A Direct Interaction between Cytoplasmic Dynein and Kinesin I May Coordinate Motor ActivityLee A. Ligon, Mariko Tokito, Jeffrey Finklestein et al.|Journal of Biological Chemistry|2004 Cytoplasmic dynein and kinesin I are both unidirectional intracellular motors. Dynein moves cargo toward the cell center, and kinesin moves cargo toward the cell periphery. There is growing evidence that bi-directional motility is regulated in the cell, potentially through direct interactions between oppositely oriented motors. We have identified a direct interaction between cytoplasmic dynein and kinesin I. Using the yeast two-hybrid assay and affinity chromatography, we demonstrate that the intermediate chain of dynein binds to kinesin light chains 1 and 2. The interaction is both direct and specific. Co-immunoprecipitation experiments demonstrate an interaction between endogenous proteins in rat brain cytosol. Double-label immunocytochemistry reveals a partial co-localization of vesicle-associated motor proteins. Together these observations suggest that soluble motors can interact, potentially allowing kinesin I to actively localize dynein to cellular sites of function. There is also a vesicle population with both dynein and kinesin I bound that may be capable of bi-directional motility along cellular microtubules. Cytoplasmic dynein and kinesin I are both unidirectional intracellular motors. Dynein moves cargo toward the cell center, and kinesin moves cargo toward the cell periphery. There is growing evidence that bi-directional motility is regulated in the cell, potentially through direct interactions between oppositely oriented motors. We have identified a direct interaction between cytoplasmic dynein and kinesin I. Using the yeast two-hybrid assay and affinity chromatography, we demonstrate that the intermediate chain of dynein binds to kinesin light chains 1 and 2. The interaction is both direct and specific. Co-immunoprecipitation experiments demonstrate an interaction between endogenous proteins in rat brain cytosol. Double-label immunocytochemistry reveals a partial co-localization of vesicle-associated motor proteins. Together these observations suggest that soluble motors can interact, potentially allowing kinesin I to actively localize dynein to cellular sites of function. There is also a vesicle population with both dynein and kinesin I bound that may be capable of bi-directional motility along cellular microtubules. Intracellular transport along microtubules is driven by the motors kinesin and cytoplasmic dynein. Kinesin I is a plus end-directed motor, consisting of two heavy chains and two light chains (1Vale R.D. Cell. 2003; 112: 467-480Google Scholar). Cytoplasmic dynein is a minus end-directed motor, composed of two globular heads each formed from a single dynein heavy chain and a base formed from intermediate, light intermediate, and light chains (1Vale R.D. Cell. 2003; 112: 467-480Google Scholar). The accessory complex dynactin binds to dynein and may be required to increase the processivity of dynein-driven movement (2Karki S. Holzbaur E.L.F. J. Biol. Chem. 1995; 270: 28806-28811Google Scholar, 3King S.J. Schroer T.A. Nat. Cell Biol. 2000; 2: 20-24Google Scholar). In the cell, kinesin powers anterograde transport, moving vesicles and organelles toward the cell periphery. Cytoplasmic dynein drives the retrograde transport of organelles and proteins toward microtubule minus ends. In addition to its role in increasing the processivity of dynein-driven motility, dynactin has been shown to mediate interactions of cytoplasmic dynein with some of its intracellular cargos (4Holleran E.A. Ligon L.A. Tokito M.K. Stankewich M.K. Morrow J.S. Holzbaur E.L.F. J. Biol. Chem. 2001; 276: 36598-36605Google Scholar, 5Muresan V. Stankewich M.C. Steffen W. Morrow J.S. Holzbaur E.L.F. Schnapp B.J. Mol. Cell. 2001; 7: 173-183Google Scholar). The mechanisms of cargo coupling to kinesin are yet to be fully understood, but at least some cargos have been shown to bind directly to kinesin light chains (6Hollenbeck P.J. J. Cell Biol. 2001; 152: 25-28Google Scholar). Although in vitro motility assays clearly indicate that kinesin and cytoplasmic dynein can function independently to produce motility in opposite directions along microtubules, multiple studies have suggested that the activities of these two motors are coordinately coupled in the cell. This coordination may be most clear in axonal transport. Studies in extruded squid axoplasm have indicated that the specific inhibition of either kinesin (7Brady S.T. Pfister K.K Bloom G.S. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 1061-1065Google Scholar, 8Stenoien D.L. Brady S.T. Mol. Biol. Cell. 1997; 8: 675-689Google Scholar) or dynein/dynactin (9Waterman-Storer C.M. Kuznetsov S.A. Karki S. Tabb J.S. Weiss D.G. Langford G.M. Holzbaur E.L.F. Proc. Natl. Acad. Sci. U. S. A. 1997; 94: 12180-12185Google Scholar) results in a bidirectional block in the transport of vesicles along microtubules. Interdependence of anterograde and retrograde motors has also been noted genetically. Martin et al. (10Martin A.A. Iyaduri S.J. Gassman A. Gindhart J.G. Hays T.S. Saxton W.M. Mol. Biol. Cell. 1999; 10: 3717-3728Google Scholar) identified dominant genetic interactions between the kinesin, cytoplasmic dynein, and dynactin genes in Drosophila. Mutations in either motor result in an inhibition of axonal transport and the accumulation of organelles in axonal swellings that stain for markers of both anterograde and retrograde motility. Here we provide biochemical evidence for a direct interaction between conventional kinesin and cytoplasmic dynein, mediated by an interaction between dynein intermediate chain (DIC) 1The abbreviations used are: DIC, dynein intermediate chain; KLC, kinesin light chain; GST, glutathione S-transferase; BSA, bovine serum albumin; PIPES, 1,4-piperazinediethanesulfonic acid; TPR, tetratricopeptide repeat; MgAMP-PNP, Mg-5′-adenylyl imidodiphosphate; KHC, kinesin heavy chain. and kinesin light chains (KLCs). Immunoprecipitation studies indicate that this interaction is physiologically relevant. Although the interaction is most robust between soluble motors, co-localization studies indicate that some vesicles have both motors bound, allowing for bi-directional motility. The results reported here provide further evidence for a key role of KLCs in mediating interactions between motors and cargo in the cell and suggest that cytoplasmic dynein may represent one of the cellular cargos of kinesin I. Yeast Two-hybrid Interaction Screen—We used the LexA yeast two-hybrid system to screen an oligo(T)-primed human fetal brain library (Stratagene) for proteins interacting with DIC. The majority of clones isolated were found to encode p150Glued, so we further screened positives by Southern blot to identify novel clones, one of which was identified by data base searching as encoding a TPR protein. Binding and domain mapping studies between kinesin and dynein were performed by subcloning full-length or partial sequences, as noted, of DIC, KLC1, and KLC2 into either the pB42Ad or pJK202 vectors (Stratagene) and scoring for growth on medium lacking adenine in comparison with the appropriate empty vector, as previously described (4Holleran E.A. Ligon L.A. Tokito M.K. Stankewich M.K. Morrow J.S. Holzbaur E.L.F. J. Biol. Chem. 2001; 276: 36598-36605Google Scholar). Affinity Chromatography and Immunoprecipitations—Mouse cDNA clones encoding kinesin light chains 1 and 2 were used to generate GST fusion constructs, which were expressed in Escherichia coli and purified (4Holleran E.A. Ligon L.A. Tokito M.K. Stankewich M.K. Morrow J.S. Holzbaur E.L.F. J. Biol. Chem. 2001; 276: 36598-36605Google Scholar). Purified recombinant DIC or rat brain cytosolic extracts were fractionated over affinity columns generated from either GST fusion proteins or GST alone bound to glutathione-Sepharose, followed by elution of bound proteins with glutathione (4Holleran E.A. Ligon L.A. Tokito M.K. Stankewich M.K. Morrow J.S. Holzbaur E.L.F. J. Biol. Chem. 2001; 276: 36598-36605Google Scholar). The binding of kinesin to DIC was examined by fractionating either purified recombinant KLCs, in vitro translated truncation constructs, or rat brain cytosol over affinity columns generated from either purified recombinant DIC or BSA covalently linked to a Sepharose matrix, followed by elution of bound proteins with 1 m NaCl (2Karki S. Holzbaur E.L.F. J. Biol. Chem. 1995; 270: 28806-28811Google Scholar). Western blots were probed with antibodies to: kinesin heavy chain (Chemicon monoclonal antibodies 1613 and 1614), KLCs 1 and 2 (Chemicon monoclonal antibodies 1616 and 1617 and monoclonal antibody 63-90 (generously provided by Dr. Scott Brady of the University of Illinois at Chicago), cytoplasmic dynein (monoclonal antibody MAB1618 to DIC from Chemicon, affinity-purified polyclonal antibodies UP1467 and UP1468 to DIC generated in our laboratory, and polyclonal antibodies to heavy chain and the Tctex-1 light chain generously provided by Drs. Richard Vallee of Columbia University and Stephen King of the University of Connecticut), and dynactin (affinity-purified polyclonal antibody UP235 to p150Glued generated in our laboratory). Immunoprecipitations were performed from rat brain cytosol (the supernatant from a 100,000 × g centrifugation of rat brain homogenate) in 50 mm PIPES, 50 mm HEPES, 1 mm EDTA, 2 mm MgSO4, pH 7.0, in the absence or the presence of 1% Triton X-100 and 0.5% Igepal. Immunoprecipitations were also performed from the 100,000 × g pellet resuspended in 50 mm Tris, 50 mm KCl, with 1% Triton X-100, and 0.5% Igepal. Immunoprecipitates were washed eight times in 50 mm Tris, 50 mm KCl with 1% Triton X-100 (4Holleran E.A. Ligon L.A. Tokito M.K. Stankewich M.K. Morrow J.S. Holzbaur E.L.F. J. Biol. Chem. 2001; 276: 36598-36605Google Scholar). Control immunoprecipitations were performed in parallel with protein A beads alone. Further purification of rat brain vesicles was performed by fractionating the membranes from the 100,000 × g pellet by flotation upward though a sucrose step gradient, as described (4Holleran E.A. Ligon L.A. Tokito M.K. Stankewich M.K. Morrow J.S. Holzbaur E.L.F. J. Biol. Chem. 2001; 276: 36598-36605Google Scholar). Soluble proteins were retained in the bottom layer of 2 m sucrose, and the vesicles were isolated from the 1.5/0.6 m sucrose interface. The vesicles were solubilized in 1% Triton X-100 for immunoprecipitation experiments. The vesicles purified by flotation were incubated with microtubules polymerized from purified tubulin (Cytoskeleton) in the absence or presence of 5 mm MgATP or 10 mm MgAMP-PNP and then centrifuged through a cushion of 40% sucrose at 8,000 × g. Gel samples made from the supernatant and the pellet fractions were analyzed by SDS-PAGE and Western blot. Immunocytochemistry and Immunofluorescence Microscopy—Cytoplasmic dynein and kinesin were localized in cultured PtK2, Rat2, and Cos7 cells by immunocytochemistry (11Ligon L.A. Karki S. Tokito M.K. Holzbaur E.L.F. Nat. Cell Biol. 2001; 3: 913-917Google Scholar), using all the antibodies to kinesin and dynein described above in all possible combinations. The cells were counterstained with antibodies to tubulin (clone DM1A from Sigma or YL1/2 from Serotec) and visualized with Alexa 350-, 488-, and 594-conjugated secondary antibodies (Molecular Probes). The vesicles purified by flotation were bound to taxol-stabilized microtubules, fixed with glutaraldehyde, pelleted onto poly-l-lysine-coated coverslips (12Ligon L.A. Shelly S.S. Tokito M. Holzbaur E.L.F. Mol. Biol. Cell. 2003; 14: 1405-1417Google Scholar), and processed for immunocytochemistry with antibodies to kinesin, dynein, and tubulin. The images were acquired on a Leica DMIRBE microscope with a 63× or 100× Planapo objective using OpenLab software (Improvision) and an Orca ER CCD camera (Hamamatsu). We performed a yeast two-hybrid screen for proteins interacting with DIC and isolated a clone encoding TPR1, a human protein with three tetratricopeptide repeats (TPR). The TPR motif is a 34-residue degenerate sequence that forms two antiparallel α-helices; tandem arrays of TPRs form an amphipathic channel that mediates protein-protein interactions (13Blatch G.L. Lassle M. Bioessays. 1999; 21: 932-939Google Scholar). The function of TPR1 is not yet known; therefore, the significance of this interaction is not clear. However, the demonstration of an interaction between DIC and a TPR protein, coupled with previous observations of functional and genetic interactions between dynein and kinesin, led us to test for an interaction between DIC and kinesin light chains. KLC1 and KLC2 each have six TPR motifs that are involved in motorcargo interactions (6Hollenbeck P.J. J. Cell Biol. 2001; 152: 25-28Google Scholar). KLC1 is a 61-kDa protein primarily expressed in neuronal tissues, and KLC2 is a 67-kDa protein that is more ubiquitously expressed (14Rahman A. Friedman D.S. Goldstein L.S. J. Biol. Chem. 1998; 273: 15395-15403Google Scholar). We first tested for interactions between DIC and the kinesin light chains using the yeast two-hybrid assay. Both KLC1 and KLC2 were observed to interact with DIC in this assay, but the interaction between DIC and KLC2 was qualitatively more robust (Fig. 1A). To further investigate the interaction, we generated GST-KLC1 and GST-KLC2 fusion proteins and used affinity chromatography to probe for the binding of DIC. Recombinant DIC bound to both the KLC1 and KLC2 affinity columns but not the control column (Fig. 1, B and C), suggesting that the dynein intermediate chain can bind directly to kinesin light chains 1 and 2. In the reciprocal experiment, KLC2 bound to the DIC column and not to a control BSA column. In contrast, KLC1 did not bind significantly to the DIC column (data not shown). This observation supports the qualitative analysis of the yeast two-hybrid assay, suggesting that KLC2 binds more strongly to DIC than does KLC1. We mapped the binding site for KLCs within the dynein intermediate chain polypeptide using the yeast two-hybrid assay to test constructs spanning residues 1-120, 1-283, 120-283, and 283-644 of DIC (Fig. 1D). No significant interaction was observed between KLC1 or KLC2 and the DIC fragment 283-644, indicating that the association between these polypeptides does not involve the C-terminal WD40 repeat region of DIC. The DIC constructs spanning residues 1-283 or 120-283 were consistently most positive in this assay, suggesting that the primary KLC-binding site in dynein spans residues 120-283. Residues 1-120 of DIC did not interact significantly with KLC1 but did demonstrate consistent interactions with KLC2, suggesting that there might be a secondary site of association between dynein and KLC2 that may explain the apparently higher affinity of DIC for this isoform. The inverse experiment was also performed to map the binding site for dynein within the KLC polypeptides. Affinity chromatography experiments revealed that the C-terminal TPR repeat domain of KLC2 bound significantly to a DIC affinity column and not to a BSA control column (Fig. 1E). However, we also noted binding of the N-terminal domain of KLC2 to a DIC column but not to a BSA control column (Fig. 1E). A similar result was observed with KLC1. These observations suggest that the interaction of DIC with kinesin light chains is not limited to the TPR motifs of the light chains but also involves determinants in the N terminus. These in vitro binding studies with purified proteins indicate that DIC can bind directly to kinesin light chains. To test whether these interactions occur with endogenous proteins, we loaded a rat brain cytosolic extract onto a DIC affinity column. Although both KLC1 and KLC2 were present in the cytosolic extract, only KLC2 was retained on the DIC affinity column (Fig. 2A). No significant binding of either KLC1 or KLC2 to a BSA control column was observed. Kinesin heavy chain was also specifically retained by the DIC affinity column (Fig. 2A), indicating that intact kinesin interacts with DIC. We then performed the reciprocal experiment and fractionated rat brain cytosolic extract over GST-KLC1, GST-KLC2, or GST control columns. Dynein intermediate chain, dynein heavy chain, and a dynein light chain (Tctex-1) were all specifically retained by both the KLC1 and KLC2 columns but not by the control GST column (Fig. 2, B and C), indicating that the intact dynein complex interacts with kinesin light chains. Dynactin was also retained on both the KLC1 and KLC2 columns (Fig. 2C); however, dynactin binds to dynein (2Karki S. Holzbaur E.L.F. J. Biol. Chem. 1995; 270: 28806-28811Google Scholar), and thus the association with kinesin may be indirect. These experiments indicate that the interaction between kinesin and dynein is both direct, because binding is observed between purified recombinant proteins, and specific, because dynein is specifically retained from brain cytosol by a KLC affinity column, and kinesin is specifically retained on a DIC affinity column. To test the physiological relevance of the association, we looked for the co-immunoprecipitation of cytoplasmic dynein and kinesin from rat brain cytosol. We observed that a fraction of cytoplasmic dynein was co-precipitated using two different monoclonal antibodies to KHC (Fig. 3A and data not shown). We also observed the co-immunoprecipitation of dynein and kinesin from cytosol using two independent monoclonal antibodies to KLC (Fig. 3B, left panel, and data not shown). Most of the dynein did not co-precipitate with kinesin, indicating that only a fraction of the total cytosolic pools of the motors are associated under these conditions. We observed only a very limited co-immunoprecipitation of dynactin, not significantly above background (Fig. 3B). There are significant pools of both soluble and vesicle-associated kinesin and dynein in the cell. Therefore, we probed for an interaction between cytoplasmic dynein and kinesin in a vesicle-enriched fraction from rat brain. As shown in Fig. 3B, immunoprecipitation of kinesin from fractions using a monoclonal antibody to KLC led to the robust co-immunoprecipitation of kinesin heavy and light chains as (Fig. 3B, In contrast, dynein dynactin was consistently observed to with kinesin from Control experiments indicate that the co-immunoprecipitation of dynein and kinesin from the soluble fraction was not by the of required to proteins from the We further purified fractions by flotation upward through a sucrose step gradient, and immunoprecipitations were performed using monoclonal antibodies to either kinesin heavy chain or light chain. We did not significant co-immunoprecipitation of dynein from these fractions (Fig. and data not shown). These immunoprecipitation data indicate that the interaction between dynein and kinesin in the cell. This interaction may be more robust between soluble than vesicle-associated motor proteins or may be the of vesicles to To on these biochemical we performed immunocytochemistry in and cell using a of monoclonal and polyclonal antibodies to dynein and Both motors are in the and the of co-localization is to in this of the cell. the cell organelles and vesicles can more be we partial co-localization of the two motor proteins to (Fig. However, of the vesicles and organelles to be with either kinesin or dynein similar to previous observations with a more limited of antibodies Pfister Cell Scholar). we noted of these vesicles along microtubules (Fig. We also observed partial co-localization of the two motors at of in cells (Fig. Dynein is localized to in cells (11Ligon L.A. Karki S. Tokito M.K. Holzbaur E.L.F. Nat. Cell Biol. 2001; 3: 913-917Google we kinesin these dynein This of kinesin to sites of may a role for the motor in the of S. S. M. A. Mol. Biol. Cell. Scholar). To further the interaction of vesicle-associated dynein and kinesin, we incubated vesicles with microtubules in the absence or presence of either MgATP or We isolated the vesicles by and analyzed the results by both and Both dynein and kinesin with microtubules the fractions were incubated with microtubules. The of was in the presence of indicating that the observed were the the of motors that with microtubules in this assay (Fig. Immunofluorescence analysis that vesicles were clearly with antibodies to either dynein or kinesin (Fig. However, of the vesicles were with antibodies to both motors (Fig. and as as vesicles with both motors bound were all observed to with microtubules. of vesicles with microtubules was more in the presence of and the addition of MgATP significant These studies are consistent with results from indicating that the most robust association between dynein and kinesin in the soluble of motors, there is a population of vesicles with both motors these observations identify a direct interaction between cytoplasmic dynein and kinesin, mediated by kinesin light chains. Although light chains may not be for all some interactions to be mediated by direct binding to KLCs A.A. J. Cell Sci. 2000; Scholar, A. A.A. Mol. Biol. Cell. 1998; Scholar), and the TPR motifs of KLCs have been shown to interact with proteins as protein and of the N-terminal A. Goldstein L.S. 2000; in P.J. J. Cell Biol. 2001; 152: 25-28Google Scholar). results suggest that kinesin I may also interact with a novel the motor cytoplasmic dynein, Although both KLCs can bind to DIC, the most interaction was between KLC2 and DIC. Although the of KLC1 and KLC2 have yet to be fully studies have suggested that KLCs are involved in the of interactions in the cell. There are two possible for the role of these direct interactions between plus and minus end-directed motor The first is that the interaction a transport The most robust interaction between dynein and kinesin was observed in the soluble pools of these motors, suggesting that one motor as a cargo for the Kinesin may be required to transport dynein to the of the cell, the of the or to the plus of the axonal transport is by kinesin and dynein both and consistent with the of the motors, as as with the that dynein be the to transport to the cell Pfister Bloom G.S. Brady S.T. J. Cell Biol. Scholar, J. Cell Biol. 1990; Scholar). dynein was observed to in in in kinesin light chain in J.G. S. Goldstein L.S. J. Cell Biol. 1998; Scholar). transport of dynein may also explain the kinesin of the robust of dynein to microtubule plus observed in J. S. Mol. Biol. Cell. 2003; 14: Scholar), the of kinesin chains in this that interaction mechanisms may be However, we also noted some co-localization of cytoplasmic dynein and kinesin to suggesting that there may be motor that bi-directional motility. The direct interaction between kinesin I and cytoplasmic dynein that we have described may thus be similar to the observation that kinesin can bind to dynactin through its to the that dynactin may the bi-directional motility of M. I. J. Cell Biol. 2003; Scholar). Although I and are cellular motors, is that in both specific interactions with the minus end-directed motor dynein have been in one a direct interaction with dynein and in the an interaction with dynactin M. I. J. Cell Biol. 2003; Scholar). The interaction we have observed between kinesin I and dynein may also have a Kinesin light chains have been reported to kinesin motor an form of the motor in which the of kinesin and the motor domain in the D.L. Schnapp B.J. T.A. J. Cell Biol. 1998; Scholar). Binding of light chains to cargo may a that this be of to whether the binding of dynein to the kinesin light chain a similar from The binding of kinesin to dynein may also have some on dynein motor function. a provide for a regulated interaction between oppositely oriented motors that for the transport of cytoplasmic dynein the in its of dynein at the might from kinesin, dynein to the motility of cargo to the cell Together the direct coupling of the motors for the of dynein to its site of which is an in cells with axonal that can a in The demonstration of biochemical interactions between oppositely oriented motors a for observed and for further We the of Shelly and of the University of We also Scott Richard and Stephen provided antibodies to kinesin and dynein, and provided cDNA clones of KLCs 1 and 2.