Biochemical Journal

Research article

EHD2 mediates trafficking from the plasma membrane by modulating Rac1 activity

Sigi Benjamin, Hilla Weidberg, Debora Rapaport, Olga Pekar, Marina Nudelman, Daniel Segal, Koret Hirschberg, Shulamit Katzav, Marcelo Ehrlich, Mia Horowitz

Abstract

EHDs [EH (Eps15 homology)-domain-containing proteins] participate in different stages of endocytosis. EHD2 is a plasma-membrane-associated EHD which regulates trafficking from the plasma membrane and recycling. EHD2 has a role in nucleotide-dependent membrane remodelling and its ATP-binding domain is involved in dimerization, which creates a membrane-binding region. Nucleotide binding is important for association of EHD2 with the plasma membrane, since a nucleotide-free mutant (EHD2 T72A) failed to associate. To elucidate the possible function of EHD2 during endocytic trafficking, we attempted to unravel proteins that interact with EHD2, using the yeast two-hybrid system. A novel interaction was found between EHD2 and Nek3 [NIMA (never in mitosis in Aspergillus nidulans)-related kinase 3], a serine/threonine kinase. EHD2 was also found in association with Vav1, a Nek3-regulated GEF (guanine-nucleotide-exchange factor) for Rho GTPases. Since Vav1 regulates Rac1 activity and promotes actin polymerization, the impact of overexpression of EHD2 on Rac1 activity was tested. The results indicated that wt (wild-type) EHD2, but not its P-loop mutants, reduced Rac1 activity. The inhibitory effect of EHD2 overexpression was partially rescued by co-expression of Rac1 as measured using a cholera toxin trafficking assay. The results of the present study strongly indicate that EHD2 regulates trafficking from the plasma membrane by controlling Rac1 activity.

  • endocytosis
  • Eps15 homology (EH)-domain-containing protein 2 (EHD2)
  • Rac1 activity

INTRODUCTION

Eukaryotic cells utilize endocytosis for critical cellular processes such as nutrient uptake, down-regulation of signalling, development and synaptic transmission [1,2]. A number of internalization routes have been characterized and include the clathrin-mediated pathway and several non-clathrin pathways, such as phagocytosis and caveolin-mediated endocytosis. It has been suggested that actin polymerization participates in the early stages of clathrin-mediated endocytosis [3], in caveolin-mediated endocytosis [4] and in phagocytosis [5]. Recent studies, mostly using live-cell imaging, have highlighted the kinetics of actin remodelling and the recruitment of endocytic proteins during early stages of internalization [6]. Actin reorganization is required for coated-pit formation, vesicle constriction and scission [6,7]. Dynamin serves to link endocytosis with actin dynamics by interacting with actin-binding proteins such as syndapin [3]. Proteins like syndapin enhance the ability of N-WASP (neuronal Wiskott–Aldrich syndrome protein) to activate Arp2/3 (actin-related protein 2/3), which, in turn, results in the nucleation of actin filaments [3]. N-WASP is activated by Rho GTPases that determine the site and duration of actin reorganization and membrane trafficking. The activity of small GTPases is regulated by two protein families, GEFs (guanine-nucleotide-exchange factors), which stimulate the exchange of GDP to GTP, and GAPs (GTPase-activating proteins), which mediate GTP hydrolysis [8,9].

All stages of endocytosis are regulated through numerous protein–protein interactions mediated by specific modules such as the EH (Eps15 homology) domain. EHDs (EH-domain-containing proteins) harbour an N-terminal G domain, followed by a helical domain and a C-terminal EH domain [1012]. One EHD orthologue exists in Drosophila and Caenorhabditis elegans [13,14], two genes exist in plants [15] and four orthologues are known in vertebrates [12]. Despite their high homology (approximately 70%), the mammalian EHDs (EHD1–4) differ in localization and function. EHD1 was localized to the endocytic-recycling compartment, a feature that was dependent on an intact nucleotide-binding domain [11,16]. Additionally, EHD1 was shown to regulate recycling of different ligands through expression of dominant-negative mutant forms of the protein or by abolishing its expression [1621]. EHD3 was localized to tubular structures of the endocytic recycling compartment [22] and loss of its expression inhibited transport from the early endosome to the endocytic recycling compartment [23]. Another role for EHD3 in regulating endosome-to-Golgi transport and in maintaining Golgi morphology has been shown [24]. EHD4 was localized to the plasma membrane, and dominant-negative forms of the protein disrupted NGF (nerve growth factor) receptor internalization in PC12 cells [2527].

An EH–NPF (Asn–Pro–Phe) interaction was reported between EHD1, EHD3, EHD4 and syndapins. Syndapins are proteins which connect vesicle formation with the cortical actin cytoskeleton and participate in receptor recycling [28,29].

EHD2 was localized to the plasma membrane [3032] and was shown to interact with phospholipids [10,32]. It interacted with EHBP1 (EHD2-binding protein 1), a putative actin-binding protein, and its overexpression and down-regulation led to inhibition of internalization. Its overexpression also led to actin reorganization [30]. In addition, a role for EHD2 in recycling has been suggested [31,33]. It was shown that EHD2 has a role in nucleotide-dependent membrane remodelling and that its ATP-binding domain is involved in dimerization, which creates a membrane-binding region. Nucleotide binding was important for association of EHD2 with the plasma membrane, since a nucleotide-free mutant (EHD2 T72A) failed to associate. The crystal structure of EHD2 showed that it exists as a dimer. Dimerization is mediated by a highly conserved, mostly hydrophobic, interface in the G domain [10].

In the present study we show that EHD2 interacts with Nek3 [NIMA (never in mitosis in Aspergillus nidulans)-related kinase 3] and associates with Vav1, a known GEF for Rac1. Overexpression of EHD2 down-regulates Rac1 and overexpression of Rac1 partially rescues the inhibitory phenotype.

EXPERIMENTAL

Antibodies

The primary antibodies used in the present study were: rabbit anti-GFP (green fluorescent protein; sc-8334, Santa Cruz Biotechnology); mouse anti-Myc (9B11, Cell Signaling Technology); mouse anti-human Vav1 for Western blot analysis (UBI); rabbit anti-human Vav1 for immunoprecipitations (sc-132, Santa Cruz Biotechnology); mouse anti-HA (haemagglutinin; sc-7392, Santa Cruz Biotechnology); mouse anti-Rac1 (610650, BD Transduction Laboratories); and anti-NEK3 (ab37636) and anti-EHD2 (ab23935) (Abcam). Secondary antibodies used were: Cy3 (indocarbocyanine)-conjugated goat anti-mouse, Cy5 (indodicarbocyanine)-conjugated goat anti-mouse, and HRP (horseradish peroxidase)-conjugated goat anti-mouse and goat anti-rabbit (Jackson ImmunoResearch).

Cell Lines

HeLa, HEK (human embryonic kidney)-293, BSC-1 cells and stable Vav1-expressing fibroblasts were grown in DMEM (Dulbecco's modified Eagle's medium; Gibco) supplemented with 10% FBS (fetal bovine serum; Beit-Haemek). All cells were grown at 37°C in the presence of 5% CO2.

Plasmids

To isolate human EHD2 cDNA, a human fetal brain cDNA library in Charon BS was screened, using a PCR fragment representing part of the 3′ UTR (untranslated region) as a probe. Plasmids bearing the human cDNA inserts were rescued by digestion with NotI and subsequent self-ligation [34]. The ORF (open reading frame) of human EHD2 was cloned into the XhoI-BamHI sites of pEGFP-C1 (Clontech), into the Ecl136II-SalI sites of pEYFP-C1 (Clontech) and the XhoI-EcoRV sites of the pCMV-NEO-BAM-myc vector (a gift from Professor Sima Lev, Weizmann Institute of Science, Rehovot, Israel). To create mutant forms of EHD2, in vitro site-directed mutagenesis was performed using the QuikChange® site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. GT–CFP contains a CFP (cyan fluorescent protein) fused to trans-Golgi-galactosyltransferase, a marker for the Golgi apparatus. HA–Rac1 was a gift from Professor Yoel Kloog, Tel Aviv University, Tel Aviv, Israel. To create a prey vector expressing EHD2 for a yeast two-hybrid screen, the ORF of EHD2 was PCR-amplified using the primers 2hybEHD2EcoRI (5′-CGGTGTGCAGAATTCATGTTCAGCTGGCTG-3′) and 2hybEHD2XhoI (5′-GGAGGGGGCTCGAGCTCACTCGGCGGAGCC-3′). The resulting fragment was digested with EcoRI and XhoI and cloned into the corresponding sites of the ‘bait’ pLexA vector (Interaction Trap Matchmaker System, Clontech). The full-length cDNA of Nek3 was produced by PCR from cDNA [created by RT (reverse transcription)–PCR using RNA isolated from HeLa cells] using the primers Nek3-4 (5′-CGGGGAAGCGGTTTGGGAGAGCC-3′) and Nek3-5NotI (5′-CTGTGTGGCGGCCGCAGCATGGATG-3′). The PCR fragment was treated with the Klenow fragment of DNA polymerase I (Fermentas) and digested with NotI, and the resulting fragment was cloned into the EcoRV-NotI sites of the pCMV-NEO-BAM-myc vector. To create prey and bait vectors expressing Vav1, the ORF of Vav1 was PCR-amplified using the primers Vav1EcoRI (5′-CGGGAATTCATGGAGCTGTGGCGCC-3′) and Vav1XhoI (5′-CCGCTCGAGTCAGCAGTATTCAGAAT-3′). The resulting fragment was digested with EcoRI and XhoI and cloned into the corresponding sites of the pLexA ‘bait’ and the pB42AD ‘prey’ vectors.

Transfections

Transfection of HeLa and BSC-1 cells was performed using FuGENE® 6 transfection reagent (Roche Diagnostics) according to the manufacturer's instructions. Stable Vav1-expressing fibroblasts were transfected using Lipofectamine™ 2000 (Invitrogen) according to manufacturer's instructions. HEK-293 cells were transfected using calcium phosphate. A mixture of DNA in 250 μl of 250 mM CaCl2 was dropped into a tube containing 250 μl of HBSX2 solution [50 mM Hepes, 280 mM NaCl and 1.5 mM Na2HPO4 (pH 7.09)] and incubated for 20 min at room temperature (25°C). The mixture was then added dropwise to subconfluent cells for 6 h to overnight after which the medium was replaced.

SDS/PAGE and Western blotting

Cell monolayers were washed three times with ice-cold PBS and lysed at 4°C in immunoprecipitation lysis buffer [10 mM Hepes (pH 8), 100 mM NaCl, 1 mM MgCl2 and 0.5% Nonidet P40] containing 10 μg/ml aprotinin, 0.1 mM PMSF and 10 μg/ml leupeptin. Lysates were incubated on ice for 10 min and centrifuged at 9300 g for 15 min at 4°C. Samples were electrophoresed by SDS/PAGE (10% gel) and electroblotted on to a nitrocellulose membrane (Schleicher and Schuell BioScience). Membranes were blocked with 5% dried skimmed milk and 0.1% Tween 20 in TBS [Tris-buffered saline (20 mM Tris/HCl, 4 mM Tris base, 140 mM NaCl and 1 mM EDTA, pH 7.4)] for 1 h at room temperature and incubated with the primary antibody overnight at 4°C or 2 h at room temperature. The membranes were then washed three times in 0.1% Tween 20 in TBS and incubated with the appropriate secondary antibody for 1 h at room temperature. After washing, membranes were reacted with ECL (enhanced chemiluminescence) detection reagent (Santa Cruz Biotechnology) and analysed using a luminescent image analyser (Kodak X-OMAT 2000 processor). Where indicated, the blots were scanned using an Image Scan scanner (Amersham Pharmacia Biotech) and the intensity of each band was measured using ImageJ software (http://rsbweb.nih.gov/ij/).

Immunoprecipitation

At 48 h after transfection, cells were washed three times with ice-cold PBS and lysed as described above. The corresponding supernatants were pre-cleared for 2 h at 4°C with Protein A–agarose (Roche Diagnostics) after which they were centrifuged at 2300 g for 1 min at 4°C. The supernatants were incubated overnight at 4°C with the desired antibody immobilized on Protein A–agarose (Roche Diagnostics). Following four washes with lysis buffer containing protease inhibitors (10 μg/ml aprotinin, 0.1 mM PMSF and 10 μg/ml leupeptin), proteins were eluted for 5 min at 100°C with 5×SDS loading buffer, electrophoresed by SDS/PAGE (10% gel) and immunoblotted. The corresponding blot was reacted with the appropriate antibodies.

Transferrin endocytosis

Transfected cells grown on coverslips were incubated for 30 min in starvation medium [DMEM containing 0.1% BSA and 20 mM Hepes (pH 7.2)] to deplete transferrin present in the serum. Following incubation with 2.5 μg/ml Alexa Fluor® 546-conjugated transferrin (Molecular Probes) at 37°C, cells were rapidly cooled to 4°C, washed with ice-cold PBS, incubated with citrate buffer [25.5 mM citric acid, 24.5 mM sodium citrate, 280 mM sucrose and 0.01 mM deferoxamine (pH 4.6)] for 2 min and fixed with 4% paraformaldehyde (Merck). The fixed cells were mounted for microscopy using galvanol mounting solution (Mowiol 4-88, Calbiochem).

CT-B [CT (cholera toxin) subunit B] endocytosis

Continuous uptake of CT-B was performed by loading cells with 15 ng/ml Alexa Fluor® 555-conjugated CT-B (Molecular Probes) in DMEM supplemented with 20 mM Hepes and 1 mg/ml BSA at 37°C for 2 h. Cells were washed with PBS, incubated with citrate buffer for 2 min and fixed with 4% paraformaldehyde, followed by mounting using galvanol mounting reagent.

Immunofluorescence and confocal microscopy

For immunofluorescence and confocal microscopy, cells were grown on coverslips (Marienfeld) and transfected with the desired plasmids. After 16–20 h, the cells were washed with PBS and fixed with 4% paraformaldehyde for 15 min at room temperature, followed by additional PBS washes. Permeabilization was performed with 0.1% Triton X-100 in 50 mM Tris (pH 7.2) for 3 min, after which cells were washed with PBS and incubated in blocking buffer (20% normal goat serum and 1% BSA in PBS) for 30 min at room temperature. Cells were incubated with the appropriate antibody that was diluted in PBS containing 1% BSA for 1 h at room temperature, followed by three PBS washes. The coverslips were then incubated with the secondary antibody for 1 h at room temperature. Following washes with PBS, the cells were mounted using galvanol mounting reagent.

Cells were examined using Zeiss 510 or Zeiss 510 META laser-scanning confocal microscopes. For quantitative studies, all images for a given experiment were exposed and processed identically. Captured images were analysed using ImageJ software (http://rsbweb.nih.gov/ij/). The pixel intensity was used to quantify fluorescence in the indicated experiments.

Yeast two-hybrid assay

The EGY48 yeast strain, transformed with the pLexA/EHD2 plasmid, served to screen a human brain cDNA library cloned into pB42AD plasmid (Matchmaker LexA two-hybrid system, Clontech). Large-scale transformation with the two-hybrid library was performed according to the manufacturer's instructions. The yeast cells were plated first on glucose/histidine/uracyl/tryptophan selection plates which only permit growth of transformed yeast. The yeast were then collected and replated on galactose/raffinose/histidine/uracyl/tryptophanleucine/X-gal plates, which permit growth of colonies that express the LEU2 reporter gene. Growth of blue colonies indicated the expression of the lacZ reporter gene.

Rac1 activation assay

Cells transfected with a Rac1-expressing plasmid, with or without other plasmids, were lysed in p21 lysis buffer [25 mM Hepes (pH 7.4), 150 mM NaCl, 1% Nonidet P40, 10% glycerol, 10 mM MgCl2 and 1 mM EDTA] containing 10 μg/ml aprotinin, 0.1 mM PMSF and 10 μg/ml leupeptin. Samples containing 0.5 mg of total protein, as determined using the Bradford assay, were interacted with glutathione–agarose beads (G4510, Sigma) bound to 15 μg of GST (glutathione transferase)–PBD [Rac1-binding domain of PAK1 (p21-activated kinase 1)] [35] for 1 h at 4°C. The beads were washed three times in p21 lysis buffer, and the proteins were eluted for 5 min at 100°C with 5×SDS loading buffer, after which they were electrophoresed by SDS/PAGE (10% gel) and immunoblotted, as described above.

RESULTS

EHD2 interacts with Nek3 and associates with Vav1

Different publications have reported several roles for EHD2 in endocytic trafficking. Guilherme et al. [30] showed that EHD2 overexpression and its knockdown led to inhibition of transferrin internalization. Other studies have indicated that EHD2 regulates the exit of cargo from the recycling endosome [31,33]. In an attempt to confirm the role of EHD2 in endocytosis, in the present study we aimed to isolate novel EHD2-interacting proteins that might hint to its function.

We screened for possible interactors using the two-hybrid assay. EHD2 fused to the LexA DNA-binding domain was used as a bait to screen a human brain cDNA library. A fragment of Nek3, encoding the 176 C-terminal amino acids of Nek3 containing part of its putative coiled-coil domain but lacking its kinase catalytic domain, was isolated (Figures 1A and 1B). Nek3, a serine/threonine kinase, belongs to the NIMA protein family. NIMA kinase was first identified in A. nidulans. Temperature-sensitive mutants of NIMA were arrested in late G2-phase at restrictive temperatures, whereas overexpression of the NIMA protein caused the premature onset of mitotic events, including depolymerization of cytoplasmic microtubules, formation of aberrant mitotic spindles and chromatin condensation [36]. Neks share high amino acid sequence identity with the kinase domain of NIMA and have been reported in diverse eukaryotic species. They are highly similar to each other in their N-terminal catalytic domain, but show little, if any, conservation within their C-terminal non-catalytic domain. Although members of this family play a crucial role at the G2-to-M cell-cycle transition, Nek3 differs in its cytoplasmic localization and in its lack of effect on cell-cycle progression [37]. Since Nek3 does not contain any NPF motifs, we assume that its interaction with EHD2 is mediated through its coiled-coil region that was present in the isolated fragment (Figure 1B).

Figure 1 EHD2 interacts with Nek3

(A) The EGY48/p8op-lacZ yeast strain was transformed with pLexA/EHD2 or pLexA/glucocerebrosidase (as a non-relevant protein) or pLexA alone, all listed under ‘Bait’ in the Figure. These strains were further transformed with pB42AD/Nek3 (a prey plasmid containing the 3′ fragment of Nek3, which was isolated from the human brain cDNA library during the initial screen) or pB42AD alone (as a negative control), all listed under ‘Prey’ in the Figure. Four isolated colonies from each transformation were seeded on selection plates containing galactose/raffinose/histidine/uracyl/tryptophane/X-gal. Dark colonies represent protein–protein interactions. (B) Schematic illustration of Nek3 domains, as well as the C-terminal Nek3 fragment that was isolated in the yeast two-hybrid screen using EHD2 as the bait. (C) HEK-293 cells were transfected with GFP–EHD2 (EHD2) or Myc–Nek3 (Nek3) alone or together for 48 h. The cells were lysed in immunoprecipitation lysis buffer (see the Experimental section) and the corresponding lysates were subjected to immunoprecipitation with an anti-Myc antibody overnight at 4°C. Washed precipitates (and samples containing 5% of the lysates as input control) were processed by SDS/PAGE and Western blotting. The resulting membrane was reacted with anti-GFP antibodies to reveal GFP–EHD2 and with anti-Myc antibody to reveal Myc–Nek3. As illustrated in the Figure, Myc–Nek3 and GFP–EHD2 co-purified in the same complex. (D) For endogenous expression and co-immunoprecipitation, anti-EHD2 and anti-Nek3 antibodies were used. The molecular mass in kDa is indicated. IP, immunoprecipitation; NT, non-transfected; Wes, Western blot; Y2H, yeast two-hybrid.

The interaction between EHD2 and Nek3 was validated using co-immunoprecipitation experiments. The two proteins were found in the same complex, purified from HEK-293 cells expressing GFP–EHD2 and Myc–Nek3 (Figure 1C). We repeated the co-immunoprecipitation experiment using the endogenous proteins. As shown in Figure 1(D), endogenous EHD2 and Nek3 were purified in the same complex.

EHD2 is a serine phosphoprotein [38]. Therefore it was interesting to test whether Nek3 phosphorylates EHD2. To this end, we compared the tryptic phospho-peptide maps of EHD2, phosphorylated in vitro by Nek3, with that of EHD2 phosphorylated in cells. The results strongly suggested that Nek3 is not an EHD2 kinase (results not shown).

An interaction has been described between Nek3 and members of the Vav protein family, Vav1 and Vav2 [39,40]. Vav proteins are GEFs for the small Rho GTPases Rac1, RhoA and Cdc42 (cell division cycle 42). The interaction between Nek3 and Vav2 was dependent on prolactin induction, and Nek3 expression increased Vav2 phosphorylation. Nek3 also regulated Vav2 GEF activity towards Rac1. Nek3 siRNA (small interfering RNA) attenuated prolactin-mediated actin reorganization and activation of Rac1 GTPase. Nek3 expression also controlled paxillin serine phosphorylation, an activity that is closely related to the regulation of actin polymerization [41]. On the basis of these data, we tested a possible association between EHD2 and Vav1. Moreover, we tested the dependence of this association on binding of EHD2 to ATP. For that we used an EHD2 variant, mutated in its P-loop, EHD2T72N. The T72N mutant is supposed to mimic a NDP-bound state [8] based on alignment with sequences of other P-loop-containing proteins (Figure 2A). The results showed that Myc-tagged EHD2 and its P-loop mutant EHD2T72N co-immunoprecipitated Vav1 (Figure 2B). Namely, EHD2 and Vav1 are present in the same complex and their association does not depend on ATP binding. To test for a direct interaction between EHD2 and Vav1 we exploited the yeast two-hybrid system. The results showed that EHD2 does not directly bind Vav1 (Figure 2C).

Figure 2 EHD2 associates with Vav1

(A) Multiple alignment of the P-loop domain of different nucleotide-binding proteins. The alignment was performed using the ClustalW program (http://www.ebi.ac.uk/Tools/clustalw2/index.html) and the shading was performed using the MacBoxShade program. Proteins used for alignment: EHD1 (GenBank® accession number NP_006786.2), EHD2 (GenBank® accession number NP_055416.2), EHD3 (GenBank® accession number NP_055415.1), EHD4 (GenBank® accession number NP_644670.1), dynamin (GenBank® accession number NP_004399.2), Ras (GenBank® accession number NP_005334.1) and Arf6 (GenBank® accession number NP_001654.1). (B) Stable Vav1-expressing NIH 3T3 cells were transfected with Myc–EHD2 (EHD2) or Myc–EHD2T72N (T72N) for 48 h. The corresponding lysates were immunoprecipitated using an anti-Myc antibody overnight at 4°C. The washed precipitates (and samples containing 5% of the lysates as input control) were subjected to SDS/PAGE and Western blot analysis. The resulting membrane was reacted with an anti-Myc antibody to verify immunoprecipitation of Myc–EHD2 and Myc–EHD2T72N. The membrane was further reacted with anti-Vav1 antibodies. The results indicate co-immunoprecipitation of Vav1 with the Myc-tagged proteins. The molecular mass in kDa is indicated. (C) The EGY48 yeast strain was transformed with the plasmids pLexA/EHD2, pLexA/Vav1 or pLexA/EHD3 as listed under ‘Bait’. These strains were further transformed with constructs expressing pB42AD/Vav1, pB42AD/EHD2, pB42AD/EHD3 or pB42AD/Nek3 that were isolated as described in the legend to Figure 1(A), as listed under ‘Prey’. Four isolated colonies from each transformation were seeded on to selection plates containing Gal/Raff/Xgal/CMUraHisTrp. Growth of blue colonies indicates a protein–protein interaction. Ab, antibody heavy chain; IP, immunoprecipitation; NT, non-transfected; Wes, Western blot.

In summary, EHD2 directly binds Nek3 (Figure 1A) and is found in the same complex with Vav1. Since Nek3 and Vav1 were shown to directly interact with each other [39,40], we assume that their association is mediated by Nek3.

EHD2, but not its P-loop mutant variants, regulates Rac1 activity

The Nek3–Vav1 complex has been shown to participate in the regulation of actin reorganization [39,40] by modulating Rac1 activity. It was also shown that Vav1 regulates phagocytosis by mediating GDP/GTP exchange on Rac1 [42]. Therefore we decided to test whether EHD2 affects Rac1-GTP levels. To do this, HEK-293 cells were transfected with HA–Rac1, with or without EHD2 and its P-loop mutant variants EHD2G65R and EHD2T72N. The G65R mutant resembles the dominant-negative C. elegans RME-1G81R and mouse EHD1G65R mutants, which completely lost their affinity to membranes [13,16]. The resulting lysates were tested in pull-down assays using the Rac1/Cdc42 effector-binding domain of PAK3 fused to GST (GST–PBD), which specifically binds Rac1-GTP [43]. The results (Figures 3A and 3B) showed that the level of transfected HA–Rac1-GTP was reduced in the presence of EHD2, whereas, in the presence of mutant EHD2 variants, Rac1 activity was higher. Under the same experimental conditions, overexpression of EHD2 led to a reduction in the level of endogenous Rac1-GTP, whereas the P-loop mutants did not present the inhibitory effect (results not shown). These results imply that overexpression of EHD2 down-regulates Rac1 activity and that this function depends on the ability of EHD2 to bind ATP.

Figure 3 EHD2, but not its P-loop mutants, regulates Rac1 activity

(A) HEK-293 cells were transfected with HA–Rac1 (Rac1) or co-transfected with HA–Rac1 (Rac1) and either GFP–EHD2 (EHD2), GFP–EHD2G65R (EHD2G65R) or GFP–EHD2T72N (EHD2T72N) for 48 h. Lysates were prepared as described in the Experimental section and samples containing 0.5 mg of protein were used in a pull-down assay using GST–PBD. The pull-down products (I) were washed and processed by SDS/PAGE and Western blot analysis. Also, 5% of each lysate was subjected to SDS/PAGE and Western blot analysis as a loading control for Rac1 (II). The corresponding membranes were reacted with anti-Rac1 and anti-GFP antibodies as indicated. As depicted, EHD2 reduced Rac1-GTP levels compared with the mutants. The molecular mass in kDa is indicated. (B) To quantify the Rac1 pull-down results, band intensities were measured using ImageJ software. Rac-GTP in each lane was divided by its total Rac amount, and the ratio for ‘Rac1’ alone was considered as 1 (n=3 independent experiments). Significance was tested using Student's t test (P<0.0003). Rac1-GTP levels were the lowest when GFP–EHD2 was expressed. NT, non-transfected; PD, pull down; Wes, Western blot.

Guilherme et al. [30] showed that overexpression of EHD2 or disruption of its function by RNAi (RNA interference) inhibits endocytosis of transferrin to EEA1 (early endosome antigen 1)-positive endosomes [30]. They also showed that EHD2 localizes to cortical actin filaments and that high expression of EHD2 mediates extensive actin reorganization. Taking the previously published results [30] and the results of the present study together, we speculate that EHD2 modulates actin reorganization by regulating the level of active Rac1. If this is so, the inhibitory effect of overexpression of EHD2 on vesicle trafficking from the plasma membrane could be rescued by overexpression of Rac1.

EHD2 inhibits CT-B trafficking from the plasma membrane to the Golgi aparatus

It has been reported that disruption of actin dynamics inhibits CT-B trafficking from the plasma membrane to the Golgi apparatus [44]. Since we have shown that EHD2 affects Rac1 activity, which in turn regulates actin polymerization, we decided to test whether overexpression of EHD2 affects CT-B trafficking to the Golgi apparatus. CT, which binds GM1 on the plasma membrane, internalizes via clathrin-dependent and clathrin-independent mechanisms [45]. It does not recycle, but reaches the Golgi apparatus, from where it traffics to the cytoplasm through the ER (endoplasmic reticulum). To monitor the effect of EHD2 on CT-B internalization, continuous CT-B uptake was performed for 2 h in BSC-1 cells, expressing either GFP-tagged EHD2 or mutant EHD2. GFP–EHD2 inhibited CT-B trafficking to the Golgi apparatus, rendering its staining punctate and peripherally dispersed (Figure 4A-I). The dispersed CT-B staining highly co-localized with EHD2, suggesting that this cargo protein is attenuated in structures associated with the plasma membrane. The P-loop mutants had no effect on CT-B internalization, which reached the Golgi as in non-transfected cells (Figures 4A-II and 4A-III). To show that the peripheral localization of CT-B does not reflect redistribution of the Golgi apparatus, we followed Golgi staining in EHD2-overexpressing cells using the Golgi marker GT-CFP. As shown in Figure 4(B), the GT-CFP marker presented a typical Golgi localization in EHD2-overexpressing cells (I) as in non-transfected cells (II). Since the Golgi marker did not seem dispersed, it was concluded that the dispersed CT-B staining in EHD2-overexpressing cells did not reflect a redistribution of Golgi. Therefore it is possible that this mislocalization is a consequence of a trafficking problem. To quantify the results, the CT-B-staining pattern in cells expressing one of the three proteins tested and in non-transfected cells, was categorized as a Golgi or dispersed localization. As shown (Figure 4C), all non-transfected cells displayed Golgi staining, whereas ~75% of cells expressing EHD2 displayed a dispersed CT-B localization. On the other hand, in ~85% of the cells that expressed mutant GFP–EHD2, CT-B reached the Golgi apparatus.

Figure 4 Overexpression of EHD2, but not of its mutant variants, inhibits CT-B internalization

(A) BSC-1 cells, grown on coverslips, were transfected with GFP–EHD2, GFP–EHD2G65R or GFP–EHD2T72N for 16 h, after which they were incubated with Alexa Fluor® 555-conjugated CT-B for 2 h at 37°C, as described in the Experimental section. The cells were incubated in citrate buffer for 2 min, washed with ice-cold PBS and fixed with 4% paraformaldehyde. Mounted cells were imaged in a LSM 510 META confocal microscope. Representative images show (I) GFP–EHD2 (EHD2)-, (II) GFP–EHD2G65R (EHD2G65R)- and (III) GFP–EHD2T72N (EHD2T72N)-expressing cells. Arrowheads denote co-localization of EHD2 and CT-B on the plasma membrane. 1, a non-transfected cell in which CT-B is localized in the Golgi; 2, a transfected cell in which CT-B is dispersed and is co-localized with EHD2. (B) BSC-1 cells co-expressing YFP (yellow fluorescent protein)–EHD2 and GT-CFP (I) or GT-CFP alone (II) for 16 h were fixed, mounted and imaged using an LSM 510 META confocal microscope. (C) The dispersed compared with Golgi localization was counted in 70–100 cells expressing each protein. The percentage of each staining pattern out of the total cell count is graphically represented in a histogram (n=4 independent experiments). Significance was tested using Student's t test (P<0.001). Scale bars, 10 μm. NT, non-transfected.

Taken together, the results strongly support our suggestion that overexpression of EHD2, but not of its P-loop mutants, inhibited trafficking of CT-B from the plasma membrane to the Golgi apparatus.

ATP-bound EHD2, but not mutant EHD2 variants, inhibits transferrin trafficking from the plasma membrane

To confirm that overexpression of ATP-bound EHD2, but not its ADP-bound variants, inhibits trafficking from the plasma membrane, we followed internalization of transferrin in HeLa cells, transfected with either GFP-tagged EHD2 or with the EHD2 P-loop mutants. In agreement with previously published data [30], GFP–EHD2 inhibited transferrin internalization (Figure 5A-I). Interestingly, the two P-loop mutants internalized transferrin at much higher levels compared with wt (wild-type) EHD2 (Figures 5A-I–5A-III). Under the same experimental conditions, overexpression of GFP–EHD1 did not interfere with transferrin uptake (Figure 5A-IV). Quantification of transferrin intensity in EHD2-overexpressing cells showed an overall 50% reduction in transferrin uptake compared with non-transfected cells. Both mutants internalized much higher amounts of transferrin compared with EHD2-transfected cells (Figure 5B).

Figure 5 Overexpression of EHD2, but not of its mutant variants, inhibits transferrin internalization

(A) HeLa cells, grown on coverslips and transfected with either (I) GFP–EHD2 (EHD2), (II) GFP–EHD2G65R (EHD2G65R), (III) GFP–EHD2T72N (EHD2T72N) or (IV) GFP–EHD1 for 16 h, were iron-starved for 30 min. Following 5 min of uptake of Alexa Fluor® 546-conjugated transferrin at 37°C, the cells were cooled to 4°C, incubated in ice-cold citrate buffer (see the Experimental section) for 2 min, washed in ice-cold PBS and fixed with 4% paraformaldehyde. Mounted cells were imaged in an LSM 510 confocal microscope. As depicted in (I), GFP–EHD2 inhibited transferrin uptake when overexpressed (cell number 1), whereas GFP–EHD2G65R (cell number 2) (II), GFP–EHD2T72N (cell number 3) (III) and GFP–EHD1 (cell number 4) (IV) did not inhibit transferrin internalization. (B) ImageJ quantification of transferrin pixel intensities measured in 60–80 cells per each protein and in non-transfected cells. The histogram shows averaged intensities as a percentage of non-transfected cells (n=4 independent experiments). Transferrin was imaged using the same conditions. Significance was tested using one-way ANOVA (P<0.0001). Scale bars, 10 μm.

The results show an ATP-dependent inhibition of transferrin trafficking from the plasma membrane by EHD2 overexpression.

Rescue of the EHD2 phenotype by overexpression of Rac1

Since actin dynamics is regulated by Rac1, and since EHD2 reduces Rac1 activity, we attempted to rescue the EHD2 phenotype towards CT-B trafficking by overexpressing Rac1 together with EHD2. Although in cells overexpressing GFP–EHD2 CT-B exhibited a dispersed punctate localization (Figure 4A-I), in cells overexpressing GFP–EHD2 and HA–Rac1 CT-B reached a perinuclear Golgi-like localization (Figure 6A). To quantify this effect, cells expressing GFP–EHD2 and HA–Rac1 were analysed for dispersed compared with Golgi localization of CT-B. As shown (Figure 6B), there was an increase in trafficking of CT-B to the Golgi apparatus in cells overexpressing EHD2 and Rac1 in comparison with cells overexpressing EHD2 alone, implying that overexpression of Rac1 overcame the block in internalization caused by overexpressing EHD2.

Figure 6 Overexpression of HA–Rac1 in EHD2-expressing cells rescues the attenuated CT-B transport to the Golgi apparatus

(A) BSC-1 cells were transfected with GFP–EHD2 (EHD2) and HA–Rac1 (Rac1) for 16 h. Alexa Fluor® 555-conjugated CT-B was added for 2 h at 37°C as described in the Experimental section. The cells were rinsed with citrate buffer for 2 min, washed with ice-cold PBS and fixed with 4% paraformaldehyde. Following immunostaining with an anti-HA antibody, the cells were stained with a Cy5-conjugated secondary antibody as described. Mounted cells were imaged on an LSM 510 META confocal microscope. The cells in the inset express GFP–EHD2 and HA–Rac1 and present Golgi localization of CT-B. (B) The peripheral compared with Golgi localization of CT-B was counted in 83 cells of the described experiment. The percentage of each staining pattern out of a total cell count is graphically represented in a histogram (n=3 independent experiments). Significance was tested using Student's t test (P<0.0005). (C) BSC-1 cells, co-expressing HA–Rac1 (Rac1) and GT-CFP (I) or GT-CFP alone (II) for 16 h, were fixed and processed for immunofluorescence. The cells were stained with an anti-HA antibody following staining with a Cy3-conjugated secondary antibody. Scale bars, 10 μm. Overexpression of Rac1 rescued the attenuated CT-B transport to the Golgi apparatus.

The Golgi structure in cells overexpressing HA–Rac1 seemed to differ from that in non-transfected cells (Figure 6A). This resulted from the overexpression of Rac1 as depicted by the Golgi marker GT-CFP, which showed a change in Golgi morphology in cells expressing HA–Rac1 compared with non-transfected cells (compare Figure 6C-I with Figure 6C-II).

Taken together, the results strongly imply that overexpression of EHD2 inhibited trafficking of transferrin and CT-B from the plasma membrane by regulating Rac1 activity, through a Nek3 interaction and association with Vav1.

DISCUSSION

It has been shown previously that EHD2 connects endocytosis to the actin cytoskeleton through interaction of its N-terminal domain with the membranes and its C-terminal domain with EHBP1 [30]. To further understand the role of EHD2 in endocytosis and actin remodelling, in the present study we aimed to isolate interacting proteins that would unravel cellular complexes in which EHD2 is involved. We found that EHD2 associates with Nek3 and the GEF Vav1. Vav1 has already been shown to regulate Rac1 activity [42], which prompted us to test whether EHD2 affects endocytosis by regulating Rac1 activity. The results of the present study complement the previously published data [30] and further show that EHD2 regulates Rac1-GTP levels. Rac1-GTP levels modulate actin remodelling [46].

We have shown that overexpression of EHD2 inhibited trafficking of CT-B or of transferrin from the plasma membrane. It has already been shown that overexpression of EHD2 down-regulates transferrin internalization to the early endosome [30]. Interestingly, the EHD2 P-loop mutants G65R and T72N failed to attenuate trafficking from the plasma membrane and had no effect on Rac1 activity. Since it has previously been shown that EHD2 binds ATP [10], the results of the present study imply that the effect of EHD2 on Rac1 activity depends on ATP binding. It is worth noting that overexpression of EHD2 P-loop mutants or of EHD1 did not inhibit trafficking, arguing against an artefact resulting from overexpression of EHD2. Moreover, the same phenomenon of trafficking inhibition was observed when the plant orthologue of EHD2 was overexpressed in tomato [15]. Thus absence [30] or overexpression of EHD2 attenuates internalization, namely, overexpression of EHD2 has a dominant-negative effect on internalization.

Puzzled by the functional differences between wt and P-loop mutants of EHD2, we noticed that whereas EHD2 displays a mostly punctate membrane localization with some even distribution, the mutants were mostly evenly spread on the plasma membrane, exhibiting lesser abundance in punctate structures (Supplementary Figure S1 at http://www.BiochemJ.org/bj/439/bj4390433add.htm). To examine whether the differences in organization and distribution of EHD2 and its ADP-bound mutants reflect alterations in the mode of association of these proteins with the plasma membrane, we expressed GFP–EHD2 and GFP–EHD2G65R in cells and subjected them to FRAP (fluorescence recovery after photobleaching) analysis (Supplementary Figure S2 at http://www.BiochemJ.org/bj/439/bj4390433add.htm). In cells expressing EHD2, less of the bleached signal was recovered as compared with the mutant (26% in EHD2-expressing cells in comparison with 41% in the mutant-expressing cells). Furthermore, in the bleached areas fluorescence recovered only as evenly distributed signal (Supplementary Figure S2B), arguing that EHD2 in the punctate structures was less mobile. Since the pattern exhibited by GFP–EHD2G65R was mainly of even distribution, the level of its fluorescence recovery was higher than that of EHD2. This difference in the physical properties of wt and mutant EHD2 explains the functional differences, exemplified by their different abilities to attenuate internalization.

Nek3 was shown to affect the prolactin-mediated phosphorylation of Vav2 as well as its GEF activity. The Nek3–Vav1 complex was shown to participate in the regulation of actin reorganization [39,40].

Vavs are GEFs that couple cell-surface receptors to various effector functions. There are three known evolutionarily conserved Vavs which differ in their expression pattern [47,48], but contain the same characteristic structural motifs. These include a DH (dbl homology) region, which exhibits a GEF activity towards the Rho family GTPases, a PH (pleckstrin homology) domain, which interacts with polyphosphoinositides, and a CH (calponin homology) region, which functions as an actin-binding domain in other proteins [4952].

Vav1 was originally isolated as an in vitro-activated oncogene [53]. Nucleotide sequence analysis of the Vav1 oncogene revealed that it was activated in vitro by replacement of 67 residues of its N-terminus with sequences of pSV2neo, co-transfected as a selectable marker [54]. Previously, molecular lesions in the N-terminus of Vav2 and Vav3 were shown to render these Vav family members transforming [52,55,56]. It is conceivable that Vav oncogene-induced transformation is the result of deregulated GEF activity, which leads to disruption of actin filaments and a decrease in focal adhesions, common features of cellular transformation by various oncogenes [57,58].

Ablation of Vav1 in macrophages inhibited early stages of phagocytosis, Arp2/3 recruitment to phagosomes and actin polymerization. This phenotype could be rescued by overexpression of active Rac1 [59].

On the basis of the work described above and of the interactions described in the present study, we suggest that EHD2 is part of an actin-regulatory complex along with Vav1 and Nek3. We suggest that, under normal expression levels, EHD2 is necessary for assembly of the Nek3–Vav1–EHD2 complex, which modulates actin reorganization at sites of endocytosis in the plasma membrane.

Since endocytosis is an essential process in the normal life of a cell, its attenuation should have a noticeable effect. In fact, we observed that increasing amounts of EHD2 were deleterious to cells. Thus when we tried to establish lines of CHO (Chinese-hamster ovary) cells stably expressing EHD2 using the DHFR (dihydrofolate reductase)-MTX (methotrexate) system [60], we could obtain EHD2-overexpressing colonies in the presence of up to 10 μM MTX. All colonies grown at 100 μM MTX lost their EHD2 overexpression, arguing that high EHD2 levels are toxic to the cells (D. Rapaport and M. Horowitz, unpublished work).

In summary, in the present study we show that EHD2 interacts with Nek3, associates with Vav1 and modulates Rac1 activity. We assume that, by doing so, it modulates actin reorganization at internalization sites during the first stages of endocytosis.

AUTHOR CONTRIBUTION

Sigi Benjamin and Mia Horowitz designed the experiments. Sigi Benjamin conducted the experiments. Hilla Weidberg conducted the yeast two-hybrid screen. Marina Nudelman performed the EHD2–Vav1 yeast two-hybrid analysis. Debora Rapaport and Olga Pekar performed the endogenous co-immunoprecipitation experiments. Koret Hirschberg and Marcelo Ehrlich collaborated on live imaging experiments and provided scientific advice. Shulamit Katzav provided crucial materials. Daniel Segal provided scientific advice. Sigi Benjamin and Mia Horowitz wrote the paper. Debora Rapaport proofread the paper.

FUNDING

This work was supported the Jacqueline Seroussi Memorial Foundation for Cancer Research; the German-Israeli BioDisc (to M.H.); and the Marine Biological Laboratory (Woods Hole, MA, U.S.A). Part of this work was performed at the Marine Biological Laboratory, Woods Hole, MA, U.S.A.

Acknowledgments

We thank Dr A. Zundelevich for valuable advice.

Abbreviations: Arp2/3, actin-related protein 2/3; Cdc42, cell division cycle 42; CFP, cyan fluorescent protein; CT, cholera toxin; CT-B, CT subunit B; Cy3, indocarbocyanine; Cy5, indodicarbocyanine; DMEM, Dulbecco's modified Eagle's medium; EH, Eps15 homology; EHD, EH-domain-containing protein; EHBP1, EHD2-binding protein 1; GEF, guanine-nucleotide-exchange factor; GFP, green fluorescent protein; GST, glutathione transferase; HA, haemagglutinin; HEK, human embryonic kidney; MTX, methotrexate; Nek, NIMA (never in mitosis in Aspergillus nidulans)-related kinase; NIMA, never in mitosis in Aspergillus nidulans; NPF, Asn–Pro–Phe; N-WASP, neuronal Wiskott–Aldrich syndrome protein; ORF, open reading frame; PAK, p21-activated kinase; PBD, Rac1-binding domain of PAK1; TBS, Tris-buffered saline; wt, wild-type

References

View Abstract