The Kinesin-like Motor Protein KIF1C Occurs in Intact Cells as a Dimer and Associates with Proteins of the 14-3-3 Family*

Proteins of the kinesin superfamily are regulated in their motor activity as well as in their ability to bind to their cargo by carboxyl-terminal associating proteins and phosphorylation. KIF1C, a recently identified member of the KIF1/Unc104 family, was shown to be involved in the retrograde vesicle transport from the Golgi-apparatus to the endoplasmic reticulum. In a yeast two-hybrid screen using the carboxyl-terminal 350 amino acids of KIF1C as a bait, we identified as binding proteins 14-3-3 β, γ, ε, and ζ. In addition, a clone encoding the carboxyl-terminal 290 amino acids of KIF1C was found, indicating a potential for KIF1C to dimerize. Subsequent transient overexpression experiments showed that KIF1C can dimerize efficiently. However, in untransfected cells, only a small portion of KIF1C was detected as a dimer. The association of 14-3-3 proteins with KIF1C could be confirmed in transient expression systems and in untransfected cells and was dependent on the phosphorylation of serine 1092 located in a consensus binding sequence for 14-3-3 ligands. Serine 1092 was a substrate for the protein kinase casein kinase IIin vitro, and inhibition of casein kinase II in cells diminished the association of KIF1C with 14-3-3γ. Our data thus suggest that KIF1C can form dimers and is associated with proteins of the 14-3-3 family.

Kinesin-like proteins (KLPs), 1 also known as kinesin family proteins (KIFs), form a superfamily of microtubule-based mechanochemical motors. All KLPs share a conserved motor domain of ϳ340 amino acids, and similarities between this domain have been used to group the known members of the kinesin superfamily into a number of subfamilies. Currently, 7 subfamilies and several outgroups comprise more than 250 KLPs from different organisms. The members of each subfamily share a common domain organization, exhibit sequence homology outside of the motor domain, and have similar motility properties and cellular functions (for recent reviews see Refs. 1 and 2). Nevertheless, the sequences outside of the conserved motor domain are responsible for the different cellular roles of the KLPs. KLPs are involved in cell division, in the formation and elongation of the mitotic spindle, the separation of the chromosomes, and in the transport of membranous vesicles along microtubules. The identification of distinct cargoes for distinct KLPs is emerging but proceeds more slowly for vesicle transporting KLPs than for the mitotic or meiotic KLPs (3). Identified cargoes include mitochondria as the cargo of KIF1B (4) and KHC (5), lysosomes of KHC (6) and KIF2␤ (7), and synaptic vesicle precursors of KIF1A (8). The membranebound organelles transported by the heterotrimer KIF3A-KIF3B-KAP3 (9 -11) as well as the cargo of KIF4 (12) or KIF2 (13,14) have not yet been conclusively characterized.
Little is known about proteins binding to KLPs, although these proteins most likely mediate the association with the cargo and may be involved in the regulation of the motor activity. This has been shown for the best understood motor protein, kinesin itself, that forms a heterotetramer consisting of the dimerized heavy chains, which contain the motor activity, and two kinesin light chains, which are believed to be involved in vesicle binding and the regulation of the motor activity of the kinesin protein complex (15). So far, five different mammalian kinesin light chains are known, and it is suggested that these different light chains are responsible for the different functions of kinesin within the cell (16). For the members of the KIF3 subfamily (17), the KAP3 proteins are described as binding partners and probably mediate the association of the KIF3 dimers with vesicles (9). Other proteins found to associate with KLPs are the serine phosphatase PP2A binding to Drosophila melanogaster KLP38B (18), the serine/threonine kinases MLK2/3 interacting with KIF3X (19), Rab6 binding to  and the tyrosine phosphatase PTPD1 associating with KIF1C (21). However, for none of these protein interactions a distinct function has been described yet.
Recently, we cloned and characterized KIF1C, a member of the KIF1/Unc104 subfamily (21). KIF1C is involved in the vesicle transport between the Golgi apparatus and the endoplasmic reticulum. At present, details on the vesicle binding mechanism and regulation of KIF1C or the other members of the KIF1/Unc104 subfamily are not known. Since the kinesin light chain proteins associate with the carboxyl terminus of kinesin and this seems to be the region that generally mediates association of kinesins with their cargo, we wanted to identify proteins that bind to the carboxyl terminus of KIF1C.
By using the carboxyl-terminal 350 amino acids of KIF1C as a bait in a yeast two-hybrid screen, we isolated different 14-3-3 isoforms from human skeletal muscle and brain cDNA libraries. The association of KIF1C with 14-3-3␥ was confirmed by coimmunoprecipitation in native human and murine fibroblasts and myoblasts and the 14-3-3␥-binding site identified. Furthermore, we have shown that in contrast to the related proteins KIF1A and KIF1B, the carboxyl terminus of KIF1C mediates the formation of KIF1C dimers.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screen-The carboxyl terminus of KIF1C (amino acids 755-1103) was cloned into the LexA fusion protein vector pBTM116 (kindly provided by S. Hollenberg and J. Cooper) and transformed into Saccharomyces cerevisiae strain L40 (MATa trp1 leu2 his3 LYS2::lexA-HIS3 URA3::lexA-lacZ) generating the L40 lexA-KIF1C CT strain. A human skeletal muscle cDNA library fused to the GAL4 activation domain in the pGAD10 vector and a human brain cDNA library cloned into the similar pACT2 vector (CLONTECH) were transformed into the L40 lexA-KIF1C CT strain, and 5 ϫ 10 6 transformants were screened for interaction as described by Hollenberg et al. (22). Yeast plasmid DNA was isolated from His ϩ ␤-galactosidase ϩ colonies, rescued into Escherichia coli HB101, retransformed into L40 lexA-KIF1C CT , and assayed for ␤-galactosidase activity and growth on dropout medium with complete supplement lacking Trp, Leu, Ura, Lys, and His (Bio 101). The specificity of the interaction between the carboxyl terminus of KIF1C and potential candidates was proven by transforming the candidate plasmid also into the L40 LexA and the L40 LexAlaminin strain.
Antisera-The KIF1C rabbit polyclonal antiserum was raised against a glutathione S-transferase (GST) fusion protein of KIF1C (amino acids 362-725) and has been described earlier (21). The antiserum against GST was produced by immunizing rabbits with bacterially expressed GST protein. The antisera against the different 14-3-3 isoforms were obtained from Santa Cruz Biotechnology.
Construction of Expression Plasmids-For transient expression in 293 cells, all cDNAs were cloned by standard procedures into the cytomegalovirus immediate early promotor-based expression plasmid pRK5 (23). The GST-KIF1C construct contains the GST of the pGEX1vector cut with the restriction endonucleases HincII and BamHI and cloned in frame to a sequence coding for the hemagglutinin peptide followed by the open reading frame of the KIF1C-cDNA. The aminoterminal deletion mutant of KIF1C, KIF1C-⌬NT 570 , contained the amino acids 570 -1103, and the carboxyl-terminal deletion mutants KIF1C-⌬CT 622 , KIF1C-⌬CT 802 , and KIF1C-⌬CT 1042 contained the amino acids 1-622, 1-802, and 1-1042, respectively. For generation of the KIF1C-SA mutant, the codon of serine 1092 was mutated to a codon for alanine using the method of Kunkel et al. (24).
Cell Lines and Cell Culture-NIH3T3, RD, and C 2 C 12 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Life Technologies, Inc.) and 2 mM glutamine. 293 cells were maintained in F-12/Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 2 mM glutamine. The inhibitor of protein kinase CKII 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (DRB, Biomol) was dissolved in dimethyl sulfoxide and used at a concentration of 30 M.
Transient Expression, Cell Lysis, and Immunoprecipitation-Transfection of 293 cells was performed using the method of Chen and Okayama (25). Treatment and lysis of the cells was performed as described (23). Briefly, the cells were lysed in 1 ml of lysis buffer/10-cm plate (1% Triton X-100, 50 mM HEPES, pH 7.5, 10% glycerol, 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM EGTA, 10 mM sodium pyrophosphate, 100 mM NaF, 1 mM sodium orthovanadate, 10 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride), and the lysates were precleared by centrifugation at 13,000 ϫ g for 15 min at 4°C. The lysates were adjusted for equal protein concentration; the appropriate antibody (2 l of serum or 0.2-2 g of antibody) and protein A-Sepharose (Amersham Pharmacia Biotech) were added, and the lysate was incubated for at least 3 h at 4°C. The immunoprecipitates were washed with HNTG buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 0.1% Triton X-100, 10% glycerol, 10 mM NaF, 1 mM sodium orthovanadate), separated on a 7.5, 8, or 10% SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and incubated with the appropriate antibody. Bound antibodies were visualized using the ECL system (Amersham Pharmacia Biotech) in conjunction with horseradish peroxidase-conjugated goat ␣-mouse or goat ␣-rabbit antibodies (Sigma).
In Vitro Binding Assays-For in vitro binding assays, 293 cells transiently expressing KIF1C were lysed and incubated for 4 h at 4°C with GST or the GST-14-3-3␥ fusion protein immobilized on glutathione-Sepharose. After washing with HNTG, the proteins were separated by SDS-PAGE and analyzed by Western blot. The GST-14-3-3␥ fusion protein was purified by standard procedures from E. coli transformed with a pGEX1-14-3-3␥ plasmid.
In Vitro Kinase Assay-Immunoprecipitated KIF1C and KIF1C-SA from transfected 293 cells were dephosphorylated by calf intestinal alkaline phosphatase (10 units, 45 min room temperature and 2 h at 4°C). After washing the precipitate with PHEM (50 mM HEPES, 50 mM PIPES, 1 mM EDTA, 1 mM MgCl 2 ) the beads were resuspended in PHEM and distributed into 6 aliquots. One aliquot was used to determine the KIF1C protein concentration by Western blotting and KIF1C immunodetection; 4 aliquots were incubated with 250 M Mg-ATP, 10 Ci of ␥-[ 32 P]ATP (Amersham Pharmacia Biotech), 1000 units of CKII (Biomol), and 1ϫ CKII buffer (20 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM MgCl 2 ) in a total volume of 20 l (26). An additional tube containing no CKII was included to serve as a control. Reactions were terminated by addition of Laemmli buffer and subsequent boiling. The samples were resolved by SDS-PAGE, and the dried gel was exposed to x-ray film. The extent of phosphate incorporation was calculated by densitometric analysis of the x-ray film using the software Scion Image, version 2b.
Covalent Cross-linking-Cells were lysed in a buffer containing 1% Triton X-100, 50 mM HEPES, pH 7.5, 10% glycerol, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, and a 30 -50-l aliquot of the lysate was treated for 2 h at 4°C with the cross-linking reagents. 1,5-Difluoro-2,4dinitrobenzene (DFDNB, Pierce) was dissolved in acetone and used at a concentration of 1 mM. Cross-linking with disuccinimidyl suberate (DSS, Pierce) was performed at identical conditions except that the reagent was dissolved in dimethyl sulfoxide and used at concentrations of 0,5 mM. The reactions were terminated by adding Tris buffer, pH 7.5, to a final concentration of 50 mM Tris.

RESULTS
Dimerization of KIF1C-To identify proteins that associate with the carboxyl terminus of KIF1C, we employed the yeast two-hybrid system. The carboxyl-terminal amino acids 755-1103 were fused in frame to the DNA-binding domain of LexA (pBTM116) and used to screen a human skeletal muscle cDNA library subcloned into the GAL4 transcription activation domain expression vector pGAD10 and a human brain cDNA library subcloned into the similar pACT2 vector (CLONTECH). Approximately 5 ϫ 10 6 transformants were screened for expression of the reporter genes HIS3 and LacZ, and five different types of clones were found to interact specifically with KIF1C. Four types of clones represented members of the 14-3-3 family of proteins (see below), and one type encoded a peptide consisting of the carboxyl-terminal amino acids 814 -1103 of KIF1C. This result was rather unexpected since the KLPs of the KIF1/Unc104 subfamily have been described as monomers (4,8).
To investigate the dimerization of KIF1C in intact cells, we first applied coimmunoprecipitation analysis. We constructed a fusion protein between GST and KIF1C that differs in size from wild-type KIF1C by 26 kDa and thus allows a separation of GST-KIF1C and KIF1C by SDS-PAGE. Coexpression of KIF1C and GST-KIF1C in 293 cells resulted in the formation of a dimer between these two proteins. As shown in Fig. 1, immunoprecipitation using the ␣GST antibody, which did not react with KIF1C, resulted in a coprecipitation of KIF1C with GST-KIF1C. Since both proteins were detected by Western blot analysis using the ␣KIF1C antibody, the amount of coprecipitated KIF1C almost equaled the amount of the immunoprecipitated GST-KIF1C. These data demonstrate that under the conditions of transient overexpression, the full-length KIF1C proteins do form dimers with GST-KIF1C.
Since a dimerization of endogenous KIF1C cannot be proven by coimmunoprecipitation, we tested two cross-linking reagents for their ability to react with KIF1C. We applied DSS and DFDNB, which differ in their reactivity against certain amino acids and their spanning distance. DSS is an amineselective cross-linking reagent, which can span approximately 1.13 nm. The aryl halide DFDNB spans only over 0.3 nm but in addition to primary amines also reacts with thiol and phenolate groups as they occur in cysteine and tyrosine, respectively. Since our coimmunoprecipitation experiments clearly showed a dimerization of overexpressed KIF1C, we first applied the two reagents to lysates of 293 cells overexpressing KIF1C. As depicted in the right panel of Fig. 2, DFDNB generated an additional KIF1C-containing protein complex of approximately 270 kDa, which is twice the size of monomeric KIF1C. In these cells even more KIF1C protein was detected in the 270-kDa complex as was found as a monomer. Since under the conditions of transient overexpression it is unlikely that an endogenous protein of the cells with approximately the same size as KIF1C is expressed at these high levels, we conclude that the 270-kDa complex represents the KIF1C dimer. Interestingly, the DSS cross-linking reagent, although it can span more than three times the distance of DFDNB and is therefore considered to be less selective, reacted only weakly with KIF1C. This finding can only be explained by the structure of the KIF1C dimer in which not lysines but tyrosines and cysteines get in close contact to each other.
We next applied the DSS and DFDNB reagents to lysates of untransfected 293 cells expressing only endogenous KIF1C. As shown in the left panel of Fig. 2, the lysate incubated with DFDNB contained a KIF1C protein complex of the same size as the KIF1C-overexpressing cells. DSS treatment did not result in any detectable complex formation of KIF1C supporting the idea that the 270-kDa protein complex consists of the same proteins as the one observed in the 293 cells transiently overexpressing KIF1C. In contrast to 293 cells overexpressing KIF1C, where most of the KIF1C is in the dimeric form, under physiological conditions only a small part of the KIF1C protein was detected as a dimer. We thus suggest that the dimer formation of KIF1C under physiological conditions is regulated, whereas the overexpressed proteins dimerize spontaneously.
Dimerization between KLPs occurs by the formation of a coiled-coil structure that is formed by the carboxyl-terminal stalk region of the proteins (1). The probability of a sequence to form coiled coils can be predicted by the algorithm of Lupas (27). We analyzed the sequence of KIF1C, its closest relative KIF1B, and the dimeric KHC for the probability to form coiled-coils (Fig. 3). KIF1C and KIF1B have only small areas that are predicted to form coiled-coils when compared with KHC. In KIF1C, four 40 -60 amino acids containing regions around the amino acids 370, 450, 640, and 850 are highlighted by the algorithm. These short stretches are also found in KIF1B. In the yeast two-hybrid system the dimerizing KIF1C polypeptides included only the last of the 4 predicted coiled-coil elements. Therefore, even the small stretch of coiled coil formation seems to be sufficient for dimerization of KIF1C in yeast and may also mediate the dimer formation of KIF1C in mammalian cells.
Association of 14-3-3␥ with the Carboxyl Terminus of KIF1C-The additional clones that were found interacting with the carboxyl-terminal part of KIF1C coded for the proteins ␤, ␥, ⑀, and of the 14-3-3 family. In our studies we concentrated on the 14-3-3␥ isoform. To confirm the interaction of KIF1C with this 14-3-3 family protein, in vitro association studies were performed. The cDNA of 14-3-3␥ was cloned into the pGEX1 vector, and the GST-14-3-3␥ fusion protein was bacterially expressed and purified. We immobilized 0.5 g of GST-14-3-3␥ or as a control 0.5 g of GST protein on glutathione-Sepharose and incubated it with 200 l of lysate from 293 cells overexpressing KIF1C. As shown in Fig. 4A, KIF1C specifically associated with the GST-14-3-3␥ fusion protein but not with GST.
These data suggested a high affinity of 14-3-3␥ for KIF1C, and we therefore investigated the association between the two proteins in untransfected fibroblasts and skeletal muscle cells. After lysis of human 293 or mouse NIH3T3 fibroblasts, we immunoprecipitated 14-3-3␥, separated the proteins by SDS-PAGE, and detected KIF1C and 14-3-3␥ in the precipitates by Western blot analysis (Fig. 4C). This association was also detected in the skeletal muscle C 2 C 12 and rhabdomyosarcoma RD cell line (data not shown). To investigate whether the 14-3-3␤, -⑀, and -isoforms also interact with KIF1C in intact cells, we constructed cDNAs that were expressed with a carboxyl-terminal Myc tag. Co-overexpression of KIF1C and these 14-3-3 proteins revealed that all isoforms associated with similar affinities with KIF1C (data not shown). These data thus clearly demonstrate an association between the motor protein KIF1C and the 14-3-3␥ protein, and it is likely that the ␤, ⑀, and isoforms act similarly.
Localization of the 14-3-3␥ Binding Site in KIF1C-By using the amino acids 755-1103 as a bait in the yeast two-hybrid screen, we limited the 14-3-3␥ interaction site of KIF1C to the carboxyl-terminal part of the protein. To locate the binding site more accurately and to exclude additional binding sites in KIF1C we coexpressed deletion mutants of KIF1C with 14-3-3␥ in 293 cells. The deletion mutants of KIF1C were either the amino-terminal truncated protein, KIF1C-⌬NT 570 , or the different carboxyl-terminal truncated proteins KIF1C-⌬CT 622 , KIF1C-⌬CT 802 , and KIF1C-⌬CT 1042 . After lysis and immunoprecipitation with the ␣14-3-3␥ and ␣KIF1C antibodies, we analyzed the immunocomplexes for coprecipitated KIF1C and 14-3-3␥, respectively. In KIF1C immunoprecipitates 14-3-3␥ was only detected by Western blot analysis in cells that were expressing either the wild-type or the amino-terminal truncated KIF1C protein. Consequently, if 14-3-3␥ was immunoprecipitated from these cell lysates, only wild-type KIF1C and KIF1C-⌬NT 570 coprecipitated (Fig. 5A). From these data we confined the binding site of 14-3-3␥ to KIF1C to the last 60 amino acids of the protein.
Recently, the binding of 14-3-3 proteins to their ligands has been described to be dependent on a phosphorylated serine residue in the amino acid context of RRXRpS(p)XP (where X indicates any amino acid; pS indicates phosphoserine (28)). This sequence motif occurs also in the very carboxyl terminus of KIF1C and is located around the serine residue Ser 1092 (RRQRSAP). To test if this serine indeed is involved in 14-3-3 binding, we mutated Ser 1092 to alanine so that the association of KIF1C with 14[ hyphen]3-3 proteins should be abolished. Immunoprecipitation of 14-3-3␥ from lysates of 293 cells transiently expressing the KIF1C-SA mutant did not result in a coprecipitation of the KIF1C-SA (Fig. 5B) although the protein was expressed (data not shown). This indicates that Ser 1092 indeed is mediating the association between KIF1C and 14-3-3␥.
Phosphorylation of KIF1C by CKII Is Required for the Association with 14-3-3␥-Since the binding of 14-3-3 proteins normally depends on a phosphorylation of the serine residue, we tried to identify the responsible kinase. Analysis of the amino acid sequence of KIF1C identified Ser 1092 as a potential substrate sequence for the protein kinase CKII. CKII can be inhibited by 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (DRB), which was shown to be specific for CKII and does not inhibit other serine/threonine kinases like protein kinase A (29,30). If CKII indeed is involved in the phosphorylation of Ser 1092 , an incubation of cells with DRB should reduce its phosphorylation and thus diminish the association of KIF1C with 14-3-3␥. To test this hypothesis, 293 cells were either left untreated or incubated with 30 M DRB for 18 h, lysed, and the 14-3-3␥ protein immunoprecipitated. The immunocomplexes were separated by SDS-PAGE and analyzed for the presence of KIF1C. As shown in Fig. 6, the amount of KIF1C coprecipitating with 14-3-3␥ was strongly reduced in cells treated with the CKII FIG. 4. Association of 14-3-3␥ with KIF1C. A, lysates from 293 cells transiently overexpressing KIF1C were incubated with either 0.5 g of GST-14-3-3␥ or GST immobilized on glutathione-Sepharose. After washing the complex, the proteins were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted using the affinity purified ␣KIF1C antibody. Since the ␣KIF1C antibody was raised against a GST-KIF1C fusion protein, it also reacts with the GST-14-3-3␥ and GST proteins. The positions of KIF1C, GST-14-3-3␥, and GST are indicated by an arrow. B, immunoprecipitates (IP) of KIF1C or 14-3-3␥ from lysates of 293 cells expressing KIF1C and 14-3-3␥ were separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted (IB) with ␣KIF1C (upper panel) and ␣14-3-3␥ antibody (lower panel). C, NIH3T3 and 293 fibroblasts were lysed and the lysates divided into two aliquots. From 1 aliquot, 14-3-3␥ was immunoprecipitated with the ␣14-3-3␥ antibody (␥). The other aliquot was used as a control and immunoprecipitated with non-immune serum (NS). The immunocomplexes were analyzed as described in B. The molecular mass markers are shown in kDa.
inhibitor DRB. The reduction was dependent on the incubation time of the cells in the presence of DRB. In cells incubated for more than 16 h a significant reduction of KIF1C bound to 14-3-3␥ was detected (data not shown), indicating that phosphorylation of Ser 1092 is rather stable. These data show an involvement of CKII in the regulation of the association of KIF1C with 14-3-3␥ and provide additional evidence for the specificity of the association of 14-3-3␥ with KIF1C.
To investigate if KIF1C is indeed a direct substrate of CKII or if the kinase CKII is regulating another kinase, which in turn then phosphorylates KIF1C, we performed an in vitro kinase assay with CKII. KIF1C and KIF1C-SA were expressed in 293 cells and purified by immunoprecipitation. To remove any phosphorylation that had occurred in 293 cells, the immunoprecipitates were incubated with alkaline phosphatase before starting the in vitro kinase assay. Then the immunoprecipitates were incubated with CKII in the presence of 25 Ci ␥-[ 32 P]ATP for different times as described under "Experimental Procedures." To ensure that the observed phosphorylation represents only CKII activity and does not result from a kinase coprecipitated with KIF1C, 1 aliquot was incubated under the same conditions without adding the CKII enzyme. In Fig. 7A the phosphorylation of KIF1C by CKII is shown. It indicates that KIF1C is readily phosphorylated by purified CKII, whereas KIF1C-SA is phosphorylated to a lower extent. The densitometric analysis of the amount of incorporated phosphate into KIF1C and KIF1C-SA is shown in Fig. 7B. According to these data, in addition to Ser 1092 CKII phosphorylates KIF1C at one more amino acid residue. DISCUSSION Since the purification of kinesin in 1985 (31) more than 250 different KLPs have been cloned from different organisms. They all contain a motor domain responsible for the binding of microtubules and the hydrolyzation of ATP for the generation of force, allowing these proteins to move along the microtubules. Although some progress has been made in the understanding of the mechanochemical basis of this movement, the regulation of motor activity and vesicle binding remains to be elucidated. In this report we describe the association of KIF1C with 14-3-3 proteins and also provide evidence for a dimerization of KIF1C in intact cells.
Dimerization of KIF1C-Whereas all characterized KLPs do form either homo-or heterodimers, the members of the KIF1/ Unc104 subfamily are believed to exist as monomers only. Recombinant KIF1A and KIF1B were purified as monomers, by applying a sucrose gradient centrifugation, gel chromatography, and rotary-shadowing electron microscopy (4,8). Under these conditions other KLPs like kinesin or KIF3 were found as dimeric complexes. The proteins of the KIF1/Unc104 subfamily were thus believed to represent the first class of monomeric microtubule-based motor proteins that must obtain processivity by a different mechanism than kinesin.
By using the carboxyl terminus of KIF1C as a bait in the yeast two-hybrid system, we have found that KIF1C can dimerize. We confirmed this result in fibroblasts with overexpressed as well as endogenous KIF1C in coimmunoprecipitation and chemical cross-linking experiments. Analysis of the sequence of KIF1C revealed a probability to form coiled-coil structures only for short stretches that was similar to that of KIF1B (Fig. 3), which has been predicted not to dimerize (4). However, O'Shea et al. (32) described a dimerization of the GCN4 protein mediated by coiled-coils formed by stretches of only 28 -35 amino acids. Thus, it remains possible that KIF1C dimerizes like the other KLPs by formation of coiled-coils, but compared with kinesin the dimerization should be rather weak.
Association of KIF1C with 14-3-3 Proteins-In this study we identified the ␤, ␥, ⑀, and members of the 14-3-3 protein family as binding partners of KIF1C. In a transient expression system the four isoforms coimmunoprecipitated with similar affinities with KIF1C. The isoform investigated in more detail, 14-3-3␥, associated with KIF1C also in untransfected cells. Despite an extensive analysis of the clones obtained in the yeast twohybrid screen, no interaction occurred in this system between KIF1C and 14-3-3, although this isoform is expressed in the brain and therefore should be represented in the library used (33).
The binding sequence of 14-3-3␥ in KIF1C was identified as the motif including the phosphorylated serine residue 1092. Mutation of this residue abolished the association of overexpressed KIF1C-SA with 14-3-3␥ in 293 cells. This indicates that the sequence RRQRS(p)AP, which is very similar to the peptide sequences reported by Yaffe et al. (28), efficiently binds 14-3-3 proteins. Ser 1092 is located in a consensus sequence for phosphorylation by the protein kinase CKII and was indeed a substrate of CKII in vitro. In addition to Ser 1092 , at least one more serine/threonine residue of KIF1C was phosphorylated by CKII under these conditions. The inhibition of CKII in intact cells by the specific inhibitor DRB resulted in a diminished association of KIF1C with 14-3-3␥. Although we cannot exclude that the effect of CKII is only indirect, the in vitro phosphorylation data provide strong evidence for a direct interaction between KIF1C and CKII. Like KIF1C and 14-3-3 proteins, the protein kinase CKII is a ubiquitous serine/threonine kinase with established roles in cell proliferation and signal transduction. Although the majority of CKII has been found in the nucleus, CKII activity was also detected at the Golgi apparatus and the endoplasmic reticulum (34,35). Identified substrates of CKII include transcription factors and enzymes involved in transcription and translation, receptor tyrosine kinases like the insulin receptor, and cytoskeletal and structural proteins like ␤-tubulin or clathrin (for review see Ref. 36). Recently, Karki et al. (26) identified cytoplasmic dynein in a complex with CKII and showed that CKII was also able to phosphorylate the dynein intermediate chain in vitro. Although the function of this CKII-derived phosphorylation of dynein is not known yet, serine/threonine phosphorylation was described to regulate the cargo association of KLPs and dynein superfamily motor proteins. For example, kinesin showed a reduced association with synaptic vesicles when phosphorylated by protein kinase A. However, the mechanism and the protein interactions affected by this phosphorylation remain to be elucidated (37). The 14-3-3 protein family consists of highly conserved members and contains several isoforms with overlapping functions. Like KIF1C, the ␤, ␥, and isoforms have been found at the Golgi apparatus (38,39). Since the structure of the 14-3-3 dimer allows simultaneous binding to two proteins (40,41), it was suggested that 14-3-3 proteins provide a scaffold on which other proteins interact. We identified KIF1C in intact cells as a dimer that associates with the also dimeric 14-3-3 proteins. It is therefore tempting to imagine a KIF1C-14-3-3 heterotetramer in analogy to the KHC⅐KLC complex. Unfortunately, it was not possible to chemically cross-link 14-3-3 proteins with KIF1C, and thus this hypothesis remains only speculative. In contrast to the kinesin complex, which is formed constitutively, the generation of a KIF1C-14-3-3 complex is dependent on the serine phosphorylation of KIF1C. In mammals, five different KLC isoforms are known so far that have homologous aminoterminal sequences for binding to KHC but vary in their carboxyl-terminal sequences that are supposed to be involved in vesicle binding (16). Thus, the attachment of KHC to different vesicles and organelles is probably regulated by the corresponding KLC. Four different isoforms of the 14-3-3 family proteins, which like the KLC also differ in their properties, have been shown to interact with KIF1C. It remains to be elucidated whether they differentially regulate the activity of KIF1C. The ␥ and isoforms, which in contrast to the ␤ and ⑀ isoform bind phospholipids (42), could act as cofactors for the association of KIF1C with the membranes of the Golgi apparatus or the transported vesicles. In earlier studies 14-3-3 proteins have been found at the membrane of synaptic vesicles (43) where they stimulate calcium-dependent exocytosis (44,45). In addition, the yeast homologues of 14-3-3, BMH1, and BMH2 are also involved in vesicular trafficking as well as in the Ras signaling pathways (46).
Recent data support the idea of a cross-talk between vesicle transport controlled by motor proteins and the stress-induced signaling pathway leading to an activation of the c-Jun aminoterminal kinase and the p38 kinases. Nagata et al. (19) de- scribed a complex of the mixed lineage kinases MLK2 and MLK3 with the motor protein KIF3X and 14-3-3 proteins. The ⑀ and isoform of the 14-3-3 proteins were described to interact with the MEK1-3 kinases upstream of the c-Jun amino-terminal kinase and ERK1/2 kinases (38). Since MEKK2 is localized at the Golgi apparatus (38), future experiments should explore a connection between the regulation of KIF1C involved in the retrograde Golgi-to-endoplasmic reticulum membrane transport and the cellular response to extracellular stimuli by mitogen-activated kinase/extracellular signal-regulated kinase kinases.
In conclusion, we have demonstrated a dual role for the carboxyl-terminal 350 amino acids in KIF1C as follows: a motif around Ser 1092 interacts in a regulated manner with members of the 14-3-3 family, and most likely a stretch of 50 amino acids around position 850 forms coiled-coil structures and is likely to mediate dimerization with another KIF1C protein. Future experiments will clarify the physiological role of these interactions.