Myosin’s powerstroke transitions define atomic scale movement of cardiac thin filament tropomyosin

Changing levels of sarcoplasmic Ca²⁺ control muscle contraction by modulating the effects of the troponin–tropomyosin system located on actin-containing thin filaments (Gordon et al., 2000; Lehman, 2016). In this process, troponin and tropomyosin regulate the access of myosin heads to their actin-binding sites and thus the myosin–crossbridge cycle on actin that powers contractility. However, muscle contraction is not regulated by a simple, reversible on-off switching mechanism as often depicted in textbooks, and, in fact, the regulatory mechanism involves multiple steps, widely considered to be allosteric (McKillop and Geeves, 1993; Vibert et al., 1997; Poole et al., 2006; Geeves, 2012; Lehman, 2017). The first step in the regulatory process involves Ca²⁺ binding to TnC of the troponin complex, dragging TnI away from its low Ca²⁺ B-state inhibitory position on actin and tropomyosin (Yamada et al., 2020; Risi et al., 2021a; Lehman et al., 2021; Lehman and Rynkiewicz, 2023). This reorganization of the thin filament allows tropomyosin to pivot on actin to its C-state position but only partially expose myosin-binding sites; as a consequence, the released troponin–tropomyosin–imposed steric inhibition remains incomplete following Ca²⁺ activation (Vibert et al., 1997; Pirani et al., 2005; Poole et al., 2006; Lehman and Rynkiewicz, 2023). Hence, residual interference of myosin binding to actin must be relieved for full activation and contraction to ensue. Since myosin binding and action on actin is a multiple-step process itself, the corresponding regulatory checkpoints for any tropomyosin repositioning during actomyosin interaction are unlikely to involve a simple binary attachment/release mechanism. At a minimum, the process must entail an interplay between additional tropomyosin movement on actin and myosin transitioning from a weakly bound pre-powerstroke association to a strongly bound post-powerstroke M-state conformation that accompanies actin-activation of myosin’s enzymatic and motor function. In sum, tropomyosin and myosin configurational changes on actin must be coupled to each other for the chemomechanical events involved in crossbridge-linked contractility to operate. All must be done in concert with the catalytic domain of the myosin motor processing nucleotide and myosin’s lever arm motion (Coureux et al, 2003, 2004; Llinas et al, 2015; von der Ecken et al, 2016; Doran and Lehman, 2021; Pospich et al, 2021; Wang and Raunser, 2023).

Cryo-EM reconstructions of S1-decorated cardiac thin filaments have shown that Loop-4, at the tip of the myosin-S1 head, appears to force tropomyosin across actin filaments in conjunction with powerstroke-driven crossbridge movement (Behrmann et al., 2012; von der Ecken et al., 2016; Doran et al., 2020; Doran and Lehman, 2021; Risi et al., 2021b; Doran et al., 2023a). All reconstructions of tropomyosin-containing thin filaments show that Loop-4 of either nucleotide-free or ADP-saturated myosin displaces tropomyosin from its default binding interactions with basic residues Lys 326 and Lys 328 on actin (Fig. 1). Such tropomyosin movement might possibly involve Loop-4–tropomyosin charge repulsion (Doran et al., 2020, 2023a). However, current cryo-EM approaches only have captured structures of tropomyosin on myosin-free C- or B-state configured thin filaments (PDB accession nos. 7UTI, 7UTL) and on myosin-labeled “post-powerstroke” M-state filaments (PDB accession no. 8EFH), i.e., at the beginning and at the end of the myosin-induced tropomyosin translocation. By contrast, neither the initial weak pre-powerstroke binding of muscle myosin to C-state filaments has been identified by current cryo-EM protocols nor have intermediates between pre- and post-powerstroke modes been resolved. Thus, the behavior of tropomyosin movement during the C- to M-state transition is uncertain. It follows that inferences about the myosin-induced C-state to the M-state tropomyosin translocation on actin and the accompanying release of troponin–tropomyosin steric inhibition are conjectural. Given the lack of experimentally determined cryo-EM models of pre-powerstroke muscle actomyosin, we have attempted to close the current information gap using well-established computational methods to simulate and track tropomyosin during its C- to M-state translocation driven by pre- to post-powerstroke myosin transitions. Our intent has been solely to rationalize tropomyosin movement on actin and the involvement of myosin Loop-4 in the process.

Previous computational descriptions of C- to M-state tropomyosin movement relied on post-powerstroke configurations without accounting for pre-powerstroke structures (Kiani et al., 2019; Baldo et al., 2021) or without having used recently refined thin filament models (PDB accession nos. 7UTI and 8EFH). In the current study, we generated a homology model for the cardiac pre-powerstroke actomyosin structure inspired by recent work done on non-muscle myosin V structure. Our objective has been to improve on prior muscle filament modeling (Kiani et al., 2019; Baldo et al., 2021) by linking the influence of the pre- and post-powerstroke myosin transition to the translocation of tropomyosin between its C- and M-state positions on thin filaments. Our goal here is limited to this aspect of the thin filament regulation. We anticipate that understanding the involvement of myosin in this context will no doubt provide additional new directions needed to characterize mutations in the sarcomeric proteins connected to cardiomyopathies.

The approach we have taken to study myosin-induced tropomyosin movement has been guided, in part, by recent groundbreaking time-resolved cryo-EM determinations of both the pre-powerstroke and post-powerstroke structures of cytoplasmic myosin V on actin (Klebl et al., 2024, Preprint). Unlike the behavior of muscle myosin II isoforms, the distinctive tight binding and slow kinetics of myosin V on actin have facilitated the identification of initial pre-powerstroke binding (see 3-D models in Klebl et al, 2024, Preprint), which to date has not been resolved for muscle myosins. Indeed, modeling of actin-bound myosin V in EM reconstructions supports previously made inferences that the initial, sometimes named “primed,” pre-powerstroke actin–myosin interaction is common to all myosins. As previously hypothesized, the pre-powerstroke actin–myosin interaction involves the conserved helix-loop-helix (HLH) motif present on the very end of the lower 50K (L50K) domain of the myosin head inserting into a hydrophobic gap located between neighboring actin subunits (Behrmann et al., 2012; Fujii and Namba, 2017; von der Ecken et al., 2016; Doran and Lehman, 2021). Recent computational approaches to elucidate cardiac myosin interactions with troponin–tropomyosin–free actin came to the same conclusion (Ma et al., 2023).

The new cryo-EM data on myosin V describe virtually the same hydrophobic interactions occurring between the HLH motif on the myosin L50K domain and actin both in pre- and in post-powerstroke structures. In each case, the HLH motif inserts between the barbed end of one actin subunit and the pointed end on the next subunit along the thin filament. These results suggest that the interaction is both an early step and then a constant structural feature of actin–myosin binding. Further, they indicate that the HLH interaction with actin persists without detectable change at the ∼4 Å resolution achieved during the pre- to post-powerstroke structural transition of the myosin head on actin. This implies that L50K–actin interaction is maintained while the upper 50K (U50K) domain of myosin reconfigures to bind actin tightly. At the same time, the well-described U50K–L50K cleft closes to facilitate nucleotide product release after ATP hydrolysis by the myosin catalytic center (Coureux et al., 2003, 2004; Behrmann et al., 2012; Preller and Holmes, 2013; Llinas et al., 2015; Várkuti et al., 2015; von der Ecken et al., 2016; Doran and Lehman, 2021; Wang and Raunser, 2023).

It is noteworthy that much of the sequence of the HLH motif is conserved across the myosin-motor superfamily (Doran and Lehman, 2021; Wang and Raunser, 2023). Not surprisingly, x-ray crystallography carried out on isolated myosin S1 has shown that the motif also displays a strongly conserved structural identity among a plethora of myosin isoforms studied (Coureux et al., 2004). Hence, interactions of the motif with equally conserved sequences of thin filament actin are expected to describe domain associations common to actomyosin regardless of isoform origin (Doran and Lehman, 2021). Thus, our working premise is that HLH motif interactions of muscle myosin II with actin emulate those of myosin V, both in pre- and post-powerstroke modes. This perspective prompted our use of the HLH motif of myosin II as the linchpin in docking models of both pre- and post-powerstroke cardiac muscle S1 onto actin filaments in the presence and absence of tropomyosin. The modeled structures were then analyzed by steered molecular dynamics (MD) to determine likely residue-to-residue connections between tropomyosin and myosin as the myosin-head transitions between pre-powerstroke and post-powerstroke conformations on F-actin. During this process, tropomyosin moved freely from C- to M-states. By identifying key residues involved, we addressed a long-standing goal of many research groups to relate tropomyosin sequence specialization to thin filament regulatory dynamics (McLachlan and Stewart, 1976; Brown and Cohen, 2005; Hitchcock-DeGregori, 2008; Li et al., 2011).

The computational approach taken to investigate progressive myosin-induced tropomyosin movements during the corresponding pre- to post-powerstroke transition relies on reference structures acquired from cryo-EM reconstructions defining tropomyosin’s C-state and its myosin-induced M-state positions (Yamada et al., 2020; Doran et al., 2020; Risi et al., 2021a; Doran et al., 2023a, 2023b). Our studies highlight the complexity of this undertaking given the well-known non-uniformity of the tropomyosin molecule and troponin complex lying along thin filament actin subunits.

Tropomyosin itself is a stereotypical dimeric coiled-coiled molecule built from an uninterrupted series of up to 40 7-residue-long (heptad) repeats (residues a, b, c, d, e, f, g) (McLachlan and Stewart, 1976; Brown and Cohen, 2005). The C- and N-terminal ends of tropomyosin interact head-to-tail to form 9–12 residue-long 4-helix bundle overlapping domains, thereby spawning tropomyosin polymers and generating continuous cables that lie along thin filaments (Greenfield et al., 2006; Holmes and Lehman, 2008; Hitchcock-DeGregori, 2008). All tropomyosins evolved to yield additional sequence specialization characterized by higher-order periodicity consisting of six or seven 39–40-residue-long similar but non-equivalent pseudorepeating modules (McLachlan and Stewart, 1976; Brown and Cohen, 2005). Each successive module is designed to bind to a neighboring actin subunit along thin filaments to form a series of repeating units of six or seven actin subunits for each tropomyosin.

Despite the tropomyosin pseudorepeat non-uniformity, each actin pseudorepeat unit can be contacted by single myosin S1 motorheads during actin–myosin binding interactions in vitro (Vibert et al., 1997; Doran et al., 2020, 2023a, 2023b). Hence, C- to M-state movement of tropomyosin on actin can be induced either when single or multiple myosin heads bind to actin (Vibert et al., 1997; Desai et al., 2015; Smith et al., 2021). However in vivo, the localized presence and non-uniform component distribution of the bulky troponin complex could interfere with the initial weak pre-powerstroke myosin binding to specific sets of actin subunits, thereby reducing potential actin targets along thin filaments (Yamada et al., 2020; Pavadai et al., 2020; Risi et al., 2021a, 2022; Rynkiewicz et al., 2022).

As a rule, cellular actin-tropomyosin-based filaments rarely are free of additional accessory actin-binding proteins (Pollard, 2016; Manstein et al., 2020), which could hinder potential local actin–myosin interactions along thin filaments. For example, in striated muscle thin filaments, the troponin core domain, where TnT, TnI, and TnC converge, lies over actin subunits associated with pseudorepeat 5 of the seven pseudorepeat-long tropomyosin coiled-coil (Marston and Zamora, 2020; Yamada et al., 2020; Rynkiewicz et al., 2022). The bulky domain might obstruct adjacent myosin binding sites along thin filaments. Furthermore, parts of TnT interact with the relatively immobile tropomyosin overlapping domain (pseudorepeats 1 and 7), potentially limiting myosin binding (Pavadai et al., 2019; Yamada et al., 2020; Risi et al., 2022). Finally, the C-terminal domain of TnI binds to actin and tropomyosin at pseudorepeats 3 and 4 to sterically block actin–myosin binding, but only at low-Ca²⁺ concentration (Yamada et al., 2020; Lehman et al., 2021). Thus, these troponin structures may obscure binding sites on actin and therefore myosin-binding to thin filaments, although differently under low- and high-Ca²⁺ conditions (Fig. 1). Hence, the local aperiodic binding of troponin and tropomyosin over individual actin subunits belies notions of a uniform 7 actin:1 troponin:1 tropomyosin muscle “regulatory” unit repeating along thin filaments, complicating analysis.

At high Ca²⁺ concentrations in activated muscles, the third and most of the fourth tropomyosin pseudorepeats lie distal to the troponin core and to the tropomyosin overlapping domain. This arrangement leaves these myosin S1 binding targets on actin–tropomyosin unhindered in activated muscle and thus open to myosin head binding (cf. PDB accession no. 7UTI and 3-D models shown in Yamada et al., 2020; Pavadai et al., 2020; Risi et al., 2021a, 2022; Rynkiewicz et al., 2022). We therefore suggest that segments of actin–tropomyosin associated with pseudorepeats 3 and 4 in troponin-regulated muscle thin filaments are a potential myosin “target zone” that is optimized for initial myosin binding during the crossbridge cycle. In addition, we anticipated that actin subunits connected to tropomyosin pseudorepeat 2 and to pseudorepeat 6, which are adjacent to the tropomyosin-overlapping domain but distant to the troponin core domain, also may be myosin target zones on thin filaments. In the work reported here, we therefore paid particular attention to the response of tropomyosin pseudorepeats 2, 3, 4, and 6 to myosin transitions between pre-powerstroke and post-powerstroke conformations on actin. We did not investigate how a powerstroke occurs (lever arm movement, ADP, and/or Pi release); rather, we examined the interface between myosin, actin, and tropomyosin and, in turn, the influence of myosin cleft closure and the effect of low-to-high affinity myosin transitions on tropomyosin translocation over actin.

The starting model for steered MD simulations consisted of three actin monomers and their accompanying ADP and magnesium ions along with tropomyosin residues from the associated pseudorepeat (residues 15–106 for pseudorepeat 2, residues 50–148 for pseudorepeat 3, residues 92–190 for pseudorepeat 4, residues 130–225 for pseudorepeat 5, and residues 169–267 for pseudorepeat 6) from the refined cryo-EM high-Ca²⁺ structure (PDB accession no. 7UTI). The cryo-EM coordinates were aligned to place the filament axis coincident with the system z-axis. An extended model comprising 14 actin monomers and their associated tropomyosin chains was also created starting from this aligned 7UTI structure.

The structure of ADP-bound human cardiac myosin S1 in its post-powerstroke conformation (PDB accession no. 8EFH) was then superimposed on the actin near a specific pseudorepeat of tropomyosin of the three-actin model to obtain the coordinates for initial alignment of the pre-powerstroke state myosin as well as the target structure for steering. A comparable three-actin model containing a 7-glycine (7G) myosin S1 mutant, i.e., with a seven-glycine substitution on the tip of Loop-4 (Doran et al., 2023a), was constructed from the coordinates of the cryo-EM actin/tropomyosin/myosin post-powerstroke rigor structure (PDB accession no. 8ENC). For the 14-actin simulations, the target S1 was aligned to either pseudorepeat 3 or to both pseudorepeats 3 and 6 of tropomyosin.

A simulation of the human pre-powerstroke myosin S1 ADP/P_i complex was performed in explicit solvent in Amber (Case et al., 2023; Childers and Regnier, 2024) to generate potential starting conformations for the pre- to post-powerstroke steering. The human homology model included residues not included in the crystal structure (PDB accession no. 5N6A) (Planelles-Herrero et al., 2017; Auguin et al., 2023, Preprint), which were built by ab initio structure prediction using the program Modeller (Webb and Sali, 2016), and then fused to the crystal structure (PDB accession no. 5N6A). Three simulations of 500 ns each were run at 300 K. K-means clustering of the resulting structures based on the backbone atoms of residues 362–378, 401–416, and 519–556 resulted in 10 top clusters. These were then superimposed by the backbone of the HLH motif onto the aligned target ADP–myosin post-powerstroke structure. One of these was selected for further analysis based on its low number of clashes with tropomyosin or actin after alignment. However, the coordinates were adjusted manually in Chimera (Petterson et al., 2004) to alleviate clashes prior to MD simulation. The phosphate was removed from the model prior to simulation since this ligand does not appear in the target structure, and the timing and pathway of exit for phosphate from the active site during myosin cleft closure is not clear. However, no difference in the behavior of tropomyosin in response to the steering of pre- to post-powerstroke myosin was observed in control simulations when identical models of pre-powerstroke myosin contained phosphate in the myosin catalytic site. To create the myosin mutant where Loop-4 residues were replaced with glycine, this pre-powerstroke starting conformation was modified by removal of the side chain atoms of residues 366–372 and reassigning them to glycine.

The system was solvated with a 20-Å cushion and neutralized, and Na⁺ and Cl⁻ ions were added at a concentration of 0.15 M in VMD (Humphrey et al., 1996). During the next steps, the tropomyosin was restrained using the collective variables module in NAMD (Phillips et al., 2005, 2020) to maintain the helical nature of the end residues by harmonically constraining the backbone psi angles to known α-helical values as well as applying a distance constraint between the n, n+4 backbone nitrogen and oxygen atoms, keeping the hydrogen bonds intact. To maintain tropomyosin at a known distance above the actin surface, a harmonic constraint was applied to tropomyosin residues near residues Lys 326 and Lys 328 on each actin so that the radius to the actin axis of the backbone atoms was kept close to 39.5 Å (Behrmann et al., 2012, von der Ecken et al., 2016; Doran et al., 2020, 2023a, 2023b). For radial values <39.5 Å, the force constant was 0.1 kcal/mol Ų while the force constant was 0.6 kcal/mol Ų for values >39.6 Å.

Next, the water structure was minimized for 500 steps fixing all protein atoms. This was followed by constrained minimization using a 5 kcal/mol Ų force constant on backbone atoms and 1 kcal/mol Ų on side chains. The constraints were slowly removed over the next 12,000 steps of minimization. The system was then heated to 300 K at constant volume with the same constraints as above using the minimized coordinates as the reference coordinates. After heating, the constraints were slowly removed over the next 1 ns of simulation at constant pressure using the Nosé-Hoover Langevin barostat in NAMD.

The steering of the myosin S1 pre-powerstroke state to the post-powerstroke configuration was performed in three phases. In the first phase (“Dock HLH”), the HLH motif was steered to its position on actin in the post-powerstroke ADP reference structure to remove the manual adjustments made to alleviate some minor clashes between myosin and tropomyosin or actin. This steering was accomplished using a collective variable defined by the root mean square deviation (r.m.s.d.) of the α-carbons of the HLH motif (residues 525–560) (Florin et al., 2013). A harmonic bias was then applied to the r.m.s.d. collective variable centered at the initial r.m.s.d. value after heating/equilibration and using a force constant of 10 kcal/mol Ų. The center of the harmonic bias was then reduced in a linear fashion over 10 ns of simulation time to a final target value of 1.5 Å and then held at 1.5 Å for another 10 ns (“Hold HLH”). In the second phase, the resultant coordinates had a new collective variable defined for the r.m.s.d. of the remaining α-carbons in the cryo-EM structure (residues 6–197, 216–524, 561–623, 639–728, and 738–781). A second harmonic bias was applied to this collective variable with a force constant of 10 kcal/mol Ų. The center of the harmonic bias was then reduced in a linear fashion over 20 ns of simulation time to a final target value of 1.5 Å, thus steering the myosin head from pre- to post-powerstroke state configuration (“steer cleft closure”). In the last phase, the force constant of the harmonic bias was linearly increased from 10 to 200 over 20 ns of simulation time to continue to drive the motor domain fully into its cryo-EM-determined post-powerstroke ADP conformation (PDB accession no. 8EFH) (“complete cleft closure”). Control simulations including Ca²⁺ ions in a concentration up to 0.01 M showed no impact of the cation on the transitions studied.

Coordinates in PDB format are provided for the initial pre-powerstroke modeled thin filament structure (Data S1) and for the final structure reached following targeted MD simulations (Data S2). Here, myosin S1 alone was steered over a segment of our C-state thin filament model containing tropomyosin pseudorepeat 3 (taken from PDB accession no. 7UTI) to an M-state target position associated with S1-decorated actin (PDB accession no. 8EFH). Data S1 corresponds to the pre-powerstroke model shown in Fig. 2, A and C. Data S2 provides coordinates for the post-powerstroke image in Fig. 2, B and D. Video 1 shows S1 transitions over an actin subunit associated with tropomyosin pseudorepeat 3. Video 2 shows S1 transitions over an actin subunit associated with tropomyosin pseudorepeat 4. Video 3 shows S1 transitions over an actin subunit associated with tropomyosin pseudorepeat 6.

Our objective is to better define how myosin contributes to reversing tropomyosin’s steric interference during muscle activation. We used a combination of classical and steered MD to dissect the likely movement of C-state tropomyosin on actin following pre-powerstroke binding of myosin S1. As we describe below, we cataloged the likely reconfiguration of tropomyosin during the powerstroke. Our modeling is predicated on the assumption that stereospecific actin–myosin binding is initiated by localized interactions between the HLH motif at the end of the L50K domain myosin motor head and neighboring actin subunits. Our modeling also relies on HLH–actin interaction, once initiated, persisting during the pre- to post-powerstroke transition. Alternative trial models (Preller and Holmes, 2013; and our own) formed by docking the myosin U50K domain to C-state thin filaments to actin yield untenable structures with major clashes between the CM-loop, Loop-4, and tropomyosin, and to our knowledge such models lack experimental support. Our study is limited to examining the interface between S1, tropomyosin, and actin, and how myosin cleft closure during the pre- to post-powerstroke transition of S1 affects tropomyosin positioning. Exploring the characteristics and effects of allostery within the myosin head is not the point of this investigation, nor would it be rational to do so during the time scale chosen for our simulations.

We first examined an ensemble of pre-powerstroke conformers of cardiac β-myosin containing ADP and taken from MD simulations of crystal structures of pre-powerstroke cardiac myosin S1 motorheads (PDB accession no. 5N6A) (Planelles-Herrero et al., 2017). We chose to study pre-powerstroke conformers that best aligned to C-state actin-tropomyosin model without obvious local clashes with actin or tropomyosin during the preliminary S1 docking to actin. We then constructed pre-powerstroke reference structures by steered MD to best fit to models of three actin-long actin–tropomyosin segments. Here, the HLH motif of pre-powerstroke S1 conformers was targeted to actin–tropomyosin to assume locally conserved conformations homologous to those in experimentally identified, cognate post-powerstroke, HLH–actin configuration (Doran and Lehman, 2021), while the rest of the pre-powerstroke structure trailed by default. The steered MD of pre-powerstroke S1 brought about by docking the HLH domain to actin resulted in an average initial 8 Å movement of tropomyosin toward the inner actin domain due to charge–charge repulsion between myosin Loop-4 and tropomyosin and possible van der Waals repulsion between the myosin cardiomyopathy loop and tropomyosin (Videos 1, 2, and 3).

As mentioned above, all muscle tropomyosin isoforms are modular proteins consisting of seven successive pseudorepeats, each linked to a single actin protomer along thin filaments; hence, each actin–tropomyosin unit potentially is a target for myosin head binding interactions. Given that the respective 39–40-residue-long tropomyosin pseudorepeats are similar but not identical (hence the term pseudorepeat), the steering of cardiac S1 between pre- and post-powerstroke conformations on actin–tropomyosin was performed separately for each individual pseudorepeat location examined. In all cases, tropomyosin azimuthal and axial motions were unconstrained to simulate its C- to M-state movement while, as mentioned, pre-powerstroke myosin S1 was first targeted to actin via its HLH motif and then the rest of myosin steered to its target post-powerstroke ADP configuration (PBD accession no. 8EFH) to measure the response of tropomyosin. By default, the procedure was accompanied by the myosin head U50K/L50K domain cleft closure. Cleft closure during steering forces the U50K domain of myosin to translate across actin and Loop-4 to move axially in the direction of the actin-pointed end. Cleft closure was accompanied by CM loop and Loop-4 binding to actin. While tropomyosin movement was conspicuous, the HLH-motif–actin interaction, which we regard as common to both pre- and post-powerstroke conformations, did not show obvious dynamic alteration during the process (with an average r.m.s.d. of 1.5 Å between the beginning and end HLH–actin structures).

We began our analysis by examining pre- and post-powerstroke myosin S1 docked to the middle actin of a three actin–tropomyosin C-state segment extracted from PDB accession no. 7UTI and centered over cardiac tropomyosin pseudorepeat 3. As mentioned, this pseudorepeat is distal to the troponin core domain on native thin filaments and also well-separated from the tropomyosin overlap nexus and TnT–actin interactions. Therefore, tropomyosin pseudorepeat 3, unlike other repeats, best provides unbridled access of the myosin head to actin–tropomyosin. Moreover, tropomyosin residues 85–126 contained in pseudorepeat 3 are completely conserved among muscle tropomyosin isoforms (Barua et al, 2011, 2018), and hence their response to myosin-binding likely is stereotypical.

Previous cryo-EM reconstructions of the post-powerstroke decorated actin-tropomyosin (i.e., the target structure in our study) have shown (Behrmann et al., 2012; Doran et al., 2020, 2023a, 2023b; Risi et al., 2021b) that Arg 369 at the tip of Loop-4 on myosin’s U50K domain closely approaches M-state tropomyosin on actin. For example, cryo-EM models of myosin S1 and Ca²⁺-treated filaments show the side chain of Arg 369 dips under Arg 90 and 91 on tropomyosin approaching the space that separates the tropomyosin coiled coil and the actin surface, while the orthogonally oriented CM loop binds actin (Fig. 1) (Doran et al., 2020, 2023b). In marked contrast, our modeling suggests the organization of pre-powerstroke S1 on actin differs where Loop-4 and the CM loop lie more removed from their M-state binding sites on actin. Here, Loop-4 and the CM loop on the U50K domain localize ∼20 Å closer to the barbed end of actin than observed in the post-powerstroke structure, while by design the pre-powerstroke HLH domain linkage to actin is largely unchanged from that in the post-powerstroke model.

Arg 369 and Lys 367 on the tip of Loop-4 in the pre-powerstroke conformation face positively charged Arg 101, Arg 105, and Lys 112 on tropomyosin, while neighboring Loop-4 Glu 370 faces Asp 100 and Glu 104. It follows that during the pre- to post-powerstroke transition, Arg 369 and Lys 367 need to sweep by Arg 105 and Arg 101 to reach the post-powerstroke rigor conformation located near Arg 90 and 91 on tropomyosin. Moreover, Glu 370 must traverse Asp 100 and Glu 104 as well as Glu 97 and Glu 98 during this process (Fig. 2 A, Data S1, and Data S2). Note that these potential encounters will be repulsive and are expected to drive tropomyosin away from myosin and toward the M-state configuration. We tested this premise by guiding S1 between pre- and post-powerstroke states using steered MD while monitoring the tropomyosin response to the changing S1 conformation. Regardless of the timing of cleft closure during pre- to post-powerstroke transitions and their connection to the allostery within the myosin head, side chain–side chain contacts between Loop-4 and tropomyosin undoubtedly occur, and steered MD is a means to assess likely outcomes.

The effect of pre-powerstroke myosin head binding to actin–tropomyosin and subsequent Loop-4-based myosin-induced movement of tropomyosin pseudorepeat 3 during steered MD can be best visualized in Video 1. The results illustrate a series of steps involving an ordered sequence of repulsive side-chain interactions driving tropomyosin in the direction of the M-state, accompanied by a brief interval of attractive electrostatic interactions (Fig. 3). Here, side chains of charged residues Lys 367, Arg 369, and Glu 370 extending from the tip of Loop-4 can be thought of as points on the ends of a three-pronged fork variously prodding and then constraining surface charges on tropomyosin and in the process causing local tropomyosin movement to the M-state. Accordingly, once prodded, tropomyosin movement is unidirectional.

Further inspection of Video 1 reveals that docking pre-powerstroke myosin S1 to actin causes an initial 8° tropomyosin movement toward the M-state during the first 8 ns of the movie. Tropomyosin maintains this shifted position during equilibration owing to repulsion between residues Lys 367 and Arg 369 of S1 and Arg 101, Arg 105, and Lys 112 of tropomyosin, and secondarily due to the effect of residue Glu 370 of S1 on Glu 104 of tropomyosin (cf. Fig. 2 A and Fig. 3). During the next 16 ns of steered MD driving the pre-powerstroke myosin-head toward its post-powerstroke configuration, tropomyosin continues to shift in the direction of the M-state over the surface of the actin filament. Here, Arg 369 of S1 continues to encounter Arg 101 of tropomyosin and Glu 370 of S1 faces Asp 100 of tropomyosin. Further steering of myosin during U50K/L50K cleft closure then leads the Arg 369/Glu 370 pair to move beside tropomyosin residues Arg 101 and Glu 97/Glu 98, again with arginine residues adjacent to each other and glutamates similarly aligned. At the movie’s midpoint, transient arginine–glutamate (Arg 369-Glu 97/Glu 100) attractive interactions are observed between the myosin-head and tropomyosin before breaking (Fig. 3). The transient coupling of opposite charges might act in vivo as an intrinsic steering mechanism to guide Loop-4 along tropomyosin. As myosin cleft closure is completed, Loop-4 sweeps across Glu 97 and Glu 98 on tropomyosin, with tropomyosin having shifted 12–14° from its initial pre-powerstroke location. With cleft closure complete as well as CM loop binding to actin done, Loop-4 Arg 369 now has approached tropomyosin Arg 90 and Arg 91 in its post-powerstroke M-state configuration. This final post-powerstroke arginine–arginine electrostatic repulsion appears to stabilize the tropomyosin at its M-state endpoint and pins tropomyosin against a ridge on the inner edge of actin subunits formed from actin residues 222–235. Thus, further tropomyosin translocation beyond the M-state configuration on actin likely is limited topologically.

The time course of the electrostatic attraction and repulsion between myosin and tropomyosin during the steering is plotted in Fig. 3. Coulombic interaction values generally are near zero or positive, except when Arg 369 crosses Glu 97/98. The overall tropomyosin movement no doubt reflects charge repulsion, increasing the distance between the tropomyosin and the myosin S1 head. Thus, the separation of tropomyosin and S1 minimizes the Coulombic interaction energy measured.

We repeated the above protocol an additional two times with the same outcome. We also carried out a control simulation in which phosphate ion was included in the myosin catalytic center as in PDB accession no. 5N6A. The presence of phosphate did not change the model of pre-powerstroke S1 used in our simulations or the docking of S1 onto actin–tropomyosin. Here again, the effect of steering the pre-powerstroke myosin toward the post-powerstroke configuration simulation involved the same myosin Loop-4–tropomyosin side-chain interactions and the same tropomyosin shift. In addition, we tested the impact of myosin S1 on the positioning of a skeletal muscle tropomyosin isoform in place of its cardiac homolog. In this case, a three-actin/tropomyosin homology model was built from skeletal muscle thin filament components containing tropomyosin Tpm3.12, in place of the Tpm1.1 cardiac model. The movement of the tropomyosin accompanying S1 steering from pre- to post-powerstroke positions was the same for the two isoforms and involved identical electrostatic perturbations.

Each of the charged tropomyosin residues mentioned above, namely Arg 90, 91, 101, and 105, Lys 112, Glu 97, 98, and 104, as well as Asp 100, which are involved in attractive and repulsive interactions with cardiac muscle myosin, are conserved in muscle tropomyosins across the animal kingdom. This becomes evident when comparing tropomyosin sequences of often-studied muscle types, for example, those from human or chicken cardiac, skeletal, and smooth muscles (UniProt IDs P09493-1, P04268, P19352-2). Similarly, sequence annotation of tropomyosin from distantly related striated muscles in Limulus (horseshoe crab), Drosophila (fruit fly), Loligo (squid), and C. elegans (nematode) (UniProt IDs A5D6H8, P09491, F1A0N3, Q22866) is the same at these residue positions except for one to three conservative amino acid replacements (Asp↔Glu, Arg↔Lys) in these species. Thus, not only are the residues on the tip of myosin Loop-4 precisely conserved (Doran and Lehman, 2021) but so are their partners on tropomyosin. This specificity in tropomyosin and myosin sequences no doubt indicates the evolutionary pressure involved in maintaining strict myosin–tropomyosin residue-to-residue reciprocity.

As mentioned, we propose that charged residues at the tip of myosin Loop-4 have a profound effect on the azimuthal movement of tropomyosin from its C-state position to its myosin-targeted M-state location. It follows that the transition between states will be aberrant in myosin mutants lacking corresponding Loop-4 residues. To test this premise experimentally, we previously expressed and then examined the structure of mutant myosin S1-decorated filaments in which seven polar residues on the outermost edge of Loop-4 were replaced with glycine (7G-S1), thus neutralizing potential electrostatic repulsion of tropomyosin normally induced by Arg 369, Lys 367, and Glu 370 at the Loop-4 tip (Doran et al., 2023a). Corresponding cryo-EM reconstructions of the post-powerstroke configuration of the mutant 7G-S1•F-actin–tropomyosin complex showed abnormally positioned tropomyosin on actin, with bound myosin S1 and tropomyosin located 3–4 Å closer to each other than in wild type (Doran et al., 2023a). In the current work, we used our computational approach to test if the C/M-state pathway for the mutant also is perturbed.

The effect of pre-powerstroke myosin head binding to actin–tropomyosin and subsequent myosin-induced movement of tropomyosin by the 7G-S1 mutant was tested by steered MD as described above for the wild-type (WT) model. In this case, the post-powerstroke target tested contained tropomyosin and 7G-S1 fitted to the cryo-EM reconstruction of 7G-S1•F-actin-tropomyosin (PDB accession no. 8ENC), rather than to the WT configuration (PDB accession no. 8EFH). A pre-powerstroke model was then generated as done for the WT containing tropomyosin and troponin (PDB accession no. 7UTI), but now using a pre-powerstroke 7G-S1 homology model of WT-S1. Our analysis again was constructed to assess the movement of tropomyosin pseudorepeat 3 over a three-actin long filament segment during steering of pre-powerstroke 7G-S1 to its post-powerstroke M-state position; thus, the 7G-S1 track was meant to mimic the WT-S1 pathway described above.

Our steered MD showed that the 7G-S1-driven azimuthal movement of tropomyosin differed qualitatively from that of the above-described WT transition. During the simulation, tropomyosin was found to localize two to three times closer to the glycine-containing Loop-4 backbone than was measured for the WT loop (Fig. 4 A and Fig. 5). Following myosin cleft closure, the resulting narrow 3–5 Å separation between tropomyosin side chains and the glycine carbonyl oxygens and backbone nitrogens suggests that hydrogen bonding interactions, largely precluded for the WT Loop-4, now accompany the aberrant myosin-induced positions (Fig. 4, B and C; and Table 1) . Thus, transient formation and then disruption of favorable but weak interactions between the mutant Loop-4 and tropomyosin accompany the transition as the myosin head traverses along tropomyosin. The energetic cost involved may place a drag on the 7G-S1-driven pre- to post-powerstroke transition. In addition, the diminished azimuthal movement of tropomyosin induced by 7G-S1 leaves attractive electrostatic interactions between tropomyosin and actin residues 326 and 328 intact during the aberrant M-state transition, which normally break during the WT transition. These residual interactions are also likely to impede the mutant powerstroke transition and slow myosin cleft closure. In addition, the removal of the Loop-4 side chains of 7G-S1 appears to alter the rotation of M-state tropomyosin which could modify relaxation and thin filament cooperativity (Fig. 4, D and E) . Finally, the subtle conformational distinctions between WT and mutant myosin appear to affect the disposition of the cardiomyopathy loop on actin, bringing oppositely charged pre-powerstroke CM-Loop Glu 409 very close to Arg 90 on tropomyosin and potentially placing a further drag on tropomyosin/myosin movement. These conclusions are consistent with the results of corresponding electrostatic interaction energy between 7G-S1 and tropomyosin, showing generally negative values; hence attractive associations between S1 and tropomyosin are unlike those for WT which are repulsive.

The atomic model of C-state thin filaments (PDB accession no. 8EFH) built to match cryo-EM coordinates (EMD-0729) shows that Arg 369 on Loop-4 of post-powerstroke myosin dips under a single arginine residue on tropomyosin at position 133 of pseudorepeat 4 (Doran et al., 2023a, 2023b). Thus, despite neighboring the troponin core domain, the behavior of tropomyosin pseudorepeat 4 during the pre- to post-powerstroke transition may resemble that observed for pseudorepeat 3. Our steered MD protocol was used to test this possibility. Indeed, even though pseudorepeat 4 is located on actin subunits at the edge of the troponin core domain, respective myosin binding sites lie sufficiently distant from the core domain to prevent clashes that might possibly interfere with pre- or post-powerstroke myosin binding.

A three-actin filament segment was extracted from PDB accession no. 7UTI but now centered on pseudorepeat 4 in order to track the effect of myosin S1 gliding past pseudorepeat 4 during steered MD, and again designed to emulate the pre- to post-powerstroke transition for this actin-tropomyosin segment. Steered MD of myosin over pseudorepeat 4 of tropomyosin once more suggests charge–charge repulsion likely participates in driving tropomyosin from C- to M-state positions. In this instance, Loop-4 Glu 370 traversing tropomyosin Glu 142 and then Glu 139 during MD appears to be the major source of tropomyosin repulsion driving the coiled coil toward the M-state. Additional repulsion may occur between Arg 369 and Lys 367 on the S1 Loop 4 tip, and lysine residues 140 and 136 on tropomyosin may add to the effect before Arg 369 then faces tropomyosin Arg 133 and Lys 136 once the myosin cleft closure is completed (see Fig. 6, A and B; and Video 2).

The troponin core domain lying over tropomyosin pseudorepeat 5 on actin as well as TnT and the tropomyosin overlap nexus associated with tropomyosin pseudorepeats 1 and 7 may impede initial myosin-to-actin binding. In contrast, the region of filaments between these domains containing tropomyosin pseudorepeat 6 appears to be freely accessible to the myosin-head binding. Cryo-EM-based models of post-powerstroke S1 decorated filaments show Loop-4 approaching tropomyosin pseudorepeats 6 tangentially, and, unlike interactions along pseudorepeats 3 and 4, Loop-4 does not dip under the M-state tropomyosin coiled-coil on actin (Fig. 2 F). In fact, residue-residue profiles during S1 steering indicate a mixed pattern of pre- to post-powerstroke repulsive and attractive interactions between Loop-4 and multiple charged residues on tropomyosin pseudorepeat 6 (Fig. 6, C and D). On balance, repulsion appears to dominate during the steered docking and associated myosin-induced transition, given that tropomyosin moves to the M-state position (see Video 3).

Corresponding analysis of the myosin-induced translocation of tropomyosin pseudorepeats 1, 2, 5, and 7 was not performed. As mentioned, troponin may obstruct pre-powerstroke binding sites on actin linked to pseudorepeats 1, 5, and 7, limiting myosin-head binding. On the other hand, the cryo-EM structure in the region of tropomyosin pseudorepeat 2 of thin filaments is diffuse (Yamada et al., 2020), and therefore modeling tropomyosin in C- and M-state positions is imprecise.

The above simulations tested the effect of the pre- to post-powerstroke myosin transition on the positions of single tropomyosin pseudorepeats spanning three actin-long thin filament segments. We also carried out steered MD over a much larger two-sided 14-actin thin filament model to test the effect of local tropomyosin movement on the neighboring regions of actin-tropomyosin distal to the bound myosin. Here, we docked pre-powerstroke myosin heads to actin subunits linked to both pseudorepeats 3 and 6 on one side of the filament, steered the pair to their post-powerstroke targets, and assessed the effect on tropomyosin repositioning. The initial docking of the pre-powerstroke myosins produced little or no (0–3 Å) azimuthal tropomyosin translation toward the actin inner domain over S1-linked pseudorepeats 3 and 6. Hence, tropomyosin did not impose obvious steric hindrance interfering with pre-powerstroke S1 docking, while the initial tropomyosin shift observed during pre-powerstroke docking on the shorter 3-actin system was largely absent. Moreover, the docking step did not disrupt tropomyosin C-state side-chain linkages to residues 326 and 328 on the underlying actin subunits nor was tropomyosin movement noted on surrounding pseudorepeats 2, 4, 5, or 7. As expected, further steered MD of the S1 to its post-powerstroke conformation caused interactions between tropomyosin and actin residues 326 and 328 to break at the level of pseudorepeats 3 and 6. Most striking, myosin-induced tropomyosin movement over pseudorepeats 3 and 6 was propagated over the rest of the tropomyosin on successive actin subunits along the filament. In fact, now C-state the tropomyosin–actin salt bridges linkages between actin 326 and 328 on all intervening, myosin-free, tropomyosin pseudorepeats were disrupted (Fig. 7). Here, the entire tropomyosin chain moved as a unit to the M-state position. Once myosin was removed from the post-powerstroke positioned tropomyosin and then MD was launched, the tropomyosin moved back toward the C-state position. We repeated the procedure studying the effect of steering a single pre-powerstroke myosin head to its post-powerstroke conformation. Remarkably, the tropomyosin movement induced by steering S1 over pseudorepeat 3 propagated to all neighboring actin–tropomyosin contacts (Fig. 7). These results confirm previous observations (Vibert et al., 1997) that local myosin-induced shifts in tropomyosin position on actin are propagated along the length of the semi-rigid tropomyosin cable.

Numerous studies have shown that salt bridge and van der Waals interactions between adjacent amino acid side chains dominate actin–tropomyosin linkage in B- and C-state thin filaments (Lorenz et al., 1995; Brown and Cohen, 2005; Holmes and Lehman, 2008; Hitchcock-DeGregori, 2008; Li et al., 2011; Orzechowski et al., 2014; Pavadai et al., 2020; Rynkiewicz et al., 2022). Here, actin residues Lys 326 and Lys 328 attract oppositely charged acidic residues projecting from each of the tropomyosin pseudorepeat modules along filaments. In contrast, we suggest that actin–myosin binding events during muscle activity cause myosin Loop-4 residues Arg 367, Arg 369, and Glu 370 to perturb high-Ca²⁺ C-state actin-tropomyosin linkages by means of electrostatic repulsion and not by residue-to-residue attraction. Such repulsion appears to cause a displacement of actin-linked tropomyosin away from its default C-state position and results in movement to the fully activated M-state. Thus, our premise is that tropomyosin translocation to the M-state is an electrochemical process resulting from charge–charge repulsion rather than an establishment of a new set of attractive linkages between actin and tropomyosin or between tropomyosin and the myosin S1 head that pulls the coiled-coil across the actin surface. It also follows that the displacement and translocation of tropomyosin from the C-state is not a strict consequence of competitive binding between tropomyosin and myosin for common sites on actin. Instead, it is a case of intersubunit dislocation based on near-neighbor domain charge incompatibility.

We suggest that myosin head binding to actin and concomitant Loop-4—tropomyosin repulsion drives tropomyosin to a relatively energetically unfavorable M-state position on actin during the pre-powerstroke–post-powerstroke transition. In fact, interaction energy measurements performed during steered MD confirm that tropomyosin-actin linkage weakens while tropomyosin is displaced from its favorable C-state linkages to actin residues 326 and 328 during myosin-based movement (also see Baldo et al., 2021). This energetic loss, however, is coupled with favorable actomyosin-binding energetics associated with the powerstroke transition.

Our study highlights the responses of tropomyosin to its changing residue-to-residue-based chemical milieu on actin. We have suggested that tropomyosin movement is likely to facilitate unencumbered myosin head interactions on actin during muscle activity at low energy cost and across low energy barriers, a characteristic emblematic of thin filament regulation (Lehman et al., 2000). At the same time, the structural mechanics of the semirigid tropomyosin coiled coil are likely to be affected during C- to M-state azimuthal movements on actin that accompany the pre- to post-powerstroke transition of myosin (Holmes and Lehman, 2008; Li et al., 2010). Indeed, tropomyosin undergoes a local counterclockwise twisting motion as it transitions to its M-state configuration, which is likely to add to its superhelical strain (Fig. 1). We suggest that release of the torque imposed on M-state tropomyosin following ATP-induced dissociation of myosin from actin will cause tropomyosin to snap-back to the default C-state. Then, during low-Ca²⁺ muscle relaxation, tropomyosin will be attracted to the C-terminal domain of TnI producing the inhibitory B-state configuration. Notably, the twisting of the tropomyosin coiled-coil associated with the myosin-induced movement of tropomyosin now determined computationally is nearly identical to that previously detected by cryo-EM experimentally (Doran et al., 2020, 2023a, 2023b), lending credence to these conclusions.

In terms of cardiac contractile function, any strain-induced snap-back of tropomyosin from the M-state could be part of a negative cooperativity mechanism that helps make way for thin filament shut down and thereby facilitates rapid relaxation at the end of cardiac systole. Conversely, at the beginning of systole, the residue-to-residue repulsion between Loop-4 and tropomyosin could be part of a positive cooperativity mechanism accompanying force development. Thus, tropomyosin is likely to play a role in both rapid activation/force development in isovolumic systole and rapid relaxation/deactivation in isovolumic diastole.

Multiple players are involved in thin filament regulation of muscle activity, and all appear to participate in fine-tuning muscle activation and relaxation (Tardiff, 2011). In the current study, we have focused on the role played by myosin to control thin filament activity. Our insights should serve to better understand the effects of disease-bearing mutations that are likely to affect the pre- to post-powerstroke crossbridge transition and its regulation in cardiac and skeletal muscles.

All data are available in the article itself and in its supplementary materials. Source data for figures is available from the corresponding author upon request.

Henk L. Granzier served as editor.

Computational work was carried out in-house and using resources provided by the Massachusetts Green High Performance Computing Center.

These studies were supported by National Institutes of Health grants R01HL036153 to W. Lehman, P30AR074990 and R01HL128368 to M. Regnier, and T32HL007828 to M.K. Childers, as well as a European Union Horizon grant EU7877204 to M. Regnier and M.A. Geeves.

Author contributions: The general approach taken evolved from extensive discussions involving all authors. M.K. Childers carried out MD simulations on isolated cardiac S1; all other MD simulation was done by M.J. Rynkiewicz and O. Karpicheva. M.J. Rynkiewicz and W. Lehman analyzed the data and prepared figures. W. Lehman wrote the manuscript.