The Ras/Rac Guanine Nucleotide Exchange Factor Mammalian Son-of-sevenless Interacts with PACSIN 1/Syndapin I, a Regulator of Endocytosis and the Actin Cytoskeleton*

Mammalian Son-of-sevenless (mSos) functions as a guanine nucleotide exchange factor for Ras and Rac, thus regulating signaling to mitogen-activated protein kinases and actin dynamics. In the current study, we have identified a new mSos-binding protein of 50 kDa (p50) that interacts with the mSos1 proline-rich domain. Mass spectrometry analysis and immunodepletion stud-ies reveal p50 as PACSIN 1/syndapin I, a Src homology 3 domain-containing protein functioning in endocytosis and regulation of actin dynamics. In addition to PACSIN 1, which is neuron-specific, mSos also interacts with PACSIN 2, which is expressed in neuronal and nonneuronal tissues. PACSIN 2 shows enhanced binding to the mSos proline-rich domain in pull-down assays from brain extracts as compared with lung extracts, suggesting a tissue-specific regulation of the interaction. Proline to leucine mutations within the Src homology 3 domains of PACSIN

In addition to the PRD, mSos contains a CDC25 homology domain, encoding Ras GEF activity (3)(4)(5)14), and a Dbl homology domain, endowed with GEF activity for Rac, a member of the Rho superfamily of GTPases (15,16). In fact, mSos is the prototype member of a family of bifunctional GEFs, including Ras-GRF1 and Ras GRF-2, having dual specificity for Ras and Rac (16). Through its Dbl homology domain, mSos binds directly to Rac (17). However, the GEF activity of mSos toward Rac appears to be unique relative to other Rho family GEFs in that it catalyzes guanine nucleotide exchange as part of a macromolecular complex with Eps8 and E3b1, two proteins functioning in growth factor signaling (18,19). Rac activation has multiple effects in cells, the most prominent being alterations in the actin cytoskeleton leading to membrane ruffling and lamellipodia formation (20). Thus mSos, through its ability to activate Rac, is thought to play a functional role in growth factor-mediated regulation of actin dynamics (18,21).
To screen for novel mSos binding partners, we performed overlay assays of brain extracts with fusion proteins encoding the PRD of mSos1. A major mSos1-binding protein of 50 kDa was detected and purified by affinity chromatography. Mass spectrometry analysis identified the protein as PACSIN 1/syndapin I. PACSIN 1 was originally identified based on its differential expression in intact and lesioned mouse brain (22), and syndapin I was independently identified through its SH3 domain-dependent interaction with dynamin 1 (23). We will use the name PACSIN throughout to collectively refer to PAC-SIN and syndapin. Whereas PACSIN 1 is neuron-specific, its closely related homologue PACSIN 2 is expressed in brain and several nonneuronal tissues (24,25). Interestingly, the PACSINs appear to be involved in regulation of endocytosis and the actin cytoskeleton. Through a C-terminal SH3 domain, the PACSIN isoforms interact with the endocytic regulatory enzymes dynamin 1 and synaptojanin 1, as well as with N-WASP, a stimulator of Arp2/3-mediated actin nucleation and assembly (23,25,26). Overexpression of full-length PACSIN stimulates cortical actin assembly, leading to filopodia formation, and the PACSINs localize to sites of high actin turnover, such as filopodia and lamellipodia (25).
After identifying PACSIN 1 as a mSos1 binding partner, we confirmed the interaction in vitro and used co-immunoprecipitation analysis to demonstrate the interaction in vivo. Interestingly, mSos1 co-distributes with PACSIN 1 in the growth cones and filopodia of cultured hippocampal neurons, and both proteins co-localize with actin in the filopodia from growth cones of dorsal root ganglia neurons in culture. Further, PACSIN 1 and mSos1 are co-localized in growth factor-induced membrane ruffles in COS-7 cells, and their interaction is regulated by mSos1 phosphorylation. Together, these data provide further evidence for a role for mSos in regulation of the actin cytoskeleton.
DNA Constructs and Recombinant Proteins-His 6 -tagged rat syndapin I (25) and Flag-tagged mouse Sos1 (28) in mammalian expression vectors were generous gifts of Dr. Regis Kelly (University of California, San Francisco) and Dr. Jeffrey Pessin (University of Iowa), respectively. A protein construct encoding the SH3 domain of rat syndapin I (residues 376 -441) was generated by polymerase chain reaction with Vent DNA polymerase (New England Biolabs) using full-length cDNA as template and the following primers: forward, 5Ј-CGCCTCGAGCG-GATCCAACCCCTTCGAGGACGATGC-3Ј; reverse, 5Ј-CGGAATTC-CTATATAGCCTCAACGTAGTTG-3Ј. The resulting polymerase chain reaction product was digested with BamHI and EcoRI and cloned inframe into the corresponding sites of pGEX-2T. Mouse Sos1 cDNA was used as a template to generate the following GST fusion proteins: GST-NT (residues 1111-1228) and GST-CT (residues 1223-1341). GST-NT was generated with the forward primer 5Ј-GCGGATCCTCT-GGCACCTCCAGCAAC-3Ј and the reverse primer 5Ј-GCGGAATTCT-CAATCAGGTGTCCTCACAGG-3Ј. GST-CT was generated with the forward primer 5Ј-GCGGGATCCCCTGTGAGGACACCTGATG-3Ј and the reverse primer 5Ј-GCGGAATTCTCAGGAAGAATGGGCATTC-3Ј. The resulting polymerase chain reaction products were digested with BamHI and EcoRI and cloned in-frame into the corresponding sites of pGEX-2T. GST fusion proteins encoding full-length mouse PACSIN 1 and 2 were prepared as described (26). PACSIN isoforms containing single amino acid changes in the SH3 domains were derived using the mutation oligonucleotides P1-P434L (5Ј-GGCCTCTATCTCGCGAAC-TACGTTG-3Ј) for PACSIN 1 and P2-P478L (5Ј-GGCCTATACCTCGC-GAACTATGTCG-3Ј) for PACSIN 2 on the corresponding wild-type cDNAs in combination with the Transformer TM site-directed mutagenesis kit (CLONTECH).
Tissue and Subcellular Fractionation-Various adult rat tissues, including brain, were homogenized in Buffer A (10 mM HEPES-OH, pH 7.4, 0.83 mM benzamidine, 0.23 mM phenylmethylsulfonyl fluoride, 0.5 g/ml aprotinin, and 0.5 g/ml leupeptin). A postnuclear supernatant was obtained by centrifugation for 5 min at 800 ϫ g max , and the extracts were then separated into cytosolic and membrane fractions by ultracentrifugation at 205 000 ϫ g max for 30 min at 4°C. In some cases, the postnuclear supernatant was incubated with 1% Triton X-100 for 30 min prior to ultracentrifugation, leading to a crude Triton-soluble lysate. Differential centrifugation of rat brain extracts, leading to the defined subcellular fractions in Fig. 2, was performed as described previously (29).
Overlay Assays-Overlay assays with GST fusion proteins were performed as described (30). Briefly, protein fractions were resolved by SDS-PAGE and transferred to nitrocellulose. Membranes were blocked in 5% nonfat dry milk powder in phosphate-buffered saline (PBS) (20 mM NaH 2 PO 4 , 0.9% NaCl, pH 7.4) for 1 h and incubated overnight at 4°C with 10 or 20 g of GST fusion protein diluted in Tris-buffered saline (20 mM Tris-Cl, 150 mM NaCl, pH 7.4) containing 3% bovine serum albumin, 0.1% Tween-20, and 1 mM dithiothreitol. Bound fusion protein was subsequently detected using affinity-purified antibodies directed against GST.
Affinity Chromatography-Cytosolic or crude Triton-soluble lysates were prepared from different tissues as described above. For the cytosolic samples, Triton X-100 was added to 1%. The samples were then incubated for 2 h or overnight at 4°C with GST fusion proteins precoupled to glutathione-Sepharose beads. After incubation, samples were washed three times in Buffer A containing 1% Triton X-100, and bound proteins were resolved by SDS-PAGE and processed for Western blot analysis or stained with Coomassie Blue. For precipitation experiments with full-length fusion proteins of wild-type and mutant PACSINs, mouse brains were homogenized in Buffer B (10 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EGTA, 0.1 mM MgCl 2 ,) containing 1% CHAPS and a protease inhibitor mixture (Sigma). The homogenates were centrifuged for 30 min at 21,000 ϫ g, the supernatant was decanted and recentrifuged, and Triton X-100 was added to the resulting supernatant at a final concentration of 0.05%. The preparation was dialyzed overnight against Buffer B and centrifuged as before. The resulting supernatant was incubated overnight at 4°C with GST-PAC-SINs precoupled to glutathione-Sepharose. The beads were subsequently washed extensively in Buffer B containing 0.1% Triton X-100, and bound proteins were resolved by SDS-PAGE and processed for Western blot analysis. For phosphorylation experiments, COS-7 cells were serum-starved overnight and preincubated with 50 M PD-098059 (Calbiochem, La Jolla, CA) in Me 2 SO or Me 2 SO alone for 30 min at 37°C. Cells were then treated with 100 ng/ml epidermal growth factor (EGF) for 5 min at 37°C in the absence or presence of PD-098059 and lysed in ice-cold Buffer C (10 mM HEPES-OH, pH 7.4, 5 mM EGTA, 5 mM EDTA, 50 mM sodium fluoride, 20 mM sodium pyrophosphate, 1 mM sodium vanadate, 0.83 mM benzamidine, 0.23 mM phenylmethylsulfonyl fluoride, 0.5 g/ml aprotinin, and 0.5 g/ml leupeptin). Cell lysates were solubilized with 1% Triton X-100 and centrifuged in a Beckman TLA 100.2 rotor at 75,000 rpm for 15 min. Cell extracts were incubated for 1 h at 4°C with GST fusion proteins (ϳ5 g/ml of each protein) prebound to glutathione-Sepharose beads. After incubation, samples were washed three times in Buffer C containing 1% Triton X-100, and bound proteins were resolved by SDS-PAGE and processed for Western blot analysis.
Immunoprecipitation Analysis-Cytosolic or crude Triton-soluble lysates were prepared from different tissues as described above. For the cytosolic samples, Triton X-100 was added to 1%. The sample were then precleared by incubation with protein A-Sepharose, and an aliquot of the precleared extract was incubated overnight at 4°C with normal rabbit serum or different antibodies precoupled to protein A-Sepharose. Beads were washed in Buffer A containing 1% Triton X-100, and proteins specifically bound to the beads were eluted and processed for SDS-PAGE. For immunodepletion experiments, an identical protocol was used except that 100 g of cytosolic extract was added to the beads and the material that did not bind to the beads was processed for SDS-PAGE.
Primary Cell Culture-Dissociated cell cultures were prepared from the CA3 and dentate regions of hippocampi from P1 rat pups as described (31). Following 2 days in culture, the hippocampal neurons were fixed with PBS containing 4% paraformaldehyde and 4% sucrose for 15 min at room temperature. The cells were permeabilized with 0.1% Triton X-100 for 5 min and blocked with PBS containing 1% normal goat serum. For explant cultures, dorsal root ganglia were dissected from E15 Harlan Sprague-Dawley rats. The ganglia were cut in half and placed on coverslips coated with 10 g/ml natural mouse laminin (Life Technologies, Inc.). The cultures were incubated in F-12 culture medium containing 10% fetal calf serum supplemented with 50 ng/ml 7S nerve growth factor. Following 1-2 days in culture, the cells were fixed with PBS containing 3% paraformaldehyde and 0.3 M sucrose for 30 min at room temperature. The cells were permeabilized with 0.2% Triton X-100 for 3 min and blocked with PBS containing 5% bovine serum albumin and 5% normal goat serum. Following blocking, both culture types were processed for immunofluorescence with antibodies 2704 and C23. In some cases, filamentous actin was detected with phalloidin-Alexa 488 (Jackson ImmunoResearch).
Immunofluorescence on COS-7 Cells-COS-7 cells were plated on poly-L-lysine-coated coverslips, transfected with LipofectAMINE 2000 (Life Technologies, Inc.), serum-starved overnight, and then left untreated or stimulated with 100 ng/ml EGF for 2 min at 37°C. Cells were washed twice in ice-cold PBS and processed for immunofluorescence as previously described (32) using antibodies 2704 or C23. Filamentous actin was detected with phalloidin-tetramethylrhodamine isothiocyanate (Sigma).

RESULTS AND DISCUSSION
We previously demonstrated that mSos interacts through its PRD with the endocytic protein intersectin, suggesting that intersectin may target mSos to Ras on the endocytic pathway (9,10,13). As the PRD of mSos contains multiple SH3 domainbinding consensus sites, we sought to identify additional mSos binding partners. Overlay of adult rat brain extracts with a GST fusion protein encoding the N-terminal half of the mouse Sos1 PRD (GST-NT) (amino acids 1111-1228) identified three proteins of 120 (p120), 90 (p90), and 50 (p50) kDa (Fig. 1A). None of the bands were detected with a GST fusion protein encoding the mouse Sos1 PRD C-terminal half (GST-CT) (amino acids 1223-1341) or with GST alone (Fig. 1A). Previously, Leprince et al. (8) identified amphiphysin II as a mSos1 binding partner. To determine whether p120 and p90 correspond to amphiphysin I (120 kDa) and amphiphysin II (90 kDa), respectively, we used GST-NT, GST-CT, or GST alone in pull-down assays with soluble rat brain extracts. Western blots of the pull-downs demonstrated that both amphiphysin I and II bind specifically to GST-NT (Fig. 1B), suggesting that they represent p120 and p90. Intersectin also bound selectively to GST-NT, whereas Grb2 bound equally well to GST-NT and GST-CT (data not shown). Surprisingly, neither Grb2 nor intersectin was detected on the overlay assays, possibly due to lower levels of expression in brain extracts than the amphiphysins or p50. The abundant SH3 domain-containing protein, endophilin 1, which is readily detected on overlays with the PRDs of synaptojanin 1 (33) and dynamin 1 (34), was not seen on the overlays with the PRD of mSos1, further demonstrating the specificity of the interactions detected.
To characterize p50, the major mSos1 binding partner identified, we performed overlays with GST-NT on tissue extracts. p50 was detected in brain but not in a variety of nonneuronal tissues ( Fig. 2A). This is consistent with the distribution of mSos1 that is expressed at higher levels in brain than in other tissues (9). Within brain, subcellular fractionation revealed p50 in both soluble and particulate fractions (Fig. 2B). The greatest enrichment was seen in the second lysed supernatant fraction (Fig. 2B, LS 2 ), which contained soluble proteins generated from the lysis of crude synaptosomes. This distribution is similar to that previously described for mSos1 (9), as well as that of the presynaptically enriched endocytic regulatory enzymes dynamin 1 and synaptojanin 1 (29).
To identify p50, we used the mSos1 GST fusion proteins to affinity purify mSos1-binding proteins from a soluble rat brain extract. As determined by Coomassie Blue staining, a 50-kDa band that bound to GST-NT but not to GST-CT or to GST alone was the major affinity-selected protein (Fig. 3). Minor bands at 120, 90, and 70 kDa were also weakly detected. The 50-kDa band was excised from the gel and subjected to trypsin digestion, and the fragments were analyzed by matrix assisted laser desorption ionization mass spectrometry at the W. M. Keck Foundation Biotechnology Resource Laboratory at Yale University. A ProFound search of the peptide masses provided a tentative identification for p50 as PACSIN 1/syndapin I. PAC-SIN 1 was identified based on its up-regulation during neuronal differentiation in mouse (22), whereas its rat orthologue syndapin I was identified through its SH3 domain-dependent interaction with dynamin 1 (23). PACSIN 1, which contains an SH3 domain at its C terminus, is a neuron-specific protein with a predicted molecular mass of 50 kDa, consistent with its identification as p50. To support this identification, we performed overlay assays with GST-NT or GST alone on cells transfected with a cDNA encoding His 6 -tagged full-length PACSIN 1. GST-NT specifically interacted with a protein species that perfectly co-migrated with PACSIN 1, as detected with an anti-His 6 Western blot (Fig. 4A), demonstrating that FIG. 1. Overlay analysis of mSos1-binding proteins. A, proteins of cytosolic fractions from rat brain were separated by SDS-PAGE, transferred to nitrocellulose, and overlaid with GST fused to amino acids 1111-1228 (GST-NT) or amino acids 1223-1341 (GST-CT) of the PRD of mouse Sos1 or with GST alone. The migratory positions of three major bands that bind to GST-NT and that of the molecular weight standards are indicated on the right and left, respectively. B, a crude Triton X-100-soluble rat brain extract was incubated with GST, GST-NT, or GST-CT conjugated to glutathione-Sepharose beads. Proteins specifically bound to the beads (B) along with aliquots of the brain extract (starting material (SM)) and equal amounts of the unbound material (void (V)) were processed for Western blot with an antibody that recognizes both amphiphysin I (amph I) and amphiphysin II (amph II). The molecular masses of amphiphysin I and II are indicated on the left .   FIG. 2. Tissue and subcellular distribution of p50. A, proteins of crude Triton X-100-soluble extracts from various adult rat tissues (200 g/tissue) were separated by SDS-PAGE, transferred to nitrocellulose, and overlaid with GST-NT. B, proteins of brain subcellular fractions (100 g/fraction) were separated by SDS-PAGE, transferred to nitrocellulose, and overlaid with GST-NT. Subcellular fractions were prepared as described (28). H, homogenate; P, pellet; S, supernatant; LP, lysed pellet; LS, lysed supernatant. For both A and B, the migratory position of p50 is indicated by the arrow on the right. PACSIN 1 directly interacts with mSos1. The identity of p50 as PACSIN 1 was confirmed with the demonstration that immunodepletion of PACSIN 1 from brain extracts using an anti-PACSIN 1 antibody completely depletes p50, as determined by GST-NT overlay (Fig. 4B).
In addition to PACSIN 1, a second member of the PACSIN family, referred to as PACSIN 2, has been recently described (24,25). In contrast to PACSIN 1, which is expressed exclusively in neurons, PACSIN 2 is expressed in brain and nonneuronal tissues (24,25) (Fig. 5A). Because mSos1 is enriched in brain but is also expressed in nonneuronal tissues (9), we hypothesized that mSos-PACSIN 2 interactions may occur both within and outside the nervous system. To explore this question, we performed pull-down assays from Triton X-100-solubilized extracts prepared from brain, as well as from lung, which expresses high level of PACSIN 2 ( Fig. 5A) (26). As expected, we detected binding of PACSIN 2 from both tissues to the mSos1 PRD (Fig. 5B). In fact, PACSIN 2 is likely to represent the 70-kDa band that was weakly detectable on the Coomassie blue stained pull-downs from brain extracts using mSos GST-NT (Fig. 3). Surprisingly, the level of PACSIN 2 recovered on the mSos1-PRD fusion protein was consistently greater when using brain versus lung extracts, even though PACSIN 2 was more abundant in the extracts from lung (Fig. 5, A and B). Moreover, a second, slightly smaller band that reacted with the PACSIN 2 antibody was detected in the GST-NT pull-downs from brain extracts but not from lung extracts (Fig. 5B). This protein may represent the short splice variant of PACSIN 2 previously described in rat tissues (25). Comparable amounts of the 70-kDa protein were pulled down from lung and brain extracts with an anti-PACSIN 2 antibody, suggesting that PACSIN 2 is equally accessible in both tissues (Fig. 5B). Thus, the differential binding of the long form of PACSIN 2 to the mSos1 PRD in brain compared with lung reveals a tissuespecific regulation of the interaction. The reason for this observation is currently unknown. However, it is possible that a tissue-specific posttranslational modification of PACSIN 2 alters its affinity for the mSos PRD.
To address whether the observed interactions are dependent on the classical SH3 domain binding interface, we generated point mutations in the SH3 domains of PACSIN 1 and 2 that converted proline to leucine (P434L for PACSIN 1; P478L for PACSIN 2). Comparable mutations in the Caenorhabditis elegans Grb2 homologue sem-5 cause a lethal phenotype by preventing sem-5 interactions with its PRD-containing binding partners (35). Using wild-type and mutated PACSIN 1 and 2 expressed as GST fusion proteins, we performed pull-down assays from brain extracts (Fig. 5C). mSos1 was found to interact with both wild-type fusion proteins, whereas the proline to leucine mutations abolished mSos1 binding to both PACSINs, demonstrating that the interactions are specifically mediated through the PACSIN SH3 domains.
To explore the potential interaction between PACSIN 1 and mSos1 in situ, we performed co-immunoprecipitation experiments from rat brain extracts. Immunoprecipitation of PACSIN 1 led to co-immunoprecipitation of mSos1 (Fig. 6). The interaction was specific, as no mSos1 precipitated in the presence of normal rabbit serum, and the abundant brain protein tubulin was not detected in the anti-PACSIN 1 immunoprecipitates (Fig. 6). Only a limited percentage of the total mSos1 in the brain extract co-immunoprecipitated with PACSIN 1. This is not surprising given that PACSIN 1 interacts through its SH3 domain with multiple binding partners, including the abundant brain proteins dynamin 1 and synaptojanin 1 (23). In fact, dynamin 1 was found to strongly co-immunoprecipitate with PACSIN 1 (Fig. 6). As the interactions between PACSIN 1 and its various SH3 domain-binding partners are likely to be competitive, the PACSIN-mSos interaction may be restricted to specific subcellular domains that are enriched for mSos1 relative to other PACSIN binding partners. Alternatively, the PACSINs may be at the core of large protein complexes in which they simultaneously interact with multiple binding partners. Consistent with the later possibility, it has been recently demonstrated that the PACSINs can self-associate to form homo-and hetero-oligomers (26).
To further examine the potential for interactions between PACSIN 1 and mSos1 in situ, we sought to determine whether the proteins were co-distributed in neurons. Immunofluorescence analysis of hippocampal neurons at 2 days in vitro with polyclonal antibodies against each protein revealed strong staining in the neuronal cell bodies, with fluorescent punctae observed along the length of the neurites and in growth cones (Fig. 7, A and C). Staining for both proteins extended into filopodia emanating from the growth cones (Fig. 7, B and D). To examine the localization within growth cones in more detail, we performed immunofluorescence analysis of primary rat dorsal root ganglia neurons maintained in culture for 1 day. Similar to hippocampal neurons, both PACSIN 1 and mSos1 were de-tected in dorsal root ganglia growth cones and were seen to extend into filopodia (Fig. 8). Interestingly, co-staining with phalloidin, which reveals filamentous actin, demonstrated that both mSos1 (Fig. 8A) and PACSIN 1 (Fig. 8B) were strongly co-localized with actin filaments at the plasma membrane and throughout the length of the filopodia. PACSIN has been demonstrated to localize at sites of high actin turnover, including filopodia and filopodial tips in nonneuronal cells (25). Further, overexpression of full-length PACSIN causes filopodia formation in an N-WASP-dependent manner, although the mechanism of this activation is unknown (25). The co-localization of mSos1 with PACSIN 1 in filopodia suggests that mSos1 may cooperate with PACSIN 1 in filopodia formation or function. This role may be particularly relevant during neuronal development, as actin-dependent filopodial dynamics are critical in the response of neuronal growth cones to extracellular guidance cues (36). Filopodia formation is dependent on N-WASP (37), which is activated by interactions with SH3 domains (38), as well as by binding to phosphatidylinositol FIG. 5. Specific interaction of mSos1 with the SH3 domains of PACSIN 1 and 2. A, proteins of crude Triton-soluble extracts from various adult rat tissues (200 g/tissue) were separated by SDS-PAGE, transferred to nitrocellulose, and processed for Western blot using antibodies against PACSIN 2. B, GST-NT conjugated to glutathione-Sepharose beads (GST-NT) and anti-PACSIN 2 antiserum precoupled to protein A-Sepharose beads (PACSIN 2 i.p.) were incubated with crude Triton-soluble extracts from rat lung and brain. Proteins specifically bound to the beads (bead) along with aliquots of the lung and brain extract (starting material (SM)) were subjected to SDS-PAGE and processed for Western blot using antibodies against PACSIN 2. C, a mouse brain extract was incubated with glutathione-Sepharose beads conjugated to GST alone or GST fused to full-length wild-type (wt) PACSINs or PACSINs with point mutations in their SH3 domains (PACSIN 1, proline 434 mutated to leucine (P434L); PACSIN 2, proline 478 mutated to leucine (P478L)). Material specifically bound to the beads was resolved by SDS-PAGE along with an aliquot of the brain extract (starting material (SM)) and processed for Western blot using antibodies against mSos1.
We next investigated the possibility of direct co-localization of PACSIN 1 and mSos1 in actin-rich structures. Fibroblasts are well established to form membrane ruffles upon treatment with growth factors (43,44). We therefore co-transfected COS-7 cells with FLAG-tagged mSos1 and His 6 -tagged PACSIN 1 and examined their distribution before and after EGF treatment. Interestingly, both PACSIN 1 and mSos1 relocalized from a predominantly cytoplasmic distribution to become concentrated and co-localized at membrane ruffles following EGF treatment (Fig. 9A). Co-staining of PACSIN 1 transfected cells with anti-PACSIN 1 antibody and fluorescent phalloidin confirmed that the structures at which PACSIN 1 and mSos1 were co-localized were actin-rich membrane ruffles (Fig. 9B). Growth factor-induced ruffle formation is mediated by Ras-dependent Rac activation (43). Recent data suggest that mSos plays an important dual role in coupling Ras to Rac (18). Through its CDC25 homology domain, mSos activates Ras, which in turn activates phosphatidylinositol 3-phosphate kinase (45). The products of phosphatidylinositol 3-phosphate kinase catalytic activity stimulate the GEF activity of mSos toward Rac, causing Rac activation (17,18). Indeed, mSos has been demonstrated to function directly in growth factor-induced membrane ruffle formation (18,21). Our observation that full-length mSos1 targets to sites of Rac activation in response to EGF stimulation is consistent with its role in Rac-dependent actin reorganization. The co-localization of PACSIN 1 with mSos1 suggests that through direct protein interactions, PACSIN may regulate mSos function in ruffle FIG. 10. Regulation of the PACSIN 1-mSos1 interaction by phosphorylation. Serum-starved COS-7 cells were treated with 100 ng/ml of EGF for 5 min at 37°C (EGF ϩ). When indicated, prior to the EGF treatment, cells were preincubated with PD-098059 (PD ϩ). Lysates from cells were incubated with GST-PACSIN 1 SH3 coupled to glutathione-Sepharose beads, and material specifically bound to the beads was processed for SDS-PAGE, along with an aliquot of cell lysate (starting material (SM)). The migratory position of mSos1 and the upwardly shifted phosphorylated mSos1 (phospho-mSos1) and that of GST-PACSIN 1 SH3 domain are indicated on the right.  were treated with 100 ng/ml EGF for 2 min (ϩEGF). PACSIN 1 was detected with an anti-PACSIN 1 rabbit antibody, and mSos1 was detected with anti-FLAG monoclonal antibody. F-actin was detected using phalloidin-tetramethylrhodamine isothiocyanate (phalloidin). Magnification, ϫ 630.
formation. In fact, PACSIN 1 interacts with mSos1 within the same region that mediates mSos1 interactions with Eps8 and E3b1 (18). The binding of these proteins to mSos stimulates mSos GEF activity toward Rac (18). Thus, PACSIN interactions with mSos may play a comparable modulatory role.
We next sought to determine whether the interaction between PACSIN 1 and mSos1 was subject to regulation in response to growth factor stimulation. Several laboratories have demonstrated that MEK-dependent feedback phosphorylation of mSos leads to Ras desensitization by inducing the dissociation of mSos from Grb2 and Shc (10, 28, 46 -49). We thus examined mSos1-PACSIN 1 interactions following MEK-induced mSos1 phosphorylation. Treatment of COS-7 cells with EGF led to a small but highly reproducible upward shift in mSos1 mobility (Fig. 10). This shift is characteristic of MEKinduced mSos1 phosphorylation (28,49) and was blocked by preincubation with the MEK inhibitor PD-098059 (Fig. 10). Interestingly, phosphorylated mSos1 showed less binding to the SH3 domain of PACSIN 1 than did nonphosphorylated mSos1 ( Fig. 10; representative of three separate experiments). In contrast, the binding of N-WASP to the PACSIN 1 SH3 domain was identical under the various treatment conditions (data not shown). Thus, phosphorylation of mSos1 negatively regulates its interaction with PACSIN 1. Interactions between SH3 domain-containing proteins and their proline-rich binding partners are often regulated by phosphorylation. For example, phosphorylation of dynamin 1 and synaptojanin 1 reduces their interactions with the SH3 domains of the amphiphysins, allowing for regulation in their targeting to clathrin-coated pits (50).
The activation of Ras and the formation of membrane ruffles occurs rapidly following EGF treatment of COS-7 cells (both can be detected within 30 s). 2 In contrast, MEK-dependent feedback phosphorylation of mSos and the dissociation of the mSos-Grb2 complex is detectable only after several minutes (48). Thus, in parallel with the mSos-Grb2 interaction, it is likely that mSos and PACSIN are initially in association following their EGF induced translocation to membrane ruffles. Feedback phosphorylation via the Ras/MEK-dependent pathway, which appears to be a general mechanism to attenuate Ras signaling, would subsequently terminate the PACSIN-mSos interaction.
It is well established that Rac functions downstream of Ras in Ras-mediated signaling to alterations in the actin cytoskeleton (51). The ability of mSos to function as a GEF for Rac is central to the transduction of signals from Ras to Rac (16) and suggests the need for adaptor proteins involved in targeting mSos to sites of actin dynamics. Interestingly, Eps8, which activates the mSos GEF activity toward Rac following its indirect binding to mSos, also binds to actin (16). Thus, it is tempting to speculate that the PACSINs, which localize to sites of actin turnover and regulate actin cytoskeletal dynamics, may play a role in targeting mSos to sites of actin dynamics. Alternatively, interactions between PACSIN and mSos may allow for coordinated activities of the two proteins in regulation of the actin cytoskeleton. Future experiments will be aimed at testing these various hypotheses.