Inhibition of focal adhesion kinase increases myofibril viscosity in cardiac myocytes

The coordinated generation of mechanical forces by cardiac myocytes is required for proper heart function. Myofibrils are the functional contractile units of force production within individual cardiac myocytes. At the molecular level, myosin motors form cross‐bridges with actin filaments and use ATP to convert chemical energy into mechanical forces. The energetic efficiency of the cross‐bridge cycle is influenced by the viscous damping of myofibril contraction. The viscoelastic response of myofibrils is an emergent property of their individual mechanical components. Previous studies have implicated titin‐actin interactions, cell‐ECM adhesion, and microtubules as regulators of the viscoelastic response of myofibrils. Here we probed the viscoelastic response of myofibrils using laser‐assisted dissection. As a proof‐of‐concept, we found actomyosin contractility was required to endow myofibrils with their viscoelastic response, with blebbistatin treatment resulting in decreased myofibril tension and viscous damping. Focal adhesion kinase (FAK) is a key regulator of cell‐ECM adhesion, microtubule stability, and myofibril assembly. We found inhibition of FAK signaling altered the viscoelastic properties of myofibrils. Specifically, inhibition of FAK resulted in increased viscous damping of myofibril retraction following laser ablation. This damping was not associated with acute changes in the electrophysiological properties of cardiac myocytes. These results implicate FAK as a regulator of mechanical properties of myofibrils.

factors controlling the mechanical properties of myofibrils continues to be elucidated.
Titin has long been established as a major contributor to the myofibrillar mechanical properties required for proper heart function. The PEVK domain of titin behaves like a molecular spring, generating passive tension within the myocardium against stretch, as experienced in the heart during peak diastole (Granzier & Irving, 1995;Granzier & Labeit, 2004). Furthermore, interactions between actin and the titin PEVK domain lead to viscous damping of myofibrils by impeding actin filament sliding (Kulke et al., 2001;. At longer length scales, viscous damping in the myocardium is thought to be provided by collagen fibrils in the extracellular matrix (ECM) (Granzier & Irving, 1995). Previous studies have demonstrated that faster shortening of myofibrils has a higher energetic cost (Gibbs & Chapman, 1985;He, Bottinelli, Pellegrino, Ferenczi, & Reggiani, 2000). Thus, viscous damping reduces this energetic cost of contraction and contributes to an overall increased efficiency of energy consumption in the heart (Caporizzo, Chen, Salomon, Margulies, & Prosser, 2018).
Cardiomyocytes interact with the ECM at specialized sites called costameres, which anchor the Z-lines closest to the cell membrane to the ECM (Samarel, 2005). We have previously established a role for cell-ECM adhesion in myofibril assembly, using human iPSC-derived cardiomyocytes (hiCMs) as a model system (Taneja, Neininger, & Burnette, 2020). Cell-ECM sites transmit mechanical forces from immature myofibrils (muscle stress fibers) to the substrate, promoting their maturation.
Focal adhesion kinase (FAK) is a key regulatory protein present at cell-ECM sites which regulates adhesion formation and turnover (Parsons, Martin, Slack, Taylor, & Weed, 2000). A role for FAK in contributing to heart mechanics has also been well established in vivo.
Cardiac specific knockout of FAK results in an ability to respond to increased load in the heart (DiMichele et al., 2006;Peng et al., 2006).
On a mechanistic level, FAK performs both scaffolding and signaling functions (Sulzmaier, Jean, & Schlaepfer, 2014). FAK phosphorylates a variety of targets, including itself at Tyr397, which is required for adhesion disassembly (Sulzmaier et al., 2014). Inhibition of this autophosphorylation using the inhibitor PF-228 results in focal adhesion stabilization (Slack-Davis et al., 2007;Taneja et al., 2016). Treatment of hiCMs with PF-228 results in stabilization of focal adhesions, longer Z-line lengths and increased incorporation of titin (Taneja, Neininger, & Burnette, 2020). However, the effect of focal adhesion stabilization by FAK on the mechanical properties of myofibrils remains unknown.
In nonmuscle cells, FAK regulates actin cytoskeleton stability and cell stiffness (Fabry, Klemm, Kienle, Schäffer, & Goldmann, 2011;Mierke et al., 2017). Stress fibers that are attached on either end to the ECM through focal adhesions have higher tension compared to those that are not directly coupled to the ECM (Lee, Kassianidou, & Kumar, 2018). Given the role of FAK in regulating cellular mechanics, and its role in load adaptation in the heart, here we specifically investigate its role in modulating the mechanical properties of cardiac myofibrils at the cellular level. Using laser-assisted dissection of myofibrils, we show that inhibition of FAK results in increased myofibril viscosity.
We further show that these changes in mechanics are not mediated by changes in electrophysiology.

| RESULTS AND DISCUSSION
Laser ablation is a widely used biophysical tool to assay the mechanical properties of cytoskeletal systems, such as stress fibers, epithelial cell-cell junctions and myofibrils (Fernandez-Gonzalez, Simoes, Röper, Eaton, & Zallen, 2009;Lee et al., 2018;Roman et al., 2017;Taneja, Neininger, & Burnette, 2020). We have previously used this technique to interrogate cortex mechanics during cell division, as well as testing mechanical coupling between cell-ECM adhesions and myofibrils Taneja, Neininger, & Burnette, 2020). Myofibrils are considered complex mechanical systems. Conceptually, they can be thought of as a series of springs and dashpots (sarcomeres), which lend both elasticity and viscosity to the mechanical response, respectively. We used an existing mechanical framework to estimate these properties from the dynamics of myofibril retraction following laserassisted dissection (Lee et al., 2018).
We have previously described the dynamics of myofibril assembly using hiCMs as a model system (Fenix et al., 2018). hiCMs are transcriptionally similar to neonatal or embryonic cardiac myocytes (DeLaughter et al., 2016;Kuppusamy et al., 2015). As such, they lose their myofibrils upon trypsinization, and reform them within 24 hr after plating (Fenix et al., 2018). Thus, we plated hiCMs on glass coverslips coated with fibronectin and allowed them to spread overnight prior to performing laser-assisted dissection of preformed myofibrils Using a custom MATLAB script based on edge detection, we segmented the retracting ends of the ablated myofibril and fit the trajectories to a standard Kelvin-Voigt viscoelastic solid model widely used for retracting stress fibers and cell-cell junctions in nonmuscle cells (Fernandez-Gonzalez et al., 2009;Kumar et al., 2006;Lee et al., 2018). Using this framework, we identified two material parameters that describe the viscoelastic behavior: (a) the asymptotic distance D a is the maximal distance between the two retracting ends as they reach a steady-state separation and is proportional to stored elastic energy or tension within the myofibril and (b) the relaxation time constant τ describes the shape of the time-dependent retraction profile and is proportional to myofibril viscosity; for the Kelvin-Voigt model, the relaxation time is mathematically equivalent to 0.632 D a ( Figure 1c).
We also obtained estimates for two additional parameters related to the retraction dynamics-linear velocity (v L ) and instantaneous velocity (v I ) (Figure 1c). Instantaneous velocity was calculated as the derivative of the fitted retraction profile at the time of ablation and is equivalent to the ratio of asymptotic distance to relaxation time. Thus, instantaneous velocity is proportional to tension and inversely proportional to viscosity and represents a combined effect of the viscoelastic parameters. Linear velocity is an estimate of the average rate at which myofibrils reach steady-state separation following ablation.
As a proof-of-concept, we first turned to investigating the effect of the inhibition of actomyosin contractility by blebbistatin on the mechanical properties of myofibrils. Actomyosin contractility is known to be the primary driver of active tension generation in both muscle and nonmuscle systems (Gordon, Homsher, & Regnier, 2000;Murrell, Oakes, Lenz, & Gardel, 2015). Toward this end, we expressed Lifeact-mApple (actin filament marker) in hiCMs and allowed them to spread overnight prior to treatment with the myosin II ATPase inhibitor blebbistatin (Straight et al., 2003). We noted that myofibril contraction was immediately attenuated upon addition of blebbistatin, as previously demonstrated (Fedorov et al., 2007). Furthermore, we found that 66.7% of cells contained at least one myofibril that adopted a bent or buckled shape upon blebbistatin treatment ( Figure 2a). Lasermediated dissection resulted in an initial separation followed by progressive retraction of the dissected myofibril ends until reaching a steady-state distance by 66.5 ± 1.9 s for control and 60.1 ± 3.1 s for blebbistatin treated myofibrils (Figure 2b,c). We observed a significant increase in the initial separation distance D 0 immediately following dissection (2.37 ± 0.23 μm for control versus 3.08 ± 0.22 μm for blebbistatin, p = .036).
Consistent with the idea that myosin II ATPase activity is required for active tension generation, we found a significant decrease in the asymptotic distance D a of separated myofibrils following blebbistatin treatment. As D a is proportional to myofibril tension, this suggests blebbistatin reduces active tension generation and elastic energy dissipation following ablation. Linear velocity was also significantly reduced indicating a slower average separation speed. We also observed a significant reduction in the relaxation time constant τ, suggesting blebbistatin reduces viscous damping within the myofibril.
That the reduction in asymptotic distance is more substantial than the reduction in relaxation time following treatment (92% reduction vs. 81% reduction, respectively), suggests the change in instantaneous velocity and energy dissipation immediately after ablation is primarily driven by reduced myofibril tension (Figure 2c,d). Taken together, these data indicate that our laser mediated dissection approach recapitulates current models of myofibril contractility.
We next investigated the role of FAK inhibition of the mechanical properties of myofibrils. We expressed Lifeact-mEmerald in hiCMs and allowed the cells to spread overnight prior to treatment with multiple doses of two different inhibitors of FAK activity, PF-228 and Defactinib (Kang et al., 2013;Slack-Davis et al., 2007). Expression of Lifeact tagged with mApple or mEmerald did not have any effects on any of the measured parameters in control hiCMs. Neither PF-228 nor Defactinib treatment attenuated hiCM beating, and we found no significant difference in the initial separation distance D 0 immediately following dissection. While PF-228 treated myofibrils reached the steady-state retraction distance in a similar time as control (46.2 ± 3.4 s for control, 44.2 ± 3.0 s for 1 μm PF-228 and 54.2 ± 2.9 s for 3 μm PF-228), Defactinib-treated myofibrils took significantly longer to reach steady state (95.1 ± 4.6 s for 2.5 μM Defactinib and 78.9 ± 5.4 s for 5 μM Defactinib).
Analysis of separation distance profiles revealed no significant changes in the asymptotic distance between treatment groups, relative to control (Figure 3a A previous study in nonmuscle cells has shown that PF-228 can activate BK Ca channels (So, Wu, Liang, Chen, & Wu, 2011). We therefore wanted to confirm that the observed changes in cardiac properties of the myofibril rather than to any potential effects of FAK inhibition on the electromechanical properties of hiCMs. That is, we wanted to evaluate whether FAK inhibition alters myocyte excitability to an extent which could affect sarcomere contraction through excitation-contraction coupling and ultimately lead to the observed differences in retraction dynamics. To that end, we performed electrical recordings of hiCMs using the Microelectrode Array (MEA) technology ( Figure 4a). This noninvasive label free approach allows the measurement of field potentials in thousands of cells (Asai, Tada, Otsuji, & Nakatsuji, 2010). Using this technique, we found that inhibition of FAK activity by treatment with either PF-228 or Defactinib had no effect on the beat period, spike amplitude or field potential duration (Figure 4b-d). Taken together, our results suggest that FAK inhibition alters the viscoelastic response of myofibrils through altered biophysical properties rather than changes in the electrophysiological properties of the cells.
In this study, we used laser-mediated myofibril dissection as a tool to assess the viscoelastic properties of myofibrils. The interpretation of such data requires mechanical descriptions of myofibril contraction. As such, we have employed a standard Kelvin-Voigt viscoelastic solid model commonly used to describe stress fibers or cell-cell junctions in nonmuscle cells (Fernandez-Gonzalez et al., 2009;Kumar et al., 2006). Using this framework, we found a good agreement between model fits and measured retraction dynamics of control and treated hiCMs (R 2 > .97 for all groups).
We further validated our approach by performing myofibril dissection experiments on blebbistatin-treated hiCMs. We found a marked decrease in both myofibril tension and viscosity, as indicated by significantly decreased asymptotic distances and relaxation times.
Thus, blebbistatin treatment attenuated active tension generation by actomyosin contraction and significantly reduced the viscous damping of the myofibril. As such, the myofibril showed minimal instantaneous velocity and end-to-end separation following the initial ablation. Interestingly, we also found that blebbistatin treated hiCMs displayed bent conformations of myofibrils that relaxed immediately following laser-assisted dissection. This relaxation correlated with an increase in the initial separation distance D 0 following myofibril dissection. It is well appreciated that myofibrils have passive elasticity that is independent of actomyosin contractility and largely dependent on titin . The instantaneous relaxation of the myofibril upon dissection, followed by a lack of any subsequent retraction, is suggestive of this passive elastic response.
In this study, we have identified a role for FAK signaling in determining the viscoelastic properties of the myofibril. Using multiple concentrations of two known FAK inhibitors-PF-228 and Defactinib-we found that ablated myofibrils showed no difference in asymptotic distance but a significant increase in relaxation time. This would suggest FAK inhibition does not alter active tension generation of the myofibril, but instead increases viscous damping of the retraction. Indeed, the observed decrease in instantaneous velocity and energy release following ablation is driven largely by the increased myofibril viscosity. However, the precise mechanism by which FAK inhibition increases myofibril viscosity remains unclear.
Previous studies have implicated actin-titin interactions, collagen fibrils, and microtubule tyrosination as potential regulators of myofibril viscosity (Caporizzo et al., 2018;Granzier & Irving, 1995;Kulke et al., 2001). We have previously shown that FAK inhibition results in stabilization of cell-ECM adhesions, which leads to increased incorporation of titin (Taneja, Neininger, & Burnette, 2020). Therefore, one possible mechanism could be due to increased actin-titin interactions mediated by increased cell-ECM adhesion. Alternatively, FAK inhibition may be altering microtubule stability by altering tyrosination, as has been reported previously in fibroblasts (Palazzo, Eng, Schlaepfer, Marcantonio, & Gundersen, 2004). These hypotheses are not mutually exclusive and future studies will be required to elucidate the exact mechanism(s) by which FAK regulates myofibril mechanics. Medium in a cell culture incubator at 37 C and 5% CO 2 . Media was exchanged every 2 days as recommended by the manufacturer. Exogenous expression was performed using Viafect (Promega, Cat. E4981).
Then, 200 ng of plasmid DNA was used to transfect one well of a 96-well culture plate according to the manufacturer's instructions.
For ablation experiments, one well of hiCMs (50,000 cells) was replated onto a 35 mm dish with a 10 mm glass bottom well (CellVis, Cat. D35-10-1.5-N), as described in detail previously (Fenix et al., 2018). The growth substrate was coated with 10 μg/ml fibronectin (Corning, Cat. 354,008) for one hour prior to use. Briefly, cells were washed twice with 100 μl washes of PBS, followed by incubation in 40 μl of 0.1% Trypsin for 2 min at 37 C. Cell detachment was confirmed using a bright-field microscope. Following trypsinization, 160 μl of Maintenance Media was added to the well and cells were collected for centrifugation. Cells were spun at 200g for 3 min. The supernatant was removed, and the pellet was resuspended in 200 μl of Maintenance Media and plated on the fibronectin-coated coverslip. Cells were allowed to spread for 16 hr prior to ablation. At this time point, myofibrils are on the dorsal surface of the cell (Fenix et al., 2018).
For experiments using the Axion bioanalyzer, hiCMs were thawed onto a 48-well CytoView MEA plate (Axion Biosystems, Cat. M768-tMEA-48B) coated with 50 μg/ml fibronectin at a concentration of 20,000 cells per well. Cells were cultured as outlined above for 10 days prior to experimental treatments.

| Laser-assisted dissection
Focused laser mediated myofibril dissection was performed as described previously . Myofibril dissection was performed on a Nikon Spinning Disk confocal microscope equipped with a ×60 1.4 NA objective and an Andor iXON Ultra EMCCD camera, provided by the Nikon Center of Excellence at Vanderbilt University. Dissection was performed using a 100 mW UV laser (Coherent technologies) at 50% power, using a dwell time of 500 μs for a total period of 1 s using a Stimulation Line ROI of 2 μm length. Cells were maintained at 37 C and 5% CO 2 using a Tokai Hit stage incubator. Three preablation images were acquired at 2 s intervals. Following ablation, images were acquired using continuous acquisition at a net acquisition rate of 138 ms per frame. For the Defactinib experiments, the frame rate was 1 s. Images were acquired until retraction reached a steady-state distance as determined by visual inspection.

| Myofibril retraction analysis
To quantify myofibril mechanical properties, individual myofibrils were subjected to laser-assisted dissection to induce separation. Following can also be mathematically expressed as 0.632 D a . Additionally, any initial damage induced by the laser was accounted for by defining an initial separation distance D 0 that was measured at the time of ablation and subtracted from the measured retraction curve prior to fitting.
We also calculated two additional parameters-instantaneous velocity and linear velocity-to characterize the dynamics of myofibril retraction. Instantaneous velocity (v I ) represents the initial release of stored energy upon ablation and was calculated as the local derivative of the retraction (i.e., dL(t)/dt) at the time of ablation (t = 0). Defined in this way, the instantaneous velocity is equivalent to the ratio of the asymptotic distance to the relaxation time (e.g., v I = D a /τ) and represents a measure of the combined mechanical effect of the two fit parameters on the retraction response. Finally, the linear velocity (v L ) is calculated as the slope of the line connecting the first and last points of the retraction profile and represents the average rate at which myofibrils retract toward the steady-state separation distance.

| Impedance assays
The Axion Biosystems analyzer was used to measure contractility and impedance in hiCMs as described previously .
Recordings were taken for 5 min at baseline and after 2 hr of drug treatment. Cells were maintained at 37 C and 5% CO 2 during recording. Cells were assayed using the standard cardiac analog mode setting with 12.5 kHz sampling frequency to measure spontaneous cardiac beating. The Axion instrument was controlled using Maestro

| Statistical analysis
Statistical analysis was performed in GraphPad and Excel. The comparison of retraction parameters was performed using the Student's t test while comparison of electrophysiological properties was performed using two-way analysis of variance. All error bars represent SEM.

ACKNOWLEDGMENTS
The authors thank Nikon Center of Excellence at Cell Imaging Shared