The Sos1-Rac1 signaling. Possible involvement of a vacuolar H(+)-ATPase E subunit.

We have purified and identified a 32-kDa protein interacting with the Dbl oncogene homology domain of mSos1(Sos-DH) from rat brains by glutathione S-transferase-Sos-DH affinity chromatography. Peptide sequencing revealed that the protein is identical to a positive regulatory E subunit (V-ATPase E) of a vacuolar H(+)-ATPase, which is responsible for acidification of endosome and alkalinization of intracellular pH. The interaction between V-ATPase E and Sos-DH was confirmed by yeast two-hybrid assay. A coimmunoprecipitation assay demonstrated that a V-ATPase E protein physiologically bound to mSos1, and the protein was colocalized with mSos1 in the cytoplasm, as determined by immunohistochemistry. mSos1 was found in the early endosome fraction together with V-ATPase E and Rac1, suggesting the functional involvement of mSos1/V-ATPase E complexes in the Rac1 activity at endosomes. Overexpression of V-ATPase E in COS cells enhanced the ability of mSos1 to promote the guanine nucleotide exchange activity for Rac1 and stimulated the kinase activity of Jun kinase, a downstream target of Rac1. Thus, the data indicate that V-ATPase E may participate in the regulation of the mSos1-dependent Rac1 signaling pathway involved in growth factor receptor-mediated cell growth control.

The guanine nucleotide exchange factor mSos1 regulates cell growth, transformation, and differentiation (1,2). mSos1 catalyzes the exchange of GDP for GTP on Ras through a carboxylterminal domain homologous to a yeast CDC25. In addition to this carboxyl-terminal catalytic domain for Ras, the aminoterminal region of mSos1 contains a domain homologous to Dbl oncogene (DH) 1 and a domain homologous to pleckstrin (PH). Dbl oncogene proteins were shown to have the guanine nucle-otide exchange activity for Cdc42 and RhoA, which belong to the Rho GTPase family (3,4). Similarly, the DH domain of Dbl-related molecules (5) appears to catalyze the GDP-GTP exchange reaction on Rho-like small GTP-binding proteins. Recently, the DH domain of mSos1 (Sos-DH) was reported to promote the GDP-GTP exchange reaction on Rac1, another member of the Rho GTPase family, and thereby induce the activation of Jun kinase (JNK) activity and membrane ruffling (6). Thus, mSos1 appears to have a dual role, activating Ras through the CDC25 domain and Rac1 through the DH domain. However, the function of the Sos-DH domain has not been fully established. In the conventional model, the binding of Grb 2 to mSos1 at its COOH terminus and the subsequent translocation of mSos1 to the plasma membrane have been thought to occur in response to growth stimulation (7). Although Grb 2 relieves an inhibitory effect of the mSos1 COOH terminus, several lines of evidence reveal that binding of mSos1 to Grb 2 alone is not sufficient for activation of mSos1 (8 -10). The overexpression of the NH 2 terminus of mSos1 interfered with serum-, plateletderived growth factor-, and EGF-dependent cellular DNA synthesis and suppressed the mitogen-activated protein kinase activation (11). This strongly supports the idea that the interaction between the mSos1-NH 2 -terminal sequence and a membrane target(s) could be required for the mSos1 activity and membrane targeting and that the NH 2 -terminal sequence is critical for physiological function of mSos1. However, to date, no protein has been identified that regulates the mSos1 activity through its interaction with the NH 2 terminus of mSos1.
The vacuolar H ϩ -ATPase (V-ATPase) is a multimer enzyme complex composed of V 1 , a complex of at least seven different cytosolic components (A to H), which is responsible for the ATPase activity, and Vo, a complex of at least four different subunits (␣ to ␦), which functions in proton translocation (12). The V-ATPase plays an important role in various cellular process, including receptor-mediated endocytosis and maintenance of cytosolic pH (12). For example, the active V-ATPase pumps protons into the lumen of the vesicle and, thus, acidifies the vacuolar compartment. The low pH within endosomes generated by the V-ATPase is essential for ligand-receptor dissociation and receptor recycling. The V-ATPase also presents in the plasma membrane and exports protons out of the cells, leading to alkalinization of the cytosol. The study on the biological role of the V-ATPase subunits has been performed by using lower organisms. Disruption of the B subunit resulted in a larvallethal phenotype in Drosophila (13) and disruption of the VMA4, a yeast homologue of E subunit of the mammalian V-ATPase caused defects in bud morphology and actin distribution (14). The V-ATPase has been implicated in cell trans-formation (15,16) and has been shown to be activated in neutrophils in response to growth stimuli including granulocyte colony stimulating factor and phorbol myristate acetate (17). This implies the possible role of the V-ATPase in cell growth control through maintenance of pH homeostasis. In the present study, we demonstrate that the Sos-DH domain interacts with a V-ATPase E subunit and that this biochemical link may lead to the Sos1-dependent Rac1 activation.

MATERIALS AND METHODS
Plasmids-The mouse V-ATPase E subunit cDNA was provided by G. Dean. V-ATPase E expression vectors were constructed by inserting the cDNA into BamHI and EcoRI sites of pcDNA3-myc plasmid (Invitrogen, San Diego, CA) or by inserting the cDNA into KpnI and BamHI sites of pcGN-hemagglutinin (provided by M. Tanaka). GST-V-ATPase E was created by inserting V-ATPase E cDNA into BamHI and EcoRI sites of pGEX4T (Amersham Pharmacia Biotech). pSR␣3 carrying a full-length mSos1 and pGEX-Rac1 were obtained from A. Aronheim (18) and K. Kaibuchi (19), respectively. The Sos-DH domain expression vector was constructed by inserting the cDNA fragment corresponding to residues 261-457 of mSos1 into BamHI and EcoRI sites of pcDNA3-myc plasmid or pGEX4T plasmid. A 7-residue cluster mutation (amino acid residues 351 IIIRDII 357 for 351 LHYFELL 357 in mSos1) was introduced into the Sos-DH domain inserted into pGEX4T plasmid by using successive mutagenesis as described (11). Site-directed mutagenesis of V-ATPase E (where an asparagine 142 is changed to a glycine) was performed according to the method published (20).
Transfection Experiments-NIH3T3 or COS1 cells were transfected with various expression vector DNAs by using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's protocol. Transfected cells were grown in the culture medium for 36 -42 h. Transfectants were then subjected to immunoblotting assay of expressed proteins, GDP/GTP exchange activity assay for Rac1, in vitro kinase assay, in vitro reconstitution experiment, and immunofluorescence analysis as described below.
Purification of GST Fusion Protein-GST fusion proteins were expressed in DH5␣ cells and induced by the addition of 0.5 mM isopropylthio-D-galactopyranoside. The harvested bacteria were homogenized in extraction buffer (PBS, 1% Triton X-100, 5 g/ml leupeptin, 5 g/ml pepstatin), and extracts were mixed with 200 l of glutathione-Sepharose 4B (Amersham Pharmacia Biotech) for 1 h. The Sepharose beads were washed with extraction buffer and used for GST pull-down assay or GST affinity chromatography.
GST Pull Down Assay with 35 S-Labeled Cells and Transfected Cells-NIH3T3 cells (4 ϫ 150-mm dishes) were serum-starved in methionine/cysteine-free medium including 0.5% fetal bovine serum for 1 h and metabolically labeled for 4 h with [ 35 S]cysteine/methionine (4 mCi) (ICN, Costea Mesa, CA). The cells were washed once in ice-cold PBS, disrupted in 1 ml of homogenization buffer (10 mM Tris-HCl, pH 7.5, 5 mM EGTA, 5 mM MgCl 2 , 1 mM dithiothreitol, 10% sucrose, 10 M phenylmethylsulfonyl fluoride, and 1 g/ml leupeptin) and centrifuged at 1,000 rpm for 5 min. After centrifugation at 40,000 rpm for 10 min, the supernatant was saved as a cytosol fraction. The pellets were suspended in 0.5 ml of homogenization buffer. Equal amount of 4 M NaCl was added to the suspension followed by gentle shaking for 1 h. After centrifugation at 40,000 rpm for 10 min, the supernatant was dialyzed against buffer A (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 5 mM MgCl 2 , 1 mM dithiothreitol, 10 M phenylmethylsulfonyl fluoride, and 1 g/ml leupeptin) and used as a membrane fraction. For GST pull-down assay, GST and GST-Sos-DH fusion proteins coupled to glutathioneagarose beads (20 l) were incubated with the cytosol or membrane extracts for 2 h at 4°C. After extensive washing in buffer A plus 20 mM NaCl, proteins retained to the beads were resolved by SDS-PAGE and visualized by fluorography. For analysis of mSos1-V-ATPase E association, COS cells were transfected with mSos1 and disrupted in lysis buffer (see "Immunoprecipitation and Kinase Assay") 36 h after transfection. Lysates were incubated with GST-V-ATPase E-coupled resins for 2 h at 4°C, and bound proteins were analyzed by immunoblotting.
Affinity Chromatography-Membrane fractions prepared from one rat brain as described above were pre-cleaned by incubation with 200 l of GST-bound agarose beads and applied to a GST-Sos-DH affinity column (200 l). The column was washed with buffer A and buffer A containing 50 mM NaCl. Then, p32 proteins were eluted by buffer A containing 200 mM NaCl and visualized by silver staining after gel electrophoresis. For a large scale purification of p32, the membrane fraction was prepared from 30 rat brains.
Amino Acid Sequencing of p32-Purified P32 proteins were resolved by SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and digested by endopeptidase Lys-C. The generated peptides were fractionated by C18 reversed phase column chromatography and subjected to amino acid sequencing (21).
Yeast Two-hybrid Experiments-The coding sequences for the Sos-DH domain and V-ATPase E were created by polymerase chain reaction and fused to a Lex A DNA binding domain of pEG202 and to a B42 transcriptional activation domain of pJG4 -5, respectively. Yeast two-hybrid screening was conducted according to the company's protocol (CLONTECH, Palo Alto, CA). Briefly, pEG202-Sos-DH, pJG4 -5-v-ATPase E, and p8oplacZ were plated onto a glucose agar plate with medium lacking His, Ura, Trp (SD/ϪHUW: master plate). Six colonies were picked up from the master plates and streaked on either galactose and rafinose-containing agar plates with the medium lacking His, Ura, Trp, and Leu (SD/Gal/Raf/ϪHUWL) or glucose-containing agar plates with the same medium (SD/Glu/ϪHUWL).
Antibody Production-The rabbit anti-V-ATPase E antibody was raised against a carboxyl-terminal peptide (ALFGANANRKFLD) of the mouse V-ATPase E as described (21).
Immunocytochemistry-Cells on coverslides were fixed in 4% paraformaldehyde in PBS in 30 min and permeabilized with 0.2% Triton X-100 for 10 min. Immunostaining was performed with mouse anti-Myc or anti-hemagglutinin antibodies (Eastman Kodak Co.) and rabbit anti-mSos antibodies (Upstate Biotechnology, Inc., Lake Placid, NY). Antibodies were visualized with anti-mouse IgG labeled with ). Then each fraction was incubated with GST or GST-DH-coupled to glutathione-agarose beads. Proteins bound to the resins were resolved by SDS-PAGE and visualized by fluorography. Note that protein bands at the molecular mass region of around 28 and 60 kDa were compressed because of the overloaded unlabeled GST and GST-DH fusion proteins, respectively. M.W., molecular mass. fluorescein isothiocyanate (Sigma) and anti-rabbit IgG antibodies labeled with fluorescein isothiocyanate or tetramethylrhodamine B isothiocyanate (Sigma). To detect F-actin, cells were stained with tetramethylrhodamine B isothiocyanate-labeled phalloidin (10 g/ml) (Molecular Probes, Eugene, OR) for 1 h.
Separation of Endosomal Fractions-Endosomal fractions were isolated as described by Di Guglielmo et al. (23). Briefly, COS cells were homogenized in 0.25 M sucrose, and the homogenates were centrifuged at 1,500 rpm for 10 min. The supernatant was centrifuged at 100,000 rpm for 30 min. The microsomal pellets were resuspended to 1.15 M sucrose. The resuspension was overlaid with 1.00 and 0.25 M sucrose cushions and centrifuged at 100,000 rpm for 1.5 h. The endosome fraction at the 0.25-1.00 M sucrose interface was collected. Early and late endosomal fractions were prepared as described (24). Briefly, for preparation of early endosomal fractions, cells (4 ϫ 10-cm dishes) were washed with ice-cold PBS and incubated for 5 min at 37°C in 2.5 ml of Dulbecco's modified Eagle's medium supplemented with 10 mM glucose and 10 mM HEPES, pH 7.4 (internalization medium). Then the cell monolayers were quickly returned at 4°C and scraped in ice-cold PBS. The monolayers were incubated with internalization medium for 30 min at 37°C to isolate late endosomal fractions. The harvested cells were homogenized, and the homogenates were fractionated by ultracentrifugation on a three-step gradient consisting of 40.6, 16, and 10% sucrose in D 2 O as described (24). The early and late endosomal fractions collected were analyzed by immunoblotting.

Identification of a Sos-DH Domain Binding Protein(s)-To
test whether the Sos-DH domain interacts with some cellular proteins other than Rac1 GTPase, we bacterially expressed the Sos-DH domain fused to GST and used it for affinity chromatography. When the GST-DH fusion protein bound to resins was incubated with cytosol or membrane fractions from NIH3T3 cells metabolically labeled with [ 35 S]cysteine/methionine, 32-and 19-kDa proteins were detected to bind to GST-DH but not GST (Fig. 1). These proteins are localized in the membrane but not cysotol. Occasionally, the 40-kDa protein bound to the GST-DH. However, when bound proteins were dissociated with glutathione, 32-and 19-kDa proteins but not a 40-kDa protein were eluted reproducibly (data not shown). The 40-kDa protein appeared to interact with GST-DH resins nonspecifically. When rat brain crude extracts were tested for the binding study, the 32-and 19-kDa proteins were identified as GST-DH binding proteins (data not shown). Because the 32-kDa protein more prominently bound to GST-DH than the 19-kDa protein, we focused on the 32-kDa protein (hereafter indicated as p32) in this study.
To further characterize p32, we purified p32 proteins from rat brain membrane fractions by using GST-Sos-DH affinity FIG. 2. Purification of rat brain p32 proteins. A, isolation of rat brain p32 proteins and its reduced binding to the mutated GST-DH. A seven-residue cluster mutation was introduced into the DH domain (indicated as GST-DH*). Rat brain membrane fractions were fractionated by GST-DH or GST-DH* affinity chromatography. The fractions eluted from the column by 200 mM NaCl were resolved by SDS-PAGE and visualized by silver-staining. The 35-kDa protein nonspecifically bound to both GST-and GST-DH resins. B, Western blotting of the purified p32 with an anti-V-ATPase E antibody. Membrane fractions from rat brains applied to a GST-DH affinity chromatography column. The fractions eluted by 200 mM NaCl were analyzed by Western blotting using an anti-V-ATPase E antibody. M.W., molecular mass. The Sos1-Rac1 Signaling chromatography. As shown in Fig. 2A, p32 proteins were eluted by buffer containing 200 mM NaCl from a GST-Sos-DH affinity column. Because a cluster of seven substitution mutations in the Sos-DH domain ( 351 IIIRDII 357 for amino acid residues 351 LHYFELL 357 ) was reported to reduce the transforming activity of myristoylated Sos1 and the growth response of cells to EGF and platelet-derived growth factor (11), we tested whether the mutation in the Sos-DH domain affects its ability to bind to p32. The amount of p32 proteins bound to GST-DH with a cluster of substitution mutation was reduced as compared with that bound to GST-DH ( Fig. 2A, lower panel). This suggests that the region affects conformation of the domain essential for specific binding of Sos-DH to p32 proteins and that p32 may participate in the cell growth control through the interaction with Sos-DH.
p32 Is an E Subunit of the V-ATPase-To clarify the molecular identity of p32, purified p32 peptides were subjected to amino acid sequencing. Amino acid sequences of three peptides showed high similarity to the mouse vacuolar ATPase E subunit (hereafter designated as V-ATPase E) (25) ( Table I). The molecular mass of mouse V-ATPase E with 228 amino acid residues is reported to be 31 kDa on SDS-PAGE, and its elec-trophoretic mobility is very close to that of the p32 protein. To confirm that p32 is V-ATPase E, the antibody against a COOHterminal peptide derived from V-ATPase E was raised and used for immunoblotting analysis. The protein at the 32-kDa region obtained from GST-DH affinity chromatography reacted with an anti-V-ATPase E antibody, indicating that p32 is indeed the rat V-ATPase E (Fig. 2B). Interestingly, the sequences of peptide 2 and 3 showed similarity to the corresponding region of V-ATPase E, but their amino-terminal sequences were slightly different (Table I), indicating that there might be isoforms of V-ATPase E. Consistent with our observation, microheterogeneity in V-ATPase E has previously been detected in the twodimensional gel electrophoresis (26).
Interaction of Sos and Sos-DH with V-ATPase E-To further confirm the result of in vitro interaction between Sos-DH and GST-V-ATPase E, we tested whether Sos-DH interacts with V-ATPase E in vivo using a Lex A-based two-hybrid system. V-ATPase E was fused to a Lex DNA binding domain and tested for interaction with the Sos-DH fused to a B42 transcriptional activation domain. Coexpression of a V-ATPase E-Lex A fusion with a Sos-DH-B42 fusion caused the growth of many cell transformants with increased ␤-galactosidase (Gal) activ- FIG. 3. A, interaction of Sos-DH with V-ATPase E in the yeast two-hybrid system. pEG202-Sos-DH was cotransfected into yeast cells with pJG4 -5-V-ATPase E. All six transformants grew well and had strong ␤-galactosidase activity on the selective synthetic medium with galactose and rafinose (Gal/Raf), but no growth of these transformants was observed on the same medium with glucose (Glu). plexA-LexA/GAL4, pLexA-Human lamin C, pEG202-Sos-DH, and pJG4 -5-V-ATPase E were transfected into yeast cells together with the p8oplac Z reporter plasmid for the negative, positive, and selfactivation controls, respectively. The positive control induced LacZ expression (blue color), but no expression of LacZ was detected in either the negative or selfactivation controls. The presence of both fusion proteins in each yeast strain was confirmed by Western analysis. B, coimmunoprecipitation of mSos1 with V-ATPase E. Lysates from COS1 cells were immunoprecipitated with anti-V-ATPase E (VE) antibodies or nonimmune rabbit IgG (cont), and the immunoprecipitates (IP) were probed with anti-mSos1 antibodies (I). Reciprocally, the lysates were immunoprecipitated with anti-mSos1 antibodies or nonimmune rabbit IgG (cont), and the immunoprecipitates were probed with anti-V-ATPase E antibodies (II). IB indicates immunoblotting. C, association of mSos1 with V-ATPase E. Cell lysates from COS1 cells (1 ϫ 10 cm) transfected with pSR␣3-mSos1 were prepared and incubated with GST-V-ATPase affinity resins. Proteins retained onto the resins were analyzed by immunoblotting with rabbit anti-mSos1 antibodies. D, interaction of mSos1 and V-ATPase in situ. NIH3T3 cells were cotransfected with mSos1 and the Myc-V-ATPase E DNAs. V-ATPase E was detected by mouse anti-Myc antibodies and tetramethylrhodamine B isothiocyanate-labeled anti-mouse IgG (a). mSos1 was detected by rabbit anti-mSos1 antibodies and fluorescein isothiocyanate-labeled anti-rabbit IgG (b). c shows a merged picture. The arrows indicate the colocalization of mSos1 and V-ATPase E. ity (Fig. 3A), indicating that V-ATPase E interacts with Sos-DH. In the control experiments, the individual construct did not self-activate in this system. To determine whether mSos1 physiologically interacts with V-ATPase E, we immunoprecipitated endogenous mSos1 from COS cells and performed immunoblotting with antibodies against V-ATPase E. mSos1 coimmunoprecipitated with V-ATPase E (Fig. 3B). Reciprocally, V-ATPase E immunoprecipitated with its antibodies was complexed with mSos1 (Fig. 3B). In addition, GST-V-ATPase E fusion proteins-coupled resins were created and incubated with cell lysates from COS cells overexpressing wt mSos1. Immunoblot analysis demonstrated that mSos1 bound to the resins, indicating the interaction of mSos1 with V-ATPase E (Fig. 3C). We next examined the subcellular localization of mSos1 and V-ATPase E. When NIH3T3 cells were cotransfected with mSos1 and the Myc-epitope-tagged V-ATPase E, mSos1 was partially colocalized with V-ATPase E in the cytoplasm (Fig.  3D), which is in agreement with the result of physical interaction between two proteins as described above. Preliminary studies using deletion mutants of V-ATPase E showed that mutants lacking amino acids 1-91 were defective in interaction with mSos1, implying that the Sos binding region may reside in the NH 2 -terminal region of V-ATPase E. 2 Colocalization of Sos with V-ATPase E at Endosomes-It is known that after growth stimulation of cells, ligand-growth factor receptor complexes are delivered to the early endosome where the V-ATPase-catalyzed acidification of this compartment leads to dissociation of the complexes (12). Because mSos1 appears to interact with V-ATPase E, anchoring on the peripheral domain of the V-ATPase at the endosome membrane, we examined whether mSos1 is detectable at the endosome upon growth stimulation. To this end, EGF receptormediated endocytosis was examined. 15 min after stimulation with EGF, endosome fractions were isolated and analyzed for the presence of mSos1 and the EGF receptor by immunoblotting. The data show that mSos1 and the EGF receptor were contained in endosome fractions (Fig. 4A). By using floatation on a sucrose-D 2 O gradient, we next analyzed whether mSos1, V-ATPase E, and Rac1 are distributed in the early and late endosome fractions. Postnuclear cell lysates were fractionated, and the early and late endosomal fractions were subjected to gel electrophoresis and immunoblotting. Cells were incubated with horseradish peroxidase as a general marker for the endosome content (23), and the endosomal fractions were confirmed by the distribution of Rab5 and Rab7 as general markers for endosome content on the gradient. Fig. 4B shows that mSos1 was detected in the early endosomal fraction but not in the late endosomal fraction. By contrast, both V-ATPase E and Rac1 were detected in the early and late endosomal fractions. These data demonstrate that mSos1, V-ATPase E, and Rac1 were present in the early endosome, which is consistent with the biochemical observation that mSos1 binds to V-ATPase E (Fig.  3, B and C) and regulates the activity of Rac1 as described below.
Effects of Overexpression of V-ATPase E on the Rac1 and JNK Activities-Overexpression of the DH domain of Sos1 in COS cells has been reported to enhance the guanine nucleotide exchange activity of Rac1, resulting in the activation of JNK, a downstream target for Rac1 (6). Activation of JNK induces phosphorylation of transcription factors including C-Jun and ATF2, which regulate gene expression (27). To test whether overexpression of V-ATPase E also affects the Rac1 exchange activity and the JNK activity, we performed these assays by using a cell-free system. As in the case of mSos1 overexpression (Fig. 5A), cell lysates from COS cells overexpressing V-ATPase E increased the amount of GDP dissociated from Rac1, suggesting that V-ATPase E stimulates the activity of a guanine nucleotide exchange factor for Rac1 (Fig. 5B). To further confirm a mediating role of V-ATPase E in the regulation of mSos1 activity, a mouse V-ATPase E mutant with Asp-142 Gly was created. This aspartate is well conserved among various species from yeast to human. Substitution of Asp-145 with Gly in yeast V-ATPase E is known to disrupt its normal function, causing defects in bud morphology, actin distribution, and cytokinesis. As shown in Fig. 5C, the level of the nucleotidereleasing activity for Rac1 was higher in wild type V-ATPase Eand mSos1-cotransfected cells than that observed in mSos1transfected cells. In contrast, a V-ATPase E Asp-142 significantly suppressed the mSos1-dependent nucleotide exchange activity. Furthermore, immunodepletion of endogenous mSos1 with anti-mSos1 antibodies completely abolished the ability of lysates from V-ATPase E-transfected cells to stimulate the guanine nucleotide exchange reaction on Rac1 (Fig. 5D). Thus, the data indicate that a nucleotide exchange factor involved in the V-ATPase E-induced activation of Rac1 is mSos1. Finally, we detected that cell lysates from the V-ATPase-transfected 2 J. Mitsushita and T. Kamata, unpublished data.

FIG. 4. Analysis of intracellular localization of mSos1 and V-ATPase E during endocytosis.
A, presence of mSos1 in endosomes. Serum-deprived Rat 6 cells were incubated with EGF (100 ng/ml) for 15 min and washed. Endosome fractions were isolated by differential centrifugation ("Materials and Methods"), and EGF receptor (EGFR) and mSos1 proteins in endosomes were detected by immunoblotting using antibodies directed to the EGF receptor and mSos1, respectively. B, distribution of mSos1, V-ATPase E, and Rac1 proteins in endosomal fractions. Early and late endosomes from Rat 6 cells were prepared by centrifugation on a sucrose-D 2 O gradient as described. 100 g of proteins of each fraction was subjected to SDS-PAGE and analyzed by immunoblotting analysis with antibodies against mSos1, V-ATPase E, and Rac1. Recovery of early and late endosome fractions was confirmed by immunoblotting with anti-Rab 5A and Rab 7 antibodies, respectively. M.W., molecular mass. cells potentiated the ability of JNK to phosphorylate c-Jun proteins (Fig. 6). Similarly, ATF2 phosphorylation by JNK was enhanced by V-ATPase E overexpression (data not shown). Overexpression of Sos-DH as a positive control activated the JNK activity, as reported previously (6) (Fig. 6). Furthermore, actin stress fiber was dramatically disrupted, and cells exhibited a condensed morphology when V-ATPase E was overexpressed in NIH3T3 and COS cells (Fig. 7). This indicates that V-ATPase E contributes to actin cytoskeleton organization closely associated with the growth state of cells. Taken together, the results suggest that V-ATPase E regulates the mSos1-Rac1 signaling. DISCUSSION We have identified V-ATPase E as a protein that specifically interacts with the DH-domain of mSos1 by means of GST-Sos-DH affinity chromatography and a yeast two-hybrid assay. mSos1 coimmunoprecipitated with V-ATPase E, associated with V-ATPase E in a cell-free system, and partially colocalized with V-ATPase E. Furthermore, mSos1 were present at endosomes after EGF stimulation of cells and cosedimented with V-ATPase E and Rac1 in the early endosome fraction. These data suggest that mSos1 can bind to V-ATPase E on the outer surface of endosomes during the process of growth factor receptor-mediated endocytosis. To our knowledge, this is the first report describing the interaction of mSos1 with a component of vacuolar type ATP-driven proton pump through its aminoterminal DH domain. The data are in agreement with the previous observation that activation of EGF receptor can lead to the localization of mSos1⅐Grb 2 complexes at endosomes (23). Very recently, mSos1 has been shown to interact with the SH3A domain of intersectin, an endocytotic protein that is involved in clathrin-mediated endocytosis (28). Intersectin competes with Grb 2 for binding to the carboxyl-terminal proline-rich domain of mSos1, and overexpression of intersectin results in the attenuation of EGF-induced Ras activation.
Originally, a subunit E of the V-ATPase has been characterized as an important regulatory protein in V-ATPase function (29). In in vitro reconstitution experiments, V-ATPase E promotes the ATPase activity of the cytosolic V 1 subcomplex of a vacuolar type ATPase. Furthermore, the temperature-sensitive yeast V-ATPase E mutant caused a decrease in the V-ATPase activity, which is associated with the defects in actin distribution, bud morphology, and cytokinesis (14). The present data suggest that in addition to the regulatory function for the V-ATPase, V-ATPase E may exert a role in the regulation of mSos1-Rac1-mediated growth signaling. This conclusion can be drawn from the following evidences. First, overexpression of V-ATPase E stimulated the mSos1-mediated release of GDP from Rac1, and a loss of function mutant of V-ATPase E was defective in this activity. Moreover, an increased level of V- ATPase expression activated phosphorylation of a transcription factor c-Jun by JNK, a downstream effector of Rac1. Distribution of actin stress fiber was altered upon overexpression of V-ATPase E. Because small G-proteins are key elements in the reorganization of the actin cytoskeleton induced by growth factors, V-ATPase E might affect cytoskeletal organization through small G-proteins including Rac1. We propose a model regarding the mechanism for mSos1-dependent Rac1 activation. After growth stimulation, ligand-induced growth factor receptor activation leads to its tyrosine phosphorylation and recruitment of Grb 2 ⅐mSos1 complexes. Then ligand-mediated receptor internalization causes translocation of receptor-Grb 2 ⅐mSos1 complexes to the endosome membrane (27,28). Subsequently, the DH domain of Sos1 binds to V-ATPase E localized at the endosome, and this interaction may aid in activating the catalytic activity of mSos1, resulting in transmission of signals to Rac1. In this scheme, the DH domain of Sos1 is postulated to act as a binding site for V-ATPase E as well as a catalytic domain for the nucleotide exchange reaction on Rac1 (6). Presumably, the different amino acid residues within the peptide region are responsible for these two distinct functions. The noncatalytic, regulatory function of Sos-DH has also been implicated in forming a complex of the Sos1 NH 2 terminus and a cellular machinery (11). The V-ATPase is associated with plasma membranes in some types of cells including macrophages and neutrophils (29), and granulocyte colony stimulating factor and phorbol esters up-regulate the V-ATPase at the plasma membrane of neutrophils, which exports protons out of the cells, resulting in alkalinization of the cytosol (17). In our study, the plasma membrane of COS cells also contained V-ATPase E, 3 implying the functional role of the plasma membrane-bound V-ATPase. Therefore, at present we do not rule out the possibility that the V-ATPase localized at the plasma membrane also participates in the regulation of the mSos1-Rac1 signaling.
Considering that V-ATPase E is involved in the endocytic function, it would be interesting to hypothesize that mSos1 may also play a role in endocytosis, perhaps through the mSos1-Rac1-signaling pathway. Various clathrin-mediated trafficking events are triggered after activation of small GTPbinding proteins by guanine nucleotide exchange factors. For example, Ral is required for the internalization of the EGF receptor (30). As for Rac1, Rac1 regulates transferrin receptormediated clathrin-coated vesicle formation (31), EGF-induced 3 K. Miura and T. Kamata, unpublished data.
FIG. 6. Activation of JNK by Sos-DH and V-ATPase E proteins. COS1 cells were transfected with the myc-Sos-DH or myc-V-ATPase E expression vector. After 36 h, cells were serum-starved for 12 h. Endogenous JNK1 was immunoprecipitated, and immune complexes were processed for the kinase assay using GST-c-Jun as a substrate. Aliquots of lysates were analyzed for expression of JNK1, Sos-DH, and V-ATPase E by immunoblotting using an anti-Myc antibody and polyclonal anti-JNK1 antibody. The common band (*) in all lanes indicates a protein nonspecifically bound to horseradish peroxidase-labeled anti-mouse IgG. endocytosis via synaptojanin2 (32), and recycling of the Fc receptor (33). Whether the mSos1-V-ATPase E interaction is a necessary step for Rac1-mediated endocytosis remains to be determined.
Finally, our finding suggests that endocytosis may couple to a growth-signaling pathway mediated by small G proteins. Recently, it has been shown that membrane trafficking not only attenuates growth factor receptor signaling but also establishes and regulates the signaling pathways (34). For example, the importance of this intracellular process has been recognized by the findings that receptor tyrosine kinase-mediated signaling to mitogen-activated protein kinase involves endocytic trafficking (35). Thus, V-ATPase E appears to have a dual role in both endocytosis as well as signal transduction pathways.