Sunday Driver/JIP3 binds kinesin heavy chain directly and enhances its motility

To determine if syd's interaction with kinesin‐1 is entirely dependent on KLC, we performed GST‐pulldown assays combined with syd deletion analyses. N‐terminal syd (N‐syd), but not C‐terminal syd (C‐syd) was able to pull down both KLC and the neuron‐specific kinesin heavy chain KIF5C, hereafter referred to as KHC (Figure 1A and B). VAMP3, an endosomal protein that does not bind to kinesin‐1, is used as a negative control. Deletion of the LZ domain (N‐sydΔLZ) resulted in loss of interaction with KLC1 and KLC2 (Figure 1C), consistent with previous observations (Verhey et al, 2001; Kelkar et al, 2005; Nguyen et al, 2005). However, N‐sydΔLZ retained the ability to interact with KHC (Figure 1C). Deletion of the JNK binding domain or coiled‐coil domains 2 and 3 did not prevent syd's interaction with either KLC or KHC (Figure 1C). Since KLC1 and KLC2 have almost identical TPR regions (Rahman et al, 1998) and interact with wild‐type and mutant syd in a similar manner (Bowman et al, 2000), we focused our study on KLC1. To define the region interacting with KHC, we tested three N‐syd fragments for their interaction with KHC and KLC. syd240–540, which contains the LZ domain, interacted with both KLC and KHC (Figure 1D). syd541–772, which lacks the LZ domain, did not interact with KHC or KLC (Figure 1D). However, syd3–239, which lacks the LZ domain, was not associated with KLC, as expected, but retained the ability to interact with KHC (Figure 1D), indicating that this syd fragment interacts with a KHC dimer lacking KLC.

To determine if a pool of KHC lacking the light chain exists in brain extract, we performed sucrose density centrifugation analysis. Similarly to what has been reported previously in cultured mammalian cells (DeLuca et al, 2001), we detected a population of KHC lacking the light chain in brain extracts (Figure 1E). The light chain was not degraded in these extracts and was detected in fractions of higher density along with a subset of KHC (Figure 1E). Quantification of KHC in the two distinct pools indicates that about 14% of KHC lacks the light chain. These results suggest that syd can associate with the pool of KHC that lacks KLC.

To determine the residues required for syd interaction with KHC, we generated a series of truncation mutants. This analysis revealed that aa50–80 was required to promote syd's interaction with KHC (Figure 2A and B). To determine if aa50–80 (hereafter referred to as Kinesin‐1 Heavy Chain Binding Domain, KBD) represents the minimal domain promoting syd–KHC interaction, we performed a deletion analysis. Deletion of the LZ domain abolished syd's interaction with KLC but not with KHC (Figures 1C and 2C). Deletion of the KBD did not prevent syd's interaction with tetrameric kinesin‐1, consistent with the LZ domain interacting with KLC (Figure 2C). Since GFP–sydΔLZ binds to KHC dimer and GFP–sydΔKBD interacts with tetrameric kinesin‐1 via KLC, deletion of both the KBD and LZ domains was required to prevent syd's interaction with kinesin‐1 (Figure 2C), indicating that the KBD is necessary for binding to KHC.

The C‐terminal tail domain of KHC (aa807–956) mediates its interaction with several cargoes (reviewed in Hirokawa et al, 2009). To test whether syd binds the KHC tail domain, we conducted a GST‐pulldown assay using a GST‐tagged KHC tail (aa807–956). syd interacted with the GST–KHC tail, but not with GST alone (Figure 2D). To test whether syd directly binds to the KHC tail domain, we performed an in vitro binding assay using GST‐tagged KHC tail (aa807–956), and His‐tagged syd fragments. All three syd fragments containing the KBD interacted directly with the GST–KHC tail (Figure 2E). To determine whether the affinity between syd and KHC is similar to what has been previously measured between syd and KLC (Bowman et al, 2000), we performed a binding assay. Similarly to what has been reported (Bowman et al, 2000), we measured a Kd of 0.33±0.08 μM between syd and KLC1‐TPR (Figure 2F). The syd–KHC tail interaction displayed a lower affinity, with a Kd of 1.83±0.43 μM (Figure 2G). Given the lower affinity of syd for KHC tail compared with KLC‐TPR, syd may bind to a kinesin‐1 tetramer containing both KHC and KLC, and syd binding to KHC may relieve or alter its autoinhibitory activity. However, kinesin‐1 does not always exist as a tetramer, suggesting that syd can bind directly to a population of KHC that lacks KLC. Together, these results reveal an alternative and KLC‐independent interaction between syd and kinesin‐1.

To determine whether syd's interaction with KHC is functional, we analysed the dynamic properties of GFP‐tagged syd by total internal reflection fluorescence microscopy (TIRF), as previously described (Blasius et al, 2007). COS cells were chosen for their low amounts of endogenous kinesin‐1 (Cai et al, 2007) and were used to express the desired GFP–syd constructs or Flag–KHC. Cell lysates containing either Flag–KHC or GFP–syd were mixed before the motility assay, thereby ensuring identical amounts of KHC in each condition (Figure 3A). When lysate expressing GFP–syd wild‐type (wt) was mixed with Flag–KHC lysate, motile events were observed (Figure 3B–D; see Supplementary Movie S1). However, when non‐transfected cell extract was used in place of the Flag–KHC lysate, the absence of Flag–KHC dramatically reduced the number of motile events (Figure 3C and D; see Supplementary Movie S2), indicating that the low level of endogenous kinesin‐1 in COS cells does not have a significant role under these conditions. The movement of GFP–syd observed in these conditions is thus mainly mediated by its interaction with Flag–KHC. We measured a velocity of 0.62±0.02 μm/s for GFP–syd wt (Figure 3E), which is similar to the velocity reported for kinesin‐1 under comparable conditions (Blasius et al, 2007). The average run length we measured (3.21±0.24 μm; Figure 3F) was greater than previously reported for kinesin‐1 under comparable conditions (Blasius et al, 2007), but similar to purified recombinant kinesin‐1 (Dixit et al, 2008). We then tested whether GFP–syd3–239 binding to KHC was sufficient for motility. GFP–syd3–239 displayed motile events, albeit with reduced frequency compared with GFP–syd wt (Figure 3D; Supplementary Movie S3). Surprisingly, we observed that in addition to a reduced frequency of motile events, the velocity and run length decreased significantly for GFP–syd3–239, to 0.39±0.03 μm/s and 2.11±0.17 μm/run, respectively (Figure 3E and F). The reduced run length could be attributed to GFP–syd3–239 detaching from KHC sooner as compared with GFP–syd wt. However, the reduced velocity observed for GFP–syd3–239 compared with GFP–syd wt suggests that other syd structural elements may participate in KHC regulation.

To directly determine whether syd regulates KHC motility, we next observed the movement of KHC–mCit in the presence or absence of Flag‐tagged syd wt (Flag–syd wt; Figure 4A). When mixed with non‐transfected extracts, KHC–mCit displayed few motile events (Figure 4B and C), probably reflecting activation by endogenous adaptors in COS cells. We, however, collected sufficient motile events to analyse KHC–mCit motile properties. We measured a velocity of 0.33±0.03 μm/s and a run length of 1.94±0.39 μm/run (Figure 4D and E; see Figure 4F and G for the raw distribution; and Supplementary Movie S4), which is comparable to what has been reported previously for full‐length kinesin‐1 motility in similar assays (Sung et al, 2008; Loiseau et al, 2010). In the presence of Flag–syd wt, we observed a dramatic increase in the number of motile events (Figure 4B and C). Remarkably, we also observed a significant increase in velocity (1.01±0.03 μm/s) and run length (5.54±0.28 μm/run) (Figure 4D and E; see Figure 4F and G for the raw distribution data and Supplementary Movie S5). The velocity and run length of Flag–KHC in the presence of GFP–syd wt (Figure 3E and F) were found to be lower than that for KHC–mCit in the presence of Flag–syd wt (Figure 4D and E). This difference may result from the position of the tag, rendering Flag–KHC not as efficient as KHC–mCit. However, importantly, in both cases the presence of syd wt increased KHC motility. This increase in motility is likely due to syd binding to the KHC tail, since Flag–sydΔKBD only minimally affected KHC motility (Figure 4D and E; see Figure 4F and G for the raw distribution data; Supplementary Movie S6). This velocity is comparable to what has been measured for purified tail‐less kinesin‐1 (Hackney et al, 2003; Yildiz et al, 2008; Shastry and Hancock, 2010), indicating that syd's binding to the KHC tail is effective in relieving the inhibition of the motor domain by the KHC tail. However, the enhanced run length in the presence of syd is significantly greater than the run length reported for purified tail‐less kinesin‐1 (Dixit et al, 2008; Yildiz et al, 2008; Shastry and Hancock, 2010). Similarly to what we observed with GFP–syd3–239 motility (Figure 3E and F), we observed that Flag–syd 3–239 was able to activate KHC motility, but with reduced velocity and run length compared with Flag–syd wt (Figure 4D and E; see Figure 4F and G for the raw distribution data; Supplementary Movie S7). Together, these results indicate that binding of syd to KHC activates kinesin‐1 for microtubule‐dependent motion, enhancing both KHC velocity and run length, and that syd structural elements beyond the KHC binding site likely participate in KHC regulation. It is also possible that a small proportion of the observed motile events is mediated by a KHC–KLC tetramer. Indeed, we found that a small fraction of expressed KHC–mCit in COS cells can interact with endogenous KLC (Supplementary Figure S1). The relatively small increase in KHC velocity provided by Flag–sydΔKBD (Figure 4D) may thus represent an effect of Flag–sydΔKBD on a kinesin‐1 tetramer via its interaction with KLC.

To determine whether syd binding to KHC is sufficient for transport in neurons, we examined the localization of GFP–syd wt and mutants in cultured hippocampal neurons. Previous evidence suggests that syd localizes predominantly to axons during development and mostly to the cell bodies and dendrites in adult brain (Akechi et al, 2001; Setou et al, 2002). In agreement with these previous observations, we found that in cultured embryonic hippocampal neurons, endogenous syd is highly enriched at the tips of developing axons (Figure 5A). Similarly, expressed GFP–syd wt accumulated mostly at the tip of axons (Figure 5B, upper panel and 5E). GFP–sydΔLZ, which binds KHC, or GFP–sydΔKBD, which interacts with tetrameric kinesin‐1 via KLC was transported equally to the axon tip (Figure 5B middle two panels and Figure 5E). GFP–syd wt, GFP–sydΔLZ, and GFP–sydΔKBD all displayed a similar low level distribution along dendrites (Figure 5C and E). In contrast, GFP–sydΔΔ failed to exit the cell body (Figure 5B and C bottom panel and Figure 5E) and was confined to the perinuclear region in proximity of the Golgi apparatus (Figure 5D). Lack of transport due to GFP–sydΔΔ misfolding is unlikely since GFP–sydΔΔ retains the ability to oligomerize with JIP family members in cells (Figure 6). Therefore, syd's interaction with KHC does not appear to specify axonal versus dendritic transport. Together, these results suggest that syd's interaction with KHC can mediate transport in neurons in the absence of the KLC binding site, as deletion of both KHC and KLC binding sites is required to prevent syd transport to axon tips.

syd/JIP family members are known to homo‐ or hetero‐oligomerize (Yasuda et al, 1999; Kelkar et al, 2000, 2005; Kristensen et al, 2006; Hammond et al, 2008). To determine whether oligomerization with endogenous JIP family members bound to KLC mediates the transport of syd mutants, we first tested the ability of GFP‐tagged syd mutants to oligomerize with Flag–syd wt. Flag–syd wt was co‐transfected with GFP, GFP–syd3–239, GFP–syd wt, or GFP–sydΔΔ in N2A cells and immunoprecipitation was performed with anti‐GFP antibodies. Flag–syd wt co‐immunoprecipitated with GFP–syd3–239, GFP–sydΔΔ and GFP–syd wt but not with GFP alone, revealing that both GFP–syd3–239 and GFP–sydΔΔ can oligomerize, similarly to syd wt (Figure 6A). However, despite its ability to oligomerize with Flag–syd wt, GFP–sydΔΔ failed to exit the cell body (Figure 5C and D). Therefore, oligomerization of GFP–sydΔΔ with endogenous syd does not appear to be sufficient for its transport in neurons. We next examined the localization of mCherry–syd3–239 in cortical neurons prepared from mice lacking syd (Kelkar et al, 2003). Immunofluorescence of cultured cortical neurons (Figure 6B) and western blot analysis of brain lysates (Figure 6C) confirmed the lack of syd expression in syd−/− animals. We observed an enrichment of the syd mutants that bind KHC but not KLC, mCherry–syd3–239 and GFP–sydΔLZ, at the axonal tips of neurons lacking syd (Figure 6D–F). These results further exclude that oligomerization with endogenous syd mediates transport of syd mutants that bind KHC but not KLC.

syd can form hetero‐oligomers with some JIP family members. In particular, syd can oligomerize with JIP2 and JIP1 (Kelkar et al, 2000; Hammond et al, 2008), but it does not oligomerize with JIP4 (Kelkar et al, 2005). While GFP–syd wt and GFP–sydΔΔ retained the ability to oligomerize with Flag–JIP2, GFP–syd3–239 failed to do so (Figure 6G). In contrast, all three constructs, GFP–syd wt, GFP–syd3–239 and GFP–sydΔΔ displayed the ability to oligomerize with myc–JIP1. Yet, despite its ability to oligomerize with myc–JIP1, GFP–sydΔΔ failed to exit the cell body (Figure 5C and D). This is in agreement with the results from Hammond et al (2008), showing that the formation of a JIP1/syd/KLC complex is necessary for efficient JIP1 or syd transport in neuronal cells. Thus, GFP–sydΔΔ and GFP–syd3–239 may be unable to be stably incorporated in a complex with JIP1 and KLC. Oligomerization with JIP1 may thus not fully account for the transport of syd lacking the KLC binding domain (GFP–syd3–239, Figure 6E). Although we cannot exclude that other yet unknown kinesin‐1 binding proteins may be involved in syd transport in neurons, these results, together with our motility experiment, suggest that syd's interaction with KHC promotes transport in neurons.

Kinesin‐1 is a processive motor, which takes multiple steps along microtubules before dissociating. How kinesin‐1 activation for microtubule transport is controlled in live cells is not well understood, but recent studies couple kinesin‐1 activation for microtubule transport with the binding of cellular partners (Blasius et al, 2007; Cho et al, 2009; Loiseau et al, 2010). A proposed regulatory mechanism for kinesin‐1 activation is the transition from a ‘folded’ inactive state to an ‘open’ active state. In the inactive folded conformation, the KHC tail domain interacts with both the motor domain and the microtubules to prevent kinesin motion (Friedman and Vale, 1999; Hackney and Stock, 2000; Cai et al, 2007; Dietrich et al, 2008; Watanabe et al, 2010). Activation of kinesin‐1 for transport requires a conformational change in which motor and tail domains are separated and the motor domains come closer together (Cai et al, 2007). Our in vitro motility results indicate that syd, but not sydΔKBD, which lacks the KHC binding domain, increased the number of motile events (Figure 4), suggesting that syd binding to the KHC tail domain efficiently relieves the inhibition by the KHC tail domain, activating or opening KHC to bind microtubules for long‐range motility. These results place syd alongside Pat1 (Loiseau et al, 2010) and RanBP2 (Cho et al, 2009) as cellular regulators of kinesin‐1 activity.

In the case of tetrameric kinesin‐1, it has been proposed that binding of both KLC and KHC is required for motor activation (Blasius et al, 2007). It will be interesting to determine in future studies whether syd may fulfill activation of a KHC/KLC complex via its ability to interact with both KHC and KLC. Alternatively, similar to the JIP1–Fez1 complex (Blasius et al, 2007), syd may require additional interacting partners for the activation of tetrameric kinesin‐1.

Kinesin‐1 stepping along microtubules is believed to involve concerted conformational change and diffusive movement of the tethered head to the next binding site. The precise mechanisms regulating the speed and the distance that kinesin‐1 can achieve are fairly well understood for purified kinesin‐1. Yet less is known about how kinesin‐1 motility is regulated in a cellular environment. The observation that different kinesin‐1 cargoes move at different rates in a cellular environment illustrates the complexity of in vivo motor regulation (Araki et al, 2007). The increase in KHC velocity and run length in the presence of full‐length syd and syd3–239 suggests that syd binding to the KHC tail relieve the ‘brake’ provided by the tail binding to the microtubule (Dietrich et al, 2008; Seeger and Rice, 2010; Watanabe et al, 2010), thus allowing efficient forward movement. A recent study indicates that the Drosophila Pat1 protein interacts with KHC and functions as a positive regulator of KHC motility for the transport of oskar mRNA and dynein in Drosophila oocytes (Loiseau et al, 2010). In the absence of Pat1, both kinesin‐1 velocity and run length are reduced (Loiseau et al, 2010). The Ran binding protein 2 (RanBP2) activates the ATPase activity of KHC (Cho et al, 2009), suggesting that RanBP2 may also regulate kinesin‐1 velocity and processivity. These observations support the notion that adaptors contribute to regulate kinesin‐1 motile properties, in addition to their roles in recruiting cargoes.

Our results indicate that syd is capable of associating with both tetrameric and dimeric kinesin‐1. This result is consistent with studies indicating that kinesin‐1 exists and functions as a dimer of two heavy chains lacking the light chains, in addition to its conventional tetrameric conformation. KLC‐independent transport has been reported for mitochondria (Cai et al, 2005; Glater et al, 2006), syntaxin‐containing vesicles (Su et al, 2004), and RNA particles (Palacios and St Johnston, 2002; Kanai et al, 2004; Loiseau et al, 2010), in agreement with earlier studies reporting that KHC dimers can bind membrane organelles in the absence of KLC (Skoufias et al, 1994). Furthermore, a small pool of KHC not associated with KLC has been found in cultured HeLa cells and bovine brain (Hackney et al, 1991; DeLuca et al, 2001; Gyoeva et al, 2004) and we obtained similar results from mouse brain (Figure 1E). Degradation of KLC during kinesin‐1 isolation can be excluded, since we detected KLC in our experiment (Figure 1E). Furthermore, significant molar excess of KHC over KLC has been reported in CV‐1 cells (Gyoeva et al, 2004). It is thus conceivable that spatially and temporally, KHC and KLC do not always fully colocalize. Indeed, it has been shown that KLC is absent in photoreceptor cells and that throughout the retina, KLC does not fully colocalize with KHC (Mavlyutov et al, 2002), suggesting that at the cellular and subcellular levels KHC localization does not fully overlap with KLC. In addition, during brain development, KHC and KLC decline after the first week of postnatal life, but the decline in KLC appears to be more pronounced (Morfini et al, 2001). Therefore, it is possible that syd interacts with tetrameric kinesin‐1 via KLC, while in cells or subcellular regions where KLC is absent syd instead interacts and regulates the activity of a kinesin‐1 dimer. Our observation that syd interacts with KHC in addition to and independently from its known interaction with KLC is not inconsistent with the data published so far. In C. elegans, the localization of UNC‐16 (syd) depends on both UNC‐116 (KHC) and KLC (Sakamoto et al, 2005). Furthermore, syd transport to neurite tips in differentiated CAD cells is mostly, but not completely dependent on KLC (Verhey et al, 2001; Hammond et al, 2008). Similarly to syd, the FMRP was found to bind KLC (Dictenberg et al, 2008), and also KHC in a KLC‐independent manner (Ling et al, 2004).

Kinesin‐1 drives different sets of cargoes to axons or dendrites, but how kinesin‐1 distinguishes between axonal or dendritic cargoes for directional sorting remains poorly understood. Previous studies suggested that there was a dendritic preference for KHC cargoes and an axonal preference for KLC linkage (Setou et al, 2002; Hirokawa and Takemura, 2005). Our results showed that regardless of its mode of association with kinesin‐1, syd is predominantly targeted to axon tips (Figure 5). It is thus more likely that the uploaded cargoes, and not the adaptor–motor complexes, determine the destination of kinesin‐1 and its adaptor. We have recently shown that syd mediates the transport of at least two distinct types of vesicles in axons: endosomes and small anterogradely moving vesicles (Abe et al, 2009). It will be interesting to define in future studies the nature of the syd cargoes transported by KLC‐dependent or KLC‐independent interaction. Indeed, KLC isoforms have been proposed to mediate targeting of KHC to proper cargo (Stenoien and Brady, 1997; Khodjakov et al, 1998; Gyoeva et al, 2000; Wozniak and Allan, 2006), but KHC can also associate with membranous cargo in the absence of KLC (Skoufias et al, 1994).

The propensity of JIP family members to form homo or hetero‐oligomers (Yasuda et al, 1999; Kelkar et al, 2000, 2005; Kristensen et al, 2006; Hammond et al, 2008; Koushika, 2008) suggests that syd may be transported via oligomerization. Indeed, Hammond et al (2008) reported that syd and JIP1 require each other for efficient transport of JIP1 or syd in non‐neuronal cells. Such cooperative transport is due to an interaction between JIP1 and syd as well as distinct binding sites on the KLC‐TPR domain. We found that despite its ability to oligomerize with myc–JIP1, GFP–sydΔΔ failed to exit the cell body (Figure 5C and D). Thus, GFP–sydΔΔ may be unable to be stably incorporated in a complex with JIP1 and KLC. Oligomerization with JIP1 may thus not fully account for the transport of the syd mutant GFP–syd3–239 lacking the KLC binding domain. Although we cannot exclude that other yet unknown kinesin‐1 binding proteins may be involved in syd transport in neurons, our results suggest that syd's interaction with KHC may promote transport in neurons and that oligomerization may provide additional layers of regulation of syd‐dependent transport.

In summary, we identified syd as an adaptor for kinesin‐1 heavy chain and we determined that syd enhances KHC motility along microtubules. Future studies are needed to determine the precise mechanisms by which syd regulates kinesin‐1 activation and processivity, and examine whether the distinct modes of syd interaction with kinesin‐1 provide specificity for cargo selection and delivery to particular subcelullar destinations.

syd antibody was previously described (Bowman et al, 2000). The antibodies are as follows: anti‐GFP (Invitrogen), anti‐VAMP3 (Synaptic System), anti‐kinesin heavy chain (KIF5C; Xia et al, 2003), anti‐KLC1 (Santa Cruz), anti‐KLC1–KLC2 (63–90) (gift from Dr Scott Brady), anti‐His₆ (Qiagen), anti‐Flag (Sigma), anti‐β‐tubulin (Sigma), anti‐myc (Cell Signaling), anti‐β‐actin (Chemicon), anti‐giantin (Abcam), anti‐MAP2 (Chemicon), and anti‐tau (Millipore). syd knockout mice were previously described (Kelkar et al, 2003). For wild‐type animals, C57B6 mice were used.

KHC–mCit and myc–JIP1 were obtained from Dr Kristen Verhey and were previously described (Cai et al, 2007; Hammond et al, 2008). Full‐length mouse syd cDNA was purchased from Open Biosystems (Huntsville, AL). Constructs for GST fusion proteins (N‐terminal syd (aa 3–772), C‐terminal syd (aa 773–1337), syd3–239, syd240–540, syd541–772, syd3–80, syd3–160, syd50–239, syd101–239, syd150–239, 201–239, syd429–459, syd50–82, KIF5C 807–956) were generated by inserting their coding sequences in‐frame into the EcoRI and XhoI sites of the pGEX‐4T‐1 vector. Full‐length clone of KHC (KIF5C) was isolated from a mouse cDNA library generated from adult mouse brain and inserted in‐frame downstream of sequences encoding Flag in the pcDNA3 vector. To generate poly‐histidine‐tagged KLC1, KLC1‐TPR (KLC1 177–413), syd3–80, syd3–160, syd3–239, KHC 807–956, the individual coding sequences were inserted into XhoI and EcoRI sites of pET‐28a(+) vectors. GFP–syd and GFP–syd3–239 were obtained by ligating the coding sequences into XhoI and EcoRI sites of the pEGFP‐C1 vector. mCherry–syd3–239 was generated by cloning syd3–239 into the mcherry‐C1 vector. Flag–syd wt and mutants, and Flag–syd3–239 were generated by inserting their coding sequence in‐frame downstream of sequences encoding a Flag tag in the pcDNA3 vector. A PCR‐based overlap extension method (Pérez‐Pinera, 2006) was used to construct the in‐frame deletion mutants N‐sydΔLZ (Δ429–459), N‐sydΔKBD (Δ50–82), N‐sydΔΔ (Δ50–82 and Δ429–459), N‐sydΔJBD (Δ200–218), N‐sydΔ266–315, N‐sydΔ384–424, GFP–sydΔLZ, GFP–sydΔKBD, and GFP–sydΔΔ. All plasmids were sequenced and confirmed.

GST and His₆ fusion constructs were expressed in E. coli BL21 cells. For GST purification, following IPTG (0.5 mM) induction of protein expression, cells were harvested, resuspended in ice‐cold PBS with 5.0 mM DTT and 1.0 mM AEBSF, and lysed with sonication. After addition of Triton X‐100 (0.2%), the lysates were mixed at 4°C for 40 min, and the insoluble material was removed by centrifugation at 16 000 g for 30 min at 4°C. The clear lysates were then mixed with Glutathione Sepharose 4B beads (GE Healthcare) for 30 min at room temperature. The beads were washed with PBS three times, and GST fusion proteins were eluted with 20 mM reduced glutathione in Tris buffer (50 mM, pH 8.0). His₆‐tagged proteins were purified using PrepEase His‐tag resin (USB Corporation, Cleveland, OH). Briefly, cells were resuspended in LEW buffer (NaH₂PO₄ 50 mM, NaCl 300 mM, AEBSF 1.0 mM, pH 8.0), sonicated and incubated with Triton X‐100 (0.2%) for 40 min at 4°C. The lysates were clarified by centrifugation at 16 000 g for 30 min at 4°C and then mixed with resin at room temperature for 30 min. After three washings with LEW buffer, the bound His₆‐tagged proteins are eluted with 250 mM imidazole in LEW buffer. The eluted recombinant proteins were dialysed in PBS or ultrafiltered for buffer exchange before their application.

GST pulldown. Mouse brains were homogenized in ice‐cold lysis buffer (Tris–HCl 20 mM, NaCl, 200 mM, EDTA 1 mM, 0.5% Nonidet P‐40, pH 8.0). The homogenate was centrifuged at 16 000 g for 20 min at 4°C, and the supernatant (1–2 mg) was incubated with Glutathione Sepharose 4B and equal molar amounts of purified GST fusion proteins or purified GST as negative control overnight at 4°C. The beads were collected by centrifugation at 500 g for 5 min at 4°C and washed three times with ice‐cold PBS. Bound material was eluted by boiling beads in sample buffer for 5 min and analysed by western blot.

Immunoprecipitation. Dynabeads^® Protein A Magnetic Beads (Invitrogen) were coated with anti‐GFP antibody or rabbit IgG as a negative control, or with anti‐myc with mouse IgG as a negative control. N2A cells were lysed in ice‐cold lysis buffer 24 h after transfection and the lysate was incubated with coated magnetic beads for 2 h at 4°C on a rotating wheel. Beads were then washed three times with ice‐cold PBS. The bound material was eluted by boiling beads in sample buffer for 5 min and analysed by western blot.

Protein binding assay. GST‐tagged proteins were incubated with Glutathione Sepharose 4B beads for 30 min at room temperature, washed with PBS, and incubated with His₆‐tagged proteins at 4°C on a rotating wheel for 90 min. Beads were washed three times with ice‐cold PBS and bound material was eluted by boiling beads in sample buffer for 5 min and analysed by western blot. This assay was used to determine the dissociation constant between syd and KLC1‐TPR and between syd and KHC tail, as previously described (Bowman et al, 2000). We used a fixed concentration of recombinant GST–syd N‐terminal (50 nM) and various concentrations of the ligands His–KLC‐TPR or His–KHC tail over a 3000‐fold range. The relative amount of bound KLC1‐TPR or KHC tail was analysed by western blot and quantified using ImageJ. The dissociation constants were measured by plotting the band intensity of bound kinesin for each concentration and using the first order binding equation: fraction of bound=[kinesin]/(Kd+[kinesin]) to fit the data.

Mouse brains were homogenized in PIPES buffer (50 mM PIPES, 1 mM EGTA, 1 mM MgSO₄, and protease inhibitors, pH 6.9). The homogenate was then centrifuged at 100 000 g for 30 min at 4°C to obtain a clear brain lysate. Glycerol (20%) and GTP (1 mM), taxol (20 μM) and AMP‐PNP (2.5 mM) were added to brain lysates and incubated for 30 min at 37°C. The lysates were centrifuged at 100 000 g for 30 min at 22°C to pellet down microtubules and the associated kinesin motor. The resulting pellet was rinsed once with PIPES buffer with AMP‐PNP (0.1 mM) and Taxol (20 μM). To release the motor from microtubules, the microtubule pellets were resuspended in ATP‐containing buffer (100 mM PIPES, 1 mM EGTA, 1 mM MgSO₄, and ATP 20 mM, pH 6.9), and incubated for 1 h on ice. The re‐suspension was then centrifuged as described above to collect the supernatant. The supernatant containing released motor proteins was loaded onto a sucrose gradient (5–50% in PIPES buffer), and centrifuged with a SW41 swinging bucket rotor for 14 h at 200 000 g at 4°C. Fractions were collected for western blot analysis with KHC and KLC antibodies.

Twenty‐four hours after transfection with the indicated constructs, COS cells were lysed with PIPES buffer with the addition of 1 mM EGTA and 0.1% Triton X‐100. The supernatant was collected by centrifugation at 16 000 g for 15 min at 4°C. The in vitro motility assay was conducted as previously described (Dixit et al, 2008). Briefly, flow chambers were assembled from a glass slide and silanized coverslip attached using double‐sided adhesive tape (chamber volume ∼15 μl). In all, 250 nM microtubules were introduced in a flow chamber coated with 0.5% monoclonal anti‐β tubulin antibody (Tub 2.1 clone, Sigma, St Louis, MO) and then blocked with 5% Pluronic F‐127 (Sigma). COS cell extracts supplemented with 2 mM ATP, an oxygen scavenging system containing glucose oxidase, catalase, and glucose, and 10 mM DTT were subsequently flowed into the chamber. Single syd–GFP or KHC–mCit molecules were visualized at 25°C using TIRF microscopy outfitted on an inverted microscope (Olympus IX81). TIRF excitation was achieved using 488 and 532 nm diode‐pumped solid‐state lasers (Melles Griot) to visualize GFP and rhodamine, respectively. Low power laser (2 mW) allowed to track fluorescent particles over long distances. Images were captured with a back‐thinned electron multiplier‐CCD camera (ImagEM, Hamamatsu) at 1 s intervals. Non‐transfected cell lysates were used for control experiments in parallel on the same day. The motion of GFP–syd or KHC–mCit was analysed using kymographs generated with Slidebook (Intelligent Imaging Innovations). Vertical lines in kymographs were defined as non‐motile events, while diagonal lines were defined as motile events. Only lines spanning more than two frames were counted as events. For non‐motile events, only vertical lines with a fluorescence intensity similar to motile, diagonal lines were counted, to avoid inclusion of aggregates bound to microtubules.

N2A and COS‐7 cells were grown in DMEM medium supplemented with 10% FBS, 1 mM sodium pyruvate, 2 mM l‐glutamate, 0.1 mM non‐essential amino acids and 1% Pen/Strep. Cells were grown on T25 flasks overnight before transfection using Lipofectamine 2000 (Invitrogen). Primary hippocampal and cortical cultures were prepared from mouse embryos at embryonic day 18 (E18). Hippocampi or cortices were treated with papain and DNase for 30 min, and triturated in neurobasal medium with 0.1% FBS. Dissociated neurons were cultured on coverslips coated with poly‐d‐lysine in Neurobasal medium containing 2% B27, 0.5 mM l‐Glutamax and 1% Pen/Strep. About 5–6 × 10⁴ neurons were plated per well in 24‐well plates. Amaxa^® Nucleofection^® was used to transfect neurons before plating. After 5–6 days in culture, neurons were fixed in 4% paraformaldehyde in PBS with 4% sucrose for 10 min. Neurons were permeabilized and blocked with 10% goat serum, 0.1% Triton X‐100 in PBS and incubated with the indicated primary antibodies and Alexa Fluor‐conjugated secondary antibodies. Nuclei were stained by DAPI in the Prolong anti‐fade mounting medium (Invitrogen). The images were acquired with fluorescence microscopy (Nikon, Eclipse TE 2000‐E) and analysed with Nis‐Element software.

Statistical analyses were performed using Student's t‐test, except that χ² analysis was used to determine the statistical differences in motile frequency. For the motility experiments, the Kolmogorov–Smirnov test was used to test for normality of distribution. If the distribution was not normal, then statistical analysis was performed with the Kolmogorov–Smirnov test.

Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).