Kinectin Anchors the Translation Elongation Factor-1δ to the Endoplasmic Reticulum

Kinectin has been proposed to be a membrane anchor for kinesin on intracellular organelles. A kinectin isoform that lacks a major portion of the kinesin-binding domain does not bind kinesin but interacts with another resident of the endoplasmic reticulum, the translation elongation factor-1 delta (EF-1δ). This was shown by yeast two-hybrid analysis and a number of in vitro and in vivo assays. EF-1δ provides the guanine nucleotide exchange activities on EF-1α during elongation step of protein synthesis. The minimal EF-1δ-binding domain on kinectin resides within a conserved region present in all the kinectin isoforms. Overexpression of the kinectin fragments in vivo disrupted the intracellular localization of EF-1δ proteins. This report provides evidence of an alternative kinectin function as the membrane anchor for EF-1δ on the endoplasmic reticulum and provides clues to the EF-1 complex assembly and anchorage on the endoplasmic reticulum.

Kinectin has been proposed to be a membrane anchor for kinesin on intracellular organelles. A kinectin isoform that lacks a major portion of the kinesin-binding domain does not bind kinesin but interacts with another resident of the endoplasmic reticulum, the translation elongation factor-1 delta (EF-1␦). This was shown by yeast two-hybrid analysis and a number of in vitro and in vivo assays. EF-1␦ provides the guanine nucleotide exchange activities on EF-1␣ during elongation step of protein synthesis. The minimal EF-1␦-binding domain on kinectin resides within a conserved region present in all the kinectin isoforms. Overexpression of the kinectin fragments in vivo disrupted the intracellular localization of EF-1␦ proteins. This report provides evidence of an alternative kinectin function as the membrane anchor for EF-1␦ on the endoplasmic reticulum and provides clues to the EF-1 complex assembly and anchorage on the endoplasmic reticulum.
Kinectin has initially been proposed to be a membrane anchor for kinesin on intracellular organelles (1,2). It is an integral membrane protein largely confined to the endoplasmic reticulum (ER) 1 (1,3). Kinectin consists of a 120-kDa polypeptide and a 160-kDa polypeptide interacting through an ␣-helical coiled-coil domain to form a heterodimer (4). The 120-kDa polypeptide is the truncated version of the 160-kDa polypeptide, lacking the first 232 amino acids, in the N terminus (4). The N terminus of the 160-kDa polypeptide consists of a transmembrane domain that anchors kinectin to organelle membranes, potentially with the help of the 7 myristylation sites throughout the molecule (3,4). The C terminus of kinectin consists of two functional domains. The kinesin-binding domain can interact with the cargo-binding site of the conventional kinesin and enhance the kinesin's microtubule-stimulated ATPase activity (5). A separate domain interacts with small G-proteins such as Rho A and Rac 1 (6,7) and plays a key role in mediating the microtubule-dependent Rho G activity (8). The role of kinectin in organelle motility is further supported by antibody inhibition and kinectin-domain overexpression studies, in which the organelle motility is adversely affected both in vitro and in vivo (2,5).
The role of kinectin as a universal membrane anchor for kinesin has been questioned when kinectin's restricted intracellular and phylogenetic distributions were discovered. Kinectin is not detected in axons of cultured neurons where kinesin is the major motor responsible for fast anterograde transport (9). Furthermore, the kinectin gene is not found in Caenorhabditis elegans or Drosophila genomes, where conserved conventional kinesin heavy chain gene is present (10). These findings suggest that additional or alternative membrane anchors for kinesin must exist for organelle motility. Recent yeast twohybrid and biochemical studies have identified a few new kinesin-interacting partners on different organelles. For examples, Sunday Driver interacts with the tricopeptide repeats of the kinesin light-chain subunit of kinesin-1 and mediates the axonal transport of post-Golgi vesicles (11). Kinesin-1 was proposed to link to a class of transport vesicles via the JIP-1 and JIP-2 scaffolding proteins that bind to members of the low density lipoprotein receptor family (12). Another scaffold protein, 14-3-3 proteins, may act as a membrane anchor for KIF1C (13). The transmembrane amyloid precursor protein is another potential membrane anchor for kinesin-1, which directly binds the tricopeptide repeats of the kinesin light chain (14). The AP-1 clathrin-associated adaptor complex, which mediates the transport of clathrin-coated vesicles from the trans-Golgi network to plasma membrane, binds KIF13A (15). There are studies revealing a potential interaction between KIF3 and fodrin (brain spectrin) on neuronal vesicles (16). The recently identified dendrite-specific kinesin KIF17 can also interact directly with the PDZ domain of mLin-10 (17). Therefore, a paradigm has emerged that motor proteins utilize different membrane anchors and the same motor protein, such as kinesin-1, can bind to different membrane anchors on different organelles (18). Since kinectin is primarily localized to the ER (1, 3), sparingly on the lysosome (8) and mitochondria 2 but not Golgi apparatus, its primary functions may be restricted to these organelle compartments.
Different kinectin isoforms with combinations of variable domains (vd) have been reported in human, mouse and fox genomes (19 -21). 2 There are at least five small (23-33 amino acid residues) variable domains scattered throughout the C terminus of kinectin. Two variable domains (vd3: amino acid residues 1177-1200 and vd4: amino acid residues 1229 -1256) overlap the kinesin-binding domain on kinectin (5). This implies that the kinectin isoforms lacking either vd3 or vd4 can-not serve as the membrane anchors for kinesin. Such isoforms without vd3 or vd4 have indeed been identified in cells (Fig. 1A) (20). 2 We have investigated the function of one such kinectin isoform lacking vd4 in ER.
We have isolated from a human fetal brain cDNA library a naturally occurring kinectin isoform that contains variable domains vd1, vd2, vd3, vd5, and vd6 but lacks vd4 (Fig. 1A). This kinectin isoform did not interact with the conventional kinesin heavy chain in a yeast two-hybrid analysis (Fig. 1A). However, it interacted with another resident of the ER, the translation elongation factor-1 delta, in a series of in vitro and in vivo experiments.
EF-1␦ is a subunit of the protein complex that participates in the elongation step during the translation of mRNA. The EF-1 complex consists of four subunits, namely, ␣, ␤, ␦, and ␥. The GTP binding protein, EF-1␣, is responsible for the transfer of aminoacyl-tRNA to ribosome with the hydrolysis of GTP. The EF-1␤ and EF-1␦ subunits are responsible for the nucleotide exchange activity, which converts the EF-1␣⅐GDP to the active EF-1␣⅐GTP (22)(23)(24). Both ␤ and ␦ sequences appear unrelated other than the C-terminal domain, which is responsible for the nucleotide exchange activity (25). The exact function of EF-1␥ is not yet known, although there are studies indicating that EF-1␥ can enhance the guanine nucleotide exchange activity of EF-1␤, and its association with tubulin suggests that EF-1␥ may help to anchor EF-1␤ to cell structures like microtubules and membranes (26,27).
We report here that kinectin interacts with EF-1␦ to anchor it to ER. Disruption of the kinectin-EF-1␦ interaction interferes with the EF-1␦ distribution in ER. The results provide us with a better understanding of the kinectin function in the ER.

EXPERIMENTAL PROCEDURES
The restriction enzymes used were from Promega (Madison, WI), and all of the other reagents were purchased from Sigma unless otherwise stated.
Construction of the Kinectin Bait D-A pair of oligonucleotides flanking 2539 -4143 base pair (bp) of the human kinectin gene (GenBank TM accession number Z22551) was designed to isolate different kinectin isoforms. A standard polymerase chain reaction (PCR) (28) was carried out using the human fetal brain, fetal liver, adult brain, and prostate libraries (Clontech). The amplified PCR products were subcloned into pTOPO vector (Invitrogen) by TA cloning. The different kinectin isoforms were grouped by restriction mapping, and their sequences were verified by sequencing. The region between 3142 and 4139 bp (amino acid residues 1024 -1356) of an isoform was subcloned in-frame to the GAL4 DNA-binding domain (BD) of the pAS2-1 vector (Clontech) and renamed as bait D.
Yeast Two-hybrid Screening-Yeast two-hybrid screening (Clontech) was performed according to the supplied protocol. The expression of baits A, B, and D fusion proteins in yeast were determined by immunoblotting the yeast extract with GAL4 DNA-BD monoclonal antibody (Clontech). 3 The kinectin bait D was used to screen a human fetal brain Matchmaker cDNA library fused to the GAL4 DNA-activation domain (AD) of the pACT2 vector (Clontech). The bait and the amplified library were sequentially transformed into yeast strain Y190 (Clontech) using the lithium acetate method (29). The transformants were assayed for their expression of histidine (HIS3) reporter gene by plating them onto the synthetic medium deficient in tryptophan, leucine, and histidine. 25 mM 3-amino-1,2,4-triazole was included to limit the number of false HIS3 positives (30). Colonies that grew successfully in the selective medium were further screened for their expression of the lacZ reporter gene by colony lift assay (31). The cDNAs in the AD vector of the persistent positive clones were isolated. False positives were further eliminated by testing whether the clones could activate both HIS3 and lacZ reporter genes with control baits (baits A, B, and BD vector alone).
The primary sequences of the positive clones were determined with an ABI PRISM Big Dye TM Terminator Cycle Sequencing Ready Kit using an ABI PRISM TM 377 DNA sequencer, according to the manu-facturer's instructions. The results were analyzed with the LASER-GENE (DNASTAR) and the BLAST 2.0 (National Library of Medicine) software.
In Vitro GST Pull-down Assay-The glutathione S-transferase (GST) fusion constructs with the kinectin baits A, B, and D were made by subcloning the corresponding cDNA into the expression vector pGEX4T-1 (Amersham Biosciences). A His 6 -tagged fusion construct with EF-1␦ was made by subcloning the EF-1␦ cDNA into pRSET expression vector (Invitrogen). The fusion proteins were expressed in Escherichia coli strain BL21/pLysS (DE3) (generous gift from Niovi Santama, Cyprus Institute of Neurology and Genetics, University of Cyprus). Protein expression was induced by the addition of isopropyl-␤-D-thiogalactoside. Intact bacterial cells containing the expressed fusion proteins were collected by centrifugation at 6000 ϫ g for 10 min. Protein extracts were obtained by freezing and thawing the cell pellet and resuspending the pellet in GST purification buffer (1ϫ phosphatebuffered saline (PBS), 50 mM Tris-HCl, pH 8, 0.5 mM MgCl 2 , 0.1% (v/v) Triton X-100) plus 5 mM dithiothreitol and a mixture of protease inhibitors (1 g/ml each of aproptinin, leupeptin, pepstatin A and 1 mM phenylmethylsulfonyl fluoride). Equal amounts of the GST fusion proteins with baits A, B, and D were immobilized onto glutathione-agarose beads by end-over-end mixing in a 1.5-ml microcentrifuge tube for 2 h at 4°C. The beads were washed three times with GST purification buffer to remove unbound proteins. The expressed His 6 -EF-1␦ (36 kDa) fusion protein extract was allowed to interact with the GST-bait fusion protein-coated beads for 2 h at 4°C with end-over-end mixing. Any unbound proteins were removed through extensive washings. The proteins that remained bound to the immobilized GST-baits were released by boiling in SDS gel sample buffer, analyzed by SDS-polyacrylamide gel (12%) electrophoresis (PAGE), and immunoblotting with mouse RGS-His antibody (Qiagen). Antibody binding was detected with goat anti-mouse secondary antibody, coupled to horseradish peroxidase according to the manufacturer's instructions (Sigma).
Co-immunoprecipitation of Kinectin with EF-1␦-African green monkey kidney cells, COS7, were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal calf serum (FCS), L-glutamine, penicillin, and streptomycin. For transient transfection assays, GST-tagged kinectin bait D (64 kDa) and FLAG-tagged EF-1␦ (36 kDa) were generated by subcloning the corresponding cDNA into the pXJ-GST and pXJ-FLAG mammalian expression vectors, respectively (32). Both constructs were co-transfected into COS7 cells cultured in a 60-mm dish using TransFast TM transfection reagent (Promega). The transfected cells were washed with ice-cold 1ϫ PBS at 48 h after transfection. The cells were isolated into 1 ml of the ice-cold GST purification buffer containing protease inhibitors (as described above). Cell lysate was prepared by passing the cells through 25-gauge needle for 10 times. Intact cells and debris were removed by centrifugation at 1000 ϫ g for 10 min at 4°C. 60 l of the equilibrated 50% glutathioneagarose bead suspension was added to the cell lysate and incubated for 2 h at 4°C. Immunoprecipitates were detected by immunoblotting with a mixture of anti-GST (Santa Cruz Biotechnology) and anti-FLAG M2 monoclonal antibodies. Antibody binding was detected as described above.
Anti-EF-1␦ Polyclonal Antibody Preparation-The EF-1␦ recombinant protein was expressed in E. coli BL21 using pRSET vector (Qiagen). The fusion protein was purified using the nickel-nitrilotriacetic acid agarose bead affinity column (Qiagen). The purified EF-1␦ protein was injected into rabbit by standard protocol (33). The antiserum was affinity-purified using the first 100 amino acids of the EF-1␦ protein coupled to cyanogen bromide-activated Sepharose 4B (34).
In Vitro Binding of the Endogenous EF-1␦ (32 kDa) with Kinectin Baits-Bacterially-expressed kinectin baits A, B, and D were coupled onto glutathione-agarose beads as described above. HeLa cells were cultured in DMEM supplemented with 10% FCS, L-glutamine, penicillin, and streptomycin. The cells were harvested and re-suspended with one volume of the GST purification buffer in the presence of protease inhibitors. The cell lysate was prepared as described above and was added to the beads coated with GST-bait fusion protein and incubated with end-over-end mixing for 2 h at 4°C. Immunoprecipitates were detected by immunoblotting with the purified anti-EF-1␦ polyclonal antibody.
In Vitro Binding of the Endogenous Kinectin with EF-1␦-Bacterially expressed EF-1␦ were coupled onto glutathione-agarose beads. The HeLa cell lysate was prepared as described above and was added to the beads coated with GST-EF-1␦ and incubated with end-over-end mixing for 2 h at 4°C. Immunoprecipitates were detected by immunoblotting with CT-1 monoclonal antibody (generous gift from Dr. Martin Krönke, University of Cologne, Germany). Antibody binding was detected as described above.
Real-time Biomolecular Interaction Analysis-The GST-tagged kinectin bait D (64 kDa) and His 6 -tagged EF-1␦ (52 kDa) recombinant proteins were expressed in E. coli BL21 using pGEX4T-1 and pET32 (Novagen) vectors, respectively. The GST-tagged and His 6 -tagged recombinant proteins were purified according to manufacturers' protocols (Amersham Biosciences & Novagen). The biomolecular interaction analysis was based on surface plasmon resonance (SPR) technology, which was performed with BIACORE 3000 (BIAcore AB). HBS buffer (10 mM HEPES-NaOH pH 7.4, 150 mM NaCl, 3.4 mM EDTA, 0.005% surfactant P20) was used as flow buffer, and the flow rate was kept at 5 l/min. 1 mg/ml of GST-only and GST-bait D proteins were coupled to CM5 sensor chip surfaces to yield a signal of 4000 response unit (RU) using standard amine coupling procedure (35). A control reference flow cell was included where no ligand bound. The surfaces were then inactivated with 1 M ethanolamine. 50 l of 1 M of the analyte was injected for 10 min and the binding kinectics was monitored for 16 min. After each interaction cycle, the surfaces were regenerated with 10 mM glycine, pH 2.5. The analytes were injected in triplicates in random order. All interaction experiments were performed at 25°C. All binding curves were corrected for refractive index changes and nonspecific binding by subtraction of the control reference flow cell. The association and dissociation rate constants were deduced from the initial signal courses of the association and dissociation phases, assuming 1:1 Langmuir binding and employing the BIAevaluation 3.0 software (BIAcore AB).
Mapping of the EF-1␦ Binding Domain on Kinectin Bait D-The EF-1␦ cDNA obtained from the yeast two-hybrid screening was subcloned into the BD vector. A series of truncated clones of the kinectin bait D (D1 to D18) were constructed (Fig. 5) by PCR. The amplified fragments were subcloned into the AD vector, and their interaction with the EF-1␦ was assayed using the yeast two-hybrid analysis as described above. A total of 50 yeast transformants were picked for each truncated clone to assay for the activation of HIS3 and lacZ reporter genes. Clones were considered as positive only when more than 90% of the transformants activated both HIS3 and lacZ reporter genes.
Immunofluorescence Assay-African green monkey kidney cells, CV1 cells, were seeded on coverglass overnight and maintained in DMEM supplemented with 10% FCS, L-glutamine, penicillin, and streptomycin. The cells were fixed with 3.7% paraformaldehyde in PBS for 10 min at 37°C and permeabilized with 0.01% saponin in PBS for 5 min at room temperature. The fixed cells were blocked with 10% FCS in PBS for an hour prior to incubation with the primary antibodies, CT-1 monoclonal antibody, and anti-EF-1␦ polyclonal antibody overnight. The cells were washed three times and followed by 1 h of incubation with the secondary antibodies, DIG-conjugated sheep anti-mouse IgG and Biotin-conjugated swine anti-rabbit IgG (Dako). The cells were incubated for another 1 h with the tertiary antibodies, FITC-conjugated sheep anti-DIG IgG and TRITC-conjugated Extravidin after the three washes with PBS. The coverglasses were washed thoroughly, mounted in FluorSave TM (Calbiochem) and imaged using a Carl Zeiss LSM510 laser-scanning confocal microscope.
Overexpression Studies-The EF-1␦ binding domains on kinectin, baits D and D12, were subcloned into the pEGFP-C vector (Clontech) for optimal expression in mammalian cells. The pEGFP-C vector alone was used as control. CV1 cells were cultured and transfected as described above. 48-h post-transfection, the cells were treated with 1 g/ml of nocodazole at 37°C for 2 h and followed by another 3 h in the absence of nocodazole. Cells were subsequently fixed and stained with anti-EF-1␦ polyclonal antibody prepared as described above. Transfected cells were identified by green fluorescence protein expression and scored either with distinct or disrupted EF-1␦ distribution. A total of 30 cells were observed for each clone during each experiment. The data tabulated were based on three independent experiments.

The Translation Elongation Factor-1 Delta Interacts with
Kinectin in Yeast Two-hybrid Screenings-To identify binding partners for kinectin isoforms lacking either vd3 or vd4 (kinesin-binding domain), we first isolated naturally occurring kinectin isoforms as baits. A PCR strategy was designed to amplify the region between 2539 and 4143 bp of the human kinectin gene using a pair of primers in the conserved sequences. The PCR reaction was carried out using human fetal brain, fetal liver, adult brain and prostate libraries. The amplified PCR bands were subcloned into pTOPO vector by TA cloning. A total of 7 unique naturally occurring kinectin isoforms were isolated and classified into two categories, namely either with or without vd4. One of the kinectin isoforms that contains vd1, 2, 3, 5, and 6 but lacks vd4 (Fig. 1A) was chosen for yeast two-hybrid screenings. The region between 3142 and 4139 bp (amino acid residues 1024 -1356), which contains vd2, vd3, and vd5, was subcloned into the BD vector and renamed as bait D. Bait D does not contain the predicted leucine zipper motifs (amino acid residues 934 -962) in the core of a predicted ␣-helical coiled-coil. Baits A, B, and D showed no interaction with the conventional kinesin heavy chain in yeast two-hybrid analysis (Fig. 1A). This is consistent with our previous findings that the vd3 and vd4 of kinectin are required for kinesin binding (5).
In order to identify interacting partners of the kinectin isoforms that lack the kinesin-binding domain, we performed a yeast two-hybrid screening of the human fetal brain cDNA library. A total of 17 unique positive-clones were identified from screening ϳ2.35 ϫ 10 7 transformants. Among them, one of the positive clones has sequence identical to the published human translation elongation factor-1 delta sequence (EF-1␦, GenBank TM accession number NM001960). The plasmid obtained has an insert of 1200 bp in length and encodes the entire open reading frame of the EF-1␦. EF-1␦ interacts specifically only with bait D, but not baits A or B in the yeast two-hybrid analysis (Fig. 1A). Baits A and B proteins were produced with the predicted molecular mass in yeast. 3 GST-Kinectin Bait D Specifically Interacts with EF-1␦ in Vitro-To verify that kinectin bait D binds to the full-length EF-1␦ in vitro, we first performed a GST pull-down analysis. GST fusion constructs with the kinectin baits A, B, and D were made and expressed in E. coli. Equal amounts of the individual fusion protein extracts were immobilized onto glutathione-agarose beads. The beads were washed extensively to remove unbound proteins such that only the kinectin baits remained bound to the beads. The His 6 -tagged EF-1␦ construct was expressed in E. coli, and the cell extract was allowed to interact with the immobilized GST-bait fusion proteins. The proteins remained bound to the immobilized GST-baits after extensive washings were analyzed by immunoblotting with anti-His antibody (Fig. 1B). Association of the EF-1␦ in vitro with the kinectin bait D, but not with baits A or B, is consistent with the interactions observed in the yeast two-hybrid analysis.
Kinectin Interacts with EF-1␦ in Mammalian Cells-To assess whether the interaction of EF-1␦ and kinectin occurs in mammalian cells, GST-tagged bait D and FLAG-tagged EF-1␦ were co-transfected into COS7 cells. Cell lysate was prepared 48 h after transfection and precipitated with glutathione-agarose beads. The protein complexes pulled down by the beads were analyzed by immunoblotting with anti-GST and anti-FLAG antibodies. The FLAG-tagged EF-1␦ indeed co-precipitated with the GST-bait D and not with GST-only (control) (Fig.  2), indicating that kinectin bait D interacts specifically with EF-1␦ in COS7 cells.
Interaction Analysis with Endogenous Proteins-To further confirm their interaction, we investigated whether the ectopically expressed kinectin would interact with the endogenous EF-1␦ in HeLa cells. Polyclonal anti-EF-1␦ antibody was produced using purified His 6 -tagged protein containing the first 100 amino acids of the human EF-1␦ as antigen. The affinitypurified antibody recognizes two polypeptide bands (ϳ32 kDa) on SDS-PAGE in HeLa cell lysate (Fig. 3A). The N terminus of the EF-1␦ protein has shown no significant homology with other proteins (36). The purified antibody is specific for EF-1␦ and has exhibited no cross-reactivity with EF-1␤ (25) (Fig. 3B).
The two bands that the purified antibody recognizes are the un-phosphorylated and phosphorylated forms of EF-1␦ (37, 38). When phosphorylated in vitro by cdc2 kinase, the un-phospho-rylated EF-1␦ undergoes a shift in electrophoretic mobility to the phosphorylated form (34) (Fig. 3C). The HeLa cell lysate was incubated with the immobilized GST-bait fusion proteins. The endogenous proteins that remained bound to the baits were analyzed by immunoblotting with the purified anti-EF-1␦ antibody. The association of the endogenous EF-1␦ with the kinectin bait D, but not with baits A or B, supports the observation of specific kinectin-EF-1␦ interaction (Fig. 3D) in vivo. Similarly, association of the endogenous kinectin with EF-1␦ was observed when the HeLa cell lysate was incubated with the immobilized GST-EF-1␦ proteins (Fig. 3E).
Real-time Biomolecular Interaction Analysis-To evaluate the kinectin-EF-1␦ interaction quantitatively, we used a biosensor instrument, BIAcore TM . BIAcore analysis is based on the surface plasmon resonance (SPR) measuring changes in refractive index upon binding of soluble analyte to an immobilized ligand (39). The purified recombinant proteins used for the analysis were shown in Fig. 4A. GST-only and GST-bait D were separately immobilized onto the sensor chip and 1 micromolar of either His 6 -EF-1␦ or His 6 -only recombinant proteins were injected over the sensor chip surface at a constant flow rate. The SPR signal is expressed as a sensorgram, in RU plotted as a function of time (40).
We found no association of both His 6 and His 6 -EF-1␦ with the immobilized control GST-only ligand (Fig. 4B); however, His 6 -EF-1␦ showed a gradual association (association phase) with the immobilized GST-bait D (Fig. 4C). When His 6 -EF-1␦ was replaced with flow buffer, dissociation of the EF-1␦ from GST-bait D was seen (dissociation phase). The association rate for the binding of EF-1␦ was determined to be 6.55 ϫ 10 3 M Ϫ1 s Ϫ1 and the dissociation rate to be 2.23 ϫ 10 4 s Ϫ1 . Kinetic analysis based on the sensorgram yielded the apparent dissociation constant of 34 nM. No significant binding was observed for His 6 to the GST-bait D. 46 -444), B (amino acid residues 444 -1049), and D (amino acid residues 1024 -1356) were constructed by cloning in-frame into a GAL4 DNA-BD vector. All the baits were assayed for interaction with conventional kinesin heavy chain and EF-1␦ using yeast two-hybrid analysis. (Ϫ) and (ϩ) indicate no interaction and positive interaction, respectively. B, equal amount of the EF-1␦ protein extracts was incubated with the beads immobilized with 3 g of GST-bait fusion proteins. After extensive washings, bound proteins were eluted and fractionated by SDS-PAGE and analyzed in an immunoblot with anti-His antibody. EF-1␦ specifically binds to bait D, but not baits A or B. TP, total protein; FT, flow-through; W, wash; E, eluate.

FIG. 2. Co-immunoprecipitation of kinectin bait D with EF-1␦.
Lysate of COS7 cells transfected with GST-tagged kinectin bait D and FLAG-tagged EF-1␦ were immunoprecipitated with glutathione-agarose beads. GST-only protein was used as negative control. After extensive washings, the total cell lysate preparation (A) and precipitates (B) were analyzed by SDS-PAGE and immunoblotted with both anti-GST and anti-FLAG monoclonal antibodies. EF-1␦ co-immunoprecipitates kinectin bait D but not with GST-only.
Characterization of the EF-1␦ Binding Domain on Kinectin-In order to identify the minimal sufficient domain on kinectin that interacts with EF-1␦, we have constructed a series of 18 truncated fragments of kinectin by both N-and C-terminal deletions (Fig. 5). Their interactions with EF-1␦ were examined with the yeast two-hybrid analysis. Among the 18 clones, 12 were positive as indicated by the activation of both HIS3 and lacZ reporter genes. The proteins synthesized from clone D1, D4, D5, D6, D7, D8, D9, D11, D12, D14, D15, and D16 could interact with the EF-1␦-bait, whereas D2, D3, D10, D13, D17, and D18 could not.
All the positive clones were confirmed experimentally for their interaction with EF-1␦ in the GST pull-down assay. The proteins synthesized from clone D1, D4, D5, D6, D7, D8, D9, D11, D12 but not D14, D15, and D16 could interact with EF-1␦. The minimal sufficient domain for EF-1␦-binding on kinectin is the clone D12 (amino acid residues 1116 -1175), which encodes a 60-amino acid fragment in the conserved region of the kinectin sequence (Fig. 6). The binding affinity of D14, D15, and D16 to the EF-1␦ could be weak or transient and therefore was only detected in the yeast two-hybrid analysis but not in the GST pull-down assay. Therefore, D12 was used in further experiments to confirm its functional significance.
Intracellular Localization of Kinectin and EF-1␦-For kinectin to interact with EF-1␦ in vivo, they need to encounter each other (i.e. co-localize in cells). We used confocal immunofluorescence to study the intracellular localization of kinectin and EF-1␦ proteins. Indirect double immunofluorescence staining was performed using monoclonal anti-kinectin antibody (41) and affinity-purified polyclonal anti-EF-1␦ antibody, on CV1 cells. Endogenous kinectin displayed an ER-like staining pattern, which overlapped significantly with that of the EF-1␦ (Fig. 7). Thus, the similar subcellular distributions of kinectin and EF-1␦ in the ER-like network suggest that the two proteins can interact with each other in vivo, with the possible function of kinectin anchoring EF-1␦ to the ER.
Kinectin Anchors EF-1␦ to the ER-To test the hypothesis that kinectin might anchor EF-1␦ to the ER, we examined the effect of the kinectin fragments (D and D12) overexpression on the EF-1␦ distribution in vivo. The kinectin baits D and D12 were cloned into a pEGFP-C vector and introduced into CV1 cells. 48 h post-transfection, the cells were treated with nocodazole for 2 h to perturb the microtubules and allowed the ER network (42) and EF-1␦ to retract toward the cell center. 3 The EF-1␦ distribution in the cell was imaged by confocal immunofluorescence microscopy with anti-EF-1␦ antibody. The control cells (untransfected cells or cells transfected with pEGFP-C vector alone) exhibited the typical ER-like staining pattern in CV1 cells (Fig. 8). 89% of the pEGFP-C vector transfected control cells and all the untransfected control cells exhibited the ER-like EF-1␦ network (Table I). However, only 26 and 30% of cells transfected with baits D and D12, respectively exhibited the ER-like EF-1␦ network. It is possible that the overexpressed kinectin fragments disrupted the ER, or EF-1␦ anchors to ER. Nevertheless, kinectin seems to be important for maintaining the distinct EF-1␦ distribution in ER-like network manner.
We further investigated if the overexpression of the kinectin fragments disrupted the ER. The cells, transfected with baits D and D12, were stained with an antibody against calnexin, which is a marker protein for ER. All the transfected cells showed similar staining pattern as the untransfected control glutathione-agarose beads and allowed to interact with the HeLa cell lysate. Endogenous kinectin was pulled down by EF-1␦. TP, total protein; FT, flow-through; W, wash; E, eluate.  2). B, GST-EF-1␦ and GST-EF-1␤ proteins were immunoblotted with anti-EF-1␦ and anti-GST antibodies. The anti-EF-1␦ antibody is specific for EF-1␦. C, HeLa cell lysate was in vitro phosphorylated by cdc2 kinase and immunoblotted with anti-EF-1␦ antibody. As a consequence of phosphorylation, the EF-1␦ protein undergoes a partial shift in electrophoretic mobility. D, 3 g each of the GST-kinectin baits A, B, or D was immobilized on glutathione-agarose beads. HeLa cell lysate was allowed to interact with the kinectin baits. After extensive washings, bound proteins were eluted and analyzed by SDS-PAGE and immunoblotting with affinity purified anti-EF-1␦ antibody. Endogenous EF-1␦ bound to the kinectin bait D, but not baits A or B. E, 3 g of GST-EF-1␦ was immobilized on  (Fig. 8C), indicating that the ER network remained intact when kinectin fragments were overexpressed. In another word, the disruption of the EF-1␦ network is likely due to the detachment of EF-1␦ from the intact ER network when the overexpressed kinectin fragments disrupted the kinectin-EF-1␦ interaction. Therefore, kinectin is important for anchoring EF-1␦ to ER.

DISCUSSION
Our previous work has implicated at least two categories of the kinectin isoforms. One category of kinectin contains both vd3 and vd4, which are essential for kinesin binding and important for organelle motility (5). 2 The other category does not contain either vd3 or vd4 with unknown functions. 2 In the present study, we have attempted to identify binding partners for kinectin at the region between 3142 and 4139 bp (amino acid residues 1024 -1356), which contains vd2, vd3, and vd5 but lacks vd4 (i.e. a kinectin isoform lacking the kinesin-binding domain). A search of a human fetal brain library, using the two-hybrid screening, has identified EF-1␦ as a candidate binding partner of kinectin. The results from subsequent GST pulldown, co-immunoprecipitation, and BIAcore real-time binding analysis, have confirmed that EF-1␦ is indeed a binding partner of kinectin. The observed interaction of kinectin bait D with EF-1␦ is unlikely to be due to nonspecific interaction, commonly seen between coiled-coil domains spanning leucine zipper motifs as the kinectin bait D does not contain such motifs. Our circular dichroism analysis of the kinectin baits A, B, and D indicated that all three baits exhibit the typical ␣-helical structures as previously predicted (3,4,41), suggesting that the differences in the ability of the kinectin baits to interact with EF-1␦ are unlikely due to gross misfolding of the baits, even though we cannot rule out the possibility of minor misfolding. 3 We have constructed a series of deletion mutants of kinectin to characterize the minimal sufficient domain for the kinectin-EF-1␦ interaction. The EF-1␦-binding domain on kinectin resides near the C terminus and is in close proximity to the kinesin-binding domain (5). The resolution of the fine mapping of the EF-1␦-binding domain on kinectin was restricted by the qualitative nature of the yeast two-hybrid assay. Despite Ͼ90% of the D14, D15, and D16 transfected yeast cells can activate both HIS3 and lacZ reporter genes, a larger fragment D12 was deduced to be the minimal binding domain. D12 is the minimal domain on kinectin that can interact with the EF-1␦ protein in both yeast two-hybrid and GST pull-down assay. The minimal EF-1␦-binding domain on kinectin is located within the conserved region, between vd2 and vd3, indicating that EF-1␦ can interact with all kinectin isoforms. Unlike EF-1␦, the conventional kinesin heavy chain can only interact with the kinectin isoforms containing both vd3 and vd4 (5). Our findings suggest that all kinectin isoforms are capable of interacting with EF-1␦ and only a subset of isoforms, which contains vd3 and vd4, is likely to play an additional role in organelle motility (5). In preliminary in vitro binding studies, we have not observed an interference of the kinesin-kinectin interaction by the addition of excess recombinant EF-1␦. 3 Further studies are needed to elucidate the relationship between the two adjacent binding domains on kinectin.
Both, transient expression of the full-length kinectin and immunolocalization of the endogenous kinectin in cells, display an ER-like staining pattern (3,41). Previous studies have also demonstrated that EF-1␦ clearly co-localizes with the proteindisulfide isomerase, which is one of the most abundant ERresident proteins (34). We have shown here that kinectin and EF-1␦ co-localize significantly in CV1 cells, and display an ER-like staining pattern. Such a co-localization of kinectin with EF-1␦ is consistent with the hypothesis that kinectin anchors EF-1␦ to the ER.
The hypothesis has been confirmed in the kinectin fragment overexpression study, where a disruption of the EF-1␦ network FIG. 5. EF-1␦ binding domain on kinectin. The kinectin bait D was truncated by both N-and C-terminal deletions. The interaction of the truncated clones with EF-1␦ was assayed by the yeast two-hybrid analysis. The solid line indicates the cDNA clones whose gene products activated HIS3 and lacZ reporter genes in Ͼ90% of the transformants, whereas the dashed line indicates the clones that activated the reporter genes in Ͻ10% of the transformants. was observed. The overexpression of the kinectin fragments containing the EF-1␦-binding domain but lacking the transmembrane N terminus competed with the EF-1␦ binding to the native kinectin on ER. ER network remained intact in this overexpression study suggesting that the observed disruption of the EF-1␦ network was likely due to the disruption of the EF-1␦ anchorage to the intact ER network via the kinectin residing on ER. The anchorage of EF-1␦ and EF-1 complex by kinectin on ER might be important for coupling protein synthesis and membrane translocation machineries on ER.
The only known kinectin-interacting protein in the EF-1 complexes identified from yeast two-hybrid screening is EF-1␦. In addition, no direct interaction of kinectin with EF-1␣, ␤, or ␥ subunit observed from one-to-one yeast two-hybrid analysis. 3 The exact mechanism of how kinectin may affect protein synthesis will need to be address in the future. The current findings indicate that kinectin seems to play an important role in supporting protein synthesis by anchoring the EF-1 complex via EF-1␣.
The precise mode of the interaction of the EF-1␤␥␦ complex with ER is unknown (34). Our current findings suggest two possible models on how EF-1 complexes are anchored to the ER membrane. The first model suggests that the anchorage of the EF-1␤␥␦ complex to ER is via kinectin. In Artemia, EF-1␦ was found to co-elute with the heterodimer EF-1␤␥ during purification (45). It has also been reported that EF-1␥ contains 2 internal repeats (VFGEXNXS) at amino acid residues 35-42 and 355-362. It is conceivable that the region of these two octapeptides of the EF-1␥ could be part of binding motif for EF-1␤ and EF-1␦, respectively, hence suggesting that EF-1␥ resides between the EF-1␤ and EF-1␦ in the complex (44,46). Since kinectin is an integral membrane protein on ER, it can interact with EF-1␦ and anchor the entire EF-1␤␥␦ complex onto ER. The C-terminal part of EF-1␤ interacts with EF-1␣ to support the guanine nucleotide exchange activity of the EF-1␣  A and B). Transfected cells were identified by GFP expression (green channel in B, C, and insert A). Two phenotypes were observed: (A) typical cells that overexpressed GFP-tagged kinectin baits D and D12 displayed reduced, disrupted EF-1␦ network; or (B) typical cells that overexpressed pEGFP-C vector alone or untransfected cells displayed distinct ER-like EF-1␦ network. The cells overexpressing GFP-tagged kinectin baits D & D12 were also stained with anti-calnexin monoclonal antibody for ER localization (C). The ER networks were intact in the cells overexpressing kinectin baits D and D12. Bar, 10 m. in protein synthesis (46). The restricted localization of kinectin to ER could be related to its primary role as the membrane anchor for the translation elongation factors. The concentration of the translation elongation factors to ER should allow efficient protein synthesis by preventing the intermittent diffusion of the translation elongation factors into the cytoplasm (47). An alternative model suggests that kinectin anchors the EF-1␦ onto specific regions of the ER membrane while the EF-1␤␥ complex interacts with other regions of the ER membrane in kinectin-and EF-1␦-independent manners. EF-1␥ was found to have specific affinity for membranes and tubulin (45,48). EF-1␤ was also found to be an actin-binding protein associating the protein synthesis apparatus with the cytoskeleton (49). This model can explain the presence of two different guanine-nucleotide exchange proteins in the EF-1 complex in cells. The C termini of both EF-1␤ and EF-1␦ contain domains that are responsible for the guanine nucleotide-exchange activities on EF-1␣ but their N termini exhibit no significant sequence homology (25,36). It is possible that the EF-1␤␥ complex and the EF-1␦-kinectin complex serve similar functions on different regions of the ER membrane. Even though the first model seems better supported by our and other reported findings, further investigations are needed to clarify the detailed mechanism of the anchorage of EF-1 complex to ER.
In conclusion, our findings reveal a new role of kinectin in cells in addition to its involvement in organelle motility. This report provides evidences suggesting that kinectin acts as a membrane anchor of EF-1␦ onto ER.