Alterations in Multi‐Scale Cardiac Architecture in Association With Phosphorylation of Myosin Binding Protein‐C

TG Models of Cardiomyopathy Associated With Modifications of MYBPC3

Generation and phenotypic expression in TG mouse lines expressing modified phospho‐regulatory serine sites (Ser‐273, Ser‐282, and Ser‐302) to nonphosphorylatable alanines (MYBPC3^AllP−) or phospho‐mimetic aspartic acids (MYBPC3^AllP+) has been previously described.4, 8, 9 MYBPC3^(t/t) (Provided by Drs Christine E. Seidman and Jonathan G. Seidman, Harvard Medical School, Boston, MA) demonstrated replacement of myosin and titin binding domains (exon 30) in MYBPC3 by novel residues and expressed less than 10% of normal MYBPC3 level.11 The small amount of protein that is present in these animals was considered truncated and thereby unstable. Homozygous MYBPC3^(t/t) mice displayed a dilated cardiomyopathy phenotype with contractile dysfunction, low systolic stiffness, arrhythmia, impaired Ca²⁺ cycling, and disordered myofibrils. Mice were in the FVB/N background and were between 10 and 12 weeks of age. The present animal experiments were approved by the institutional animal care and use committees at Loyola University Chicago (Maywood, IL) and followed the policies of the Guide for the Use and Care of Laboratory Animals published by the National Institutes of Health.

Echocardiography to Assess Cardiac Mechanics

Cardiac mechanical function and organ morphology were characterized using a Visual Sonics Vevo2100 echocardiography station with an MS550D 22‐ to 55‐MHz transducer (FujiFilm, Toronto, Canada).18, 19 Statistics were performed on echocardiography data using a 1‐way repeat‐measure ANOVA with Tukey's multiple comparison post‐test using Prism software (version 6.0; GraphPad Software Inc., La Jolla, CA), with significance defined as P<0.05. Echo parameters were compared on the basis of ranks (Kruskal–Wallis test) with CIs for the median of slope differences computed (Wilcoxon rank sum).

Imaging Cardiac Myoarchitecture With GQI

The theory and methods associated with the use of GQI for imaging myoarchitecture in humans (in vivo) and rodents (ex vivo) were previously described.16 Restricted nuclear magnetic resonance diffusion can be used to resolve anisotropy at the voxel scale, based largely on the influence of impediments to random translational motion of protons.20 The net outcome of these measurements is a probability distribution function (PDF) that displays diffusion in terms of a three‐dimensional (3D) orientation function multiplied by the quantitative spin density, derived from the Fourier transform of diffusion‐weighted image data.21 Employing GQI, multiple gradient pulses were applied with varying orientations and diffusion sensitivities in order to evaluate signal attenuation in 3D space (Figure S1). Imaging experiments were conducted at 7 Tesla using a Bruker Biospec equipped with a CryoProbe (Bruker Corp, Billerica, MA) utilizing closed‐cycle radiofrequency coil refrigeration. MR scanning was accomplished on thawed mouse hearts (n=3–5 samples for each condition) in fluorinated oil (Krytox; Dupont, Wilmington, DE) by a multishot echo planar imaging pulse sequence with the following imaging parameters, obtained through simulations,16 to achieve optimal angular separation: 512 gradient directions; B value of 750 s/mm²; 8 k‐space segments; and voxel size of 100×100×300 μm, with scan times of ≈8 hours/sample depending on the number of slices required to achieve complete sampling. The resulting PDF was assessed to discern the orientation and magnitude of the dominant fiber populations, with 3D directionality of the fiber populations encoded to a red‐green‐blue scale. Intervoxel fiber tracts were constructed employing streamline methods based on the alignment of the maximal diffusion vectors in the PDF.

Quantitative Analysis of Cardiac Myoarchitecture

The insertion of particular pdfs into user‐defined regions of interest (ROIs) was performed in individual slices and in multislice arrays using an anisotropy map (DSI Studio).21 Larger ROIs were drawn and positioned in the LV at multiple transmural depths (ie, endocardium, mid‐wall, and epicardium) or in the right ventricle (RV). Multivoxel tract generation was carried out using a trilinear streamlining algorithm under Eulerian integration, modified to make use of multiple fiber orientations.21, 22 Myofiber tracts were generated that intersected ROIs through intervoxel angular coherence at a range of ±17.5 degrees as derived from PDF maxima. This approach has been substantively validated through correlation with microscopic imaging methods.14, 23 Tractography coherence (ie, tract length) was then compared between MYBPC3^WT and MYBPC3 mutants. For this analysis, the common maximum tract length was identified (50 mm) and then set as a length constraint to spatially identify the longest coherent tracts in individual hearts. The longest coherent tracts were extracted from data sets of normal and mutant hearts and colored coded on the basis of 3D orientation. These common maximum tracts were then overlaid in red onto fibers measured from the respective heart data sets (overall tract length variation of 2–55 mm; displayed in silver). Macroscopic cardiac fiber orientation was analyzed for helix angle as a function of depth of the myocardium sampled employing TrackVis. For helix‐angle analysis, ROIs were placed transmurally along the depth of the heart, and helix angle at specific locations was quantified and plotted for each heart category and samples (n=3–5 hearts for each genetic type). Relative transmural depth (RTD) was calculated for each ROI in a sample using the ratio of the number of ROI positions from the innermost tissue wall divided by the total number of ROIs required to traverse the cardiac wall. RTD for each ROI in a sample was consequently calculated and displayed as a percentage. Statistical comparisons were performed employing the t test and ANOVA of the helical orientation data, and through comparison of slopes following linear regression. The computed RTD to helix angle slopes were compared on the basis of ranks (Kruskal–Wallis test) with CIs for the median of slope differences computed (Wilcoxon rank sum).

Analysis of Sarcomere Morphology Through Electron Microscopy

We analyzed sarcomeric morphology using a Hitachi 7600 transmission electron microscope, equipped with an Advanced Microscopy Techniques digital camera (Hitachi, Tokyo, Japan).24 Sections were cut from 2 to 3 mice of mixed sexes, and 50 to 100 fields were observed for each TG condition. Methods of geometric analysis were adapted from past studies of cardiac sarcomere anisotropy25 and morphology.26

Geometric Model of LV Pump Efficiency Derived From an Analysis of Myofiber Helicity

An analytic model of LV pump efficiency as a function of fiber helicity was developed based on the method of Grosberg et al., where EF is related to energy expenditure (nondimensionalized work).27 This model considers a cylinder with n=4 fibers, under the constraints of either circumferential or helical fiber geometry. A combined model, incorporating both circular and helically aligned fibers, was developed, assuming that circular and helically aligned fibers are equally weighted and that these fibers do not interact mechanically (eg, they slide past one another without resistance).

Normal Cardiac Myoarchitecture Derived Through GQI

We employed GQI to track the myoarchitecture of excised MYBPC3^WT mouse hearts (Figure 1A). Through this approach, we observed that the LV possessed helical fibers with a prototypical transmural distribution, with fiber angulation ranging from −90/+90 degrees, whereas the RV principally displayed largely circularly oriented fibers (Figure 1B). We further examined these patterns by reducing the number of tracts to display a “wire‐frame” outline of the myoarchitecture (Figure S2), confirming LV helical fiber patterns, with RV circular fibers inserting into the LV. Tracking these insertions, we identified intravoxel crossing branch points present in the wall of the septum (S) and at the interface of the LV and RV, with a close‐up view of this branch point. The demonstrated LV and RV fiber coalignment at the point of the mid‐myocardium is further described in Figure S3, implying whole‐heart functional connectivity.

Effect of the Constituitive Modification of MYBPC3 Phosphorylation on Cardiac Myoarchitecture

We compared cardiac morphology and function for TG mice using echocardiography, as shown in Table. The MYBPC3^WT, along with the phospho‐mimetic MYBPC3^AllP+/(t/t), demonstrated a normal anatomic and mechanical phenotype, whereas the MYBPC3^{AllP−/(t/t)} and the homozygous MYBPC3^(t/t) TG models displayed moderate and severe reduction of cardiac mechanics, respectively.

We performed a phenotypic comparison of MYBPC3^{AllP−/(t/t)} and MYBPC3^WT hearts in relation to an architectural template of fiber orientation and helix angle employing GQI (Figure 2). Intervoxel fiber coherence was depicted with selection of long (50 mm) tracts. In so doing, the transmural variation of fiber tract helicity in MYBPC3^WT was substituted by a loss of helicity in the MYBPC3^{AllP−/(t/t)} and conversion to a set of circularly orientated fiber tracts, similar to those observed in MYBPC3^WT mouse RV. This resulted in red and green encoded X‐Y orientated fibers (Figure 2E) and a loss in the blue‐color Z‐transmitting fiber orientations, as observed in MYBPC3^WT (Figure 2C), and modeled by reducing the helix angle over parametric equations (Figure 2A). Helicity loss in the MYBPC3^{AllP−/(t/t)} mutant was confirmed by mapping helix angles in the transverse section of the LV with rotations displayed to enable visualization of the transmural wall (Figure 2G through 2J). MYBPC3^AllP+/(t/t) hearts analyzed employing models of long‐tract coherence, fiber orientation, and LV transmural helix angle (Figure S4) were equivalent to MYBPC3^WT, with long coherence tracts distributed throughout the heart from apex to base, established coalignment between the LV and RV, and the subtle difference of thickening at the mid‐myocardium of the LV. The MYBPC3^(t/t), a homozygous mutation resulting in unstable protein targeted for degradation, displayed minimal organ‐scale architectural organization and abnormal transmural helical transitions (Figure 3). Architectural analysis depicted myofiber tracts that traversed the heart in a coiled manner, without a consistent transmural pattern. Interestingly, the effective absence of MYBPC3 protein was associated with formation of local helical transitions, as observed in the transverse section of the LV wall (Figure S5), and at various locations throughout the LV wall.

We performed a quantitative analysis of myocardial helix angle as a function of transmural depth through the use of a “pin‐wheel” display that depicted the entire array of helix angles from a single view oriented in the endocardium (Figure 4). Using this method, we observed fiber angle progression in MYBPC3^WT hearts with a smooth transition and consistent presence of transmural helices from −90/+90 degrees, along with an equivalent transition in MYBPC3^AllP+/(t/t), albeit at a slightly reduced degree of helix angle progression. In contrast, MYBPC3^{AllP−/(t/t)} hearts displayed an abnormal helix angle progression, with the mid‐myocardium exhibiting horizontal fiber orientations (approximately circular with a helix angle of 0 degrees) and abrupt transitions. Quantification of the null phenotype, MYBPC3^(t/t), demonstrated helical architectural disarray, and a loss in transmural helical transitions. Linear regression was performed to estimate transmural helix angle progression, along with statistical comparison of the data sets (Figure 5). We observed that the straight‐line slope of the null phenotype displayed a propensity toward the central mean, with reduced helicity. Employing the Student t test, MYBPC3^(t/t) helical transitions were significantly reduced compared to MYBPC3^WT and MYBPC3^WT/(t/t) (Figure 5A). With MYBPC3^AllP+/(t/t), helical transitions were significantly reduced, as mediated through the increased presence of circular‐oriented fibers at the mid‐myocardium, compared to MYBPC3^WT and MYBPC3^{AllP−/(t/t)} (Figure 5B).

Sarcomere Morphology as the Basis for Architectural Disarray

Genetic modifications of MYBPC3 were related with sarcomere ultrastructure utilizing electron microscopy (EM). Figure 6A depicts a schematic of myofilament elements of the cardiac sarcomere with respect to MYBPC3. Morphological features of the cardiac sarcomere analyzed were the aspect ratio (width to length ratio) and shear angle measured at the M‐line (Figure 6B). Using this method of analysis, we observed morphologic differences between TG variants and MYBPC3^WT (representative EM shown in Figure 6C). Using violin plots of morphological differences (Figure 6D), we observed that MYBPC3^WT and MYBPC3^WT/(t/t) had normal aspect ratios, whereas tMYBPC3^AllP+/(t/t) demonstrated reduced sarcomere length, and MYBPC3^{AllP−/(t/t)} and null MYBPC3^(t/t) demonstrated increased width, each resulting in reduced aspect ratios. MYBPC3^{AllP−/(t/t)} and null MYBPC3^(t/t) mutants demonstrated significantly increased sarcomere shear angles, whereas MYBPC3^WT/(t/t) was structurally similar to MYBPC3^WT, and MYBPC3^AllP+/(t/t) was significantly decreased.