|Home | About | Journals | Submit | Contact Us | Français|
Myosin VI is an unconventional motor protein with unusual motility properties such as its direction of motion and path on actin and a large stride relative to its short lever arms. To understand these features, the rotational dynamics of the lever arm were studied by single-molecule polarized Total Internal Reflection Fluorescence (polTIRF) microscopy during processive motility of myosin VI along actin. The axial angle is distributed in two peaks, consistent with the hand-over-hand model. The changes in lever arm angles during discrete steps suggest that it exhibits large and variable tilting in the plane of actin and to the sides. These motions imply that, in addition to the previously suggested flexible tail domain, there is a compliant region between the motor domain and lever arm that allows myosin VI to accommodate the helical position of binding sites while taking variable step sizes along the actin filament.
Myosin VI is an unconventional motor protein of the myosin superfamily that serves as an anchor and transporter in cellular processes such as vesicular membrane traffic, cell migration, and mitosis (Buss et al., 2004; Cramer, 2000; Frank et al., 2004; Hasson, 2003; Sweeney and Houdusse, 2007). Similar to other myosins, the myosin VI heavy chain is composed of an N-terminal motor domain that binds actin and ATP, a calmodulin (CaM) binding region, and an α-helical domain leading to the cargo binding region (Hasson and Mooseker, 1994). One CaM subunit is bound very tightly to a peptide sequence unique to myosin VI and a second CaM is bound to a consensus CaM binding IQ domain (Bahloul et al., 2004; Morris et al., 2003). The motility of myosin VI exhibits several unusual properties that have made it the subject of many structural (Ménétrey et al., 2005; Wells et al., 1999) and single molecule (Ali et al., 2004; Altman et al., 2004; Balci et al., 2005; Iwaki et al., 2006; Nishikawa et al., 2002; Ökten et al., 2004; Park et al., 2006; Rock et al., 2005; Rock et al., 2001; Yildiz et al., 2004; Yildiz and Selvin, 2005) investigations. It has been shown that myosin VI moves processively in a hand-over-hand manner toward the minus end of actin filaments, opposite to other classes of myosin. The reported measurements of myosin VI step size range between 25 nm and 36 nm with high variability (e.g. ±~12 nm (Rock et al., 2001)). Ali et al. (2004) have shown with an experimental geometry that avoids surface hindrance that full length dimeric myosin VI carries bead cargos either straight along actin or on a right-handed spiral, suggesting an average step size even larger than 36 nm.
Another well studied dimeric processive motor, myosin V, also exhibits hand-over-hand processive motility with a tightly distributed 36 nm stride. Myosin V contains six CaM or CaM-like subunits on each of its lever arms, giving these structures enough length (~24 nm each) for the two heads to bind simultaneously to actin monomers separated along the filament by 36 nm (Cheney et al., 1993; Forkey et al., 2003; Mehta et al., 1999). Only limited distortion of the structure is observed in EMs of myosin V bound to two actin monomers separated by this distance (Walker et al., 2000) and the two heavy chains join to form a strong coiled-coil α-helix at a head-stalk junction just beyond the IQ motifs (Hasson and Mooseker, 1994; Mehta et al., 1999). For myosin VI, on the other hand, the two CaM subunits would give, at most, a 7.2 nm length of lever arm, not sufficient for the dimeric molecule to span over 25–36 nm without substantial structural rearrangements. A C-terminal extension of the heavy chain has been suggested to increase the lever arm length to 10 nm (Rock et al., 2005). It has been puzzling how myosin VI achieves the large step size in spite of its short lever arm. The proximal tail domain (PTD) has been suggested to unwind to allow the molecule to stretch out to span over the filament distance of its stride (Altman et al., 2004; Ökten et al., 2004; Rock et al., 2005; Yildiz et al., 2004). This idea, though, introduces the problem that a random segment of sequence would not support a mechanical load, whereas optical trap experiments indicate that myosin VI can exert up to 2 pN without slowing (Altman et al., 2004).
Debate has also taken place over whether myosin VI is monomeric or dimeric in vivo. Expressed native myosin VI is a non-processive monomer (Lister et al., 2004); while artificially dimerized myosin VI shows robust processivity (Ökten et al., 2004; Yildiz et al., 2004). Native myosin VI monomers may dimerize via clustering through their cargo binding domain, and these dimers behave similarly to the artificially dimerized myosin VI and show processive movement (Park et al., 2006). Monomeric myosin VI also moves processively when it binds cargo (Iwaki et al., 2006).
As with other myosins, rotations, tilting and twisting of the functional domains are expected to be essential motions of myosin VI. In myosin V and II, for instance a rotational stroke of the lever arm, within the axial plane containing the actin filament, is the main structural change that moves the cargo (Forkey et al., 2003; Howard, 1997; Purcell et al., 2002; Tyska and Warshaw, 2002; Yildiz et al., 2003). Out of plane motions are used by myosin V to alter its path, possibly to avoid obstacles on its track (Syed et al., 2006). For myosin VI, if PTD is flexible, then the azimuthal angle of the myosin VI lever arm around the actin filament and the path of the molecule are expected to be very variable.
In the present work, we used single molecule polarized Total Internal Reflection Fluorescence Microscopy (polTIRF) (Forkey et al., 1999; Forkey et al., 2000; Forkey et al., 2003; Rosenberg et al., 2005) to measure angular changes of fluorescent labeled CaM in the myosin VI lever arm domain. A short smooth muscle myosin II coiled-coil region attached to the tail domain of myosin VI (Nishikawa et al., 2002) ensured dimerization of the coiled-coil domain of myosin VI, leading to robust processivity. Large tilting motions, expected from the lever arm model, were detected at the rate of stepping. The azimuthal angle changes of the leading and trailing lever arms suggest that in addition to walking straight, myosin VI often steps left and right around an actin filament. The large variability of angle, accompanied by large tilting and twisting motions outside of the actin axial plane explain the high variability of myosin VI step size and imply that there is a compliant region near the head-lever arm junction.
The single molecule polTIRF setup was described previously (Forkey et al., 2005; Rosenberg et al., 2005). Briefly, bifunctional rhodamine (BR) probes in myosin VI molecules bound to actin were excited by polarized evanescent waves generated by total internal reflection (TIR) microscopy. The polarized fluorescent emission from single rhodamine probes was used to determine its orientation in terms of its axial angle (β) relative to the actin filament, the azimuthal angle (α) around the actin filament, and the amplitude of μs wobbling (δs). In the present work, and to be described more fully elsewhere (Beausang et al., 2007), eight combinations of time-multiplexed incident directions and polarizations (s, p, 45° and 135° polarizations in both incident beams) allow unambiguous 3D resolution of individual probe dipole orientations within a hemisphere during each 80 ms detection cycle. See Materials and Methods and Suppl. Mater. Fig. S1 for definitions of the excitation paths and polarizations. The remaining ambiguity is due to the dipolar nature of the fluorescent probe. The coordinate frame of Fig. 1A is defined relative to actin, with the z axis pointing in the direction of motility, the pointed (−) end of actin for myosin VI. Because actin filaments are attached firmly to the slides in the experimental assay, myosin VI most likely binds and walks on an actin filament in the hemispherical space that faces the objective (0<βlever<180° and 0°<αlever<180°, Fig. 1). Given that assumption, the appropriate hemisphere defining the corresponding range of probe angles is tilted by the relative angle between the myosin VI molecule and the probe, defined by its two linked Cys residues. Using the available crystal structure of myosin VI (Ménétrey et al., 2005) docked onto actin (Fig. 1B) by superimposing it onto that of myosin II in the actomyosin near-rigor complex (Rayment et al., 1993), the azimuthal angle of the probe relative to that of the bound actin monomer is approximately −60°. Thus, the probe hemisphere corresponding to exclusion of the myosin VI lever from the glass slide is given by (0<βprobe<180° and −60°<αprobe<120°). One end of the probe dipole remains in this hemisphere as the lever arm rotates and the other end is restricted to the opposite hemisphere.
Fig. 2 shows three typical traces from single molecule polTIRF experiments on myosin VI at 150 μM ATP. In each figure, the top panel shows the 16 polarized fluorescence intensities from the 8 time-multiplexed incident polarizations detected by the two APDs. The BR probe photobleached to the background level at 7, 6 and 2 s respectively. The bottom panels show the angles obtained from fitting a wobble-cone model of the probe fluorescence to the 16 intensities. Sample traces recorded 500 μM ATP are given in Suppl. Mater. Fig. S2). Most (>95%) of the β (red) traces from processively stepping myosin VI molecules showed clear alternating up and down transitions, the lever arm tilting characteristic of hand over hand stepping. The azimuthal angle (α, green traces), often, but not exclusively, changed value at the same time as β, indicating large azimuthal swaying. The μs wobble (δs, bottom blue traces) varied, apparently randomly, between 30° and 60°.
Angular transitions were picked manually from the median filtered (window size = 5 samples) β traces taking the approximate angular resolution of the measurement (10°) as a lower limit for selecting steps and avoiding spikes of the traces that deflected β for only one measurement cycle, see Fig. 2. We also applied an automated step finding algorithm (Kerssemakers et al., 2006) to select transitions in the α and β traces, and the steps were very similar to those selected manually (see Suppl. Mater. Fig. S2). The distributions of dwell times (Suppl. Mater. Fig. S3) were fitted with a kinetic model having two reaction steps: a first-order step (k1, presumably ADP release from the trailing head) and subsequent ATP binding (k2), to obtain apparent rate constants for the motility process. Simultaneous fitting of the dwell times at 150 μM and 500μM ATP gave k1 = 9.84 s−1 and k2 = 0.019 μM−1s−1, similar to other reported values (Altman et al., 2004; De La Cruz et al., 2001; Park et al., 2006; Robblee et al., 2004; Yildiz et al., 2004). Dwell time data at ATP concentrations ranging down to 20 μM were also compatible with the two-step kinetic model. The stride length per angular transition of β can be calculated from product of the dwell time and the average velocity, which was measured as 81.7 nm/s at 150 μM ATP and 154.3 nm/s at 500 μM ATP. Thus the amount of translocation along actin per tilt of the CaM (the apparent stride length) of myosin VI is 37 nm and 32 nm at 150 μM and 500 μM ATP, respectively.
The distributions of the probe β values just prior to upward and downward β transitions (Fig. 3A) show two clear peaks. In the coordinate frame of Fig. 1, the trailing head lever arm has the smaller β angle and the leading head the larger β. The peak of β at the smaller angle (56°), corresponding to the trailing head, is considerably narrower (σβ_trail = 17°, comparable to that of myosin II S1 (Quinlan et al., 2005)), than the β peak at larger angle 117° (leading head, (σβ_lead = 27°,), suggesting that the leading head is more disordered or has a wider range of angles than the trailing head.
In contrast to the difference in width between the leading and trailing β distributions, the azimuthal probe angles around the actin filament, α, measured just before upward and downward β transitions show similar broad distributions (Fig. 3B). The change of azimuthal angle per step (Δα) also has large variance of 43 − 48° (Fig. 3C), 1.5–2 fold larger than that of myosin V measured under similar conditions (data not shown). The standard deviations of βprobe and αprobe angles of myosin VI bound to actin filaments in the absence of ATP (rigor probe angles) are about 8° and 12° respectively, considerably smaller than the breadth of the distributions during stepping. Thus, myosin VI stepping involves large azimuthal variations. The δs distributions were the same for the leading and trailing positions as shown in Fig. 3D.
The number of upward and downward transitions of β are almost equal (Table 1), consistent with the lever arm model alternating back and forth between leading and trailing positions. As shown in Figure 2, the abrupt changes of β were often accompanied by transitions of α. At the time of β transitions, α either increased, decreased or did not change appreciably (deflection <10°). For both up and down β transitions, there is about equal percentage of up, down, and null α transitions (Table 1). Thus, both the stepping head (trailing to leading position) and the non-stepping head (leading to trailing) showed large, essentially random, azimuthal changes. In Table 1, Δβ and Δα show the differences between angles before and after the steps. For both increasing and decreasing β, the Δβ values were noticeably larger for transitions without any change of α (Δβ = 60° and −55°) than for ones with changes in α (Δβ = 40° – 49°). Thus, larger azimuthal motions were correlated with smaller changes of β.
If the lever arm alternates between leading and trailing positions, then it returns to its original state after two transitions, i.e. leading → trailing → leading, or trailing → leading → trailing. Because the relative orientation between labeled lever arm and the BR probe is fixed, the measured angular change of the probe after two steps, termed 2Δβ and 2Δα, should be same as the angular change of the lever arm, irrespective of the relative angle between the two. When 2Δα is zero or small, the molecule is walking straight along the axis of actin. Distributions of 2Δβ and 2Δα (Fig. 4A and B) show that 2Δβ is rather narrow (σ = 20.8°), whereas 2Δα is distributed quite broadly (σ = 54.7°), about twice as much as myosin V (data not shown). Thus myosin VI walks much more wiggly than myosin V.
Fig. 4C and 4D show the proportions of different classes of 2Δα when β goes through down-up-down transitions and up-down-up transitions, respectively. In approximately 2/3 of the examples, myosin VI wiggles left and then right (or vice versa) during two consecutive steps (wiggling pattern), while in about 1/3 of the examples, it follows the same handedness over two consecutive steps (partial twirling pattern). The fractional difference between the two stepping patterns is significant (p < 0.05). Thus the likelihood for any particular azimuthal change is not completely random, but it seems to depend on the direction of the prior step.
A newly developed assay using polTIRF to monitor torque and helical motions of molecular motors detects the azimuthal rotation of sparsely labeled actin in a gliding filament assay (Beausang et al., 2006; Rosenberg et al., 2005). F-actin, sparsely labeled with tetramethyl-rhodamine at residue Cys374, actively glides on myosin bound to the microscope slide surface. In the present case, myosin VI is bound to the fused silica slide using poly-lysine. The 3D orientation of an individual fluorophore in the actin is monitored by polTIRF during the gliding to determine if it tracks straight or with a helical twist. In this “twirling assay”, the axial angle β of the actin probe is expected to be a constant (Forkey et al., 2005; Otterbein et al., 2001; Rosenberg et al., 2005), and therefore the appropriate hemisphere to quantify the angles for this type of experiment is the front (or rear) hemisphere relative to the direction of motility. If the filament twirls, then the azimuthal angle (α) of the actin changes steadily (Fig. 5), due to a net torque exerted on it by the myosin molecules. During translocation by myosin VI, approximately 20% of the observed actin filaments follow a helical path with rotation around the filament axis. In those twirling filaments, myosin VI induces a right-handed twirling motion with pitch 1.3 ± 0.1 μm (mean ± s.e.m., n = 23).
Processive motility by two-headed dimers of myosin VI has been shown to operate by the hand-over-hand mode (Balci et al., 2005; Ökten et al., 2004; Yildiz et al., 2004). The biggest puzzle with myosin VI motility is its large stride, averaging 25–36 nm in various studies (Altman et al., 2004; Balci et al., 2005; Nishikawa et al., 2002; Ökten et al., 2004; Park et al., 2006; Rock et al., 2005; Rock et al., 2001; Yildiz et al., 2004). The length of the myosin VI putative lever arm, which contains a converter domain, a unique heavy chain insert bound to a CaM, an IQ motif with a second CaM, and possibly a rigid extension of the heavy chain, is no more than 10 nm long (Rock et al., 2005). Single-headed constructs of myosin VI produce steps of up to 18 nm, requiring postulation of a ~180° rotation of the lever arm (Bryant et al., 2007; Lister et al., 2004), and in addition, a considerable diffusive search to complete the stride. The results presented in this paper suggest that the trajectory of the working stroke (taking place on the attached head during a step) may be highly variable for myosin VI. We used a construct containing a segment of the smooth muscle myosin II coiled-coil region attached to the tail domain of myosin VI to ensure dimerization (Nishikawa et al., 2002). The effects of the short myosin II segment on the stepping behavior of myosin VI were expected to be minimal because the motor domain, neck domain, and entire coiled-coil domain of myosin VI were preserved in the construct. The velocity, stepping rate and ATP dependence of the motility (Nishikawa et al., 2002; Suppl. Mater. Fig. S3) are, in fact, consistent with results from studies on other constructs (Park et al., 2006; Yildiz et al., 2004). The dynamic angular measurements support the previously proposed hypothesis of a large power stroke rotation of nearly 180° and also suggest that the leading head has considerable flexibility to find actin monomers over a large range of azimuthal positions. We conclude that variable motions of both the leading and trailing heads contribute to the variation of path and step size of myosin VI.
The kinetics of waiting times between β transitions over a range of ATP concentrations (Suppl. Mater. Fig. S3) are appropriate to the measured translocation velocities for 30–36 nm average stride implying that the tilting motions are associated with each step.
Since the probe is rigidly bound to CaM, which is bound tightly to the lever arm, the two peaks of β (Fig 3A) are expected to correspond to the pre- and post-power stroke orientations of the lever arm in the leading and trailing positions. Docking the myosin VI nucleotide-free crystal structure onto actin (Fig. 1B) results in a lever arm that is approximately parallel to the actin filament with a probe β angle of 57° (black double-headed arrow in Fig. 3A) for our bifunctional rhodamine at CaM residues 66 and 73. The peak in the β distribution at 56° is in good agreement with this angle, thus enabling assignment of this peak to the trailing (post-stroke) myosin VI head.
Crystal structures of the myosin VI leading head (pre-power stroke) are not available for a similar calculation of the lever arm angle, so we estimate it using the measured probe angles in the leading lever arm. We assume that the pivot for the myosin VI working stroke is at the joint between the converter domain and the nucleotide/actin-binding domain, and that the plane of power stroke rotation is approximately parallel to the actin filament as in other myosins (Coureux et al., 2004; Dominguez et al., 1998; Smith and Rayment, 1996). Fig. 6A shows the predicted probe β values (βprobe) and the change of azimuthal probe angle (|Δαprobe|, relative to the probe in the trailing position) as a function of the leading lever arm orientation (βlever). The 99% confidence interval of measured mean βprobe in the leading lever arm from Fig. 3A is shown as the red horizontal bar, which intersects the curve of βleadingprobe in two regions: 81° – 94°, and 182° – 196° as denoted by the black dots in Fig. 6A. These two ranges of lever arm rotation correspond, respectively, to a power stroke (approximately L·[cos(βleadinglever) − cos(βtrailinglever)]. L is the length of the lever arm) along actin of ~12 nm and ~18 nm, both of which have been reported (Bryant et al., 2007; Lister et al., 2004; Rock et al., 2005). A lever arm rotation in the 81° – 94° range, however, results in a very small azimuthal angular change (<10°, green curve in Fig. 6A), which is inconsistent with the Δα distribution in Fig. 3C. The green bar in Fig. 6A shows the 99% confidence interval of |Δαprobe| during a step (data from Fig. 3C.; instead of Δαprobe, |Δαprobe| is used here to included both trailing-to-leading and leading-to-trailing transitions) supports the larger powerstroke rotation with βlever between 182° and 190° (grey vertical bar in Fig. 6A). Therefore, our data imply that myosin VI may have a large power stroke rotation close to 180° as has been suggested from other studies (Bryant et al., 2007; Lister et al., 2004; Park et al., 2007). The slight mismatch between the zones of βlever intersecting the 99% confidence intervals of the mean βprobe and |Δαprobe| may be due to azimuthal flexibility of myosin VI.
Note that the broader distribution of βleadingprobe than that of βtrailingprobe (Fig. 3A) indicates more distortion in the lever arm. Similarly, in a FIONA experiment with myosin VI, Yildiz et al. (Yildiz et al., 2004) found that the fluorescent probe in the leading position had larger spatial variation than in the trailing position, suggesting an uncoupling of the lever arm from the motor head. In later work, those authors found that an engineered construct of myosin VI with the unique insert 2 removed moves in the opposite direction but with a similar step size compared to the native molecule; a result which is less favorable to the uncoupling idea (Park et al., 2007). In the present experiment, uncoupling between the lever arm and motor head would result in an increase of the wobbling motion (δs) of the probe in the leading state, which was not observed (Fig. 3D). Instead, we speculate that the larger variance of βleadingprobe is due to the leading lever arm adopting different orientations when it binds to different actin monomers (Fig. 6B).
The most straightforward explanation of the results is that the angle of the leading lever arm is determined by the interaction between its own flexibility and the force exerted on it by the other head through their linkage, as illustrated in Fig. 6B. Shortly after binding to actin, the leading head’s active site presumably contains ADP or is empty, and if there were no intramolecular force between the heads, then the lever arm angle would still be directed toward the minus end of actin (dashed yellow lever), in a configuration similar to what we assumed for the trailing head. Such a scenario, however, would predict only one peak in the β distribution instead of the two peaks that we found. Since we detect the regular switching of β angles at the appropriate rate for the molecules’ velocity (Fig. 2 and Suppl. Mater. Fig. S3), we conclude that there is a large force between the two heads of the molecule. We hypothesize that the leading head is flexible (green segment in Fig. 6B), possibly in the converter/unique insert region or at a ‘pliant’ region at the base of one of the CaM subunits (Houdusse et al., 2000). Another likely compliant region in the molecule is the proximal tail domain (PTD, ~80 residues after the CaM-bound IQ domain) that has been suggested to be unfolded on the basis of myosin VI’s large step (Altman et al., 2004; Ökten et al., 2004; Rock et al., 2005). If these two compliances (stretching of the PTD and rotation of the leading head lever arm) are comparable, the strain would be partitioned between them, and the leading lever arm would be bent backwards as in Fig. 6B (solid yellow lever). When a leading head binds to an actin monomer closer to the trailing head (cyan motor domain and lever), the smaller extension leads to a smaller βleadinglever. Multiple actin binding sites thus result in multiple βleadinglever angles and the observed greater variance of βleadingprobe (Fig. 3A). Our proposed role for strain in pulling the lever arm backwards, and the consequent higher angular variability of the leading head compared to the trailing head, suggests that gating of the ATPase cycle would predominantly be exerted on the leading head (Sweeney and Houdusse, 2007). There are also implications of these ideas for changes in α, as described next.
If myosin VI simply walks straight with the a ~180° power stroke rotation in the plane of actin (grey bar in Fig. 6A and yellow leading head in Fig. 6B), then the changes of β and α should be anti-correlated, i.e. when βprobe increases from 56° to 117° during the power stroke, α would decrease by ~60° and vice versa (Fig. 6A). In addition to anti-correlation between Δβ and Δα, we also observed correlation (both angles increase or decrease) between Δβ and Δα and non-correlation (e.g. steps of β, without change of α), see Table 1. The large variation of βleadingprobe (Fig. 3A), Δα (Fig. 3C), power stroke size (Altman et al., 2004; Rock et al., 2001) and step size (Ökten et al., 2004; Park et al., 2006; Yildiz et al., 2004) of myosin VI, all suggest large azimuthal flexibility of myosin VI that binds to widely varying actin monomers. For example, the actin monomer 6 subunits along the short-pitch helix from a starting point is tipped 83° azimuthally rightward around actin and the 7th monomer is tipped leftward −83°. The different types of Δβ − Δα correlation in individual molecules (Table 1) suggest that both lever arms may flex or rotate to accommodate the azimuthal difference between the two heads.
As indicated in Fig. 6A, steps giving changes of β less than 100° would be expected to be accompanied by |Δα| of less than 15°, so the much larger changes of α observed are indicative of large azimuthal motions of the head binding to different monomers along the actin helix. In a theoretical calculation by Lan and Sun, 2006, the energy cost for myosin VI binding to monomers 6, 7, 9 and 11 are comparable, whether the stroke is either 12 nm or 20 nm. Lastly, molecules could differ from each other in the local probe orientation if some of the tightly bound CaM subunits at the unique insert exchanged with labeled CaM.
After two steps of β, the lever arm should return to its former position, e.g. leading→trailing→leading. As explained in Results, the probe azimuthal angle change, 2Δα, following two consecutive steps indicates helical (two successive α changes occur in the same direction) or wiggly (successive α changes occur in opposite directions) motion irrespective of the relative orientation between the probe and the lever. The much broader width of the distribution of 2Δα values relative to 2Δβ (Fig. 4) confirms that large azimuthal rotations are common, but α reverses its change (wiggles) more often than it changes twice in the same direction (Fig. 4 bottom panels) contributing to a relatively straight overall path or gradual spiral. Strain in an extreme azimuthal step might introduce intramolecular torque which would tend to restore the path in the following step.
The actin filaments in the single molecule polTIRF experiments were attached rather firmly to the microscope slide by biotin-streptavidin linkages, probably restricting motions of the myosin VI molecules from squeezing between the filament and the slide. The broad and almost evenly distributed α angles (Fig. 3B) imply that the molecules landed on the filaments over a broad range of angles and did not exhibit a strong preference for overall left-handed or right-handed motion.
In experiments designed to determine the path of myosin VI in the absence of physical restrictions around the actin filament, Ali et al. (Ali et al., 2004) studied motions on actin filaments suspended away from the slide surface. In that geometry, 80% of the myosin VI molecules walked straight, and 20% followed a right-handed helical path with a gradual pitch of 2.3 μm. Since the long pitch helix of the actin filament is right-handed, they concluded that their results implied an even longer step than the 36 nm cross-over distance of the two long-pitch strands. The twirling experiments reported here, measuring the path of filaments in a gliding assay, are compatible with Ali et al.’s results: many filaments went straight and ~20% twirled with a gradual right-handed helical motion (Fig. 5). A right-handed helical motion suggests that myosin VI leading head binds actin monomer 6 or 15 more often than actin monomer 7, 9 or 11. This relationship is energetically unfavorable (Lan and Sun, 2006). However, if rotation plane of the power stroke is slightly tilted toward the right side of the actin filament, the diffusive head would have better chance to bind actin monomer 6 and 15 than 7 and 11, thus causing a right handed helical motion with a long pitch.
Most of the time, myosin VI molecules walk in an overall relatively straight path made up of chaotic left-right wiggling. The tilting motions measured here by polTIRF are compatible with a surprisingly large (around 180°) rotation during the power stroke, but other angles are produced sometimes. When the heads bind to actin monomers at markedly different azimuths, both of the lever arms seem to be distorted accommodating the extra intermolecular stress. These features enable myosin VI to walk relatively straight while sampling among several possible actin binding sites.
The proximal tail domain and the postulated compliant region between the motor domain and the lever arm junction provide great flexibility for motions and alternative configurations of myosin VI binding to actin. Partition of strain between the two compliances would lead to geometries that vary the angle of the leading lever arms shown in Fig. 6, and allow biochemical gating of the leading head for functional roles of processive transport and anchoring (Buss et al., 2004; Sweeney and Houdusse, 2007). In cells, actin forms diverse structures such as parallel bundles and meshes. The flexible nature of myosin VI that makes it possible to find several favorable binding sites on single actin filaments presumably enables it to bind to sites on two nearby filaments and thereby explore pathways, routes and anchoring positions on the diverse actin structures in cells.
G-actin was obtained from rabbit skeletal muscle and purified as described by Pardee and Spudich (Pardee and Spudich, 1982). Biotinylated, Alexa647-labeled F-actin was prepared from G-actin, Alexa647 actin (Molecular Probes, Carlsbad, CA), and biotin-actin (Cytoskeleton, Denver, CO) at 1 μM total actin monomer concentration with a ratio of 5:1:1 of G-Actin: Alexa647: biotin and stabilized with 1.1 μM phalloidin (Molecular Probes, Carlsbad, CA). 0.3 % rhodamine labeled F-Actin was prepared from 6′-IATR rhodamine-actin (Corrie and Craik, 1994) and stabilized with 1.1μM phalloidin. The M6HMMsRod Myosin VI construct, a truncated myosin VI with a fragment of smooth myosin II rod, was expressed in SF9 cells and purified as described previously (Nishikawa et al., 2002). This construct contains the mouse myosin VI sequence, residues Met1 – Leu1023, including the motor domain, neck domain, and coiled-coil domain, part of the chicken smooth muscle tail, residues Gln1111 – Asp1566, and a 6-His affinity tag. The portion of smooth muscle rod ensures dimerization of the expressed myosin VI, but is too short to form filaments.
Residues Pro66 and Ala73 of chicken calmodulin (CaM) were mutated to cysteine and labeled with bifunctional rhodamine (BR-I2) as described by Forkey et al. (Forkey et al., 2003). BR-I2 was a generous gift from Dr. J.E.T Corrie (Corrie et al., 1998). Myosin VI was labeled by exchanging endogeneous CaM with exogeneous BR-I2 labeled mutant CaM at low stoichiometry.
M6BH buffer (pH = 7.6) contains 25 mM KCl, 20 mM Hepes, 2 mM MgCl2, and 1 mM EGTA in deionized water. M6BH+ buffer, the motility buffer for single molecule motility assays, is M6BH plus 10 mM dithiothreitol (DTT) and 100 μg/ml wild-type CaM (WT-CaM, expressed in bacteria as described by Putkey et al. (Putkey et al., 1985)). The motility buffer for actin twirling assays is M6BH+ buffer plus 500–1000μM ATP, 10 mM phosphocreatine (Sigma P-7936), 0.3 mg/ml creatine phosphokinase (prepared daily from powder, Sigma C3755), and 50 mM DTT.
In the single molecule polTIRF setup described previously (Forkey et al., 1999; Forkey et al., 2000; Forkey et al., 2005; Forkey et al., 2003; Rosenberg et al., 2005), time multiplexing between two incident paths, each polarized s and p relative to the scattering plane, resulted in ambiguity of the deduced orientations outside 1/8 of a sphere. In the present work, time-multiplexed 45° and 135° polarizations were added in each of the incident directions in order to break these symmetries. Eight combinations of time-multiplexed incident directions and polarizations and two simultaneously recorded emission intensities are accumulated in each 80 ms interval, giving 16 different polarized fluorescence intensity traces (see Suppl. Mater. Fig. S1 for details). This arrangement allows unambiguous 3D resolution of individual probe dipoles within a hemisphere. With the current setup, the average orientation of the probe during each 80 ms detection cycle is resolved to within approximately 10°. Procedures for calibration and data collection can be found in Forkey et al., 2003; Quinlan et al., 2005; and Rosenberg et al., 2005.
A pre-cleaned fused silica slide (Quartz Scientific) was freshly treated in an ion plasma cleaner for 5 min. and spin-coated with 2 mg/ml PMMA (poly(methyl methacrylate)) (Aldrich Chemical, no. 37,003-7) in methylene chloride. The PMMA coated slide was assembled into a 10–20 μl flow chamber with a glass coverslip and double-sided adhesive tape. Actin was adhered to the surface and flow-aligned with the microscope x axis by successive incubations with 1 mg/ml biotinylated BSA (Sigma, A-8549), 0.5 mg/ml streptavidin (Sigma, S-4762), and 100 nM biotinylated, Alexa647-labeled F-actin, each followed by washes with M6BH+ buffer. Myosin VI, containing BR-labeled CaM, was introduced into the sample chamber at 10 – 1,000 pM in M6BH+ and ATP as indicated.
A flow chamber was assembled as above. 20 μl of 3.5 mg/ml poly-L-lysine HBr (MP Biomedicals, ImmunO 71120G) was flowed into the chamber, incubated for 1 min. and washed with M6BH+ buffer. 20 μl of ~0.2 mg/ml unlabeled myosin VI was introduced into the sample chamber and incubated for 2 min. Exposed poly-L-lysine and fused silica was blocked by 2 × 20 μl washes with 5 mg/ml BSA. 2 × 20 μl of pre-sheared, unlabeled F-actin was added in the absence of ATP to block any inactive myosin heads, and excess actin was removed by addition of M6BH+ buffer at 2 mM ATP, followed by 2 washes of M6BH+ without ATP. Actin filaments, sparsely (0.3%) labeled with rhodamine, were added in M6BH+ buffer, and then motility buffer was added to initiate filament gliding and polTIRF measurements were made of the rhodamine orientation.
The work was funded by NIH grant AR26846 and NSF grant NSEC DMR04-25780. We acknowledge helpful discussion and comments by John H. Lewis, Drs. Jennifer L. Ross and Jody A. Dantzig.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.