Tether-scanning the kinesin motor domain reveals a core mechanical action

Significance

Kinesin superfamily members have evolved a variety of N- and C-terminal linkers, which tether the motors to their cargoes. The docking of these linkers against the motor domain exerts force and in kinesin-1 dimers is required for molecular walking. What remains unclear is whether linker docking is the only mechanism by which kinesin motor domains exert force or whether an underlying, more fundamental force-generating mechanical cycle exists, whose action is amplified by linker docking. Here, we show that kinesin motor domains tethered via double-stranded DNAs linked to surface loops drive robust microtubule gliding. Our data reveal a basal, force-generating mechanical cycle of the kinesin motor core that in wild-type motors underlies, drives, and is amplified by, linker docking.

Abstract

Natural kinesin motors are tethered to their cargoes via short C-terminal or N-terminal linkers, whose docking against the core motor domain generates directional force. It remains unclear whether linker docking is the only process contributing directional force or whether linker docking is coupled to and amplifies an underlying, more fundamental force-generating mechanical cycle of the kinesin motor domain. Here, we show that kinesin motor domains tethered via double-stranded DNAs (dsDNAs) attached to surface loops drive robust microtubule (MT) gliding. Tethering using dsDNA attached to surface loops disconnects the C-terminal neck-linker and the N-terminal cover strand so that their dock–undock cycle cannot exert force. The most effective attachment positions for the dsDNA tether are loop 2 or loop 10, which lie closest to the MT plus and minus ends, respectively. In three cases, we observed minus-end-directed motility. Our findings demonstrate an underlying, potentially ancient, force-generating core mechanical action of the kinesin motor domain, which drives, and is amplified by, linker docking.

The wild-type kinesin-1 molecular motor is a twin-headed bionanomachine that transduces the chemical energy of ATP hydrolysis to haul cargo toward the plus end of microtubules (MTs) (1, 2). In classical experiments, the wild-type dimeric kinesin-1 motor was truncated to establish the minimal functional motor domain. It was shown that lawns of monomeric kinesin-1 heads, tethered to a cover glass and supplied with ATP, are sufficient to drive MT gliding (3).

In most such motility assays, kinesin heads (now called motor domains) are tethered to a surface via a ~13 residue C-terminal neck-linker (NL) sequence (β9-β10). In the context of wild-type kinesin-1 dimers, this sequence connects each motor core (β1-α6) to the coiled-coil tail of the motor (4, 5). During force generation, the NL cyclically docks and undocks to the catalytic core (6, 7). The N-terminal cover strand (β0) partially overlies the docked NL (8, 9). Both the NL and cover strand sequences are strongly conserved across a large fraction of the kinesin superfamily (10). Mutating the NL, or its docking station, or restricting NL motion by chemical cross-linking, profoundly inhibits motility, for both dimeric and monomeric kinesin-1 (11–16). Moreover, NL docking is known to play roles other than in force generation, especially in facilitating unbinding from the MT via ADP trapping (14). There is thus clear evidence for a central role for the NL docking in the wild-type mechanochemical cycle of both dimers and monomers of kinesin-1. Nonetheless, it remains possible that the NL docking cycle overlies and amplifies a more fundamental mechanical stroke generated by the core kinesin-1 motor domain.

To test this point, we ran in vitro polarity-marked MT-gliding assays in which kinesin-1 motor domains were anchored to coverslips not by their NL or cover strand but instead via surface-exposed loops of their catalytic motor core. We screened all available surface loops, using double-stranded DNA (dsDNA), or in certain cases polyethylene glycol (PEG), to bridge the glass substrate to the attachment point. We compared MT-gliding velocities for multiple different loop-tethered constructs with those for the wild-type NL or cover strand connectivity. Additionally, we tested for an ongoing functional role for NL and cover strand docking in the context of surface loop tethering. Surface loop tethering disconnects the NL and cover strand so that their dock–undock cycle cannot itself exert force. Our approach reveals that indeed the kinesin-1 core motor domain can drive directional MT gliding with both its NL and cover strand disconnected.

Results

We developed a modified in vitro MT-gliding assay in which defined amino acid residues within surface loops of the kinesin-1 motor domain are tethered to a cover glass. We engineered a series of constructs based on a “cysteine-light” kinesin-1 monomer (Rattus norvegicus KIF5C; residues 1 to 340) (17), in which a cysteine residue is introduced at the N terminus, C terminus, or into a surface-exposed loop (Fig. 1 A and B). One end of a dsDNA oligomer (40 base pairs, ~14 nm) was reacted to the inserted cysteine residue (SI Appendix, Fig. S1), and its opposite end was anchored via biotin to a streptavidin-coated cover glass surface (Fig. 1C). For two constructs we used a smaller and more flexible molecule, PEG, as an alternative tether (SI Appendix, Fig. S1). For lawns of kinesin-1 motor domains tethered via their C-terminal NL (termed NL-tethered WT), the NL forms the physical connection between the substrate and the catalytic motor core, so that gliding force generated by the MT-bound motor domain must necessarily be transmitted through the NL (Fig. 1 C, Left). For lawns of cover strand–tethered WT motor domains, tethering is via the N-terminal cover strand and the NL can no longer directly transmit force (Fig. 1 C, Middle), but force might still be transmitted via cover-neck bundle (CNB) formation (18). By contrast, for motor domains tethered via surface loops (loop-tethered WT), neither the NL nor the CNB can exert or transmit force (Fig. 1 C, Right).

Fig. 1.

MT-gliding assay using NL-, cover strand–, or loop-tethered monomeric kinesin-1. (A) Positions of cysteines introduced into the cysteine-light kinesin for labeling with thiol-reactive dsDNA oligomer or PEG, shown as red spheres on the crystal structure of rat kinesin (PDB: 3KIN). The catalytic motor core (residues 10 to 325; light brown) is flanked by the N-terminal cover strand (residues 2 to 9, magenta) and the C-terminal neck-linker (residues 326 to 338, blue). Mutations of three residues to alanine (I327A, K328A, and N329A) in the neck-linker are shown in orange. The kinesin is rotated approximately 90° about the horizontal axis. (B) Monomeric kinesin-1 constructs used in this study. A cysteine residue is introduced into the N terminus, the C terminus, or a loop region of the cys-lite motor domain (residues 1 to 340). Color coding as in (A). (C) Topology of the MT-gliding assay. Motor domains are tethered to the streptavidin-coated glass surface via dsDNA or PEG covalently linked to the introduced cysteine in C terminus (Left, blue), N terminus (Middle, magenta), or a loop region (Right, orange).

Both NL-Tethered and Cover Strand–Tethered Kinesin-1 Monomers Drive Plus-End-Directed Motility.

We first assayed the motor activity of lawns of NL-tethered WT and cover strand–tethered WT kinesin-1 monomers by attaching them to a cover glass and allowing them to drive gliding of polarity-marked MTs (Fig. 2A). Our choice of relatively stiff dsDNA linkers is based on our own earlier demonstration that dsDNA linkers can provide space between the cover glass surface and MTs without significantly affecting the motility driven by kinesin-1 monomers (19). Consistent with this previous work, WT monomers that are NL-tethered using a 40 bp (~14 nm) dsDNA drove plus-end-directed MT gliding at 38 nm s⁻¹ (Fig. 2B, L-N and Table 1 and Movie S1). Truncating the dsDNA to 20 bp (~7 nm) did not change the MT-gliding velocity (SI Appendix, Fig. S2). Cover strand–tethered WT drove plus-end-directed MT gliding with a velocity of 10 nm s⁻¹ (Fig. 2C, L-N). A previous study measured MT gliding by Drosophila kinesin-1 truncated WT monomers fused at their N termini to GST and found plus-end-directed motility, but at only ~1% of that driven by NL-tethered truncated monomers (3). In our experiment, lawns of cover strand–tethered WT monomers drove plus-endward MT gliding at velocities that were approximately 25% of those produced by NL-tethered WT monomers, presumably because of improved geometry in the motility assays. Nitta and colleagues found that gliding velocity driven by an N terminus–tethered kif1A (kinesin-3) mutant (C351-N) was approximately one-third that of C terminus–tethered kif1A mutant (C351-C) (see their figure 7a) (20).

Fig. 2.

Motor activity of NL-, cover strand–, and loop-tethered kinesin motor domains. (A) Scheme of the in vitro polarity-marked MT-gliding assay. Single-headed kinesins tethered to a biotinylated dsDNA (40 bp) are anchored to biotinylated BSA via biotin–streptavidin. (B–K) Typical kymographs of polarity-marked MT gliding driven by (B) NL-tethered WT, (C) cover strand (CS)-tethered WT, (D) L1-, (E) L2-, (F) L3-, (G) L5-, (H) L6-, (I) L8-, (J) L10-, and (K) L12-tethered WT are shown. L8- and L12-tethered WT were anchored to a glass surface via PEG and others via dsDNA. NL-, cover strand (CS)-, L1-, L2-, L6, and L10-tethered WT glide MTs with their bright minus-ends leading, indicating plus-end-directed motor activity, whereas L3-, L5-, and L12-tethered WT glide with their dim plus-ends leading, indicating minus-end-directed motor activity. (L) Typical trajectories of polarity-marked MT gliding driven by NL-, cover strand (CS)-, and loop-tethered monomers. (M) Beeswarm plots of velocities of MT gliding as driven by NL-, cover strand (CS)-, or loop-tethered WT monomeric kinesins. Horizontal black lines indicate means; error bars indicate SD. (N) MT-gliding velocities versus tether position. The color (scale bar) indicates the MT-gliding velocity. The plus (+) and minus (−) signs refer to the plus-end-directed and minus-end-directed motor activity. Means ± SD. values and sample sizes are tabulated in Table 1.

Table 1.

MT-gliding velocities and MT-activated ATPase activities

Construct	Linker	Directionality	Gliding velocity (nm s−1)	k cat (s−1 per head)	Km (50% MT) (µM)
Cover strand-tethered WT	dsDNA	+ (148/148)	10.0 ± 1.1 (148)	57 ± 2.0	0.6 ± 0.1
L1-tethered WT	dsDNA	+ (112/125)	3.6 ± 0.7 (112)	41 ± 1.8	1.6 ± 0.3
L2-tethered WT	dsDNA	+ (111/113)	7.4 ± 0.7 (111)	40 ± 1.9	1.0 ± 0.2
L3-tethered WT	dsDNA	− (108/114)	−0.4 ± 0.1 (108)	54 ± 2.1	1.2 ± 0.2
L5-tethered WT	dsDNA	− (100/104)	−0.1 ± 0.04 (100)	45 ± 2.1	0.9 ± 0.2
L6-tethered WT	dsDNA	+ (104/112)	0.5 ± 0.1 (104)	65 ± 1.6	0.6 ± 0.1
L8-tethered WT	PEG	+ (70/84)	2.8 ± 1.0 (70)	41 ± 1.1	1.5 ± 0.2
L10-tethered WT	dsDNA	+ (109/110)	4.3 ± 0.6 (109)	41 ± 1.8	1.6 ± 0.3
L12-tethered WT	PEG	− (37/38)	−0.3 ± 0.2 (37)	27 ± 1.1	1.6 ± 0.3
NL-tethered WT	dsDNA	+ (107/107)	38.1 ± 4.4 (107)	47 ± 1.7	0.8 ± 0.1
Cover strand-tethered IKN	dsDNA	+ (136/138)	1.1 ± 0.1 (136)	20 ± 0.5	0.5 ± 0.1
L2-tethered IKN	dsDNA	+ (117/120)	2.7 ± 0.4 (117)	19 ± 0.7	0.9 ± 0.1
L10-tethered IKN	dsDNA	+ (131/135)	0.6 ± 0.1 (131)	20 ± 0.6	0.9 ± 0.1
NL-tethered IKN	dsDNA	+ (115/115)	3.9 ± 0.5 (115)	22 ± 0.6	0.5 ± 0.1
Cover strand-tethered del8	dsDNA	+ (119/126)	8.1 ± 0.7 (119)	56 ± 2.3	0.5 ± 0.1
NL-tethered del8	dsDNA	+ (105/108)	41.6 ± 2.1 (105)	46 ± 1.3	0.7 ± 0.1

Values of the denominator in parentheses are the number of MTs analyzed, and values of the numerator are the number of polarity-marked MTs which moved as indicated (+ or −). Only MTs that were >1 µm in length and did not cross each other were analyzed. Gliding velocity data are given as mean ± SD. The plus (+) and minus (−) signs refer to MT plus-end- or minus-end-directed motion of the motor. MT-activated ATP turnover rates (k_cat) and apparent Michaelis–Menten constants (K_{m. 50% MT}) were determined by the best fit to a hyperbolic curve at varying polymerized tubulin concentrations. Values are given as mean ± SEM.

Loop-Tethered Kinesin-1 Monomers Drive Robust MT Gliding.

To test whether kinesin motor domains can generate directional force independently of NL or cover strand tethering, we examined the motility of loop-tethered kinesin-1 motor domains (Fig. 2). We introduced a cysteine residue into loop 1 (K28C), loop 2 (Q43C), loop 3 (N56C), loop 5 (D102C), loop 6 (E125C), loop 8 (A169C), loop 10 (T220C), or loop 12 (T273C) of the cysteine-light kinesin-1 monomer (Fig. 1) and reacted these with the maleimide moiety of a DNA oligomer. Reaction efficiency was 30 to 70%, except for loop 8 (A169C) which was unreactive (SI Appendix, Fig. S1), probably because the cysteine residue is buried in the motor core. For lawns of L12-tethered WT in an in vitro gliding assay, in the presence of nonhydrolyzable ATP (AMPPNP), fluorescent MTs were observed attached to the kinesin-coated surface, while in the presence of ATP, no MTs were observed, likely because attaching the DNA tether at loop 12 sterically blocks the binding of kinesin to tubulin. We therefore utilized a smaller and more flexible molecule, PEG, as an alternative tether at loops 8 and 12 (SI Appendix, Fig. S1B). Loop 12 (known as the K-loop in kinesin-3) is intimately involved in MT binding (21–24).

For all the loop-tethered kinesin motor domains, ATPase activity measurements and MT-gliding assays were conducted. ATPase activities for kinesins with inserted cysteines that were not coupled to dsDNA did not differ greatly (Table 1 and SI Appendix, Fig. S3). Coupling to dsDNA did not substantially change ATPase activities, though K_m values were higher compared to the corresponding uncoupled kinesins, indicating reduced MT affinity (SI Appendix, Fig. S4). All the loop-tethered motor domains we examined drove MT gliding (Fig. 2 D–L and Movie S1). The fastest gliding velocity we observed was with the motor domain tethered via loop 2 (L2-tethered WT) (Table 1 and Fig. 2 M and N). MT-gliding velocity driven by L2-tethered WT was approximately 90% of cover strand–tethered WT and 20% of NL-tethered WT. L2-tethered WT labeled with PEG instead of dsDNA also drove MT gliding, although the gliding velocity was approximately half that of L2-tethered WT labeled with dsDNA (SI Appendix, Fig. S5). Our finding that loop-tethered kinesin motor domains drive MT gliding clearly demonstrates an underlying force-generating mechanical cycle in the catalytic motor core that is independent of both the C-terminal NL stroke and the CNB formation.

Certain Loop-Tethered Kinesin-1 Monomers Reverse the Direction of MT Gliding.

Among loop-tethered monomeric kinesins (Fig. 2A), we found that L1-, L2-, L6-, L8-, and L10-tethered WT were plus-end-directed motors (Fig. 2 D, E, H–J, and L–N and Movie S2), consistent with earlier evidence that plus-end-directionality is intrinsic to the catalytic motor core of kinesin superfamily members in general (25–28). By contrast, MT gliding driven by L3-, L5-, and L12-tethered WT was minus-end-directed (Fig. 2 F, G, K, and L). Minus-end-directed gliding was robust, but very slow (Fig. 2 M and N). ATPase activities for L3-, L5-, and L12-WT with unlabeled dsDNA, measured in solution, were not greatly reduced (Table 1 and SI Appendix, Fig. S3). Our findings indicate that the direction of MT movement driven by lawns of the same monomeric kinesin motor domain can be reversed by tethering to a specific subset of surface loops.

NL Mutations Inhibit MT Gliding for NL-, Cover Strand–, and Loop-Tethered Motor Domains.

A large body of previous work has shown that intact NL function is crucial for kinesin-driven motility. We investigated the role of NL docking in the mechanical cycle of the loop-tethered motor core. Loop tethering disconnects the NL so that its dock–undock cycle cannot directly drive MT gliding. We engineered NL mutations in NL-, cover strand–, L2-, and L10-tethered WT kinesin monomers (Fig. 1 A and B). The IKN mutation disrupts the residue contacts that ordinarily stabilize the docked state of the NL (11). We confirmed that NL-tethered IKN, a triple mutant (IKN>AAA) in the NL close to its junction with the α6 helix of the catalytic motor core produced inhibition of motility (Fig. 3A), as previously reported (11). We found that, in all other cases, MT-gliding velocities driven by IKN mutants were approximately 10 to 30% compared to the corresponding tethered kinesin constructs (Fig. 3 B–D, G, and H, Table 1, and Movie S3) while the ATP turnover rates (k_cat) in solution were approximately 30 to 50% and the Michaelis–Menten constants (K_m, _{50% MT}) were approximately 50 to 90% (Table 1 and SI Appendix, Fig. S3). The slightly tighter binding to the MT of the IKN mutants was accompanied by reduction of the k_cat and a dramatic decrease of MT-gliding velocity. These data confirm that a docking-competent NL is required to allow lawns of kinesin-1 monomers to drive rapid MT gliding. Our data extend the earlier work (11) by showing that NL docking remains functionally crucial even when the NL no longer connects the motor to its cargo or substrates.

Fig. 3.

Mutations that disrupt neck-linker docking inhibit the motor activity of NL-, cover strand-, and loop-tethered motor domains. (A–D) Typical kymographs of polarity-marked MT gliding driven by (A) NL-tethered IKN, (B) cover strand (CS)-tethered IKN, (C) L2-tethered IKN, (D) L10-tethered IKN, (E) NL-tethered del8, and (F) cover strand (CS)-tethered del8, anchored to a glass surface via dsDNA. (G) Typical trajectories of polarity-marked MT gliding. (H) Beeswarm plots of velocities of MT gliding. Horizontal black lines indicate means; error bars indicate SD. The values and sample sizes are tabulated in Table 1.

Cover-Strand Truncation Has Only Minor Effects on MT Gliding Driven by NL- and Cover Strand–Tethered Motor Domains.

To probe the role of CNB formation in the plus-endward mechanical stroke of the kinesin-1 motor domain, we engineered mutant kinesins that lacks most of the N-terminal cover strand (deleting residues 2 to 8). We compared MT-gliding velocities for NL-tethered del8 and cover strand–tethered del8 (Fig. 1 A and B) with those driven by wild-type NL- and cover strand–tethered motor domains. For NL-tethered kinesins, cover strand truncation increases gliding velocity by ~10% while for cover strand–tethered kinesins, cover strand truncation reduces gliding velocity by ~20%. Both effects are slight, but significant (Fig. 3 E–H and Table 1). ATPase activities were not detectably affected (Table 1 and SI Appendix, Fig. S3). We interpret our findings to indicate that for lawns of WT motor domains under our conditions, CNB formation contributes only modestly to the mechanical stroke that drives MT gliding (Discussion).

Subtilisin Cleavage of the Tubulin E-Hook Modulates MT Gliding in All Three Tethering Schemes.

The C-terminal E-hook domain of β-tubulin interacts electrostatically with the kinesin motor domain, in particular via loop 12 (29, 30). These interactions tend to constrain tethered diffusion of weakly bound kinesin heads into 1D, so supporting faster diffusional scanning and larger diffusional excursions. To test whether these E-hook interactions differentially affect NL- and loop-tethered (L2 and L10) kinesins, we cleaved the E-hooks with subtilisin and remeasured MT-gliding velocities (SI Appendix, Fig. S6). Gliding velocity for NL-tethered WT slightly increased by approximately 10% of intact MTs, while velocities for L2- and L10-tethered WT decreased to approximately 60 to 70%. Possibly, dsDNA tethering positions at the NL, close to loop12, might inhibit the E-hook/L12 interaction due to the charged dsDNA, whereas tethering via L2 or L10 locates the dsDNA further from the L12 and so might interfere less.

Discussion

The experiments reported here explore the fundamental force-generating mechanism of lawns of tethered kinesin monomers. We have three salient findings. First, our tether-scanning approach shows that loop anchoring drives slow but robust MT gliding, revealing an underlying, presumably ancient, mechanical cycle of the motor core that is harnessed and amplified by the natural N- or C-terminal tethers. Second, we find that depending on the anchoring position, otherwise-identical kinesin-1 monomers can exhibit either plus- or minus-end-directed motility. Third, we find that even if the NL is not connected in line between the catalytic motor core and the glass substrate, mutations that influence NL docking reduce the MT-gliding velocity. What do these findings tell us about the mechanism by which lawns of monomeric kinesins generate force?

A Basal Force-Generating Cycle of the Motor Core.

By showing that loop-tethered monomeric kinesin drives MT gliding (Fig. 2), our data reveal that the catalytic motor core has its own mechanical cycle and that this cycle drives forceful motions of exposed surface loops. Our tethering scheme connects to these loops and harnesses their motions to drive MT gliding. MT-gliding velocities for L1-, L2-, and L10-tethered WT were ~10 to 20% of that of NL-tethered WT (Fig. 2N and Table 1). The kinesin motor domain binds within α/β tubulin dimers (31, 32) such that loops 1 and 2 are located closer to the α-tubulin (toward the minus end of the MT), while loop 10 is located closest to the plus end of the MT and further from the MT surface (23, 24) (Figs. 1A and 2N). Tethering from these loops is therefore not expected to interfere with MT-binding or nucleotide turnover. Our data unambiguously demonstrate an underlying force-generating conformational cycle of the catalytic motor core of kinesin.

CNB Formation Plays a Minor Role in Driving MT Gliding At Low Loads.

MT-gliding velocity driven by cover strand–tethered WT was 25% of that of NL-tethered WT (Fig. 2 M and N and Table 1). Crystal structures and molecular dynamics simulations show that the N-terminal cover strand forms part of the docking site for the NL, implying that the cover strand undergoes an undocked-to-docked transition, coincident with that of the NL (18). The cover strand anneals to the outside surface of the proximal part of the docked NL. Formation of this CNB has been proposed to constitute kinesin’s power stroke (33). However, cover strand-tethered del8, which lacks the N-terminal 2-8 residues (β0, cover strand), can no longer form a CNB, yet in our assays, this deletion has only minor effects on MT-gliding velocity (Fig. 3H and Table 1). Under our conditions, therefore, for both loop-tethered and linker-tethered motor domains, docking of the cover strand appears to play a minor role in the generation of directional force for MT-gliding motion. This is consistent with work showing that while CNB formation is required for the stepping of kinesin dimers against substantial loads, it is not required for directional stepping at low load (16).

Loop Tethering Reveals Both Plus- and Minus-End-Directed Components of the Basal Mechanical Cycle.

L3-, L5-, and L12-tethered WT kinesin-1 motor domains show minus-end-directed motor activities, reversing the usual polarity of kinesin-1 monomer surfaces (Fig. 2 F, G, K, and L). Gliding was robust, but very slow (<1 nm s⁻¹) (Fig. 2 M and N and Table 1). For loop tethering in general, it is possible that directional tension exerted via the tether influences the binding affinity for MTs, as it does for WT kinesin (34, 35). For the direction-reversing linker-attachment positions in loops L3, L5, and L12, the attachment points are positioned directly above the nucleotide pocket near switch I (for loops 3 and 5) or in switch II (for loop 12), close to the center of the MT interface. Exactly why the intrinsic directionality of the motor core is reversed for this set of loops is unclear (see below), but our data do unequivocally demonstrate that the potential to be a minus-end-directed motor is present in the core kinesin motor domain.

NL Docking Controls ATPase Activity and Detachment from the MT.

Our data (Fig. 3) show that although NL connectivity is not required for the motor core to be able to drive MT gliding, rapid gliding does require that the NL is competent to dock and undock, even when it is mechanically disconnected. To explain this, we speculate that inhibiting NL docking causes drag, due to posthydrolysis heads failing to release promptly from the MT. As previously noted (6), when the kinesin head is attached to MTs, the docking station for the NL lies almost exactly parallel to the MT longitudinal axis (Fig. 1A). Measurements made by dragging kinesin-coated beads along MTs (36) show that the detachment rate constant of ADP-bound kinesin heads depends acutely on the direction and magnitude of the force experienced by the head such that a minus-end-directed pull prolongs binding. The alignment of the axis of the docked NL with the MT axis, combined with the property that NL docking accelerates unbinding from the MT, predicts that axial load on the NL will bias both MT binding and MT unbinding.

A Model for MT Gliding Driven by Teams of Kinesin Monomers.

How do lawns of surface-loop tethered kinesin motor domains generate force? Directional force might originate in conformational changes of the motor domain, or in directionally biased binding, or in directionally biased unbinding. Our data suggest a scheme with elements of all three processes.

Structural studies have identified a conformational cycle of the kinesin motor domain consisting of a sequence of subdomain motions, driven by ATP turnover and by MT binding. The detailed boundaries of kinesin subdomains are assigned slightly differently in different studies, but a central feature is that subdomain motion opens or closes a hydrophobic pocket between the C termini of the α4 and α6 helices of the motor domain. Occupancy of this pocket by a conserved IKN triplet at the root of the docked NL wedges the subdomains apart and allosterically stabilizes the active site in a hydrolytically competent conformation. The same docking event also stabilizes the K.ADP state of the active site (37), thereby provoking MT release.

The ATP-driven cycle of subdomain motions also shifts the positions of surface loops, with different loops moving differently. Much of the motion is a rocking perpendicular to the longitudinal axis of the MT, but there is also an on-axis component of rocking (23, 24, 38–40) (Movie S4). It is possible that these surface loop motions themselves supply motive force. We find however no obvious correlation between the direction and amplitude of these surface loop motions (SI Appendix, Fig. S7) and the speed and direction of MT gliding. Instead, we propose that MT gliding is predominantly due to biased binding and unbinding of the motor to the MT (Fig. 4A).

Fig. 4.

Strain-sensitive asymmetric binding/unbinding mechanism for directional motion of the single-headed kinesin head, depending on the tethering position in the motor domain. Key mechanical events for (A) WT (B) loop-tethered kinesin motor domains. (i) The kinesin head undergoes tethered diffusion; NL (orange) is docked on average in the detached ADP-bound head (8, 41). (ii) Initial diffusion to capture by the MT, with a bias toward the plus end. (iii) Following binding, backward (minus-endward) strain acts via either (A) the NL or (B) a surface loop. The backward strain accelerates ADP release, which stabilizes the motor head in a strong binding conformation and produces a slight displacement. The nucleotide cleft between p-loop and switch II loop is open, while the docking cleft is closed, with the NL disordered (24). NL docking is also disfavored because the NL does not have the conformational freedom to anneal into its docking station. (iv) MT gliding, driven largely by the actions of other engaged motors. (v) ATP binding and processing are favored once the NL has the conformational freedom to dock. (vi) NL docking in the ADP state triggers prompt detachment from the MT. Return to state (i). (vii) Kinesin heads that do not promptly dock the NL (for example, our NL mutants) overstay strong binding and are dragged by continuing MT gliding into a negatively strained state. These drag heads ultimately have to be pulled off by the action of other heads. An equivalent mechanical cycle with L3-, L5-, or L12-tethering strains the tether in the opposite direction (not shown).

In biased binding models for lawns of NL-tethered WT kinesin monomers, thermally activated tethered diffusion (42) carries the motor domain to high positive offsets (43, 44), where it is captured by the MT and transitions into stereospecific binding, with coupled ADP release (45). Diffusion to capture in the plus end direction is favored because backward load on the NL tends to inhibit its docking, strengthening MT binding (34) and inhibiting nucleotide binding and processing (36, 46–48). Conversely, diffusion to capture in the minus-end direction is disfavored, because minus-end-directed excursions tend to favor, rather than disfavor, NL docking.

For our lawns of loop-tethered kinesin monomers (Fig. 4B), docking of the IKN sequence of the disconnected NL is expected to continue to stabilize the hydrolytically competent state of the active site and to stabilize the detaching trapped-ADP state, thereby reducing drag. Both effects will drive ATP turnover, but there is no reason to think either effect would be directionally biased. To explain directional bias, we speculate that tethered diffusion to capture of the loop-tethered motor core loads the surface loops so as to bias the open-close cycle of the IKN docking station, allowing NL docking to continue to sense the direction of the load acting on the motor domain, by an indirect mechanism (Fig. 4B). The differing speeds and directions of gliding due to tethering by different loops might then relate to differential effects on the IKN docking station. An additional element of biased unbinding might be due to a differential effect of directional load on the MT interface itself.

Whether or not this model is correct, our results reveal a potentially ancient core mechanical cycle of the kinesin motor domain that is amplified by the dock cycles of the WT NL and cover strand. Our data reaffirm linker docking as a central player in the kinesin mechanism, both for force generation and as a sensor and controller of biased unbinding, while clarifying that linker docking is an amplification strategy that leverages a more basal, underlying, potentially ancestral mechanical cycle that can itself drive robust directional motion.

Materials and Methods

Construction of Plasmids.

To make “cys-lite” rat kinesin 340-residue monomers, solvent-exposed cysteines (C7S, C66S, C169A, and C296A) of a rat kinesin (R. norvegicus KIF5C) (17) rk340-His₆ construct were mutated to Ser/Ala. Buried cysteines at positions C13, C303, and C304 were retained. To allow tether attachment, single surface-exposed residues in this cysteine-light rk340-His₆ construct were mutated to cysteine. Residues mutated were at the N terminus (2C), in loop 1 (K28C), loop 2 (Q43C), loop 3 (N56C), loop 5 (D102C), loop 6 (E125C), loop 8 (A169C), loop 10 (T220C), and loop 12 (T273C), and at the C terminus (341C). For del8 deletion constructs, cysteines were introduced at the new N terminus (8C) or at the C terminus (341C) of the cys-lite rk340 sequence, after deleting residues 2 to 8 of the cover strand. For IKN mutant constructs, 3 residues within the NL of cys-lite rk340-His₆ were mutated to alanine (I327A-K328A-N329A), and cysteines introduced at the N terminus (2C), in loop 2 (Q43C), or loop 10 (T220C), or at the C terminus (341C). All constructs were cloned into the pColdIII vector (Takara, Shiga, Japan) and were confirmed by DNA sequencing.

Expression and Purification of Kinesins.

Wild-type and mutant cys-lite rat kinesin-1 monomers were expressed in Escherichia coli and purified as described previously (28). Expression plasmids for His-tag-fused monomeric kinesins were transformed into E. coli strain BL21 Star (DE3) cells (Invitrogen). After expression in BL21 Star (DE3) cells using 0.1 mM IPTG (Wako, Osaka, Japan) for 24 h at 15 °C for pColdIII vectors, cells were pelleted then resuspended in lysis buffer (pH 7.4: 80 mM PIPES-KOH, 1 mM MgCl₂, 1 mM EGTA, 100 µM ATP (Sigma-Aldrich), 1 mM DTT (Roche), 0.1% CHAPS (Dojindo, Kumamoto, Japan), 0.1% Tween20 (Nacalai, Kyoto, Japan), 10% glycerol, protease inhibitors, and 500 mM NaCl) and sonicated for 20 min on ice. Then, the lysate was clarified by centrifugation (20 min, 305,000 g, 2 °C). Expressed His-tagged proteins were purified by immobilized metal affinity chromatography using a HisTrap HP column (Cytiva). Pooled fractions were exchanged into a desalting buffer (80 mM NaCl, 20 mM potassium phosphate, 1 mM MgCl₂, 1 mM DTT, and 20 µM ATP, pH 7.4) using a HiTrap Desalting column (Cytiva) and then further purified by affinity binding to polymerized MTs. Taxol-stabilized MTs were mixed with purified proteins supplemented with 1 mM AMPPNP (Sigma-Aldrich) and incubated for 15 min at 25 °C. After removing the unbound proteins by centrifugation (20 min, 305,000 g, 23 °C), the MT-bound kinesin was eluted with ATP-containing buffer (10 mM ATP, 10 mM MgCl₂, 200 mM potassium acetate, 20 µM taxol (Wako, Osaka, Japan) in M buffer [20 mM PIPES-KOH, 4 mM MgCl₂, 10 mM potassium acetate, and 1 mM EGTA, pH 7.4]) by incubation for 10 min at 25 °C. MTs were finally removed by centrifugation (20 min, 305,000 g, 23 °C).

Labeling Kinesin with dsDNA or PEG.

A 40-base-pair dsDNA oligomer was synthesized with a primary amine at the 5′ end of one strand (5′-amine- CGTTG GACAC GACAG GTCCT TAGAC TGAAA TTAGT TAGTA-3′) and biotin at the 5′ end of one strand (5′-biotin- TACTA ACTAA TTTCA GTCTA AGGAC CTGTC GTGTC CAACG -3′) (Integrated DNA Technologies). We affixed these labels to the DNA using hydrocarbon linkers (6 carbon atoms long) intended to serve as swivels. For L8- and L12-tethered WT, biotin-PEG6-maleimide (Tokyo Chemical Industry, Tokyo, Japan) was used instead of a short DNA oligomer. DNA was activated with sulfo-SMCC (Pierce) before incubation with kinesin. A solution of 100 µM DNA and 2 mM sulfo-SMCC was incubated for 1 h at 37 °C in 25 mM potassium phosphate buffer (pH 7.2) containing 100 mM NaCl. The sample was passed over a desalting column (Zeba, Pierce) three times to remove unreacted sulfo-SMCC. A mixture of ~20 µM kinesin and ~60 µM DNA or ~350 µM PEG was incubated for 1 h at 27 °C. ATP was added to bring the concentration to ~50 µM. We quenched the reaction with 1 mM reduced cysteine (Tokyo Chemical Industry, Tokyo, Japan). Labeled kinesin molecules were purified by using an Amicon Ultra filter (Merck) with a 30 or 50 kDa cutoff for 15 min at 14,000 g and the resulting volume was made up to ~40 µL. The concentrations of kinesins were estimated by SDS-PAGE on 10% acrylamide gels using BSA standards (Thermo Fisher Scientific) loaded on the same gel (49). Gels were stained with Quick-CBB PLUS (Wako, Osaka, Japan) and imaged using a CCD camera (CSFX36BC3, Toshiba-teli, Tokyo, Japan). The bands containing kinesins and BSA standards were quantified using ImageJ (NIH) (50).

Purification of Tubulin.

Tubulin was purified from the porcine brain through four cycles of temperature-regulated polymerization and depolymerization in a high molarity PIPES buffer to remove contaminating MT-associated proteins (51). Purified tubulin was flash-frozen and stored in liquid nitrogen.

Polarity-Marked MT-Gliding Motility Assays.

Cy5-labeled polarity-marked MTs were prepared as described (52). First, bright short MTs (labeled tubulin: unlabeled tubulin = 1: 2) were polymerized with 0.5 mM GMPCPP (a nonhydrolyzable GTP analog, Jena Bioscience, Germany); then, plus-end elongation of dim long segments (labeled tubulin: unlabeled tubulin = 1: 9) was achieved by inclusion of NEM treated tubulin which inhibits minus-end polymerization. MT-gliding assays were performed in flow chambers assembled from a KOH-cleaned cover glass (24 × 36 mm and 18 × 18 mm, Matsunami Glass. ind., Osaka, Japan) attached using double-sided tape (Scotch W-12, 3M) as described (52). For motility assay, 1 flow chamber volume (5 µL) of diluted (20 k-fold) microbeads (0.1-µm diameter carboxylate-modified polystyrene microbeads (FluoSphere, red fluorescent [580/605]), Thermo Fisher Scientific) was introduced into the flow chamber for drift compensation. Then, 1 flow chamber volume of 5 mg mL⁻¹ biotinylated-BSA (Sigma-Aldrich) was introduced into the flow chamber, incubated for 5 min, and then rinsed with 3 volumes of M buffer. The flow chamber was filled sequentially with 1 volume of 5 mg mL⁻¹ streptavidin (Wako, Osaka, Japan) for 5 min, 3 volumes of 0.4 mg mL⁻¹ casein (Nacalai, Kyoto, Japan) for 30 s and 1 volume of either 1 µM DNA-tethered monomeric kinesins or 1 µM PEG-tethered monomeric kinesins for 5 min and then 3 volumes of ~10 µg mL⁻¹ Cy5-labeled polarity-marked MTs in BRB80 buffer (80 mM PIPES-KOH, 1 mM MgCl₂, and 1 mM EGTA, pH 6.8) containing 0.4 mg mL⁻¹ casein and 100 µM taxol. Finally, 5 volumes of motility buffer (M buffer containing 3 mM ATP, 100 µM taxol, ATP regeneration system, the oxygen scavenger system, 1 mM DTT, and 0.4 mg mL⁻¹ casein) were applied to the chamber. All chambers were sealed with grease (Apiezon M Grease, M&I Materials, Manchester, UK). Assays were carried out at 25 ± 1 °C. MT gliding was observed using a fluorescence microscope (Eclipse Ti-E equipped with Perfect Focus System, Nikon) with a stable stage (KS-N, Chukousya Seisakujo, Tokyo, Japan) and a stage controller (QT-CM2-35, Chuo Precision Ind., Tokyo, Japan), using illumination from a LED light source (D-LEDI, Nikon), 100 × /1.49 NA, Plan-Apochromat objective lenses (Nikon) and Cy5 filter set (Semrock). Images were recorded by EM-CCD camera (iXon X₃ DU-897E-COO-#BV, Andor Technology). Temperature-control equipment (53) (F-12, Julabo, Germany) was used to suppress positional drift, which is mainly caused by temperature variations. In the absence of dsDNA/PEG or streptavidin, no MTs were observed, confirming that only dsDNA/PEG-tether kinesin anchored to the cover glass via biotin–streptavidin. In a preliminary screening for conditions that support MT gliding, we found large differences in gliding velocity between constructs. We adjusted observation times to track gliding MTs over approximately equal distances. Constructs with velocities exceeding 8 nm s⁻¹, namely NL-, cover strand–tethered WT and NL-, cover strand–tethered del8 were observed for 600 s. Those with velocities between 3.8 and 8 nm s⁻¹, namely L2-, L10-tethered WT, and NL-tethered IKN, were observed for 2,000 s. All other constructs had velocities below 3.8 nm s⁻¹ and were observed for 4,000 s. The position of each MT was tracked by custom automated tracking software (MARK 2.6) using a two-dimensional Gaussian fitting algorithm (54), and the x and y positions at each frame were exported. Using the tracked positions, the distance from the first point was calculated and MT-gliding velocity was determined by fitting a linear curve to the time-distance plot. Only MTs that were >1 µm in length and did not cross each other were analyzed. For each tethering scheme, the bracketed pairs of numbers in the following list represent N, the total number of polarity-marked MTs observed for that condition, along with the number that glided. Cover strand-tethered WT (150,148), L1-tethered WT (136,125), L2-tethered WT (123,113), L3-tethered WT (115,114), L5-tethered WT (123,104), L6-tethered WT (113,112), L8-tethered WT (87,84), L10-tethered WT (116,110), L12-tethered WT (48,38), NL-tethered WT (115,107), cover strand-tethered del8 (133,126), NL-tethered del8 (109,108), cover strand-tethered IKN (138,138), L2-tethered IKN (130,120), L10-tethered IKN (135,135), or NL-tethered IKN (116,115). A fluorescent bead (FluoSphere, red fluorescent [580/605]) fixed to the cover glass was tracked to distinguish the displacement of MTs by drift from the MT-gliding driven by slow motor activity. Directionality and velocities were determined using measurements from at least three independent assays for each construct.

Measurement of MT-Stimulated ATPase Rates.

The MT-stimulated ATPase activities of individual kinesins were measured using a pyruvate kinase/lactate dehydrogenase-linked assay as described previously (55). These assays used purified cys-lite kinesins with inserted cysteines but without dsDNA coupling. The activity of ATP hydrolysis coupled to NADH oxidation was monitored using 0.1 µM kinesins under the various MT concentrations ranging from 0.15 µM to 35 µM in BRB80 buffer containing 20 µM taxol, 4.5 mM PEP (Sigma-Aldrich), 2.7 mM ATP, 0.3 mM NADH (Sigma-Aldrich), and 2.5% PK-LDH (Sigma-Aldrich). The ATPase activities of individual kinesins were measured from at least three independent experiments under the same conditions. The turnover rate and the Michaelis constant were obtained by fitting the Michaelis–Menten equation to the kinetic data using MATLAB software (MathWorks Inc.).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Movie S1.

Polarity-marked microtubule gliding driven by NL-, or Cover-strand-tethered WT monomeric kinesin-1. Cover-strand (N)-tethered WT (a) and NL (C)-tethered WT (b) drive gliding microtubules with their minus ends leading, indicating a plus-end-directed motor activity. These movies are 2 seconds long at ×300 speed. Each image shows a region 22 × 22 μm. Scale bar: 5 μm.

Movie S2.

Polarity-marked microtubule gliding driven by Loop-tethered WT monomeric kinesin-1. L1-tethered WT (a), L2-tethered WT (b), L6-tethered WT (e), L8-tethered WT (f) and L10-tethered WT (g) drive gliding microtubules with their minus ends leading, indicating a plus-end-directed motor activity, whereas L3-tethered WT (c), L5-tethered WT (d), L12-tethered WT (h) drive gliding with their plus ends leading, indicating a minus-end-directed motor activity. These movies are 2 seconds long at ×2000 (a), ×1000 (b), ×2000 (c), ×2000 (d), ×2000 (e), ×2000 (f), ×1000 (g) and ×2000 (h) speed, respectively. Each image shows a region 22 × 22 μm. Scale bar: 5 μm.

Movie S3.

Polarity-marked microtubule gliding driven by monomeric kinesin-1 with del8 or IKN mutation. Cover-strand (N)-tethered del8 (a), NL (C)-tethered del8 (b), Cover-strand (N)-tethered IKN (c), L2-tethered IKN (d), L10-tethered IKN (e) and NL (C)-tethered IKN (f) drive gliding microtubules with their minus ends leading, indicating a plus-end-directed motor activity. These movies are 2 seconds long at ×300 (a), ×300 (b), ×2000 (c), ×2000 (d), ×2000 (e) and ×1000 (f) speed, respectively. Each image shows a region 22 × 22 μm. Scale bar: 5 μm.

Movie S4.

A morphing animation of monomeric kinesin-1 transitions between Apo-state and ATP-state. The structures of kinesin-1 motor domain in complex with tubulin heterodimers (purple, α-tubulin; green, β-tubulin) were obtained from the Protein Data Bank (PDB) with entries 3J8X and 3J8Y for the Apo- and ATP-like states, respectively. A conformational morph animation was generated in PyMOL between Apo-state and ATP-state, with apo-kinesin shown throughout in a lighter color. The target amino acids used in this study are highlighted in red.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.