Conversion of Unc104/KIF1A Kinesin into a Processive Motor After Dimerization

See allHide authors and affiliations

Science  27 Sep 2002:
Vol. 297, Issue 5590, pp. 2263-2267
DOI: 10.1126/science.1073386


Unc104/KIF1A belongs to a class of monomeric kinesin motors that have been thought to possess an unusual motility mechanism. Unlike the unidirectional motion driven by the coordinated actions of the two heads in conventional kinesins, single-headed KIF1A was reported to undergo biased diffusional motion along microtubules. Here, we show that Unc104/KIF1A can dimerize and move unidirectionally and processively with rapid velocities characteristic of transport in living cells. These results suggest that Unc104/KIF1A operates in vivo by a mechanism similar to conventional kinesin and that regulation of motor dimerization may be used to control transport by this class of kinesins.

Caenorhabditis elegansUnc104 and the mouse ortholog KIF1A are kinesin motors that transport synaptic vesicle precursors along microtubules from the neuronal cell body to the nerve terminal (1–3). For such long-range transport to be efficient, organelles that encounter a microtubule must move processively. Conventional kinesin, which belongs to a different subfamily of vesicle-transporting kinesins, is dimeric and uses its two motor domains in a coordinated manner to take successive, unidirectional 8-nm steps along the microtubule without dissociating (4). However, KIF1A (2) and Unc104 (5) are monomeric in solution and are thought to operate using a different motility mechanism, because a single KIF1A motor domain has been shown to undergo biased diffusional movement along the microtubule (6). A novel processivity mechanism was proposed that involves an electrostatic interaction between a highly basic loop (the K loop) on the motor domain and the disordered, negatively charged COOH-terminus of tubulin (7,8); a nucleotide-induced rotation of the motor domain is believed to create directional bias (9). However, the observed net motion (0.14 μm/s) (6) is ∼eight times slower than that observed for full-length Unc104/KIF1A-induced movement in vivo (3) and in vitro (2). In addition, biased diffusion was not observed for the C. elegans Unc104 motor domain (5) and may be an artifact of particular motor constructs and buffer condition (10). Hence, the normal mode of motility for this kinesin class [often referred to as the “monomeric” kinesins (11)] has not been clear (12).

Sequence analysis of Unc104/KIF1A (and other members of this kinesin class) revealed potential coiled-coil regions adjacent to the motor domain, the first of which aligns with the neck coiled-coil of conventional kinesin (11), a region believed to play a role in head-head coordination (4). However, the probability of coiled-coil formation is low and may require high motor concentrations for dimerization. We recently found that concentration of Unc104 into lipid rafts facilitated liposome transport along microtubules and that a mutation predicted to destabilize the neck coiled-coil inhibited liposome movement (13). Collectively, these findings raised the possibility that Unc104/KIF1A might dimerize and operate by a two-headed mechanism, and we sought to test this idea using single-molecule analyses.

Our hypothesis predicted that constitutively dimerized Unc104/KIF1A motors would show processive unidirectional motion along the microtubule and not biased diffusion. We created two constitutive Unc104 dimers [green fluorescent protein (GFP)–tagged] either by fusing the GCN4 leucine zipper (LZ) after the proposed Unc104 neck coiled-coil (U389-LZ-GFP) or by joining the kinesin neck coiled-coil and stalk to Unc104 motor domain (U356-Kstalk-GFP) (Fig. 1A) (14). In contrast to two monomeric Unc104 constructs (U653-GFP and U371-GFP), the two constitutive Unc104 dimers showed robust processive movement that was smooth and unidirectional in a single-molecule, total internal reflection fluorescence (TIRF) motility assay (Fig. 1B and movie S1). The run lengths were several microns long, which is comparable to or greater than those reported for conventional kinesin (Fig. 1A), and the velocities were similar to those measured for Unc104 in the gliding assay (Fig. 1A) (5) and in vivo (3). It was previously suggested that the fast velocities produced by Unc104/KIF1A (three- to fourfold greater than those produced by conventional kinesin) required many motors interacting simultaneously with a microtubule (5, 15). However, our results indicate that these fast velocities are intrinsic to the motor domains of a single dimerized Unc104 molecule (16).

Figure 1

Single-molecule motility of artificially dimerized Unc104-GFP. (A) Single-molecule velocities and run lengths were measured with TIRF microscopy (at <20 nM motor concentrations), and multiple-molecule velocities were determined with a microtubule gliding assay (21). The velocity of U403-Kstalk2-GFP was determined during the portions of traces without pauses. See (14) for further details. (B) Kymographs showing frequency, velocity, and processivity of dimerized Unc104 constructs and a conventional kinesin construct (K560-GFP) along axonemes. Unidirectional movement appears in this plot as a diagonal line. Arrows show transient pauses during movement of U403-Kstalk2-GFP. (C) The movement of a bead coated with U389-LZ-GFP at predicted single motor density and U371-GFP at 100-fold higher motor density at 1 mM ATP in an optical trap (14) (left), the force-velocity curve of U389-LZ-GFP (upper right), and a raw trace of a U389-LZ-GFP–coated bead at 10 μM ATP showing successive ∼8-nm steps (lower right).

Because the GCN4 region of U389-LZ-GFP could induce coiled-coil formation in the adjacent Unc104 sequence, another dimeric Unc104 was created that separated the putative Unc104 neck coiled-coil and the kinesin stalk with a ∼35–amino acid unstructured region. This U403-Kstalk2-GFP protein showed unidirectional movement with a similar velocity to U389-LZ-GFP (Fig. 1A), suggesting that the Unc104 neck can form a coiled-coil. However, unlike U389-LZ-GFP, U403-Kstalk2-GFP often paused during the movement (Fig. 1B, arrows), possibly due to a temporary unwinding of the native Unc104 neck coiled-coil.

Using an optical trap, we also measured the mechanical properties of single U389-LZ-GFP molecules attached to 1-μm beads. U389-LZ-GFP displayed a linear force-velocity curve, stalled when opposed by a force of ∼6 pN, and displayed 8-nm steps (Fig. 1C). These mechanical properties are similar to those of conventional kinesin (17), implying a similar motility mechanism. In contrast, U371-GFP, which had most of the coiled-coil region removed, did not show movement at single-molecule densities. At 10- to 100-fold higher motor densities, U371-GFP–coated beads moved, but the stall forces (<3 pN) and velocities (<0.6 μm/s) were both low (Fig. 1C). These results reveal that many truncated monomers can work cooperatively to move a cargo, but the stall forces and velocities are lower than those produced by a single dimer.

To test whether coiled-coil formation in the Unc104 neck region is required for processivity, we used a mutant in which two hydrophobic residues in the first predicted heptad repeat were changed to charged residues [Ile362 to Glu362 and Leu365 to Lys365 (I362E/L365K) (Fig. 2A)] (13, 18). This neck destabilization mutation in U653-GFP and U403-Kstalk2-GFP decreased the velocity of microtubule gliding to closer to that of U371-GFP (Fig. 2B), suggesting conversion to a monomer mechanism. Consistent with this idea, this mutation abolished the unidirectional movement of single U403-Kstalk2-GFP molecules, which instead showed diffusional movement with a small plus-end–directed bias (Fig. 2C). This mutation had no effect on the gliding velocity of U371-GFP (Fig. 2B), indicating that the catalytic core was not affected. These observations reveal that coiled-coil formation in the neck region is required for head-head coordination in an Unc104 dimer, as is true for conventional kinesin.

Figure 2

A neck coiled-coil destabilizing mutation inhibits the fast velocity and processivity of Unc104. (A) Sequence alignments of human conventional kinesin and C. elegans Unc104 reveal a putative neck coiled-coil sequence in Unc104 (a and d positions of heptad repeats are shown). Single-letter 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) Single-molecule velocities, run lengths [derived from the Gaussian fit shown in (C)], and microtubule-gliding velocities of the neck destabilizing mutation (I362E/L365K) are shown in the constructs depicted in Fig. 1A. (C) Kymograph of U403-Kstalk2-GFP I362E/L365K showing diffusional movements (left) and run length distributions of U403-Kstalk2-GFP without and with I362E/L365K mutation (14) (right).

Previous studies with a native construct (U653-GFP) reported no processive movement (5); however, these single-molecule assays require low concentrations (<20 nM) of fluorescently labeled motor to reduce background fluorescence, which may be incompatible with motor dimerization because of the low coiled-coil probability (19). Supporting the idea that U653 might form dimers at higher concentrations (3 μM), we observed dimeric U653 species by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) after addition of a zero-length chemical crosslinker (Fig. 3A). To test whether U653 might dimerize into a processive motor, we increased the motor concentration in the single-molecule fluorescence assay without increasing the background fluorescence by adding an excess of non-GFP–labeled U653 (0.1 to 7 μM) to a fixed concentration of GFP-tagged motor (7 nM). With >1 μM “dark” motor, single U653-GFP molecules now showed fast unidirectional movement (Fig. 3, B and C, and movie S2) with movement properties (table S1) nearly identical to those of the artificially dimerized Unc104 motors (Fig. 1A). In contrast, the movement frequency of conventional kinesin (K560-GFP) did not increase upon addition of dark motor (K560) (Fig. 3B), and the monomer U371-GFP never moved unidirectionally even after addition of 7 μM dark U371 (Fig. 3B) or U653 (not shown). We also obtained similar results using rat KIF1A (fig. S1 and movie S3).

Figure 3

Fast processive movement of single U653-GFP motors at high motor concentrations. (A) SDS-PAGE after treatment of 3 μM U653, U371, and K560 with and without the zero-length cross-linker 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC). Arrows show the U653 and K560 dimers. U371 did not cross-link into a dimer-sized band. (B) Frequencies of the movement of U653-GFP, U371-GFP, and K560-GFP (at 7 nM) along axonemes as a function of concentration of added non–GFP-labeled motors. (C) Kymographs showing unidirectional movements (left) and distributions of velocities (right) of U653-GFP with 2 μM dark U653, K336-Ustalk-GFP with 2 μM dark K336-Ustalk, and K336-Ustalk-GFP with 2 μM dark U653 (also see table S1). (D) Kymographs (left) and distributions of spot intensities (right) of Ni-NTA liposomes transported by U653- or U371-GFP-His6, and of K560-GFP (the latter without liposomes) moving along axonemes visualized with TIRF microscopy. Insets in the kymograph show fluorescence images of the moving liposomes (or K560-GFP alone). A.U., arbitrary units.

To confirm that high motor concentrations cause dimerization through the Unc104 stalk region, we made a chimeric motor consisting of the motor domain of conventional kinesin fused to the coiled-coils and the forkhead-associated (FHA) domain of Unc104 (termed K336-Ustalk-GFP). At low concentrations, this chimeric motor did not move, but upon addition of >1 μM of dark K336-Ustalk, processive movement occurred at a rate characteristic of conventional kinesin (Fig. 3C and table S1). Processivity of conventional kinesin requires dimerization (4), which in this case must be mediated by the Unc104 sequence. When non-GFP–labeled U653 was added to GFP-labeled K336-Ustalk, we also observed processive movement of GFP-labeled motors but at a higher velocity of 0.78 μm/s as compared to the velocity after the addition of K336-Ustalk (P value of 10−10) (Fig. 3C). This higher velocity is most likely due to a heterodimer composed of a conventional kinesin motor domain and an Unc104 motor domain joined by an Unc104 stalk (additional information in table S1).

To examine whether a single dimeric Unc104 is sufficient to move membranes processively along microtubules, we used an in vitro vesicle transport assay in which U653-GFP-His6 molecules were bound to 1-μm-diameter liposomes containing 10% Ni–nitrilotriacetic acid (NTA)–conjugated lipid. This provided a stronger motor-membrane attachment than the Unc104-phosphoinositide interaction that we described previously (13). The Ni-NTA–containing liposomes moved smoothly along microtubules at velocities of 1.68 ± 0.14 μm/s. We then measured the number of U653 molecules spanning between moving liposomes and the microtubule using TIRF microscopy. Although several GFP-tagged U653 molecules may be bound to the 1-μm liposomes, only the molecules within ∼150 nm of the slide can be clearly visualized in the evanescent field. The fluorescence intensities of moving U653-GFP-liposomes varied, but the minimum intensity peak was similar to that of single conventional kinesin molecules tagged with GFP (Fig. 3D and movie S4). In contrast, the truncated monomer U371-GFP-His6 supported liposome movement only at three- to fourfold higher motor concentrations than U653-GFP-His6 motor concentrations, and these moving liposomes displayed higher fluorescence intensities and slower velocities (0.38 ± 0.09 μm/s) with frequent pausing (Fig. 3D and movie S5). These findings suggest that many monomers are required to move a vesicle, but a single U653 dimer can transport a membrane at maximal velocity.

The positively charged K loop on the motor domain was reported to be essential for the biased diffusion of monomeric KIF1A (7,8). When the K loop of U356-Kstalk-GFP was replaced with the noncharged loop of conventional kinesin (K0), processive movement was observed with velocities similar to that of the wild type (WT) (WT and K0, 2.24 ± 0.36 and 1.72 ± 0.52 μm/s, respectively), but the run lengths were dramatically decreased (WT and K0, >9.5 ± 2.2 and 1.75 ± 0.12 μm, respectively) (fig. S2, A to C). The maximum force produced by the K0 mutant was within 15% of that of the wild type (fig. S2D). Therefore, in dimeric Unc104, the K loop is not needed for either processivity or force production but enhances the run length under low load conditions, as has been described for positively charged residues in the neck coiled-coil of conventional kinesin (20).

We have shown that Unc104/KIF1A can dimerize through a weak coiled-coil region and that upon dimerization, single Unc104/KIF1A molecules move processively along microtubules. The processive mechanism of dimeric Unc104/KIF1A appears to be similar to that of conventional kinesin, except that the long run length of Unc104/KIF1A is reliant on its unique microtubule-binding K loop. Because single Unc104/KIF1A dimers move at the maximal velocities recorded for this motor in vivo, we propose that Unc104/KIF1A transports its cargo as dimers in living cells. Dimer-driven vesicle transport in vitro is also smoother, faster, and more efficient than that driven by many monomeric motors with low duty ratios. Our findings also imply that cargo motility by this “monomeric” class of motors can be regulated by promoting or inhibiting the transition to a functional dimer. This transition may be facilitated by concentrating motors into lipid raft domains (13) or by enhancing neck coiled-coil stability, possibly by factors binding to the FHA domain located adjacent to the Unc104/KIF1A neck coiled-coil (fig. S3).

Supporting Online Material

Materials and Methods

Figs. S1 to S3

Table S1


Movies S1 to S6

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


Stay Connected to Science

Navigate This Article