Force-producing ADP state of myosin bound to actin

Abstract

Molecular motors produce force when they interact with their cellular tracks. For myosin motors, the primary force-generating state has MgADP tightly bound, whereas myosin is strongly bound to actin. We have generated an 8-Å cryoEM reconstruction of this state for myosin V and used molecular dynamics flexed fitting for model building. We compare this state to the subsequent state on actin (Rigor). The ADP-bound structure reveals that the actin-binding cleft is closed, even though MgADP is tightly bound. This state is accomplished by a previously unseen conformation of the β-sheet underlying the nucleotide pocket. The transition from the force-generating ADP state to Rigor requires a 9.5° rotation of the myosin lever arm, coupled to a β-sheet rearrangement. Thus, the structure reveals the detailed rearrangements underlying myosin force generation as well as the basis of strain-dependent ADP release that is essential for processive myosins, such as myosin V.

Keywords: force generation; molecular motor; myosin V; transducer.

Fig. 1.

Actomyosin ATPase cycle. Scheme of the myosin motor domain with its subdomains [U50, L50, N-terminal (dark gray), converter (green)] and its interaction with F-actin along the ATPase cycle. The strong-binding ADP state is the only state that exhibits high affinity for both actin and nucleotide.

Fig. S1.

Transducer in the Rigor state and comparison of the myosin V Rigor-like X-ray structure to the Rigor F-actin–myosin model from CryoEM. (A) The myosin transducer consists of a seven-stranded β-sheet (first strand on the right, distal from actin) and associated loops. The transducer corresponds to the central region at the interface between the U50, L50, and the N-terminal subdomains that mediates the closure of large cleft between the U50 and L50 subdomains and thus greatly influences the actin-binding interface. The transducer is also in direct contact with the nucleotide-binding site. Highlighted are the nucleotide-binding loops Switch I (purple), Switch II (orange), and the P-loop (green). (B) The Rigor-like structure (2) (gray; PDB ID code 1OE9) is a very good model of the Rigor state (blue), although the myosin was not bound to F-actin (light yellow) in the crystal. Larger deviations occur only distal from F-actin starting at the pliant region (green) of the lever arm. The particular position of the lever arm in the crystal is stabilized by the packing in the crystal.

Fig. 2.

CryoEM image of actomyosin and reconstructed 3D density maps of actomyosin in the ADP and Rigor states, both with a fitted atomic model. (A) Image acquired by transmission electron cryoEM of F-actin decorated with myosin V_a motor domains with truncated lever arm, in the nucleotide-free Rigor state. The defocus value is 1.7 μm, and the value in the Inset is 5 μm. (B) The reconstructed density maps show a ∼10° difference in the lever arm position, as highlighted in the Inset, with the ADP map in orange. The protein backbones of the atomic models obtained by MDFF are shown in ribbon representation for six actin subunits (light blue in ADP; light yellow in Rigor) and four myosin motor domains (bright yellow in ADP; blue in Rigor). Also see Movies S1 and S3.

Fig. S2.

Comparison of different starting models of myosin for the MDFF procedure. (A) Myo5a Post-Rigor model (red; PDB ID code 1W7J) used for the fitting in the Rigor Myo5 EM density map. Coordinates prior to and after fitting (red and blue, respectively) are compared. (B) Myo5a Rigor-like model (violet; PDB ID code 1OE9) used for the fitting in the Rigor Myo5 EM density map. Coordinates prior to and after fitting (violet and green, respectively) are compared. (C) Final coordinates for the rigor map from two independent flex-fitting experiments: from the initial Post-Rigor (PDB ID code 1W7J) or Rigor-like (PDB ID code 1OE9) starting models are compared (blue and green, respectively); note the similarity of position of the subdomain, in particular, the N-terminal and β-sheet. (D) This N-terminal region is shown in detail (for the two final models of the Rigor flex fitting (blue and green) and compared with the starting Post-Rigor coordinates (red). Note how the β-sheet (arrow) and helices of the N-terminal domain converge in the two Rigor models, regardless of the position the β-sheet and helices started from (either initial Rigor-like or Post-Rigor coordinates).

Fig. S3.

Comparison of different actin models and F-actin symmetries. All of the actin subunits correspond to F-actin structures determined by cryoEM. (A) Note the similarity between the ADP state (orange) and the Rigor actin models (blue). The rmsd for the C_α atoms of residues 5–371 is 0.8 Å. (B) The Rigor actin model (blue) is compared with actin from undecorated F-actin (20) (PDB ID code 3MFP; rmsd 1.6 Å). (C) The Rigor actin is compared with actin from a F-actin–Tropomyosin–myosin 1 Rigor complex (AM1-TM) (25) (PDB ID code 4A7L; rmsd 1.9 Å). (D) The Rigor actin is compared with actin from a F-actin–Tropomyosin complex (A-TM) (21) (PDB ID code 3J8A; rmsd 1.3 Å). Note the small deviation found for our model compared with the most recent structures solved at high resolution (PDB ID code 3J8A). (E) Although the actin subunits are quite similar, the symmetries of their respective F-actin structures are differing slightly. Shown are the estimated azimuth angles of the different F-actin complexes. Rigor and ADP state refer to our myosin V data. The azimuth angle for the F-actin/Cofilin complex is also shown for comparison (68).

Fig. 3.

Resolution assessment of the Rigor and ADP state reconstructions, examples of details, and differences in the maps. (A) Surface coloring of the reconstructions [Rigor (Upper); ADP state (Lower)] according to local resolution, as determined by ResMap (28). Color coding is as follows: dark blue, better than 7 Å; cyan, better than 8 Å; green, better than 9 Å; and yellow, better than 10 Å. As expected for helical objects, their azimuthal alignment error results in lower resolution at a higher radius. Note the 7- to 8-Å resolution of actin and the myosin motor domain, which is an ideal starting point for the MDFF. (B) Detail of the HCM loop (green circle) and Loop 2 (yellow circle). Whereas for the HCM loop, a complete density is observed, Loop 2 density is observed only for its starting and ending α-helices at the surface of the myosin motor domain. ResMap assigns the local resolution as less than 15 Å (gray surface). (C) Detail of the density of the Rigor and ADP reconstructions for the transducer β-sheet (red circle). Looking down the β-sheet across the strands, one can see a layer of density, which is clearly tilted between the Rigor (blue) and ADP (yellow) states. Also see Movie S4.

Fig. S4.

Residual density for Loop 2 at the actin interface of the Rigor state. The density of Loop 2 (light blue, residues Q594–T635) is located close to F-actin (yellow), and the density on both ends that has been modeled as helices on either end of the loop clearly indicates the orientation of the loop. Because this large loop (∼40 residues in myosin V_a) is not resolved by crystallography, we modeled ab initio the loop into the density, followed by MD simulations. Note that a similar position for Loop 2 is also observed in the map of the Strong-ADP state (not shown). The C-terminal basic residues of the loop are positioned close to aspartates D24 and D25 (red) of subdomain 1 of actin. Our Loop 2 model differs greatly in structure and position on F-actin compared with the model built for Loop 2 of myosin II (24) (also see Fig. S5).

Fig. S5.

Comparison of actin–myosin interfaces. (A, C, and D) Rigor interfaces between F-actin and myosin V (A), myosin II (24) (C), and myosin I (25) (D) The actin-binding loops of myosin are highlighted as follows: green, HCM loop; red, Loop 2; orange, Loop 4; and yellow, Loop 3. The DNase1-binding loop of actin is shown in cyan. (B) Overlay of myosin V (light colors) and myosin II structures superimposed with the L50 subdomain (lower part with Loop 3), oriented to expose its actin-binding site on the right. Not only does the Loop 2 conformation differ greatly but so does the position of the U50 subdomain and the subdomain’s surface loops HCM and Loop 4. Furthermore, note the different positioning of the lever arm between myosin V and myosin II (helix in C, running through the lower right corner). Comparison reveals that the Rigor myosin II lever arm direction is closer to the myosin V ADP state than to its Rigor position.

Fig. S6.

Resolution of the cryoEM-reconstructed densities. FSC curves of different areas of interest created by masking. Both densities show a resolution gradient with highest resolution for actin and lowest resolution for the light chain (LC) density at largest radius from the filament axis (compare ResMap coloring of reconstruction as shown in Fig. 3). According to the 0.5 FSC criterion, the Rigor density has a resolution of about 9 Å (A) and the ADP state has a density of about 9 Å (B) for the myosin motor domain and actin. The density of the light chain region in less well-resolved, especially in the ADP state. It should be noted that the 0.5 FSC criterion gives lower resolution than the real space maximum-likelihood analysis by ResMap (28).

Fig. 4.

Comparison of myosin models in Rigor and ADP state (from cryoEM maps of Actomyosin), with X-ray myosin structures in either Weak-ADP or Post-Rigor state. (A) The myosin models (blue, Rigor state; orange, ADP state with yellow nucleotide) as found in the complex with F-actin (three actin subunits in yellow) show large deviations distal from actin. (B) Zoom into the nucleotide-binding site. On the left, coordination of the nucleotide in Strong-ADP state compared with the Rigor conformation (light gray). Note the light gray P-loop rigor conformation that is found remote from Switch I. In Strong-ADP (color), both Switch I and P-loop are positioned to participate in ADP coordination. On the right, coordination of ADP in the low-actin affinity Post-Rigor structure (11) (color, PDB ID code 1W7J, with superimposed N-terminal subdomain) differs greatly compared with that in the Strong-ADP state (light gray). Also see Fig. S8. (C and D) Nucleotide-binding site densities and fit of myosin models (ADP stick model in yellow). (C) Density of the ADP state. (D) Rigor density. The densities are depth-cued and clipped to emphasize the shape of the active site. Note that in the ADP state, the density indicates full occupancy of nucleotide, but it does not in Rigor. Note also that the map density supports the position of the P-loop being different in the two structural states. Also see Movie S5.

Fig. S7.

Comparison of different models of myosin resulting from MDFF procedure. Models of myosin V after flex fitting in the ADP EM map (red and pink) and in the Rigor EM map (blue and green) are compared. (Upper) Overall views show how the U50 and L50 subdomains superimpose nicely for all four models, whereas consistent shifts in the Nter, converter, and lever arm helix are found. (Lower) Zoom to highlight the differences in the Nter subdomain position and on the differences in the nucleotide-binding pocket that result in changes in the Nter and converter subdomains orientations.

Fig. S8.

(A) Loop 1 in Rigor (Left) and Strong-ADP (Right) density maps. The structures of ADP state and Rigor are shown as in Fig. 4B (gray, Rigor; colored, ADP state). The Rigor (Left) and ADP (Right) density maps differ for Loop 1 and helix HF. This supports the existence of a conformational change in this region upon ADP release. (B) Change of the transducer (seven-stranded β-sheet and associated loops) near the active site, together with nearby structural elements (blue and colored, Rigor; orange, ADP state). There is a difference in bending of the transducer. Also see Fig. 3C, Fig. S9, and Movies S4 and S6. (C) The Post-Rigor transducer (gray, with superimposed U50 subdomain) differs in conformation compared with Rigor (blue). The SH1 helix and the relay communicate these changes to the converter and lever arm.

Fig. S9.

β-Sheet characterization of the transducer in the Strong-ADP state compared with the Rigor state after alignment of the entire atomic models. (A) rmsd of the C_α atoms of the seven β-strands between the ADP and Rigor states (strand 1, distal to actin; strand 7, proximal to actin). (B) Twist analysis of the β-strands. Shown is the angle between the vectors running through the strands in the ADP and Rigor states, which indicates a change in twisting of the strands. The twist is gradually decreasing in two groups: from strand 1 to 4 and from strand 5 to 7. Note that the two states are compared after alignment of F-actin to which the motors are bound to. (C and D) Analysis of the sheet curvature via polygonal lines connecting selected C_α atoms of the strands as illustrated in E (line 1 connects strand 1 and 2...; blue, Rigor; orange, ADP state; yellow, nucleotide in its orientation in front of the sheet in the ADP state). (C) The angle of the connecting lines between the ADP and Rigor states indicates a change in curvature. Besides changes at both ends of the sheet, the bending direction is changing between β-strands 3 and 4 (connecting line 3), resulting in a larger deviation of strands 1, 2, and 3, as indicated by the rmsd in A. (D) Likewise, the angle with the first connecting line shows the different curvature of the β-sheet in ADP state compared with Rigor. The dotted lines depict an angular deviation from lines 3 to 4, which implies a change in curvature around β-strand 4. Structurally, strand 4 is connected directly to the P-loop and belongs together with strands 1, 2, and 3 to the N-terminal subdomain, whereas strands 5, 6, and 7 belong to the U50 subdomain. Thus, strand 4 is located directly at the interface of the rotating N-terminal subdomain upon ADP release and the U50 subdomain, which is acting less dynamically during the transition from ADP state to Rigor.

Fig. 5.

Change within the N-terminal subdomain, consequences for the lever arm, PPS and extrapolated full length, two-step powerstroke. (A) The N-terminal and lever arm changes upon ADP release from ADP state (opaque) to Rigor (transparent) resulting from a rigid body rotational movement connected to the differences in transducer conformation (yellow, relay; red, SH1 helix). Also see Movie S7. (B) Large converter and lever arm swing from PPS (myosin V_c, opaque) to Rigor (transparent). Also see Movie S8. (C) Docking of the full-length myosin V six IQ lever arm (67) (PDB ID code 2DFS) by superimposing the converter and the first IQ motif. We extrapolate a two-step powerstroke with a larger first step [PPS (green) to ADP (orange) state], which is greatly influenced by the docking of the PPS to actin and thus exhibits larger inaccuracies than the second step [ADP to Rigor state (blue)]. The lever arm was assumed to stay as a rigid body.

Fig. S10.

Prepowerstroke structures of different myosins. Myosin V_c (green) is compared with smooth muscle myosin II (30) (PDB ID code 1BR1) (A), myosin VI (32) (PDB ID code 2V26) (B), and myosin I (33) (PDB ID code 4BYF) (C). The structures are very similar; only the position of the converter differs, leading to different directions of the lever arm. The PPS of myosin V is the closest to myosin II. The differences are larger between myosin VI or myosin I and myosin V.