Long-term, functional culture and in vitro manipulation of adult mouse cardiomyocytes

Primary adult cardiomyocyte (aCM) culture is challenged by poor survival and loss of phenotype, rendering extended in vitro experiments unfeasible. Here, we establish murine aCM culture methods that enhance survival and maintain sarcomeric structure and Ca2+ cycling to enable physiologically-relevant contractile force measurements. We also demonstrate genetic and small-molecule manipulations that probe mechanisms underlying myocyte functional performance.

Primary adult cardiomyocyte (aCM) culture is challenged by poor survival and loss of phenotype, 2 rendering extended in vitro experiments unfeasible. Here, we establish murine aCM culture methods that 3 enhance survival and maintain sarcomeric structure and Ca 2+ cycling to enable physiologically-relevant 4 contractile force measurements. We also demonstrate genetic and small-molecule manipulations that 5 probe mechanisms underlying myocyte functional performance. 6

Main text
7 In vitro primary cell culture is a valuable tool to complement in vivo physiological investigation of 8 many tissues. In vitro studies of myocardium are limited by the challenges of adult cardiomyocyte (aCM) 9 culture. Although neonatal and pluripotent stem cell-derived cardiomyocytes have high survival rates in 10 culture, they do not fully replicate the adult phenotype in terms of morphology, mature protein isoform 11 expression, action potential component currents, Ca 2+ transient dynamics, contractile force production, 12 or metabolic substrate preference 1,2 . Therefore, these immature cells fall short of replicating the 13 physiology of mature CMs required for detailed mechanistic study 1 . While primary aCM isolation remains 14 a crucial tool for acute studies 2 , primary aCMs cultured in single-cell format detach and rapidly lose 15 physiological function 2-4 , with surviving cells typically assuming a 'fetal' phenotype after 2-3 weeks in 16 culture 2 . Finally, aCM contractility has been primarily measured using shortening velocity of cells in 17 suspension or isometric force production of intact or permeabilized cells glued to a force transducer. 18 Neither of these methods represent a physiological setting. aCMs respond in real time to their physical 19 environment 5,6 , and commonly used contractile metrics of shortening velocity of aCMs in suspension are 20 not ideal because they do not provide physiological resistance nor estimate contractile force directly. 21 Here, we demonstrate a culture protocol that maintains primary murine aCM function, enables 22 physiological contractile force measurement, and is amenable to genetic and small-molecule treatments 23 with phenotype-altering effects. 24 Typically, primary CMs are often cultured on a purified matrix and supplemented with specific 25 growth factors added to the media. Since aCM function depends on various integrin-ECM protein 26 interactions to allow for precise regulation of downstream signaling pathways 7,8 , we first explored altering 27 the matrix constituents to promote aCMs. Basement membrane preparations including Geltrex provide a 28 complex ECM protein mixture including collagens, laminins, fibronectins, and entactins among others 9 , 29 recapitulating the diversity of composition of intact myocardial ECM 10 . Secondly, the strength of aCM 30 contractions result in substantial stresses on the membranes of these cells. In aCM culture, butanedione 31 monoxime (BDM) was initially adopted as a myosin ATPase inhibitor, however BDM also impairs Ca 2+ 32 cycling through L-type channels 11,12 , modulates cardiac ryanodine receptor (RyR) flux in a Ca 2+ -dependent 33 manner 13 , and inhibits the oxidative metabolism upon which aCMs are reliant 14 ; all of which are 34 deleterious to maintaining cellular myocyte homeostasis 15 . A non-BDM protocol to inhibit contractility, 35 using myosin ATPase inhibitor blebbistatin, has shown enhanced aCM longevity and function in culture 3 . 36 Previous methods attempting prolonged aCM survival have shown complete loss of initial 37 sarcomere structure by day 8 (d8) 2 , followed by regressionto a neonatal morphology. However, CMs 38 isolated from adult mice using our modified protocol combining Geltrex and blebbistatin demonstrated 39 high adhesion, survivability and continued functionality in sustained culture ( Figure 1). CMs retained α-40 actinin-positive sarcomeres and an ordered sarcoplasmic reticulum (SR) with both longitudinal and 41 cisternal components (Fig. 1A). DHPR-positive t-tubules exhibited a mild but progressive degree of 42 disorder starting even on d1 after isolation; overall sarcolemmal organization as evidenced by STIM1 43 imaging revealed a similar trend. Furthermore, in the presence of blebbistatin the use of Geltrex resulted 44 in a 32 ± 11% decrease in Akt phosphorylation compared to a standard laminin coating (p = 0.02, Fig 1B), 45 indicating potentially altered signaling in pathways associated with differentiation, proliferation, and 46 growth. This was reflected in the survival and morphology of aCMs, which remained largely viable and 47 only declined slowly from an initial yield of 90-95% rod-shaped cells (Fig. 1C); surviving aCMs exhibited 48 morphology similar to their initial state immediately post-isolation for up to 1 week in culture and neither 49 hypertrophied nor lost phenotype, as is common in standard aCM culture methods. 50 Spontaneous Ca 2+ transient properties were analyzed using Fluo-4 AM fluorescent microscopy 51 ( Figure 1D-E) after 1 week in culture. of CellROX-active ROS production through live cell imaging increased steadily in culture to 149 ± 10% of 64 baseline at d7 (p < 0.0001; Fig. 1F). This increase coincided with a significant decrease in FCCP-uncoupled 65 maximal oxygen consumption rate (p < 0.0001, Fig. 1G), suggesting growing metabolic inefficiency over 66 increased culture time. However, increased maximal to basal OCR ratio (4.3 vs. 3.0) was noted at d1 67 compared to standard Langendorff isolation and culture methods 16 , suggesting significant spare 68 respirometric capacity early in culture with this protocol. 69 Traction force microscopy (TFM) has been carried out in many cell types, including neonatal and 70 iPSC-derived cardiomyocytes 17 . However, its use in primary isolated aCMs has been hampered by poor 71 cell attachment and functionality post-isolation. In this study, due to the significant improvement of cell 72 isolation, we applied widefield TFM techniques to aCMs to measure auxotonic contractions in a 73 mechanically-relevant environment. Spontaneous cell-associated total stresses ranging from 1.1-6.1 kPa 74 were used to calculate single-cell contractile forces of 1.4-9.6 µN cell -1 or estimated cross-sectional forces 75 of 16.5 ± 2.5 mN mm -2 (mean ± SEM). These cross-sectional force values are in agreement with other 76 auxotonic measurements of individual cells obtained with considerably higher experimental difficulty 18 . 77 Cells plated on 2 kPa gels produced significantly lower peak forces than cells on gels of 11 kPa (Fig. 1H). 78 When electrically paced, cells showed a positive force-frequency response, peaking between 2.5-4 Hz and 79 180% of baseline 1 Hz force (Fig. 1I) and approximating a quadratic curve. The resulting Bowditch curve 80 was significantly steeper and left-shifted compared to in vivo measurements of mice 19 , ostensibly due to 81 the effects of isolation and culture. Similarly, cells failed to reliably pace at field stimulation rates above 7 82 Hz while murine myocardium can pace in vivo to c.a. 14 Hz. Transmission electron microscopy ( Fig. 1J) 83 revealed preserved sarcomeric structures and mitochondrial networks. 84 Finally, we demonstrated functional manipulation of aCMs entirely in vitro ( Figure 2). In 85 phospholamban (PLN)-knockout CD1 mice we showed lentiviral transfection of WT-PLN and the human 86 pathogenic variant R9C-PLN was successful ( Fig. 2A), and clearly visualized FLAG-PLN expression (Fig. 2B). 87 Furthermore, TFM analysis revealed differences between peak contractile forces, likely due to different 88 levels of SERCA2a regulation between treatments (Fig. 2C). These experiments clearly highlight the 89 feasibility of utilizing these aCMS for biochemical and functional assays in genetic studies. In parallel 90 experiments, we showed that we can knockdown gene expression in these cells, further emphasising the 91 utility of the cellular isolation methodology. Specifically, as the SR adaptor protein Reep5 is known to be 92 essential in maintaining cardiac function 20 , AAV9-mediated Reep5 shRNA knockdown ( In conclusion, this study demonstrates a significant innovation to primary murine aCM cultures 100 with increased cell longevity and durable physiological function in vitro, notably allowing non-acute 101 experimental manipulations in aCM culture for the first time. By combining the physiological benefits of 102 blebbistatin and Geltrex, the loss of aCM phenotypes and cell death is reduced, with aCM hallmarks 103 retained over at least 7 days of culture. Importantly, we have demonstrated the ability to model pathology 104 through both genetic modulation and small-molecule administration due to the enhanced timescales and 105 experimental flexibility afforded by these methods. Finally, to our knowledge, we also describe the first 106 use of live-cell TFM for physiological, treatment-sensitive force measurement of primary aCM auxotonic 107 contraction. Together, these tools will allow for the application of experimental treatments and 108 measurements previously restricted to in vivo settings to an in vitro model, with the mechanistic control 109 afforded Isolation of primary murine adult cardiomyocytes 123 The surgical and perfusion methods closely followed those of a recent optimization of a 124 Langendorff-free preparation 2 ; we encourage readers to refer to Ackers-Johnson et al. for detailed 125 considerations in surgical methods and culture optimization. We have provided media recipes in 126 Supplementary Table 1. Briefly, male CD1 mice of 8 weeks or older were euthanized by open drop 127 exposure to isoflurane followed by cervical transection. The chest cavity was opened, the descending 128 aorta severed, and 7 mL of EDTA buffer containing 15 µmol L -1 blebbistatin (Toronto Research Chemicals, 129 Toronto ON) was injected into the right ventricle. The heart was hemostatically clamped at the ascending 130 aorta, excised from the chest cavity, and placed into a fresh dish of EDTA buffer with blebbistatin while 9 131 mL of the same buffer was slowly injected into the apex of the left ventricle. After the heart was free of 132 blood, it was moved to a dish of perfusion buffer with 15 µmol L -1 blebbistatin, and injected with 3 mL of 133 fresh 15 µmol L -1 blebbistatin through the same hole previously used in the LV. Finally, the heart was 134 moved to a dish containing 475 U mL -1 collagenase type II (Worthington Biochemical Corporation, 135 Lakewood NJ) in perfusion buffer with 15 µmol L -1 blebbistatin, of which 20 mL more was injected through 136 the existing LV hole. We found that the use of 27G, ½ length needles minimized mechanical damage to 137 the heart, allowing for maintained pressure during perfusion and optimal coronary circulation of 138 collagenase and thus digestion of the myocardium. Additionally, we observed batches of higher specific 139 activity (IU/mg) collagenase type II to be more amenable to cell survival, even at the same final activity 140 concentration. 141 At this point, the tissue was minced in 3 mL of fresh collagenase buffer with two pairs of forceps, 142 and gently triturated with a wide-bore 1 mL pipette. The collagenase activity was inhibited with addition 143 of 3 mL of perfusion buffer with blebbistatin and 10% FBS (Wisent)). The isolate was then passed through 144 a 70 µm strainer and rinsed with 3 mL additional stop buffer. The filtrate was divided between two 15 mL 145 Falcon tubes which were left standing upright for 15 min. The rod-shaped viable cardiomyocytes gravity-146 settled to form a deep red pellet, while rounded, nonviable CMs and other cell types remained in 147 suspension. The use of 2 tubes prevented oxygen or nutrient gradients forming in the cell pellets, while 148 the use of steep-walled 15 mL Falcon tubes allowed for the best recovery of the pellet over successive 149 washes. The supernatant was carefully removed, and the cells resuspended in a mixture of 75% perfusion 150 buffer and 25% culture media, containing 15 µmol L -1 blebbistatin. Cells were allowed to settle 15 min, 151 and the process repeated two more times with mixtures of 50%:50% and 25%:75% perfusion 152 buffer:culture media respectively, all containing 15 µmol L -1 blebbistatin. The final cell pellet was 153 resuspended in culture medium containing 5% FBS and 15 µmol L -1 blebbistatin, which was then plated 154 for culture in the required format. After having been plated for 3 h, dishes were gently washed and 155 replaced with culture medium containing 15 µmol L -1 blebbistatin (no FBS) to avoid serum toxicity. Cells 156 were then cultured typically up to 7 DPI for functional analyses, although we noted substantial CM survival 157 3 weeks post-isolation without full phenotypic characterization. 158 Cell culture and treatments  (TR-1003,  167 Millipore-Sigma), were treated for a further 21 h with a lentiviral vector containing PLN-WT or PLN-R9C, 168 or an AAV9 vector containing Reep5 or scrambled shRNA, prepared as previously described 21  analysis, cells were lysed in lysis buffer (8 mol L -1 urea, 10% (v/v) glycerol, 20% (w/v) SDS, 1 mol L -1 194 dithiothreitol, 1.5 mol L -1 Tris-HCl, pH 6.8, 1X cOmplete™ Mini protease inhibitor cocktail (4693159001,195 Roche)) with an 18-gauge needle. Lysates were centrifuged at 15,000 g for 15 min at 4°C. 196 SDS-soluble supernatants were added to 2X loading buffer and subjected to SDS-PAGE in a 12% 197 polyacrylamide gel with 6% stacking gel at 100 V for 20 min, then 120 V for 1 h. Semi-dry transfer to a 198 PVDF membrane occurred at 70 V for 1 h. Membranes were blocked in 5% BSA in TBS + 0.05% Tween-20 199 for 1 h at room temperature, then incubated overnight at 4°C in primary anti-Akt, anti-pAkt (Ser473), anti-200 Reep5, anti-FLAG, or anti-α-tubulin antibodies (described above), then in secondary antibodies (1:2500 201 dilution) for 1 h at room temperature. ECL detection was performed with a ChemiDoc™ Touch (Bio-Rad 202 Laboratories, Hercules CA). Uncropped images are provided in Figure S1  203 Confocal microscopy 204 Cultured cells were fixed with 4% paraformaldehyde for 10 min on ice, followed by 90% ice-cold 205 methanol for 10 min. Next, cells were incubated with permeabilization buffer (0.5% Triton X-100, 0.2% 206 Tween-20 in PBS) for 30 minutes at 4 degrees. Blocking buffer (5% FBS in 0.1% Triton-X-100 in PBS) was 207 then added and incubated for 30 minutes at room temperature. Cells were incubated with primary 208 antibodies (listed above) in blocking buffer (SERCA2a -1:500, PLN -1:1000 coverslip, which was placed in one well of a 12-well plate. Gels were washed 3x with PBS. Protein 231 conjugation to the surface of the gel was accomplished as previously described 22 . Briefly, N-232 sulfosuccinimidyl-6-(4'-azido-2'-nitrophenylamino) hexanoate (sulfo-SANPAH, CovaChem, Loves Park, IL) 233 was solubilized in DMSO (0.25% final concentration) before diluting to 500 mmol L -1 in 50 mmol L -1 HEPES 234 pH 7.4. The solution was immediately added (2 mL) to the well containing the gel, and exposed to 365 nm 235 UV light for 10 min. This process was repeated with a fresh aliquot of sulfo-SANPAH solution. Gels were 236 rinsed three times with 50 mM HEPES. Gels were then incubated overnight at 4°C in 1 mL 1:50 Geltrex in 237 PBS. Gels were rinsed 3x with PBS before being plated with cells as previously described. 238 CMs were plated on the gels as described above. For contractile analysis, wells were rinsed and 239 then replaced with blebbistatin-free culture media and incubated 5 min at 37 °C. Spontaneous 240 contractions as visualized by displacement of the FluoSpheres were then recorded on an IX71 inverted 241 widefield fluorescent microscope (Olympus Corporation, Tokyo, Japan) with a Texas Red filter cube at 242 timelapse series with exposures of 55 ms. Brightfield images of the contracted cell were also taken for 243 integration of the strain vectors (described below). For force-frequency curves, cells on 11 kPa gels were 244 field-stimulated using two carbon electrodes of c.a. 1.5 cm length and 1 cm distance, soldered to copper 245 leads that were then insulated with silicone rubber. For force-frequency curves, an S48 physiological 246 monophasic square wave stimulator (Grass Technologies, Warwick, RI) was used to pace contraction from 247 1-6 Hz at 50 V, 5 ms duration. 248 Frames of peak contraction and relaxation were analyzed using a particle image velocimetry (PIV) 249 plugin for ImageJ (v1.51j8) 23 . Interrogation windows of 64 x 64 pixels in 128 x 128 pixel search windows 250 with a 0.60 correlation threshold were used to generate a displacement field, which was then used as the 251 basis for Fourier transform traction cytometry (FTTC) as calculated by a separate ImageJ plugin 23 . A 252 Poisson's ratio of 0.48, Young's modulus of 11.0 kPa or 2.0 kPa, and a unitless regularization parameter 253 (λ) of 4.7 x 10 -10 ). The resulting stress matrix was integrated within the borders of the cell as captured by 254 brightfield microscopy using a custom Matlab 2018a script. This sum was multiplied by the area of the cell 255 to produce a total force scalar, and divided by 2 assuming both uniaxial force production by the cell, and 256 null net force production given a complete reversion to the pre-contraction state. Final force values were 257 expressed in terms of whole-cell peak force. Representative cells across N = 5 animals were measured for 258 the 2 kPa vs. 11 kPa comparison and Bowditch curve. Duplicate separately-treated wells (one cell per well) 259 for N = 3 animals were assessed for each treatment of the PLN-knockout experiment. TFM workflow is 260 visualized in Figure 4. 261

262
CMs were incubated 30 min in culture medium with blebbistatin, containing 5 µmol L -1 CellROX 263 Green (Thermo-Fisher) and counterstained 5 min with 1:1000 Hoechst 33342, both according to 264 manufacturer directions. Cells were washed in fresh media and immediately imaged with a 40X objective 265 on an inverted IX71 widefield microscope and using MicroManager acquisition software with exposures 266 of 200 ms with a FITC filter and 50 ms with a DAPI filter. Fluorescent intensity per unit area was normalized 267 to an equivalent area of adjacent background using ImageJ. 268

269
Freshly isolated CMs were plated in 24-well Seahorse XFe™ (Agilent Technologies Inc, Santa Clara, 270 CA, USA) plates and cultured as described previously for timepoint analysis. Mitochondrial respiration was 271 assessed 0, 1, 3, and 7 DPI in a Seahorse XFe24 bioanalyzer. Cells were incubated in DMEM XF assay media 272 (#102353-100, Agilent) supplemented with 5 mmol L -1 glucose, 1 mmol L -1 pyruvate, and 2 mmol L -1 273 glutamine at 37°C in a CO2-free incubator for 1 h prior to assay. Injector ports were loaded to provide final 274 concentrations of 1 µmol L -1 oligomycin, 0.5 µmol L -1 FCCP, and 1 µmol L -1 rotenone and 2 µmol L -1 275 antimycin-A together, respectively according to the mitochondrial stress test protocol provided by Agilent. 276 Maximal respiration was calculated and normalized to total protein concentration as measured by a 277 Bradford assay. Optimal oligomycin and FCCP concentrations were previously determined by titration as 278 shown in Supplemental Figure 3. 279 Fluo-4 AM (Thermo-Fisher Scientific), reconstituted in DMSO and frozen at -20 °C, was incubated 281 in the dark with cells at 37 °C for 30 min at a final concentration of 5 µmol L -1 . Spontaneous Ca 2+ waves 282 were recorded using an IX71 widefield microscope and MicroManager acquisition software at 55 ms 283 exposures with a FITC filter and expressed normalized to baseline fluorescence (F/F0) using ImageJ 284 processing. 285 Transmission electron microscopy 286 Isolated adult mouse cardiomyocytes were fixed in 2.5% glutaraldehyde in 0.1 mol L -1 phosphate 287 buffer at 4°C overnight. Samples were post-fixed in 1% osmium tetroxide buffer and processed through 288 graded alcohols and embedded in Quetol-Spurr resin. Sections of 90-100 nm were cut and stained with 289 uranyl acetate and lead citrate and imaged at 20 000X magnification using a Hitachi TE microscope at the 290 Department of Pathology, St. Michael's Hospital (Toronto, Canada). 291 dSTORM super-resolution imaging 292 Direct stochastic optical reconstruction microscopy was carried out as previously described 24 . 293 Briefly, stochastic photoswitching of immunostained samples was initiated with a buffer containing 50 294 mmol L -1 2-mercaptoethylamine (M9768, Sigma-Aldrich), 40 µg mL -1 catalase (C3155, Sigma-Aldrich), 500 295 µg mL -1 glucose oxidase (G7141, Sigma-Aldrich), 50% (w/v) D-glucose (Sigma-Aldrich), in PBS pH 7.4. A 643 296 nm laser set at 20 mW was used to inactivate the AlexaFluor 647 fluorophore into an off-state prior to 297 stochastic reactivation over the acquisition period. A super-resolved composite of 10 000 images acquired 298 over a period of 300 s at 30 ms exposures was then reconstructed using the ThunderSTORM v1.3 ImageJ 299 plugin, using a linear least square localization method. Coordinates of single emitters were filtered based 300 on localization precision and photon count to discard electronic noise (0 nm < localization precision < 7 301 nm) and sample noise (localization precision >60 nm). 302

303
All experiments were replicated at least 3 times. Statistical analysis was conducted with Prism 5 304 (GraphPad Software Inc.), except for respirometric analyses by JMP 11 (SAS Institute, Caly, NC, USA). All 305 treatments were tested using the D'Agostino and Pearson omnibus normality test before comparison with 306 an unpaired t-test. Respirometric timepoint data was assessed by 1-way split-plot ANOVA, followed by 307 Tukey-Kramer HSD. Differences between timepoints and PLN cDNA knock-ins were assessed by 1-way 308 ANOVA followed by Tukey's LSD, except for Seahorse experiments which were assessed by 1-way 309 repeated measure ANOVA followed by Tukey's LSD. induced protein folding stress (5 µmol L -1 for 24 h) as seen by the disruption of calcium-handling protein 424 expression patterns. Scale bars represent 10 µm, magnified inset images are 4x4 µm. All data expressed 425 as mean ± SEM; N denotes biological replicates; significance indicated by * (p<0.05), ** (p<0.01), and *** 426 (p<0.001). 427   Preparation of a widefield-compatible TFM gel using 500 nm fluorescent microbeads. B) A strain field was 443 constructed from frames of relaxation and maximal contraction before translation to a stress field using 444 a Young's modulus of 11 kPa for the polyacrylamide gel. The resulting uniaxial stress vectors were 445 integrated within the projected area of the cell and divided by 2 to obtain a whole-cell net traction force. 446