We first sought to determine whether any of the five mutations impacted the expected 1:1:1 stoichiometry of myosin heavy chain:ELC:RLC of our purified proteins. To test this, we used a Coomassie SDS-PAGE gel-based assay that quantified the ratio of myosin heavy chain to ELC and RLC (Figure 2A, B). For this assay, we used a two-headed myosin construct that contains the myosin head, full lever arm, and the first two heptads of the S2 tail, followed by a GCN4 leucine zipper moiety to ensure dimerization, a GFP tag, and a C-terminal affinity clamp peptide tag (2-hep myosin [Nag et al., 2017]). When compared to the paired wild-type (WT) 2-hep control, none of the mutants had significantly changed ELC loading (p vs WT for D778V = 0.085, L781P=0.17, S782N=0.53, A797T=0.78, F834L=0.95) or RLC loading (p vs WT for D778V = 0.14, L781P=0.28, S782N=0.32, A797T=0.82, F834L=0.78), suggesting that at least for these five mutations in the context of our purified system, light chain loading is not significantly impacted (Figure 2C).

We then examined the effects of each of the five mutations on myosin’s actin-activated ATPase activity of 2-hep β-cardiac myosin. We have previously shown that 2-hep myosin produces similar ATPase rates to S1 constructs, albeit with a slightly tighter apparent affinity for actin (K_app; Adhikari et al., 2019). Of the five mutations, only D778V produced a significantly different actin-activated ATPase rate as compared to WT, increasing the turnover number k_cat by 16 ± 9% (p=0.041, average of three technical replicates of each of two biological replicates, compared to WT controls; Figure 3A; see raw values in Supplementary file 1 and raw plots in Figure 3—figure supplement 1). In contrast, L781P, S782N, A797T, and F834L all had actin-activated ATPase curves that very closely resembled their matched WT controls (Figure 3B–E; see raw values in Supplementary file 1).

We next assessed the effects of lever arm mutations on actin gliding velocity in an in vitro actin gliding motility assay (Aksel et al., 2015). For these experiments, we again used the 2-hep myosin construct, attached to the surface via interaction of its C-terminal affinity clamp to SNAP-PDZ. In the pliant region, the D778V mutation increased actin gliding velocity by 46 ± 8% (mean ± SD, p<0.0001), the L781P mutation reduced velocity by 30 ± 8% (mean ± SD, p<0.0001), and the S782N mutation had no significant effect on velocity (p=0.19) as compared to actin gliding velocities of same day WT 2-hep controls (Figure 3F). In contrast, mutations in the light chain-binding domains had little to no effect on gliding velocities, as the A797T mutation reduced velocity by only 5 ± 6% (mean ± SD, p=0.033) and the F834L mutation had no significant effect on velocity (p=0.72; Figure 3F). Actin gliding velocity is primarily a function of detachment kinetics and step size (Liu et al., 2018). Thus, it appears that as a whole, mutations in the pliant region may have more impact on myosin’s motor properties (i.e., attachment and detachment kinetics and step size), while the mutations in the light chain-binding regions may impact myosin’s activity primarily through a different mechanism. Accordingly, we next sought to further dissect the effects of pliant region mutations using harmonic force spectroscopy (HFS).

We have previously described an optical trap setup called HFS that allows for the measurement of myosin’s detachment rate from actin as a function of external load force (Liu et al., 2018; Sung et al., 2015; Sung et al., 2017) as well as its step size (Vander Roest et al., 2021). Myosin’s detachment rate is reduced in the presence of resistive load forces and increased in the presence of assistive load forces, in accordance with the force-velocity relationship of contracting heart muscle (Sung et al., 2015; Greenberg et al., 2014). Here, we used HFS to measure the load-dependent detachment rates and step size of a short S1 (sS1) myosin construct containing the motor domain of myosin plus the first half of the lever arm (amino acids 1–808), allowing for the binding of the ELC but not the RLC, followed by a GFP tag.

We measured detachment kinetics of several molecules each of WT, D778V, L781P, and S782N and fit their behavior to the Arrhenius equation with a harmonic force correction as described previously (Liu et al., 2018; Sung et al., 2015; Sung et al., 2017), where k_B is the Boltzmann constant and T is temperature:(1)kdet (F, ΔF)= k0I0ΔFδkBTexp-FδkBT

This equation is a function of k₀, the detachment rate at zero load force, δ, a measure of the force sensitivity, and F, the average external load force applied. Since the molecule experiences sinusoidal force, the equation is corrected with a zero-order Bessel function I₀ which is a function of ΔF, the amplitude of the sinusoidal force. We then averaged the individual molecules (WT n=10 molecules, D778V n=12, L781P n=22, and S782N n=11) to obtain the characteristic k₀ and δ values of WT and each lever arm mutant myosin. Interestingly, each pliant region mutation resulted in significant changes to the load force vs. detachment rate curve (Figure 4A and Figure 4—figure supplement 1), consistent with previous studies showing that mutations frequently impact force-dependent kinetics (Vander Roest et al., 2021; Liu et al., 2018). The fitted values for k₀ (WT = 147.0 ± 6.8 s^–1 (mean ± SEM); D778V = 315.6 ± 14.0 s^–1, p<0.0001; L781P = 101.7 ± 5.8 s^–1, p<0.0001; and S782N = 145.3 ± 6.7 s^–1, p=0.88; Figure 4B) roughly correlate to the changes we observed in motility velocity, consistent with the fact that motility velocity is detachment rate-limited (Liu et al., 2018). Interestingly, fitted values for δ demonstrate that all three pliant region mutations reduce force sensitivity (WT = 1.04 ± 0.05 nm (mean ± SEM); D778V = 0.68 ± 0.05 nm, p<0.0001; L781P = 0.81 ± 0.03 nm, p=0.0006; S782N = 0.83 ± 0.05 nm, p=0.0039; Figure 4C). We also measured the step size d of each mutation, and notably only L781P had any measurable impact on step size (WT = 4.3 ± 1.4 nm (mean ± SD); D778V = 4.5 ± 1.4 nm, p=0.66; L781P = 2.5 ± 1.4 nm, p=0.0032; S782N = 4.3 ± 1.2 nm, p=0.97; Figure 4D).

Given measured values for k_cat, k₀, step size, and the load-dependent detachment rate k_det, duty ratio r as a function of load force F can be calculated as follows Liu et al., 2018:(2)r(F)=kattachkattach+kdet(F)

where k_attach is the force-independent attachment rate, described by:(3)kattach=1kcat-1k0-1

The resulting force-dependent duty ratio is plotted in Figure 5A. Here, we plot only resistive forces since the ensemble behavior of heart muscle is to move against a load (blood pressure) during contraction. Duty ratio is increased when myosin spends more of its catalytic cycle in the force-producing state, bound to actin. Thus, mutant motors with faster detachment rates have lower duty ratios. In this case, the effect of reduced δ in pliant mutations had the strongest effect on duty ratio: even for S782N, which only affected force sensitivity, duty ratio was lowered at resistive forces. For D778V, this effect was compounded by the marked increase in k₀, which was not offset by the relatively smaller increase in k_cat. For L781P, the effect of reduced δ dominated the decreased k₀, except at high resistive force.

Using duty ratio, we can also calculate the average force that an individual myosin would exert in an ensemble system, which is again dependent on the load force (Liu et al., 2018):(4)FavF=F r(F)

Results from this calculation are plotted in Figure 5B. The overall trends for average force resemble that of duty ratio because average force is just load force (the x-axis) multiplied by duty ratio.

Finally, power output is a function of force × velocity. Velocity itself a function of detachment rate k_det and step size d:(5)velF=kdetF×d

such that power output is calculated by:(6)Pav(F)=velF×Fav(F)

Results from this calculation are plotted in Figure 5C. In this case, the increased velocity of D778V myosin dominates over the reduced average force, resulting in an increase in power output, particularly at high resistive forces. For L781P, the reduction in both step size and detachment rate results in a decrease in power output. S782N, by contrast, is relatively similar to WT in its power output.

To assess whether lever arm mutations impact myosin S2 tail-based autoinhibition, we introduced the mutations into a 25-hep myosin construct that is identical to 2-hep myosin, except that it contains the first 25 heptads of the S2 tail, rather than only the first 2 heptads. Thus, the myosin tail is long enough in the 25-hep construct for the myosin heads to fold back onto the tail. We have previously shown that this 25-hep myosin has a reduced k_cat as compared to the 2-hep myosin, such that a significant proportion of the 25-hep myosin is autoinhibited by the presence of the S2 tail (Nag et al., 2017; Trybus et al., 1997). This autoinhibition could arise from the IHM structure, where some of the myosin heads are sterically unable to bind actin. Mutations that disrupt autoinhibition show smaller differences between the 2-hep and 25-hep k_cat’s (Adhikari et al., 2019; Sarkar et al., 2020; Vander Roest et al., 2021).

Here, we found that all five of the lever arm mutations showed smaller differences between 2-hep and 25-hep as compared to WT (Figure 6, see full values in Supplementary file 1 and raw plots in Figure 3—figure supplement 1). The mutations all resulted in smaller differences between 2- and 25-hep myosin (WT 2-hep/25-hep ratio = 0.61 ± 0.10; D778V = 0.81 ± 0.15, p vs. WT ratio = 0.0116; L781P = 0.85 ± 0.10, p=0.007; S782N = 0.81 ± 0.12, p=0.033; A797T = 1.19 ± 0.16, p=0.0001; F834L = 0.82 ± 0.10; p=0.0281), with reductions comparable to some previous mutations we have measured (D382Y [Adhikari et al., 2019], R403Q, and R663H [Sarkar et al., 2020]).

To determine the effect of lever arm mutations on the SRX, we next measured the single-turnover ATPase activity of the myosins in the absence of actin. In this experiment, myosin is loaded with a roughly equimolar amount of fluorescent mant-ATP, then chased with excess unlabeled ‘dark’ ATP (Anderson et al., 2018; Stewart et al., 2010; Rohde et al., 2018). The fluorescence of the sample decays, and the decay rate can only be adequately fit with a double-exponential function (it does not fit a single exponential), suggesting the presence of two distinct structural states. We have previously correlated the slower SRX rate to a folded back state, whereas the faster DRX rate appears to be correlated with a structure where the heads are not bound to the S2 tail (Anderson et al., 2018; Rohde et al., 2018).

Here, we performed the single-turnover ATPase experiment with mutant and WT 25-hep myosins. Surprisingly, the pliant region mutations showed no significant reduction in the SRX state (Figure 7A–D and G; WT = 57 ± 10% (mean ± SD); D778V = 63 ± 10%, p=0.20; L781P = 61 ± 11%, p=0.42; and S782N = 68 ± 5%, p=0.018; Figure 7—figure supplement 1). In fact, S782N showed a statistically significant, though functionally small, increase in the SRX phase. This unanticipated result is in opposition to the actin-activated ATPase result comparing the 2-hep vs 25-hep constructs, which suggested that the autoinhibition was disrupted. In contrast, the light chain-binding region mutations did show a statistically significant reduction in SRX compared to WT (Figure 7E–F and G; A797T = 26 ± 6%, a 31% decrease, p<0.0001; and F834L = 16 ± 4%, a 41% decrease, p<0.0001), in keeping with their increases in 25-hep actin-activated ATPase rates. None of the mutations affected the rates of either the fast or slow phases (Supplementary file 2).

As a control, we assayed the single-turnover ATPase rates of the 2-hep constructs, and we did not see any statistically significant changes in the percent SRX turnover for any of the mutations as compared to WT 2-hep (Figure 7—figure supplement 2; WT = 21 ± 5%; D778V = 16 ± 9%, p=0.19; L781P = 23 ± 2%, p=0.23; S782N = 21 ± 4%; p=0.99; A797T = 18 ± 6%, p=0.08; F834L = 26 ± 5%, p=0.12 [mean ± SD]). This suggests that the increased SRX turnover for both WT and the pliant mutations is S2 tail-dependent and not based on any changes specific to the motor domain. In contrast, the A797T and F834L 2-heps had a similar fraction of SRX turnover as compared to the 25-hep constructs, suggesting that these mutations virtually extinguish the S2 tail-dependent autoinhibition measured by the single-turnover assay.

Recombinant human β-cardiac myosin constructs described within, including sS1, 2-hep (short-tailed), and 25-hep (long-tailed), were purified as described previously (Nag et al., 2017; Sommese et al., 2013) with some minor modifications. Heavy chain myosin (MYH7) was co-expressed with human ELC (MYL3) containing an N-terminal FLAG tag followed by a TEV protease site in the differentiated mouse myoblast C2C12 cell line (ATCC) using adenovirus generated in HEK293T cells (ATCC) using the AdEasy Vector System (Qbiogene Inc, Carlsbad, CA, USA). The sS1 construct used in this study contained a C-terminal eGFP tag, while the 2-hep and 25-hep constructs contained both eGFP and PDZ C-peptide on their C termini, respectively. C2C12 cells were infected with adenovirus constructs 48 hr after differentiation and harvested 4 days after infection in a lysis buffer containing 50 mM NaCl, 20 mM MgCl₂, 20 mM imidazole pH 7.5, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 3 mM ATP, 1 mM PMSF, 5% sucrose, and Roche protease inhibitors. Cells were then immediately flash frozen in liquid nitrogen. Note that this lysis buffer composition differs from our previously published methods—salt concentration and sucrose are lowered and MgCl₂ concentration is raised to encourage the native mouse myosin to form filaments, reducing contamination. Pellets were stored up to 6 months at –80°C, then thawed at room temperature for purification. HEK293 and C2C12 cell lines were tested for mycoplasma contamination using the Mycoalert plus kit (Lonza).

Cells were lysed with 50 strokes of a dounce homogenizer and clarified by spinning at 30,000 RPM in a Ti-60 fixed angle ultracentrifuge rotor for 30 min. Supernatant was bound to anti-FLAG resin for 1–2 hr at 4°C. The resin was then washed with a wash buffer containing 150 mM NaCl, 5 mM MgCl₂, 20 mM imidazole pH 7.5, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 3 mM ATP, 1 mM PMSF, 10% sucrose, and Roche protease inhibitors. For both the 2-hep and 25-hep constructs, native mouse RLC was depleted with a buffer containing 20 mM Tris pH 7.5, 200 mM KCl, 5 mM CDTA pH 8.0, and 0.5% Triton-X-100 for 1 hr at 4°C, and human RLC (purified from Escherichia coli as previously described [Nag et al., 2017]) was added to the resin with wash buffer and incubated for >2.5 hr at 4°C. The resin was then incubated overnight with TEV protease at 4°C to cleave the ELC-myosin complex off of the resin. The next day, the supernatant was further purified using a HiTrap Q HP column on an fast protein liquid chromatography (FPLC) with a gradient of 0–600 mM NaCl over 20 column volumes in a buffer containing 10 mM imidazole pH 7.5, 4 mM MgCl₂, 10% sucrose, 1 mM DTT, and 2 mM ATP. Pure fractions (determined by Coomassie staining on 10% SDS PAGE) were collected and concentrated to 5–50 uM using Amicon Ultra 0.5 mL centrifugal filters with a 50 or 100 kDa cutoff, which aids in removing any unbound ELC or RLC. The myosin was then used directly for ATPase assays, buffer exchanged for single turnover assays (described below), or flash frozen in liquid nitrogen for in vitro motility or optical trapping experiments.

For further analysis using in vitro motility or optical trapping, the myosin was subjected to a ‘deadheading’ procedure to remove any myosin that bound irreversibly to actin. After thawing on ice, myosin was first incubated with a >10× excess of F-actin for 5 min on ice. 2 mM ATP was then added to the mixture, and it was further incubated for 3 min on ice. The F-actin was then pelleted by ultracentrifugation at 95 K RPM in a TLA-100 rotor, and the supernatant containing active myosin was collected and used. This procedure was applied to all 2-hep myosin samples used for in vitro motility, while we found empirically that it was only necessary to deadhead the D778V sS1 myosin for optical trapping (the WT and other mutants were not deadheaded). This is because in a single molecule experiment, a ‘dead’ myosin head does not generate useful data; instead, it binds to the actin irreversibly and destroys the actin dumbbell. This rarely occurred for the WT, L781P, and S782N myosins, but was more frequent for D778V. Thus, we deadheaded only D778V to reduce dumbbell loss.

Motility measurements of 2-hep WT and mutant myosins were collected as described previously (Aksel et al., 2015). Multichannel flow chambers were constructed on microscopy slides using double-sided tape and coverslips coated in 0.1% nitrocellulose/0.1% collodion in amyl acetate. SNAP-PDZ (purified from E. coli as described previously [Aksel et al., 2015]) was first flowed into each channel at a concentration of 3 μM in assay buffer (AB: 25 mM imidazole pH 7.5, 25 mM KCl, 4 mM MgCl₂, 1 mM EGTA, and 10 mM DTT) and incubated for 2 min at room temperature. Then, each channel was blocked with assay buffer plus 1 mg/mL BSA (ABBSA) twice for 2 min each. Myosin, diluted to 50–100 nM in ABBSA, was then flowed into each channel and incubated for 2 min. Each channel was then washed with ABBSA, then loaded with final GO buffer containing 2 mM ATP, 2.5 nM TMR-phalloidin-labeled actin, and an oxygen-scavenging system (0.4% glucose, 0.11 mg/mL glucose oxidase, and 0.018 mg/mL catalase) in ABBSA.

The slide was then imaged on a Nikon Ti-E inverted microscope with a 100 × TIRF objective at a rate of 2 Hz with an exposure of 300 ms for 30 s on an Andor iXon + EMCCD camera. Each channel was imaged in three separate frames, and that data was combined for analysis. Movies were analyzed using Fast Automated Spud Trekker (Aksel et al., 2015) for filament tracking and velocity measurement using the following parameters: window size n=5, path length P=10, percent tolerance pt=20, and minimum velocity for stuck classification minv=80 nm/s. Filtered mean velocity was used as unloaded velocity. At least five replicates of each mutant were performed across 13, 3, 3, 4, 3, and 4 independent protein preparations for WT, D778V, L781P, S782N, A797T, and F834L 2-heps, respectively (biological replicates), with additional technical replicates for mutants with smaller or no difference from WT (to confirm little or no variation from WT). Temperature varied from 21 to 23°C during imaging; these temperature variations result in some variation in velocity between slides, but every mutant channel was imaged on the same slide at the same temperature as a WT control. Each mutant was then normalized to the WT control from the same slide, which negates the effect of slight variations in temperature. p-Values were determined using a paired t-test comparing mutant velocities to their paired WT control velocities.

To obtain load-dependent detachment rates of single myosin-actin interaction events, HFS measurements were performed in a dual-beam optical trap. The details of the instrumental setup and the method are similar to what has been previously described (Liu et al., 2018; Sung et al., 2015; Sung et al., 2017). The sample chambers for this experiment are made by sticking a No. 1.5 coverslip onto a 1 mm thick microscope slide with the use of double-sided tape (Figure 9A, bottom). These coverslips are spin-coated with silica beads (diameter ~1.6 micrometer, Bangs Laboratories, Fishers, IN, USA) and then with an amyl acetate solution containing 0.1% nitrocellulose and 0.1% collodion. After forming the chamber, the surface of the coverslip and the platform beads are functionalized by flowing anti-GFP antibody (product #ab1218, Abcam, Cambridge, United Kingdom) solution in AB buffer (25 mM imidazole at pH 7.5, 25 mM KCl, 4 mM MgCl₂, 1 mM EGTA, and 10 mM DTT) into the chamber for 2–3 min. The concentration of the anti-GFP antibody solution is kept low (1 nM) to ensure only a stochastic presence of the molecules on the surface. BSA (1 mg/mL) solution in AB buffer (ABBSA) is then flown into the chamber for 2–3 min to passivate/block the remaining uncovered glass surfaces. A 50 nM solution of GFP functionalized (C-terminal) short construct of either WT or mutants of human β-cardiac myosin (sS1-eGFP) in ABBSA buffer is then flown into the chamber for 2–3 min. These sS1-eGFP molecules are thus translationally immobilized on the surface by binding to the anti-GFP antibody (Fig. S7A, top). The unattached free myosin molecules are then washed away with ABBSA buffer. Finally, a solution containing filaments of 0.3 nM TMR-phalloidin-labeled biotinylated actin (Cytoskeleton, Denver, CO, USA), 0.4% glucose, 0.11 mg/mL glucose oxidase, 0.018 mg/mL catalase, 2 mM ATP, and 1-micrometer-diameter streptavidin-coated polystyrene beads (Bangs Laboratories) diluted (~5000 times) in ABBSA is flown in, prior to the sealing of the chamber with vacuum grease. This entire sample chamber preparation is done at 23°C.

The stiffnesses of the two optical traps were kept between 0.08 and 0.10 pN/nm for all experiments. Each trap was calibrated by trapping single polystyrene beads in them and then by using the power spectrum method, as described elsewhere (Liu et al., 2018; Sung et al., 2015; Sung et al., 2017). Corrections were made to rectify the effect of the anti-aliasing filter in the system. The contributions from the surface effects were also corrected during calibration. At this point in the experiment—after calibration—each trap contained one streptavidin-coated polystyrene bead. Next, an actin filament was snared between the two trapped beads using a biotin-streptavidin linkage by moving the chamber (i.e. the microscope stage) with respect to the position of the optical traps in 3D. The filament is then stretched by pulling the two beads from their ends (by steering the optical traps away from each other) to form a ‘dumbbell’ (Figure 9B). At this point in the experiment, the actin filament is stretched and held in the solution. On the surface, the myosin molecules (in the presence of ATP) are ready to initiate stroking events upon actin interaction. The actin dumbbell is then lowered toward the platform beads on the surface (keeping the stage under 200 Hz oscillation) anticipating potential interactions with single myosin molecules (Figure 9B). The stage oscillation imparts different amounts of assistive or resistive external forces during each stroking event based on the stochastic binding of myosin. In the time-trace data, the myosin-actin interactions and stroking events are identified by an expected change in the phase and the amplitude of the bead-oscillation in both the traps (Liu et al., 2018). Upon detachment after the stroking events, the oscillation in the position of the two trapped beads returns to its initial value. Several stroking events for each myosin molecule are recorded. These stroking events are then identified and binned based on the extent of external force. Then, using maximum likelihood estimation, detachment rates for every force range are obtained from the durations of the events for each molecule (Liu et al., 2018). The external load/force (F) dependent change in the detachment rates (k_d) is exponential in nature:(7)kd F,∆F=k0I0∆FδkBTe-FδkBT

where k_B is the Boltzmann constant, T is the temperature, k₀ is the detachment rate at zero external force, I₀ is a correction for the harmonic force with ΔF amplitude, and δ is the measure of force sensitivity of the myosin molecule. k₀ and δ are the parameters that vary in mutant myosins as compared to WT. These values can be obtained by fitting the spread of the k_d values at different external forces with the equation S1. During each stroking event, the positions of the dumbbells are shifted accordingly and therefore, the step sizes for individual myosin molecules can also be obtained by analyzing the same HFS data (Vander Roest et al., 2021). In the HFS experiments, however, the presence of several compliant-elements—e.g., rotation of the beads, biotin-streptavidin linkage between the beads and the actin, surface attachment of the myosin etc.—makes the step size measurement complicated. Our reported data of step-sizes do not have compliance correction and the values of the step-sizes are, therefore, smaller than what is expected for myosin’s working stroke. The data reported in this paper are gathered from the interactions of multiple different individual myosin molecules with different actin dumbbells (technical replicates) during independent experiments done over several different days with two separate protein preparations for each WT and mutant protein (biological replicates). p-Values for WT vs each mutant k₀, δ, and step size were determined using an unpaired t-test with Welch’s correction (correcting due to varying SDs by mutant), where each molecule was treated as an individual replicate. Molecules came from 2, 2, 2, and 3 unique protein preparations for WT, D778V, L781P, and S782N sS1, respectively (Fig. S4).

Actin-activated ATPase rates were measured using an NADH-coupled assay as previously described (Vander Roest et al., 2021; De La Cruz and Ostap, 2009). Actin was prepared as described previously (Spudich and Watt, 1971) and dialyzed 4× into assay buffer: 5 mM KCl, 10 mM imidazole pH 7.5, 3 mM MgCl₂, and 1 mM DTT. Actin was then mixed with a 1:50–200 molar ratio of gelsolin (prepared from E. coli as described previously [Dawson et al., 2003]), mixed thoroughly, and incubated on ice for >30 min. In a clear 96-well plate with 100 uL final volume, actin × gelsolin was mixed with assay buffer to achieve 0–80 μM final concentrations and enough myosin to achieve 25 nM final concentration. To measure the basal ATPase rate, myosin was added without actin at final concentrations of 75–125 nM. The plate was then incubated at room temperature for 10 min with constant shaking. To initiate the reaction, 20 uL of a 5× coupling solution containing 100 U/mL lactate dehydrogenase (product #L1254, Sigma-Aldrich, St. Louis, MO, USA), 500 U/mL pyruvate kinase (product #500–20, Lee Biosolutions, Maryland Heights, MO, USA), 2.5 mM phospho(enol) pyruvate (Sigma P0564), 10 mM ATP, and 2 mM NADH (Sigma N8129). The plate was again incubated for 2–5 min at room temperature with shaking before reading absorbance at 340 nm every 15–30 s for 15–25 min. A standard curve of ADP from 0 to 300 μM was created to convert the absorbance values to concentration of ADP produced.

Rates for each concentration of actin were calculated from the slope of a plot of (ADP) produced over time. For each concentration of actin, technical triplicates were performed on the same plate using the same proteins. These rates were divided by the concentration of myosin in each well, plotted against the concentration of actin, and fitted to Michaelis-Menten kinetics to obtain the values in columns 4 and 5 of Supplementary file 1, where the error reported is the SE of the fit. Each mutant 2-hep and 25-hep was prepared in tandem with a paired WT 2-hep control, and technical triplicates were measured for two independent biological replicates, where biological replicates use proteins made from different C2C12 cell batches and prepared freshly on separate days. The two biological replicates are used to validate each other; we have previously observed that additional biological replicates do not typically differ when the two biological replicates match. Within technical triplicates, individual measurements were rejected if they differed from the other two values by >50%; this resulted in rejecting less than 4% of total measurements. This is necessary because actin is inherently viscous, sometimes resulting in pipetting errors that lead to erroneous rates.

To obtain plots in Figures 2 and 5, rates from one biological replicate were normalized to the k_cat’s of the same-day WT 2-hep and mutant 2-hep, respectively. A t-test was used to compare the WT 2-hep to the mutant 2-hep rates. To obtain the data in Supplementary file 1, each individual biological replicate was fitted to standard Michaelis-Menten kinetics to obtain k_cat and K_app, where errors reported represent the SE of the fit. Errors reported in the four rightmost columns of Supplementary file 1 (mutant 2-hep/WT 2-hep k_cat ratio, mutant 25-hep/mutant 2-hep k_cat ratio, average mutant 2-hep/WT 2-hep k_cat ratio, and average mutant 25-hep/mutant 2-hep k_cat ratio) are calculated using SE propagation methods for ratios and averages, respectively, propagating the SE from the fit of k_cat.

To determine data reported in Figure 6, Fig. S2, and Supplementary file 2a single mant-ATP turnover assay was used as described previously (Anderson et al., 2018). Briefly, myosin was buffer exchanged 5× into assay buffer containing 100 mM KOAc, 10 mM Tris pH 7.5, 1 mM DTT, 4 mM MgCl₂, and 1 mM EDTA in a 50 or 100 kDa cutoff 0.5 μL Amicon filter. Myosin was then mixed with appropriate volumes of assay buffer containing 0 mM KOAc and 100 mM KOAc to achieve final salt concentrations of 5 mM (for 25-hep myosin) or 25 mM (for 2-hep myosin—2-hep does not show a dependence on salt concentration in this assay [Anderson et al., 2018]) KOAc and myosin concentrations of 200–900 nM with a 100 μL final assay volume. This was added to a 96-well black plate, where only a single well was measured at a time. 2′-(or-3′)-O-(N-methylanthraniloyl) adenosine 5′-triphosphate (mant-ATP, Thermo-Fisher Scientific, Waltham, MA, USA) was serially diluted to concentrations of 5–15 μM in assay buffer and added to the myosin at a final concentration of 1×–1.2×. Within 10–20 s, excess unlabeled ATP (4 mM final concentration) was added to the myosin + mant-ATP mixture, and the fluorescent signal (470 nm Em/405 nm Ex) was measured every ~2 s for 16 min. The ‘dead time’ between adding unlabeled ATP and the first fluorescence measurement was recorded for each well and added to the time measured by the plate reader. Fluorescence signal vs time was plotted for each replicate and fitted to a five-parameter biexponential decay. Ambiguous fits were discarded. Average fast rates, slow rates, percent fast phase decay, and percent slow phase decay (with their SEMs) are presented in Supplementary file 2. For Figure 6A–F, representative curves for each protein were normalized to the fitted Y₀=1.0 and plateau value=0.0 and fitted again to a five-parameter biexponential decay. They are plotted alongside simulated single exponential curves that have the fast rates and slow rates, respectively. A range of 4–10 technical replicates of each protein construct were performed across 8, 3, 4, 3, 3, and 3 independent protein preparations for WT, D778V, L781P, S782N, A797T, and F834L 2-heps, respectively, and 4, 4, 5, 4, 5, and 4 independent protein preparations for WT, D778V, L781P, S782N, A797T, and F834L 25-heps, respectively (biological replicates). p-Values for WT vs each mutant were determined using unpaired t-tests across all technical and biological replicates.

Loading of the ELC and RLC was determined using denaturing SDS-PAGE. Before analysis, WT and mutant 2-hep myosins were buffer exchanged 5× in a 50 or 100 kDa 0.5 μL Amicon filter to remove any light chains that were unbound to the heavy chain. Myosin samples were loaded in a dilution series across the gel at 10 pmol, 5 pmol, 3 pmol, 2 pmol, and 1 pmol per lane. After separation by gel electrophoresis, the gel was stained in Coomassie, destained, and its fluorescence was scanned at 700 nm with a LI-COR Odyssey imaging system. Each band was quantified using Fiji (Schindelin et al., 2012). A plot of raw integrated density vs pmol protein loaded was generated for each light chain and the heavy chain of each protein sample. Non-linear points were removed (due to the much higher molecular weight of the heavy chain as compared to the light chains, the linear range does not fully overlap—e.g., the 10 pmol load is generally non-linear with the 5, 3, 2, and 1 pmol loads for the heavy chain), and a linear fit was generated for each light chain and the heavy chain of each protein sample separately. The slope of each light chain fit was divided by the slope of the heavy chain fit for each protein sample, respectively, to give raw data as presented for WT 2-hep in Fig. S1B. Expected ratios based on the molecular weights of the light chains and heavy chain are noted in the legend of Fig. S1, but Coomassie staining is biased by amino acid identity in each protein sample, thus it is not unusual that measured ratios deviate somewhat from expected values based on molecular weight alone. Fig. S1 shows light chain ratios for each mutant 2-hep normalized to their same-day WT 2-hep controls. Three biologically independent protein preparations were analyzed for each mutant along with nine biologically independent protein preparations for WT 2-hep. p-Values were determined using a paired t-test, where each mutant 2-hep was paired with its same-day WT 2-hep control.