Enzymes for Microtubule-dependent Motility*

Many of the motions by and within cells are dependent on me- chanochemical enzymes that interact with cytoplasmic filaments to generate force. For example, myosin interacts with actin filaments to generate the forces for muscle contraction, cytokinesis in animal cells, and some cases of vesicle motion in plant cells (1). Cilia and flagella are constructed from microtubules (MTs)’ and accessory proteins assembled to form the “axoneme,” and ATPases called dynein (2) drive axonemal bending. Like myosin, they use the chemical energy derived from ATP binding and hydrolysis to force a sliding of adjacent filaments (3, 4). Recently, assays based on visualizing the motion of MTs and small particles with high contrast optics (5, 6) have led to the discovery of enzymes that promote movement over the surfaces of cytoplasmic MTs; kinesin (7) and a cytoplasmic form of dynein (8, 9) have now been isolated from diverse types of cells. This paper will review recent work on dyneins and kinesins and discuss the roles these enzymes may play in cell motility. The majority best “arms” “A subfiber” of (the a complete in cross-sections). the


rangement of the enzyme in axonemes is still in dispute (3).
Knowledge about axonemal proteins in general and about dynein in particular has profited from a genetic dissection of flagellar motility in Chlamydomonas (11). This approach has corroborated the biochemical complexity of dynein. Some mutations that inactivate only a single polypeptide result in the absence of an entire dynein arm; electrophoresis shows that 12 polypeptides are then missing from the axoneme (13). The importance of dynein in regulation of the flagellar wave form is also being analyzed by genetics Axonemal dynein's nucleotide binding is specific for Mg-ATP (14). Its ATPase activity is strongly inhibited by vanadate (VOi-) (15) but is stimulated by MTs at high concentrations (16). The mechanism of ATP hydrolysis has been examined for outer arm dynein from the cilia of Tetrahymena (10). This enzyme binds tightly to MTs in the absence of ATP but is rapidly dissociated following ATP binding to its heavy chain; ATP hydrolysis then proceeds more slowly on the dissociated enzyme. The release of reaction products is rate-limiting in the absence of MTs, but the addition of MTs stimulates product release, resulting in an activation of ATPase activity (16).
light in the presence of ADP and V 0 4 (3,17). A different UV-Dynein heavy chains can be cleaved at specific sites by near U V stimulated cleavage pattern is observed without nucleotide at higher concentrations of vanadate (3). These properties have been used, together with nucleotide and antibody binding, to construct onedimensional maps of the dynein heavy chains (Fig. 2.4) (3, 18). The refinement of these maps and an understanding of their relationships to both the electron microscope images and the binding sites for the other dynein polypeptides are active areas of current work. Computer processing of light microscopic images (5, 6) has made it possible to assess the movement of small objects in cell extracts (3, 19), spawning an assay for MT motility (7) that has been used to analyze the mechanical action of MT-dependent motors in vitro.
Axonemal dynein will adsorb to glass and move MTs over that surface in an ATP-dependent manner (20, 21). The polarity of motion observed in vitro is the same as that shown for dynein in situ (21,22); axonemal dynein walks along a MT toward its "minus" end, i.e. the end that is slow to add tubulin and that lies proximal to the basal body of the axoneme. The motion is Mg-ATP-specific, vanadatesensitive, and proceeds at a velocity comparable to that observed in situ. Sea urchin outer arm dynein is dissociable by dialysis against low ionic strength into a soluble complex of one heavy chain plus one intermediate chain (HC-@/IC-1) and an insoluble residue. The soluble moiety will move MTs over glass, so a single heavy chain complex (Le. one head) is sufficient for dynein-driven MT motility (23).
The frequent observation of MT-dependent motility outside an axoneme has led to speculations about the existence of non-axonemal or "cytoplasmic" dyneins. Early work on such enzymes focused on FIG. 1. Dyneins and kinesin prepared for electron microscopy by J. Heuser (Washington University, St. Louis, MO) using fast freezing and etching (X 250,000). A, protozoan axonemal dynein prepared by U. Goodenough (Washington Univ.). The heads corresponding to the dynein heavy chains are evident. The stems and bases probably correspond to the intermediate and light chains that associate to form the active complex. B , chicken brain dynein prepared by E. Steuer (Washington Univ.). C, chicken brain kinesin prepared by T. Schroer  T I and Tz are tryptic cleavage sites in order of sensitivity; VI and V2 are the ADP-requiring and -independent vanadate cleavage sites, respectively. The arrows indicate the positions of binding of the chemicals shown. The N terminus is acetylated. Fragment A is a proteolytic cleavage product that retains ATPase activity, while fragment B has none (redrawn from Ref. 18). E , cytoplasmic dynein heavy chain from brain (9), though the diagram is approximately applicable to all cytoplasmic dyneins known. C, kinesin heavy chain from Drosophila (arrows as in A ) .
echinoderm eggs, which are convenient sources of a cytoplasm programmed for mitosis, vesicle movements, and morphogenesis. Dynein-like activities have been identified in egg extracts by both biochemical and immunological criteria, and dynein-like proteins have been isolated by several methods (for review, see Ref. 24). There are at least two isoforms of dynein in sea urchin eggs: one is primarily soluble but is recognized by antibodies to sperm axonemal dynein; the other binds MTs in uitro but is not recognized by the same antiaxonemal dynein (58). The roles of egg dynein are still controversial, but it seems likely that one isozyme is a maternal contribution to the formation of embryonic cilia, while the other may be important for egg functions such as granule movement.
To obviate axonemal contamination, dynein-like enzymes have been sought in cells that do not make cilia or flagella. Early efforts with brain tissue yielded ambiguous results (24), and indeed any quest for a non-axonemal dynein in a tissue or organism that forms motile cilia or flagella will lead to questions about the possibility of axonemal contamination. A search for high molecular weight ATPases in brain, HeLa cells, and insect eggs identified 20 S enzymes containing polypeptides that co-migrated with dynein heavy chain on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, but these enzymes showed CTPase > ATPase (25). Since axonemal dynein has low CTPase, the relationship of these activities to dynein was unclear. A dynein-like, MT-dependent, mechanochemical enzyme has, however, been isolated from Caerwrhabditis elegans, a nematode worm that never makes motile axonemes at any stage of its life cycle (8). This protein shows ATP-sensitive binding to MTs and a 20 S vanadatesensitive Mg-ATPase activity. It contains polypeptides that co-migrate with axonemal dynein heavy chain and are cleaved by near UV light in the presence of ADP and vanadate. It moves MTs over glass in an ATP-dependent manner, and its motility is blocked by vanadate-ADP-UV cleavage. Such a cluster of properties is a clear demonstration that dynein-like proteins exist outside axonemes.
Similar dynein-like proteins have now been described in vertebrate brain (9), testis (26), in HeLa cells (27), Drosophila eggs (28), the amoeba, Reticulomyxa (29), and the slime mold, Dictyostelium (30). It seems likely that cytoplasmic dynein is ubiquitous among the eukaryotes, though work on plant cells has yet to be reported. These enzymes are all very similar to dynein from C. ekgam, and recent work has greatly expanded our knowledge of their characteristics. The brain enzymes have been visualized in the electron microscope, where they appear as two globular heads connected by stalks to a common base (Fig. 1B) (27,31). Cleavage of their heavy chains by ADP-VO,-UV results in polypeptides that differ slightly in size from those obtained from axonemal dynein ( Fig. 2E) (8,9,24). Most cytoplasmic dyneins appear to include several chains of lower molecular weight (9, 31). Some have been shown to form ATP-sensitive bundles of MTs (27), and some hydrolyze CTP faster than ATP, though this nucleotide will not support motility (32). It seems likely, therefore, that the high molecular weight CTPases mentioned above (25) were cytoplasmic dyneins. Data comparing the enzymology and motility of axonemal and cytoplasmic dyneins are summarized in Table I.
The localizations and functions of cytoplasmic dynein are receiving much current attention. The presence of the enzyme in neurons, where many vesicles move from the axon terminus toward the cell body, is consistent with its serving as a motor for organelle movement toward the minus ends of MTs, e.g. "retrograde" axonal transport (33, 34). Dynein from chick embryo fibroblasts can promote such motion of cell-derived vesicles in uitro (35).

Kinesin
During 1985 a factor was identified in the cytoplasm of squid giant axons that could bind MTs to glass and move them along their axes in the presence of ATP (7,24). This motile activity was frozen by the ATP analog, AMP-PNP (36), and AMP-PNP-stimulated binding to MTs was soon used to purify the motility factor from brain (7, 37) and sea urchin eggs (38). This protein, called kinesin, will bind to latex spheres, glass, and some cytoplasmic vesicles; it will move them toward a MT's plus end at about 0.5 pm/s (7, 24).
Kinesin comprises at least two polypeptides with relative molecular masses of about 120 and 65 kDa; it migrates on gel filtration columns with a solution molecular mass of about 360 kDa, suggesting that the native molecule contains two copies of each polypeptide (39, 40). Affinity labeling of native kinesin with azido-ATP (24, 40) and MT binding to kinesin synthesized in vitro (41) show that both the nucleotide and the MT-binding sites reside on the heavier chain. Electron microscopy reveals that kinesin has two small heads at one end connected by a rodlike segment to a feathery tail (Fig. 1C) (24,42,43). Some monoclonal antibodies to the larger kinesin polypeptide will block its motility but stimulate its MT-activated ATPase activity. These antibodies recognize a 45-kDa chymotryptic fragment that retains ATP-sensitive MT binding (44). They bind to native kinesin at or near its heads, suggesting that this region of the molecule, like subfragment 1 of heavy meromyosin, contains both the nucleotide and the fiber binding sites (43). Kinesin's feathery tail is the location of at least one epitope on its lower molecular weight component(s) (24).
Polyclonal antibodies have been used to identify the single gene that encodes kinesin's heavy chain in Drosophila (41,451. This gene has now been sequenced (46), revealing a consensus nucleotide binding site near the N terminus of the gene product. Toward the C terminus there are two successive strings of heptad repeats separated by proline, suggestive of two distinct regions of a-helical coiled coil. The synthesis in uitro of polypeptides encoded by cloned fragments of this gene has demonstrated that a 50-kDa N-terminal polypeptide displays ATP-sensitive MT binding (41), consistent with the behavior of the proteolytic fragments described above (Fig. 2C). These facts are assembled in the model for kinesin proposed in Yang et al. (46) and redrawn in Fig. 3.
Kinesin's pathway for ATP binding and hydrolysis is similar to the cross-bridge cycle previously described for myosin and dynein, but there is an interesting difference. Like the other mechanochemical enzymes, kinesin is dissociated from its fiber by the addition of ATP (7), and the nucleotide triphosphate is hydrolyzed rapidly. The steady state reaction rate is limited by the release of products from the enzyme's active site (47, 48), but unlike myosin, kinesin releases Pi rapidly and holds tightly to ADP. The addition of MTs stimulates ADP release and thereby activates the ATPase (47).
Kinesins from different sources appear to be rather similar (24).
Their enzymatic and motility properties are summarized in Table I. More divergence is seen, however, in the plus-directed MT motors isolated from Acanthmoeba (49) and Dictyostelium (50). These do not show enhanced binding to MTs in the presence of AMP-PNP, but functional criteria suggest that they too are probably kinesins. These rate constants were calculated by measuring the initial slopes of the data describing MT activation of the enzyme's ATPase activity (Fig. 6 in Ref. 48 and Fig. 1

FIG. 3. Diagram of kinesin, incorporating the information from Goldstein's, Scholey's, Heuser's, Bloom's, Brady's, and Hirokawa's laboratories (see text for references).
Kinesin's role in cells is a subject of active current investigation. Its prevalence in neurons and its capacity to bind small spheres or vesicles to MTs and move them suggest that it plays a role in particle movement in vivo (7). Kinesin is probably a motor for the motion of cytoplasmic vesicles toward the plus ends of MTs, e.g. "orthograde" axonal transport (24,33), and it may link MTs with the membranes of endoplasmic reticulum, stretching them to the edge of the cytoskeleton (51). The localization of kinesin in uiuo could help to identify kinesin's role in cells, but this issue is still unresolved. Most kinesin is soluble in cell extracts (7), but some is fixed by treatment of cells with aldehydes or cold methanol. Kinesin is concentrated in the mitotic apparatus of sea urchin eggs (38) and appears to be localized at the spindle poles of several cultured cell types (24). Other studies with different antibodies and cell types have found fixed kinesin on tiny cytoplasmic spheres (52) or diffusely distributed with some concentration along MTs (27). All these patterns can be attributed to membrane or MT binding, consistent with the enzyme's playing a role in membrane-cytoskeleton interactions, but the diversity of localizations precludes strong inferences about kinesin's function. Probes that perturb specific aspects of kinesin's action will therefore be needed to define kinesin's roles in cells. Since motility-blocking antibodies are now available (44) and mutants with modified kinesins are being sought (27), we can expect information about this motor's role in cell behavior rather soon.

Functional Comparison of Dynein and Kinesin
Polarity of Motor Action-For most cases so far studied, dynein moves toward the minus end of a MT (21, 341, while kinesin moves toward the plus (24,33). One apparent exception was the assignment of plus-directed motility to nematode dynein (S), but this direction has subsequently been corrected (24). The misassignment was a result of dynein's differential effect on tubulin disassembly from the plus and minus ends of MTs (27). There is, however, an organism in which dynein-like molecules appear to be involved in both plus-and minus-directed movements. MT-dependent vesicle transport in Reticubmyxa is bidirectional, and both directions of movement are revived in detergent-extracted cell models by the addition of dynein prepared from this organism (29). Both directions of movement are inhibited by treatments with vanadate or with vanadate-ADP-UV (27), suggesting that dynein-like molecules in this organism contribute to motility in both directions. It remains to be seen whether there are dynein isozymes with opposite directions of action, whether dynein can interact with a "gear box" to reverse its polarity of motion, or whether dynein associates with a kinesin-like molecule to make a bidirectional complex whose function is sensitive to inactivation of only one of the component motors. A monoclonal antibody to kinesin has been found to block both directions of granule motion in extracts of squid axoplasm (53), consistent with the latter idea.
An Analogy with Actin-dependent Motors-The existence of two MT-dependent motors with opposite polarities of motion raises the intriguing question of whether there is an actin-dependent motor that works in the direction opposite to myosin's. The flexibility of individual microfilaments may be too great to support transport toward a free filament end, so it may be only MTs that can sustain a bidirectional motor system. However, many plant cells show bidirectional streaming in their transvacuolar strands, and this appears to be an actin-based motility (1).
Patterns of Movement-Dynein and kinesin differ in the details of their motile behavior. Image processing of light micrographs showing kinesin-coated microspheres moving on MTs reveals a 4-nm step (54). At high ATP concentrations, the bead motion is smooth, and at all ATP concentrations the motion is largely parallel to the MT axis.
Microspheres coated with cytoplasmic dynein show a more circuitous motion with a significant component of motion perpendicular to the MT axis and occasional large steps (27). A t low ATP or in [ADP] = [ATP], dynein-driven motion of MTs over glass is jittery with steps back and forth as large as a micrometer.
Motor Enzymology and the Value of Inhibitors-Cytoplasmic dynein hydrolyzes high concentrations of CTP faster than ATP, but this hydrolysis appears to be analogous to the increased ATPase activity shown by some axonemal dyneins after treatment with Triton X-100 or by kinesin after binding to some monoclonal antibodies; the enzymes hydrolyze NTP more rapidly but become mechanochemically inactive. Cytoplasmic dynein does not waste the cell's CTP because of the high K,,, for this reaction (27).
Both axonemal and cytoplasmic dyneins show MT activation of their ATPase activities (9, 16,32). The effect appears more dramatic for the cytoplasmic isozyme because of both a lower basal activity and a 10-fold higher rate constant for MT binding ((32) Table I).
The same factors give kinesin an even stronger MT activation than cytoplasmic dynein (47,48). It is not surprising that the activities of cytoplasmic motors are strongly coupled to MT binding, since ATP hydrolysis without polymer binding cannot promote motility. The motors from axonemes are not as tightly constrained, because the concentration of MTs in cilia or flagella is about 60-fold higher than in cytoplasm, and structural considerations make the effective tubulin concentration for axonemal dynein even higher.
The dyneins are particularly sensitive to inhibition by vanadate and N-ethylmaleimide, while kinesin is more strongly inhibited by AMP-PNP (Table I). These properties provide important tools for distinguishing processes that depend on the different motors in vivo, but it is noteworthy that the pharmacology of axonal transport (19) and granule motion in chromatophores (reviewed in Ref. 55) do not correspond exactly with the properties of the MT-dependent motors described in vitro. This inconsistency is probably due to the multiple factors that contribute to kinesin or dynein-driven movement in uiuo, e.g. the levels of phosphorylation of associated proteins. The discrepancies provide strong motivation for finding even more specific probes for each motor's action. Under the conditions used, vanadate-ADP-UV cleavage seems to be specific for dynein in vitro (3, 17, 24), but the efficacy of this treatment in cells remains to be determined. Function-blocking antibodies, antisense RNA, mutations, and transformation of cells with genes making products that poison motor action would appear to be profitable avenues for further work.
Motor Diuersity-The structural complexity of dynein or kinesin may permit a functional diversity that is significant for the organism. The action of either motor in vivo will depend upon the binding of its parts that do not associate with MTs in an ATP-sensitive way. This binding will define the "payload" that the enzyme can carry along a MT and thus the function that its motile action will perform. Structural data for both motors suggest that their lower molecular weight components contribute to these binding activities and as such may play a key role in the specificity of motor action.
Even though plural binding capacities may make kinesin and dynein multifunctional, there are probably other MT-dependent motors yet to be discovered. A MT-activated ATPase that is neither kinesin nor dynein has been described in sea urchin eggs (56), and an analogous activity has now been found in brain (27). Many of the movements associated with mitotic spindles do not correspond well with the properties of either kinesin or dynein (24), so we expect that additional MT-dependent motors remain to be discovered when the right assays can be found. This probable plethora of motor activity is consistent with the diversity in speed, control, and function seen for MT-dependent motions in cells.

Control of Motor Function
We are just beginning to understand how cells control the activities of their MT-dependent motors. Most axonemes work nonstop, but control is exercised over their wave forms by the concentration of free Ca2+ and the activities of both protein kinases and phosphatases (3). Axonal transport also runs continuously, and its regulation appears to be at the level of deciding which components will move. One can imagine a model in which the vesicles that move neurotransmitter from a nerve cell body to its axon terminus bind active kinesin to their exterior surfaces and sequester dynein in an inactive form. Such particles would then be carried to the distal tip of an axon and unloaded, making available both the neurotransmitter and the motor essential for retrograde motility. Alternatively, one can posit a complex of dynein and kinesin that can manifest only one activity at a time. The factors that bind this complex to vesicles might define the active motor and hence the direction in which the vesicle would move. The frequent reversal of direction by some cytoplasmic vesicles is more consistent with the latter model than the former. Indeed the shuttling character of pigment granule movement in chromatophores suggests that both plus and minus motors are always available, and studies on lysed cell models indicate that phosphorylation plays a key role in selecting the motor that will be most active at any moment (57).
The binding of motors to the objects they will move suggests a second level of regulation on motor enzymology. One would expect that motors without loads would show a low affinity for MTs and hence a low level of ATP hydrolysis. Since purified kinesin shows a dramatic MT activation of its ATPase, and cytoplasmic dynein something of the same, it seems likely either that the factors important for this level of control have been purified away from the current motor preparations or that when a load binds to a motor it greatly enhances the motor's affinity for MTs. Recent studies with MTs, motors, and vesicles combined in vitro show that some additional soluble component is essential for the formation of a vesicle-moving complex (24), and future studies on motor enzymology may reveal this higher level of control on ATPase activity as well. It is clear that we still have much to learn about the ways cells handle MT-dependent motors, but it is an exciting time for practitioners in this field who have so many avenues for productive work.