Distinguishing Inchworm and Hand-Over-Hand Processive Kinesin Movement by Neck Rotation Measurements

See allHide authors and affiliations

Science  01 Feb 2002:
Vol. 295, Issue 5556, pp. 844-848
DOI: 10.1126/science.1063089


The motor enzyme kinesin makes hundreds of unidirectional 8-nanometer steps without detaching from or freely sliding along the microtubule on which it moves. We investigated the kinesin stepping mechanism by immobilizing a Drosophila kinesin derivative through the carboxyl-terminal end of the neck coiled-coil domain and measuring orientations of microtubules moved by single enzyme molecules at submicromolar adenosine triphosphate concentrations. The kinesin-mediated microtubule-surface linkage was sufficiently torsionally stiff (≥2.0 ± 0.9 × 10−20 Newton meters per radian2) that stepping by the hypothesized symmetric hand-over-hand mechanism would produce 180° rotations of the microtubule relative to the immobilized kinesin neck. In fact, there were no rotations, a finding that is inconsistent with symmetric hand-over-hand movement. An alternative “inchworm” mechanism is consistent with our experimental results.

The motor enzyme kinesin moves membrane-bound organelles along microtubules in eukaryotic cells (1). Microtubule-based movements of organelles in vivo may be driven by as few as one motor enzyme molecule (2). Observations of the movement of single kinesin molecules in vitro demonstrate that the enzyme is well adapted to functioning as an isolated single molecule in living cells. First, the enzyme is processive: The kinesin undergoes multiple catalytic turnovers without detaching from the microtubule (3, 4), facilitating efficient organelle transport over long distances (5). Second, the duty ratio of kinesin is high: The enzyme cannot freely slide in the direction of the microtubule axis during most or all of its enzymatic cycle (6, 7) and thus is able to move forward even when opposed by the substantial elastic forces imposed by mechanical obstructions to organelle movements inside cells.

The mechanism by which single kinesin molecules achieve processive, high–duty-ratio movement is not well understood. Both of the enzyme's two identical head domains are required for such movement: the kinesin one–headed homolog KIF1A is processive but has low duty ratio (8–10), and truncated kinesin constructs with only one head have low duty ratios and little or no processivity [(7, 11, 12) but see also (13)]. These observations suggest that kinesin retains its grip on the microtubule through a mechanism in which head movement is coordinated so that at least one of the two heads is tightly bound to the microtubule at every stage of the catalytic cycle. It has been hypothesized that the two heads alternately move past each other (Fig. 1A). In such a symmetric hand-over-hand mechanism, the three-dimensional structure of the kinesin-microtubule complex is identical at the beginning of each adenosine triphosphate (ATP) hydrolytic cycle, except that in each cycle the two subunits (and therefore the two heads) swap places (14–18). Consequently, the neck coiled-coil domain, which links the heads together, rotates 180° around its axis for every 8-nm step (Fig. 1A, magenta arrows) (18). Analogous alternating-sites mechanisms have been hypothesized to explain the behavior of other dimeric processive motor enzymes such as myosin V (19) and Rep helicase (20). An alternative hypothesis is that kinesin head movement is coordinated through an “inchworm” mechanism (Fig. 1B) (21–23) in which the structure of the kinesin-microtubule complex is again identical at the beginning of each cycle but the two heads do not swap places. Thus, there is no net neck rotation in each cycle (Fig. 1B, magenta arrows). Such a mechanism differs fundamentally from the symmetric hand-over-hand type in that the two identical subunits of the kinesin homodimer are maintained in different environments and therefore have nonequivalent enzymatic cycles.

Figure 1

Examples of two alternative classes of mechanisms for processive, high–duty ratio movement by kinesin. Kinesin hydrolyzes one molecule of ATP and moves 8 nm in a single catalytic cycle. In both examples, the enzyme moves along a microtubule protofilament consisting of alternating α (gray) and β (white) tubulin subunits in such a way that at least one of the two identical head domains (colored red and blue for identification) is bound to the microtubule at all times. In a symmetric hand-over-hand mechanism (A), the two heads undergo identical chemomechanical reaction sequences out of phase with each other so that the heads alternate in the leading and trailing positions at the beginnings of consecutive cycles. Consequently, each cycle changes the orientation (magenta arrow) of the neck coiled-coil domain by 180°. By contrast, in an inchworm mechanism (B), the two heads retain their nonequivalent positions at the beginnings of each successive cycle (in the example shown here, the red head is always in front of the blue). The cycle of chemomechanical reactions is thus different in the two heads, and the neck orientation does not change from the beginning of one cycle to the next. Postulation of particular structures or mechanical properties for the intermediates formed during the cycles (for example, those shown in brackets) is not necessary to differentiate between the two types of mechanisms because the types are defined only by the structures of the kinesin-microtubule complex at the beginning of successive cycles.

To differentiate between these two types of mechanisms, we immobilized kinesin molecules by the distal end of the neck and examined the extent of microtubule rotation relative to the immobilized neck (Fig. 2A). The biotinated kinesin derivative K448-BIO (7, 11, 15,24) was attached to a streptavidin-coated glass cover slip (25) at low surface density, and the orientations of microtubules interacting with the surface-attached enzyme molecules were then observed by light microscopy, a modification of the method used by Hunt and Howard (26) to study kinesin torsional stiffness. In the presence of 1 to 4 mM concentrations of the kinesin inhibitor adenylyl imidodiphosphate (AMP-PNP), 100% of the microtubules that remained bound at the surface pivoted around a single point (Fig. 2B), demonstrating that they are attached to single molecules of kinesin (3, 27).

Figure 2

Torsional Brownian motion of a microtubule bound to a single surface-attached kinesin molecule in the presence of 1 mM AMP-PNP. Schematic (A) illustrates the experimental sample in which a microtubule (cylinder) is bound to the head(s) of a kinesin molecule specifically attached to a streptavidin (white squares)–coated glass cover slip (gray) through one or both biotin (black circles) moieties incorporated at the distal end of the neck coiled coil (27). Light microscope images (33.3-ms acquisition time; 1-s interval between images shown) of a 1.8-μm-long microtubule (B) demonstrate that it pivots around a single point (cross) on the surface through a restricted range of angles (36) (C). MT, microtubule.

All microtubules pivoted over a limited range of orientations (Fig. 2C); σθ, the root-mean-square orientation angle, was 44° ± 18° (mean ± SD) (Table 1). Assuming that the linkage of the microtubule to the surface through kinesin and streptavidin behaves like a simple torsion spring, the linkage torsional stiffness (K) can be calculated from the equipartition theorem (26, 28) as 〈kTθ 2〉 = 2.0 ± 0.9 × 10−20 N m rad−2, where k is the Boltzmann constant, T is the absolute temperature, and the angle brackets indicate that the quantity is the average over the sample of microtubules analyzed (29). When the kinesin molecules were instead bound nonspecifically to the surface, the microtubules exhibited Brownian rotation over >360° (Fig. 3), as observed previously with full-length kinesin (26), suggesting that the restricted torsional Brownian motion observed with specifically attached enzyme molecules is not due to obstruction of microtubule movement by surface irregularities. Taken together, the results illustrated in Figs. 2 and3 suggest that the head-microtubule connection, the neck-streptavidin connection, and the enzyme molecule itself all are torsionally stiff.

Figure 3

Rotational Brownian motion in 1 mM AMP-PNP of a microtubule bound to a kinesin molecule nonspecifically adsorbed to the cover slip surface (37). Nonspecific attachment was achieved by saturating the streptavidin-coated surface with excess biotin (27) and then increasing the enzyme concentration 80-fold and the incubation time threefold over that used for specific attachment. A range of rotation larger than 360° was measured in 10 of 10 such microtubules studied. MT, microtubule.

Table 1

Movement of microtubules interacting with single kinesin molecules.

View this table:

To determine whether single catalytic cycles of kinesin entail net rotation of the neck coiled coil relative to the microtubule, we examined orientations of microtubules moved by single enzyme molecules at 400 nM ATP, a concentration 115-fold lower than the Michaelis constant (K m) (15). Under these conditions, the kinesin spends >99% of its time poised at the beginning of its catalytic cycle, waiting for substrate to bind. Individual catalytic turnovers (which have a mean duration of ∼10 ms) are well separated by long intervals (which have a mean duration of >1 s) in which the enzyme remains bound to a fixed position on the microtubule (15). In these experiments, microtubules glide past a single pivot point on the cover slip surface (Fig. 4A) and release from the surface when the microtubule end reaches this same single pivot point (3). The width of the distribution of microtubule orientations in 400 nM ATP was similar to that seen in AMP-PNP (Table 1). Thus, the cover slip–microtubule linkage has a well-defined equilibrium orientation and does not freely swivel when the kinesin molecule is poised at the beginning of the cycle. The calculated rigidity of the linkage is sufficiently high that the torsional relaxation time for a 2-μm-long microtubule pivoting around its center to reach its equilibrium position is <1 s (30). We collected movement records corresponding to >1000 8-nm steps (Table 1); if the catalytic cycle entails a 180° rotation of the neck relative to the microtubule, the short relaxation time dictates that these records would contain numerous observations of 180° microtubule rotations. Strikingly, no such rotations were observed; the microtubules were always oriented in approximately the same direction (Fig. 4B). This observation directly conflicts with the behavior predicted for a symmetric hand-over-hand mechanism.

Figure 4

Movement of microtubules by specifically surface-attached single kinesin molecules in the presence of ATP. (A) Overlaid traces at 1-s intervals of a microtubule (arrows) in 400 nM ATP pivoting around a single point (cross) on the cover slip surface. Panels show the same microtubule in two different time periods separated by 104 s (37). Scale bar, 100 nm. (B and C) Displacement and orientation records of two microtubules in 400 (B) and 5 (C) nM ATP. Displacement is measured as the length of the microtubule segment between the trailing end and the pivot point. Displacement records are filtered with 40-point (B) and 100-point (C) mean filters and give linear fits (lines) with slopes of 2.6 (A) and 0.1 (B) nm s−1. Orientation records are not filtered. Note different time scales in (B) and (C).

To further improve our ability to detect even small (<<180°) neck rotations associated with kinesin steps, we repeated the experiments at 5 nM ATP, a condition in which kinesin steps are separated by an average interval of 89 s (Fig. 4C and Table 1). Again, no 180° rotations were observed, even though we examined microtubule movements corresponding to >300 8-nm steps. No experimentally significant discontinuities of any size were detected in most microtubule orientation records, indicating that step-associated rotations, if any, are considerably smaller than the 〈σθ〉 of 31° (Table 1) for these records.

For data at both 5 and 400 nM ATP, the fraction of measured orientations falling within ± 90° from the mean is >99% (Table 1), as would be expected from a mechanism in which no rotation occurs, not ∼50%, as predicted by the symmetric hand-over-hand mechanism.

No systematic changes in σθ with time were observed in individual microtubule movement records, even for those at 400 nM ATP in which the microtubule moved from pivoting near the microtubule center to pivoting at or near its end. These observations are consistent with our proposal that the extent of pivoting is limited by the surface linkage torsional stiffness; they are inconsistent with a possible alternative interpretation in which pivoting is restricted by collision of the microtubule ends with the surface. To confirm that the limited rotation we observed was not a consequence of some unknown geometrical constraint specific to the microtubule pivoting experiments used here, we independently measured the extent of kinesin neck rotation from one 8-nm step to the next at subsaturating ATP concentrations by observing the motion of asymmetric bead aggregates moved by kinesin along immobilized microtubules (31). Again, step-associated neck rotation was found to be small or zero. The same result was obtained in further experiments in which the rotation of spherical fluorescent beads coupled to single kinesin molecules moving on immobilized microtubules or demembranated axonemes was monitored by fluorescence polarization techniques (32).

The observation that the kinesin neck coiled coil does not rotate 180° from the beginning of one step to the beginning of the next is inconsistent with the symmetric hand-over-hand model (Fig. 1A), in which the two heads swap leading and trailing positions in consecutive cycles. This conclusion depends only on the essential feature of this model: that the three-dimensional structure of the kinesin-microtubule complex is identical (except for translation along the microtubule lattice and the interchange of the two heads in alternate steps) at the beginning of each cycle. Thus, the conclusion is independent of any assumptions about the details of the movements that occur transiently during the cycle. For example, it is immaterial whether one assumes that the trailing head always passes to the same side of the leading one (Fig. 1A, green arrows), alternates the sides on which it passes, or chooses sides randomly (18). Our observations instead are consistent with the inchworm type of mechanism (Fig. 1B), in which the structure of the kinesin-microtubule complex is identical at the beginning of every cycle and the heads do not swap places.

Both mechanisms shown in Fig. 1 adhere to the simplifying assumption that the three-dimensional geometry of the kinesin-microtubule complex is identical at the beginning of each 8-nm step. However, if we drop this assumption, we can consider yet a third type of mechanism (21), in which the neck-linker domain (or any other structure through which the heads are attached to the surface in our experiments) exists in two distinct, stable conformations that alternate in successive enzymatic cycles. In that case, hand-over-hand alternation of the head positions in successive cycles can produce the ∼0° microtubule rotation with each step observed in our experiments, provided that the two conformations differ in precisely such a way as to cancel the 180° reorientation induced by head alternation. We call this type of mechanism asymmetrichand-over-hand to emphasize that the three-dimensional structures at the beginning of consecutive 8-nm steps are different (i.e., not symmetry related). A concrete example of such a mechanism is that proposed by Hoenger et al. (33). To be consistent with the known properties of kinesin movement, the two postulated linker conformations must satisfy stringent criteria in addition to nearly exact compensation for the rotation caused by head interchange: (i) The angle between the microtubule axis and the coiled coil must not differ in the two conformations; otherwise, consecutive steps observed in bead movement experiments would not be uniformly 8 nm as observed (14, 15, 34). (ii) The equatorial angle of the coiled coil around the microtubule circumference also must not change, because beads moved along immobilized microtubules by single kinesin molecules do not wobble from side to side in alternate steps (11, 31). The radii of the beads used in the cited experiments are sufficiently large (>50 nm) that changes of even ∼5° in either angle would likely be detected (11). (iii) A high-energy barrier must block any interconversion of the two conformations that is not accompanied by catalytic turnover. Even spontaneous interconversion rates of 10−3 s−1 would produce detectable 180° rotations of microtubules in both the AMP-PNP and the limiting ATP experiments summarized in Table 1. (iv) The two stable structures must unfailingly (>99% of the time) alternate with each adenosine triphosphatase (ATPase) turnover; otherwise, 180° rotations would be observed in the 400 nM ATP experiments (Table 1). Thus, although our experimental results do not rigorously exclude an asymmetric hand-over-hand mechanism, we regard as improbable the existence of two structures that simultaneously satisfy all of the requirements outlined above.

In an inchworm mechanism, the two heads of kinesin remain in different environments (one always leading, the other always trailing) during continuous processive movement. The difference in environment implies that the chemical reactions taking place in the ATPase active sites of the two heads need not be identical. Indeed, the reactions cannot be identical—in a single cycle of the inchworm mechanism, each of the two heads moves forward 8 nm, yet only a single molecule of ATP is consumed (Fig. 1B) (14, 15). Thus, the inchworm mechanism makes the unorthodox prediction that only one of kinesin's two heads is an active ATPase during processive movement. Observations of ATP-stimulated adenosine diphosphate (ADP) release from kinesin-microtubule complexes [for example, in (17)] are sometimes taken as supporting hand-over-hand mechanisms, in which both heads hydrolyze ATP. However, such results are also consistent with processive movement driven by ATP hydrolysis in only one of the two heads, given that ATP-stimulated ADP release has been demonstrated only in the pre–steady-state reactions that occur immediately upon mixing of kinesin with ATP and/or microtubules. Such reaction steps are thus not proven to be part of the catalytic cycle for steady-state processive movement. The data presented here, taken together with that from previous studies of kinesin function, strongly support an inchworm mechanism for the processive, high–duty-ratio movement of kinesin dimers. Because hand-over-hand mechanisms proposed for other processive motor enzymes may be based in part on analogy to kinesin, the failure of kinesin to conform to predictions of the hand-over-hand hypothesis suggests that reevaluation of the evidence supporting hand-over-hand movement by other motor enzymes may be necessary.

  • * To whom correspondence should be addressed. E-mail: gelles{at}


View Abstract

Stay Connected to Science

Navigate This Article