Association of insulin-like growth factor 1 receptor with EHD1 and SNAP29.

Ligand-induced receptor-mediated endocytosis plays a central role in regulating signaling conveyed by tyrosine kinase receptors. This process depends on the recruitment of the adaptor protein 2 (AP-2) complex, clathrin, dynamin, and other accessory proteins to the ligand-bound receptor. We show here that besides AP-2 and clathrin, two other proteins participate in the endocytic process of the insulin-like growth factor receptor (IGF-1R); they are EHD1, an Eps15 homology (EH) domain-containing protein 1, and SNAP29, a synaptosomal-associated protein. EHD1 and SNAP29 form complexes with alpha-adaptin of AP-2 and co-localize in endocytic vesicles, indicating a role for them in endocytosis. EHD1 and SNAP29 interact directly with each other and are present in complexes with IGF-1R. After IGF-1 induction, EHD1 and IGF-1R co-localize intracellularly. Overexpression of EHD1 in Chinese hamster ovary cells represses IGF-1-mediated signaling, as measured by mitogen-activated protein kinase phosphorylation and Akt phosphorylation, indicating that EHD1 plays a role as a down-regulator in IGF-1 signaling pathway.

endophilin, and dynamin are recruited to the plasma membrane to promote the formation of clathrin-coated vesicle (9 -12). The dynamic assembly of these complexes is mediated through various protein recognition modules. One of the modules is the EH domain that was first identified as a 100-amino acid sequence repeated 3 times in the N terminus of the EGF receptor pathway substrate Eps15 (13)(14)(15)(16)(17)(18). The central region of Eps15 shares the characteristic heptad repeats of coiled-coil proteins, and the C terminus contains a prolin-rich region and a repeated DPF motif (16,17). Intracellular localization to plasma membrane pits and vesicles, interaction with prominent proteins of the basal endocytic machinery like ␣-adaptin of AP-2, and interactions through the EH domain with other proteins harboring the NPFXD sequence like epsin have implicated Eps15 in receptor-mediated endocytosis (18 -22). Moreover, functional inhibition of Eps15 by antibodies or dominant negative mutants abrogated endocytosis of EGF and transferrin, therefore supporting the notion that Eps15 mediates endocytosis of the EGF receptor (21,23). Recently a family of EH domain-containing proteins harboring four members, EHD1-EHD4, was identified (24,25). 2 These proteins share an EH (Eps15 homology) domain in their C terminus including an EF hand, Ca 2ϩ binding motif, a central coiled-coil region, and a nucleotide binding consensus site at the N-terminal domain. Northern analysis and immunohistochemical studies indicated specific expression of EHD1 in tissues and cell types that depend on IGF-1 for differentiation and maintenance (24). In cells in tissue culture EHD1 was localized in transferrin-containing endocytic vesicles. Structural resemblance to Eps15 and intracellular localization of this protein in endocytic vesicles proposed a role for EHD1 in endocytosis. Here we present data indicating that EHD1 associates with proteins of the endocytic machinery as well as with IGF-1R and SNAP29. The results strongly suggest that EHD1 and SNAP29 participate in IGF-1-induced IGF-1R endocytosis.

EXPERIMENTAL PROCEDURES
Antibodies-Polyclonal antibodies raised against the IGF-1R ␤-chain (C-20, sc-713) were purchased from Santa Cruz Biotechnology. ACM11, mouse monoclonal anti-␣-chain of AP-2 antibodies, were a kind gift from Dr. M. S. Robinson (26). Anti-clathrin heavy chain antibodies (X22) were purchased from American Type Culture Collection. Monoclonal anti-IGF-1R ␣-chain antibodies (2C8, sc-463) for immunoprecipitation were purchased from Santa Cruz Biotechnology. Monoclonal anti-IGF-1R ␣-chain antibodies for immunofluorescence were purchased from Oncogene. Secondary antibodies conjugated to horseradish peroxidase were purchased from The Jackson Laboratory. Anti-phospho-MAP kinase monoclonal antibodies (442705-S) and anti-MAP ki-* This research is supported by grants from the Israel Cancer Research Foundation (ICRF), the Ministry of Health, and the Ela Kodesh Institute for Research on Cancer Development and Prevention (to M. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Funded by a post-doctoral fellowship from the Valazzi-Pikovsky foundation.
Affinity Purification-One grain of rat tissues was homogenized in 4 ml of 20 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM MgCl 2 , 0.2% Triton X-100 buffer supplemented with protease inhibitor mixture (Sigma, P8340) and phosphatase inhibitor mixture (Sigma, P5726). The homogenates were centrifuged for 15 min at 14,000 rpm in SS-34 rotor. The extract was affinity-purified on rEHD1 bound to CNBr-activated Sepharose 4B (Amersham Pharmacia Biotech) according to manufacturer's manual. Column-bound material was washed three times with PBS, and three elutions were performed with 0.1 M glycine, pH 2.5, which was titrated immediately by 1 M Tris, pH 8.8. Starting material, flow-through, washes, and elutions were resolved on SDS-PAGE and subjected to Western analysis. For affinity purification of rat proteins on a rSNAP29-nickel column, rat tissues were homogenized as described above. Half the extracts were incubated with 3 g/ml purified rSNAP29 for 2 h at 4°C. All the extracts were further incubated for 2 h at 4°C with 30 l of PBS-washed (50% slurry) nickel beads (Novagen). The beads were washed three times with 0.2% Triton X-100 Hepes buffer and boiled in Laemmli sample buffer containing 5% ␤-mercaptoethanol (Sigma).
In Vitro Complex Formation and Immunoprecipitation-For in vitro complex formation, 2 ϫ 10 6 NIH/IGF-1R cells were serum-starved for 10 h in DCCM-1 medium without insulin or IGF-1 supplemented with 2 mM L-Glutamine solution and PEN-STREP solution (Biological Industries). The cells were treated with 100 ng/ml human IGF-1 (PeproTech) for different time intervals. After the treatment, the cells were lysed in 20 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM MgCl 2 , 0.2%Triton X-100 buffer supplemented with a protease inhibitor mixture (Sigma, P8340) and a phosphatase inhibitor mixture (Sigma, P5726). The extracts were incubated with 3 g/ml purified rEHD1 for 2 h at 4°C and subjected to immunoprecipitation using anti-IGF-1R ␣-chain antibodies conjugated to immobilized rProteinA (Repligen Corp.). The proteins were resolved on SDS-PAGE, and Western blot analysis was performed. For immunoprecipitation of proteins derived from rat tissues or tissue-cultured cell lines, the tissues and cells were lysed in 0.2% Triton X-100, Hepes buffer as described above. The extracts were cleared by centrifugation, and supernatants were subjected to immunoprecipitation using the relevant antibodies conjugated to immobilized rProteinA. The beads were washed three times in the same buffer containing 0.05% Triton X-100 and boiled in Laemmli sample buffer supplemented with 5% ␤-mercaptoethanol (Sigma) before loading on SDS-PAGE.
Western Blot Analysis-Proteins were resolved on SDS-PAGE and transferred to a nitrocellulose blot, and the immunoblot was decorated with the relevant antibodies for 2 h at room temperature or overnight at 4°C. Detection was carried out using horseradish peroxidase conjugated to a specific secondary antibody followed by enhanced chemiluminescence reaction (Amersham Pharmacia Biotech).
MAP Kinase and Akt Phosphorylation-CHO cells and CHO cells overexpressing EHD1 (CHO/EHD1) were serum-starved for 10 h and treated with 100 ng/ml human IGF-1 (PeproTech) for different time intervals. The cells were extracted in Laemmli sample buffer, and the proteins were resolved on SDS-PAGE and subjected to Western analy-sis using anti-phospho-MAP kinase and anti-MAP kinase antibodies or anti-phospho-Akt and anti-Akt antibodies. The autoradiographs were quantitated by phosphorimaging, and the values for the phosphorylated-MAPK or phosphorylated Akt were normalized to the MAPK or Akt values, respectively.
Two-hybrid System Library Screen-Human EHD1 and mouse EHD3 open reading frames were individually cloned in-frame into the EcoRI and the XhoI restriction sites in the "bait" pLexA vector (Interaction trap matchmaker system, CLONTECH) to provide the pLexA/ EHD1 and pLexA/EHD3, respectively. Mutant form of EHD1 missing 0.7 kilobases of the N terminus was constructed by cloning a BamHI-XhoI fragment of EHD1 cDNA into pLexA to provide the pLexA/EHD1/ NTM. A mutant form of EHD1 missing 0.3 kilobases of the C terminus encompassing the EH domain was constructed by cloning an EcoRI-HincII fragment of the EHD1 cDNA into pLexA to provide the pLexA/ EHD1/EHM. All the constructs were used to transform the EGY48 yeast strain and were tested for protein expression, entrance to the nucleus, and inability to activate the system. Yeast harboring the pLexA/EHD1 served to screen human brain library according to the manufacturer's instructions (Interaction trap matchmaker system). pB42AD vector encoding the SNAP29 gene was rescued and tested for interaction with yeast strains expressing the pLexA fusion proteins described above. PB42AD vector expressing a mutant form of SNAP29 (SNAP29/NTM), missing the first 30 amino acids of the N terminus, was constructed by amplifying a SNAP29 cDNA fragment with the sense primer 5Ј-gcggacagggaattccagcagtac-3Ј and the antisense primer 5Ј-tccgtctgtctctcgagtcagagttg-3Ј. The PCR fragment was cloned into the EcoRI-XhoI sites of the pB42AD plasmid to provide pB42AD/SNAP29/ NTM. SNAP29 open reading frame was cloned into EcoRI and XhoI restriction sites of pLexA to produce pLexA/SNAP29. Human IGF-1R ␤-chain cytoplasmic tail (from amino acid 970 to the end of the IGF-1R ␤-chain) was amplified from an human placenta cDNA library (CLONT-ECH) using the sense primer 5Ј-GCAGGCTGGAATTCGGAGTGCTGT-ATGCC-3Ј and the antisense primer 5Ј-GATCCACTGAGGTACAGGA-GGCTTGTG-3Ј and was cloned in-frame into the EcoRI and the BamHI sites of the bait vector pLexA (Interaction trap matchmaker system). The human IGF-1R ␤-chain cytoplasmic tail PCR fragment was also cloned into the EcoRI and XhoI sites of the pB42AD vector. The transformed yeast colonies were first grown on glucose/histidineϪ/ura-cylϪ/tryptophanϪ plates, and four individual colonies from each transformation were assayed for growth either on galactose/raffinose/ histidineϪ/uracylϪ/tryptophanϪ/leucineϪ or galactose/raffinose/his-tidineϪ/uracylϪ/tryptophanϪ/X-gal plates. Growth on leucineϪ plates or blue-colored colonies on X-gal plates indicated interaction between the bait and the prey recombinant proteins.
Immunofluorescence and Confocal Microscopy-For SNAP29 and EHD1 localization, CHO cells grown on cover glasses for 1 day were transfected with pEGFP-C3-EHD1 (24) for 48 h. After rinsing with PBS at room temperature, cells were fixed for 30 min with 3.7% paraformaldehyde. After permeabilization with 0.1% Triton X-100 for 4 -5 min, blocking was performed in 5% normal goat serum (in PBS) for 1 h at 37°C. The cells were incubated with anti-rSNAP29 antibodies at room temperature for 1 h. After rinsing with PBS, the slides were incubated with the monoclonal indocarbocyanin (Cy3)-conjugated goat anti-rabbit antibodies (The Jackson Laboratory) for 30 min at room temperature. After washes with PBS, the cover glasses were mounted on a glass slide in mounting reagent (Galvanol). For EHD1 and IGF-1R co-localization, CHO cells grown on cover glasses for 1 day were starved overnight in a serum-free DCCM-1 medium. After treatment of 1 h with 100 ng/ml IGF-1 at 4°C, they were incubated at 37°C for 10 min, fixed, and stained with polyclonal anti-EHD1 antibodies and monoclonal anti-IGF-1R antibodies. The cells were decorated with the corresponding, fluorescently labeled, secondary antibodies.

Interaction of EHD1 with Proteins of the Endocytic Machinery and IGF-1R-
To explore the proteins that associate with EHD1, rat testes extract was affinity-purified on a rEHD1 affinity column. Eluted proteins were subjected to Western analysis. As depicted in Fig. 1a, ␣-adaptin of AP-2 and the heavy chain of clathrin, proteins characteristic of the endocytic machinery, were found in complexes with rEHD1. REHD1 formed complexes with three proteins with molecular masses of 63, 68, and 80 kDa, which were detected by anti-rEHD1 antibodies. The 63-kDa protein represents EHD1 and EHD3, whereas the 68-kDa represents EHD2 and EHD4 (25). Actually we have shown that EHD1 is able to undergo self-dimerization, as demonstrated for Eps15 (27) and to form hetero-oligomers with EHD3, 2 EHD2, and EHD4. 3 The proteins designated as 1 and 2, Fig. 1a, may represent other unknown proteins that share an antigenic determinant(s) with EHD1, are unable to bind EHD1, and are recognized by anti-EHD1 antibodies. The 80-kDa protein may represent another unknown EH domaincontaining protein that shares antigenic determinant(s) with EHD1, is able to bind EHD1, and is recognized by anti-EHD1 antibodies. Interestingly, EHD1 formed complexes with the ␤-chain of the IGF-1 receptor. These results indicate association of EHD1 with proteins of the endocytic machinery and with the IGF-1 receptor.
In Vitro Complex Formation between EHD1 and IGF-1R-An in vitro assay was developed to confirm the interaction of EHD1 with IGF-1R. NIH3T3 cells over-expressing the human IGF-1 receptor (NIH3T3/IGF-1R) (28) were treated with IGF-1 for different time intervals. Extracts prepared from these cells were incubated with rEHD1 and further subjected to immunoprecipitation using monoclonal anti-IGF-1R ␣-chain antibodies, and the Western blot was decorated with either anti-IGF-1R ␤-chain antibodies or anti-EHD1 antibodies. As depicted in Fig. 1b, the level of the IGF-1 receptors recognized by both anti-␣-chain and anti-␤-chain antibodies was elevated at up to 5 min of IGF-1 treatment, whereas within 10 min of treatment, their level declined. Minor changes were observed in the level of IGF-1R precursor, which migrated as a 180-kDa protein, indicating that only the mature receptors responded to IGF-1 induction, presumably by conformational changes that prevented their antibody recognition. rEHD1 was shown to associate with IGF-1R. The association increased within 5 min of IGF-1 induction and decreased to basal level after 15 min of treatment. These results demonstrate an association between rEHD1 and IGF-1R that depends on IGF-1 stimuli.
Intracellular Colocalization of IGF-1R and EHD1-To confirm the interaction between IGF-1R and EHD1, we tested their intracellular localization. CHO cells overexpressing EHD1 were serum-starved overnight, treated with IGF-1 for 1 h, fixed, and reacted with anti-EHD1 rabbit polyclonal antibodies or anti-IGF-1R monoclonal antibodies as detailed under "Experimental Procedures." The cells were visualized using confocal microscopy. As indicated in Fig. 1c, there is an intracellular colocalization of IGF-1R and EHD1, confirming their interaction.
MAP Kinase and Akt Phosphorylation in EHD1 Overexpressing Cells-IGF-1 mediates the MAPK and Akt-signaling cascades (29). To verify whether EHD1 influences the signal transduced by IGF-1 receptor, phosphorylation of MAP kinase (MAPK) or Akt was followed in CHO cells and in CHO cells overexpressing the EHD1 gene (CHO/EHD1) 4 in response to IGF-1. As depicted in Fig. 2a, treating CHO cells with IGF-1 leads to phosphorylation of MAPK within 2.5 min, which reaches high levels at 15 min of treatment. In contrast, in CHO/EHD1 cells, the phosphorylation of MAPK is retarded, and at 15 min, the phosphorylation is 2.3-fold reduced compared with the parental cells. As indicated in Fig. 2b, IGF-1 treatment leads to Akt phosphorylation within 2.5 min, which reaches high levels at 5 min of treatment. In contrast, in CHO/EHD1 cells, the phosphorylation of Akt is retarded, and at 5 min, the phosphorylation is 2.2-fold reduced compared with the parental cells. Therefore, overexpression of EHD1 reduces IGF-1-induced signaling as measured by MAPK and Akt phosphorylation, suggesting that EHD1 is a down-modu- 3 1. EHD1 interacts with AP-2, clathrin, and IGF-1R. a, extract from rat testes was affinity-purified on rEHD1 bound to CNBractivated Sepharose 4B beads. Starting materiel (L), flow-through (FT), wash (W), and elutions (E1-E3) were resolved on SDS-PAGE and transferred into nitrocellulose, and a Western blot was employed using anti-IGF-1R ␤-chain antibodies, anti-rEHD1 antibodies, antibodies directed against ␣-adaptin of AP-2, and anti-clathrin heavy chain antibodies. The arrows designated as 1 and 2 represent proteins that were recognized by anti-EHD1 antibodies in Western blot, and their identity is not known. b, NIH/IGF-1R cells were treated with IGF-1 for different time intervals. The cells were lysed, and the extracts were incubated with rEHD1 for 2 h and subjected to immunoprecipitation using anti-IGF-1R ␣-chain antibodies. The proteins were resolved on SDS-PAGE, and immunoblot was performed using anti-IGF-1R ␤-chain antibodies and anti-EHD1 antibodies. c, CHO cells overexpressing EHD1 were induced with IGF-1. The cells were stained with anti-EHD1 antibodies and anti-IGF-1R antibodies followed by relevant secondary-labeled antibodies. EHD1 (red), IGF-1R (green), and co-localization (yellow) were visualized by confocal analysis. lator of the signaling pathway.
Another approach was applied to show that overexpression of EHD1 down-regulates IGF-1-induced signaling. NIH3T3/ IGF-1R cells were co-transfected with an EHD1 expression vector (pCDNA1/EHD1) and a reporter vector harboring the CD4 cDNA under a VEGF promoter (pVEGF/CD4). After an overnight starvation, the cells were treated with IGF-1 for 18 h and harvested. RNA was extracted and used for reverse transcriptase-PCR reactions using primers specific for CD4 and EHD1 (as explained under "Experimental Procedures"). We found an ϳ10-fold reduction in the amount of CD4 RNA in cells overexpressing EHD1 versus those that do not after IGF-1 induction (Fig. 2c).
Isolation of SNAP29 as an EHD1-interacting Protein-We searched for potential binding partners of EHD1 using the yeast two-hybrid system. A bait construct comprising the EHD1 fused to the LexA protein was used to screen a human brain cDNA library. One isolated clone encoded the human SNAP29. SNAP29, a 29-kDa synaptosomal-associated protein, shares structural homology with SNAP25 and SNAP23, proteins of the SNARE complex, participating in vesicular membrane fusion events (30). Unlike the latter, SNAP29 contains at its N terminus two motifs, NPFXD sequence (amino acids 8 -12) and PXXPXXP (amino acids 32-38), which are characteristic of proteins participating in trafficking, especially in endocytosis. As shown in Fig. 3a, SNAP29 was able to interact with EHD1, EHD3, and truncated EHD1 missing either its N . The autoradiographs were quantitated by phosphorimaging scanner, and the values for the phosphorylated-MAPK or phosphorylated-Akt were normalized to the MAPK and Akt values, respectively. c, NIH/IGF-1R cells transfected with pCDNA1, pCDNA1/EHD1 (EHD1), and pVEGF/CD4 (CD4) were induced with IGF-1, as indicated. Cytoplasmic RNA was prepared, and reverse transcriptase-PCR was performed. Monitoring EHD1 and CD4 expression was executed by PCR reactions and Southern blot. 3. EHD1 forms complexes with SNAP29 and IGF-1R. a, EGY48/p8op-lacZ yeast strain was transformed with the plasmids pL-exA/EHD1, pLexA/EHD3, or plasmids harboring truncated forms of EHD1 missing the N terminus (pLexA/EHD1/NTM) or the EH domain (pLexA/EHD1/EHM) as marked to the left margins. These strains were further transformed with constructs expressing pB42AD/SNAP29 or pB42AD/SNAP29/NTM, which encodes a truncated SNAP29 as marked at the top. Four isolated colonies from each transformation were seeded on selection plates containing either galactose/raffinose/uracylϪ/histi-dineϪ/tryptophanϪ/X-gal (X-gal), galactose/raffinose/uracylϪ/histi-dineϪ/tryptophanϪ/leucineϪ (LeuϪ), or glucose/uracylϪ/histidineϪ/ tryptophanϪ (UraϪ/HisϪ/TrpϪ), as marked at the top. Growth on LeuϪ plates or blue colonies on the X-gal plates indicates proteinprotein interaction. b, extracts from rat tissues were supplemented with (ϩ) or without (Ϫ) rSANP29 for 2 h and further incubated with nickel beads. The beads were washed, and bound material was separated on SDS-PAGE and subjected to Western analysis using anti-EHD1 antibodies. c, extracts from rat tissues were immunoprecipitated using anti-IGF-1R ␣-chain antibodies. Immuno-complexes were resolved on SDS-PAGE and transferred into nitrocellulose, and a Western blot was employed using either anti-rEHD1 antibodies or anti-rSNAP29 antibodies. d, extract from NIH3T3/IGF-1R cells was subjected to immunoprecipitation using anti-IGF-1R ␣-chain antibodies. Immuno-complexes were resolved on SDS-PAGE and transferred into nitrocellulose, and a Western blot was employed using either anti-IGF-1R ␤-chain antibodies or anti-rSNAP29 antibodies. e, EGY48/p8op-lacZ Yeast strain was transformed with the plasmid pLexA/SNAP29 as marked in the left margins. This strain was further transformed with constructs expressing either pB42AD/IGF-1R ␤-chain cytoplasmic tail, pB42AD/SNAP29, or pB42AD/SNAP29/NTM, as marked in the right margins. Four isolated colonies from each transformation were seeded on selection plates containing either galactose/raffinose/uracylϪ/histidineϪ/trypto-phanϪ/X-gal (X-gal), galactose/raffinose/uracylϪ/histidineϪ/trypto-phanϪ/leucineϪ (LeuϪ) or glucose/uracylϪ/histidineϪ/tryptophanϪ (UraϪ/HisϪ/TrpϪ). Growth on LeuϪ plates or blue colonies on the X-gal plates indicates protein-protein interactions.
terminus or its C terminus, indicating it interacts with the central coiled-coil region of EHD1. On the other hand, a mutated version of SNAP29 lacking amino acids 1-38 or other non-relevant genes like Rubisco of tomato or human clathrin light chain (data not shown) were unable to interact with any of the EHD1-derived proteins (Fig. 3a). Nonetheless, the truncated SNAP29 was able to interact with SNAP29, although to a lesser extent than the wild type SNAP29 protein, as shown in Fig. 3d by growth on the LEUϪ plate. These results indicate that the interaction between SNAP29 and EHD1 is directed through modules within those 38 amino acids of SNAP29.
To further demonstrate the interaction between SNAP29 and EHD1, rat tissue extracts were affinity-purified on a rSNAP29 column. The eluates were subjected to SDS-PAGE, and the imunoblot was decorated with anti-EHD1 antibodies. As shown in Fig. 3b, EHD proteins from rat brain, heart, lung, and kidney bound to rSNAP29 and not to the control beads. This experiment indicates that SNAP29 binds to EHD proteins from various rat tissues, supporting the findings in yeast.
In Vivo Association of IGF-1R, SNAP29, and EHD Proteins-Immunoprecipitation of rat tissue extracts was performed using monoclonal antibodies directed against the ␣-chain of the IGF-1 receptor. The precipitates were resolved on SDS-PAGE and subjected to Western analysis using anti-EHD1 and anti-SNAP29 antibodies. As indicated in Fig. 3c, IGF-1 receptor formed complexes with EHD proteins derived from testes, brain, lung, and kidney. SNAP29 was found in complexes with IGF-1R in the same tissues as the EHD proteins. To confirm the interaction between SNAP29 and IGF-1R, extracts of NIH3T3/IGF-1R cells were subjected to immunoprecipitation using anti-IGF-1R ␣-chain antibodies, and the immunoblot was decorated either with anti-SNAP29 antibodies or with anti-IGF-1R ␤-chain antibodies. As depicted in Fig. 3d, IGF-1R formed complexes with SNAP29. To test whether SNAP29 interacts directly with IGF-1R, the yeast two-hybrid system was employed. Yeast expressing the pLexA/SNAP29 plasmid were transformed with either the pB42AD/IGF-1R ␤-chain, pB42AD/EHD1, pB42AD/SNAP29, or pB42AD/SNAP29/NTM constructs. As demonstrated in Fig. 3e, SNAP29 did not interact directly with IGF-1R, whereas it strongly interacted with itself to form homodimers. It weakly interacted with SNAP29 mutant missing amino acids 1-38, as demonstrated by growth on LeuϪ plates. These results indicate that SNAP29 does not bind to IGF-1R directly but most probably through a mediator (s).
Association of SNAP29 with ␣-Adaptin of AP-2 Complex-To test the possible association of SNAP29 with proteins of the endocytic machinery, immunoprecipitation of rat tissue extracts was performed using anti-SNAP29 antibodies. The precipitates were resolved on SDS-PAGE and subjected to Western analysis using anti-␣-adaptin of AP-2 antibodies. The results (Fig. 4) indicate that SNAP29 forms complexes with AP-2.
Intracellular Localization of EHD1 and SNAP29 -The twohybrid analysis and the co-immunoprecipitation results indicated interaction between EHD1 and SNAP29. Moreover, both proteins interacted with members of the endocytic machinery.
To confirm these results, intracellular localization was followed. Cells were transfected with plasmid expressing the EHD1-green fluorescent protein fusion protein. The cells were fixed, interacted with anti-SNAP29 antibodies, and decorated with rhodamine-conjugated secondary antibodies. The results demonstrated (Fig. 5) that both SNAP29 and EHD1 co-localized to punctuate vesicular structures throughout the cytoplasm. Since EHD1 was previously shown to localize to transferrin-containing vesicles, it is conceivable that the vesicular structures observed herein are endocytic vesicles. DISCUSSION Structural resemblance to Eps15, which was shown to participate in endocytosis, and intracellular localization to endocytic vesicles, have already marked EHD1 as a candidate participating in endocytic processes (24). Moreover, using immunofluorescence analysis, we co-localized AP-2 and EHD1 in endocytic vesicles. 2 Here we show that EHD1 interacts with proteins considered as hallmarks of clathrin-mediated endocytosis, ␣-adaptin of AP-2 and clathrin. Eps15 was shown to couple ␣-adaptin of AP-2 directly via DPF repeats at its C terminus. EHD1 does not harbor such repeats, and further investigation is required to determine whether it binds the protein directly. Affinity purification experiments have shown that EHD1 binds 63, 68, and 80 kDa proteins that are recognized by anti-EHD1 antibodies (Fig. 1a). The 63-kDa protein represents EHD1 and EHD3, whereas the 68-kDa protein represents EHD2 and EHD4 (25). Since these proteins share antigenic determinant with EHD1 and are able to bind EHD1 (Fig. 1a), and since all four EHD proteins are able to interact with each other and this interaction is basically mediated by the coiled-coil region, 3 we propose that the EHD family of proteins and proteins that share similar structural modules form oligomers. Similarly, Tebar et al. (27) demonstrate that the Eps15 forms homodimers and oligomers in vivo and that FIG. 4. SNAP29 forms complexes with AP-2. Extracts from rat tissues were subjected to immunoprecipitation using anti-SNAP29 antibodies. Immuno-complexes were resolved on SDS-PAGE and transferred into nitrocellulose, and a Western blot was employed using anti-␣-adaptin antibodies. this interaction is mediated via the coiled-coil region of the proteins. These authors proposed that oligomerization of EH domain-containing proteins may be a general feature important for the functional activity of this family of proteins.
Results of experiments presented here implicate EHD1 in the endocytic process of the IGF-1R. IGF-1R bound to the rEHD1 column. Only a small fraction of IGF-1 receptors expressed in the rat testis was bound to rEHD1, presumably due to prerequisites like phosphorylation for such binding (Fig. 1a). It has already been shown for the EGF receptor that phosphorylation within specific motifs in the cytoplasmic tail is required for extensive recruitment of accessory proteins, among them Eps15 (4). IGF-1R and EHD1 were found in in vitro formed complexes that were immunoprecipitated with anti-␣ -chain of IGF-1R antibodies (Fig. 1b).
Complexes containing IGF-1R and EHD proteins were immunoprecipitated with anti-IGF-1R antibodies (Fig. 3c). Results of immunofluorescence analyses indicated colocalization of EHD1 in IGF-1R-containing vesicles (Fig. 1c). We also have results indicating that overexpression of EHD1 abrogates IGF-1-induced signaling. It has been shown that the IGF-1-signaling cascade induces the MAP kinase (extracellular signal-regulated kinase) and Akt (29,31). Overexpression of EHD1 reduced 2.3 and 2.2 times the magnitude of IGF-1-induced signaling through the Ras-Raf pathway and the phosphatidylinositol 3-kinase pathway, respectively (Fig. 2), demonstrating that EHD1 is a down-modulator of these pathways. It is worth mentioning that recent publications indicated that endocytosis up-regulates signaling of activated receptors like the EGF receptor and the ␤ 2 -adrenergic receptor (32). We could also show that overexpression of EHD1 down-regulates IGF-1-induced signaling by co-transfection of an EHD1 expression vector with a reporter vector harboring the CD4 cDNA under a VEGF promoter into NIH3T3/IGF-1R cells. We found a 10-fold reduction in the amount of CD4 RNA in cells overexpressing EHD1 versus those that do not after IGF-1 induction (Fig. 2c).
Using the two-hybrid analysis, we identified SNAP29 as an interacting partner of EHD1 (Fig. 3, a-c). SNAP29 was initially discovered as an interacting partner of syntaxin 3 in a two-hybrid screen and was shown to be ubiquitously expressed in mammalian tissues (30). SNAP29 displays 39% similarity to neuronal SNAP25 and 35% similarity to the ubiquitously expressed SNAP23 (30). The SNAP proteins were described as participating in the core SNARE complex that underlines membrane fusion processes (33)(34)(35). In the nerve terminal, the vesicle and plasma membrane are brought into close apposition through the formation of a four-helices-bundle structure. Vesicle-associated membranous proteins synaptobrevin/VAMP (vesicle-associated membrane protein)-vesicle SNAP receptor (v-SNARE) contribute one helix, whereas the target membrane proteins (t-SNARE) SNAP25 and syntaxin contribute three helices to form a parallel bundle referred to as the SNARE complex (33)(34)(35). Unlike SNAP25 and SNAP23, SNAP29 does not have palmitoylation sites for anchoring a distinct membrane compartment, but rather, it is able to interact with different syntaxins and VAMPs that are specifically localized to distinct compartments; therefore, it was suggested as involved in multiple transport steps (30).
SNAP29 forms complexes with IGF-1R in vivo (Fig. 3, c and  d). Moreover, SNAP29 binds directly to EHD1 (Fig. 3, a and b). This interaction is strong and is carried out through the central coiled-coil region of EHD1 (Fig. 3a). It seems that the N terminus of SNAP29 participates in the binding to EHD1 (Fig. 3a). However, we cannot exclude the possibility that the truncated form of SNAP29 missing this region has lost the conformation required for the binding. This mutant was able to bind SNAP29 (Fig. 3e), although the binding was inefficient; therefore, we cannot exclude the possibility that this region is involved in binding to both EHD1 and SNAP29. The mechanism whereby EHD1 and SNAP29 are recruited to the endocytic complex of the activated IGF-1R is not yet understood. However, according to the yeast two-hybrid assay, SNAP29 does not interact directly with the cytoplasmic tail of IGF-1R ␤-chain, as demonstrated in Fig. 3e, suggesting that the interaction between SNAP29 and IGF-1R depends on accessory proteins. Indeed, SNAP29 also formed complexes with ␣-adaptin of AP-2 (Fig. 4). EHD1 and SNAP29 co-localized to endocytic vesicles (Fig. 5). All these findings indicate participation of EHD1 and SNAP29 in endocytosis of the IGF-1R. Future studies will provide detailed characterization of the involvement of SNAP29 and EHD1 along specific intracellular traffic stations of the endocytic process.
Another EH domain-containing protein termed intersectin was found in a yeast two-hybrid screen as an interacting partner of SNAP25, a hallmark of the vesicle SNAP receptor (SNARE) complex, playing a role in exocytosis at nerve terminals (36). Intersectin interacts with dynamin, a prominent accessory protein of endocytosis that participates in the pinching process of the coated pit and binds to Ras exchange factor mSos1, thus providing indication for linkage between endocytosis and signaling processes (37). In another work, Adams et al. (38) demonstrated that intersectin overexpression activated Elk-1 transcription factor in a MAPK-independent manner, and this ability resided within the EH domain. Moreover, intersectin expression was enough to induce oncogenic transformation of rodent fibroblasts. These data suggest that intersectin may link endocytosis with regulation pathways important for cell growth and differentiation. Different from EHD1, intersectins harbor several EH domains and three SH3 modules, enabling them, most probably, to participate in trafficking and signaling processes (36).
We propose that SNAP29 functions with EHD1, an EH domain-containing protein, to mediate IGF-1R endocytosis. Both these proteins were co-localized to endocytic vesicles, and both formed complexes with ␣-adaptin of AP-2 and IGF-1R. EHD1, which shares structural homology to Eps15, was found to associate with IGF-1R in response to IGF-1 stimuli and was shown to down-regulate IGF-1 signaling, suggesting it is involved in the sequestration of IGF-1R. SNAP29 and EHD1 were demonstrated to bind IGF-1R both in vitro and in vivo. The accumulated data indicate a role for these proteins in IGF-1R endocytosis.