January 2010 Biol. Pharm. Bull. 33(1) 35—39 (2010) 35

Proline-Rich Domain Plays a Crucial Role in Extracellular Stimuli- Responsive Translocation of a Cdc42 Guanine Nucleotide Exchange Factor, FGD1

Toshiyuki OSHIMA, Tomofumi FUJINO, Ken ANDO, and Makio HAYAKAWA* School of Pharmacy, Tokyo University of Pharmacy and Life Sciences; 1432–1 Horinouchi, Hachioji, Tokyo 192–0392, Japan. Received September 29, 2009; accepted October 14, 2009

We previously demonstrated that FGD1, the Cdc42 guanine nucleotide exchange factor (GEF) responsible for faciogenital dysplasia, and its homologue FGD3 are targeted by the ubiquitin ligase SCFFWD1 upon phospho- rylation of two serine residues in their DSGIDS motif and subsequently degraded by the proteasome. FGD1 and FGD3 share highly homologous Dbl homology (DH) and adjacent pleckstrin homology (PH) domains, both of which are responsible for GEF activity. However, their function and regulation are remarkably different. Here we demonstrate extracellular signal-responsive translocation of FGD1, but not FGD3. During the wound-healing process, translocation of FGD1 to the leading edge membrane occurs in cells facing to the wound. Furthermore, epidermal growth factor (EGF) stimulates the membrane translocation of FGD1, but not FGD3. As the most striking difference, FGD3 lacks the N-terminal proline-rich domain that is conserved in FGD1, indicating that proline-rich domain may play a crucial role in signal-responsive translocation of FGD1. Indeed, there is a facio- genital dysplasia patient who has a missense mutation in proline-rich domain of FGD1, by which the serine residue at position 205 is substituted with isoleucine. When expressed in cells, the mutant FGD1 with S205/I sub- stitution fails to translocate to the membrane in response to the mitogenic stimuli. Thus we present a novel mech- anism by which the activity of FGD1, a GEF for Cdc42, is temporally and spatially regulated in cells. Key words FGD1; FGD3; guanine nucleotide exchange factor; proline-rich domain; Cdc42; Rho family G protein

The Rho family small guanosine 5-triphosphate (GTP)- activation.9) However, we demonstrated that the inducible ex- binding proteins (G-proteins), consisting mainly of the Rho, pression of the full-length FGD1 led to the formation of Rac, and Cdc42 subfamilies, regulates various actin cy- striking filopodia structures whereas that of full-length FGD3 toskeleton-dependent cell functions, including cell shape induced broad sheet-like protrusions, i.e., remarkable lamel- change, cell migration, cell adhesion, and cytokinesis.1—3) lipodia structures.8) These results indicate that FGD1 and Like all G-proteins, Rho family proteins act as binary FGD3 play different roles in cells but undergo the same de- switches by cycling between an inactive (GDP-bound) and an struction pathway. active (GTP-bound) conformation state. Although higher sequence similarity between FGD1 and Guanine nucleotide exchange factors (GEFs) stimulate the FGD3 is observed in their DH and adjacent PH domains (i.e., exchange of GDP for GTP to generate active forms of Rho amino acid residue identity is 70.0% and 60.6%, respec- proteins. A family of candidate GEFs for Rho proteins has tively),9) there exist striking differences in their sequence in been identified based on the homology with CDC24, a pro- other regions; for example, most prominently, FGD3 lacks tein genetically as an upstream activator of CDC42 in yeast.4) the N-terminal proline-rich domain conserved in FGD1. Dbl, isolated as an oncogene in a focus formation assay using Proline-rich sequences are known to bind to specific pro- NIH 3T3 cells transfected with DNA from a human diffuse tein interaction modules such as Src homology 3 (SH3), B-cell lymphoma, was identified as the first mammalian Rho WW, and Ena/VASP homology 1 (EVH1) domains.10—12) GEF showing significant sequence similarity to CDC24.5) These different modules recognize different types of proline- The domain conserved in both Dbl and CDC24 is now recog- rich sequences and play important roles in various signal nized as the Dbl homology (DH) domain and almost 60 DH- transduction pathways. Phenotypic and molecular analyses domain-containing proteins have been found.6) The activities revealed that a patient with missense mutation occurred in of GEFs must be tightly regulated and each member of the the proline-rich domain in FGD1 showed typical clinical fea- GEF family is likely to have a unique mechanism for activa- tures as AAS.13) In this mutant FGD1, the serine residue at tion and inactivation. position 205 was replaced with isoleucine. Although the In our previous reports, we revealed that FGD1, a GEF for DHPH domains necessary for the GEF activity remain in- Cdc42 and known as the faciogenital dysplasia (or Aarskog- tact in FGD1 (S205/I), impaired interaction with other mole- Scott syndrome; AAS) gene product, and FGD3, a homo- cules via proline-rich domain may result in the loss of func- logue of FGD1, underwent proteasomal degradation through tion of this mutant FGD1. It is likely that the activity of the polyubiquitination by the ubiquitin ligase SCFFWD1.7,8) FGD1 may be turned on at an appropriate intracellular com- FGD3 contains adjacent DH and pleckstrin homology (PH) partment through the interaction with the upstream signaling domains, both of which are highly homologous to those of modules that are activated in a temporally and spatially regu- FGD1. Pasteris et al. reported that DHPH domains of lated manner. In this study, we examined the extracellular FGD3 induced long finger-like protrusions, filopodia, as stimuli-regulated function of FGD1. Immunostaining analy- DHPH domains of FGD1 did, suggesting that both FGD1 ses revealed that FGD1 translocates to the leading edge and FGD3 share the similar GEF activity to induce Cdc42 membrane in migrating cells. The translocation of FGD1

∗ To whom correspondence should be addressed. e-mail: [email protected] © 2010 Pharmaceutical Society of Japan 36 Vol. 33, No. 1 also occurs in cells exposed to epidermal growth factor GCAGCAGTATCA-3 (sense) and 5-GATGGCCTCAG- (EGF). By fractionating cells, we demonstrated that FGD1 CTCTGGGGAC-3 (anti-sense). For the template cDNA, translocates to the membrane fraction in cells under mito- pBluescript SK()-FLAG-FGD1(SA)7) was used. The am- genic condition while FGD3 and FGD1 (S205/I) do not. These plified sequence was verified by dideoxy sequencing, and results indicate that FGD1 is targeted to the membrane upon then FLAG-FGD1(SA+S205/I) cDNA was subcloned into the interaction with proline-binding modules that are acti- pTRE vector (Clontech) and resultant plasmid was termed vated at the specific compartment in the membrane. pTRE-FLAG-FGD1(SAS205/I). HeLa Tet-Off cells purchased from Clontech were co- MATERIALS AND METHODS transfected with pTRE-FLAG-FGD1(SAS205/I) and pTK- Hyg. The hygromycin-resistant clonal cell lines were tested Reagents Tetracycline (Tc) and anti-FLAG M2 Ab were for Tc-regulated FLAG-tagged FGD1 expression. obtained from Sigma-Aldrich. Fluorescein isothiocyanate (FITC)-conjugated AffiniPure goat anti-mouse immunoglob- RESULTS ulin G (IgG) was the product of Jackson ImmunoResearch Laboratories. Alexa Fluor 594 phalloidin was obtained from In our previous study, we demonstrated that FGD1 and Molecular Probes. ProteoExtract Subcellular Proteome Ex- FGD3 are targeted by the ubiquitin ligase SCFFWD1 upon traction Kit was purchased from Calbiochem. phosphorylation of two serine residues in the conserved DS- Cell Culture HeLa Tet-Off-derived stable cell lines that GIDS motif and then subsequently degraded by the protea- inducibly overexpress either FGD1(SA) or FGD3(SA), in some.7,8) Sequence similarity between FGD1 and FGD3 is which the destruction sequence DSGIDS is replaced with also observed in their DH and adjacent PH domain (i.e., DAGIDA were maintained in 5% fetal calf serum (FCS)/ amino acid residue identity is 70.0% and 60.6%, respec- Dulbecco’s modified Eagle’s medium (DMEM) supple- tively).9) In contrast, there are striking differences in their se- mented with 100 mg/ml G418, 200 mg/ml hygromycin, and quence in other region; most prominently, FGD3 lacks the N- 2 mg/ml Tc as described previously.8) To induce FGD1(SA) terminal proline-rich domain conserved in FGD1 (Fig. 1A). or FGD3(SA), cells were trypsinized and then cultured in the We previously established the HeLa Tet-Off-derived stable absence of Tc for 72 h. cell lines that inducibly overexpress either FGD1(SA) or Immunostaining Immunostaining analyses were carried FGD3(SA), in which the destruction sequence DSGIDS is out as follows: cells cultured on glass coverslips were fixed replaced with DAGIDA and found that cells expressing with 4% paraformaldehyde, then permeabilized with 0.25% FGD1(SA) showed representative filopodia structure whereas Triton X-100 and blocked with 10% skim milk in PBS. The cells expressing FGD3(SA) showed broad and sheet-like pro- cells were incubated with anti-FLAG M2 antibody, followed trusions called lamellipodia, demonstrating that FGD1 and by the double staining with FITC-conjugated goat anti- FGD3 induce different types of actin cytoskeleton reorgani- mouse IgG and Alexa Fluor 594 phalloidin for 1 h. Fluores- zation.8) cence images were captured by CCD camera (QICAM In the present study, we further examined the difference of FAST1394; QIMAGING) mounted on a Leica DM IRB intracellular distribution of FGD1 and FGD3. Confluent microscope using IP Laboratory imaging software (Scan- monolayer of cells expressing FGD1(SA) or FGD3(SA) were alytics). wounded and further cultured for 30 min. Cells were then im- Subcellular Fractionation Cells seeded on 35-mm munostained to detect the FLAG epitope. As shown in Fig. dishes were harvested and subcellular fractionation was car- 1B, translocation of FGD1 to the leading edge membrane ried out using ProteoExtract Subcellular Proteome Extraction was prominently observed in FGD1(SA)-expressing cells Kit. Briefly, cells were washed twice with ice-cold Wash that faced to the wound, whereas it was not obvious in Buffer and then cytosolic proteins were extracted with 0.4 ml FGD3(SA)-expressing cells. Interestingly, mitogenic stimuli- ice-cold Extraction Buffer I. Subsequently, membranes and dependent translocation of FGD1, but not FGD3, was also membrane organelles were solubilized with 0.4 ml ice-cold observed. As reported previously,8) control FGD1(SA)- Extraction Buffer II without impairing the integrity of nu- expressing cells showed long, finger-like protrusions called cleus and cytoskeleton. Next, nuclear proteins were enriched filopodia (Fig. 1C, “actin”). The treatment of cells with EGF with 0.2 ml ice-cold Extraction Buffer III. Finally, compo- induced membrane ruffles that were significantly stained nents of the cytoskeleton were solubilized with 0.2 ml room with anti-FLAG antibody (Fig. 1C, “FLAG”). In contrast, temperature Extraction Buffer IV. EGF did not change the intracellular distribution of FGD3 Immunoblotting Extracts of cytosolic, membrane, nu- (Fig. 1D). clear and cytoskeletal fractions were resolved by SDS-PAGE We next examined the intracellular distribution of FGD1 and then subjected to immunoblotting analyses. The im- and FGD3 by the use of subcellular fractionation kit. FGD1 munocomplexes on the polyvinylidene difluoride membranes was mainly detected in cytosolic fraction during the culture were visualized using enhanced chemiluminescence detec- without FCS for 24 h while a small amount of FGD1 was tion. also found in cytoskeletal fraction (Fig. 2A). In contrast, the Site-Directed Mutagenesis and Establishment of Stable amount of FGD1 in membrane fraction was significantly in- Cell Lines Expressing FGD1(SAS205/I) In order to re- creased when cells were kept cultured in the presence of 5 % place Ser205 in FGD1 with an isoleucine, we carried out FCS (Fig. 2A). Unlike FGD1, FCS did not affect the distri- site-directed mutagenesis by PCR using the following bution of FGD3 that were detected in both cytosol and mem- primers designed in inverted tail-to-tail direction to amplify brane fractions (Fig. 2B). These results suggest that FGD1 vector together with the target sequence: 5-ACTAGTTCT- but not FGD3 will change its intracellular distribution in re- January 2010 37

Fig. 1. FGD1 but not FGD3 Translocates to the Membrane in Response to Extracellular Stimuli (A) Schematic representation of FGD1, FGD3, and their mutants used in this study. FGD1(SA) and FGD3(SA) bear Ser to Ala amino acid substitutions within the SCFFWD1 pu- tative consensus binding site. In addition, a Ser to Ile substitution is made at position 205 in FGD1(SAS205/I). Domains shown are “Pro-rich” (proline-rich domain), “DH” (Dbl homology domain), “PH” (pleckstrin homology domain), “FYVE” (domain present in Fab1, YOTB, Vac1, and EEA1). (B) HeLa Tet-Off-FGD1(SA) or HeLa Tet-Off-FGD3(SA) cells were seeded on coverslips and then cultured in the absence of Tc for 72 h. Monolayers of confluent cultures were lightly scratched with a 1000-ml disposable plastic pipette tip. After having been washed with DMEM, the wound regions were incubated in 5% FCS/DMEM for 30 min. After fixing cells with 4% paraformaldehyde, cells were permeabi- lized, and then analyzed by immunostaining to detect FLAG epitope as described in Materials and Methods. Lower panels represent the enlarged images of the boxed area shown in upper panels (scale bar, 30 mm). Arrow heads indicate the membrane-translocated FGD1. (C) HeLa Tet-Off-FGD1(SA) cells were seeded on coverslips and then cultured in the ab- sence of Tc for 48 h. Cells were subjected to the serum starvation for 24 h followed by the stimulation with 20 ng/ml EGF for 30 min. Double staining to detect the FLAG epitope and the actin cytoskeleton was carried out as described in Materials and Methods. Arrow heads indicate the membrane-translocated FGD1 (scale bar, 30 mm). (D) HeLa Tet-Off- FGD3(SA) cells seeded on coverslips were stimulated with EGF and subjected to double staining to detect FLAG epitope and actin cytoskeleton as described in (C). sponse to the mitogenic stimuli. Indeed, translocation of one of the patients with typical AAS phenotype.17) Proline- FGD1 in the membrane fractions was clearly demonstrated rich regions often contain serine or threonine residues, sug- in a time-dependent manner after cells were treated with gesting that proline-rich sequence-directed protein kinase EGF (Fig. 3A). In contrast, FGD3 did not change its distribu- may bind and phosphorylate these residues. As shown in Fig. tion whether or not cells were treated with EGF (Fig. 3B). 4A, not only in human FGD1 but also in FGD1 from other Since FGD3 lacks the N-terminal proline-rich region while species, the serine residue at position 205 (position 36 in the FGD1 contains it, we next examined the role of proline-rich case of rat) is conserved. Therefore, we established a HeLa- domain of FGD1 in mitogenic stimuli-responsive transloca- Tet Off cell line that inducibly expresses a mutant mouse tion to the membrane. FGD1 protein in which the serine residue at position 205 is FGD1 was originally identified and characterized by posi- replaced with isoleucine in addition to DAGIDA replace- tional cloning in a family in which the phenotype of the de- ment. As shown in Fig. 4B, several cells expressing FGD1 velopmental disorder was associated with an X;8 transloca- (SAS205/I) showed two or three broad sheet-like lamellae at tion.14,15) Genetic analyses revealed that different types of opposite sides of cells, as in the case of FGD3-expressing point mutations occurred in FGD1 are also responsible for cells. Furthermore, in contrast to the case of wild type AAS.16,17) In one of such mutations, we focused on the muta- FGD1, FGD1 (SAS205/I) was detected in cytosolic fraction tion in proline-rich domain of FGD1 (S205/I). even in the presence of 5% FCS during the culture (Fig. 4C). Orrico et al. identified the S205/I mutation in FGD1 from These results demonstrated that S205 conserved in proline- 38 Vol. 33, No. 1

Fig. 2. Intracellular Distribution of FGD1 and FGD3 in the Presence or Absence of FCS (A) HeLa Tet-Off-FGD1(SA) cells were cultured in the absence of Tc for 48 h and subjected to serum starvation or left untreated for 24 h. Subcellular fractionation was then carried out as described in Materials and Methods. Cytosolic proteins were ex- tracted with 0.4 ml ice-cold Extraction Buffer I. Membranes and membrane organelles were solubilized with 0.4 ml Extraction Buffer II. Nuclear proteins were extracted with 0.2 ml Extraction Buffer III. Components of the cytoskeleton were solubilized with 0.2 ml Extraction Buffer IV. The volume of nuclear and cytoskeletal fractions were then adjusted to 0.4 ml with buffer III or IV, and 20 ml aliquots from each fraction were sub- jected to SDS-PAGE followed by the immunoblotting to detect FLAG epitope. (B) HeLa Tet-Off-FGD3(SA) cells were cultured in the absence of Tc for 48 h and sub- jected to serum starvation or left untreated for 24 h. Subcellular fractionation was then carried out as described in (A).

Fig. 4. Mutant FGD1 with S205/I Substitution Fails to Translocate to the Membrane in Response to the Mitogenic Stimuli (A) Alignment of amino acid sequences conserved in proline-rich domain of FGD1 from different species. Black shading indicates sequence identity. (B) HeLa Tet-Off- FGD1(SAS205/I) cells were seeded on coverslips and then cultured in the absence of Tc for 72 h. Double staining to detect the FLAG epitope and the actin cytoskeleton was carried out as described in Materials and Methods. Arrow heads indicate the mem- brane-translocated FGD1 (scale bar, 30 mm). (C) HeLa Tet-Off-FGD1(SAS205/I) cells were cultured in the absence of Tc for 48 h and subjected to serum starvation or left un- treated for 24 h. Subcellular fractionation was then carried out as described in Fig. 2.

rich domain is critical for the membrane translocation of FGD1 in response to the mitogenic stimuli.

DISCUSSION

Fig. 3. EGF Induces Membrane Translocation of FGD1 but not FGD3 The Rho-family small G-proteins are defined by the pres- (A) HeLa Tet-Off-FGD1(SA) cells were cultured in the absence of Tc for 48 h and ence of a Rho-type GTPase-like domain and to date 22 subjected to serum starvation for 24 h. Cells were then stimulated with 20 ng/ml EGF 18) for indicated times. Subcellular fractionation was then carried out as described in Mate- human genes have been described. For outnumbering Rho rials and Methods. Protein assay was performed to normalize the protein content of proteins, more than 60 GEFs and approximately 80 GTPase- each sample: Cytosol, 4.3 mg/lane; Membrane, 2.5 mg/lane; Nucleus, 7.0 mg/lane; Cy- activating proteins (GAPs) are encoded in human toskeleton, 8.3 mg/lane, respectively. Samples were then subjected to SDS-PAGE fol- 6,19) lowed by the immunoblotting to detect FLAG epitope. (B) HeLa Tet-Off-FGD3(SA) genome. Therefore, the activities of individual Rho cells were cultured in the absence of Tc for 48 h and subjected to serum starvation for family proteins in cells should be regulated by these over- 24 h. Cells were then stimulated with 20 ng/ml EGF for indicated times. Subcellular fractionation was then carried out as described in Materials and Methods. Protein assay abundant GEFs and GAPs in upstream signals-dependent was performed to normalize the protein content of each sample: Cytosol, 3.9 mg/lane; manners. Membrane, 2.7 mg/lane; Nucleus, 8.0 mg/lane; Cytoskeleton, 8.5 mg/lane, respectively. Samples were then subjected to SDS-PAGE followed by the immunoblotting to detect In our previous study, we revealed that two homologous FLAG epitope. GEFs, FGD1 and FGD3, conserved the same destruction se- January 2010 39 quence “DSGIDS” that lead to the down-regulation of both to translocate FGD1 to membrane ruffles. molecules by GSK-3b/SCFFWD1/the proteasome-regulated We previously focused on the phosphorylation of two ser- pathway.7,8) When expressed as truncated forms consisting of ine residues in the sequence of DS283GIDS287, which leads to only DHPH domains that are necessary for the GEF activ- the destruction of FGD1.7) We should next focus on the phos- ity, both FGD1 and FGD3 induced similar morphological phorylation of S205, which may result in the stabilization of changes; i.e., large numbers of filopodial extensions were FGD1 at specific intracellular compartments in response to formed out of cell periphery while neither stress fiber nor the “motogenic” stimuli. lamellipodia were observed.9,20) However, we found that FGD1 induces remarkable filopodial extensions while FGD3 Acknowledgements We thank Hiroyuki Noguchi and induces thin and sheet-like protrusions known as lamelli- Norihisa Onodera for their excellent technical assistance. podia when expressed as the full length GEFs.8) This work was supported in part by a Grant from the Japan In our previous report, we described that the broad sheet- Private School Promotion Foundation. like protrusions recognized as lamellipodia in FGD3(SA)- expressing cells contained many microspikes.8) Interestingly, REFERENCES at the tips of several microspikes, FGD3 was clearly stained whereas actin filaments were not stained at all.8) Interconver- 1) Van Aelst L., D’Souza-Schorey C., Genes Dev., 11, 2295—2322 sion between microspikes, filopodia and retraction fibers may (1997). 2) Etienne-Manneville S., Hall A., Nature (London), 420, 629—635 be occurring at lamellipodial protrusions induced by FGD3. (2002). These broad sheet-like lamellae are often observed at oppo- 3) Malliri A., Collard J. G., Curr. Opin. Cell Biol., 15, 583—589 (2003). site side of a cell. The “lamellipodium” is sometimes re- 4) Bender A., Pringle J. R., Proc. Natl. Acad. Sci. U.S.A., 86, 9976—9980 ferred as the “leading edge” or the “leading lamella.”21) In (1989). the case of FGD3(SA)-expressing cells having two broad 5) Eva A., Aaronson S. A., Nature (London), 316, 273—275 (1985). 6) Schmidt A., Hall A., Genes Dev., 16, 1587—1609 (2002). sheet-like lamellae oriented in opposite directions, it is there- 7) Hayakawa M., Kitagawa H., Miyazawa K., Kitagawa M., Kikugawa fore difficult to know which of them represents the “leading K., Genes Cells, 10, 241—251 (2005). lamella.” In this study, we demonstrated that FGD1 was 8) Hayakawa M., Matsushima M., Hagiwara H., Oshima T., Fujino T., strongly stained at the membrane ruffles formed in cells Ando K., Kikugawa K., Tanaka H., Miyazawa K., Kitagawa M., Genes stimulated with EGF or in cells facing to the wound-healing Cells, 13, 329—342 (2008). 9) Pasteris N. G., Nagata K., Hall A., Gorski J. L., Gene, 242, 237—247 front (Figs. 1B, C). Translocation of FGD1 to the leading (2000). edge membranes was induced in an extracellular stimuli-de- 10) Kay B. K., Williamson M. P., Sudol M., FASEB J., 14, 231—241 pendent manner (Figs. 2A, 3A). Therefore, we can assume (2000). that FGD1, which contains signal-responsive proline-rich do- 11) Macias M. J., Wiesner S., Sudol M., FEBS Lett., 513, 30—37 (2002). main, responds to the “motogenic” stimuli and participates in 12) Li S. S., Biochem. J., 390, 641—653 (2005). 205 13) Orrico A., Galli L., Cavaliere M. L., Garavelli L., Fryns J. P., Crushell leading lamella formation. It should be noted that the S E., Rinaldi M. M., Medeira A., Sorrentino V., Eur. J. Hum. Genet., 12, residue conserved in proline-rich domain of FGD1 plays an 16—23 (2004). important role in the function of FGD1, since the S205/I 14) Pasteris N. G., Cadle A., Logie L. J., Porteous M. E., Schwartz C. E., mutation affected the patients with the characteristic AAS Stevenson R. E., Glover T. W., Wilroy R. S., Gorski J. L., Cell, 79, 205 669—678 (1994). phenotype. Indeed, S /I mutation significantly impaired 15) Pasteris N. G., Buckler J., Cadle A. B., Gorski J. L., Genomics, 43, the signal-responsive translocation of FGD1 (Fig. 4C). 390—394 (1997). Hou et al. described that FGD1 directly interacted with 16) Schwartz C. E., Gillessen-Kaesbach G., May M., Cappa M., Gorski J., SH3 domains of mouse cortactin and actin-binding protein 1 Steindl K., Neri G., Eur. J. Hum. Genet., 8, 869—874 (2000). (mAbp1) through the SH3-binding region located in FGD1 17) Orrico A., Galli L., Cavaliere M. L., Garavelli L., Fryns J. P., Crushell 22) E., Rinaldi M. M., Medeira A., Sorrentino V., Eur. J. Hum. Genet., 12, proline-rich domain. Using yeast two-hybrid and in vitro 16—23 (2004). binding techniques, they identified the FGD1 SH3-binding 18) Wennerberg K., Der C. J., J. Cell Sci., 117, 1301—1312 (2004). region as the amino acid sequence P157KPQVPPKP165.22) 19) Moon S. Y., Zheng Y., Trends Cell Biol., 13, 13—22 (2003). They also showed that full length FGD1, but not the trun- 20) Olson M. F., Pasteris N. G., Gorski J. L., Hall A., Curr. Biol., 6, cated FGD1 lacking proline-rich region (residues 1—358), 1628—1633 (1996). 21) Small J. V., Stradal T., Vignal E., Rottner K., Trends Cell Biol., 12, translocated to the ruffling membranes in response to PDGF. 112—120 (2002). Since the N-terminal proline-rich region (1—358) also con- 22) Hou P., Estrada L., Kinley A. W., Parsons J. T., Vojtek A. B., Gorski J. tains S205, we should emphasize that S205 as well as the SH3- L., Hum. Mol. Genet., 12, 1981—1993 (2003). binding region (residues 157—165) may be critical in order