Cayman ataxia protein caytaxin is transported by kinesin along neurites through binding to kinesin light chains

Cayman-type cerebellar ataxia is a rare autosomal recessive disease. Patients show marked psychomotor retardation and prominent none-progressive cerebellar dysfunction manifested by nystagmus, intention tremor, dysarthric speech and a wide-based ataxic gait (Brown et al., 1984; Jonson et al., 1978; Nystuen et al., 1996). Recently, the gene responsible for this disease has been identified and named ATCAY, which encodes a protein termed caytaxin (Bomar et al., 2003). Three different mutant alleles, jittery, hesitant and sidewinder, of the homologous mouse Atcay gene cause similar neurological abnormalities. Hesitant mice show mild ataxia and dystonia whereas jittery and sidewinder mice have severe symptoms and die within 4 weeks of birth (Gilbert et al., 2004; Kapfhamer et al., 1996). In rats, deficiency of caytaxin results in generalized dystonia (dt) (Xiao and Ledoux, 2005). Cerebellectomy rescues the severe motor symptoms and prolongs the lifespan of dt rats (LeDoux et al., 1993). Electrophysiological and biochemical studies of dt rats define the olivocerebellar pathway, particularly the climbing fiber projection to Purkinje cells, as main sites of dysfunction. The cerebellar cortex of the dt rat shows abnormal expression of genes involved in phosphatidylinositol signaling pathways, calcium homeostasis and extracellular matrix interaction (Xiao et al., 2007). Caytaxin is therefore crucial for cerebellar function. In addition, the range of symptoms seen in human patients, including psychomotor retardation, suggests that caytaxin is also essential for cerebral functions.

Caytaxin protein shows the highest homology to BNIP-2, which was originally isolated as an interacting protein for the anti-apoptotic Bcl-2 protein and the adenovirus E1B 19 kDa protein (Boyd et al., 1994). C-terminal regions of caytaxin and BNIP-2 contain a conserved domain known as a SEC14 domain in yeast and a CRAL-TRIO domain in mammals. The CRAL-TRIO domains in general bind lipophilic small molecules such as retinal, vitamin E, squalene and phosphatidylinositol (Panagabko et al., 2003). However, the CRAL-TRIO domain of BNIP-2, also known as the BCH domain, has been shown to interact with Cdc42 and p50RhoGAP (Cdc42GAP) (Zhou et al., 2005). Overexpressed BNIP-2 induces cell elongation, membrane protrusion and subsequent apoptosis of a variety of non-neuronal cultured cells. This activity depends on the interaction with Cdc42. Although it is not known whether the CRAL-TRIO domain of caytaxin binds small lipophilic molecules, the caytaxin CRAL-TRIO domain has been reported to associate with a neurotransmitter-producing enzyme, kidney-type glutaminase (KGA) and the peptidyl-prolyl isomerase Pin1 in neurons (Buschdorf et al., 2006; Buschdorf et al., 2008). The physiological and etiological significance of this binding remains to be elucidated. To understand the function of caytaxin, we conducted yeast two-hybrid screening and identified kinesin light chains (KLCs) as binding partners for the N-terminal region of caytaxin. KLC is a component of kinesin-1, which is a tetrameric motor protein composed of two KLCs and two kinesin heavy chains (KHCs). Kinesin-1 is involved in axonal and dendritic transport of mitochondria; vesicles containing amyloid-β precursor protein (APP) or apolipoprotein E receptor 2 (ApoER2); the c-Jun N-terminal kinase interacting proteins (JIPs); other protein complexes; and mRNA granules (Hirokawa and Takemura, 2005; Salinas et al., 2008). Here we show that caytaxin is transported by kinesin in neurites. Defects in axonal and dendritic transport could cause ataxia and dystonia in caytaxin-deficient animals.

Yeast two-hybrid screening using full-length caytaxin as bait isolated KLC1 most frequently from a mouse adult brain cDNA library (20 clones out of three million clones screened). Two clones of KLC2 and a single clone of KLC3 were also isolated. Other proteins isolated several times were RNA polymerase II polypeptide G (four clones) and Bcl-2-associated transcription factor (two clones). Considering the nuclear localization of these two proteins, it seemed implausible that they were interacting partners for caytaxin, which is a cytoplasmic protein (Buschdorf et al., 2006; Hayakawa et al., 2007). Another 17 genes were isolated only once. Quantitative two-hybrid β-galactosidase assay using the caytaxin N-terminal fragment (amino acids 1-170) [caytaxin(1-170)] as bait showed that the most isolated clones including KLCs bound to caytaxin(1-170). One exception was inositol-1,4,5-trisphosphate 3-kinase A (IP3KA), which did not interact with the N-terminal caytaxin fragment (data not shown) and seemed to bind to the C-terminal CRAL-TRIO domain of caytaxin.

The KLC1 cDNA clones isolated in this screen ranged in size from those encoding full-length protein to those with various truncations in the N-terminal heptad-repeat region, which recognizes the KHC molecule (Hirokawa and Takemura, 2005). All the clones retained the intact C-terminal tetratricopeptide (TPR) motifs and a KLC2 clone contained only the TPR motifs. To confirm the binding of caytaxin to the TPR region, the full-length KLC1, the KLC1 N-terminal 1-212 amino acid fragment and the C-terminal 177-542 amino acid fragment were produced as GST-fusion proteins in E. coli and analyzed by GST pull-down assay for binding to full-length caytaxin produced in HEK293T cells (Fig. 1A). Full-length KLC1 and the C-terminal TPR motif fragment bound to caytaxin but the N-terminal heptad repeat fragment did not. Therefore, caytaxin binds to the TPR region of KLC1. Binding of KLC1 to caytaxin was also confirmed in HEK293T cells expressing Myc-tagged KLC1 and FLAG-tagged caytaxin by immunoprecipitation (Fig. 1B). Furthermore, when endogenous caytaxin was immunoprecipitated from the lysate of newborn rat brains, endogenous KHC was detected in the precipitate (Fig. 1C), suggesting the in vivo interaction of caytaxin with tetrameric kinesin-1. To confirm the binding of caytaxin and tetrameric kinesin-1, full-length FLAG-KLC1and a Myc-tagged KHC derivative (Myc-KIF5A-DN) which lacked the N-terminal motor domain but retained the ability to associate with KLC (Kimura et al., 2005) were produced in HEK293T cells and the call lysate was analyzed by GST pull-down assay for binding to GST-caytaxin(1-170) (Fig. 1D). Binding of caytaxin(1-170) and KLC1 was confirmed by this assay. KIF5A-DN was pulled down only when it was co-expressed with KLC1, indicating that caytaxin binds to tetrameric kinesin.

To narrow down the KLC1 binding motif within caytaxin(1-170), a series of deletion mutants were produced as GST-fusion proteins and their binding to KLC1 was assayed by GST pull-down. Deletion of an 18 amino acid segment from residues 109 to 126 abolished the caytaxin binding to KLC1 (Fig. 2A). Replacement of the three successive amino acids in this segment by alanine residues further narrowed down the KLC1 interacting region to an ELEWED sequence. Replacement of ELE to AAA greatly diminished, and that of WED to AAA completely abolished the binding of caytaxin(1-170) to KLC1 (Fig. 2B). However, other substitutions did not affect the interaction between KLC1 and caytaxin(1-170). When a full-length caytaxin protein with the WED to AAA mutation and KLC1 were expressed in HEK293T cells, the mutant caytaxin did not co-immunoprecipitate with KLC1 from the lysate (Fig. 2C), further substantiating the essential function of this sequence in the interaction with KLC1. The ELEWED motif is highly conserved among caytaxin homologues from various animals as well as among BNIP-2 family proteins, such as BNIP-2, BNIP-Sα (BNIPL) and BNIP-XL (Fig. 7). This motif is also conserved in KLC binding regions of the vaccinia envelope protein A36R and the neural vesicular transmembrane protein calsyntenin (alcadein), which are known to be transported by kinesin-1 (Araki et al., 2007; Konecna et al., 2006; Ward and Moss, 2004).

Overexpression of BNIP-2 induces cell elongation and membrane protrusion in non-neural cells (Zhou et al., 2005). Overexpression of BNIP-Sα results in cell rounding and apoptosis (Zhou et al., 2006). These effects have been attributed to the C-terminal CRAL-TRIO domains of these proteins. We examined the effects of caytaxin overexpression on the shape of hippocampal neurons (Fig. 3). Primary hippocampal neurons were transfected with GFP-caytaxin expression vectors on the day of preparation. On day 2 of in vitro culture (2 DIV), caytaxin-expressing cells were analyzed for the length of the longest neurites. The population of cells having longer neurites (length ≥50 μm) was increased by caytaxin overexpression from 35.5% to 71.0% or 61.3% depending on the position of the fused GFP polypeptide (GFP-caytaxin, a caytaxin fusion with GFP at the N-terminus; caytaxin-GFP, a caytaxin fusion with GFP at the C-terminus, respectively) (Fig. 3B). The average length was significantly increased by the expression of caytaxin from 41.0±16.5 μm (mean ± s.d.) to 60.0±17.1 μm for GFP-caytaxin or to 57.2±22.5 μm for caytaxin-GFP (Fig. 3B). Overexpression of caytaxin did not affect the number of neurites or their characteristics as axons or dendrites (data not shown). By contrast, knockdown of caytaxin expression by RNAi-inhibited elongation of neurites (Fig. 3C). Two RNAi molecules designed from different regions of caytaxin had essentially the same effect (data not shown), and an unrelated RNAi for the LacZ gene had no effect on the neurite extension (Fig. 3C).

So far, no biological activities have been assigned to the N-terminal regions of caytaxin, BNIP-2 and BNIP-Sα (Buschdorf et al., 2006; Qin et al., 2003; Zhou et al., 2005; Zhou et al., 2006). However, if the binding of caytaxin to KLCs has physiological significance, overexpression of the caytaxin N-terminal fragment might exert dominant-negative effects on the endogenous caytaxin. Actually, when caytaxin(1-170) was overexpressed in primary hippocampal neurons, neurite extension was compromised (Fig. 3D). However, the WED-to-AAA mutant of caytaxin(1-170) did not show this inhibitory activity. All these data indicate that caytaxin has a supportive effect on the elongation of neurites and binding to KLCs is essential for this activity, suggesting that the transportation by kinesin-1 is an integral feature of caytaxin.

To determine whether caytaxin was transported by kinesin-1, we expressed caytaxin fused with GFP in rat primary hippocampal neurons and analyzed cells by time-lapse fluorescence microscopy (Fig. 4). Expressed GFP-caytaxin protein showed a granular distribution along the neurites with accumulation at the distal region of neurites (Fig. 4A). A typical movement of a caytaxin-containing granule along an axon is shown as a movie in supplementary material Movie 1. During most of the observation period, the granule in an axon moved in the anterograde direction, but sometimes the granule stopped moving and then changed direction (Fig. 4B). For ∼72% of the time, migrating granules moved in the anterograde direction and for 13% of the time they moved in the retrograde direction (Fig. 4C). The average velocity of the anterograde movements was 1.23±0.87 μm/second (Fig. 4D), with the most frequent velocity range around 1.0-1.5 μm/second (Fig. 4E). The retrograde velocity (0.842±0.51 μm/second) did not differ very much from the anterograde velocity. These velocities were consistent with the speed of the fast axonal flow as well as with the speed of microtubule-dependent movement of kinesin motors (Hirokawa and Takemura, 2005).

To prove that the transport of caytaxin is mediated by kinesin-1, we overexpressed a dominant-negative form of KLC1 (KIF5A-DN) (Kimura et al., 2005) and measured endogenous caytaxin accumulated in growth cones (Fig. 4F). Caytaxin in growth cones was significantly decreased by the expression of KIF5A-DN.

To elucidate the destination and cargos of caytaxin transportation, colocalization of endogenous caytaxin with intracellular structures in neurites was analyzed by immunofluorescence staining of primary hippocampal neurons. Endogenous caytaxin was distributed both in axons that were stained with anti-tau-1 antibody, and in dendrites stained with anti-MAP2 antibody (Fig. 5A-F). Within the axonal growth cone, part of endogenous caytaxin colocalized with microtubules, which serve as tracks for kinesin motors (Fig. 5G-J). Caytaxin also colocalized with KHCs in the neurite and the growth cone (Fig. 5K-N). All these data support the idea that caytaxin is transported by kinesin.

The presynaptic marker synapsin colocalized with caytaxin (Fig. 5O-R). We therefore were transfected hippocampal neurons with GFP-shank3, which specifically localizes in the postsynapse (Boeckers et al., 2002). In contrast to the presynaptic marker synapsin, the shank-3-positive postsynapses were localized in close proximity to the caytaxin puncta but they did not exactly coincide (Fig. 5S-V). We can thus confirm that caytaxin is preferentially concentrated in the presynaptic region, as reported previously (Hayakawa et al., 2007).

The well-known kinesin-1 cargo protein APP (amyloid precursor protein) and the adaptor protein JIP-1b did not colocalize with caytaxin-positive granules (Fig. 6A-H), indicating that caytaxin is transported in structures other than vesicles containing APP or JIP-1b. Other cargoes of kinesin-1 are membranous organelles such as the endoplasmic reticulum and mitochondria. To visualize the endoplasmic reticulum, hemagglutinin-epitope (HA)-tagged JPDI was expressed in hippocampal neurons (Hosoda et al., 2003). Double staining with anti-HA and anti-caytaxin antibodies showed that caytaxin did not colocalize with the endoplasmic reticulum (Fig. 6I-L). By contrast, some of the caytaxin-positive puncta colocalized with mitochondria, which were stained with MitoTracker (Fig. 6M-P). We found that the inner plexiform layer of mouse retina expressed the highest level of caytaxin (T.A., unpublished results). Immunoelectron microscopy of the inner plexiform layer showed localization of caytaxin on the surface of mitochondria (Fig. 6Q), suggesting that caytaxin tethers mitochondria to kinesin. In fact, when caytaxin expression was suppressed by RNAi expression, the number of MitoTracker-stained particles in the distal part of neurites within 25 μm of the neurite tip was significantly diminished (Fig. 6R,S). These data suggest that caytaxin is involved in the distribution of mitochondria to the distal ends of neurites.

We have shown that caytaxin is bound to KLCs, transported along the neurites, and partially concentrated in the presynapse. The KLC TPR domain, which interacts with cargos of kinesin (Hirokawa and Takemura, 2005), recognizes the ELEWED sequence in the N-terminal region of caytaxin. The ELEWED sequence is highly conserved among caytaxin, BNIP-2, BNIP-Sα and BNIP-XL in a wide range of animals (Fig. 7). Considering the high homology of the CRAL-TRIO domains, it is quite reasonable to conclude that these four proteins belong to the same family of proteins whose functions depend on the interaction with the kinesin motor protein. Sequences similar to ELEWED are also conserved in other proteins that are known to bind KLC1 and are transported by kinesin-1 (Fig. 7). Mouse calsyntenin, a neural vesicular membrane protein, has two cytoplasmic KLC1-TPR-binding segments (KBS1 and KBS2) that contain amino acid sequences EMDWDD (KBS1) and QLEWDD (KBS2) (Konecna et al., 2006). Konecna and co-workers showed that amino acid residues M, W and two Ds following W in KBS1 were crucial for the interaction with KCL1, but the first E and the third D were less important. Vaccinia virus A36R protein has a SLIWDN sequence within the KLC1-TPR interaction domain (Ward and Moss, 2004). Taken together with the conserved motif in the four BNIP-2 family members, a novel consensus motif for interaction with the TPR regions of KLCs may be defined as xϕxW(E/D)x where ϕ is a hydrophobic amino acid and x is a preferably acidic amino acid. This `WED' motif could be reminiscent of WD40 motifs, which also recognize TPR motifs of various proteins (van der Voorn and Ploegh, 1992).

A TPR-containing ubiquitin E3 ligase, CHIP, interacts with caytaxin and efficiently ubiquitylates caytaxin, at least in vitro (Grelle et al., 2006). It has not yet been determined which domains of caytaxin and CHIP interact each other, but it is likely that the WED motif of caytaxin recognizes the TPR region of CHIP. If that is the case, it raises the possibility that the competition between KLC and CHIP and the consequent ubiquitylation and degradation of caytaxin exert an intriguing regulation of caytaxin functions.

No biological activities had been assigned to the N-terminal regions of caytaxin, BNIP-2 or BNIP-Sα (Buschdorf et al., 2006; Qin et al., 2003; Zhou et al., 2005; Zhou et al., 2006). When overexpressed in non-neural cells such as MCF-7 cells, BNIP-2 and BNIP-Sα induce dramatic morphological changes and apoptosis (Zhou et al., 2002; Zhou et al., 2005; Zhou et al., 2006). The C-terminal CRAL-TRIO domains of these proteins are thought to be solely responsible for these activities. Overexpression of caytaxin in MCF-7 cells also induced cell elongation that was indistinguishable from that caused by BNIP-2 (Hayakawa et al., 2007). A caytaxin C-terminal fragment that contained only the CRAL-TRIO domain showed virtually the same effects (T.A., unpublished data). These results had suggested seemingly dispensable function of the N-terminal region of the BNIP-2 family proteins. However, it should be noted that all these data were derived from overexpression of the BNIP-2 family proteins, and exact physiological functions of endogenous BNIP-2 and BNIP-Sα have been largely unknown. Overexpression of proteins could complement the requirement for intracellular trafficking mediated by kinesin motors. We showed that the N-terminal fragment of caytaxin inhibited neurite extension of primary hippocampal neurons in a WED-motif-dependent manner (Fig. 3D), suggesting that binding of caytaxin to KLCs should have physiological significance. Another recent finding supports the importance of the N-terminal region of caytaxin. A newly found chemically induced Atcay mutation allele, wobbly, which shows mild neurological symptoms, has a T to A transversion in the splice donor site of intron 4. This wobbly mutation is assumed to cause alternative splicing that deletes the exon 4 of the Atcay gene and causes in-frame splicing between exon 3 and exon 5, resulting in deletion of amino acids 46-119 (see http://mutagenetix.scripps.edu/). Because this deletion removes the ELEWED sequence but keeps the CRAL-TRIO domain intact, the wobbly phenotype could be caused by aberrant interaction with kinesin that results in impairment of axonal and dendritic transport.

We showed co-immunoprecipitation of KHCs with caytaxin from rat brain homogenate (Fig. 1C) and indirect binding of caytaxin to KHCs in vitro (Fig. 1D), suggesting the binding of caytaxin to tetrameric kinesin-1, which is the entity that moves along microtubules. Caytaxin colocalized with both microtubules and KHCs in the growth cone (Fig. 5G-N), and moves in neurites, mostly in the anterograde direction at a speed that is compatible with kinesin transport (Fig. 4B-D). Furthermore, the dominant-negative KHCs inhibit accumulation of caytaxin in the neurite tips (Fig. 4F). These data indicate that caytaxin is transported by kinesin-1. In support of this, Bushdorf and colleagues reported that overexpression of caytaxin in neurons relocalized KGA from mitochondria to neurite terminals (Buschdorf et al., 2006). Since KGA is a mitochondrial enzyme (Curthoys and Watford, 1995), relocalization of this enzyme could be accounted for by the transport of mitochondria harboring this enzyme with the caytaxin-kinesin complex. We showed that a proportion of caytaxin in neurites is colocalized with mitochondria, which are major cargos of kinesin-1 (Hirokawa and Takemura, 2005) (Fig. 6M-Q). Knockdown of caytaxin expression by RNAi redistributed mitochondria away from the distal part of neurite (Fig. 6R and S). It is therefore possible that caytaxin binds to mitochondria directly or indirectly via mitochondrial membrane proteins and tethers them to kinesin for transport.

Caytaxin-containing granules moved along the neurites mostly in the anterograde direction, but occasionally they stopped moving and switched to the retrograde direction. This behavior is compatible with that observed for calsyntenin-containing organelles (Araki et al., 2007; Konecna et al., 2006). The distribution (%) of caytaxin granules moving anterogradely, retrogradely and those in the stationary phase was also congruous with that reported for calsyntenin-containing organelles (Araki et al., 2007).

APP and JIP-1 are other major cargos for kinesin-1 (Hirokawa and Takemura, 2005). Caytaxin did not colocalize with APP or JIP-1, nor with calsyntenin-containing vesicles (T.A., unpublished data), suggesting that caytaxin-containing structures are specific vesicles or unique large protein complexes.

Considering the interaction of the caytaxin CRAL-TRIO domain with seemingly unrelated proteins such as KGA and Pin1, and the association of caytaxin directly or indirectly with mitochondria, it is probable that caytaxin functions as an adaptor that tethers various cargos to KLCs for transport by kinesin-1. In our two-hybrid screening, we isolated IP3KA that seemed to bind to the caytaxin CRAL-TRIO domain (T.A., unpublished data). Differential expression analysis between dt and wild-type rats identified various genes that encode proteins in the phosphatidylinositol-calcium signaling pathways, such as a phosphatidylinositol-binding protein, inositol-polyphosphate phosphatases and calcium-dependent ATPase-4, being dysregulated in the cerebellar cortex of the dt rats (Xiao et al., 2007). Taken together, caytaxin might function as a scaffold for signaling proteins in the phosphatidylinositol-calcium signaling pathways. In this sense, caytaxin might have functions similar to those of JIPs, which also bind to KLCs. JIP-1 associates with ApoER2 and APP and tethers vesicles containing these proteins to kinesin-1 via KLC. JIP-1 is also known to function as a scaffold for kinases of the JNK MAP kinase cascade (Whitmarsh and Davis, 1998). To clearly understand the physiological function of caytaxin, it will be essential to determine the complete repertoire of factors that interact with the caytaxin CRAL-TRIO domain.

Mouse and human caytaxin cDNAs were obtained from the mammalian gene collection, NIH (cDNA clone MGC:56757) and Kazusa DNA Research Institute (KIAA1872), respectively. The full-length caytaxin cDNA was modified by PCR amplification using primers that added a BamHI site before the initiation codon. The resulting fragment was cloned into a pcDNA3 derivative to make pcDNA3-FLAG-caytaxin, which expresses caytaxin protein with a FLAG epitope tag at the N-terminus. The same cDNA was cloned into pEGFP-C1 (Clontech) to make a vector that expresses a GFP-caytaxin fusion protein with GFP at the N-terminus of caytaxin. The full-length caytaxin cDNA was also amplified by PCR using primers that changed the termination codon into an XhoI site. The resulting fragment was cloned into a pcDNA3 plasmid derivative that contained GFP cDNA. The product plasmid expresses a caytaxin-GFP fusion protein with GFP at the C-terminus of caytaxin.

A fragment of cDNA encoding a mouse caytaxin N-terminal 1-170 amino acid fragment was PCR amplified using primers that replaced codon 171 into a stop codon and added a SalI site downstream of the new stop codon. The amplified fragment was cloned into pcDNA3 to make pcDNA3-FLAG-caytaxin(1-170). The same fragment was cloned into pGEX-5X-3 to express caytaxin(1-170) as a GST-fusion protein. Caytaxin (1-150), (1-126), (1-108) or (1-90) were similarly amplified by PCR. A fragment encoding caytaxin(55-170), (91-170), (109-170) or (127-170) was amplified using primers that introduced a new ATG codon with a BamHI site upstream. These deletion mutant fragments were cloned in pGEX-5X-3. Mutation of successive three amino acids in the region of residues 109-126 of caytaxin was achieved with QuikChange Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer's protocol.

pcDNA3-FLAG-KCL1 and pcDNA3-FLAG-TPR that express full-length KLC1 (1-542) and a KCL1 C-terminal fragment (177-542) containing the TPR region, respectively, as FLAG-tagged proteins, were kind gifts from Katsuji Yoshioka (Kanazawa University, Kanazawa, Japan). An N-terminal KLC1 fragment encoding residues 1-212 was PCR-amplified using pcDNA3-FLAG-KLC1 as template and cloned into pGEX-5X-3. To add Myc epitopes at the C-terminus of KLC1, the KLC1 cDNA was PCR amplified using primers that changed the stop codon to a BamHI site. The resulting fragment was inserted into a pcDNA3 derivative that contained three tandem sequences encoding the Myc epitope. The final plasmid product expressed the full-length KLC1-Myc, which has three Myc epitopes at the C-terminus. pCAGGS-Myc-KIF5A-DN, which expresses a dominant-negative form of KIF5A, was obtained from Kozo Kaibuchi (Nagoya University, Nagoya, Japan).

Rabbit polyclonal anti-caytaxin antibody was raised against a human caytaxin N-terminal fragment (residues 1-131). cDNA encoding the caytaxin(1-131) fragment was inserted in frame into pET32-b (Novagen) and the recombinant protein was expressed in E. coli BL21 cells. The protein was purified with a Ni-NTA agarose column (Qiagen). The purified protein was digested with 1 unit of thrombin per 100 μg of protein to remove the histidine tag. The digested sample was applied to a Ni-NTA agarose column. The pass-through fraction was collected and dialyzed against PBS. New Zealand White rabbits were injected subcutaneously with the purified protein. After three boosts, the whole blood was withdrawn to prepare serum. The serum was affinity purified using the GST-caytaxin(1-131) protein immobilized on a HiTrap NHS-activated HP column (Amersham) according to the manufacturer's protocol.

Monoclonal antibodies against KHC (H2), Tau-1 and APP were purchased from Chemicon; anti-MAP2 and anti-FLAG (M2) were from Sigma; anti-T7-tag was from Novagen; anti-synapsin was from StressGen. Rhodamine- and FITC-conjugated antibodies against mouse and rabbit immunoglobulin (Ig) were from Wako Chemicals and Jackson ImmunoResearch. Horseradish-peroxidase-conjugated antibodies against mouse and rabbit Ig were from GE Healthcare.

Yeast CG1945 cells were transformed with the pGBT9 vector (Clontech) containing the full-length mouse caytaxin cDNA and DNA prepared from an adult mouse brain MatchMaker cDNA library in pACT2 (Clontech). Yeast two-hybrid screening of three million yeast clones was carried out according to the Clontech manual. Clones growing on (–) histidine SD plates in the presence of 10 mM 3-amino-1,2,4-triazole were picked and analyzed for β-galactosidase activity. Plasmid DNA were recovered from 45 β-galactosidase-positive clones, transformed into E. coli HB101 cells to select for pACT2 plasmid and purified for DNA sequencing by ABI310 sequencer. To select for proteins binding to the caytaxin N-terminal region, the HF7c yeast cells were transformed with the pGBT9-caytaxin(1-170) plasmid and a pACT2 clone DNA isolated by the screening. The transformed cells were selected on SD selection plates containing 10 mM 3-amino-1,2,4-triazole and transferred to a liquid SD medium. After culturing overnight, the cells were harvested, disrupted by glass beads, and the cell extract was measured for β-galactosidase activity according to the Clontech liquid culture protocol.

HEK293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS and antibiotics in 10 cm dishes. After 24 hours of plating 1.2×10⁶ cells per dish, cells were transfected with 20 μg plasmid DNA by the calcium phosphate precipitation method (Chen and Okayama, 1987). Cells were disrupted 2 days later with 1 ml per dish of lysis buffer [50 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1.5 mM MgCl₂, 10% glycerol, 1% Triton X-100 and a protease inhibitor cocktail (2 μg/ml leupeptin, 2 μg/ml pepstatin, 2 mM benzamide, 40 μg/ml bestatin, 1 mM phenylmethylsulfonyl fluoride)]. The lysate was centrifuged at 100,000 g for 30 minutes and the supernatant recovered for further analysis by immunoprecipitation and GST pull-down assay.

KLC1, caytaxin(1-170), and their derivatives were produced as GST-fusion proteins in E. coli BL21 cells and purified by glutathione-Sepharose 4B (GE Healthcare). The purified GST-protein (5 μg) was mixed with 200 μl cell lysate of HEK293T cells expressing either FLAG-KLC1, Myc-KKIF5A-DN, caytaxin or their derivatives and incubated for 4 hours at 4°C. Then, 10 μl glutathione-Sepharose 4B beads were added to the mixture and the incubation continued for 4 hours at 4°C. The beads were collected and washed with lysis buffer five times. The proteins bound to the beads were separated by SDS-acrylamide gel electrophoresis, transferred to PVDF membrane and subjected to western blot analysis. The membrane was blocked with TBST [20 mM Tris-HCl, pH 7.5, 150 mM NaCl and 0.1 % (v/v) Tween20] containing 5% skimmed milk for 1 hour at room temperature and then incubated with the anti-FLAG antibody 2000-fold diluted in TBST containing 5% skimmed milk for 3 hours at room temperature. After washing extensively with TBST, membrane was incubated with horseradish-peroxidase-conjugated anti-mouse Ig antibody (GM Healthcare, 1:2000 dilution). The bound peroxidase was detected with ECL (GM Healthcare).

The V1 membrane fraction was prepared from newborn rat brains by differential centrifugation as described previously (Konecna et al., 2006; Morfini et al., 2001). In brief, brains were homogenized in two volumes of a solution containing 0.32 M sucrose, 10 mM HEPES buffer, pH 7.4, 5 mM EDTA, 4 mM N-ethylmaleimide and the protease inhibitor cocktail. The homogenate was centrifuged at 12,500 g for 8 minutes. The supernatant was centrifuged at 40,000 g for 40 minutes. The resulting supernatant was centrifuged at 100,000 g for 1 hour. The pellet (V1 membrane fraction) was suspended in the IP buffer (50 mM Tris-HCl, 75 mM NaCl, 5 mM EDTA, 10 mM CHAPS, 5 mM N-ethlymaleimide and protease inhibitor cocktail).

Protein-A-Sepharose (20 μl) was incubated with 1 μg anti-FLAG, anti-Myc (9E10.2) or purified anti-caytaxin antibody for 30 minutes at 4°C. The beads were washed with PBS and then mixed with the HEK293T cell lysate (400 μl) or the V1 membrane fraction (400 μl). After 3 hours of incubation at 4°C, the beads were washed with lysis buffer five times. Proteins bound to the beads were resolved on SDS polyacrylamide gel and electrotransferred to PVDF membrane. The membrane was subjected to western blot analysis using the following mouse monoclonal antibodies: anti-FLAG (1:2000 dilution), anti-Myc (1:2000 dilution) and anti-KHC (1:1000 dilution) antibodies. The membrane was incubated with horseradish-peroxidase-conjugated anti-mouse Ig antibody (GM Healthcare, 1:2000 dilution). The bound peroxidase was detected with ECL (GM Healthcare).

Preparation of hippocampal neurons from embryonic day 18 (E18) rat embryos was carried out as described (Inagaki et al., 2001). In short, hippocampi were resected from E18 rat embryo brains and cells were dispersed by treatment with 0.5 mg/ml papain. Dispersed cells were suspended in Neurobasal medium (Gibco) supplemented with 10% FBS, and plated on poly-D-lysine-treated laminin-coated coverslips placed in wells of a 24-well plate at 0.5×10⁵ cells/well. Three hours after plating, the medium was replaced with 0.4 ml/well complete neuron medium (Neurobasal medium supplemented with B-27 supplement (Gibco), 1 mM glutamate and 2.5 μM cytosine β-D-arabinofuranoside). Half of the medium was changed every 3 days. One hour before transfection, the coverslips were transferred to new wells containing 0.5 ml fresh complete neuron medium. Lipofectamine 2000 (2 μl) was diluted with 50 μl of Neurobasal medium. Plasmid DNA (0.5 μg total) was mixed with 50 μl Neurobasal medium. These two solutions were mixed and incubated for 20 minutes at room temperature. The mixture was added to the well and the cells were incubated for 90 minutes. The cells were then washed twice with DMEM and culture continued in the complete neuron medium (0.5 ml/well).

For observation of hippocampal neurons earlier than 3 days in vitro (3 DIV), neurons were transfected before plating as follows. Dispersed neurons (6×10⁵ cells) were suspended in 0.3 ml Neurobasal medium supplemented with 10% FBS. Plasmid DNA (1 μg) and Lipofectamine 2000 (2 μl) were mixed as described above and added to the cell suspension. The cell suspension was placed in a CO₂ incubator for 90 minutes. The cells were washed with Neurobasal medium containing10% FBS and the cells (1.5×10⁵ cells) were plated on a coated coverslip. After 90 minutes, the medium was changed to complete neuron medium. To obtain efficient expression of RNAi, the transfected neurons washed with Neurobasal medium containing 10% FBS were suspended in 1.7 ml complete neuron medium and cultured as floating cells in an uncoated well for 24 hours. Cells were then plated on a coated coverslip.

For examination of caytaxin localization in neurons expressing the dominant-negative KHC (KIF5A-DN), neurons were transfected with pCAGGS-Myc-KIF5A-DN as described above on 2 DIV. An expression vector for FLAG-tagged β-galactosidase (LacZ) was used for control. After 48 hours of culture, the cells were incubated with 5 μM CellTracker blue CMAC (Invitrogen) for 2 hours at 37°C just before fixation. Then, the cells were fixed and stained with anti-Myc or anti-FLAG antibody using rhodamine-conjugated anti-mouse Ig as the secondary antibody or with anti-caytaxin antibody using FITC-conjugated anti-rabbit Ig as the secondary antibody.

Cultured hippocampal neurons were fixed with 4% paraformaldehyde in PBS for 30 minutes at room temperature. The cells were washed with PBS and permeabilized with methanol for 15 minutes at –20°C. After washing with PBS, the cells were incubated with a primary antibody diluted with PBS containing 1% bovine serum albumin for 1 hour at room temperature. After washing with PBS, the cells were incubated with a diluted FITC- or rhodamine-conjugated secondary antibody for 1 hour at room temperature. After washing with PBS, the coverslip was mounted on a glass slide with PermaFluor (Beckman Coulter). Mitochondria were stained with MitoTracker Orange CMTMRos (Invitrogen).

Knockdown by plasmid-based RNAi was carried out using the pcDNA6.2-GW/EmGFP-miR vectors (Invitrogen) that expressed two different rat caytaxin RNAi sequences purchased from Invitrogen (Rmi647864 and Rmi647866). The pcDNA6.2-GW/EmGFP empty vector and the vector expressing β-galactosidase (LacZ) RNAi were used as controls. The efficiency of RNAi was assayed by measuring the intensity of endogenous caytaxin immunofluorescence in GFP-positive hippocampal neurons in culture. Expression of Rmi647864 and Rmi647866 siRNAs suppressed endogenous caytaxin levels to 29±7.5% (s.d., n=12) and 35±9.5% (n=26), respectively, compared with cells expressing GFP alone (100±2.8%, n=16) or LacZ siRNA (100.4±10.7%, n=22). These two caytaxin siRNAs gave essentially the same results and the results with Rmi647864 were described in the text.

Eyes were removed from 4-week-old ICR mice which were perfusion-fixed with 4% paraformaldehyde and 0.3% glutaraldehyde in PBS. The retina was resected and further fixed overnight with the same solution. The retina was immersed in PBS overnight on ice and then dehydrated with an ethanol series and embedded in LR white (Nisshin EM Corporation). Ultrathin sections 70 nm thick were mounted on uncoated Ni #300 grids. The sections were treated with 3% H₂O₂ for 5 minutes, rinsed with H₂O and blocked with 2% normal goat serum and 2% normal donkey serum in PBS for 30 minutes. Sections were incubated with anti-caytaxin antibody and then reacted with 15-nm-gold-conjugated goat anti-rabbit IgG (Biocell Research Laboratories). After immunolabeling, sections were stained with 2% hafnium chloride in methanol.

To observe fine structures, the fixed retina was treated with 1% OsO₄ in BPS and dehydrated with an ethanol series and embedded in Spurr resin (Nisshin EM Corporation). Ultrathin sections of 70 nm were mounted on Cu/Rh #300 grids and stained with 2% hafnium chloride followed by staining with a mixture containing 1% lead (II) citrate, 1% lead (II) nitrate, 1% lead (II) acetate, 2% Na citrate and 0.18 M NaOH. Samples were visualized with a transmission electron microscope (H-7100, Hitachi).

Fluorescence images were captured with an Olympus BX50 fluorescence microscope equipped with a Nikon DS-2MBWc CCD camera using NIS-elements software (Nikon) or with a Carl Zeiss LSM510 confocal microscope. The images were processed by Photoshop (Adobe). The length of the longest neurite and the fluorescence intensity were measured using ImageJ (NIH). Time-lapse images were taken with Deltavision 2 (Applied Precision) equipped with Axiovert S100 (Carl Zeiss) or with a fluorescence microscope (IX71, Olympus) equipped with a UIC-GE CCD camera (MDS Analytical Technologies) using MetaMorph software. The images were taken at 2- to 5-second intervals and compiled into a movie using Windows movie maker (Microsoft). To measure the anterograde and retrograde velocities, the images of GFP-caytaxin granules (n=30) were recorded at 2 second intervals for 30 seconds and the distance that the granule moved between two frames were divided by two. Statistical significance was analyzed by Student's t-test. The results shown in figures represent means ± s.d.

We thank Naoyuki Inagaki, Nara Institute of Science and Technology (NAIST) for kind help in experiments with primary culture of hippocampal neurons and observation with time-lapse fluorescent microscopes; Megumi Iwano, NAIST, for the help in immunoelectron microscopy; Kenji Kono, NAIST, for providing us with JPDI cDNA; Tobias M. Boeckers, University of Ulm, for Shank3 cDNA; Hisashi Koga and Takahiro Nagase, Kazusa DNA Research Institute, for mouse calsyntenin and caytaxin cDNAs, respectively; Katsuji Yoshioka, Kanazawa University and Kozo Kaibuchi, Nagoya University, for kinesin plasmids.