Human hypertrophic cardiomyopathy mutation R712L suppresses the working stroke of cardiac myosin and can be rescued by omecamtiv mecarbil

Hypertrophic cardiomyopathies (HCMs) are the leading cause of acute cardiac failure in young individuals. Over 300 mutations throughout β-cardiac myosin, including in the motor domain, are associated with HCM. A β-cardiac myosin motor mutation (R712L) leads to a severe form of HCM. Actin-gliding motility of R712L-myosin is inhibited, despite near normal ATPase kinetics. By optical trapping, the working stroke of R712L-myosin was decreased 4-fold, but actin-attachment durations were normal. A prevalent hypothesis that HCM mutants are hypercontractile is thus not universal. R712 is adjacent to the binding site of the heart failure drug omecamtiv mecarbil (OM). OM suppresses the working stroke of normal β-cardiac myosin, but remarkably, OM rescues the R712L-myosin working stroke. Using a flow chamber to interrogate a single molecule during buffer exchange, we found OM rescue to be reversible. Thus, the R712L mutation uncouples lever arm rotation from ATPase activity and this inhibition is rescued by OM.

Over 300 mutations throughout the entire coding region of MYH7 have been 46 associated with HCM [3], with many of these occurring in regions predicted to affect 47 mechanochemical activity. Some mutations are clustered in a region of the myosin 48 motor that interacts with the thick filament, termed the "mesa," that stabilizes a 49 biochemically "off" state [4]. A widely cited model relating myosin function to disease 50 proposes that HCM arises from myosin mutations that enhance activity yielding 51 hypercontractile myocytes (for review, see [5]). Many MYH7 mutations examined in 52 biophysical and physiological studies have been found to cause changes in individual 53 kinetic steps that impact the fraction of intermediates in force producing states. While 54 some confer apparent hypercontractile activity, no uniform kinetic signature for HCM 55 has emerged from these studies (e.g., [6][7][8]). Thus, it is not clear that all HCM 56 mutations in the myosin motor conform to the hypercontractile hypothesis. It is 57 therefore important to perform experiments that assess the biochemical and mechanical 58 In the present work, we show that the R712L mutation in recombinant human -82 cardiac myosin does not directly confer gain-of-function, but, rather results in inhibited 83 motility due to a 4-fold decrease in working stroke amplitude, while only marginally

96
The β-cardiac myosin HCM mutant, R712L, has impaired actin filament motility 97 Human β-cardiac myosin wildtype (WT-myosin) and R712L mutant (R712L-98 myosin) HMM constructs were expressed in C2C12 myoblasts and purified (Figure 1-99 supplement 1). We first used in vitro gliding assays to measure the ability of WT-and 100 R712L-myosins to move actin filaments. Myosins were adsorbed to a nitrocellulose-101 coated glass coverslip, and the proportion of filaments that moved and their speeds 102 were determined as a function of the myosin concentration. R712L-myosin propelled 103 actin filaments more slowly than WT-myosin at all concentrations tested, with maximum 104 velocities of 1.46 ± 0.11 µm·s -1 for WT-myosin and 0.29 ± 0.02 µm·s -1 (mean ± SD) for 105 R712L-myosin at loading concentrations 100 µg·mL -1 (Figure 1B, C, Table 1, and 106 Movie 1). Although a substantial fraction of actin filaments was immobile at any given 107 time in the presence of R712L-myosin, nearly all filaments were motile at some point 108 during the assay with both constructs (Figure 1B and Movie 1). Thus, the mutant 109 motors are able to power actin gliding, but at a substantially inhibited rate. 110 As discussed above, the side chain of R712 forms a highly conserved salt bridge 111 with E497 near the fulcrum of the motor's lever arm. A conserved acidic mutation 112 (E497D) at this site also causes HCM in humans, but we found that an E497D-HMM 113 construct powers actin gliding at nearly wildtype rates (1.41 ± 0.19 µm·s -1 ). Thus, we 114 focused our efforts on characterizing R712L.  (blue, 708-806), and the essential light chain (red) are shown. The box indicates the region expanded to the right showing the E497-R712 salt bridge located at the fulcrum of the lever arm. (B) Distribution of individual filament gliding speeds from motility assays. WT-myosin (black) has a higher average motility rate compared to R712L-myosin (red). (C) Increasing loading concentrations of myosin were added and the average filament speed of fluorescently labelled actin filaments was assessed. Higher concentrations of R712L-myosin were required to achieve motility. β-cardiac myosin R712L has normal ATPase activity and attachment durations 117 We studied the biochemical kinetics of WT-and R712L-myosin to determine how 118 the mutation affects actin-activated ATPase activity (Table 1-supplement 1-5). The 119 R712L mutation has only minor effects on the individual rate constants of the ATPase 120 cycle. There is a 2-fold increase in the actin-activated Pi release rate (Table 1-121 supplement 1), which is presumably the rate-limiting step of the ATPase cycle.

122
Notably, there was a ~2-fold increase in the rate of ADP release from actin-bound 123 R712L myosin (142 s -1 ) compared to WT-myosin (73 s -1 ) ( Table 1-supplement 2). 124 This result is surprising, since normally the rate of ADP dissociation limits unloaded 125 shortening velocity of the intact muscle, and thus a > 2-fold increase in ADP release 126 would be expected to produce higher velocity in the gliding assay; yet, R712L-myosin 127 has a 5-fold slower actin gliding rate ( Figure 1B, C, and Table 1). These considerations 128 suggest that ADP release may not limit actin gliding velocity for R712L-myosin, or there 129 is a structural modification in the mutant motor that changes the linkage between the 130 ATPase and mechanical activities. 131 132 R712L-myosin HMM has a defective, single-step working stroke 133 The conserved salt bridge involving E497-R712, which is disrupted in the mutant, 134 is located at a mechanically crucial region that links myosin's relay-helix with the 135 converter/lever arm ( Figure 1A) [10][11][12]19]. We performed all-atom molecular 136 dynamics simulations to ascertain the effect of the R712L mutation on the equilibrium 137 structure of the MgADP state of -cardiac myosin (PDB 6FSA). The position of the 138 myosin lever-arm and converter fluctuate during a 100 ns simulation, but remain close 139 to the orientation found in the crystal structure (Figure 2A)     Green bars indicate binding events identified by decreases in bead covariance (gray traces; see Methods). An averaging window of 30 ms was used for covariance traces, and the position traces shown were smoothed to 5 ms to clarify the displacements. (B) Binding events were synchronized at their beginnings and ends and were averaged forward or backward in time, respectively. The average working stroke of R712L-myosin is substantially smaller than WT-myosin. WT-myosin has 2 clear steps in its working stroke, whereas substeps could not be resolved in R712L-myosin. Yellow lines are single exponential fits to the data. (C and D) Cumulative distributions of attachment durations for WT-(C) and R712L-(D) myosin at 1 µM and saturating MgATP. Inset shows the same data on a semi-log scale. For (C) and (D) yellow lines are fitted exponential distributions, where the 1 µM ATP data were well fit by single exponentials, and the saturating 1 & 4 mM ATP data were best described by the sum of two exponentials.
is consistent with a second displacement associated with ADP release ( Figure 3B; reversed ensemble averages leading up to detachment (4.6 ± 0.04 s -1 ; Figure 3B) is 192 consistent with the biochemical rate of ATP binding to nucleotide-free actomyosin at 1 193 µM ATP (Table 1-supplement 3-4).

194
In the corresponding experiment with R712L-myosin, the data traces revealed durations were adequately fitted by single exponential functions for WT-myosin (6.9 s -1 ) 208 and R712L-myosin (7.6 s -1 ) in the presence of 1 µM MgATP ( Figure 3C, D, and Table   209 3). These rates are reasonably close to the biochemical rates (4.2 and 4.5 s -1 ) expected 210 for 1 µM MgATP binding in solution (Table 1-

211
(1 or 4 mM) the distributions of event lifetimes were best described by the sum of two 212 exponentials as determined by log-likelihood ratio testing ( Figure 3C, D, and actin attachment durations was also well described by the sum of two exponentials, with 219 the predominant component (70 s -1 ; Figure 3D and WT-myosin, but comprised a larger fraction of the total ( Table 3) Crystal structures of myosin [11,12] reveal that R712L is located near the 230 binding pocket for omecamtiv mecarbil (OM), a drug in phase-3 clinical trials that 231    Given the unexpected result that OM rescues gliding motility, we measured the 247 R712L-myosin working stroke displacement and kinetics in the optical trap.  Table 2). Detachment rates were independent of the OM concentration.

252
Because the biochemical rate constants of the ATPase cycle for R712L-myosin were 253 largely unchanged in the presence of OM (Table 1- Table 2) consistent with ATP binding rates ( Table 1).

Rescue of R712L by OM is reversible 270
To test whether rescue of the R712L working stroke by OM is reversible, we designed a   alternates between a state having a normal displacement and a state with a defective 332 working stroke. However, we disfavor this possibility given that the ensemble averages 333 reveal a one-step working stroke, rather than the sum of two reduced displacements 334 ( Figure 3D and Table 3).    (Table 1). 356 Ten-fold higher OM concentrations were required to achieve rescue of R712L-357 myosin than to inhibit WT-myosin, suggesting the R712L mutation alters the OM affinity.

358
Structural studies with WT-myosin found two different OM binding conformations that 359 depend on whether the motor is in a pre-powerstroke or a post-powerstroke state, with 360 a ten-fold tighter affinity for the pre-powerstroke state [11,12]  Purified wildtype human β-cHMM and R712L HCM variants were routinely 483 analyzed by SDS PAGE (Figure 1-supplement 1). The purified proteins, which we Steady-state ATPase activity was measured by an NADH coupled assay as described 529 previously [45]. Addition of DMSO (0.25-2%) had no effect on the rates.

530
Measurements with thin filaments were carried out at pCa < 4 (100 µM Ca) and the KCl Optical trap data analysis 626 We analyzed the optical trap data from force signals of the 2 beads as previously This window was adjusted to be as low as possible while maintaining separation 636 between bound and unbound peaks such that the unbound peak mean minus its 637 standard deviation was greater than the bound peak plus its standard deviation.

638
Molecules where this separation could not be achieved were not analyzed further.

639
Actomyosin binding events were identified and refined in a two-step process: respectively. Next, as the covariance signal is a slightly delayed indicator of attachment 645 and detachment, the event start was further refined by determining when the covariance 646 trace first crossed below the value halfway between the bound and unbound peaks near 647 the initial beginning marker of each event. "Near" the originally detected event was 648 defined as within 1.5x the instrument dead time or within the duration of the detected 649 event, whichever was smaller. Event ends were refined similarly to the event 650 beginnings, that the refined event ends were marked at to the first point in time at which 651 the covariance trace crossed above 80% of the way back toward the unbound peak. For 652 display and further calculation, event starts and ends were shifted minus or plus 0.75 x 653 the dead time, respectively, to account for the effects of calculating the covariance using 654 the finite averaging window.

656
Event durations were defined as the interval between these refined start and end times.

657
Events shorter than the calculated instrument dead time were excluded from analysis.

658
Ensemble averages were performed by aligning events at their refined beginnings Attachment duration and step size parameter estimation 673 As previously described, we used MEMLET to estimate detachment rates and mean 674 step sizes [23], which allowed us to perform maximum-likelihood estimation without the 675 need for binning. Only molecules which had > 75 events were included in analysis.