Interactions between relay helix and Src homology 1 (SH1) domain helix drive the converter domain rotation during the recovery stroke of myosin II

Abstract

Myosin motor protein exists in two alternative conformations, prerecovery state M* and postrecovery state M**, on adenosine triphosphate binding. The details of the M*-to-M** transition, known as the recovery stroke to reflect its role as the functional opposite of the force-generating power stroke, remain elusive. The defining feature of the postrecovery state is a kink in the relay helix, a key part of the protein involved in force generation. In this article, we determine the interactions that are responsible for the appearance of the kink. We design a series of computational models that contain three other segments, relay loop, converter domain, and Src homology 1 (SH1) domain helix, with which relay helix interacts and determine their structure in accurate replica exchange molecular dynamics simulations in explicit solvent. By conducting an exhaustive combinatorial search among different models, we find that: (1) the converter domain must be attached to the relay helix during the transition, so it does not interfere with other parts of the protein and (2) the structure of the relay helix is controlled by SH1 helix. The kink is strongly coupled to the position of SH1 helix. It arises as a result of direct interactions between SH1 and the relay helix and leads to a rotation of the C-terminal part of the relay helix, which is subsequently transmitted to the converter domain.

Fig. 1

Cartoon explaining two alternative states of ATP-bound myosin head: pre-recovery stroke state M* and post-recovery state M**. The recovery stroke consists of two steps: 1) Closure of switch II loop in M* state and 2) The resulting cascade of conformational changes that leads to the bending of the relay helix and rotation of the converter domain in M** state. The two events occur in physically distinct locations, highlighted by squares, separated by 40Å one from another.

Fig. 2

Myosin fragment containing converter domain, relay helix and loop and SH1 helix in a) M* and M** states aligned along the N-terminal part of the relay helix and b) M* conformation with different parts colored appropriately. It is assumed that the structural elements highlighted in b) form a motif within which folding of the relay helix into two alternative conformations, M* and M**, is encoded. Interactions responsible for that folding are determined by the combinatorial search described in the text. All pictures presented in this work were generated using PyMol

Fig. 3

Residue-specific structural characterization: a) Secondary structure observed in our simulations for the isolated relay helix. Two most populated non-random structures, α-helix and 3₁₀ helix, are shown. The arrow indicates the boundary of the constrained region. b) Hydrogen-bond pattern observed for the same model in equilibrium and short simulations started from M** structure. Probability of forming a hydrogen bond between oxygen atom of residue i and nitrogen atom of residue i+4 or i+5 is shown. The α-helical pattern is broken in short simulations with the appearance of characteristic i,i+5 bonds at the site of the kink. No such bonds are seen in equilibrium sampling.

Fig. 4

α-helical content observed for the residues in the model that contains relay helix only, red line, and relay helix plus relay loop, black line. The relay loop is seen to enhance helicity in the C-terminal part of the relay helix.

Fig. 5

Most populated conformations observed in our simulations, yellow, in comparison with the experimental structure M*, orange. The relay loop is seen to fold up into a number of alternative conformations with populations ranging from 9% to 30%. The side chain of F487 is seen to occur on both sides of the loop.

Fig. 6

Most populated structures observed in our simulations for RH/RL/CD, yellow, and RH/RL, green, models, in comparison with the experimental M* fragment, orange. The relay helix is identical in all three structures. The relay loop and the converter domain adopt a different conformation than in the crystal structure.

Fig. 7

Most populated structures observed in our simulations of RH/RL/SH1 models. A) RH/RL/SH1* model shown in yellow, in comparison with the experimental M* fragment, orange. Good agreement between experiment and theory is seen. B) RH/RL/SH1** model in comparison with M** state. The most populated structure is dominated by interactions between charged residues K690, E490, E493 and K498, which is inconsistent with the crystallographic M** state.

Fig. 8

Three dominant conformations, according to the structure of their relay helix, observed in RH/RL/CD/SH1* simulations. Corresponding populations are shown above each structure. Two most populated clusters have the converter domain rotated similarly to the crystallographic M* state. In the third cluster, the rotation is similar to M** state. Experimental M* structure is shown in orange. The simulations demonstrate direct M*-to-M** transition.

Fig. 9

Same as Figure 8 but for RH/RL/CD/SH1** simulations. Four dominant conformations are shown along with their populations. Three most populated clusters have the converter domain rotated similarly to the crystallographic M** state. In the fourth cluster, the rotation is similar to M* state. The simulations demonstrate spontaneous M**-to-M* transition.

Fig. 10

Most populated structure observed in our simulations of RH/RL/CD/SH1m model, yellow, in comparison with the experimental M** fragment, orange. White lines indicate the dihedral angle Θ used to characterize the rotation of the converter domain.

Fig. 11

Free energy map (shown in units of kT, where k is the Boltzmann’s constant) as a function of the dihedral angle Θ and RMSD from the experimental M* structure over C_α atoms of the relay helix fragment obtained in the simulations of a) RH/RL/CD/SH1* model and b) RH/RL/CD/SH1** model. The location of experimental M* and M** states is highlighted. Strong correlation between RMSD and Θ is seen.

Fig. 12

Distribution function of the dihedral angle between residues K477, Q479, T742 and A748 obtained in this work for different computational models. Solid black lines indicate the angles of the crystallographic M* and M** states. Deletion of the chemical bond in RH/RL/CD/SH1m model imparts additional rotational freedom to the converter domain. The main conformation, however, remains M**-like.

Fig. 13

An illustration of how the minimal model that captures two alternative conformational states in a fragment of a protein is built. In a multi-domain protein, the conformations of the domain of interest A are determined by a) intra-domain interactions and b) interactions with other domains. By considering different combinations of domains one can establish the interactions responsible for the concerned transition.