KLP6: a newly identified kinesin that regulates the morphology and transport of mitochondria in neuronal cells

Cytoskeleton-dependent movement of mitochondria influences a variety of cellular behaviors including proliferation, differentiation and apoptotic events, as well as mitochondrial respiratory activity and inheritance (Boldogh and Pon, 2007; Frederick and Shaw, 2007; Fagarasanu et al., 2010). Defects in mitochondrial movement are often manifested in polarized cell types. Axonal transport in neurons requires the most rapid and longest mitochondrial movements. It is driven by molecular motors along the cytoskeleton, and is probably required to maintain local ATP levels at the synapse (Miller and Sheetz, 2004; Hollenbeck and Saxton, 2005; Chen and Chan, 2009). In axons, mitochondria are delivered in both the anterograde and retrograde directions by a combination of transport and stopping events (Ligon and Steward, 2000; Miller and Sheetz, 2004; Frederick and Shaw, 2007). In budding yeast, inheritance of mitochondria into the emerging bud sites is regulated by bidirectional transport along the cytoskeleton (Frederick and Shaw, 2007; Fagarasanu et al., 2010).

Mitochondrial transport is mediated by three types of molecular motor proteins that travel along cytoskeletons. In multicellular eukaryotic cells, most mitochondria move along microtubules, whereas in yeast cells actin filaments are the primary cytoskeleton for mitochondrial transport (Boldogh and Pon, 2007; Fagarasanu et al., 2010). In human a myosin, Myo19, has recently been found to function in actin-based mitochondrial movements (Quintero et al., 2008). Microtubule-dependent movement is controlled by two different types of motors, dyneins and kinesins. Dyneins deliver mitochondria toward the minus end of microtubules, whereas kinesins transport their cargo toward both the plus and minus ends of microtubules, depending on the position of the motor domain at the N- or C-terminus (Schliwa and Woehlke, 2003; Vale, 2003). Kinesins are classified into 14 families (Lawrence et al., 2004; Hirokawa and Noda, 2008). To date, two kinesins, kinesin heavy chain (KHC; also known as KIF5) and KIF1Bα, are known to be involved in mitochondrial movement (Nangaku et al., 1994; Tanaka et al., 1998). KHC/KIF5 belongs to the kinesin-1 family and is a conventional kinesin that has been conserved evolutionarily. In mice, two brain-specific isoforms (KIF5A and KIF5C) have been identified in addition to a ubiquitous isoform (KIF5B) (Aizawa et al., 1992; Kanai et al., 2000). In mouse disruption of the gene encoding KIF5B is embryonic lethal, and the null mutant cells exhibit mitochondrial clustering at the perinuclear region (Tanaka et al., 1998). KHC/KIF5 transports a variety of cargo, including mitochondria, through distinct adaptor proteins (Hirokawa and Noda, 2008). In mammals, Miro1/2, OIP106/GRIF-1 and syntabulin are adaptors that allow KHC/KIF5 to associate with mitochondria (Brickley et al., 2005; Cai et al., 2005; Fransson et al., 2006). KIF1Bα, a member of the kinesin-3 family, is one of two KIF1B isoforms that are splicing variants of the same gene and that share a motor domain at the N-terminus. KIF1Bβ delivers synaptic vesicle precursors through a distinct cargo-binding domain at the C-terminus (Zhao et al., 2001). An in vitro motility assay using purified mitochondria revealed that KIF1Bα transports mitochondria along microtubules (Nangaku et al., 1994). A two-hybrid screen showed that KIF1B interacts with KBP, a peripheral protein associated with mitochondria that is required for neuronal outgrowth (Wozniak et al., 2005; Lyons et al., 2008). Nonetheless, neuronal death in Kif1b knockout mice can be rescued by expressing only KIF1Bβ (Zhao et al., 2001), suggesting the redundancy of KIF1Bα function in mitochondrial movement.

Despite the fact that various cellular behaviors are regulated by mitochondrial dynamics, only two kinesins among more than 44 members of human kinesin-like proteins are known to contribute to mitochondrial movements (Hirokawa and Noda, 2008). In the present study, using RNA interference (RNAi) to screen for motor proteins involved in mitochondrial morphology in Caenorhabditis elegans, we found that a newly identified kinesin, kinesin-like protein KLP6, is involved in mitochondrial dynamics. KLP6 is crucial for transport of mitochondria in neuronal cells as well as maintenance of their morphology.

To identify molecular motors and the related proteins required to maintain mitochondrial morphology, 89 C. elegans genes were examined by RNAi screening (supplementary material Table S1). Bacteria expressing specific double-stranded RNAs were fed to C. elegans, both to efficiently produce loss-of-function RNAi phenocopies and to minimize physical damage to the worms. After feeding for at least 24 hours, the worms were stained with Mitotracker Red CMXRos, a membrane-potential-dependent mitochondrial vital dye. In the control wild-type worms, tubular mitochondria were observed in the body wall muscles (Fig. 1A). Among the 89 genes tested, three genes (klp-6, dlc-4 and dnc-4) reproducibly caused abnormal mitochondrial morphology when they were knocked down (Fig. 1A,C). Most of mitochondria in the body wall muscles appeared to fragment and mitochondrial aggregation was often observed (Fig. 1A, arrowheads). Similar results were obtained in intestinal cells. The predicted genes dlc-4 and dnc-4 encode putative members of the dynein light chain and dynactin families, respectively. klp-6 codes for a kinesin, and is required for polycystin localization in sensory cilia (Peden and Barr, 2005). In this study, klp-6 was subjected to subsequent analyses.

Members of the kinesin family possess motor domains that are highly homologous within the family (Lawrence et al., 2004; Hirokawa and Noda, 2008). To confirm whether knockdown of klp-6 induces changes in mitochondrial morphology, three regions corresponding to the N-terminal, central and C-terminal segments of C. elegans KLP-6 (CeKLP-6) were used to knock down expression using feeding RNAi (Fig. 1F). As expected, all three RNAi induced changes in mitochondrial morphology in the body wall muscle cells (Fig. 1A,C), indicating that klp-6 knockdown results in abnormal mitochondria. We previously generated transgenic worms that constantly express a mitochondria-targeted fluorescent protein in body wall muscles (Ichishita et al., 2008). When klp-6 was knocked down in the transgenic worms, mitochondrial fragmentation and aggregation were also observed without vital dye staining (Fig. 1B). Thus, the downregulation of klp-6 expression induced changes in mitochondrial morphology. A previous report showed that expression of CeKLP-6 lacking the motor domain at the amino terminus interferes with the CeKLP-6 function (Peden and Barr, 2005). Therefore, the same mutant (KLP-6C–GFP) was expressed in the body wall muscle, and the mitochondrial morphology in the transgenic worms was observed by Mitotracker staining (Fig. 1D,F). Ectopic expression of KLP-6C–GFP induced a significant increase in the number of worms with abnormal mitochondria, compared with the GFP-expressing worms (Fig. 1E). These results indicate that KLP-6 is involved in the formation and/or maintenance of mitochondrial morphology.

To further analyze the roles of this kinesin in mitochondrial morphology at the cellular level, we generated polymerase chain reaction (PCR) primers to clone a cDNA coding for an ortholog of CeKLP-6 from rat tissues. Sequence analysis of the reverse transcription-polymerase chain reaction (RT-PCR) products revealed a coding region for 1034 amino acid residues containing motor, forkhead-associated and three coiled-coil domains. Rat KLP6 (rKLP6) shares 53.8% and 26.9% sequence similarities with the motor and forkhead-associated domains of CeKLP-6, respectively (Fig. 2A). Comparative analysis showed that rKLP6 and its putative orthologs in vertebrates are uncharacterized kinesins and belong to the kinesin-3 family (Lawrence et al., 2004; Miki et al., 2005). Anti-rKLP6 antibody specifically recognized a single 105-kDa protein in total cell lysates of rat PC-12 and human HeLa cells (Fig. 2B). An attempt to determine the intracellular distribution of endogenous KLP6 using the antibody was unsuccessful because of the low-level expression of KLP6. Because the expression of a C-terminally hemagglutinin (HA)-tagged rKLP6 (rKLP–3HA) did not influence mitochondrial morphology (Fig. 2C), the association of KLP6 with mitochondria was analyzed in the rKLP–3HA-expressing cells. Cell fractionation indicated that rKLP6–3HA was mainly distributed in the cytoplasm, and was particularly associated with the endomembranes (Fig. 2D). Flotation analysis using a discontinuous density gradient showed that rKLP6–3HA was cofractionated with mitochondria-enriched membranes (Fig. 2D, lane 5), suggesting that a fraction of KLP6 is associated with mitochondrial membranes.

To determine whether rKLP6 is involved in the formation and/or maintenance of mitochondrial morphology, we expressed rKLP6 lacking the N-terminal motor domain (GFP–rKLP6C) in HeLa cells stably expressing a mitochondria-targeted red fluorescent protein (mito-RFP), because motor domainless forms of the kinesin-3 family members are reported to exert dominant-negative effects (Dorner et al., 1998; Lee et al., 2002; Xue et al., 2010) and overexpression of a full-length rKLP6 did not affect mitochondrial morphology (Fig. 2C). Upon the expression of GFP–rKLP6C, the number of cells with a short and cup-shaped mitochondria significantly increased (Fig. 3A, arrowhead) compared with control GFP-transfected cells: 41% of cells had cup-shaped mitochondria after 24-hours incubation following transfection (Fig. 3B), indicating that rKLP6 is involved in maintaining normal mitochondrial morphology. Expression of GFP–rKLP6C did not affect the morphology of other organelles such as the endoplasmic reticulum, Golgi apparatus and peroxisomes (supplementary material Fig. S1), or the organization of the microtubules (supplementary material Fig. S2).

Mitochondrial morphology usually depends on its intracellular movement in addition to membrane fusion and fission (Boldogh and Pon, 2007; Frederick and Shaw, 2007; Chen and Chan, 2009; Fagarasanu et al., 2010). Therefore, rKLP6 function in mitochondrial motility was assessed by time-lapse imaging in living cells. Expression of GFP–KLP6C led to a significant decrease in mitochondrial motility. The length of the track of one end of the mitochondria in a 15-min period was obviously shorter in cells expressing GFP–rKLP6C (Fig. 3C; supplementary material Movies 1 and 2). The mean velocity of randomly selected mitochondria in the cells transfected with GFP only was 19.6±2.0 nm/second, consistent with the mitochondrial velocity in A549 human epithelial cells (Quintero et al., 2008). By contrast, the mean velocity of mitochondria in GFP–rKLP6C-transfected cells was 4.3±0.6 nm/second, a 78% decrease over that of control cells (P<0.001; Fig. 3D). Thus, introduction of GFP–rKLP6C inhibits intracellular movement of mitochondria.

A previous report showed that CeKLP-6 functions in the transport of polycystin in sensory neurons (Peden and Barr, 2005). To study the roles of KLP6 in mitochondrial transport in neuronal cells, KLP6 was knocked down in mouse neuroblastoma neuro 2a cells. We designed three different RNA duplexes matched to a coding region of the mouse hypothetical gene (Gm1305), which is predicted to be an ortholog of rat Klp6. In neuro 2a cells, doublet bands were recognized reproducibly by an affinity-purified antibody to rKLP6. One of the short interfering RNAs (siRNAs) induced an approximately 62% decrease in the amount of KLP6 protein when transfected into neuro 2a cells (Fig. 4A). The siRNA for mouse ‘KLP6’ (mKLP6) was transfected into neuro 2a cells stably expressing mito-RFP, and the cells were incubated for 24 hours with low serum medium supplemented with retinoic acid to induce neuronal differentiation. In neurons, mitochondria are delivered through axons in a combination of transport and stopping events, resulting in a uniform distribution along the axon (Hollenbeck, 1996; Ligon and Steward, 2000; Miller and Sheetz, 2004). In most of the control cells, the axon-like neurites were filled with mitochondria (Fig. 4B, a), but a small fraction of these control cells had axon-like neurites in which the mitochondria were scattered. Knockdown of mKLP6 caused a twofold increase in neurites that contained only a small number of mitochondria (GFP RNAi, 11.1±1.7%; mKLP6 RNAi, 21.3±2.5%; Fig. 4B). Similar results were obtained with GFP–rKLP6C. Ectopic expression of rKLP6 suppressed the defect in mitochondrial distribution in the neurites of neuro 2a cells in which mKLP6 had been knocked down (vector, 20.9±3.0%; rKLP6 expression, 13.2±4.2%; supplementary material Fig. S3). These results indicate that downregulation of KLP6 induces a delay in transport of neuronal mitochondria.

To quantify mitochondrial motility in neuronal cells, we used time-lapse imaging in living cells, examined the kymograph and analyzed mitochondria that moved continuously for more than 3 μm. In the GFP-transfected cells, the mean velocity of mobile mitochondria during persistent anterograde movement were 0.371±0.022 μm/seconds (n=100 randomly selected mitochondria; Fig. 4C; supplementary material Movie 3). Upon the expression of GFP–rKLP6C, the mean velocity in the anterograde direction was 0.201±0.027 μm/seconds (n=31 mobile mitochondria), which was a significant reduction (P<0.0003; Fig. 4D; supplementary material Movie 4). In the control cells, the velocities of most mitochondria were distributed broadly in the range of 0.1–0.6 μm/seconds, whereas the velocities in the GFP–rKLP6C-transfected cells were concentrated around a peak of 0.1 μm/seconds (Fig. 4E). Similarly, the mean velocity of mobile mitochondria during anterograde movement in the mKLP6-knocked down cells (si-mKLP6) significantly decreased (0.212±0.016 μm/seconds; n=61 mobile mitochondria), compared with the control luciferase-knocked down cells (si-Luc; 0.386±0.026 μm/seconds, n=107 mobile mitochondria, P<0.0001; Fig. 4D). By contrast, the mean velocity during retrograde movement in the GFP–rKLP6C-transfected cells was 0.261±0.027 μm/seconds (n=26 mitochondria), which was not significantly different from that in the GFP-transfected cells (0.343±0.030 μm/seconds, n=45 mitochondria, P>0.05; Fig. 4F). Furthermore, the velocity profiles in the retrograde direction in GFP- and GFP–rKLP6C-transfected cells were similar (supplementary material Fig. S4). In addition, expression of GFP–rKLP6C resulted in a 60% decrease in the number of mobile mitochondria in axon-like neurites (GFP, 19.2±2.9%, n=520 mitochondria; GFP–rKLP6C, 7.8±2.4%, n=444 mitochondria; Fig. 4G). These results indicated that KLP6 has a role in mitochondrial transport during anterograde movement. Note that transfection with GFP–rKLP6C resulted in 80% less differentiated neuro 2a cells that transfection with GFP alone (supplementary material Fig. S5). Nonetheless, the transfection efficiency of the GFP–rKLP6C expression plasmid was only twofold lower than that of the GFP vector, suggesting that neuro 2a cells expressing GFP–rKLP6C died during differentiation. The addition of a pan-caspase inhibitor completely restored the number of transfected cells that differentiated, resulting in a 2.5-fold increase in the number of differentiated GFP–rKLP6C-transfected cells. Thus, the culture condition for differentiation (i.e. low serum medium containing retinoic acid) appears to be detrimental to GFP–rKLP6C-transfected cells, resulting in apoptotic cell death.

To date, two kinesins, KHC/KIF5 and KIF1Bα, are known to function in mitochondrial transport (Nangaku et al., 1994; Tanaka et al., 1998). To study the roles of the three kinesins, including KLP6, in transport of mitochondria in neuronal cells, we used the motor domainless GFP fusion proteins, rather than gene silencing, because of both the redundant expression of the KIF5A, -B and -C genes and the alternative splicing of the KIF1B gene in neuronal cells (Aizawa et al., 1992; Tanaka et al., 1998; Kanai et al., 2000; Zhao et al., 2001). After 24 hours incubation with low serum medium containing retinoic acid, almost all axon-like neurites of the control cells contained many clusters of mitochondria (Fig. 5A). Similarly, mitochondria were abundantly distributed in neurites of the cells expressing GFP–KIF1Bα (ΔMD) and GFP–rKLP6C (Fig. 5A). By contrast, approximately 60% of the cells expressing GFP–KIF5B(ΔMD) had neurites that completely lacked mitochondria (Fig. 5Ae,B), and the remaining cells contained only a few clusters of mitochondria in their neurites (Fig. 5Ah) suggesting that GFP–KIF5B(ΔMD) inhibited the entry of mitochondria from the cell body to the axon and neurite. It is worth noting that the expression of the motor domainless forms, GFP–KIF5B(ΔMD) and GFP–KIF1Bα (ΔMD), interfered with cell growth and differentiation in neuro 2a cells as well as GFP–rKLP6C-expressing cells. Therefore, a pan-caspase inhibitor was needed to examine mitochondrial distribution in the neurites.

To quantify mitochondrial motility, GFP–KIF1Bα (ΔMD) was introduced into neuro 2a cells stably expressing mito-RFP. In cells expressing GFP–KIF1Bα (ΔMD), the mean velocity of mobile mitochondria during persistent anterograde movement was 0.211±0.015 μm/seconds (n=54 mobile mitochondria), which was significantly lower than that in the GFP-transfected cells (0.486±0.027 μm/seconds, n=124 mitochondria, P<0.0001; Fig. 5C). This is consistent with a previous report that KIF1Bα drives microtubule-based mitochondrial movement in an in vitro reconstitution assay (Nangaku et al., 1994). It should be noted that a pan-caspase inhibitor did not affect the cell number in the control GFP-transfected cells (supplementary material Fig. S5), but caused an increase in the motility of neuronal mitochondria (compare Fig. 5C with Fig. 4D). To examine the roles of KLP6 and KIF1Bα in mitochondrial movement, GFP–KIF1Bα (ΔMD) was coexpressed with the motor domainless KLP6 (N-HA–KLP6C) mutant, in which GFP was removed to avoid interaction with GFP–KIF1Bα (ΔMD). To simultaneously express both motor domainless forms in a single cell, we used a bicistronic expression plasmid (Oka et al., 2008) that contains the internal ribosomal entry site (IRES) sequence between N-HA–KLP6C and GFP–KIF1Bα (ΔMD), driven by a single promoter. Immunofluorescence microscopy confirmed that both motor domainless forms were constantly coexpressed in single cells. As expected, the expression of N-HA–KLP6C alone also interfered with mitochondrial movement (0.176±0.013 μm/seconds, n=28 mitochondria). Simultaneous expression of GFP–KIF1Bα (ΔMD) and N-HA–KLP6C did not apparently influence the mean velocity of mitochondria (0.233±0.016 μm/seconds, n=77 mitochondria), compared with those in the cells expressing the single proteins (P>0.05; Fig. 5C). This finding raises two possibilities; one is that a single mitochondrial movement requires two independent kinesins. Another possibility is that the motor domainless forms inhibit the functions of both kinesins by competing for the common associated proteins. KBP was recently identified as a KIF1-binding protein associated with mitochondria, and its zebrafish kbp mutants exhibit a defect in the distribution of axonal mitochondria (Wozniak et al., 2005; Lyons et al., 2008). Therefore, we examined the interaction between KLP6 and KBP. KBP was co-transfected with one of the three kinesins, and immunoprecipitated. KLP6 and KIF1Bα, but not KIF5B, obviously co-immunoprecipitated with KBP (Fig. 5D). Together with the fact that KLP6 and KIF1Bα belong to the kinesin-3 family, these findings suggest that KLP6 and KIF1Bα cooperatively function via KBP in mitochondrial transport in neuronal cells.

In the present study, we identified three C. elegans genes (klp-6, dlc-4 and dnc-4) as molecular motors involved in mitochondrial morphology, using RNAi screening. This suggests that each of these microtubule-dependent motors (i.e. kinesin, dynein and dynactin) is crucial for maintaining mitochondrial morphology. Recently, human myosin Myo19 was reported to be involved in mitochondrial motility (Quintero et al., 2008). In our assay, however, none of the myosins reproducibly influenced mitochondrial morphology. A direct relation between CeKLP-6 and mitochondria has not yet been shown. Barr and colleagues reported that CeKLP-6 transports PKD-2, a transient receptor protein channel, in sensory neurons (Peden and Barr, 2005). PKD-2 is proposed to form a receptor and/or channel complex with LOV-1, in which the PLAT domain physically interacts with ATP-2, the β subunit of ATP synthase (Igarashi and Somlo, 2002; Hu and Barr, 2005). These findings suggest the involvement of CeKLP-6 in mitochondrial movements along the C. elegans sensory neurons, although we did not observe this.

We cloned a cDNA for rat KLP6, which is a novel kinesin in vertebrates. Analyses of the mouse KLP6 knockdown and the overexpression of rKLP6 mutants revealed that KLP6 is crucial for maintaining mitochondrial morphology and regulating transport of mitochondria in neuronal cells. Because KLP6 has a motor domain at the amino terminus, it is predicted to transport its cargo toward the plus end of microtubules (Schliwa and Woehlke, 2003; Vale, 2003). Strikingly, our findings indicate that the motor domainless forms and knockdown of KLP6 decrease mitochondrial velocity during anterograde movement (i.e. toward the plus end of microtubules). Large-scale microarray analysis showed that the gene encoding mouse KLP6 (known as hypothetical gene Gm1305) is ubiquitously expressed (Su et al., 2004; Wu et al., 2009), suggesting that KLP6 might be implicated in mitochondrial movement in all tissues. In HeLa cells expressing GFP–rKLP6C, membrane fusion and fission events of mitochondria appeared normal (supplementary material Movies 1 and 2), suggesting that KLP6 contributes to maintenance of mitochondrial morphology by regulating their movements. Furthermore, cup-shaped mitochondria were frequently found upon expression of GFP–rKLP6C. Treatment with CCCP, a protonophore that disrupts the membrane potential, induced severe mitochondrial fragmentation in the GFP–rKLP6C-transfected cells (unpublished observation). These results suggest that cup-shaped structures depend on membrane fusion events and are not mitochondrial aggregates. Furthermore, there is no significant difference in Mitotracker Red CMXRos staining between cup-shaped and filamentous mitochondria (unpublished observation), suggesting that the cup-shaped mitochondria maintain their membrane potential. Mitochondria have been shown to be preferentially localized on acetylated microtubules with high curvature (Friedman et al., 2010). Therefore, inhibition of mitochondrial movement may emphasize mitochondria associated with curved microtubules.

In mammals, two kinesins (KIF1Bα and KHC/KIF5) are involved in mitochondrial transport (Nangaku et al., 1994; Tanaka et al., 1998). The physiological functions of three kinesins, including KLP6, in mitochondrial movement remain obscure. Primary cultured cells prepared from Kif5B^−/− mice exhibit a perinuclear clustering of mitochondria (Tanaka et al., 1998), indicating that KIF5B has a role in anterograde mitochondrial movement from the microtubule-organizing center. This is consistent with our finding that GFP–KIF5B (ΔMD) inhibits the delivery of the mitochondria from the cell bodies to the axons and neurites. The downregulation of Miro1, a KHC/KIF5-associated mitochondrial membrane protein, however, reduced mitochondrial motility in axonal transport (MacAskill et al., 2009; Misko et al., 2010), suggesting that KHC/KIF5B is also required for axonal transport of mitochondria. Two isoforms (α and β) of KIF1B are transcribed by alternative splicing, and in neurons they transport mitochondria and synaptic vesicle precursors, respectively (Zhao et al., 2001). Deficiencies of the Kif5B^−/− neurons can be rescued by KIF1Bβ expression, suggesting that other kinesins can compensate for the loss of KIF1Bα function in neurons. One candidate is KHC/KIF5, a conventional kinesin. However, it is unlikely to directly replace KIF1Bα because KIF1B belongs to the kinesin-3 family whereas KHC/KIF5 belongs to the kinesin-1 family. In addition, there is no evidence of KIF1B binding to syntabulin and OIP106/GRIF1, which anchor KIF5 to mitochondria. Another candidate is KLP6 because it is also in the kinesin-3 family and commonly binds to KBP. Further examination is needed to determine the reason for the functional redundancy.

KBP was originally identified as a KIF1-binding protein (Wozniak et al., 2005), and its zebrafish kbp mutants have defective axonal outgrowth and maintenance (Lyons et al., 2008). Furthermore, mutations in the human KBP/KIAA1279 gene are associated with Goldberg–Shprintzen syndrome, a disorder characterized by malfunction of the central and enteric nervous systems (Brooks et al., 2005). The KIF1B gene is responsible for human Charcot–Marie–Tooth disease type 2A, an inherited peripheral neuropathy (Zhao et al., 2001). A putative gene for the human KLP6 ortholog is located next to the AHCTF1 gene on chromosome 1q44. Some chromosomal deletions mapped to this region were recently identified in patients with inherited disorders (e.g. Dandy–Walker complex) (Poot et al., 2007; van Bon et al., 2008; Aktas et al., 2010). The clinical phenomena as a consequence of KLP6 deficiency are intriguing.

The C. elegans strains used in this study, wild-type Bristol N2 and transgenic worms carrying mitochondria-targeted fluorescent proteins (Ichishita et al., 2008), were maintained at 20°C on nutrient growth medium (NGM) agar plates seeded with Escherichia coli OP50, according to standard techniques (Sulston and Hodgkin, 1988).

An RNAi screen was performed as described previously (Ichishita et al., 2008) with some modifications. The C. elegans RNAi feeding library (MRC Geneservice) was used for the RNAi screen. E. coli cells expressing double-stranded RNA specific to each target gene were grown overnight in 2× YT medium containing 25 μg/ml carbenicillin and 12.5 μg/ml tetracycline (both from Sigma-Aldrich), diluted 100-fold, grown for an additional 6 hours, and then seeded onto NGM agar plates containing 25 μg/ml carbenicillin and 5 mM isopropylthiogalactoside (Sigma-Aldrich) for the induction of double-stranded RNA. To examine mitochondrial morphology, approximately 100 L1 larvae were placed onto NGM agar plates seeded with the E. coli cells and incubated for 24 hours at 20°C. For Mitotracker staining, 150 μl of 5 μM Mitotracker Red CMXRos (Invitrogen) solution was added to the NGM plates and the worms were incubated for a further 24 hours. They were then transferred to an agar pad on glass slides, and the mitochondria were analyzed under a fluorescence microscope.

To express C. elegans KLP-6 lacking the motor domain predominantly in the body wall muscle cells, the coding region (+1177 bp to +2784 bp) of C. elegans klp-6 from the expressed sequence tag (EST) cDNA clone yk1134d3 was subcloned into pPD95.77 containing the myo-3 promoter (Ichishita et al., 2008) to construct pmyo-3–KLP-6C–GFP. A DNA solution (50 μg/ml) containing pmyo-3–KLP-6C–GFP was microinjected into young adult hermaphrodites, and transgenic worms with an extrachromosomal array were selected under a fluorescence microscope.

Rat EST clones (IMAGE 7323123; Invitrogen), coding for a part of KLP6, were sequenced. To obtain the 5′ and 3′ regions of the rat KLP6 cDNA, total RNA was prepared from rat liver and brain. RT-PCR was performed using gene-specific primers, and sequences of the PCR products were determined. These PCR products were ligated with the EST clone to construct a full-length clone of rat KLP6, and cloned into pBluescript II (Stratagene). The nucleotide sequence data reported in this paper were deposited in the nucleotide sequence databases with the accession numbers (AB573712). To construct the expression plasmid pEF1-rKLP6–3HA, the entire coding region for rKLP6 was cloned into pEF1-3HA, in which the EF1 promoter expressed the inserted gene as a three-tandem HA-tagged protein (Oka et al., 2008). A DNA fragment of the coding region (+1342 bp to +3102 bp) was subcloned into pEGFP-C3 (Clontech) for the expression of GFP–rKLP6C fusion protein.

HeLa and neuro 2a cells were maintained at 37°C in medium A (Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 4.5 mg/ml glucose and non-essential amino acids). DNA transfection was performed using FuGENE6 (Roche Diagnostics) according to the manufacturer's instructions. To obtain neuro 2a cells stably expressing mito-RFP (a mitochondria-targeted fluorescent fusion protein) cells transfected with pIRES-puro-Su9RFP (Ishihara et al., 2003) were grown in medium A containing 5 μg/ml puromycin, and individual colonies were harvested and grown to mass culture. The relative protein expression levels were determined under a fluorescence microscope, and those cell lines that expressed mito-RFP were used in subsequent experiments.

An siRNA duplex (target sequence; 5′-aaggaactcacgaggtatgga-3′) to the mouse putative gene (Gm1305), an ortholog of rKLP6, was transfected into undifferentiated neuro 2a cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. After incubation for 48 hours, the cells were further transfected with the same siRNA duplex at least once for morphological analysis. To construct pGSH1–GFP–si-mKLP6, a plasmid expressing a short hairpin RNA (shRNA) for mKLP6, oligonucleotides including the above target sequence (5′-gatccggaactcacgaggtatggagaagcttgtccatacctcgtgagttccttttttggaagc-3′ and 5′-ggccgcttccaaaaaaggaactcacgaggtatggacaagcttctccatacctcgtgagttccg-3′) was synthesized and ligated, according to the manufacturer's instructions, into the shRNA expression vector pGSH1–GFP (GeneSilencer), which simultaneously expresses both GFP and the shRNA.

Subcellular fractionation was performed as described previously (Oka et al., 2008) with minor modifications. HeLa cells expressing rKLP6–3HA were scraped off and suspended in homogenization buffer [10 mM Hepes-KOH (pH 7.4), 70 mM sucrose and 0.22 M mannitol] containing the protease inhibitor cocktail Complete EDTA-free (Roche Diagnostics). The cell suspension was homogenized by passing 30 times through a 27-gauge needle, and then centrifuged at 600 g for 5 minutes. The supernatant was homogenized and further centrifuged at 5000 g for 10 minutes to obtain a low-speed pellet fraction. The resultant supernatant was then centrifuged at 100,000 g for 60 minutes to separate the high-speed supernatant and high-speed pellet fractions. To prepare a mitochondria-enriched fraction, the low-speed pellet fraction was subjected to flotation using a 0, 20, 25 and 36% discontinuous gradient of OptiPrep (Axis-Shield) as described previously (Wood-Allum et al., 2006). After ultracentrifugation at 100,000 g for 4 hours, the proteins that migrated to the interface between the 20 and 25% OptiPrep solutions were collected as a mitochondria-enriched (Mt) fraction.

Immunoblotting was performed as described previously (Oka et al., 2004). Immuno-Star AP (Bio-Rad) was used as a substrate for chemiluminescence, and images were acquired using Luminoimage analyzer LAS 4000 (Fujifilm). Immunofluorescence microscopy was performed essentially as described previously (Oka et al., 2008). Images were acquired using a Radiance 2000 confocal microscope (Bio-Rad). Changes in mitochondrial morphology were calculated (as means ± s.d.) using data from at least three independent experiments. In each experiment, 100–200 individual cells were counted. Antibodies used for immunoblotting and immunofluorescence microscopy were as follows: rabbit polyclonal antibodies: anti-Tim17 (Ishihara and Mihara, 1998), anti-Sec61β (Upstate), anti-H450 (Ishihara et al., 1990), anti-Tom40 (Suzuki et al., 2000); mouse monoclonal antibodies: anti-HA (16B12) (Covance).

A DNA fragment of the rKLP6 coding region (+1342 bp to +2145 bp) was subcloned into pMAL-c2 (New England Biolab) and pGEX4T-3 (GE Healthcare) to prepare MBP and GST fusion proteins, respectively. MBP–KLP6 protein was expressed in E. coli BL21 cells, and purified using an amylose resin column according to the manufacturer's instructions. GST–KLP6 was expressed as an inclusion body in the E. coli cells, purified by SDS gel electrophoresis, eluted from the gels, and used for immunizing rabbits to generate antibodies. The antiserum was affinity-purified by chromatography through a column conjugated with MBP–KLP6.

Transfected cells were washed with phosphate-buffered saline, harvested and solubilized in immunoprecipitation buffer [20 mM Hepes-KOH (pH 7.4), 1% Triton X-100, 150 mM NaCl and 10% glycerol]. After incubation at 4°C for 60 minutes, the cell lysate was centrifuged to remove insoluble materials. The resultant supernatant was incubated with anti-FLAG (M2) agarose (Sigma-Aldrich) for 3 hours at 4°C, washed three times with immunoprecipitation buffer, and then washed once with detergent-free immunoprecipitation buffer. The precipitates were analyzed by SDS-PAGE and immunoblotting.

HeLa and neuro 2a cells expressing mito-RFP were transfected with either pEGFP-C3, pEGFP-C3–rKLP6C, pGSH1–GFP–si-Luc or pGSH1–GFP–si-mKLP6, and then cultured with medium A for 24–48 hours in a glass-bottomed dish. To induce differentiation of neuro 2a cells, the cells were further incubated for 15–24 hours with Dulbecco's modified Eagle's medium containing 20 nM retinoic acid (Sigma-Aldrich), 1% fetal bovine serum, 4.5 mg/ml glucose and non-essential amino acids. In some experiments, 0.1 mM zVAD-fmk, a pan-caspase inhibitor, was added during differentiation to avoid apoptotic cell death. Time-lapse images were collected using a fluorescence microscope IX81 (Olympus) with a cooled charge-coupled device camera (Roper Scientific). Images were analyzed using Metamorph software (Universal Imaging) and NIH ImageJ (National Institute of Health) as described previously (Miller and Sheetz, 2004). To calculate the mean velocities of mobile mitochondria in differentiated neuro 2a cells, neurites that were more than twice the length of their cell body were selected as axon-like neurites.

We thank Keiichi Nakayama and Michiko Shirane for the neuro 2a cells. We are also grateful to the Caenorhabditis Genetics Center (CGC) for providing the C. elegans wild-type Bristol N2 strain. This work was supported by grants from the Ministry of Education, Science and Culture of Japan, and the Naito Foundation.