A Processive Single-Headed Motor: Kinesin Superfamily Protein KIF1A

See allHide authors and affiliations

Science  19 Feb 1999:
Vol. 283, Issue 5405, pp. 1152-1157
DOI: 10.1126/science.283.5405.1152


A single kinesin molecule can move “processively” along a microtubule for more than 1 micrometer before detaching from it. The prevailing explanation for this processive movement is the “walking model,” which envisions that each of two motor domains (heads) of the kinesin molecule binds coordinately to the microtubule. This implies that each kinesin molecule must have two heads to “walk” and that a single-headed kinesin could not move processively. Here, a motor-domain construct of KIF1A, a single-headed kinesin superfamily protein, was shown to move processively along the microtubule for more than 1 micrometer. The movement along the microtubules was stochastic and fitted a biased Brownian-movement model.

A single kinesin molecule has two heads or motor domains (1) and moves processively (2–4), taking more than 100 steps before detaching from a microtubule (MT). The two-headed structure is assumed to be essential for processive movement, because engineered single-headed or monomeric molecules have motor activity but are not processive, whereas dimeric constructs have high processivity (4–9). The prevailing hypothesis for the mechanism of processive movement is the “walking model” (2, 10), in which one head anchors the molecule to the MT while the other moves to the next binding site on the MT. Thus, a single-headed molecule is not expected to move processively because the molecule is not anchored while it moves to the next binding site.

A monomeric kinesin superfamily protein, KIF1A, which has only one motor domain, supports MT gliding at ∼1.2 μm/s (11). It remains unresolved whether this naturally single-headed KIF is processive or not. A conventional, very low motor density MT gliding assay and a single-motor motility assay with fluorescently labeled full-length KIF1A protein suggested that it is processive (12). However, the results were very variable, possibly because of the instability or heterogeneity of the full-length KIF1A recombinant protein. Thus, to overcome these problems, we produced a KIF1A motor-domain chimeric construct, C351, for further analysis.

The design of C351 is based on the shortest monomeric motile, but nonprocessive, construct DmK351 made from Drosophilaconventional kinesin (4, 6). The catalytic core of the conventional kinesin motor domain of K351 was swapped with that of KIF1A (13). Thus, C351 contains the catalytic core of the motor domain of KIF1A followed by the 23–amino acid “linker” of conventional kinesin (14), which is reported to be insufficient for dimerization (8). As a control, a dimeric conventional kinesin molecule, K381, was also constructed, containing the coiled-coil neck region essential for dimerization (8,15).

The motor molecules were labeled with a red fluorescent dye, Alexa, and observed under a low background video-intensified microscope (16). When C351 molecules were applied at a relatively higher concentration (final 100 pM) in the presence of adenosine triphosphate (ATP, 2 mM), each of the MTs in the chamber was decorated with dozens of C351 molecules. They continuously moved to one end of the MT, and accumulated to create a bright spot (Fig. 1A). Almost all of the MTs (>99%) were strongly decorated only on one end. The other ends were either dimly or not at all decorated with C351. This lollipop-like decoration of the MTs was also observed with K381, as reported previously (17). In contrast, K351 did not bind to MTs even at 1000 pM in the presence of ATP, confirming the previous report that monomeric conventional kinesin does not show a processive movement in this type of assay (4).

Figure 1

Motility assay of fluorescently labeled C351. (A) High motor concentration assay of C351. C351 (red) was applied at 100 pM to MTs (green) fixed on a cover glass. The MTs were decorated with dozens of C351 molecules. The molecules moved continuously to one end of the MT (arrowhead) and accumulated at this end to create a bright spot. The other end was either dimly or not at all decorated with C351 (arrow). Scale bar, 2 μm. (B) Low motor density assay of C351. At 1 to 10 pM, MTs (green) were sparsely labeled with several fluorescence spots (red). The fluorescence intensity profile of these spots [see (C)] indicated that these spots were single C351 molecules. They attached to the MTs, moved along them for several seconds, and were finally detached from the MTs. Sometimes C351 moved backward (arrowhead), but it usually moved in one direction (arrow). Scale bar, 2 μm; frame interval, 0.5 s. (C) Intensity profile of the moving fluorescence spots. Blue (C351) and red (K381) bars show the measured distribution of the fluorescence intensities of C351 and K381 moving along the MTs. Blue and red lines show the theoretical curve assuming a Gaussian distribution. The variance was estimated from the observed values. The blue curve assumes that C351 is a monomer, and the red curve assumes that K381 is a dimer. The fluorochrome/protein ratio of the K381 used in this assay was 70%. Thus, 49% of the molecule is doubly labeled, and 42% is singly labeled, if the COOH-terminal cysteine residues are randomly labeled. These theoretical curves fit very well with the observed values, indicating that single fluorescent molecules are observed with our system, and that C351 is a monomer, whereas K381 is a dimer. A video illustrating motility data can be viewed

At lower motor molecule concentrations (1 to 10 pM) of C351 or K381, MTs were sparsely decorated with several fluorescent spots (Fig. 1B), moving along them for several seconds. Statistical analysis of the fluorescence intensities of these moving spots indicated that they are single-motor molecules (Fig. 1C). Thus, under these conditions, we could observe the behavior of single-motor molecules (18).

Single C351 molecules were observed to move processively along MTs (Fig. 1B). To quantitatively measure the processivity of the movement (mechanical processivity), we plotted the exponential distribution of the duration of movement on the MT (4, 19). The best-fit curve indicated the mean duration of movement, or the time constant (τmec) of C351 dissociation from the MTs, to be 6.1 ± 0.8 s (Fig. 2A), suggesting the high processivity of a single-headed C351 motor, compared to 2.6 ± 0.2 s for K381 (Fig. 2B). The movement of single C351 molecules on the MTs was not smooth, and actually appeared oscillatory. Although it was on average unidirectional (Fig. 2D), single C351 motors sometimes paused for a while, or moved backward for a short distance, and then moved forward (arrowhead in Fig. 1B).

Figure 2

Analysis of the movement of single fluorescently labeled C351 and K381 molecules. (A andB) Distribution of the duration of movement of motor C351 (A) and K381 (B). Bars show the measured values, and lines show the best fit with a single exponential distribution. The time constants τmec of the motor dissociation from MTs thus derived are 6.1 ± 0.8 s for C351 (A) and 2.6 ± 0.2 s for K381 (B). (C) Distribution of run length. Shaded bar indicates C351; open bar indicates K381. Curves show the best fits with a single exponential distribution, with μ = 0.84 ± 0.2 μm for C351 and 2.0 ± 0.2 μm for K381. (D and E) Typical traces of the displacement of the motor molecules (39). Each trace shows the path of an individual C351 (D) or K381 (E) motor. C351 movements are rather oscillatory or stochastic; K381 movements are smoother. (F and G) Distribution of the displacement of C351 (F) and K381 (G) plotted against time. Curves show the best fits with normal distributions (μ = 140, 230, or 430 nm, σ2 = 4 × 104, 10 × 104, or 18 × 104 nm2 forT = 0.5, 1, or 2 s, respectively, for C351; μ = 340, 690, or 1400 nm, σ2 = 8 × 104nm2 for K381). (H) MSD, ρ(t), of C351 (○) and K381(⧫) plotted against time. Curves show the best fits with ρ(t) = 2Dt +v 2 t 2 (D = 44,000 ± 1200 nm2/s, v = 140 ± 10 nm/s for C351; D = 2200 ± 1000 nm2/s, v = 710 ± 10 nm/s for K381). (I) The diffusion term or ρ(t) –v 2 t 2 of C351 (○) and K381 (⧫) plotted against time. Curves show the best fits with the same parameters as in (H).

To quantitatively analyze this apparently oscillatory or fluctuating movement of C351, we plotted the distribution of the displacement (Fig. 2F). The plot fitted well with a normal distribution whose mean and variance increased linearly against time. The linear increase of the mean indicates the unidirectional constant-velocity movement on average. The linear increase of the variance cannot be explained by the error (or noise) in the position measurement. Such noise does not accumulate and is reflected as the constant term in the variance. The linear increase of the variance indicates the linear accumulation of random noise, as found in the Brownian movement. Thus, the distribution of the displacement of C351 suggests that it is a biased Brownian movement (that is, diffusion with a drift).

In contrast, K381 moved more smoothly in a single direction (Fig. 2E), as reported previously for the longer construct K440 (4). The distribution of its displacement fitted well with a normal distribution with a linearly increasing mean and a constant variance (Fig. 2G). This constant variance can be explained by the error in the position measurement. Thus, relative to C351, the movement of K381 is much smoother, as reported previously (20).

We further analyzed the movement of C351 and K381 by plotting the mean square displacement [MSD, ρ(t)] against time, which is a convenient quantitative measure of stochastic movement (21). The MSD plot of C351 (Fig. 2H) fitted very well with biased Brownian movement: ρ(t) = 2Dt +v 2 t 2, where D is the diffusion coefficient and v is the (mean) velocity. The MSD plot of C351 fitted well withD = 44,000 ± 1200 nm2/s andv = 140 ± 10 nm/s. This estimated value ofD is much larger than the value expected from the fluctuation in the adenosine triphosphatase (ATPase) activity (<2400 nm2/s) (20), whereas it is in good agreement with the value previously reported for one-dimensional Brownian movement in MT–motor protein systems (22, 23). In contrast, the MSD plot of K381 fitted well with D = 2200 ± 1000 nm2/s and v = 710 ± 10 nm/s (Fig. 2H). These values are in good agreement with previous results (D = 1400 nm2/s, v = 670 nm/s) (20), and the fluctuation of the movement of K381 can be explained by the fluctuation in the ATPase activity (20).

The difference between C351 and K381 in the degree of fluctuation of the movement is illustrated by the plot of the diffusion term, or ρ(t) –v 2 t 2(Fig. 2I). The movement of C351 is about 20 times as diffusive as that of K381. This heavily stochastic nature of the movement cannot be explained by the error of the position measurement, by the Brownian noise that often impedes the bead-based nanometry of the motor movement, or by any heterogeneity in the motor activity. The errors of the position measurement and the Brownian noise do not accumulate, and only contribute as the constant term to the diffusion term of MSD or the variance of the displacement. Furthermore, our single-motor assay is free from the Brownian noise because the position of the fluorescent spot directly reflects the position of the motor molecule. The heterogeneity in the motor activity would produce a variance that grew quadratically with time (24). Thus, the diffusive nature of the C351 movement would reflect its inherent nature rather than the artifactual noise.

With the single-motor motility assay, only active molecules that bind and move along MTs can be observed. To avoid this bias (25), we measured the kinetic parameters of the C351 ATPase, which reflect the mean enzymatic behavior of the population. As the kinetic index of processivity, we measuredk bi(ratio), the ratio of the second-order rate constant of MT-activated ATPase activity (k bi(ATPase)) and that of MT-activated adenosine diphosphate (ADP) release (k bi(ADPrelease)), which indicates the number of ATPase cycles per motor-MT encounter (26).

The results of the kinetic assays are summarized in Table 1 (27). The kinetic processivity index k bi(ratio) of C351 was 690. On average, C351 hydrolyzes nearly 700 ATPs before detaching from the MT. This is more than five times the value for the dimeric kinesin K381, and more than 40 times that for the monomeric kinesin K351. This high kinetic processivity could be the source of the observed mechanical processivity. Dividing this value by the turnover ratek cat (110 s–1) yields a time constant for the motor-MT dissociation (τkin) of 6.3 s, showing good agreement with that obtained from the single-motor motility assay (τmec = 6.1 ± 0.8 s) (28). The agreement of τkin and τmec strongly suggests that the behavior of the fluorescent spots observed in the single-motor assays actually reflects the behavior of the single-motor molecules, and that the single-headed C351 is a highly processive motor. What, then, lends processivity to C351?

Table 1

Summary of the mechanical and kinetic assays of the motor domain constructs. s 20,w, sedimentation coefficient; N.D., no detectable movement.

View this table:

The kinetic processivity index k bi(ratio) is determined by the three kinetic parameters k cat,K MT(ATPase) 0.5, (MT concentration for half-saturation of ATPase) and k bi(ADPrelease). Among these, the K MT(ATPase) 0.5 of C351 is smaller than that of K351 by a factor of about 100, whereas the other parameters differ in magnitude by a factor of only 2 to 3 (Table 1). Thus, the data suggest that C351 binds to the MT with exceptionally high affinity, so that more than several hundred ATPase cycles occur before it is detached from the MT.

The binding of C351 to MTs is very sensitive to the ionic strength (29), which suggests that it is mediated by electrostatic interactions, as proposed for kinesin (4, 30), whose positively charged residues might bind to the negatively charged region of tubulin (31, 32). Quite interestingly, a region just upstream of the well-conserved His-Ile-Pro-Tyr-Arg-Asp motif has six extra lysines in tandem (Fig. 3A). This region corresponds to loop 12 (33), which is suggested to be involved in MT-kinesin interactions (32). Thus, this loop with six extra lysines (which we call the “K-loop”) is a good candidate to explain the exceptionally high affinity of KIF1A for the MT.

Figure 3

The source of the exceptionally high affinity of KIF1A for MTs. (A) Sequence comparisons of the region around the K-loop. Unc104/KIF1 subfamily KIFs are shown at the top, followed by other KIFs representative of the various KIF subfamilies (40). Their GenBank accession numbers and their positions in the phylogenetic tree are indicated in (40) and are available at . Conserved α helices (α4 and α5) are shown in light blue, and positively charged amino acid residues (lysine and arginine) in the K-loop are in red. Monomeric KIFs of the Unc104/KIF1 subfamily have more than five lysine residues in the K-loop, whereas other dimeric or tetrameric KIFs have fewer than two lysine or arginine residues. This suggests that the extra lysine residues in the K-loop play important roles only in monomeric KIFs. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. (B) Competitive inhibition of the MT-activated ATPase activity of C351 by oligo-lysine (K6) with K i = 80 nM.

This speculation is supported by the following three results. First, molecular phylogenetic analysis demonstrates that the K-loop is conserved only in naturally monomeric KIFs (Unc104/KIF1 subfamily) but not in dimeric (or tetrameric) KIFs (Fig. 3A), indicating its unique role in monomeric KIFs. Second, a lysine hexamer (K6) competitively inhibits the C351-MT interaction (Fig. 3B), with inhibition constant K i = 80 nM. At 1 μM, K6 completely blocks the binding to MTs by C351 in the single-motor assay. Third, a mutant C351 whose K-loop is replaced by the corresponding residues of kinesin showed less than one-fifth the affinity for the MT, and showed no binding to MTs in the single-motor assay (34). These results collectively suggest that this K-loop (and the MT region that binds to K6) plays an essential role in the high-affinity binding of KIF1A to MTs.

Thus, C351 is anchored to the MT via the K-loop so as to prevent its diffusion away from the MT. At the same time, however, it should be able to translocate along the MTs, otherwise it cannot move because the motor domain is too small [∼6 nm (33, 35)] to reach the next binding site, which is 8 nm away along the MT protofilament.

One possible model posits that extra binding sites exist on the MT and that the mechanical step size of C351 is smaller than that of conventional kinesin. Then, the small motor domain might be able to step to the next binding site. However, this model cannot explain the observed large fluctuation (variance = 400 nm2 per ATPase cycle) compared with the small mean displacement (1.3 nm per ATPase cycle) (36).

Another possibility is that C351 moves by successive dissociation from and reassociation to the MT. This can explain the observed diffusive nature of the movement. However, after complete dissociation from the MT, the motor will rapidly diffuse away from it, and the probability of reassociation will be too low to explain the observed long movement. This difficulty can be overcome by assuming that C351 is weakly anchored to the MT so as not to diffuse away from it. That is, C351 is anchored to the MT in such a way that the electrostatic potential restricts its movement away from the MT while allowing free movement along the MT. This can be represented as a groove-shaped potential along the MT (37).

The free thermal fluctuation during this “weak binding state” enables C351 to move to the distant binding site, and also explains the highly fluctuating nature of the movement. When C351 switches to the strong binding state, it will bind to the “nearest” binding site. If there is a bias toward the plus-end direction in this rebinding step, this step can bias or rectify the thermal fluctuation in the weak binding state. This bias can be thermodynamically modeled as the anisotropy in the potential of interaction between C351 and the MT. This anisotropy might be produced by the periodic polarized nature of the MT protofilament (38), or by the structural change (or the “power stroke”) in the motor domain. Similar bias toward the MT plus-end might also be produced in the dissociation step. When C351 makes a transition from the strong to the weak binding state, there might be some structural change in the motor domain, which pushes C351 toward the MT plus-end (Fig. 4). This model fits very well to the observed movement of C351 with parameters of reasonable values (38).

Figure 4

Model of the processive movement of monomeric C351. Each panel consists of three parts. At the top is a schematic representation of C351 and a MT protofilament (yellow and green depict α tubulin and β tubulin, respectively). The middle and bottom parts show the probability distributionp(x) of the position of C351, and the potentialu(x), respectively (38).

In this model, the thermal diffusion in the weak binding state is essential for the movement. The presumable structural changes in the motor (power stroke) or the anisotropy in the potential of the motor-MT interaction by themselves are too small to move the motor to the next binding site. The thermal diffusion allows the motor to reach the next binding site. The power stroke, or the anisotropy in the potential, only biases or rectifies the undirected thermal diffusion toward the MT plus-end. That is, this model envisions that the ATPase hydrolysis is used for the rectification of the thermal diffusion by the transition of the binding state or the structural changes in the motor.

In a conventional MT gliding assay, binding of other motor molecules on the glass surface anchors the whole system to the MT, and only a small fluctuation within the reach of the motor-glass linker is allowed. This can effectively reduce the possibility of the backward movement if the reach is short enough. This will enable rapid movement with less fluctuation. This mechanism might explain the discrepancy between the motile behavior of C351 in the multiple-motor assay (smooth unidirectional movement at 1.2 μm/s) and in the single-motor assay (biased Brownian movement at 0.14 μm/s).

The combination of the single-motor motility assay and the biochemical assay has conclusively demonstrated the processivity of the monomeric KIF1A motor-domain construct C351. The kinetic measurements suggest the source for this high processivity, whereas the single-motor assay demonstrates the diffusive nature of its movement. These results suggest a new model for kinesin movement. Future studies must be carried out to test these hypotheses.

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


View Abstract

Navigate This Article