Copurification of kinesin polypeptides with microtubule-stimulated Mg-ATPase activity and kinetic analysis of enzymatic properties

Determination of kinetic properties for kinesin adenosine triphosphatase (ATPase), a proposed motor for transport of membranous organelles, requires adequate amounts of kinesin with a consistent level of enzymatic activity. A purification procedure is detailed that produces approximately 2 mg of kinesin at up to 96% purity from 800 g of bovine brain. This protocol consists of a microtubule affinity step using 5'-adenylylimidodiphosphate (AMP-PNP); followed by gel filtration, ion exchange, and hydroxylapatite chromatography; and then sucrose density gradient centrifugation. The microtubule-activated ATPase activity of kinesin coeluted with kinesin polypeptides throughout the purification. Highly purified kinesin had a Vmax of 0.31 mumol/min/mg in the presence of microtubules, with a Km for ATP of 0.20 mM. The kinetic constants obtained in these studies compare favorably with physiological levels of ATP and microtubules. Variations in buffer conditions for the assay were found to affect ATPase activity significantly. A study of the ability of kinesin to utilize a variety of cation-ATP complexes indicated that kinesin is a microtubule-stimulated Mg-ATPase, but kinesin is able to hydrolyze Ca-ATP, Mn-ATP, and Co-ATP as well as Mg-ATP in the presence of microtubules. In the absence of microtubules, Ca-ATP appears to be the best substrate. Studies with several inhibitors of ATPases determined that vanadate inhibited kinesin ATPase at the lowest concentrations of inhibitor, but significant inhibition of the ATPase also occurred with submillimolar concentrations of AMP-PNP. Other inhibitors of kinesin include N-ethylmaleimide, adenosine diphosphate (ADP), pyrophosphate, and tripolyphosphate. Further characterization of the kinetic properties of the kinesin ATPase is important for understanding the molecular mechanisms for transport of membranous organelles along microtubules.     

The quaternary structure of bovine brain kinesin.

In the present work we have studied the subunit composition of kinesin, the microtubule-activated, mechanochemical ATPase, isolated from bovine brain. Polypeptides with mol. wts of 120 and 62 kd are the major components of the kinesin preparation. These polypeptides could not be separated by electrophoresis under nondenaturing conditions or by FPLC on a MonoQ column, and are therefore assumed to form a tight complex. As shown by immunoblotting with polyclonal and monoclonal antibodies to the 120-kd polypeptide and by one-dimensional peptide mapping, the 62-kd polypeptide does not appear to be a proteolytic product of the 120-kd component. Densitometric scanning of polyacrylamide-SDS gels shows that these polypeptides are present in a complex in a 1:1 molar ratio. The mol. wt of native kinesin was studied by sedimentation equilibrium and was found to be 386 +/- 14 kd. A comparison of the mol. wts of individual polypeptides with the mol. wt of the intact molecule indicates that the native molecule contains two 120-kd subunits and two 62-kd subunits.     

Isolation of a 45-kDa fragment from the kinesin heavy chain with enhanced ATPase and microtubule-binding activities

Kinesin is a microtubule-activated, mechanochemical ATPase capable of moving particles along microtubules and making microtubules glide along a solid substrate. In this study we used limited proteolysis to study the structure of bovine brain kinesin, a heterotetramer composed of two heavy (120-kDa) and two light (62-kDa) chains. alpha-chymotrypsin, trypsin, and subtilisin all produced a protease-resistant 45-kDa fragment from the kinesin heavy chain. As isolated by gel-filtration chromatography, this fragment contains both the microtubule-binding site and the ATP catalytic site of the molecule. Proteolytic cleavage stimulated microtubule-dependent Mg2+-ATPase activity 4- to 5-fold up to 75-120 mumol ATP/min/mg. Cleavage also increased the affinity of the fragment for microtubules at least 10-fold. Since the purified fragment does not support the gliding of flagellar axonemes, we propose that cleavage of the heavy chain uncouples ATPase activity from its translocator activity, which may require other parts of the molecule.     

Submolecular domains of bovine brain kinesin identified by electron microscopy and monoclonal antibody decoration

Kinesin is a microtubule-activated ATPase thought to transport membrane-bounded organelles along MTs. To illuminate the structural basis for this function, EM was used to locate submolecular domains on bovine brain kinesin. Rotary shadowed kinesin appeared rod-shaped and approximately 80 nm long. One end of each molecule contained a pair of approximately 10 x 9 nm globular domains, while the opposite end was fan-shaped. Monoclonal antibodies against the approximately 124 kd heavy chains of kinesin decorated the globular structures, while those specific for the approximately 64 kd light chains labeled the fan-shaped end. Quick-freeze, deep-etch EM was used to analyze MTs polymerized from tubulin and cross-linked to latex microspheres by kinesin. Microspheres frequently attached to MTs by arm-like structures, 25-30 nm long. The MT attachment sites often appeared as one or two approximately 10 nm globular bulges. Morphologically similar cross-links were observed by quick-freeze, deep-etch EM between organelles and MTs in the neuronal cytoskeleton in vivo. These collective observations suggest that bovine brain kinesin binds to MTs by globular domains that contain the heavy chains, and that the attachment sites for organelles are at the opposite, fan-shaped end of kinesin, where the light chains are located.     

Identification and characterization of a novel microtubule-based motor associated with membranous organelles in tobacco pollen tubes

Pollen tube growth depends on the differential distribution of organelles and vesicles along the tube. The role of microtubules in organelle movement is uncertain, mainly because information at the molecular level is limited. In an effort to understand the molecular basis of microtubule-based movement, we isolated from tobacco pollen tubes polypeptides that cosediment with microtubules in an ATP-dependent manner. Major polypeptides released from microtubules by ATP (ATP-MAPs) had molecular masses of 90, 80, and 41 kD. Several findings indicate that the 90-kD ATP-MAP is a kinesin-related motor: binding of the polypeptide to microtubules was enhanced by the nonhydrolyzable ATP analog AMP-PNP; the 90-kD polypeptide reacted specifically with a peptide antibody directed against a highly conserved region in the motor domain of the kinesin superfamily; purified 90-kD ATP-MAP induced microtubules to glide in motility assays in vitro; and the 90-kD ATP-MAP cofractionated with microtubule-activated ATPase activity. Immunolocalization studies indicated that the 90-kD ATP-MAP binds to organelles associated with microtubules in the cortical region of the pollen tube. These findings suggest that the 90-kD ATP-MAP is a kinesin-related microtubule motor that moves organelles in the cortex of growing pollen tubes.     

花粉高尔基囊泡类动蛋白的鉴定

在萌发的烟草花粉管顶端,有囊泡状的颗粒被牛脑动蛋白重链的单克隆抗体所识别。用蔗糖密度梯度离心法从榛木花粉中分离得到高尔基囊泡,体外免疫胶体金处理后可被标记。ＳＤＳ－聚丙烯酰胺凝胶电泳和免疫印迹表明,分子量为１００ｋＤ的多肽大量存在于高尔基囊泡,此多肽可与动蛋白单克隆抗体进行特异性反应,证明花粉高尔基囊泡上有关动蛋白,其重链分子量为１００ｋＤ。     

Identification of Kinesin-Like Protein on Golgi Vesicles of Pollen

Kinesin-like protein was identified on Golgi vesicles of pollen. At the tip of pollen tube of Nicotiana alata, the vesicle-like particles were recognized by monoclonal antibody against the kinesin heavy chain from bovine brain (K71s23). The Glogi vesicles isolated from the pollens of Corylus avellana by discontinious sucrose gradient ultracentrifugation, could be recognized as antikinesin, based on immuno-gold labelling. Results from SDS-PAGE and western blot, showed that the 100 kD polypeptides on Golgi vesicles were the major polypeptides of kinesin-like protein.     

Mechanical and chemical properties of cysteine-modified kinesin molecules.

To probe the structural changes within kinesin molecules, we made the mutants of motor domains of two-headed kinesin (4-411 aa) in which either all the five cysteines or all except Cys45 were mutated. A residual cysteine (Cys45) of the kinesin mutant was labeled with an environment-sensitive fluorescent probe, acrylodan. ATPase activity, mechanical properties, and fluorescence intensity of the mutants were measured. Upon acrylodan-labeled kinesin binding to microtubules in the presence of 1 mM AMPPNP, the peak intensity was enhanced by 3.4-fold, indicating the structural change of the kinesin head by the binding. Substitution of cysteines decreased both the maximum microtubule-activated ATPase and the sliding velocity to the same extent. However, the maximum force and the step size were not affected; the force produced by a single molecule was 6-6.5 pN, and a step size due to the hydrolysis of one ATP molecule by kinesin molecules was about 10 nm for all kinesins. This step size was close to a unitary step size of 8 nm. Thus, the mechanical events of kinesin are tightly coupled with the chemical events.     

Motor domain mutation traps kinesin as a microtubule rigor complex

Conventional kinesin is a highly processive, microtubule-based motor protein that drives the movement of membranous organelles in neurons. Using in vivo genetics in Drosophila melanogaster, Glu164 was identified as an amino acid critical for kinesin function [Brendza, K. M., Rose, D. J., Gilbert, S. P., and Saxton, W. M. (1999) J. Biol. Chem. 274, 31506-31514]. Glu164 is located at the beta-strand 5a/loop 8b junction of the catalytic core and projects toward the microtubule binding face in close proximity to key residues on beta-tubulin helix alpha12. Substitution of Glu(164) with alanine (E164A) results in a dimeric kinesin with a dramatic reduction in the microtubule-activated steady-state ATPase (5 s(-1) per site versus 22 s(-1) per site for wild-type). Our analysis shows that E164A binds ATP and microtubules with a higher affinity than wild-type kinesin. The rapid quench and stopped-flow results provide evidence that ATP hydrolysis is significantly faster and the precise coordination between the motor domains is disrupted. The data reveal an E164A intermediate that is stalled on the microtubule and cannot bind and hydrolyze ATP at the second head.     

The stator complex of the A1A0-ATP synthase--structural characterization of the E and H subunits

Archaeal ATP synthase (A-ATPase) is the functional homolog to the ATP synthase found in bacteria, mitochondria and chloroplasts, but the enzyme is structurally more related to the proton-pumping vacuolar ATPase found in the endomembrane system of eukaryotes. We have cloned, overexpressed and characterized the stator-forming subunits E and H of the A-ATPase from the thermoacidophilic Archaeon, Thermoplasma acidophilum. Size exclusion chromatography, CD, matrix-assisted laser desorption ionization time-of-flight mass spectrometry and NMR spectroscopic experiments indicate that both polypeptides have a tendency to form dimers and higher oligomers in solution. However, when expressed together or reconstituted, the two individual polypeptides interact with high affinity to form a stable heterodimer. Analyses by gel filtration chromatography and analytical ultracentrifugation show the heterodimer to have an elongated shape, and the preparation to be monodisperse. Thermal denaturation analyses by CD and differential scanning calorimetry revealed the more cooperative unfolding transitions of the heterodimer in comparison to those of the individual polypeptides. The data are consistent with the EH heterodimer forming the peripheral stalk(s) in the A-ATPase in a fashion analogous to that of the related vacuolar ATPase.     

Purification and characterization of DNA-dependent ATP phosphohydrolases from KB cells

DNA-dependent ATPases have been purified from logarithmically growing KB cells by chromatography on single-stranded DNA cellulose and phosphocellulose. Phosphocellulose resolved the DNA-dependent ATPases into three activities designated ATPase I, II and III, respectively. From gel filtration and sedimentation analysis ATPases II and III were found to be very similar, both with calculated molecular weights of 78,000. Due to the extreme lability these enzymes were not purified further. The molecular weight of ATPase I determined by gel filtration and sedimentation analysis was calculated to be 140,000. ATPase I was further purified by gradient elution on ATP-agarose, revealing two peaks of activity (IA and IB), and by sucrose gradient sedimentation. Analysis of the fractions from the sucrose gradient by sodium dodecylsulphate gel electrophoresis revealed only one broad polypeptide band co-sedimenting with both ATPase IA and ATPase IB. This band was composed of four closely spaced polypeptides with apparent molecular weights of 66,000, 68,000, 70,000 and 71,000. Comparison of the native molecule weight (140,000) with these results suggests that ATPase I is a dimer. ATPase IA and IB were indistinguishable in their structural and enzymatic properties and presumably represent the same enzyme. The purified enzyme has an apparent Km of 0.5 mM for ATP producing ADP + Pi. A maximum activity of 2,100 molecules of ATP hydrolyzed per enzyme molecular per minute was found. Hydrolysis of ATP requires the presence of divalent cations (Mg2+ greater than Ca2+ greater than Mn2+ greater than Co2+). A broad pH optimum (pH 6--8) was observed. The enzyme uses ATP or dATP preferentially as a substrate, while other deoxyribonucleoside or ribonucleoside triphosphates were inactive. ATPase I prefers denatured DNA as cofactor. The activity with native DNA is 40% of that with denatured DNA.     

Purification and characterization of tonoplast ATPase from etiolated mung bean seedlings

The tonoplast ATPase from etiolated seedlings of Vigna radiata L. (mung bean) was isolated using a two-step detergent solubilization modified from Mandala and Taiz (S Mandala, L Taiz [1985] Plant Physiol 78: 327-333). After ultracentrifugation on 10 to 28% sucrose gradient, the ATPase showed a 31.6-fold purification over the initial specific activity of the starting tonoplast-enriched membranes. The purified ATPase used Mg²⁺-ATP as the preferred substrate. The tonoplast ATPase was isolated in a form with characteristics similar to that on its native membrane environment. Analysis by SDS-PAGE revealed two prominent bands with molecular weights of 78,000 (α subunit) and 64,000 (β subunit). The intensity of Coomassie blue staining showed a 1:1 stoichiometry for α and β subunits. The amino acid composition of α and β subunits also confirmed the suggested stoichiometry of the subunit composition of the tonoplast ATPase. Moreover, radiation inactivation analysis yielded a functional size of 414 ± 24 and 405 ± 25 kilodaltons for soluble and membrane bound tonoplast ATPases, respectively. It is possible that the functioning tonoplast ATPase may be in a form of αβ-heteromultimer.     

Homology of egg and flagellar dynein. Comparison of ATP-binding sites and primary structure

Unfertilized sea urchin eggs contain a Mg2+-ATPase which shares physical and enzymatic characteristics with dynein, the enzyme which powers ciliary and flagellar movement. To further investigate the homology of the egg ATPase and axonemal dynein, ATP-binding subunits in preparations of each of the enzymes were identified using a photoaffinity probe of ATP, 8-azido-ATP (8-N3ATP), and three high molecular weight (HMW) polypeptide components of the two enzymes were compared by one-dimensional peptide mapping. Two heavy chains (A and B) of both the flagellar and egg ATPases bound [alpha-32P]8-N3ATP. The labeling of the HMW bands was specifically inhibited by ATP or ADP. Both the cytoplasmic ATPase and flagellar dynein utilized 8-N3ATP as a substrate indicating that the reagent binds to the active site. The two HMW ATP-binding polypeptides and one other HMW component of the egg ATPase were compared to flagellar dynein heavy chains by peptide mapping. Digestion of the egg versus flagellar HMW polypeptides with Staphylococcus V8 protease or alpha-chymotrypsin produced a highly similar group of peptides, and each pair of heavy chains was qualitatively estimated to be over 85% homologous. These data support the identification of the egg ATPase heavy chains as components of a cytoplasmic dynein and suggest that the HMW polypeptides form active enzymatic sites in flagellar and egg dynein which are substantially homologous.     

Characterization of alpha 2 beta 2 and alpha 2 forms of kinesin

Bovine brain kinesin separates into two components on sucrose density gradient centrifugation. The predominant component is a heterotetramer of two 120 kDa alpha subunits and two 64 kDa beta subunits with an sedimentation coefficient of 9.6 S and a low Vm rate of microtubule-stimulated ATPase of 1.3 +/- 0.5 sec-1 at 25 degrees, pH 7.0. The minor element is a homodimer of two alpha subunits without beta subunits with a sedimentation coefficient of 6.9 S and a higher Vm rate of microtubule-stimulated ATPase of 7.0 +/- 1.9 sec-1. Microtubules stimulate the rate of release of ADP from the active site of the tetramer, but the rate of release is not fast enough to account for the rate of steady state ATP hydrolysis. Further complexity is indicated by biphasic release kinetics. In spite of the large difference in Vm ATPase rate for the two species, both drive the sliding of sea urchin axonemes over glass surfaces at the same velocity.     

Partial resolution and reconstitution of the subunits of the clathrin-coated vesicle proton ATPase responsible for Ca2+-activated ATP hydrolysis

The clathrin-coated vesicle proton-translocating complex is composed of a maximum of eight major polypeptides. Of these potential subunits, only the 17-kDa component, which is a proton pore, has been defined functionally (Sun, S.Z., Xie, X. S., and Stone, D. K. (1987) J. Biol. Chem. 262, 14790-14794). ATPase-and proton-pumping activities of the 200-fold purified proton-translocating complex are supported by Mg2+, whereas Ca2+ will only activate ATP hydrolysis. Like Mg2+-activated ATPase activity, Ca2+-supported ATP hydrolysis is inhibited by N-ethylmaleimide, NO3-, and an inhibitory antibody and is stimulated by Cl- and phosphatidylserine. Thus, Ca2+ prevents coupling of ATPase activity to vectoral proton movement, and Ca2+-activated ATPase activity is a partial reaction useful for analyzing the subunit structure required for ATP hydrolysis. The 530-kDa holoenzyme was dissociated with 3 M urea and subcomplexes, and isolated subunits were partially resolved by glycerol gradient centrifugation. No combination of these components yielded Mg2+-activated ATPase or proton pumping. Ca2+-activated ATP hydrolysis was not catalyzed by a subcomplex containing the 70- and 58-kDa subunits but was restored by recombination of the 70-, 58-, 40-, and 33-kDa polypeptides, indicating that these are subunits of the clathrin-coated vesicle proton pump which are necessary for ATP hydrolysis.     

Light chains of sea urchin kinesin identified by immunoadsorption

Previous studies with monoclonal antibodies indicate that sea urchin kinesin contains two heavy chains arranged in parallel such that their N-terminal ends fold into globular mechanochemical heads attached to a thin stalk ending in a bipartite tail [Scholey et al., 1989]. In the present, complementary study, we have used the monoclonal antikinesin, SUK4, to probe the quaternary structure of sea urchin (Strongylocentrotus purpuratus) kinesin. Kinesin prepared from sea urchin cytosol sedimented at 9.6 S on sucrose density gradients and consisted of 130-kd heavy chains plus an 84-kd/78 kd doublet (1 mol heavy chain: 1 mol doublet determined by gel densitometry). Low levels of 110-kd and 90-kd polypeptides were sometimes present as well. The 84-kd/78 kd polypeptides are thought to be light chains because they were precipitated from the kinesin preparation at a stoichiometry of one mol doublet per 1 mol heavy chain using SUK4-Sepharose immunoaffinity resins. The 110-kd and 90-kd peptides, by contrast, were removed using this immunoadsorption method. SUK4-Sepharose immunoaffinity chromatography was also used to purify the 130-kd heavy chain and 84-kd/78-kd doublet (1 mol heavy chain: 1 mol doublet) directly from sea urchin egg cytosolic extracts, and from a MAP (microtubule-associated protein) fraction eluted by ATP from microtubules prepared in the presence of AMPPNP but not from microtubules prepared in ATP. The finding that sea urchin kinesin contains equimolar quantities of heavy and light chains, together with the aforementioned data on kinesin morphology, suggests that native sea urchin kinesin is a tetramer assembled from two light chains and two heavy chains.     

Purification and characterization of kinesin from bovine adrenal medulla

Kinesin was purified from bovine adrenal medulla. The sedimentation coefficient was 8.8 S. Sedimentation equilibrium ultracentrifugation studies showed the molecular weight of kinesin to be 300,000. The calculated axial ratio was 1:16. The Stokes radius was estimated to be 8.9 nm by gel filtration. Circular dichroism showed the alpha-helix content to be about 50%. Purified kinesin preparation contained a major polypeptide with a molecular weight of 120,000 and minor ones with molecular weights of 71,000, 68,000, and 65,000. Bovine adrenal kinesin had an ATPase activity which was stimulated severalfold by microtubules to a specific activity of about 0.1 mumol/min.mg. Kinesin molecules adsorbed to a glass slide promoted the movement of microtubules on the glass surface at a rate of about 0.5 micron/s. Immunostaining of EBTr (bovine embryonic trachea fibroblast) cells and bovine adrenal chromaffin cells in interphase with an affinity-purified antibody against the major polypeptide of kinesin showed that some kinesin was located on microtubules and the rest distributed throughout the cytoplasm in a diffuse manner. EBTr cells in mitotic phase gave a staining pattern showing that kinesin was present throughout the cytoplasm with higher concentration in the region of mitotic apparatus.     

Purification and Subunit Composition of a Cholinergic Synaptic Vesicle Glycoprotein, Phosphointermediate-Forming ATPase

A glycoprotein ATPase in cholinergic synaptic vesicles of Torpedo electric organ was solubilized with octa-ethylene glycol dodecyl ether detergent. Study of potential stabilizing factors identified crude brain phosphatidylserine, glycerol, dithiothreitol, and protease inhibitors as of value in maintaining activity. The ATPase was purified from the solubilized, stabilized material by glycerol density gradient band sedimentation velocity ultracentrifugation, and hydroxylapatite, wheat germ lectin affinity, and size exclusion chromatographies. The pure ATPase had a specific activity of about 37 mumol ATP hydrolyzed/min/mg protein. After sodium dodecyl sulfate-polyacrylamide gel electrophoresis, the purified material typically exhibited three polypeptides of molecular masses 110, 104, and 98 kilodaltons (kDa) and a fourth diffuse polypeptide of 60 kDa. This composition suggests that the ATPase is a member of the P-type, or phosphointermediate-forming, family, but it was shown to be distinct from the ouabain-sensitive Na+,K+- and CA2+-stimulated Mg2+-ATPases. The purified vesicle enzyme was rapidly phosphorylated by [gamma-32P]ATP on about 14% of the subunits with molecular weights of 98,000-110,000. About 16% of the ATPase was phosphorylated in whole-vesicle ghosts in a manner consistent with formation of a phosphointermediate, thus confirming the P-type nature of this enzyme.     

Recombinant SFD isoforms activate vacuolar proton pumps.

The vacuolar proton pump of clathrin-coated vesicles is composed of two general sectors, a cytosolic, ATP hydrolytic domain (V1) and an intramembranous proton channel, V0. V1 is comprised of 8-9 subunits including polypeptides of 50 and 57 kDa, termed SFD (Sub Fifty-eight-kDa Doublet). Although SFD is essential to the activation of ATPase and proton pumping activities catalyzed by holoenzyme, its constituent polypeptides have not been separated to determine their respective roles in ATPase functions. Recent molecular characterization of these subunits revealed that they are isoforms that arise through an alternative splicing mechanism (Zhou, Z., Peng, S.-B., Crider, B.P., Slaughter, C., Xie, X.S., and Stone, D.K. (1998) J. Biol. Chem. 273, 5878-5884). To determine the functional characteristics of the 57-kDa (SFDalpha)1 and 50-kDa (SFDbeta) isoforms, we expressed these proteins in Escherichia coli. We determined that purified recombinant proteins, rSFDalpha and rSFDbeta, when reassembled with SFD-depleted holoenzyme, are functionally interchangeable in restoration of ATPase and proton pumping activities. In addition, we determined that the V-pump of chromaffin granules has only the SFDalpha isoform in its native state and that rSFDalpha and rSFDbeta are equally effective in restoring ATPase and proton pumping activities to SFD-depleted enzyme. Finally, we found that SFDalpha and SFDbeta structurally interact not only with V1, but also withV0, indicating that these activator subunits may play both structural and functional roles in coupling ATP hydrolysis to proton flow.     

Monoclonal antibodies to kinesin heavy and light chains stain vesicle- like structures, but not microtubules, in cultured cells

Kinesin, a microtubule-activated ATPase and putative motor protein for the transport of membrane-bounded organelles along microtubules, was purified from bovine brain and used as an immunogen for the production of murine monoclonal antibodies. Hybridoma lines that secreted five distinct antikinesin IgGs were cloned. Three of the antibodies reacted on immunoblots with the 124-kD heavy chain of kinesin, while the other two antibodies recognized the 64-kD light chain. When used for immunofluorescence microscopy, the antibodies stained punctate, cytoplasmic structures in a variety of cultured mammalian cell types. Consistent with the identification of these structures as membrane-bounded organelles was the observation that cells which had been extracted with Triton X-100 before fixation contained little or no immunoreactive material. Staining of microtubules in the interphase cytoplasm or mitotic spindle was never observed, nor were associated structures, such as centrosomes and primary cilia, labeled by any of the antibodies. Nevertheless, in double-labeling experiments using antibodies to kinesin and tubulin, kinesin-containing particles were most abundant in regions where microtubules were most highly concentrated and the particles often appeared to be aligned on microtubules. These results constitute the first direct evidence for the association of kinesin with membrane-bounded organelles, and suggest a molecular mechanism for organelle motility based on transient interactions of organelle-bound kinesin with the microtubule surface.