Altered coronary artery function, arteriogenesis and endothelial YAP signaling in postnatal hypertrophic cardiomyopathy

Introduction: Hypertrophic cardiomyopathy (HCM) is a cardiovascular genetic disease caused largely by sarcomere protein mutations. Gaps in our understanding exist as to how maladaptive sarcomeric biophysical signals are transduced to intra- and extracellular compartments leading to HCM progression. To investigate early HCM progression, we focused on the onset of myofilament dysfunction during neonatal development and examined cardiac dynamics, coronary vascular structure and function, and mechano-transduction signaling in mice harboring a thin-filament HCM mutation. Methods: We studied postnatal days 7–28 (P7–P28) in transgenic (TG) TG-cTnT-R92Q and non-transgenic (NTG) mice using skinned fiber mechanics, echocardiography, biochemistry, histology, and immunohistochemistry. Results: At P7, skinned myofiber bundles exhibited an increased Ca2+-sensitivity (pCa50 TG: 5.97 ± 0.04, NTG: 5.84 ± 0.01) resulting from cTnT-R92Q expression on a background of slow skeletal (fetal) troponin I and α/β myosin heavy chain isoform expression. Despite the transition to adult isoform expressions between P7–P14, the increased Ca2+- sensitivity persisted through P28 with no apparent differences in gross morphology among TG and NTG hearts. At P7 significant diastolic dysfunction was accompanied by coronary flow perturbation (mean diastolic velocity, TG: 222.5 ± 18.81 mm/s, NTG: 338.7 ± 28.07 mm/s) along with localized fibrosis (TG: 4.36% ± 0.44%, NTG: 2.53% ± 0.47%). Increased phosphorylation of phospholamban (PLN) was also evident indicating abnormalities in Ca2+ homeostasis. By P14 there was a decline in arteriolar cross-sectional area along with an expansion of fibrosis (TG: 9.72% ± 0.73%, NTG: 2.72% ± 0.2%). In comparing mechano-transduction signaling in the coronary arteries, we uncovered an increase in endothelial YAP expression with a decrease in its nuclear to cytosolic ratio at P14 in TG hearts, which was reversed by P28. Conclusion: We conclude that those early mechanisms that presage hypertrophic remodeling in HCM include defective biophysical signals within the sarcomere that drive diastolic dysfunction, impacting coronary flow dynamics, defective arteriogenesis and fibrosis. Changes in mechano-transduction signaling between the different cellular compartments contribute to the pathogenesis of HCM.


Animal Model
Transgenic (TG) cTnT-R92Q mice were originally generated and characterized in C57Bl/6 genetic background. 1 In our previous 2 and current studies we used cTnT-

Skinned Fiber Bundles Tension Measurement Calculations
Force-Ca 2+ relation measurements were performed as previously described. 3 Mice were anesthetized with a ketamine/xylazine solution (200mg/20mg/kg body weight). Left ventricular papillary muscles were isolated, dissected into fiber bundles approximately 150-200 μm in width and 3-5 mm in length, and mounted between a micromanipulator and a force transducer. Fiber bundles were detergent-extracted in a high relaxing (HR) solution (10 mmol/L EGTA, 41.89 mmol/L K-Propionate, 100 mmol/L BES, 6.75 mmol/L MgCl2, 6.22 mmol/L Na2ATP, 10mM Na2CrP, 5 mmol/L NaN3, pH 7.0) and with 1% v/v Triton X-100 for 30 min at room temperature. Protease inhibitors (pepstatin A 2.5 µg/ml, leupeptin 1 µg/ml, PMSF 50 µl/ml) and Creatine Phosphokinase (CPK) (1 U/µL) were added to each pCa solution. The sarcomere length was set at 2.2 µm using laser diffraction patterns. The fibers were initially stimulated to generate force at pCa 4.5, and placed back into HR solution. Fibers were then subjected to sequential increases in Ca 2+ concentration; their developed force was recorded on a chart recorder. At the end of the experiment, the fiber bundles were subjected to pCa 4.5 solutions and the ratio of pre-and post-serial tension measurements were calculated. If the final max force was less than 90% of the original force the fiber was not included in calculations. At the end of the experiments, the width and diameter were measured along three points of fiber length and averaged. The cross-sectional area was calculated assuming the shape of an ellipse. Force-pCa measurements were calculated by modeling onto a modified Hill equation with nonlinear least-squares regression using the Gauss-Newton algorithm. Tension was calculated by dividing the force by the fiber bundle cross-sectional area.

SDS-PAGE and Immunoblotting
Excised heart samples were immediately frozen in liquid nitrogen and stored at -80°C. Heart samples (10-20mg) were homogenized 1:10 relative to original tissue weight in standard relax buffer (SRB; 75 mM KCl, 10  with Triton X-100 added to 0.5% (v/v) in SRB buffer. The myofibril preparation was spun clarified at 4°C, 15,000 X g for 1 min, and 500µL of SRB with 1% (v/v) Triton X-100 6 was added to the pellet and incubated on ice with intermittent vortexing for 15 min and this was repeated once. The pellet was washed with SRB without Triton X-100 and resuspended 1:5 relative to the original tissue weight in either industrial strength buffer (ISB: 8M urea, 2M thiourea, 50mM Tris pH 6.8, 3% v/v SDS, 75mM DTT, and 0.05% bromophenol blue 7 ) or 2X Laemmli buffer (BioRad, #161-0747). The whole homogenate preparations were solubilized 1:5 relative to original tissue weight in either ISB buffer or 2X Laemmli buffer. The proteins were solubilized by continuous shaking for 15 min at room temperature, sonicated in a water bath for 10 min, and underwent one freeze/thaw cycle. The samples were then heated at 100°C for 3 min and spun clarified at room temperature for 3 min at 21,000 X g, the supernatant was saved as the sample. Protein concentrations were determined with 660 nM Protein Assay (ThermoFisher, 22660) with ICDR reagent. Samples prepared for glutathionylation experiments had minor modifications to the myofibril sample preparation described above. The SRB containing buffers had 25mM NEM 8 added and 2X Laemmli buffer was used to solubilize without any reducing agents added. The homogenization with the Bead Ruptor 24 Elite (Omni International, 19-040E) used the same parameters except for the number of cycles which was decreased to two.
Whole homogenate protein samples (non-myofilament targets) were loaded (10-25µg/lane) on 12 or 15% (w/v) total acrylamide SDS-PAGE gels, with 0.5% (w/v) bisacrylamide as previously described 7 . The gels were cast in Bio-Rad's Criterion Cell for most of the experiments except for troponin T, myosin heavy chain, and regulatory light chain (RLC) separations described below. The criterion gels were run in 25mM tris, 192mM glycine, 0.1% (v/v) SDS running buffer at 200v for 1:15 hr:min at room temperature. To effectively separate the TnT variants we adapted a previously described method 9 and loaded myofibril heart samples at 4ug/lane onto 8% (w/v) total acrylamide, with 1.1% (w/v) bis-acrylamide in an SE600 gel box (16cm X 18cm X 1mm thick) (Hoefer). The stacking gel was 4% (w/v) total acrylamide with 3.4% (w/v) bisacrylamide, and the gel was run at 22mA until the dye front was at the bottom of the gel.
Myosin heavy chain isoform separation was carried out in 6% (w/v) total acrylamide SDS-PAGE as previously described 10 with 2ug/lane of myofibril heart sample loaded onto the gel and stained with Coomassie G-250 (Bio-Rad, 1610786) following the manufacture's recommendations. The RLC separations utilized Phos-tag SDS-PAGE as previously described with minor modifications 11 and with 2.5ug/lane of myofibril heart sample loaded onto the gel. RLC was separated into multiple bands corresponding to unphosphorylated (U), one (P1), two (P2), and three (P3) phosphorylation sites; all within the same lane allowing simple ratio analysis. The Phos-tag gel was 12% (w/v) total acrylamide, 3.3% (w/v) bis-acrylamide, 50uM Phos-tag, 100uM MnCl, and poured into 1mm thick Bio-Rad mini gel glass plates. The gel was run in a Bio-Rad mini gel apparatus at 20mA for 75 min at room temperature then the proteins were transferred to the immunoblot membrane.
After the SDS-PAGE was completed, immunoblot transfers were done with the Bio-Rad Criterion Blotter. The protein transfers were done as previously described with some modifications 12 . The protein containing gels were transferred onto 0.2µm polyvinylidene difluoride (PVDF) membrane in 10mM CAPS pH 11.0 without methanol at 20-30V for 90 min chilled with a blue ice pack. The TnT large gel was cut down to a region of interest and transferred as described above. The transfer of the Phos-tag gels required preincubation with 10mM CAPS pH 11.0, 5mM EDTA for 10 min repeated once and then washed once in 10mM CAPS pH 11.0 buffer before transferring at 30V for 90 min. After the transfer the membranes were blocked with either 5% (w/v) non-fat dry High quality coronary flow velocity signals were obtained from all animals at P7, P14 and P28, under isoflurane induced anesthesia, as described above. The coronary vasodilator properties of isoflurane are well known, so we strictly controlled the level of isoflurane input and heart rate to ensure the accuracy of the collected data. Coronary flow measurements were performed on a modified parasternal long-axis view as previously described 13 . From the low parasternal short-axis view, a search for diastolic color velocity in the anterior interventricular groove followed by clockwise rotation to achieve alignment of the color jet was performed. The sample volume was in a consistent position in all mice during the measurements.
Upon absence of pedal reflex, hearts were excised and placed into cold PBS where they were cleaned of extraneous tissue. The hearts were quickly sliced in basal, midpapillary and apical parts, and placed into biopsy cassettes, followed by fixation in 10% neutral buffered formalin (Milipore-Sigma, HT501128), then washed and stored in 70% Ethanol. Next, samples were paraffin embedded, and non-consectuive transverse sectioned were obtained (Research Histology Core, UIC).
The formalin fixed and paraffin embedded slides were deparaffinized with 100% xylene (2 x 7 min) followed by rehydration with incremental washes of decreasing aqueous ethanol (100% for 2 x 5 min, 95% for 5 min, 70% for 5 min, and 50% for 5 min) solutions, and washed in H2O for 20 min and used for staining.

Fibrosis Assessment
The deparaffinized slides were stained for collagen depositions (fibrosis) using Trichrome Stain kit (Abcam, ab150686) according to manufacture's instructions. The Trichrome stain kit is intended for visualization of collagenous connective tissue fibers in tissue sections. Then slides were mounted with Krystalon toluene-based mounting medium (Harleco, 64969-71). Next, images of whole heart sections were taken by Zeiss Axio Imager Z2 (Germany) brightfield microscope with a motorized stage for tiling. Tiles (region of scanning) were fused using native Zen stitching. Analysis of fibrosis levels in whole heart scans of apex/apical, and midventricular levels were done using ImageJ (NIH ver. 1.53k14) in heart sections harvested at 7, 14, and 28 days postnatally. The Trichrome-stained fibrosis images were analyzed by taking the original RGB image color channels and selecting the color channel corresponding to the trichrome stain.
The channel was then manually adjusted to the pixel threshold values that best fit the collagenous staining. The area was measured using ImageJ's Measure tool with the Limit to Threshold property enabled. The tissue/background was determined by minimal auto-thresholding of the same channel. The fraction of collagenous area was calculated by dividing the collagenous area over the tissue area. Localized fibrosis was assessed by 2048 x 2048 square pixel window selection of regions of interest (coronary artery regions -CA, right ventricular insertion -RVI, intraventricular septum-IVS, lateral free wall-LW). Levels of fibrosis were measured as percent collagenous area to tissue area (within the scanned window).

Immunohistochemistry
Antigen retrieval was performed using Tris-EDTA solution at 95° C for 1.5 hours.
Slides were then blocked in 5% BSA in TBST (0.1% Tween-20) for 1 hour at room temperature. To visualize vessels, slides were incubated in rat anti-CD31 antibody rabbit anti-YAP ) in TBST (Tween 0.1%) overnight at 4°C. Next, after three 5min washes with TBST (Tween 0.1%), slides were incubated with secondary antibodies (see Supplemental table 4). Slides were washed three times 5min and incubated with DAPI for nuclear counterstaining for 20min at room temperature. Slides were then washed in TBST, and mounted with a mounting medium preserving fluorescent signal (ThermoFisher Scientific, P10144). All slides were tile-scanned (1024 x 1024 pixel size) with z-stacking at 16-bit values and a pinhole diameter of 600.7 µm. Z-stacks were acquired to capture the full depth of the 5µm thick sections followed by maximumintensity projection. Channels with their properties include: YAP, αSMA, CD31 Detection and Classification. CD31, α-SMA, and DAPI labeled areas were segmented into masks using Trainable Weka Segmentation (ImageJ plug-in) 14 .
To adjust for local signal variation for whole 2D heart images, an adapted classification technique was necessary. Weka supervised machine learning models ("*.model" files) were used to generate the binarized images (masks) for CD31, a-SMA, and DAPI corresponding to endothelial cells, smooth muscle cells, and nuclei, respectively. The models were trained using small cutout sections from tissues of both genotypes at all ages ("*.arff" files). The masks generated from the model were saved in tiff format.