The myosin X motor is optimized for movement on actin bundles

Abstract

Myosin X has features not found in other myosins. Its structure must underlie its unique ability to generate filopodia, which are essential for neuritogenesis, wound healing, cancer metastasis and some pathogenic infections. By determining high-resolution structures of key components of this motor, and characterizing the in vitro behaviour of the native dimer, we identify the features that explain the myosin X dimer behaviour. Single-molecule studies demonstrate that a native myosin X dimer moves on actin bundles with higher velocities and takes larger steps than on single actin filaments. The largest steps on actin bundles are larger than previously reported for artificially dimerized myosin X constructs or any other myosin. Our model and kinetic data explain why these large steps and high velocities can only occur on bundled filaments. Thus, myosin X functions as an antiparallel dimer in cells with a unique geometry optimized for movement on actin bundles.

Figure 1. Myosin X dimerization region.

(a) Blueprint of the myosin X motor. (b) X-ray model of IQ3-SAH-CC. Yellow=IQ3, blue=SAH region, green=anti-parallel coiled-coil. (c) The dimerization region of the IQ3-SAH-CC structure (green) is compared with the short antiparallel coiled-coil structure (grey, 2LW9). (d) Sequence alignment of different myosin X for the IQ3-SAH-CC region. Note the variability for the SAH region. The region (E847-E884) includes a number of hydrophobic residues, unlike the more proximal portion of the SAH. Note that the boundary between the SAH and the dimerization region cannot be predicted from the sequence. The IQ3-SAH-CC structure shows that the region (E847-E884) forms a SAH, rather than possibly being part of the dimerization region, as studies of myosin X chimeras would suggest (Supplementary Table 1). (e) Structures explored during the MD simulations of the IQ3-SAH-CC/CaM complex. (f) Variation of the distance between residues 813 of chains A and B during the MD simulation, describing the end-to-end distance of the SAH-coiled-coil region. The black line is the moving average with a 125 ps window; the grey envelope shows the actual values. The time-series shows that after ∼20 ns, the distance stabilizes at 25.7 nm on average, with a s.d. of 0.22 nm over the last 80 ns.

Figure 2. Histogram and density plots of myosin X full length and HMM step-size data.

Histograms (open vertical bars) and kernel density estimate plots (dashed curves) generated by the density function of the step-size data of myosin X full length and HMM are presented along with fitting components of multiple Gaussian distributions (solid curves; See details of the fitting procedure in Supplementary Fig. 5). (a) Myosin X full length on single F-actin filaments. Two Gaussian component distributions were fitted. The values for the best two Gaussian fits are −26±15 (s.d.) nm, and 36±14 (s.d.) nm, N=214. (b) Myosin X full length on fascin-bundled F-actin filaments. Four Gaussian component distributions were best fitted. The fitted values for four Gaussian components are: −33±14 (s.d.) nm, 19±7 (s.d.) nm, 38±7 (s.d.) nm and 52±5 (s.d.) nm, N=393. (c) Myosin X HMM on single F-actin filaments. The values for the best two Gaussian fit are −33±15 (s.d.) nm, and 39±13 (s.d.) nm, N=238. (d) Myosin X HMM on fascin-bundled F-actin filaments. Best fits for four Gaussian component distributions were obtained. The values for the Gaussian components are −28±18 (s.d.) nm, 17±6 (s.d.) nm, 40±9 (s.d.) nm and 57±2 (s.d.) nm, N=178. For the HMM on a dimer, the backward steps were best fit as two populations (Supplementary Fig. 5e), with steps of −49±6 (s.d.) nm and −20±11 (s.d.) nm.

Figure 3. Pre-powerstroke and rigor states of myosin X.

(a) Overview of the myosin X pre-powerstroke state (motor domain, blue; lever arm (green, pink, yellow); N-terminal SH3 domain (green). (b) Comparison of pre-powerstroke converter positions. (c) Comparison of the SH3 domain of myosin V (green) and MyoVc (magenta); orientation as in (a–e). (d) Interactions of the converter (green) with the N-terminal subdomain (blue) stabilize the specific orientation of the myosin X pre-powerstroke converter. The relay (yellow) and the SH1 helix (red) also specify the converter position. (e) Comparison of the myosin X pre-powerstroke state with that of myosin Vc (two molecules in the asymmetric unit are shown in magenta). In (b) the converter position closest to that of myosin X is shown. (f) Cryo-EM reconstruction of the rigor myosin X motor truncated after the converter, bound to actin filaments reveals a novel position of the converter that places the lever arm helix nearly parallel to the actin helical axis. The density map and corresponding molecular model are rendered in slate blue for the core of the myosin motor, green for the converter and grey for actin. A green rod indicates the extrapolated position of the lever helix, based on the orientation of the converter. The modelled coordinates of the N-terminal SH3 domain are depicted in green. An arrow indicates the direction of myosin force production. (g) Close-up of the rigor density for the myosin X converter, with coordinates from crystallized rigor-like myosin V (pink ribbons) superimposed. (h) Similar to (g) but with a fitted model of the myosin X converter, rendered in green; a rotation of ∼30° with respect to myosin V is predicted from this fit. See also Supplementary Fig. 9 and Supplementary Movies 1 and 2. (i) View of the fitted coordinates as seen from the opposite face of myosin as panels (f–h); this view reveals that the predicted position of the lever arm helix in rigor is incompatible with the conformation of the SH3 domain found in rigor myosin V (magenta). Steric clashes of main-chain atoms are indicated by coloured spheres; (Supplementary Movie 1). (j) Low-pass-filtered rendering of the rigor myosin X density map is consistent with movement of the SH3 domain away from the lever arm helix in order to accommodate the new lever position. Modelled position of the SH3 domain in myosin X is identical to that in (e).

Figure 4. Power strokes of myosin X and myosin V derived from X-ray crystallography and cryo-electron microscopy.

Comparison of the powerstrokes of myosin V and myosin X motors in two orientations, using U50 and L50 subdomains for superimposition. Note that the powerstroke do not have a significant azimuthal component. The stroke at the end of 6 IQ of myosin V is 25 nm, while that for myosin X at the end of 3IQ-SAH is ∼48 nm.

Figure 5. How myosin X makes up to 55 nm step.

(a) Model of a two-headed myosin X using the structures solved here (SAH-CC-SAH of 24 nm as found in the crystal structure, or ∼25.7 nm, the average found in MD simulations). (b) Left: when the heads are flatten, the molecule can reach up to 66 nm as observed for myosin X in EM grids. Right: bending of the lever arm in particular in the SAH allows to explore different spacing between heads and thus find different hot spots. While 52–57 nm is favoured (molecule flat and straight) 40 nm is also possible. Twenty nanometres requires more bending and is thus less favourable. (c) Model of myosin X (rear head grey) stepping on an actin bundle. The parameters for the actin bundle were derived from ref. . The hot spots are marked with red. Note that 52 nm is favoured on the next filament, while 38 nm would occur if the lead head would step by exploring the bundle. Note that 52–60 nm is not possible on the filament in which the rear head is bound. (d) Hot spots for myosin binding on a single F-actin filament. A model of MD-3IQ is modelled as the rear head (grey) and the lead head (colour). Note that the helicity of F-actin is not favourable for the lead head attachment at 47–57 nm from the rear head, while 36–41 nm steps are possible.