Myosin filament-based regulation of the dynamics of contraction in heart muscle

Significance Cardiovascular disease continues to be the leading cause of death worldwide, and is frequently associated with heart failure. Efforts to develop better therapeutics for heart failure have been held back by limited understanding of the normal control of contraction on the timescale of the heartbeat. We used synchrotron X-ray diffraction to determine the dynamic structural changes in the myosin motors that drive contraction in the heart muscle, and show that myosin filament-based control mechanisms determine the time course and strength of contraction, allowing those mechanisms to be targeted for developing new therapies for heart disease.


Supplementary Information for
Myosin filament-based regulation of the dynamics of contraction in heart muscle. Elisabetta

This PDF file includes:
Supplementary text Figures S1 to S3 Tables S1 to S2 Legends for Movies S1 to S2 SI References

Other supplementary materials for this manuscript include the following:
Movies S1 to S2 www.pnas.org/cgi/doi/10.1073/pnas.1920632117

Interpretation of the profile of the M3 reflection in terms of myosin motor conformations.
Myosin filaments from vertebrate skeletal and cardiac muscle have a bipolar structure that is symmetrical about their midpoint at the M-band of the sarcomere (Fig. 1B). Each half-filament has 49 layers of myosin motors with a periodic repeat of about 14.5 nm, corresponding to the spacing of the M3 X-ray reflection (S M3 ), with the first layer (layer 1) starting about 80nm (the half bare zone, hbz) from the filament midpoint. The C-zone of the myosin filament, containing myosin binding protein-C (MyBP-C) extends from layers 7 to 31 in heart muscle (1); layers 1 to 7 are denoted the proximal or P-zone, and layers 31 to 49 the distal or D-zone (Fig. 1B).
Each of the 49 layers contains 3 dimeric myosin molecules, or 6 motor domains. Each motor domain is considered to be in one of four possible conformations defined by previous studies (2)(3)(4)(5). The model is constrained by the one-dimensional projection of the motor mass distribution onto the filament axis, as inferred from the axial or meridional part of the X-ray diffraction diagram. The available information about the axial mass projection can be expressed in terms of two parameters. ∆z is the center of mass of the distribution with respect to the junction between the motor and its adjoining rod domain, the point of attachment to the filament backbone, defined as positive for displacements away from the filament mid-point. The other parameter is usually expressed in terms of the width of a Gaussian distribution, either of motor mass or of the orientation of its lever arm (2)(3)(4)6). Here we use a simpler approach, representing each motor conformation as a point mass at ∆z with a weight w that represents the inverse width of the mass distribution; narrower distributions have a larger w and a greater contribution to diffracted intensity. The point-diffractor representation is an excellent approximation for the purpose of calculating the profile of the M3 reflection, since the crystallographic form factor corresponding to the Fourier transform of the axial mass distribution is effectively constant over that profile (7).
In the formalism described above the four motor conformations (Fig. 6A) are defined as follows: • F: motors that are folded back against the filament surface in the 'interacting heads motif' (8,9) or OFF state (10, 11) that makes them unavailable for interaction with actin. ∆z F is taken as -7.97 nm (2, 3) and the F state is taken as the reference structure so w F = 1. • A: actin-attached or force-generating motors, with ∆z A = +3.03 nm ('model 1' of ref. 2) and w A = 2.19. ∆z A was determined from the results of length-or load-step X-ray experiments in skeletal muscle (4,5,12,13) as described previously (2,3). w A was calculated from the width of the mass distributions of A motors determined from those experiments in comparison to the reference weight for the F conformation (3). For simplicity ∆z A and w A were assumed to be independent of mechanical conditions or time. • P: Partner (P) (2,6) or 'active detached' (3) motors, also originally characterized by length-and load-step X-ray experiments in skeletal muscle, with ∆z P = +2.07 nm and w P = 2.19. The designation 'partner' reflects the evidence from skeletal muscle that these motors constitute part of a myosin dimer with the A motors during isometric contraction, and imposes the constraint that the number of P motors is equal to that of A motors. • I: isotropic or disordered motors that do not contribute to the M3 reflection, either because their range of conformations is so large that their mass is evenly spread over the 14.5-nm axial repeat of the motor layers, or because they have lost their registration on the myosin filament backbone. Thus w I = 0.
The above formalism was applied by considering the myosin filament as a 1D lattice of point diffractors with mirror symmetry about the filament midpoint, with the first lattice point in each halffilament at the half bare zone (z 1 = hbz), followed by the other 48 layers. hbz was 87.72 nm in diastole, and was assumed to increase with increased filament stress as expected for the measured myosin filament compliance in skeletal muscle (3). The diffractor at layer m has axial periodicity S M3m , and weight w and displacement ∆z corresponding to its conformation. The contribution to the amplitude of the M3 reflection from a given conformation was calculated from eq. 1: where R is the reciprocal space co-ordinate, z m the distance of the head-rod junction of the mlayer of motors from the filament midpoint, and m1 and m2 represent the first and last layers in a contiguous array of point diffractors in each half-filament, layers outside that region making zero contribution because their motors are in the I conformation. Multiple conformations (i) were introduced with weight w i defined above and number of motors n i , to calculate the intensity of the M3 reflection as a function of the reciprocal spacing R as in eq.2: n A = n P was assumed to be zero in diastole and 29 at peak force, corresponding to 10% of the total number of motors in the myosin half-filament (Fig. 3); the values at other times were calculated under the assumption that n A is proportional to the force at each time point. Similar results were obtained with n A at peak force in the range 23 to 35 suggested by previous studies (2,14). A, P and F motors were assumed to be confined to the ordered layers, and equally distributed within those layers.
The axial profile of the M3 reflection (I(R)) was calculated for each set of values of the four adjustable parameters n F , S M3m , m1 and m2 and fitted with three Gaussian distributions of the same width, using the same fitting algorithm as for the experimental data ( Fig. 5C,F,I), giving the calculated relative intensities (Fig. 6D, cyan line) and absolute spacings (Fig. 6E, cyan line) of the LA, MA and HA components of the M3 reflection. This approach provides a precise comparison with the corresponding experimental parameters (circles) without the additional step of convolution of the point-spread functions of the X-ray beam and detector, which were themselves well-fitted by Gaussian functions. The axial profiles in Fig. 5 Fig. 6, the best fit to the relative intensities and spacings of the LA, MA and HA components (constituting five independent parameters since the total intensity I M3 was not used as explained below) at each time point was determined by a global search of n F , S M3m , m1 and m2, by minimising χ 2 calculated using the experimental SDs from the multiple Gaussian fitting algorithm. The global search for each time point yielded a unique best-fit with χ 2 close to unity. An example 2D χ 2 contour plot for the S M3c , n F parameter pair at PF is shown in Fig. S3. The SDs for n F , S M3m , m1 and m2 at PF quoted in the main text were obtained by ten runs of a Monte Carlo simulation of the fitting algorithm in which the mean of each experimental parameter was replaced by a value chosen at random from a normal distribution with the experimental SD for that parameter.
The fitting procedure described in the previous paragraph failed at very early and late time points at which the value of n A was too small to give a reliable estimate of n F . However the results for all other time points were consistent with about 30% of the motors in the ordered layers being in the I conformation (calculated by subtracting the A, P and F motors from the total) within this time range. Since it is very unlikely that there is a sudden jump in n I early after stimulation that is reversed in late relaxation, we extrapolated this conclusion to diastole, i.e. we also assumed that 30% of the motors in the ordered layers (making up almost all the layers of motors in diastole) are disordered in diastole. This assumption is supported by the approximate fit between the observed and calculated changes in the total intensity of the M3 reflection (I M3 , Fig. 6B). However the dependence of I M3 on other structural parameters, including the systematic displacements of the three motor layers in each ca 44-nm helical repeat from the ideal 14.5-nm spacing in the resting or diastolic states that is linked to the presence of the so-called forbidden reflections M1, M2, M4 etc and is reduced on activation (15), means that a more extensive analysis and a more complicated model would be required for a definitive interpretation of I M3 . For that reason I M3 was excluded from the fitting procedure described above.
To define S M3m in the above calculations, the axial periodicity of the myosin motors in the C-zone (S M3c , Fig. 6F) was constrained to be 2% shorter than that in the P-and D-zones (S M3pd ), as explained in the main text. The 2% difference was estimated as follows. S M3c at PF was obtained from the observed S M3 , when only the C-zone contributes to the M3 reflection. S M6c at PF was calculated as S M3c /2 under the assumption that, in a given zone, motors index on the local backbone periodicity. S M6pd at PF was then estimated from the experimental value of S M6 assuming that layers 1 to 46 of the myosin filament contribute to the M6 reflection, i.e. the same as the layers contributing to the M3 reflection in diastole. This assumption is supported by the observation that, within the precision of the measurements, S M3 = 2* S M6 in diastole (Table S1). As expected from the presence of two distinct periodicities differing by 2%, the profile of the M3 reflection in diastole calculated from the model is more complex than that described here in terms of LA, MA and HA peaks (Fig. 5C); a series of additional peaks should be present on the lowangle side with intensity less than 10% of that of the main peak. Such peaks would not be experimentally detectable in the present protocol and preparation with the available signal:noise.  Movie S1 (separate file). X-ray diffraction movie of the sarcomere reflections during the heartbeat. Ultra-small angle X-ray patterns in the spacing range ~1 to 0.25 µm (2 nd to 8 th order of the sarcomeric repeat), collected from one electrically-paced trabecula at 20 ms intervals with the Pilatus detector at 31m. Equivalent full beam exposure time per frame, 1.5 ms.
Movie S2 (separate file). X-ray diffraction movie of the myosin-based reflections during the heartbeat. Small angle X-ray patterns with meridional periodicities in the range 70-6 nm collected at 20 ms intervals with the Pilatus detector at 3.2 m. The black horizontal lines are gaps between detector modules, with the equatorial and the myosin-based ML1/M1 and M2 reflections in the middle module. The M3 and M6 reflections are near the edges of the outer modules. Data added from three electrically-paced trabeculae. Equivalent full beam exposure time per frame, 30 ms.