the blots (F and G) show the ratios (arbitrary units) of Ras-GTP/total Ras. Similar results (F and G) were obtained from two independent experiments.

DISCUSSION
To gain insight into the mechanism by which Gab2 regulates mast cell development in vivo, we first investigated the role of Gab2 in Kit-initiated signaling responses in BMMC ex vivo. Our results indicate that SCF-activated Kit via Tyr 567 leads to Gab2 tyrosyl phosphorylation and association with Shp-2 in a SFKdependent manner. The Gab2⅐Shp-2 complex is required for full activation of Ras and the Rac/JNK pathway, and SCFevoked mast cell proliferation. These data identify Gab2⅐Shp-2 as a new pathway downstream of Kit in controlling mast cell proliferation in vitro. Considering that Gab2 contributes to mast cell development in tissues (Fig. 7), our data strongly suggest that the Gab2/Shp2 pathway may regulate mast cell development in vivo (Fig. 8).
Previous studies identified Tyr 567 in Kit as an important site for activation of the Rac/JNK pathway required for mast cell proliferation. However, the detailed mechanism by which Tyr 567 transmits downstream signals had been unclear. Our data strongly suggest that Gab2 is one critical signal mediator from Kit Tyr 567 and provide a biochemical explanation for how Tyr 567 via SFK contributes to the activation of the Rac/JNK pathway and mast cell proliferation (4). Gab2 tyrosyl phosphorylation requires Kit Tyr 567 (Fig. 2B), a site that is required for SFK binding and activation (4) (13,14). Consistent with this FIGURE 6. Gab2 acts in a pathway parallel to Kit Tyr 719 -PI3K to regulate mast cell proliferation. BMMC were generated from mice of the following four genotypes: ϩ/ϩ (WT), KitY719F/Y719F, Gab2 Ϫ/Ϫ , and KitY719F/ Y719FϩGab2 Ϫ/Ϫ . A, BMMC with the above four genotypes were subjected to SCF-evoked [ 3 H]thymidine incorporation assay as in Fig. 1C. *, p Ͻ 0.05 and **, p Ͻ 0.02 by ANOVA. B, Gab2 and Kit Tyr 719 both contribute to SCF-evoked JNK activation. BMMC from the above four genotypes were starved, stimulated with 50 ng/ml SCF for the indicated times, and subjected to assay for JNK activation, as in Fig. 3D. C, Gab2 plays a minor role in SCF-evoked Akt activation. Lysates from BMMC as in B and C were immunoblotted with antibodies against phospho-Akt (Ser 473 ) and Akt. Numbers under the blot show the ratios (arbitrary units) for phosphorylated Akt/Akt. Data shown represent one of the three independent experiments with similar results.

Gab2/Shp-2 Activates the Rac/JNK Pathway in Kit Signaling
result, SFK inhibitor dramatically impairs SCF-evoked Gab2 tyrosyl phosphorylation and association with Shp-2. SCFevoked Shp-2 tyrosyl phosphorylation requires Gab2 (Fig. 2C). These data suggest a model in which SFK activation from Kit Tyr 567 results in Gab2 recruitment to Kit, Gab2 tyrosyl phosphorylation, and association with Shp-2 (Fig. 8). Two recent papers showed that there are Ͼ95% deceases of mast cells in the peritoneal cavity (6,43) and stomach (43) of Kit Y567F knock-in mice. The loss of mast cells in these two tissues in the Kit Y567F/Y567F mice resembles the phenotypes that were observed in Gab2 Ϫ/Ϫ mice (32). In contrast, there is ϳ75% decrease in mast cell in the peritoneal cavity of Kit Y719F/Y19F mice. These genetic data further support the model that Kit Tyr 567 signals to Gab2 (Fig. 8).
Our data strongly support a model that Tyr 567 via Shc/Grb2 brings Gab2 to Kit, where Gab2 can be phosphorylated by SFK in trans in the Kit heterodimer (Fig. 8). This mode of Gab2 recruitment to Kit is analogous to our previous finding that Shc, when tyrosyl-phosphorylated, via Grb2 recruits Gab2 to ␤c of IL-3/GM-CSFR (19). Consistent with this idea, we found that SCF-evoked Shc and Gab2 tyrosyl phosphorylation is dramatically reduced in Kit Y567F BaF3 cells (Fig. 2, B and E). Kit Tyr 567 was reported to be important for Shc phosphorylation (15). In addition, a published report suggests that Shc can bind Tyr 567 because a phosphorylated-Tyr 567 containing peptide can precipitate Shc from cell lysates (12). Result from searching Scansite (44) suggests phosphorylated Tyr 567 in Kit as a potential binding site for Shc SH2 domain. Importantly, the Gab2 ⌬Grb2 mutant lost its association with Shc and Kit in response to SCF stimulation (Fig. 2F). Consistent with the idea that Shc/Grb2recruited Gab2 can be phosphorylated by SFK in trans in the Kit heterodimer, we found that SFK inhibitor only impairs SCFevoked Gab2 tyrosyl phosphorylation whereas Gab2-associated Shc tyrosyl phosphorylation was not affected (Fig. 2G). This result suggests that Shc, upon being recruited to Kit Tyr 567 , is tyrosyl-phosphorylated mainly by kinase other than SFK.
At present, we cannot exclude the presence of other minor routes for recruiting Gab2 to Kit. For example, SFK activated through Kit Tyr 567 phosphorylates another docking site for Shc or Grb2 in Kit cytoplasmic tail. However, we did observe normal SCF-evoked Gab2 tyrosyl phosphorylation in cells expressing a mutant of the Grb2 binding site (45), Kit Y702F (Fig. 2B). It is also possible that SFK directly binds Gab2. Consistent with this idea, a previous report showed that EGF-evoked Gab2 tyrosyl phosphorylation was correlated with the weak Gab2 association with Src through the interaction between the two proline-rich motifs in Gab2 and the Src SH3 domain (46).
Our results reveal a novel function of Gab2, via Shp-2, in activating the Rac/JNK pathway critical for mast cell proliferation (Fig. 3C). How Gab2/Shp-2 regulates Rac remains unclear. Ras can activate Rac in PI3K-dependent and -independent manners (17,41). As SCF-evoked Akt is unaffected in Gab2 Ϫ/Ϫ BMMC or BMMC with acute deletion of Shp-2 (Figs. 3A and 5D), PI3K activation probably is not affected in Gab2 Ϫ/Ϫ or Shp-2 deleted BMMC. However, given that Gab2/Shp2 is important for SCF-evoked Ras activation (Fig.  5, E-G), it is possible that Gab2/Shp2 may regulate Rac activation via a Ras-dependent PI3K-independent pathway in mast cells. Consistent with this possibility, LY294002 pretreatment only partially inhibits SCF-evoked Rac activation (Fig. 3E). Notably, Ras interacts with the Rac exchange factor, which can activate Rac in a PI3K-independent manner (41). Alternatively, Gab2 via Shp-2 may inhibit the activity of a Rac GAP. In this regard, Shp-2 has been postulated to activate Rho via dephosphorylation of tyrosyl-phosphorylated p190RhoGAP, which has increased activity against Rho (47). Conceivably, Gab2/Shp2 may activate Rac by dephosphorylating a RacGAP expressed in mast cells. One attractive candidate for such a protein is GC/GAP, which is a Gab2-binding protein that has Rac GAP activity in vitro and ex vivo when overexpressed (48).
Previous genetic study in Drosophila and biochemical studies in mammalian cells indicate that Shp-2 acts at a step upstream of Ras (28). However, the Shp-2 substrate/target critical in Ras activation is still not unclear. Our data indicate that the Gab2⅐Shp-2 complex is involved in Kit-evoked Ras activation, which is consistent with the reported role of Kit Tyr 567 in Kit-evoked Ras activation (14,15). Together with our biochemical data that Kit Tyr 567 signals to Gab2 (Fig. 2), these results strongly support a model that Kit Tyr 567 via Gab2/Shp-2 regulates Ras activation. Furthermore, our result is also consistent with previous reports that Shp-2 via Gab1 activates Ras (49,50) in EGF signaling by preventing the recruitment of Ras-GAP to Gab1 (50). However, we could not detect any increased Ras-GAP association with the Gab2-⌬Shp-2, a Gab2 mutant that cannot bind Shp-2, compared with Gab2 WT in BMMC upon SCF stimulation. 6 This suggests that Gab2/Shp-2 regulates SCF-evoked Ras activation via mechanism other than inhibit-6 M. Yu, Y. P. Wang, and H. Gu, unpublished observation.

Gab2/Shp-2 Activates the Rac/JNK Pathway in Kit Signaling
ing recruitment of Ras-GAP to Gab2, which is a subject of continuing investigation in our laboratory.
Our biochemical data strongly suggest that the Gab2 and Kit Tyr 719 -PI3K activated pathways such as the Rac/JNK pathway contribute equally to mast cell development in some tissues, such as the dermis of back skin (Fig. 7A) and ears (Fig. 7C). However, analysis of mast cells in hypodermis (Fig. 7B), stomach (32), and peritoneal cavity (6,30,32,43) (data not shown) suggest that the Kit Tyr 567 -Gab2 activated pathway is more important than the Kit Tyr 719 pathway in mast cell development in these specific tissues. One possibility is that the signal from Kit Tyr 719 might be transient or weak. In these tissues, the Kit Tyr 567 -Gab2 pathway becomes more critical for the mast cell maturation and development. In this scenario, it is also possible that the small of amount of PI3K activated by Gab2 (Fig. 6D) may contribute to mast cell growth/maturation in these tissues. Consistent with this idea, p85␣ (the regulatory subunit of PI3K) knock-out mice showed no detectable mast cells in stomach (51). The other possibility is that Gab2/Shp-2 via Ras may activate other downstream effector only expressed in mast cells in specific tissues. Known Ras effectors include Raf, PI3K, Rac, and Ral (52). Lastly, it is also possible that Gab2 may be involved in other signaling system important for mast cell development in specific tissue. It has been reported that ␤1 integrin cross-linking can activate Gab2 biochemically (53). Mac-1 (a ␤2 integrin) knock-out mice were reported to have defective mast cell development in the peritoneal cavity (54). ␤7 integrin is required for mast cell progenitor (MCP) homing to intestines (55). Therefore, Gab2 could mediate integrin-initiated signals that may be important for MCP homing to specific tissues.
Decreased mast cell proliferation in Gab2Ϫ/Ϫ BMMC because of impaired activation of the Rac/JNK pathway could explain decreased mast cell numbers in various tissues. Rac is also known to play important role in regulating actin cytoskeleton remodeling and cell migration (56). Therefore, it is also possible that Gab2 Ϫ/Ϫ mast cells have impaired ability to migrate and adhere, which could also affect mast cell development in vivo. A recent study identified MCP from bone marrow as Kit ϩ as well as other cell surface makers ␤7 ϩ T1/ST2 ϩ Lin Ϫ Sca-1 Ϫ Ly6c Ϫ Fc⑀RIa Ϫ CD27 Ϫ (57). Therefore, our data suggest the possibility that Gab2 may play a role in differentiation of MCP into mature mast cells. Future studies are required to clarify the role of Gab2 in these processes.