Coordination of kinesin's two heads studied with mutant heterodimers

Abstract

A conventional kinesin molecule has two identical catalytic domains (heads) and is thought to use them alternately to move processively, with 8-nm steps. To clarify how each head contributes to the observed steps, we have constructed heterodimeric kinesins that consist of two distinct heads. The heterodimers in which one of the heads is mutated in a microtubule-binding loop moved processively, even when the parent mutant homodimers bound too weakly to retain microtubules in microtubule-gliding assays. The velocities of the heterodimers were only slightly higher than those of the mutant homodimers, although mixtures of these weak-binding mutant homodimers and the WT dimers moved microtubules at a velocity similar to the WT. Thus, the mutant head affects the motility of the WT head only when they are in the same molecule. The maximum force a single heterodimer produced in optical trapping nanometry was intermediate between the WT and mutant homodimers, indicating that both heads contribute to the maximum force at the same time. These results demonstrate close collaboration of kinesin's two heads in producing force and motility.

A conventional kinesin molecule is a homodimer with two catalytic domains (heads) linked together by a coiled-coil, and can move continuously for hundreds of steps along a microtubule (MT) track (1–3). Although monomeric constructs of kinesin can support MT movement in in vitro assays, various lines of evidence have shown that the two-headed structure is important for motility of conventional kinesins. For example, the velocities of monomeric constructs of kinesin are considerably slower than those of dimers (4, 5). Heterodimeric kinesin that lacks one of the two heads (one-headed kinesin) also showed much-decreased MT-gliding velocities (6). In addition, these monomeric or one-headed kinesins showed little or no processivity (4, 5, 7).

To move processively without dissociating from a MT, at least one of the heads is always bound to a MT. For a dimeric kinesin to stay bound to a MT, it is thought that the two heads of a kinesin coordinate their mechanochemical cycles. Previous biochemical studies have indicated the existence of functionally significant coordination between the two heads (8–12). For example, ATP binding by one of the heads triggers ADP release from the other (8–10). Also, the ATPase rate of one-headed kinesin dimers was decreased by almost 10-fold compared with that of two-headed kinesins (11). However, a large part of the mechanism for coordination is still unknown.

With the two heads, kinesin can move continuously for hundreds of nanometers against an applied load. Previous optical trapping experiments have shown that a kinesin dimer moves with discrete steps of 8 nm and can produce force up to several piconewtons (13–16). However, it is not yet known how each of the two heads contributes to the observed force. Comparing the force with that produced by monomeric or one-headed kinesins is also difficult, because monomeric or one-headed kinesins tend to detach from a MT before producing enough force to be measured.

One of the difficulties in studying the roles of each head in force production and motility comes from the fact that the two heads are identical, so that we cannot distinguish the contribution of one head from the other. To circumvent this problem, we have constructed heterodimeric kinesins that consist of two distinct motor domains. Using heterodimers in which one of the heads is mutated, we measured the velocities and the maximum force a single heterodimer produces and compared them with those of the WT and mutant homodimers. The results indicated close collaboration of kinesin's two heads in force production and motility.

Materials and Methods

Expression Constructs.

Heterodimeric kinesins were prepared by coexpressing the WT and mutant K432 (a construct of human conventional kinesin consisting of amino acid residues 1–432), tagged with GST and histidine residues, respectively. First, a pET17b vector (Novagen) containing K432 with a C-terminal eight-histidine sequence (K432-His/pET17b) was made by using a K560-His construct in a pET17b vector (K560-His/pET17b) (17) as a starting material. A DNA fragment encoding K432 with an EcoRI site, the stop codon, and a XhoI site at the 3′ end was amplified by PCR and digested with SacI and XhoI. The resulting fragment, encoding amino acid residues 366–432, was subcloned into K560-His/pET17b digested by the same two restriction enzymes to produce K432/pET17b. An oligonucleotide encoding consecutive histidine residues was introduced into the EcoRI site of K432/pET17b. Thus prepared K432-His/pET17b encodes the heavy chain B (51 kDa) as shown in Fig. 1a.

Fig 1.

Purification of heterodimeric kinesin. (a) Coexpression vector for generating heterodimeric kinesin. The two heavy chains in a kinesin heterodimer contain distinct motor domains attached to two different carrier sequences. The DNA fragments encoding the heavy chains A and B are placed side by side in a pGEX-2T-based vector. (b) The strategy for purifying heterodimeric kinesin. Three kinds of dimeric kinesins (A-B heterodimer, A-A homodimer, and B-B homodimer) form spontaneously via the coiled-coil regions. The heterodimeric kinesin with two different tags is separated from the homodimeric kinesins with two identical tags by the two-step affinity purification. (c) The SDS/PAGE patterns during the purification of WT/WT-His, after purification with glutathione Sepharose 4B/glutathione (lane 1), digestion by thrombin (lane 2), purification with Ni-NTA agarose/imidazole (lane 3), and removal of residual GST and GST-tagged protein with glutathione beads (lane 4). Lane 5 shows the WT dimer prepared by the same method but without digestion of GST (GST-WT/WT-His). The positions of K432 with a GST-tag and with a His-tag are indicated by an asterisk and double asterisks, respectively, and the cleaved GST is indicated by an arrowhead.

A plasmid encoding K432 with an N-terminal GST tag was also prepared by subcloning K432 in a vector carrying GST (pGEX-2T). A DNA fragment encoding the entire sequence of K432 was excised from K432/pET17b by digestion with NdeI and XhoI, rendered blunt with the Klenow fragment, and subcloned into the SmaI site of pGEX-2T (18). To use the same promoters for both K432-His and GST-K432 of a heterodimer, a T7 promoter was substituted for the tac promoter of pGEX-2T. The resulting construct, K432/pGEX-2T, encodes the heavy chain A (77 kDa) in Fig. 1a.

A series of mutant kinesins was prepared by using K432-His/pET17b by QuikChange mutagenesis (Stratagene). To produce the L11 mutant, the codons for K240, L248, and K252 in K432-His/pET17b were all mutated to those for alanine residues. The L12 (Y274A/R278A/K281A), L8 (E158A/K159A/R161A), and L13 (G291A/G292A) mutants were similarly prepared. To obtain heterodimeric kinesins, DNA fragments encoding the WT and mutated kinesins were placed side by side in a vector (Fig. 1a). In brief, K432-His/pET17b-based plasmids were digested by BglII, followed by a Klenow fragment treatment. After digestion by AatII, the fragments were introduced into K432/pGEX-2T treated by BsaAI/AatII. All PCR products and mutations were checked by DNA sequencing.

Protein Expression and Purification.

Proteins were expressed in BLR (DE3) pLys S (Novagen) freshly transformed with the plasmids (17). After being washed with buffer A (10 mM Tris-acetate, pH 8.0/4 mM magnesium acetate/250 mM potassium acetate/0.05% 2-mercaptoethanol), the bacterial cells were resuspended in buffer A supplemented with 1 mM ATP and protease inhibitors (17), disrupted by sonication, and then centrifuged (30,000 × g, 30 min).

To separate heterodimeric kinesins from other species, two-step affinity purification was performed (Fig. 1b). First, 1/40 volume of Glutathione 4B Sepharose resin (Amersham Pharmacia) was incubated with the supernatant on a roller for 1–2 h. The beads were collected by brief centrifugation (30 s at 500 × g) and washed with buffer A. The proteins were eluted with 20 mM glutathione (Fig. 1c, lane 1), and dialyzed against buffer A. GST was then removed by a treatment with 5 units/ml thrombin (Sigma) for 2–6 h in the presence of 2 mM CaCl₂ (Fig. 1c, lane 2). To collect the heterodimers, 1/20 volume of Ni-NTA agarose resin (Qiagen, Chatsworth, CA) was incubated with the resultant solution for 1–2 h. The beads were washed with buffer A containing 30 mM imidazole, and the proteins were eluted by applying 200 mM imidazole to the beads (Fig. 1c, lane 3). Finally, to remove a trace of residual GST-tagged kinesin and GST, Glutathione 4B Sepharose beads were incubated with the protein solution for 30 min, and centrifuged briefly. The supernatant was then concentrated (Fig. 1c, lane 4), and aliquots were stored in liquid nitrogen. To avoid loss of activity, the protein was kept at 4°C and in 100 μM ATP at all times. All of the mutant dimers were prepared and assayed in parallel with the WT kinesin. The histidine-tagged homodimeric kinesins were purified by using the Ni-NTA agarose/imidazole procedure. The concentration of the protein was determined by the Bradford method (19). Native PAGE was performed at pH 8 (20).

ATPase Assay.

Steady-state MT-stimulated ATPase rates were measured by using malachite green (21). The motors (typically 40–200 nM) and MTs (0.1–50 μM tubulin dimers) were mixed at 27 ± 0.1°C in 10 mM Tris-acetate, pH 7.5/2 mM Mg-acetate/1 mM EGTA/1 mM Mg-ATP/20 μM paclitaxel. The ATPase assay was repeated three to six times with at least two different kinesin preparations. The ATPase data were fit to the Michaelis–Menten equation to determine the K_m(MT) and k_cat values.

Multiple Motor Motility Assay.

A Penta-His Antibody (Qiagen) was fixed on the nitrocellulose-coated surface of a coverslip. The chamber was washed with 1 mg/ml BSA to remove the unbound antibody and then filled with 25 mg/ml BSA. The histidine-tagged kinesin (0.05–3 μM) was added into the flow chamber. After removal of unbound kinesin, tetramethylrhodamine-labeled MTs were introduced. Finally, the chamber was filled with the assay buffer [10 mM Tris-acetate, pH 7.5/4 mM Mg-acetate/50 mM K-acetate/1 mM EGTA/20 μM paclitaxel/0.5 mg/ml casein/0.5% 2-mercaptoethanol/1 mM ATP, and the oxygen scavenger system (22) at 25 ± 1 or 30 ± 1°C]. MTs were visualized under a fluorescent microscope equipped with a silicon-intensified target (SIT) camera (C2400-08; Hamamatsu Photonics, Hamamatsu City, Japan). Two to four independent batches of protein preparation were analyzed for each construct.

Optical Trapping Nanometry.

The apparatus of trapping nanometry was as described (23). For specific attachment to latex beads, a sequence for biotinylation was added at the C terminus of heavy chain A (Fig. 1a) (24). The oligonucleotides encoding a sequence LGSIFEAQKIEWR (25), which is biotinylated in Escherichia coli, was introduced into the EcoRI site of K432/pGEX-2T. The biotinylated kinesins, purified as described above, were incubated with streptavidin-coated 0.2-μm beads for >20 min at 25 ± 1°C. After tetramethylrhodamine-labeled MTs were bound directly to a coverslip, a kinesin-coated bead was placed in contact with a MT by using an optical trap. The displacement of kinesin was corrected by multiplying the factor [(K_t + K_p)/K_p] to the observed bead displacement, where K_t and K_p are the stiffness of the optical trap and the stiffness of bead-to-glass linkage, respectively (13, 26). The total stiffness (K_t + K_p) and the trap stiffness K_t were determined by using the equipartition method from the Brownian noise of the bead movement (13, 27). After high-pass filtering at 10 Hz, the SD values were calculated at every 0.1 s of individual traces, grouped according to bead displacement, and averaged to determine the stiffness (K_t + K_p) at given displacement values. The range of the correction factor used was 1.1–1.7.

The experiments were performed in 80 mM Pipes, pH 6.8/2 mM MgCl₂/1 mM EGTA/1 mM ATP, and an oxygen scavenging system (22), at 24–26°C. The concentration of kinesin (5–50 nM, unless indicated otherwise) was determined so that the probability of the kinesin-coated beads to move on a MT was 0.15–0.35. Under this condition, it is considered that only a single kinesin molecule is involved in the movement of a bead (14, 26). To investigate the force–velocity relationships, the velocities were calculated at several points in the time course of the displacement as in Fig. 4a, from the slope of the trace. The force–velocity data sets from different traces were grouped according to the force. At every 0.5 pN (for the WT) or 0.15 pN (for the mutant), the velocity and force were averaged within the group. The averaged velocities were plotted against the averaged force.

Fig 4.

Optical trapping nanometry for measurement of stall force. (a) Representative traces of the displacement of kinesin-coated beads along a MT. The displacement of the bead and the corresponding force are indicated on the left. For WT/L12 and L12/L12, the scale of displacement is different, because a weaker trap stiffness was used (0.007 pN/nm, instead of 0.032 pN/nm used for others). Applied force is shown on the same scale but the time scale is different in each trace. In the case of L12/L12, a higher concentration of kinesin (1–2 μM instead of 5–50 nM) was needed to observe any attachment signals (reduced vibration, as indicated by the solid line). In this case, more than one molecule is expected to be involved. (b) The force–velocity relationships of the WT kinesin (black, filled triangle), WT/L8 (blue, filled square), L8/L8 (blue, open square), WT/L11 (green, filled diamond), L11/L11 (green, open diamond), WT/L12 (red, filled circle), and L12/L12 (red, open circle). Solid lines represent the best fit of the data, and the y axis and x axis show the maximum velocity and the stall force, respectively.

Results and Discussion

Preparation of Heterodimeric Kinesins.

We have produced heterodimeric kinesins, which possess two distinct heads, by coexpressing the WT and mutant kinesin heavy chains attached to different tags: consecutive histidine residues (His) and GST (Fig. 1a). Heterodimers (A-B heterodimer in Fig. 1b) were separated from other species (A-A and B-B homodimers), first by using Glutathione Sepharose 4B, then with Ni-NTA agarose, and further purified with glutathione beads. Thus purified heterodimeric kinesin appeared as a single band by SDS/PAGE/Coomassie Brilliant blue staining (Fig. 1c, lane 4), because the A and B heavy chains after removal of GST have similar molecular weights. When GST was not removed, the SDS gel showed two bands (Fig. 1c, lane 5), corresponding to the heavy chain A attached to GST (marked by an asterisk) and the heavy chain B with histidine residues (double asterisks). Densitometry of the gel showed that these heavy chains are present in a proportion of one to one. The results indicate that the purified protein is indeed heterodimeric, and that there is no detectable contamination by homodimers.

We have also examined the possibility of recombination that might occur after the purification; native PAGE patterns of the WT dimer prepared as above without removal of GST (GST-WT/WT-His) were compared with those of the GST-WT homodimer and WT-His homodimer. The mobility of GST-WT/WT-His was intermediate between the GST-WT and WT-His homodimers, and the patterns did not change for at least 4 h after the proteins were thawed (data not shown). The native PAGE pattern of WT/WT-His (prepared by thrombin-digestion of GST-WT/WT-His) also remained constant, and there was no band corresponding to monomers that would form in the process of recombination. Because all our measurements were started immediately after the protein was thawed and finished within 2 h, contamination by recombination, if any, was negligible. Typically, heterodimeric proteins of 0.03–0.1 mg were obtained per liter of culture.

To confirm that the expression and purification using these two different tags does not affect the functions of the motors, we prepared the WT kinesin dimers by the above method (WT/WT-His) and compared their biochemical properties with those of the dimers expressed with His-tags only (WT-His/WT-His). The observed K_m(MT), k_cat, and the MT-gliding velocities of WT/WT-His were similar to those of WT-His/WT-His (Table 1), and were also in agreement with previous work (4).

Table 1.

Summary of ATPase measurement, MT-gliding assays, and single-molecule experiments of the homodimeric and heterodimeric kinesin constructs

	Construct	ATPase assay		Gliding assay	Beads assay
		kcat (s−1⋅head−1)	Km(MT), μM	Velocity, nm/s	Stall force, pN
Homodimer	WT	28.3  ± 2.5	0.4  ± 0.2	679  ± 59	6.3  ± 0.9
	WT	27.8  ± 1.7	0.5  ± 0.2	683  ± 42	6.0  ± 0.3
	L11/L11	11.1  ± 1.2	1.1  ± 0.3	179  ± 23	1.0  ± 0.2
	L12/L12	0.8	ND	0	0
	L8/L8	20.8  ± 3.1	1.2  ± 0.6	514  ± 31	4.0  ± 0.5
	L13/L13	19.8  ± 2.0	0.3  ± 0.2	5  ± 1	0
Heterodimer	WT/L11	20.2  ± 1.7	1.0  ± 0.3	202  ± 29	1.8  ± 0.3
	WT/L12	16.6  ± 2.2	2.0  ± 0.4	101  ± 25	0.8  ± 0.2
	WT/L8	22.7  ± 1.4	0.5  ± 0.2	554  ± 29	6.0  ± 0.7
	WT/L13	24.1  ± 0.7	0.2  ± 0.1	8  ± 1	0

The values of k_cat, K_m(MT), the gliding velocity, and the stall force are shown as the mean ± SD. The stall force was determined from the level of the plateau of the traces as shown in Fig. 3. ND, not determined.

^*WT kinesin homodimers with His-tags (WT-His/WT-His).

^† WT kinesin homodimers prepared with two different tags (WT/WT-His). The stall force of WT/WT-His was measured by using 1.0-μm beads.

^‡ The k_cat and K_m(MT) of L12/L12 could not be accurately measured because of low affinity to MTs. The indicated value is the ATPase rate with 50 μM tubulin.

^§ In the gliding assay, the L12/L12 homodimers did not retain MTs. In the beads assay, attachment signals were observed only when they exist in excess, but no movement was detected.

^¶ These constructs did not show processive behavior at a single-molecule level, although the beads moved continuously when more than one molecule was bound.

Mutations Used for Making Heterodimeric Kinesins.

Using this method, we made heterodimeric kinesins in which one of the two heads is altered in the functions of either MT binding [“L11” (K240A/L248A/K252A), “L12” (Y274A/R278A/K281A), and “L8” (E158A/K159A/R161A)] or mechanochemical coupling [“L13” (G291A/G292A)]. The residues to be mutated were chosen according to the previous reports by using alanine scanning and proteolysis experiments (4, 28, 29). In agreement with them, the homodimers of the first three mutants showed higher K_m values, lower k_cat, and slower MT gliding (Table 1, Fig. 2c). Of these, the homodimers of the L12 mutant were reported to have neither measurable MT-activated ATPase nor MT-gliding activity (28). In our case also, MTs did not bind to the L12/L12-coated glass surface in multiple-motor-gliding assays, even in the absence of nucleotides. On the other hand, L11/L11 and L8/L8 supported MT gliding, with velocities that are 26% and 76% of the WT homodimers, respectively. The homodimer of the uncoupling mutant, L13/L13, was previously shown to have almost normal MT-activated ATPase, but the MT-gliding velocity decreased by ≈100-fold (4). Our results were consistent with the report (Table 1).

Fig 2.

Histograms of MT-gliding velocities of the WT and L11 mutant kinesins. The assay chambers were coated with the WT kinesin (a), the wild-type and L11/L11 homodimers mixed in a proportion of 1:1 (b), L11/L11 homodimers (c), and WT/L11 heterodimers (d). The gliding velocity of the L11/L11 was much lower than the WT, but the copresence of L11/L11 did not interfere with the motility of MTs on the WT kinesin homodimers. The gliding velocity of WT/L11 was only slightly higher than that of L11/L11 homodimers. These assays were performed at 30 ± 1°C, so that the velocities were higher than the values obtained at 25 ± 1°C, as in Table 1.

Mixing Experiments of the WT and Mutant Homodimers.

Using the L11/L11 and L12/L12 homodimers, we have also examined the MT-gliding activity caused by a mixture of the WT and mutated kinesin homodimers. When the wild-type and L11/L11 homodimers were introduced into an assay chamber in a proportion of 1:1 (Fig. 2b), the averaged MT-gliding velocities were comparable with the velocity driven by the WT alone (Fig. 2a). Similar results were obtained with mixtures of the WT and L12/L12 homodimers (data not shown). These results show that the copresence of the L11/L11 or L12/L12 homodimers, which bind only weakly to MTs, do not interfere with the motility of the MTs supported by the WT homodimers.

ATPase Activities and MT-Gliding Velocities of the Heterodimers.

In contrast to the experiments using mixtures of homodimers, the MT-gliding velocities of the WT/L11 and WT/L12 heterodimers were considerably slower than that of the WT (Fig. 2d, Table 1). The percentages of the velocities relative to the WT homodimer are shown in Fig. 3. The average velocity of WT/L11 (202 ± 29 nm/s, n = 20; 30% of WT) was only slightly higher than that of L11/L11 (179 ± 23 nm/s, n = 14; 26% of WT). Although the L12/L12 homodimers did not retain MTs at the glass surface, WT/L12 was found to move MTs at 15% of the velocity of the WT (101 ± 25 nm/s, n = 16). Thus, mutant heads, when they exist as homodimers, did not slow down the movement of MTs supported by the WT homodimers but did cause a decrease in the velocity when they formed heterodimers with WT heads. The MT-gliding velocities of WT/L8 and WT/L13 were also significantly slower than the WT (Table 1, Fig. 3). The results indicate that a kinesin head needs a proper partner to produce normal velocity; the two heads must be closely cooperating in determining the velocity.

Fig 3.

The maximum MT-activated ATPase rates, the MT-gliding velocities, and the maximum forces of the homodimeric and heterodimeric kinesin constructs. Percentages relative to those of the WT homodimer are shown.

The ATPase rates of the heterodimers were intermediate between the WT and the parent mutant homodimers (Table 1, Fig. 3). For example, k_cat of L11/WT was 20.2 ± 1.7 ATP/s/head, which was approximately equal to the average of the WT and L11 homodimers [(28.3 + 11.1)/2 = 19.7]. Other heterodimers also showed k_cat values close to the average of the parent homodimers, suggesting that the maximum MT-activated ATPase rate of each head in a heterodimer is similar to that in a homodimer. To confirm the results, we made a heterodimer in which one of the heads is mutated so that it cannot hydrolyze ATP (E236A in ref. 30). The observed k_cat of the heterodimer WT/E236A (12.9 ± 1.7 ATP/s/head) was ≈1/2 of the WT homodimer, confirming that the ATPase activity of each head was not largely affected by its partner head.

These results indicate that the two heads of a heterodimer have different ATPase rates. It is difficult to explain by a simple form of hand-over-hand models, in which the two heads of a kinesin hydrolyze ATP alternately. Another unexpected result was that, although the L11/WT, L12/WT, and L13/WT heterodimers showed k_cat values >50% of the WT homodimer, their velocities were only ≈30, 20, and 1% of the WT homodimer, respectively (Fig. 3). These results are inconsistent with tightly coupled models, in which each ATP hydrolysis leads to an 8-nm step. We think that a proper coordination of the two heads was disrupted in heterodimers, which caused futile ATPase cycles.

The gliding speed of WT/L11 did not change as the surface density of the heterodimer was varied, similar to the case of the WT (data not shown). At a very low surface density of WT/L11, some MTs exhibited thermally driven pivot motions around a single contact point on the glass surface at which a single heterodimer is presumably located (31), while gliding slowly. The averaged gliding velocity of such MTs was in good agreement with the speed induced by multiple motors (271 ± 39 nm/s and 283 ± 33 nm/s at 30°C, for single and multiple motors, respectively; these velocities are higher than those in Table 1, which were measured at 25°C). The results suggest that a WT/L11 heterodimer can move processively along a MT, even though the mutant head may have only a weak affinity for MTs. Surprisingly, the WT/L12 heterodimer showed the same processive feature, even though the L12/L12 homodimers did not retain MTs at the glass surface. The average gliding speed of these MTs, presumably supported by a single molecule of WT/L12, was again similar to that of multiple motors.

Optical Trapping Nanometry.

Processive movement of the WT/L11, WT/L12, and WT/L8 heterodimers was confirmed in optical trapping experiments. When 0.2-μm beads were coated with a low density of kinesin at which only a single kinesin dimer per bead is thought to interact with a MT (see Materials and Methods), these three heterodimers, as well as the L11/L11 and L8/L8 homodimers, moved the beads (Fig. 4a). In the force–velocity curve (Fig. 4b), the velocity decreased almost linearly with increasing force. The maximum velocities of the heterodimers estimated from the force–velocity curve were in agreement with those measured in the multiple motor assays, confirming that their velocities do not depend on the number of motors. As expected, L12/L12 exhibited no signal at a single molecule level. At a higher density of L12/L12, the beads sometimes appeared to attach to the MT (indicated by a solid line in Fig. 4a), but no significant movement was observed. Hancock and Howard (6) reported that a heterodimeric kinesin that lacks one of the heads cannot move processively. The processive movement of the WT/L12 heterodimer suggests that even a mutant head that has only a weak affinity to MTs can help its partner WT head to move the molecule processively. With the L13/L13 homodimer and WT/L13 heterodimer, the bead bound briefly to a MT but did not show processive movement at a single molecule level.

With the WT and L8/L8 homodimers and the WT/L8 heterodimer, many beads moved continuously until they stalled before detaching from the MT (Fig. 4a). The averaged stall force exerted by a WT homodimer (6.3 ± 0.9 pN, n = 35; Table 1), determined from the level of the plateau of the traces, was similar to those previously reported using single, native kinesin molecules and recombinant kinesins (13–16). The stall force of L8/L8 (4.0 ± 0.5 pN, n = 22) was smaller than the WT. Although the distance moved was shorter for WT/L11, WT/L12, and L11/L11 than for the WT, enough beads moved continuously until they reached the maximum force to determine the stall force. As expected from these experiments, L11/L11 showed a considerably smaller stall force (1.0 ± 0.2 pN, n = 36) than the WT. The heterodimers WT/L11, WT/L12, and WT/L8 showed the stall forces that are intermediate between the WT and mutant homodimers [1.8 ± 0.3 pN (n = 74), 0.8 ± 0.2 pN (n = 29), and 6.0 ± 0.7 pN (n = 18), for WT/L11, WT/L12, and WT/L8, respectively]. These values were in agreement with the stall forces obtained from the force–velocity curve in Fig. 4b.

It is not known which process during the kinesin's mechanical and chemical cycle determines the stall force, although it was reported that it does not depend on the temperature (32) or the concentration of ATP (13). If each head of a dimer is alternately and independently responsible for producing force, we might expect the stall force of a heterodimer to be determined by the weaker of the two heads, i.e., the mutant. However, the observed stall forces were significantly larger than those of the mutant homodimers (P < 0.005, t test). The results indicate that both heads of a dimer must cooperate in producing the maximum force.

Conclusion

In the present work, we have succeeded in generating heterodimeric kinesins in which one of the two heads was mutated. The resulting heterodimers moved processively, even when the mutant head had only a weak affinity to a MT. This is in contrast to a one-headed dimer, which did not move processively (6). Use of processive heterodimeric kinesins enabled us to measure the force and velocity produced by a single heterodimer and thus investigate the contribution of each head in force production and motility. Our results indicated a close mechanical cooperation of the two heads: a head cannot produce the maximum force and velocity by itself. The results are supported by previous biochemical studies showing that the two heads communicate with each other (8–12). In the most generally accepted model, a kinesin dimer uses the two heads alternately, moving in a hand-over-hand fashion (8–12, 33–36). In contrast, there are other models in which one of the heads is always leading (37). In the former case, however, the time interval between successive 8-nm steps (dwell time) in heterodimers should be different every other time. With a statistical analysis of the dwell time, our method of using heterodimeric kinesins opens the way to test alternate stepping models directly.

Abbreviations

• K432, a construct of human conventional kinesin consisting of amino acid residues 1–432

• MT, microtubule