Cross, R.A [1], Crevel, I [1], Carter, N.J.[1], Alonso, M.C.[1] , Hirose, K.[2] and Amos, L.A.[3]

1. Molecular Motors Group, Marie Curie Research Institute, The Chart, Oxted, Surrey RH8 0TL
2. National Institute of Advanced Interdisciplinary Research, Tsukuba 305-8562 Japan
3. MRC Laboratory of Molecular Biology, Hills Road, Cambridge, CB2 2QH

SUMMARY
The stepping mechanism of kinesin can be thought of as a programme of conformational changes. We briefly review protein chemical, electron microscopic and transient kinetic evidence for conformational changes, and working from this evidence, outline a model for the mechanism. In the model, both kinesin heads initially trap MgADP. Microtubule binding releases ADP from one head only (the trailing head). Subsequent ATP binding and hydrolysis by the trailing head progressively accelerate attachment of the leading head, by posing it closer to its next site. Once attached, the leading head releases its ADP and exerts a sustained pull on the trailing head. The rate of closure of the molecular gate which traps ADP on the trailing head governs its detachment rate. A speculative but crucial coordinating feature is that this rate is strain-sensitive, slowing down under negative strain and accelerating under positive strain.

1. INTRODUCTION
2-headed ukinesin (ubiquitous kinesin, Kif5b) molecules are able to step processively along microtubules against an applied load (Howard et al., 1989). Elegant single molecule mechanical measurements have shown that the motor steps 8nm at a time, corresponding to the axial separation between tubulin heterodimers in a microtubule protofilament, and stalls at a retroactive force of about 6pN (Svoboda et al., 1993) (Svoboda and Block, 1994) (Kojima et al., 1997). Recent reports suggest that progress is tightly coupled to ATP consumption, approximately one ATP molecule being consumed per 8nm step (Schnitzer and Block, 1997) (Hua et al., 1997) (Coy et al., 1999) (Cross, 1997a).

At low loads, the kinesin stepping mechanism can be thought of as a programme of discrete conformational changes, each reaction step in ATP processing producing modest motions in the active site which are amplified so as to drive the motor into a new conformation. Under load, the motor tries to execute this same programme of conformational changes, but the rate constants for some or all of the transitions are altered by physical tension. The reciprocal relationship between the tension on the motor molecule and its active site chemistry is called the mechanochemical coupling. To elucidate the mechanism of mechanochemical coupling, we need to know the minimum energy conformation of the motor for each different nucleotide occupant of the active site and the rate constants for transitions between these conformations at low load, together with the influence of external strain on the structure and on these rate constants.

We have studied several kinesin superfamily members in relation to conformation and the kinetics of conformational changes. Here, we briefly discuss current evidence for conformational changes in the kinesins, and suggest that the various lines of evidence point to a general model for mechanochemical coupling, in which the key coordinating process is the ADP trapping reaction which precedes detachment. The rate constant for this step is postulated to be strain-dependent, and hence is expected to both buffer and broker internally-generated and externally-applied forces, and to provide a basis for head-head coordination in processive stepping. Such a model is testable and is supported by preliminary evidence.

2. CLASSES OF CONFORMATIONAL STATE
Like myosin, conformational states of kinesins can usefully be classified as weak ( low affinity, tending to detach) and strong (high affinity, able to hold force). All kinesin family members so far examined show a common pattern whereby the nucleotide-free state binds tightest, and analogues of intermediates in ATP processing weaken the binding in order of effectiveness ADP > ADP.AlF4 > AMPPNP. Data for various kinesin family members are summarised in the table. The ADP state is the weakest binding.

[table 1]

3. CRYO-ELECTRON MICROSCOPY OF KINESIN AND NCD
Beautiful X ray structures of the ADP conformations of several kinesin family monomers (Sablin et al., 1996) (Kull et al., 1996) (Gulick et al., 1998) (Sack et al., 1997) and dimers (Kozielski et al., 1997) (Sablin et al., 1998) have been obtained, but we do not yet have atomic-resolution views of the other conformational intermediates in nucleotide turnover. Thus far, all attempts to crystallise kinesins in states other than K.ADP have failed (J. Kull, personal communication). Electron microscopy offers the possibility to visualise these crystallographically intractable conformations in complex with microtubules, and hence to probe for changes in shape and attitude of the motor as the nucleotide in the active site is exchanged. For kinesin even the weakest binding state, the kinesin.ADP state, has a micromolar dissociation constant, and so can be populated at workable protein concentrations and visualised in the electron microscope.

Early work using negative staining of single heads indicated a plus-end directed tilt of microtubule-attached kinesin heads on ADP release (Hirose et al., 1995). Negative staining can in principle deliver very high resolution, but because of worries that the staining process can damage the specimen, subsequent work has been done using rapid freezing of unstained specimens. Using cryoEM and helical reconstruction procedures,, Hirose and Amos have obtained 3d electron density maps of the ADP, rigor and AMPPNP conformations of 2-headed recombinant rat kinesin and drosophila ncd attached to microtubules, at between 2 and 4 nm resolution. For each motor, substantial conformational differences are seen between the different nucleotide states. In all cases, only one of the two heads binds to the microtubule, the other being held clear of the microtubule surface, and so able, presumably, to probe for its next site. The tethered heads are posed in different positions in the different conformations. Figure 1 shows reconstructions of 2-headed kinesin and 2-headed ncd in three different nucleotide states.

In all three conformational states of ncd, the tethered head is held to the right of the bound head and points towards the minus end of the microtubule. In the ADP state it is held at roughly the same level as the bound head, in an arrangement which is similar but apparently not identical to that in the N.ADP dimer crystal structure (figure 3). In the nucleotide free state, the junction between the heads has moved up and a little to the right, and the tethered head has moved outwards from the microtubule surface. In the AMPPNP state, assumed to mimic the ATP state, the tethered head moves about 2nm towards the plus end, is retracted somewhat against the bound head, and twists clockwise about 10 degrees.

For kinesin, changing the nucleotide in the active site has a more substantial effect on the position of the tethered head. In ADP, the tethered head is held above and to the left of the bound head in an arrangement that mimics that in the dimeric K.ADP crystal structure (figure 3). In apyrase , the tethered head moves down and lies against the bound head, pointing towards the minus end with its long axis roughly parallel to the microtubule axis. In AMPPNP, the tethered head shifts upwards and over to the right of the bound head. The apparent mass of the tethered head is reduced in the ADP and AMPPNP states, suggesting mobility. Dynamics produces blurring which reduces the apparent mass. This problem does not seem to affect ncd, suggesting its free head may be less dynamic. Mandelkov and coworkers do not see the second head mass at all in their reconstructions of kinesin (Hoenger et al., 1998) (Thormahlen et al., 1998), and this could be due to dynamics (as we believe), or to binding of both heads to the microtubule in their specimens (as they believe). Both heads certainly can bind, but the kinetic work shows that binding of the second head in AMPPNP is ~10-fold slower than first head binding (Ma and Taylor, 1997) (Gilbert et al., 1998), so that at saturating concentrations of kinesin, one would predict that most sites would become occupied by first heads.

The attached heads of kinesin and ncd look very similar, and bind to identical sites on tubulin, consistent with earlier work showing mutually competitive binding (Lockhart et al., 1995). Conformational differences between the bound heads in different nucleotide states can nonetheless be seen. In particular, a spike on the left side of the bound head of ncd shifts upwards towards the plus end on AMPPNP binding, as was first noted in the negative staining work on kinesin single heads (Hirose et al., 1995). A protrusion at top right of the kinesin.ADP bound head is missing in the apyrase state. For both motors, nucleotide free and AMPPNP conformations of the bound heads sit closer to the microtubule surface and have a larger contact interface than do the ADP conformations, consistent with tighter binding.

The cryoEM results reported by Wade and colleagues ((Arnal and Wade, 1998) are broadly similar to our own, with the interesting difference that in their experiments the ukinesin.ADP conformation looks closer to the ukinesin.AMPPNP reconstruction of Hirose et al. We believe this may indicate that two different ADP states can occur, corresponding to different points in the conformational cycle.

4. MAPPING THE MICROTUBULE BINDING INTERFACE
In parallel with fitting studies, we and others have directly probed for the microtubule binding surface of kinesin using biochemical techniques. Woehlke et al (Woehlke et al., 1997) made an alanine scan of charged residues on the candidate microtubule binding face of human kinesin. Charged residues are distributed broadly over the candidate binding surface. Of those residues whose mutation to alanine had an effect, substitution of positive residues reduced the Km for microtubule activation of the ATPase, whilst substitution of negative residues had the opposite effect. We took an alternative approach, using proteolytic protection to probe for the microtubule binding interface of kinesin and ncd monomers, and to search for nucleotide-induced conformational changes (Alonso et al., 1998). The data are shown in figure 3. We identified a set of 18 cleavage sites in kinesin and 19 in ncd, and asked which were protected from proteolysis by microtubule binding. A substantial surface is protected by microtubule binding (figure 3). Similar surfaces are protected on kinesin and on ncd, in agreement with earlier work which showed competitive binding of ncd and kinesin heads to microtubules (Lockhart et al., 1995), and with recent fitting studies (Hirose, 1999) which suggest the bound heads of kinesin and ncd are congruent. Using the conventional orientation (figure 3), the protected surface is on the opposite side of the molecule to that containing the ATP binding site.

5. FITTING X RAY MAPS INTO EM MAPS
In the case of ncd, (Sosa et al., 1997) fitted the ncd.ADP monomer coordinates into an ncd.AMPPNP EM structure. The junctions between the heads can be clearly seen, and were used as the main constraint on the fitting. Because the tethered head of ncd does not appear to move very much between different states, this yielded an orientation which agreed with the alanine scanning data. For kinesin however, attempts to fit kinesin in the same orientation as ncd predicted unacceptable steric clashes for dimeric kinesin, indicating that something was wrong, as noted by Mandelkow and colleagues ((Hoenger et al., 1998) (Thormahlen et al., 1998).

The problem turns out to be that all this fitting was done using motor.ADP crystal structures, and motor.AMPPNP EM maps. Our ukinesin.ADP EM map turns out to be very different from the ukinesin.AMPPNP EM maps (Hirose, 1999). A fit made by Kozieslki et al (Kozielski et al., 1998) to the K.ADP EM map of Arnal et al. places sites which are protected from proteolysis in an exposed position, and cleavable sites in a protected position, and hence is inconsistent with the proteolysis data. However, fitting of the ukinesin.ADP dimer crystal structure into our newly-obtained ukinesin.ADP cryoEM map produces a different and more satisfactory result. Since the heads are shaped like slightly flattened hearts, the flattening can be used to find the best rotational attitude of the tethered heads. The new fit (Hirose, 1999) is still consistent with the alanine scan and proteolytic data but, being rotated by ~60û from the original position, avoids the steric clashes. It also brings the nucleotide binding site nearer to the microtubule and, interestingly, puts loop L9, which is thought to be analogous to a G-proteinÕs switch I loop (Vale and Fletterick, 1997), into direct contact with tubulin.

6. NUCLEOTIDE-INDUCED LOOP MOVEMENTS
The proteolytic study described above identified several sites which were protected from proteolytic cleavage by a change of nucleotide in the active site, in the absence of microtubules. Three sites, I70 and L154 in rat kinesin, and Y622 in ncd, were cleavable with ADP in the active site, but not with AMPPNP in the site. Possibly the L8 and L12 loops which contain the latter two residues are displayed so as to mediate the initial kinesin-microtubule binding interaction, but are then retracted to allow the switch to a strong binding conformation. Probe work is also beginning to reveal molecular motions (Naber et al., 1997).

7. KINETICS OF CONFORMATIONAL TRANSITIONS
ADP trapping by kinesin was first described by Hackney (Hackney, 1988), who also was the first to show that mixing two-headed kinesin with microtubules released only half of the bound ADP, so that it was necessary to add a chase of ATP to release the other half (Hackney, 1994). Subsequent work showed that AMPPNP could also chase ADP from the second head, and it was suggested that in the normal cycle of walking, ATP binding to the bound (trailing) head is the trigger for microtubule binding and consequent microtubule-activated ADP release from the leading head (Ma and Taylor, 1997) (Gilbert et al., 1998). We recently tried analogous experiments with the very fast Neurospora kinesin (nkin), which moves around fourfold faster than conventional (brain) kinesin. The Neurospora enzyme can also walk, as was evident from single molecule optical trapping experiments. Kinetic studies showed however that compared to brain kinesin, AMPPNP was relatively inefficient at triggering release of leading head ADP. ATPgS and ATP induced much faster second head ADP release, and we accordingly concluded that for nkin, hydrolysis was necessary for rapid second head ADP release. Importantly, both the kinesin and nkin data would be consistent with a single model in which progressive acceleration of leading head microtubule binding occurs as ATP is processed on the trailing head. For both kinesin and nkin, ADP binding to the trailing head induced only very slow leading head ADP release, indicating that the spontaneous detachment of the trailing head K.ADP is too slow to be on the normal pathway, and suggesting the need for facilitated detachment.

8. TOWARDS A MECHANOCHEMICAL MODEL
Figure 5 shows a mechanochemical scheme which derives from the evidence outlined above. In the scheme, the motor generates force and gains some ground by changing its conformation (its shape or tilt angle) after attachment to the track, but most of the per-head step distance of 16nm is gained by facilitated diffusion of tethered heads prior to microtubule binding. Tethered heads are assumed to diffuse randomly about a null point defined by the connection to the bound head, but have the ability to recognise and lock-in productive diffusional excursions by binding to the track and changing conformation. Reversing the directional bias of this diffusion-to-capture process would provide a straightforward way to reverse directionality in two motors which otherwise underwent similar conformational changes once bound to the track. (Lockhart and Cross, 1994). It is possible to rationalise the finding that the direction of ncd heads can be reversed by splicing them to a kinesin tail (Henningsen and Schliwa, 1997) (Case et al., 1997), and the inverse experiment reversing kinesin (Endow and Waligora, 1998) in this way (Cross, 1997b). The conformational change accompanying landing (that corresponding to microtubule activated release of trapped ADP) can contribute to the step distance via a lever arm action, though evidence to data suggests that for ukinesin at least, any such lever must be relatively short ((Inoue et al., 1997), (Romberg et al., 1998), (Mazumdar and Cross, 1998)).

In the absence of microtubules, both kinesin heads trap ADP. Collision of this complex with microtubules releases ADP from one of the two heads, but not from the other. The processive cycle starts at this point, with one head in rigor, and the other trapping ADP. ATP binding and hydrolysis then occur on the trailing head, and these progressively facilitate the diffusional search made by the tethered (leading) head for its next site, perhaps by shifting the null point (the head-head junction or neck) closer to the new site. Once the leading head locates its site, it binds and releases its ADP, concomitantly exerting a sustained pull on the trailing head. This pull accelerates closure of the gate which traps ADP on the trailing head, and the trailing head rapidly detaches.

Considerable uncertainty surrounds the Pi release step. The K.ADP.AlF4 data suggest that K.ADP.Pi is a tightly bound state, and that Pi release could therefore do work. This is an attractive idea because in myosin, Pi release is associated with a large free energy change that probably drives a lever arm motion. In the kinesin case we speculate that Pi release is also strain dependent, but there are no data. For the purposes of the present discussion, were Pi release to be strain dependent (slowing done under negative strain and accelerating under positive strain), the net effect would be the same: the rate of formation of the detaching trapped-ADP conformation would be strain dependent. This postulated strain dependence of ADP trapping would provide a mechanism for coordinating neighbouring heads, because opening of the gate which traps ADP on the leading head is synchronised with closure of the equivalent gate on the trailing head. For non-processive kinesins (we believe ncd and eg5 are essentially non-processive (Crevel et al., 1997) (Stewart et al., 1998)), this type of mechanochemical scheme can still apply, but tension signals are passed between heads via the microtubule. The key feature of the scheme, strain-sensitive ADP trapping, works equally well as a mechanism of head-head coordination when the heads are not part of the same molecule. Mechanical processivity of the ukinesin subfamily would then be a result of structural adaptions which allow dimeric motors to bridge between neighbouring binding sites, whilst retaining essentially the same strain-dependent chemistry as other family members.

Acknowledgements
We thank those present and past colleagues and collaborators who have contributed materially and mentally to this work, particularly Andrew Lockhart, Scott Brady, Sharyn Endow, Manfred Schliwa and Joel Vanderkerckhove.

Figure legends

Figure 1
Comparison of 3D cryo EM reconstructions of 3 different nucleotide states of dimeric ncd (left) and Kinesin (right). Microtubule plus ends are towards the top of the page.

Figure 2
Front and back views of (A,B) ncd and (C,D) kinesin in the conventional orientation showing proteolytic sites protected (red) and unprotected (blue) by MT binding.

Figure 3
Spacefilling and cartoon views of the newly-fitted orientation of the kinesin.ADP dimer structure (Kozielski et al., 1997)

Figure 4
Mechanochemical scheme. Binding and hydrolysis of ATP (yellow circle) to the back foot accelerates front foot attachment and associated microtubule-activated ADP (yellow semicircle) release. A resulting pull on the back foot accelerates the ADP trapping reaction which precedes detachment, possibly by accelerating Pi (black semicircle) release.

References

Alonso, M. C., van Damme, J., Vandekerckhove, J. and Cross, R. A. (1998). Proteolytic mapping of kinesin/ncd-microtubule interface: nucleotide- dependent conformational changes in the loops L8 and L12. Embo J 17, 945-51.

Arnal, I. and Wade, R. H. (1998). Nucleotide-dependent conformations of the kinesin dimer interacting with microtubules. Structure 6, 33-8.

Case, R. B., Pierce, D. W., Hom-Booher, N., Hart, C. L. and Vale, R. D. (1997). The directional preference of kinesin motors is specified by an element outside of the motor catalytic domain. Cell 90, 959-66.

Coy, D. L., Wagenbach, M. and Howard, J. (1999). Kinesin takes one 8-nm step for each ATP that it hydrolyzes. J Biol Chem 274, 3667-71.

Crevel, I. M., Lockhart, A. and Cross, R. A. (1996). Weak and strong states of kinesin and ncd. J Mol Biol 257, 66-76.

Crevel, I. M., Lockhart, A. and Cross, R. A. (1997). Kinetic evidence for low chemical processivity in ncd and Eg5. J Mol Biol 273, 160-70.

Cross, R. A. (1997a). Molecular motors: the natural economy of kinesin. Curr Biol 7, R631-3.

Cross, R. A. (1997b). Reversing the kinesin ratchet--a diverting tail [news; comment]. Nature 389, 15-6.

Endow, S. A. and Waligora, K. W. (1998). Determinants of kinesin motor polarity. Science 281, 1200-2.

Gilbert, S. P., Moyer, M. L. and Johnson, K. A. (1998). Alternating site mechanism of the kinesin ATPase. Biochemistry 37, 792-9.

Gulick, A. M., Song, H., Endow, S. A. and Rayment, I. (1998). X-ray crystal structure of the yeast Kar3 motor domain complexed with Mg.ADP to 2.3 A resolution. Biochemistry 37, 1769-76.

Hackney, D. D. (1988). Kinesin ATPase: rate-limiting ADP release. Proc Natl Acad Sci U S A 85, 6314-8.

Hackney, D. D. (1994). Evidence for alternating head catalysis by kinesin during microtubule- stimulated ATP hydrolysis. Proc Natl Acad Sci U S A 91, 6865-9.

Henningsen, U. and Schliwa, M. (1997). Reversal in the direction of movement of a molecular motor [see comments]. Nature 389, 93-6.

Hirose. (1999). congurent binding .. Mol Biol Cell

Hirose, K., Lockhart, A., Cross, R. A. and Amos, L. A. (1995). Nucleotide-dependent angular change in kinesin motor domain bound to tubulin. Nature 376, 277-279.

Hoenger, A., Sack, S., Thormahlen, M., Marx, A., Muller, J., Gross, H. and Mandelkow, E. (1998). Image reconstructions of microtubules decorated with monomeric and dimeric kinesins: comparison with x-ray structure and implications for motility. J Cell Biol 141, 419-30.

Howard, J., Hudspeth, A. J. and Vale, R. D. (1989). Movement of microtubules by single kinesin molecules. Nature 342, 154-8.

Hua, W., Young, E. C., Fleming, M. L. and Gelles, J. (1997). Coupling of kinesin steps to ATP hydrolysis. Nature 388, 390-3.

Inoue, Y., Toyoshima, Y. Y., Iwane, A. H., Morimoto, S., Higuchi, H. and Yanagida, T. (1997). Movements of truncated kinesin fragments with a short or an artificial flexible neck. Proc Natl Acad Sci U S A 94, 7275-80.

Kojima, H., Muto, E., Higuchi, H. and Yanagida, T. (1997). Mechanics of single kinesin molecules measured by optical trapping nanometry. Biophys J 73, 2012-22.

Kozielski, F., Arnal, I. and Wade, R. H. (1998). A model of the microtubule-kinesin complex based on electron cryomicroscopy and X-ray crystallography. Curr Biol 8, 191-8.

Kozielski, F., Sack, S., Marx, A., Thormahlen, M., Schonbrunn, E., Biou, V., Thompson, A., Mandelkow, E. M. and Mandelkow, E. (1997). The crystal structure of dimeric kinesin and implications for microtubule-dependent motility. Cell 91, 985-94.

Kull, F. J., Sablin, E. P., Lau, R., Fletterick, R. J. and Vale, R. D. (1996). Crystal structure of the kinesin motor domain reveals a structural similarity to myosin. Nature 380, 550-555.

Lockhart, A., Crevel, I. M. and Cross, R. A. (1995). Kinesin and ncd bind through a single head to microtubules and compete for a shared MT binding site. J Mol Biol 249, 763-771.

Lockhart, A. and Cross, R. A. (1994). Origins of reversed directionality in the ncd molecular motor. Embo J 13, 751-7.

Lockhart, A. and Cross, R. A. (1996). Kinetics and motility of the Eg5 microtubule motor. Biochemistry 35, 2365-73.

Ma, Y. Z. and Taylor, E. W. (1997). Interacting head mechanism of microtubule-kinesin ATPase. J Biol Chem 272, 724-30.

Mazumdar, M. and Cross, R. A. (1998). Engineering a lever into the kinesin neck. J Biol Chem 273, 29352-9.

Naber, N., Cooke, R. and Pate, E. (1997). Binding of ncd to microtubules induces a conformational change near the junction of the motor domain with the neck. Biochemistry 36, 9681-9.

Romberg, L., Pierce, D. W. and Vale, R. D. (1998). Role of the kinesin neck region in processive microtubule-based motility. J Cell Biol 140, 1407-16.

Rosenfeld, S. S., Rener, B., Correia, J. J., Mayo, M. S. and Cheung, H. C. (1996). Equilibrium studies of kinesin-nucleotide intermediates. J Biol Chem 271, 9473-82.

Sablin, E. P., Case, R. B., Dai, S. C., Hart, C. L., Ruby, A., Vale, R. D. and Fletterick, R. J. (1998). Direction determination in the minus-end-directed kinesin motor ncd. Nature 395, 813-6.

Sablin, E. P., Kull, F. J., Cooke, R., Vale, R. D. and Fletterick, R. J. (1996). Crystal structure of the motor domain of the kinesin-related motor ncd. Nature 380, 555-559.

Sack, S., Muller, J., Marx, A., Thormahlen, M., Mandelkow, E. M., Brady, S. T. and Mandelkow, E. (1997). X-ray structure of motor and neck domains from rat brain kinesin. Biochemistry 36, 16155-65.

Schnitzer, M. J. and Block, S. M. (1997). Kinesin hydrolyses one ATP per 8-nm step. Nature 388, 386-90.

Song, H. and Endow, S. A. (1998). Decoupling of nucleotide- and microtubule-binding sites in a kinesin mutant. Nature 396, 587-90.

Sosa, H., Dias, D. P., Hoenger, A., Whittaker, M., Wilson-Kubalek, E., Sablin, E., Fletterick, R. J., Vale, R. D. and Milligan, R. A. (1997). A model for the microtubule-Ncd motor protein complex obtained by cryo- electron microscopy and image analysis. Cell 90, 217-24.

Stewart, R. J., Semerjian, J. and Schmidt, C. F. (1998). Highly processive motility is not a general feature of the kinesins. Eur Biophys J 27, 353-60.

Svoboda, K. and Block, S. M. (1994). Force and velocity measured for single kinesin molecules. Cell 77, 773-84.

Svoboda, K., Schmidt, C. F., Schnapp, B. J. and Block, S. M. (1993). Direct observation of kinesin stepping by optical trapping interferometry [see comments]. Nature 365, 721-7.

Thormahlen, M., Marx, A., Muller, S. A., Song, Y., Mandelkow, E. M., Aebi, U. and Mandelkow, E. (1998). Interaction of monomeric and dimeric kinesin with microtubules. J Mol Biol 275, 795-809.

Vale, R. D. and Fletterick, R. J. (1997). The design plan of kinesin motors. Ann Rev Cell Dev Biol 13, 745-77.

Woehlke, G., Ruby, A. K., Hart, C. L., Ly, B., Hom-Booher, N. and Vale, R. D. (1997). Microtubule interaction site of the kinesin motor. Cell 90, 207-16.