Higher plant myosin XI moves processively on actin with 35 nm steps at high velocity

Purification of 175 kDa myosin XI

The 175 kDa myosin was isolated from cultured tobacco bright yellow-2 cells of Nicotiana tabacum (BY-2 cell) basically according to the method of Yokota et al. (1999). The purity of this myosin was examined by 5% SDS–PAGE with silver staining (Figure 1A). No significant contamination by other myosins (particularly the 170 kDa myosin) and no aggregation was detected, as reported by Yokota et al. (1999). Since plant myosins in general are labile, we used 175 kDa myosin for functional studies within 2 days of preparation.

Sequence analysis and morphology of myosin XI

cDNA encoding the 175 kDa myosin from an N.tabacum cDNA library was cloned, including 1362 amino acids starting 12 amino acids upstream of the P-loop (GESGAGKT) to the C-terminus. Sequencing the cDNA revealed that this myosin shares 60% sequence identity with MYA1 (DDBJ/EMBL/GenBank accession No. Z28389) (Kinkema and Schiefelbein, 1994) and 75% with MYA2 (DDBJ/EMBL/GenBank accession No. Z34293) (Kinkema et al., 1994) of Arabidopsis thaliana, both of which are members of myosin class XI. The amino acid sequence of the C-terminal peptide fragments (14 amino acids) of 175 kDa myosin exactly matched that deduced from the cDNA. This indicates that the amino acid sequence deduced from the cDNA really is that of 175 kDa myosin. A non-rooted phylogenetic tree was created based on the neighbor-joining method using the program in Clustal_X (Thompson et al., 1997) showing a bootstrap value of 1000 at the node where MYA2 branched to this myosin. These results support classification of this 175 kDa myosin within myosin class XI. In common with all known sequences of plant myosin XIs (Kashiyama et al., 2000; Morimatsu et al., 2000; Reddy, 2001; Yokota et al., 2001), this 175 kDa myosin is predicted to have six IQ motifs per heavy chain in the light chain-binding domain that normally bind six calmodulin molecules, and to have an α-helical coiled-coil domain leading to dimer formation (Figure 1B). In the center of this domain, 10 amino acids break the α-helical coiled-coil structure, representing a putative hinge region.

Electron microscopy of rotary shadowed 175 kDa myosin confirmed predictions from the sequence: there are two head domains including an elongated neck ∼33 nm in total length, much larger than for myosin II (Figure 1C). This feature is different from Chara myosin XI which has two heads of similar size and shape to those of myosin II (Yamamoto et al., 1995). The two neck domains join at a thin stalk extending for ∼25 nm. Two globular domains are found at the tail of the molecule. This structural organization is analogous to that of myosin V (Espreafico et al., 1992; Walker et al., 2000), whose mechanical properties have been studied extensively (Rief et al., 2000; Mehta, 2001; Veigel et al., 2002). This 175 kDa myosin hereafter is called myosin XI.

Processivity of myosin XI

In order to characterize the mechanical properties of myosin XI, we first used the in vitro motility assay (Kron et al., 1991; Yokota et al., 1999) in which actin filaments glide over glass surfaces coated with myosin at low calcium concentration (pCa >8). The direction of myosin movement was examined using actin filaments marked at the pointed end (Wells et al., 1999): these filaments moved with the minus end leading, indicating that myosin XI moves towards the plus end on actin (data not shown). Sliding velocities of fluorescently labeled actin filaments depended on ATP concentration in a Michaelis–Menten-like manner: at 20°C, K_m for ATP was 100 ± 17 µM and V_max was 4.6 ± 0.2 µm/s (Figure 2A, red triangles). This maximum sliding velocity of actin filaments was comparable with the velocities of cytoplasmic streaming observed in vivo (Yokota et al., 1999) and three times that observed for skeletal myosin under similar conditions (Anson, 1992). In addition, continuous movement of actin filaments shorter than 1 µm with constant maximal velocity was observed even at low surface densities of myosin XI (<50 molecules/µm²), indicating that it is a processive motor undergoing multiple ATPase cycles before dissociating from actin filaments. In contrast, 600 molecules/µm² of rabbit skeletal heavy meromyosin were required for the continuous movement of actin filaments longer than 1.1 µm (Toyoshima et al., 1990).

We also measured the rate at which actin filaments came out of bulk solution, bound and then moved over surfaces coated with myosin XI. The attachment rate plotted against myosin surface density (Figure 2B) is well fitted by the theoretical relationship for a single myosin molecule being sufficient for attachment and motility of actin filaments (Hancock and Howard, 1998). The fraction of actin filaments that moved further than their own length before leaving the surface gradually decreased from 100% to 0 as myosin density was lowered. Actin filaments observed at very low myosin surface densities rotated erratically about a vertical axis through a fixed point on the surface, where a single myosin molecule presumably is located, while still progressing as expected for processive motors (for details see the Supplementary data available at The EMBO Journal Online).

To assess this processivity further, we undertook steady-state kinetic analyses of the myosin XI. Actin-activated ATPase rates were measured in the same solution as that used for in vitro motility assays, and yielded values of K_app for F-actin = 9.8 ± 1.0 µM and V_max = 77.2 ± 2.8/s/head at 20°C (Figure 2C). In the presence of 48 µM F-actin, the ATPase showed Michaelis–Menten kinetics, with K_m for ATP = 81.1 ± 1.0 µM and V_max = 76.3 ± 2.5/s/head (Figure 2A, black circles). Thus the actin-activated Mg-ATPase of this myosin under steady-state conditions is characterized by a high V_max but a relatively low K_app for actin compared with skeletal myosin II (Oiwa et al., 2000).

Kinetic processivity, defined as the average number of ATPase cycles before dissociation of motors (myosin) from their track (actin), can be established using criteria derived from studies of kinesin processivity (Hackney, 1995). These criteria may be applied to this myosin XI as they were for myosin V (De La Cruz et al., 1999; Trybus et al., 1999). The second-order rate constant derived from k_cat/K_app is one such criterion (Hackney, 1995): a value of >10⁷/M/s indicates many ATPase cycles per diffusional encounter between myosin and F-actin (Trybus et al., 1999); for myosin XI it is 7.9 × 10⁶/M/s at 20°C, suggesting a processive motor. Another such criterion is the ratio of K_m for the actin-activated ATPase to that of motility (Leibler and Huse, 1993). Motors with high duty ratios have high K_m ratios close to unity, e.g. kinesin has a ratio of unity. Motors with two heads and high K_m ratios tend to be highly processive (Hancock and Howard, 1998; De La Cruz et al., 1999). For this myosin, K_m (ATPase)/K_m (motility) is 81 µM/100 µM = 0.81, again suggesting that myosin XI has a high duty ratio and is a processive motor. These calculations fit well with the expected cellular roles of myosin XI and demonstrate its processivity.

Effects of ADP on mean velocity of actin filaments moving on myosin XI-coated surfaces

Examination of the effect of addition of ADP to actively cycling myosin XI is important since it is generally believed that ADP release is the rate-limiting step in the chemomechanical cycle of processive motors, such as myosin V (De La Cruz et al., 1999; Rief et al., 2000). Analysis of the sliding velocity of actin filaments in the presence of ADP provides information about the ADP dissociation transition from this myosin XI (Figure 3). The ATP concentration was held at 20, 100 or 500 µM, and the ADP concentration was varied. The Dixon plot clearly indicates that ADP acts on the velocity as a competitive inhibitor (Figure 3, inset). The mechanism of ADP inhibition on the sliding velocity was modeled as: rebinding of ADP or binding of a new ADP to the active site of myosin competes with binding of a fresh ATP after product release. Least-squares fits to Equation 1 with fixed K_m = 100 µM (determined above),

V = V_max [ATP]/{K_m (1 + [ADP]/K_d) + [ATP]}(1)

gave a dissociation constant for ADP from the acto-myosin complex of 42 ± 4 µM, which is much smaller than for skeletal muscle myosin (Homsher et al., 1992) or smooth muscle myosin (Warshaw et al., 1991).

Stepwise movements of single myosin XI molecules

Precisely measuring the movement of individual myosin XI molecules should help to elucidate mechanisms for processive motion on actin for this myosin XI as implied from the kinetic studies above, so we performed a bead assay using an optical trap (Kojima et al., 1997; Sakakibara et al., 1999). Polystyrene beads were first coated with an excess of antibody molecules against the C-terminal sequence of myosin XI (Figure 4A). These beads were then coated with various densities of myosin XI molecules by mixing myosin solutions with beads at various molar ratios. After introducing beads into the observation cell with 1 µM–1 mM ATP, a freely diffusing bead was captured with the optical trap and brought into contact with a single actin filament bound to the glass surface. Within 30 s, beads that had bound to an actin filament commenced movement.

The relative frequency of binding, and hence movement, depends on myosin:bead stoichiometries. Assuming that one active myosin molecule attached to a bead with a geometry favorable for interaction suffices to bind that bead to an actin filament, the fraction of beads bound to actin filaments should follow Poisson statistics (Block et al., 1990). It was confirmed that single myosin molecules are sufficient to move beads by the observed statistics (Figure 4B, solid curve) with probability P(n) = (1 – e^–n/λ), where n is the number ratio of myosin molecules mixed with beads and λ is a fitting parameter, showing a Poisson distribution. The slightly lower values for moving beads (Figure 4B, circles) compared with the bound beads (Figure 4B, black dots) may reflect the fraction of the myosin molecules that can bind to actin filaments but cannot move due to unfavorable bead attachment geometry restricting motor movement. The experiments described here were conducted using beads having myosin:bead stoichiometries low enough such that there was <40% probability of binding and hence movement, so the Poisson probability that a bead carried two or more active myosin molecules was <0.09. Assuming a random distribution of myosin molecules over the bead surface and allowing a 60 nm reach for each myosin molecule’s tether (see Figure 1C), the probability that two or more myosin molecules on the same 1 µm diameter bead were able to interact simultaneously with an actin filament is <0.035.

Trap stiffness was chosen to be as small as possible in order to permit measurements with sufficiently large displacement of the bead whilst not exceeding the linear range of the sensor. For these measurements, a compliant optical trap, 4.6 fN/nm, was used. When a trapped bead was brought into contact with an actin filament, it attached and executed transient movements along the actin filament direction (Figure 5A), such that several runs occurred as a cluster. In such clusters, beads progressed for various periods and then were drawn rapidly back to the trap center, whereupon they rebound and recommenced moving. Multiple cycles of attachment, movement and release were observed with the same bead. The average distance moved was 126 ± 76 nm (mean ± SD, n = 355) at 1 µM ATP, corresponding to a trap force of 0.58 ± 0.35 pN, and 113 ± 54 nm (mean ± SD, n = 532) at 100 µM ATP, corresponding to a force of 0.52 ± 0.25 pN (Figure 5B). At 100 µM ATP, we defined the stall force as the force at which movement of the bead was stalled within ±35 nm positional fluctuation for >0.1 s. Since the beads often dissociated before slowing down substantially, most forces calculated here do not correspond to a stall force but rather a yielding force. The mean stall force was 0.60 ± 0.25 pN, n = 85 and that of yielding force was 0.50 ± 0.25 pN, n = 447 at 100 µM ATP. There seems to be no substantial difference between these forces, so that we refer to the maximal force generated in each run without distinguishing stall or yielding forces. This maximal force did not depend significantly upon ATP concentration.

The maximal sliding velocities of beads were derived by fitting lines to segments of movement with between 0 and 0.3 pN force at various concentrations of ATP in order to compare them with the actin-activated ATPase rate. In these segments, the line fitting yielded reliable velocities in spite of the stochastic nature of the stepping process at low ATP concentrations. The maximal velocity of bead movement at saturating ATP concentration was estimated to be 7.7 ± 0.5 µm/s (Figure 2A, green squares) from the fit by the Michaelis–Menten equation (Figure 2A, green curve). K_m was estimated to be 90 ± 24 µM, which was between that for ATPase (81 µM) and that for actin motility (100 µM) (Figure 2A). This maximal velocity of beads was nearly 2-fold faster than the gliding velocity, 4.6 µm/s, of actin filaments moving in the in vitro motility assay (Figure 2A), and hence closer to velocities of cytoplasmic streaming observed in vivo (Yokota et al., 1999).

We also measured the distance (run length) moved by these beads without trapping after bringing them into contact with actin filaments using the optical trap (Figure 6A). Beads having a probability of binding and hence moving between 0.28 and 0.33 were used. Beads that moved >1 µm were measured, since displacements <1 µm are difficult to distinguish from Brownian motion. Histograms of run length at 1, 10, 100 µM and 1 mM ATP show single exponential distributions with mean distances traveled of 1.11–1.43 µm (1.28 ± 0.13 µm, mean ± SEM, n = 4) (Figure 6B). No significant tendency of change in mean distance with ATP concentration was observed. Even though this myosin XI carried a bead 1 µm in diameter, the mean distance is comparable with that obtained from single molecule optical measurements of myosin V (Sakamoto et al., 2000). The probability of myosin XI detachment from an actin filament is estimated to be 0.027 per step (35 nm/1.28 µm). Therefore, the detachment of the beads from actin filaments observed under load (Figure 5A) is due mainly to a load-dependent transition.

In records of bead displacement, thermal noise decreased as the beads developed greater force (Figures 5A and 7A), indicating a non-linear elasticity, the total (glass–F-actin–myosin–bead–trap) linkage stiffness becoming greater with increasing extension. This phenomenon, a departure from Hooke’s law, was also reported for kinesin-coated (Svoboda et al., 1993) and dynein-coated beads (Sakakibara et al., 1999).

At forces >0.5 pN, stepwise movement of beads was clearly observed not only at low ATP concentrations but also at high concentrations (Figures 5A and 7A). Thus we then focused on the stepwise movement at forces >0.5 pN. The histogram of the distance moved between dwells shows that this stepwise movement occurred in 35.0 ± 4.3 nm steps, corresponding closely to the 36 nm pseudo-repeat of the actin helix (Figure 7B). At 100 µM and 1 mM ATP, stepwise movement seemed sometimes to occur at 70 nm, possibly due in part to the limited time resolution of the system (τ_rise ∼7 ms at f_c = 50 Hz); hence, the step size distribution showed a second peak (Figure 7B).

Dwell time analysis at various nucleotide concentrations

In order to correlate the ATPase cycle and the stepwise movements of single myosin molecules, the dwell times of steps were measured at three ATP concentrations (1, 10 and 100 µM). Since the dwell time became shorter than the temporal resolution of the system (τ_rise ∼7 ms at f_c = 50 Hz) as the ATP concentration increased, mean dwell time at 1 mM ATP was not measured directly but was estimated as 35 nm/(velocity at >0.5 pN). The velocity at >0.5 pN was calculated by fitting lines to segments of movement between 0.5 pN and maximal force, giving a mean value of 2.9 ± 1.2 µm/s. Thus the mean dwell time at 1 mM ATP was estimated to be ∼12 ms. The mean rate constants calculated from the mean dwell times at 1, 10 and 100 µM ATP were 0.91, 2.89 and 39.7/s, respectively, and at 1 mM ATP, 82.9/s. Thus we obtained a hyperbolic relationship between the mean rate constants and ATP concentration (Figure 8A).

Consider the following mechanism (Equation 2) in which the transition from strong (A–M) to weak binding (A + M–ATP) of acto-myosin follows ADP dissociation from acto-myosin complex and subsequent binding of ATP to myosin.

If we assume that k_–4[ADP] is negligible at low ADP concentration relative to k₊₁ and that the dwell time reflects the time for transition from strong (A–M) to weak binding (A + M–ATP), this relationship can be described by one ATP-dependent (k₊₁) and one ATP-independent rate (k₊₄) as:

k_mean = k₊₄ k₊₁ [ATP]/(k₊₄ + k₊₁[ATP])(3)

A fit by Equation 3 yields the rate constant, k₊₄ = 95 ± 4/s (±SD) (Figure 8A) and an apparent ATP-dependent second-order rate constant per mechanical step, k₊₁ = (0.6 ± 0.1) × 10⁶/M/s (±SD), comparable with the ATP-binding rate in solution for myosin II, 2 × 10⁶/M/s (Oiwa et al., 2000) and 0.9 × 10⁶/M/s for myosin V (De La Cruz et al., 1999).

Although dwell time distributions at various ATP concentrations were approximately exponential, in accordance with the kinetic scheme in Equation 2, the double exponential shown below was used for fits to the dwell time distributions:

P(t) = λ₁λ₂/(λ₂ – λ₁) [exp(–λ₁t) – exp(–λ₂t)](4)

Since the temporal response of the system does not distinguish dwell time data shorter than 10 ms, these were all categorized into a histogram bin of dwell times <10 ms. In the absence of ADP, λ₁ corresponds to k₊₄ and λ₂ corresponds to k₊₁[ATP]. Solid curves in Figure 8B represent double exponential fits with one rate constant k₊₄ fixed to 95/s, yielding 1.0 ± 0.1, 4.5 ± 0.6 and 47 ± 28/s (±SEM) at 1, 10 and 100 µM ATP, respectively.

In the presence of 200 or 400 µM ADP as well as 100 µM ATP, the dwell time distributions shifted to longer times than in the absence of ADP (Figure 8C). In the double exponential fitting using Equation 3, the product of λ₁ and λ₂ provides the constraint of the rate constants since it is the product of k₊₁[ATP] and k₊₄, which was determined above. Fitting gave the constants of 15.3 ± 1.6 and 370 ± 40/s at 200 µM ADP, and rate constants of 9.9 ± 0.9 and 580 ± 51/s at 400 µM ADP. Although λ₁ and λ₂ are not simply expressed by the rate constants in Equation 2, the mean rate constant, k_mean is expressed as:

k_mean = λ₁ λ₂/(λ₁ + λ₂) = k₊₁[ATP]/(1 + [ADP]/K_d + k₊₁[ATP]/k₊₄)(5)

Using k₊₁ = 0.6/µM/s and k₊₄ = 95/s determined by the dwell time distributions at different ATP concentrations, K_d of ADP is calculated to be 82 ± 40 and 44 ± 11 µM for 200 and 400 µM ADP, respectively.

Protein preparation

Plant myosins were isolated from cultured tobacco bright yellow-2 cells of N.tabacum (BY-2 cell) basically according to the method of Yokota et al. (1999). Briefly, cells from 5-day-old cultures were used for the isolation of myosin. BY-2 cells suspended in solution containing 0.25 M sucrose, 1% (w/v) casein, 20 mM EGTA, 6 mM MgCl₂, 100 µg/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 2 mM dithiothreitol (DTT) and 30 mM PIPES-KOH pH 7.0 were homogenized. After removal of debris by centrifugation, the supernatant was co-sedimented with 0.1 mg/ml rabbit skeletal muscle F-actin at 100 000 g for 30 min. The pellet was re-suspended in 20 ml of EMP solution containing 90 mM KCl, 5 mM EGTA, 6 mM MgCl₂, 0.5 mM PMSF, 50 µg/ml leupeptin, 1 mM DTT and 30 mM PIPES-KOH pH 7.0, and kept for 30 min on ice. After a second centrifugation at 100 000 g for 30 min, the pellet was extracted with EMP solution containing 10 mM ATP and 5 mM potassium phosphate pH 7.0 for 10 min on ice. This ATP extract, following centrifugation at 100 000 g for 30 min, was applied directly to a hydroxylapatite column pre-equilibrated with EMP solution containing 5 mM potassium phosphate pH 7.0. The column was eluted stepwise with 5, 150 and 300 mM potassium phosphate in EMP solution. The fractions eluted with 5 mM potassium phosphate buffer were dialyzed against EMP solution with 5 mM KCl on ice for 10 h. This dialysate applied to a Mono Q HR5/5 anion exchange column (Amersham Biosciences, Uppsala, Sweden) was eluted with a gradient of 180–500 mM KCl in EMP solution. The fraction was incubated with F-actin (final concentration at 0.15 mg/ml) for 30 min on ice and centrifuged at 100 000 g for 30 min. The pellet was extracted with EMP solution containing 5 mM ATP and 1 mM DTT on ice for 10 min. The supernatant after centrifugation at 100 000 g for 30 min was used directly as isolated 175 kDa myosin XI in solution. For further purification, it was fractionated by HPLC with a Mono Q HR5/5 column (Amersham Biosciences) under linear gradient operation (180–500 mM KCl with the EMP buffer). To avoid dilution and overestimation of protein concentrations due to contamination, myosin concentrations were always determined using Bradford assays on SDS–polyacrylamide gels, with rabbit skeletal muscle HMM as the concentration standard.

Production of antibodies

We previously cloned and sequenced cDNA encoding 175 kDa myosin from N.tabacum (Yokota et al., 2001). The cDNA coding the C-terminal region of the protein (87 amino acid residues) was amplified by PCR using forward primer 5′-TAGAGATCTGCAACTTTACAGAA TCTGTA-3′ with a BglII site (underlined) and reverse primer 5′-CCTAAGCTTCACTCGTGTAAAAATTGGAA-3′ with a HindIII site (underlined). The PCR product was cloned into pQE41 vector (Qiagen, Valencia, CA) using the BglII and HindIII sites. A fusion protein of the C-terminal region of the protein with dihydrofolate reductase and a histidine tag was expressed in Escherichia coli strain XL1Blue (Stratagene, La Jolla, CA), and purified through an Ni-NTA resin column according to the manufacturer’s protocol (Qiagen). The protein was dialyzed against water, lyophilized, dissolved in phosphate-buffered saline and injected into Japanese white rabbits with Freund’s adjuvant. The antiserum was prepared according to the standard protocol.

Sequence analysis

GenBank was searched using the NCBI-BLASTA program. Alignment of nucleotide sequences and comparison were performed with the DiAlign program at EMBL. After multiple alignment, phylogenetic analysis was performed with the neighbor-joining method using the program in Clustal_X (Thompson et al., 1997). Bootstrap resampling (1000 trials) was used for estimation of the degree of confidence in the branching order. Predictions for coiled-coil structure were carried out using COILS version 2.2 (Lupas, 1996).

Electron microscopy

Mono Q column-purified 175 kDa myosin was adsorbed on a freshly cleaved mica surface, stabilized with buffered uranyl acetate and washed with glycerol-containing solution. The sample was then rotary replicated with platinum–carbon at a 6° elevation angle. Details of the method were described elsewhere (Mabuchi, 1991; Katayama and Ikebe, 1995).

Kinetic experiments

Steady-state actin-activated ATPase was measured at 20°C in solutions containing 25 mM KCl, 4 mM MgCl₂, 1 mM EGTA, 1 mM DTT, 25 mM HEPES-KOH pH 7.5 and 0.4–48 µM F-actin with an enzyme-coupled assay (No. E-6646, Molecular Probes, Eugene, OR).

In vitro motility assays.

In vitro assays were performed at 20°C, basically as performed by Toyoshima et al. (1990), but with some modifications. A flow cell, volume 10 µl, was made from two nitrocellulose-coated coverslips (24 mm × 40 mm and 18 mm × 18 mm), with two slivers of polycarbonate film as spacers. A 10 µl aliquot of purified myosin (1–5 µg/ml) in buffer solution (25 mM KCl, 4 mM MgCl₂, 1 mM EGTA, 1 mM DTT and 25 mM HEPES-KOH pH 7.5) was introduced into the flow cell and incubated for 10 min. After washing the flow cell with buffer solution containing 0.5 mg/ml bovine serum albumin (BSA), 10 µl of buffer solution containing 50 nM tetramethylrhodamine-labeled actin filaments, 1 mM DTT, an oxygen-scavenging system (25 µg/ml glucose oxidase, 45 µg/ml catalase and 1% glucose) and various concentrations of ATP (from 10 µM to 1 mM) was infused into the flow cell.

In order to measure velocities of actin filaments moving on myosin-coated surfaces, movement was recorded at 30 frames/s with an intensified CCD camera. Video tapes of the actin filament movement were played back off-line, and the positions of individual actin filaments were digitized frame by frame using an Argus-20 image processor (Hamamatsu Photonics, Hamamatsu). Velocities were calculated from the change in position as a function of time. To obtain reliable velocities, only actin filaments shorter than 1 µm that moved continuously over 5 µm were analyzed.

For the attachment rate assay, a custom-built total internal reflection fluorescence microscope (Oiwa et al., 2000) was used to enable precise scoring of landed filaments as those that had touched down and moved for >0.5 µm over 2 s at 20 µM ATP at constant actin concentration (5 µg/ml). The surface density of myosin molecules was calculated assuming that every myosin molecule infused into the flow cell evenly attached to the flow cell surface without losing its activity.

Optical trap nanometry

Polystyrene carboxylated beads 1 µm in diameter were incubated on ice in assay buffer (25 mM KCl, 4 mM MgSO₄, 1 mM EGTA, 25 mM HEPES-KOH pH 7.5) for 10 h with varying concentrations of anti-C-terminal peptide antibody. Coated beads were added to myosin solutions and then applied to actin filaments fixed onto a coverslip via avidin–biotin interaction. The experimental set-up for the optical trap (Nd:YAG laser power 500 mW at 1064 nm, Spectra Physics) was the same as used previously (Kojima et al., 1997; Sakakibara et al., 1999). Displacement of the bead was measured using a quadrant photodiode system with 0.1 nm spatial (noise-equivalent value) and 0.02 ms temporal resolution. Trap stiffness (4.6 fN/nm) and bead–myosin linkage stiffness (21–140 fN/nm) were determined from variances of the Brownian fluctuations of beads. An attenuation factor, the ratio of the sum of bead–myosin linkage stiffness and trap stiffness to bead–myosin linkage stiffness, was 1.03–1.22. Displacement of beads within the trap was used in this study without conversion to displacement of myosin. Observations were made at 20°C in the same buffer as used for in vitro motility.

Accession number

Sequence data for 175 kDa myosin XI are available from DDBJ/EMBL/GenBank with accession No. AB082121.

Supplementary data

Supplementary data are available at The EMBO Journal Online.