Nucleotide-Dependent Single- to Double-Headed Binding of Kinesin

See allHide authors and affiliations

Science  26 Jan 2001:
Vol. 291, Issue 5504, pp. 667-669
DOI: 10.1126/science.291.5504.667


The motility of kinesin motors is explained by a “hand-over-hand” model in which two heads of kinesin alternately repeat single-headed and double-headed binding with a microtubule. To investigate the binding mode of kinesin at the key nucleotide states during adenosine 5′-triphosphate (ATP) hydrolysis, we measured the mechanical properties of a single kinesin-microtubule complex by applying an external load with optical tweezers. Both the unbinding force and the elastic modulus in solutions containing AMP-PNP (an ATP analog) were twice the value of those in nucleotide-free solution or in the presence of both AMP-PNP and adenosine 5′-diphosphate. Thus, kinesin binds through two heads in the former and one head in the latter two states, which supports a major prediction of the hand-over-hand model.

Kinesin is a molecular motor that transports membrane-bound vesicles and organelles toward the plus end of a microtubule in various cells including neurons (1, 2). Kinesin takes hundreds of 8-nm steps (the size of tubulin heterodimers composed of α and β subunits) (3–5) before detachment, so that the run length reaches longer than 1 μm (3,6). Each step is associated with one cycle of ATP hydrolysis (7, 8). Structural and biophysical evidence shows that stepping of kinesin is triggered by conformational changes in the ATP-bound head (9).

A “hand-over-hand” model has been proposed to explain the processive movement of kinesin (Fig. 1) (5, 9–16). To substantiate the hand-over-hand model, it is essential to determine the binding mode—either single- or double-headed binding—at each nucleotide state, and the kinetic step at which the transition between the two binding modes occurs (5,9–16). Results of image analysis by cryoelectron microscopy on the dimeric kinesin-microtubule complex have been inconclusive; either single-headed (13,14) or double-headed (15,16) binding has been found to predominate both in the absence of nucleotides and in the presence of AMP-PNP. In solution the binding stoichiometry of the kinesin head and the tubulin heterodimer in a microtubule has a molar ratio of 2:1 in both the nucleotide-free state (17–19) and in the presence of AMP-PNP (17), implying a single-headed binding. In these studies, the microtubule was fully decorated by kinesin so that the conformation of kinesin may have been constrained (16). The intramolecular interhead distance of kinesin in the crystal structure is about 5 nm (20), considerably shorter than the size of the tubulin heterodimer. The kinetics of detachment in solution also suggests the single-headed binding not only in the nucleotide-free condition but also in the coexistence of AMP-PNP and adenosine 5′-diphosphate (ADP) (21). Overall, the evidence for the model remains indirect.

Figure 1

A simplified version of the hand-over-hand model on the mechanism of kinesin motility proposed to date (5,9–16). O: nucleotide-free; D, T, and Pi: ADP, ATP, and inorganic phosphate, respectively.

To obtain direct evidence for the binding mode at each nucleotide state, we measured the mechanical properties of single kinesin molecules attached to a microtubule in three different solvent conditions (22): in the absence of added nucleotides, in the presence of 0.5 mM AMP-PNP and 1 mM ADP, and in the presence of 1 mM AMP-PNP. The first condition corresponds to the 〈O, O〉 or 〈O, D〉 states (Fig. 1). Although apyrase was added, some proportion of heads may still have bound ADP because the kinesin was purified in the presence of ADP and the detachment rate of ADP is slow (18, 19). The second condition mimics the 〈T, D〉 state. Under this condition, only one of the two heads is expected to bind AMP-PNP, while the other head is in the ADP state (21). The third condition is considered to represent the 〈T, O〉 state (23), because the binding ratio of AMP-PNP to kinesin molecules is reported to be 1:1 under the present condition (23).

An external load was imposed on a single kinesin-microtubule bead complex (Fig. 2A) (24, 25) as it was moved toward the plus (or minus) end of a microtubule with optical tweezers (26,27). We repeated unbinding force measurements at nearly the same position on the same microtubule several times for the same bead, presumably for the same kinesin molecule (Fig. 2, B to D). Upon loading toward the plus end, unbinding force in the absence of nucleotides was about 7 pN (Fig. 2D), whereas that in the presence of AMP-PNP could apparently be classified into two components at about 14 pN (major) and 7 pN (minor), the latter corresponding to that in the absence of nucleotides. Using the same data, we obtained the force-extension relation on the kinesin-microtubule complex (Fig. 2E). This relation was almost linear, so that the elastic modulus could be estimated simply from the slope. The elastic modulus could also be classified into two components (compare Figs. 2E and3, D to F). On the minus-end loading, the unbinding force for both components increased by 45% irrespective of the nucleotide states, keeping the elastic modulus unchanged. This shows that the binding is unstable for the plus-end loading compared with the minus-end loading.

Figure 2

Measurements of unbinding force and elastic modulus. (A) Fluorescence micrograph showing a kinesin-coated bead attached to a microtubule. The bead was brought onto the plus (or minus) end portion of the microtubule for plus (or minus)-end loading. Bar, 5 μm. (B and C) Time course of movement of the trap center (thin lines) and bead (circles). A kinesin-coated bead was attached to a microtubule in the absence of nucleotides (B) or in the presence of AMP-PNP (C). The bead was then pulled to the plus end of the microtubule at the constant rate of 120 nm s−1 (equal to the loading rate of 10.4 pN s−1). The unbinding force was estimated to be 8.4 and 15.1 pN, respectively, from the abrupt displacement of the beads at about 4.4 s (B) and 3.8 s (C). (D) Sequential data of unbinding force measurements for four different preparations; data shown by arrows were taken from (B) and (C). Conditions: (B) and (○) in (D), nucleotide-free; (C) and (•) in (D), + AMP-PNP. (E) Examples of the force-extension relation (○, nucleotide-free; •, + AMP-PNP). The relation was obtained from the time course of bead displacement as shown in the inset. To be strict, the force means the force component parallel to the glass surface, and the extension means the displacement of the bead in parallel to the glass surface (27). Arrows show the position of unbinding.

Figure 3

Distribution of unbinding force (A to C) and relation between elastic modulus and unbinding force (D to F) at different nucleotide-binding states. The external load was applied toward the plus end. (A and D) Nucleotide-free (n = 46); (B and E) AMP-PNP + ADP (n = 33); (C and F) + AMP-PNP (n = 43). Unbinding force (pN) ± SD: 6.7 ± 1.8 (A), 7.2 ± 2.0 (B), and 6.6 ± 1.7 (n = 14), 12.8 ± 1.6 (n = 29) (C). Elastic modulus (pN nm−1) ± SD: 0.35 ± 0.14 (D), 0.37 ± 0.16 (E), and 0.39 ± 0.17 (n = 14), 0.67 ± 0.21 (n= 29) (F). Loading rate (pN s−1): 3.5 (A and D), 6.0 (B and E), and 4.3 (C and F). A single Gaussian distribution could simulate unbinding force distribution in (A) and (B). In contrast, the unbinding force distribution in (C) was simulated by the sum of two Gaussian distributions, with S- and L-components defined as the smaller and larger unbinding force, respectively. The boundary between the S- and L-components was determined by the junction of two Gaussian distributions [shown by an arrow (C)]; the boundary in (F) was determined according to that in (C). The average values for S- and L-components are shown by asterisks in (D) to (F).

The small (∼7 pN; S-) and large (∼14 pN; L-) components of unbinding force, respectively, correspond to those of the elastic modulus (Fig. 3). The finding that the unbinding force and the elastic modulus for the L-component were twice those for the S-component strongly suggests that the S- and L-components are attributable to the single- and double-headed binding of kinesin, respectively. Thus, each kinesin head contributes equally to the elastic modulus, such that each head equally shares the external load.

This interpretation shows that the binding mode in the 〈O, D〉 state (Fig. 1A) is single-headed. Assuming that AMP-PNP is an ATP analog, single-headed binding is also predominant in the 〈T, D〉 state (Fig. 1B). It is highly probable that the attached head binds AMP-PNP (ATP) (9), whereas the detached head binds ADP (9, 21), because the attachment of the ADP-bound head was reported to be weak (17). In the 〈T, O〉 state (Fig. 1C), in contrast, double-headed binding is predominant. Here we find that, based on the bimodal distributions of unbinding force and elastic modulus at one loading rate (Figs. 2D and 3, C and F), both single- and double-headed binding exist. Additionally, we find that the proportion of the S-component decreased as the loading rate increased from 2 to 18 pN s−1 and, finally, disappeared at the highest loading rate we examined (18 pN s−1), irrespective of the loading direction. This implies that double-headed binding predominates in the absence of external load. The finding that the unbinding force for the plus-end loading was smaller than that for the minus-end loading suggests that, in the “bridge” structure of double-headed binding (see Fig. 1C), the rear head is relatively unstable so that it tends to be detached. Such an asymmetry for the loading direction regarding the stability of the attached state is favorable for kinesin motors stepping forward.

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


View Abstract

Navigate This Article