GEP100�BRAG2: Activator of ADP-ribosylation factor 6for regulation of cell adhesion and actin cytoskeletonvia E-cadherin and �-cateninToyoko Hiroi*, Akimasa Someya†, Walter Thompson, Joel Moss, and Martha Vaughan*

Pulmonary-Critical Care Medicine Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD 20892

Contributed by Martha Vaughan, May 17, 2006

GEP100 (p100) was identified as an ADP-ribosylation factor (ARF)guanine nucleotide-exchange protein (GEP) that partially colocal-ized with ARF6 in the cell periphery. p100 preferentially acceler-ated guanosine 5[�-thio]triphosphate (GTP�S) binding by ARF6,which participates in protein trafficking near the plasma mem-brane, including receptor recycling, cell adhesion, and cell migra-tion. Here we report that yeast two-hybrid screening of a humanfetal brain cDNA library using p100 as bait revealed specificinteraction with �-catenin, which is known as a regulator ofadherens junctions and actin cytoskeleton remodeling. Interactionof p100 with �-catenin was confirmed by coimmunoprecipitationof the endogenous proteins from human HepG2 or CaSki cells,although colocalization was difficult to demonstrate microscopi-cally. �-Catenin enhanced GTP�S binding by ARF6 in vitro in thepresence of p100. Depletion of p100 by small interfering RNA(siRNA) treatment in HepG2 cells resulted in E-cadherin content3-fold that in control cells and blocked hepatocyte growth factor-induced redistribution of E-cadherin, consistent with a known roleof ARF6 in this process. F-actin was markedly decreased in normalrat kidney (NRK) cells overexpressing wild-type p100, but not itsGEP-inactive mutants, also consistent with the conclusion thatp100 has an important role in the activation of ARF6 for itsfunctions in both E-cadherin recycling and actin remodeling.

adherens junction � F-actin

GEP100 (p100) was described by Someya et al. (1) as an�100-kDa guanine nucleotide-exchange protein (GEP)

that preferentially activated ADP-ribosylation factor 6 (ARF6)in vitro. The molecule itself was of interest because it contained,in addition to the ARF-activating Sec7 domain (with a nuclearlocalization sequence), a pleckstrin homology-like domain andan IQ-like motif, the functions of which remain to be demon-strated. In homogenates of human T98G neuroblastoma cells,p100 was entirely cytosolic. On confocal immunofluorescencemicroscopy, however, it appeared in punctate concentrationsscattered throughout the cytoplasm colocalized, in part, withearly endosomal antigen 1 (EEA1) in a perinuclear region andwith ARF6 near the plasma membrane (1). ARF6 is known tofunction at the cell periphery in endocytosis (2–6), exocytosis�secretion (7–9), phagocytosis (10, 11), cell migration (12–14),adherens junction (AJ) turnover (12, 15), actin cytoskeletonremodeling (16–18), and in the activation of enzymes thatmodify membrane phospholipids (19, 20). Donaldson, in a 2003review of the roles of ARF6 (21), commented on the ‘‘complexinterplay between signal transduction, membrane traffic, and thecytoskeleton.’’ It seems that p100 may be a locus of some of thatinterplay as it is now implicated in regulation of several of thoseARF6 actions.

ARF6 involvement in the internalization and recycling ofdiverse cell surface receptors has been described. Partialcolocalization of ARF6 with transferrin receptor and partic-ipation in its clathrin-dependent internalization was observed(2, 22). Fc� receptor-mediated phagocytosis by macrophagesalso was shown to be regulated by ARF6 (10, 11). ARF6

functions in the internalization of both agonist-stimulated Gprotein-coupled receptors, such as �2-adrenergic receptors via�-arrestin- and clathrin-dependent mechanisms (5, 6, 23) andM2-muscarinic acetylcholine receptors via �-arrestin- andclathrin-independent mechanisms (4). In addition, ARF6 reg-ulation of recycling to the plasma membrane of integrins (24,25), which mediate cell–cell adhesion in most solid tissues (12,15, 26) and are crucial for cell migration, indicates that it caninf luence cell shape, migration, and scattering. ARF6 activa-tion of phosphatidylinositol 4-phosphate 5-kinase (20), gen-erates phosphatidylinositol 4,5-bisphosphate, a major plasmamembrane phosphoinositide involved in membrane traffickingand actin rearrangement (18, 27–29). ARF6 also can activatephospholipase D, which catalyzes phosphatidylcholine hydro-lysis, producing phosphatidic acid, another activator of phos-phatidylinositol 4-phosphate 5-kinase (19). The activation ofboth phospholipase D and phosphatidylinositol 4-phosphate5-kinase by ARF6 can amplify phosphatidylinositol bis-4,5-phosphate signals to affect membrane ruff ling and trafficking,as well as actin rearrangement (24, 30–32).

Like all GTPases, ARF6 requires for its activation andinactivation, respectively, GEPs that accelerate GTP-bindingand GTPase-activating proteins that enhance GTP hydrolysis.We assume that ARF6 activation for its several diversefunctions depends on specific GEPs that will be present withit at the correct place and time. All of our observations areconsistent with a potential role for p100 in multiple ARF6actions. Chen et al. (33) recognized p100 as a shorter form ofa protein termed Loner that was mutated in Drosophilaembryos in which myoblasts failed to form myotubes. Theydetermined that Loner was required to activate ARF6 forproper localization of Rac to initiate formation of a membranefusion assembly. Our finding of p100 interaction with �-cate-nin in a yeast two-hybrid screen led us to investigate thepossible involvement of p100 in ARF6-inf luenced processesrelated to cell adhesion and migration. As reported here,depletion of p100 or �-catenin with specific small interferingRNAs (siRNAs) did not alter cell content of the other, butp100 siRNA treatment increased E-cadherin content 3-foldand interfered with the ‘‘scatter’’ response to hepatocytegrowth factor (HGF). In addition, overexpression of p100resulted in disappearance of F-actin, implicating p100 innormal rat kidney (NRK) cells in ARF6 regulation of adhesionjunction dynamics and actin remodeling.

Conflict of interest statement: No conflicts declared.

Abbreviations: AJ, adherens junction; ARF, ADP-ribosylation factor; GTP�S, guanosine5�-�-(thio)triphosphate; [35S]GTP�S, 5-[�-(35S) thio]triphosphate; HGF, hepatocyte growthfactor; NRK, normal rat kidney; p100, GEP100; siRNA, small interfering RNA.

*To whom correspondence may be addressed at: Building 10, Room 5N307, MSC 1434,National Institutes of Health, Bethesda, MD 20892. E-mail: vaughanm@nih.gov orhiroit@nhlbi.nih.gov.

†Present address: Department of Host Defense and Biochemical Research, Juntendo Uni-versity, School of Medicine, 2-1-1 Hongo, Bunkyo-Ku, Tokyo 113-8421, Japan.

10672–10677 � PNAS � July 11, 2006 � vol. 103 � no. 28 www.pnas.org�cgi�doi�10.1073�pnas.0604091103

ResultsInteraction of p100 and �-catenin. Yeast two-hybrid screening of ahuman fetal brain cDNA library using human p100 as baityielded 62 clones representing 29 proteins; �-catenin is 1 of 4 thatinteracted specifically with p100 in mating assays.

On confocal laser-scanning immunofluorescence microscopyof HepG2 cells, endogenous p100 and �-catenin were distributedin punctate concentrations throughout the cytoplasm (Fig. 1A),but colocalization was not detected, presumably because only afraction of total cell p100 is designated to activate the pool ofARF6 involved in regulating AJs, and most of the �-catenin isfunctioning in other molecular assemblies, e.g., related to actindynamics. Endogenous p100 was immunoprecipitated fromHepG2 and CaSki cells by antibodies (Abs) against �-catenin(Fig. 1B).

Effect of �-Catenin on Guanosine 5�-�-(Thio)Triphosphate (GTP�S)Binding to ARF6. Effects of �-catenin on GTP�S binding by ARF6in vitro were evaluated in assays without or with p100 (Fig. 2).

ARF6 alone bound a small amount of guanosine 5-[�-(35S)thio]triphosphate ([35S]GTP�S) in 20-min assays; binding by�-catenin or p100 alone was not detected. p100 dramaticallyaccelerated [35S]GTP�S binding by ARF6, and in its presence,�-catenin enhanced [35S]GTP�S binding significantly at thehighest concentration tested, which alone was without effect(Fig. 2 A). These effects were seen throughout the 20-minincubation (Fig. 2B).

Effects of p100 and �-Catenin siRNAs on These and Other IntracellularProteins. After incubation of HepG2 cells for 72 h with p100siRNA, p100 protein was 5.2 � 1.5% (n � 5) of that in lysatesof untreated (N) cells (Fig. 3). It was not significantly altered incells incubated for 72 h with vehicle alone or in those in which�-catenin was decreased �90% (9.7 � 3.6% of N) after incu-bation with �-catenin siRNA. E-cadherin in cells incubated withp100 siRNA was 309 � 79% (n � 5) of that in untreated cells,but amounts of �-catenin, ARF6, �-catenin, or GAPDH werenot significantly altered. �-Catenin content was significantlylower (40.2 � 12.1% of N) in cells incubated with �-cateninsiRNA, whereas amounts of the other proteins were not altered(Fig. 3).

On immunof luorescence microscopy, elevated amounts ofE-cadherin in CaSki cells that had been incubated with p100siRNA were concentrated at the plasma membrane as incontrol cells, but the cells appeared to be larger, perhapsbecause of spreading (Fig. 4). Incubation of cells with HGFresulted in redistribution of E-cadherin from its proximity toplasma membranes to scattered punctate collections through-out the cytoplasm. In cells incubated with p100 siRNA,however, internalized E-cadherin was not seen after HGFtreatment (Fig. 4).

Effects of p100 siRNA on HepG2 cell adhesion to collagen-coated wells were assessed by counting cells remaining inmedium at intervals after replating (Fig. 5). Cells that had beenincubated with p100 siRNA became attached to the type 1collagen-coated surface significantly more slowly than thosetreated with nontarget siRNA, or with vehicle alone, which isdifficult to relate to E-cadherin function.

Fig. 1. Localization and immunoprecipitation of p100 and �-catenin incultured cells. (A) Endogenous p100 and �-catenin in HepG2 cells reacted withmouse anti-�-catenin (�-cat; green; Left) and rabbit anti-p100 (red; Right) Abs.(Scale bar: 20 �m.) (B) Immunoprecipitation of endogenous p100 and �-cate-nin from HepG2 (Left) and CaSki (Right) cells. Samples (100 �g of protein) ofhomogenates were immunoprecipitated (IP) with anti-�-catenin Abs (lane 1)or normal mouse IgG (lane 2) and analyzed by Western blotting (IB) withanti-p100 (p100) or anti-�-catenin (�-cat) polyclonal Abs. W, 1�10 of total; ppt(precipitate), 1�3; sup (supernatant), 1�10. Data were similar in twoexperiments.

Fig. 2. Effect of p100 and �-catenin on [35S]GTP�S binding by ARF6. (A)Recombinant human ARF6 (35 pmol) and 4 �M [35S]GTP�S (2.5 � 106 cpm)without or with p100 (2 pmol) and�or �-catenin as indicated in 20 mM Tris�HCl(pH 8.0)�2 mM DTT�3 mM MgCl2�1 mM EDTA�1 mM NaN3�250 mM sucrosewith 40 �g of BSA and 10 �g of L-�-phosphatidyl-L-serine (total volume 50 �l)were incubated at 30°C for 20 min, before radioassay of bound [35S]GTP�S asdescribed (1). (B) ARF6 was incubated as in A without or with p100 and�or 25pmol of �-catenin (�-cat) for the indicated time. Data in A and B are means �SEM of values from triplicate assays. Data were similar in two experiments.

Fig. 3. Effects of p100 or �-catenin siRNA on these and other proteins inHepG2 cells. Cells were incubated without additions (N), with DharmaFECTsiRNA Transfection reagents no. 4 (M), or with p100 or �-catenin (�-cat) siRNAsfor 72 h, harvested, and homogenized in 50 mM Tris�HCl (pH 7.5)�150 mM NaClcontaining 1 mM EDTA, 1 mM PMSF, and protease inhibitor mixture. Samples(10 �g of protein) of homogenates were analyzed by immunoblotting withAbs against p100, �-catenin, E-cadherin (E-cad), ARF6, �-catenin (�-cat), andGAPDH. *, Unidentified immunoreactivity seen in all experiments. Means �SEM of values from five experiments quantified by the NIH IMAGE program andexpressed relative to that of untreated cells (N) � 100 are shown.

Hiroi et al. PNAS � July 11, 2006 � vol. 103 � no. 28 � 10673

CELL

BIO

LOG

Y

F-Actin in Cells Overexpressing p100. The disappearance of F-actinfrom NRK cells overexpressing wild-type (WT) p100–GFP wasdramatic and was not seen in cells overexpressing GFP or anyGFP-tagged p100 mutants (p100E498A, p100�Sec7, orp100�PH) (Fig. 6D). The staining pattern of F-actin in cellsoverexpressing p100�PH, i.e., without the PH domain, was,however, somewhat altered. Amounts of total actin were similarin all cells whether or not the appearance of F-actin was altered(Fig. 6B). From NRK cells overexpressing WT p100–GFP, butnot GFP alone, anti-GFP Abs precipitated endogenous �-cate-nin (Fig. 6C).

DiscussionFinding that p100 interacted with �-catenin in a yeast two-hybridscreen suggested, of course, its involvement as an activator in at

least some of the known actions of ARF6 near the cell surface,such as actin remodeling (16–18), cell migration (12–14), and AJfunction (12, 15). The dramatic increase in E-cadherin contentthat we found in HepG2 cells depleted of p100 by siRNAtreatment was not associated with changes in amount of �-cate-nin, nor did similar depletion of �-catenin alter amounts of p100or E-cadherin. E-cadherin distribution in the p100-depleted cellsresembled microscopically that in controls, although the cells didappear larger, consistent with spreading and flattening. In AJs,�-catenin interacts with E-cadherin indirectly through its asso-ciation with �-catenin, which does interact with E-cadherin. Italso binds to actin (34), and �-catenin was long considered alinker between AJs and actin fibers. Recently, it was reported(35, 36) that �-catenin binding to actin and �-catenin is, in effect,mutually exclusive, i.e., �-catenin monomers preferentially bind�-catenin and thereby the E-cadherin complex, whereas dimeric�-catenin binds actin filaments. In this way, �-catenin wouldserve as a molecular switch coordinating functions of AJs incell–cell interaction and the actin cytoskeleton in cell motility(35, 36)

Cadherins, Ca2�-dependent, homophilic, adhesion proteinspresent in the cells of most solid tissues, are critical in thestructure of AJs, which regulate cell–cell adhesion (37, 38).Cell-surface E-cadherin in epithelial cells is partially internal-ized and recycled back to the plasma membrane (39). ARF6has important functions in both cell adhesion and E-cadherinrecycling (12, 15, 26, 40). Overexpression of constitutivelyactive ARF6(Q67L) resulted in loss of AJs and ruff ling ofbasolateral membranes of Madin–Darby canine kidney(MDCK) cells (18). In contrast, overexpression of dominant-negative ARF6(T27N) blocked HGF-induced internalizationof E-cadherin, which requires ARF6–GTP (18). ARF6 andRac1 functions were both required in cell scattering, and anARF6-dependent decrease in Rac1–GTP was necessary forHGF-induced cell–cell dissociation, which depends on priordisassembly of AJs (40). ARF6 activation by p100 appeared tobe required for internalization of E-cadherin by HepG2 cells,resulting in its accumulation at the plasma membrane ofp100-depleted cells, and interfered with its internalization inresponse to HGF. We found that E-cadherin levels wereelevated in cells after siRNA depletion of p100 content, whereit appeared increased in both cell interior and plasma mem-brane, with clearly impaired redistribution in response toHGF. Along with changes in their E-cadherin localization, thep100-depleted cells became attached to collagen-coated wellsmore slowly than did cells treated with other siRNAs. Thisresult seems likely due to the requirement of p100-activatedARF6 for maintenance of a cell surface adhesion functionother than that of E-cadherin in AJs. It is consistent withknown roles of ARF6 in cell–matrix and cell–cell interactions(21), as well as with the recent report of Dunphy et al. (41)implicating BRAG2, a form of which is identical to p100, inrecycling of �1-integrin.

Those researchers found that overexpression of epitope-tagged BRAG2a�p100 (or BRAG2b) increased amounts ofARF6–GTP in cell lysates, whereas mutants incapable ofactivating ARF6 were without effect. Overexpressed BRAG2�p100 in both HeLa and MDCK cells was detected in nuclei andincreased after incubation of cells with leptomycin B (41). Thisobservation is consistent with the presence of a nuclearlocalization sequence, which had been identified in p100,although we had not seen the endogenous protein in T98G cellnuclei (1), and its nuclear function is not readily apparent.Overexpression of HA–BRAG2a or -2b (but not an E498Amutant incapable of activating ARF6) markedly decreasedamounts of F-actin seen microscopically in HeLa cells, withoutchanging total actin content (41). The findings in NRK cellsreported here are quite similar, but possible differences in the

Fig. 4. Effects of p100 siRNA on endogenous E-cadherin distribution andresponse to HGF. CaSki cells after incubation for 72 h with vehicle alone (Mock;Left) or with p100 siRNA (Right) were incubated for 4 h without or with 20 �MHGF, followed by reaction with anti-E-cadherin monoclonal Ab. (Scale bars: 20�m.)

Fig. 5. Effect of p100 siRNA on HepG2 cell adhesion. HepG2 cells incubatedfor 72 h without additions (Normal), with vehicle (Mock), or with siRNA,nontarget (Non-T), or p100 were harvested by using EDTA, dispersed in MEMcontaining 10% FBS and 0.1 mM nonessential amino acids, and distributed(5 � 104 cells per well) in collagen-coated 24-well plates. At the indicated timesthereafter, DNA in medium was quantified as an index of cell number by usingthe Quant-iT PicoGreen dsDNA assay kit (Invitrogen) and is reported relativeto that at zero time � 100%. Data are means � SEM of values from fivereplicates in one experiment representative of two.

10674 � www.pnas.org�cgi�doi�10.1073�pnas.0604091103 Hiroi et al.

appearance of F-actin when p100 lacked the PH domainrequire additional investigation, as do other potential effectsof p100 mutants. Dunphy et al. (41) described the accumula-tion of overexpressed BRAG2 (not mutant E498A) in clustersor patches at the plasma membrane where F-actin was alsoconcentrated, and those effects were enhanced by phorbolmyristoyl acetate. In their studies, ‘‘knock-down’’ of BRAG2in HeLa cells resulted in increased amounts of �1-integrin onthe cell surface, which correlated with increased cell spread-ing, whereas cell-surface �1-integrin and cell spreading weredecreased in cells treated with ARF6 siRNA. The authorssuggested that ARF6 activated by BRAG2 served specificallyfor �1-integrin internalization, whereas a different activator ofARF6 was required for its recycling (41).

The endocytosis of E-cadherin that accompanies AJ disas-sembly in MDCK cells (26) required Nm23-H1, a nucleosidediphosphate kinase (NDPK), earlier described as a suppressorof tumor metastasis (26). ARF6–GTP recruited Nm23–H1 tothe cytoplasmic face of AJs where it was necessary for dy-namin-catalyzed fission of clathrin-coated endocytic vesiclesthat remove E-cadherin and also sequestered Tiam 1, a GEPfor Rac1, thereby interfering with Rac1 activation (26). Thealtered appearance of E-cadherin that we saw in p100-depletedcells responding to HGF may ref lect a role of p100 in theactivation of ARF6 for at least some of the steps of AJdissociation and E-cadherin recycling in addition to itsendocytosis.

In addition to p100�BRAG2, EFA6 (42) and cytohesin�ARNO (43) are reported to be brefeldin A (BFA)-insensitiveactivators of ARF6. EFA6 was initially implicated in regulationof transferrin receptor endocytosis, endosomal membrane recy-cling, and actin cytoskeleton remodeling related to membraneruffling (42). Its role in E-cadherin recruitment and actinrearrangement for generation of tight junctions (TJs), whichfollows AJ assembly, and establishment of cell polarity was laterdescribed (44). EFA6 and p100 may have analogous functions inparallel pathways that regulate, respectively, TJ and AJ dynam-ics, including their interactions with the actin cytoskeleton.ARNO had been described as an activator of ARF1 (43) beforeFrank et al. (45) described its activation of ARF6 and locationat the plasma membrane, commenting on the relationship oftheir findings to the report of cytohesin-1 as a regulator of �L�2integrin function at the cell surface (46). Santy and Casanova(13) later showed that ARNO was responsible for ARF6 acti-vation in MDCK cells, where its overexpression enhanced mi-gration by inducing formation of lamellipodia and separation ofcell–cell contacts. Activation of Rac and phospholipase D alsowas observed. Inhibition of phospholipase D activity diminishedcell motility, but not Rac activation, i.e., it appeared thatphosphatidic acid production was not necessary for Rac activa-tion but was important for the migratory behavior (13). Whetheror not p100 is involved in those processes remains to bedetermined.

Proteins that contain Sec7 domains, which activate ARFGTPases by accelerating replacement of bound GDP by GTP,

Fig. 6. Effects of overexpressed WT and mutant p100-GFP on F-actin in NRK cells. (A) Positions of IQ-like motif, serine-rich region (SR), nuclear localization signal(NLS) in the Sec7 domain, and PH domain are indicated in WT and mutant p100 molecules with C-terminal GFP. In mutant E498A, Sec7 domain Glu-498, whichis required for ARF activation, is replaced by Ala; �Sec7 lacks the Sec7 domain, and �PH lacks the pleckstrin homology domain. (B) Proteins (10 �g) in homogenatesof NRK cells overexpressing GFP or GFP-tagged WT, E498A, �Sec7, or �PH p100 were analyzed by Western blot with Abs against p100, actin, and GAPDH.Untreated (Cont) and reagent-treated (Mock) cells were incubated without or with Lipofectamine PLUS (Invitrogen) reagent, respectively. Data were similar intwo experiments. (C) NRK cells, grown on 100-mm plastic dishes, were transfected with p100-GFP (WT) or GFP by using Lipofectamine PLUS (Invitrogen) accordingto the manufacturer’s instructions. Samples (100 �g of protein) of homogenates were immunoprecipitated (IP) with anti-GFP Abs (lane 1) or normal rabbit IgG(lane 2) and analyzed by Western blotting (IB) with anti-�-catenin Abs, as described for Fig. 1B. Data were similar in two experiments. (D) Confocal images ofNRK cells overexpressing GFP or GFP-tagged WT, E498A, �Sec7, or �PH p100. Fixed and permeabilized cells were reacted with Abs against GFP (Lower) andphalloidin–tetramethylrhodamine B isothiocyanate (TRITC) (F-actin; Upper). See Supporting Text, which is published as supporting information on the PNAS website, for details of construction of p100-GFP expression plasmids and transfection of NRK cells.

Hiroi et al. PNAS � July 11, 2006 � vol. 103 � no. 28 � 10675

CELL

BIO

LOG

Y

are present in all eukaryotic cells. In a phylogenetic analysis ofSec7 domain proteins, Cox et al. (47) described the ARF6-activating BRAG2�p100, EFA, and cytohesin�ARNO groups asthe only ones that are found exclusively in animals. Perhaps eachcoevolved with other proteins to serve in similarly structured,parallel pathways, through which dynamic mechanical, meta-bolic, and signaling processes associated with cell–cell andcell–matrix interactions are controlled. Those vital functions allrequire continual temporal and spatial regulation to assureprecise coordination and integration throughout the life of anorganism. It will be necessary to understand better each of theBRAG2�p100, EFA6, and cytohesin�ARNO pathways for in-side-out and outside-in signaling before we can begin to relatethem. In addition to those GEPs, BIG2 was identified as criticalin E-cadherin transport from Golgi to plasma membranes in aninvestigation of autosomal recessive mutations that caused de-fective human embryonic brain development (48). BIG2 is abrefeldin A-inhibited activator of ARF1–3, previously impli-cated in delivery of transferrin receptor from recycling endo-some to the cell surface (49). The findings of Sheen et al. (48)emphasize the major importance of vesicular trafficking indifferentiation and development.

The elevated E-cadherin content of cells depleted of p100 inour studies was striking, but the mechanism through which itoccurs is unknown, as is the presumably different processthrough which �-catenin was decreased in cells treated with�-catenin siRNA. Our data are consistent with roles for p100 inan ARF6-dependent E-cadherin recycling pathway and E-cad-herin��-catenin��-catenin-mediated effects on cell adhesionand structure. Learning how p100 levels are controlled andcharacterization of the endogenous p100 molecules are now ofmajor interest. We need to understand how the production ofalternatively spliced forms of BRAG2 described by Dunphy et al.(41) is regulated, as well as how these forms differ in function.Recognition of BRAG2�p100 involvement in functions resem-bling several better-studied actions of EFA6 and cytohesin�ARNO with ARF6, adhesion molecules, and the actin cytoskel-eton provides many new questions.

Materials and MethodsMaterials. FBS, medium, nonessential amino acids, Nupage Bis-Tris gels, and Escherichia coli-competent cells [one shot TOP10and BL21 (DE3)] were purchased from Invitrogen; HGF andprotease inhibitors were purchased from Sigma; and [35S]GTP�S(1,250 Ci�mmol; 1 Ci � 37 GBq) was purchased from NEN.cDNA encoding KIAA0763 (GenBank accession no. AB018306;GI 3882246) was kindly provided by T. Nagase (Kazusa DNAResearch Institute, Chiba, Japan).

Cell Culture. All cells (purchased from American Type CultureCollection) were incubated in an atmosphere of 5% CO2�95%air at 37°C. HepG2 human liver carcinoma cells, used in mostexperiments, were grown in minimum essential medium with10% FBS and 0.1 mM nonessential amino acids on collagen-coated plastic plates (BD Biosciences, San Diego). Humanepidermoid carcinoma CaSki cells were grown in RPMI me-dium 1640 with 10% FBS. NRK cells, used for overexpressionof GFP-tagged WT and mutant p100, were grown in Dulbec-co’s modified Eagle’s medium with 10% FBS. All mediacontained penicillin G (100 units�ml) and streptomycin (100�g�ml).

p100 and �-Catenin siRNA Experiments. HepG2 cells (50–60%confluent) were incubated for 72 h with p100 or �-cateninsiRNA (siGENOME Smartpool reagent), according to themanufacturer’s instructions (Dharmacon Research, Lafayette,CO). Control cells were incubated with vehicle (DharmaFECTsiRNA Transfection reagents no. 4) alone or with nontarget

siRNA (siGENOME Smartpool reagent), but not all controldata are shown. For immunoblotting analyses, cells were washedwith PBS, harvested, and homogenized in lysis buffer (150 mMNaCl in 50 mM Tris�HCl, pH 7.5) containing benzamidine (16�g�ml), phenanthroline, aprotinin, leupeptin, and pepstatin A(each 10 �g�ml), 1 mM phenylmethylsulfonyl f luoride (PMSF),and 1% Triton X-100 (0.2 ml per dish).

Confocal Microscopy. HepG2 and CaSki cells grown on type 1collagen-coated four-well CultureSlide (BD Biosciences) wereusually fixed (20 min, room temperature) with 3% paraformal-dehyde in PBS, washed with PBS, permeabilized (4 min) with0.1% Triton X-100 in PBS, washed with PBS, and incubated (1h, room temperature) with PBS containing 10% goat serum and3% BSA. Cells were washed with PBS and incubated (overnightat 4°C) with anti-p100 polyclonal Abs (1�20 dilution), anti-�-catenin monoclonal Abs (1�20 dilution, �-E-catenin; Santa CruzBiotechnology), or anti-E-cadherin monoclonal Abs (1�200 di-lution; BD Biosciences), washed with PBS, and incubated (1 h atroom temperature) with FITC-conjugated anti-mouse IgG (1�100 dilution; Vector Laboratories) or Texas red-conjugated orFITC-conjugated anti-rabbit IgG (1�100 dilution; Vector Lab-oratories). After washing with PBS, cells were mounted inVectashield (Vector Laboratories) and inspected with a confo-cal microscope (LSM 510; Zeiss).

Immunoblotting Analyses. Proteins (10 �g) were separated bySDS�PAGE (4–12% gel) and transferred to poly(vinylidenedifluoride) membranes. Blots were incubated with anti-p100polyclonal (1�500 dilution), anti-�-catenin monoclonal (1�500dilution), anti-E-cadherin monoclonal (1�2,500 dilution), anti-ARF6 monoclonal (1�200 dilution; Chemicon International,Temecula, CA), anti-�-catenin monoclonal (1�500 dilution; BDBiosciences), or anti-GAPDH polyclonal (1�2,000 dilution;Abcam, Inc., Cambridge, MA) Abs, followed by horseradishperoxidase-conjugated anti-mouse IgG or anti-rabbit IgG (Pro-mega) and development using SuperSignal West Pico Chemilu-minescent Substrate (Pierce). Band density was quantified by theNIH IMAGE program (http:��rsb.info.nih.gov� nih-image).

Immunoprecipitation. HepG2 or CaSki cell proteins (100 �g) wereincubated with anti-�-catenin monoclonal Ab or mouse IgG (2�g) for 1 h, then overnight with protein A�G-agarose (50 �l).Agarose was washed twice with lysis buffer, once with 300 mMNaCl in 50 mM Tris�HCl (pH 7.5), and twice with 0.1% TritonX-100 in 10 mM Tris�HCl (pH 7.5). Bound proteins, extracted byboiling beads for 3 min in loading buffer, were separated bySDS�PAGE (4–12% gel) and transferred to poly(vinylidenedifluoride) membranes for reaction with anti-p100 polyclonalAbs (1�500 dilution).

NRK cells (4 � 106 cells), 24 h after transfection, were washedtwice with PBS, suspended in lysis buffer, and, after 30 min on ice,homogenized (30 strokes) in a Dounce homogenizer. The homog-enate was centrifuged (2,000 � g, 15 min), and the supernatant (100�g) was incubated (4°C, 1 h) with 50 �l of protein A�G-agarose,which was discarded, then incubated with anti-GFP Living ColorsFull-Length Aequorea victoria polyclonal Ab (2 �g; Clontech) for1 h, followed by addition of 50 �l of protein A�G-agarose andincubation overnight. Bound proteins were eluted for immunoblot-ting with anti-�-catenin monoclonal Ab (1�500 dilution).

We thank Dr. Christian Combs (National Heart, Lung, and BloodInstitute Confocal Microscopy Core Facility) and Dr. Zu-Xi Yu (Na-tional Heart, Lung, and Blood Institute Pathology Core Facility) forinvaluable help and Dr. Julie G. Donaldson (Laboratory of Cell Biology,National Heart, Lung, and Blood Institute) for manuscript review. Thiswork was supported by the Intramural Research Program of the NationalHeart, Lung, and Blood Institute, National Institutes of Health.

10676 � www.pnas.org�cgi�doi�10.1073�pnas.0604091103 Hiroi et al.

1. Someya, A., Sata, M., Takeda, K., Pacheco-Rodriguez, G., Ferrans, V. J., Moss,J. & Vaughan, M. (2001) Proc. Natl. Acad. Sci. USA 98, 2413–2418.

2. D’Souza-Schorey, C., Li, G., Colombo, M. I. & Stahl, P. D. (1995) Science 267,1175–1178.

3. Altschuler, Y., Liu, S., Katz, L., Tang, K., Hardy, S., Brodsky, F., Apodaca, G.& Mostov, K. (1999) J. Cell Biol. 147, 7–12.

4. Claing, A., Chen, W., Miller, W. E., Vitale, N., Moss, J., Premont, R. T. &Lefkowitz, R. J. (2001) J. Biol. Chem. 276, 42509–42513.

5. Delaney, K. A., Murph, M. M., Brown, L. M. & Radhakrishna, H. (2002) J. Biol.Chem. 277, 33439–33446.

6. Hunzicker-Dunn, M., Gurevich, V. V., Casanova, J. E. & Mukherjee, S. (2002)FEBS Lett. 521, 2389–2398.

7. Galas, M. C., Helms, J. B., Vitale, N., Thierse, D., Aunis, D. & Bader, M. F.(1997) J. Biol. Chem. 272, 2788–2793.

8. Matsukawa, J., Nakayama, K., Nagao, T., Ichijo, H. & Urushidani, T. (2003)J. Biol. Chem. 278, 36470–36475.

9. Lawrence, J. T. & Birnbaum, M. J. (2003) Proc. Natl. Acad. Sci. USA 100,13320–13325.

10. Zhang, Q., Cox, D., Tseng, C. C., Donaldson, J. G. & Greenberg, S. (1998)J. Biol. Chem. 273, 19977–19981.

11. Niedergang, F., Colucci-Guyon, E., Dubois, T., Raposo, G. & Chavrier, P.(2003) J. Cell Biol. 161, 1143–1150.

12. Palacios, F., Price, L., Schweitzer, J., Collard, J. G. & D’Souza-Schorey, C.(2001) EMBO J. 20, 4973–4986.

13. Santy, L. C. & Casanova, J. E. (2001) J. Cell Biol. 154, 599–610.14. Weber, K. S., Weber, C., Ostermann, G., Dierks, H., Nagel, W. & Kolanus, W.

(2001) Curr. Biol. 11, 1969–1974.15. Paterson, A. D., Parton, R. G., Ferguson, C., Stow, J. L. & Yap, A. S. (2003)

J. Biol. Chem. 278, 21050–21057.16. D’Souza-Schorey, C., Boshans, R. L., McDonough, M., Stahl, P. D. & van Aelst,

L. (1997) EMBO J. 16, 5445–5454.17. Radhakrishna, H., Al-Awar, O., Khachikian, Z. & Donaldson, J. G. (1999)

J. Cell Sci. 112, 855–866.18. Boshans, R. L., Szanto, S., van Aelst, L. & D’Souza-Schorey, C. (2000) Mol. Cell

Biol. 20, 3685–3694.19. Brown, H. A., Gutowski, S., Moomaw, C. R., Slaughter, C. & Sternweis, P. C.

(1993) Cell 75, 1137–1144.20. Honda, A., Nogami, M., Yokozeki, T., Yamazaki, M., Nakamura, H., Wa-

tanabe, H., Kawamoto, K., Nakayama, K., Morris, A. J., Frohman, M. A. &Kanaho, Y. (1999) Cell 99, 521–532.

21. Donaldson, J. G. (2003) J. Biol. Chem. 278, 41573–41576.22. D’Souza-Schorey, C., van Donselaar, E., Hsu, V. W., Yang, C., Stahl, P. D. &

Peters, P. J. (1998) J. Cell Biol. 140, 603–616.23. Salvador, L. M., Mukherjee, S., Kahn, R. A., Lamm, M. L., Fazleabas, A. T.,

Maizels, E. T., Bader, M. F., Hamm, H, Rasenick, M. M., Casanova, J. E. &Hunzicker-Dunn, M. (2001) J. Biol. Chem. 276, 33773–33781.

24. Brown, F. D., Rozelle, A. L., Yin, H. L., Balla, T. & Donaldson, J. G. (2001)J. Cell Biol. 154, 1007–1017.

25. Powelka, A. M., Sun, J., Li, J., Gao, M., Shaw, L. M., Sonnenberg, A. & Hsu,V. W. (2004) Traffic 5, 20–36.

26. Palacios, F., Schweitzer, J. K., Boshans, R. L. & D’Souza-Schorey, C. (2002)Nat. Cell Biol. 4, 929–936.

27. Schafer, D. A., D’Souza-Schorey, C. & Cooper, J. A. (2000) Traffic 1, 896–907.28. Czech, M. P. (2003) Annu. Rev. Physiol. 65, 791–815.29. Yin, H. L. & Janmey, P. A. (2003) Annu. Rev. Physiol. 65, 61–89.30. Aikawa, Y. & Martin, T. F. (2003) J. Cell Biol. 162, 647–659.31. Vitale, N., Chasserot-Golaz, S., Bailly, Y., Morinaga, N., Frohman, M. A. &

Bader, M. F. (2002) J. Cell Biol. 159, 79–89.32. O’Luanaigh, N., Pardo, R., Fensome, A., Allen-Baume, V., Jones, D., Holt,

M. R. & Cockcroft, S. (2002) Mol. Biol. Cell 13, 3730–3746.33. Chen, E. H., Pryce, B. A., Tzeng, J. A., Gonzalez, G. A. & Olson, E. N. (2003)

Cell 114, 751–762.34. Kobielak, A. & Fuchs, E. (2004) Nat. Rev. Mol. Cell Biol. 5, 614–625.35. Yamada, S., Pokutta, S., Drees, F., Weis, W. I. & Nelson, W. J. (2005) Cell 123,

889–901.36. Drees, F., Pokutta, S., Yamada, S., Nelson, W. J. & Weis, W. I. (2005) Cell 123,

903–915.37. Takeichi, M. (1991) Science 251, 1451–1455.38. Gumbiner, B. M. (2005) Nat. Rev. Mol. Cell Biol. 6, 622–634.39. Le, T. L., Yap, A. S. & Stow, J. L. (1999) J. Cell Biol. 146, 219–232.40. Palacios, F. & D’Souza-Schorey, C. (2003) J. Biol. Chem. 278, 17395–17400.41. Dunphy, J. L., Moravec, R., Ly, K., Lasell, T. K., Melancon, P. & Casanova, J. E.

(2006) Curr. Biol. 16, 315–320.42. Franco, M., Peters, P. J., Boretto, J., van Donselaar, E., Neri, A., D’Souza-

Schorey, C. & Chavrier, P. (1999) EMBO J. 18, 1480–1491.43. Chardin, P., Paris, S., Antonny, B., Robineau, S., Beraud-Dufour, S., Jackson,

C. L. & Chabre, M. (1996) Nature 384, 481–484.44. Luton, F., Klein, S., Chauvin, J. P., Le Bivic, A., Bourgoin, S., Franco, M. &

Chardin, P. (2004) Mol. Biol. Cell 15, 1134–1145.45. Frank, S., Upender, S., Hansen, S. H. & Casanova, J. E. (1998) J. Biol. Chem.

273, 23–27.46. Kolanus, W., Nagel, W., Schiller, B., Zeitlmann, L., Godar, S., Stockinger, H.

& Seed, B. (1996) Cell 86, 233–242.47. Cox, R., Mason-Gamer, R. J., Jackson, C. L. & Segev, N. (2004) Mol. Biol. Cell

15, 1487–1505.48. Sheen, V. L., Ganesh, V. S., Topcu, M., Sebire, G., Bodell, A., Hill, R. S., Grant,

P. E., Shugart, Y. Y., Imitola, J., Khoury, S. J., et al. (2004) Nat. Genet. 36,69–76.

49. Shen, X., Xu, K. F., Fan, Q., Pacheco-Rodriguez, G., Moss, J. & Vaughan, M.(2006) Proc. Natl. Acad. Sci. USA 103, 2635–2640.

Hiroi et al. PNAS � July 11, 2006 � vol. 103 � no. 28 � 10677

CELL

BIO

LOG

Y