Spatiotemporal Stability of Neonatal Rat Cardiomyocyte Monolayers Spontaneous Activity Is Dependent on the Culture Substrate

In native conditions, cardiac cells must continuously comply with diverse stimuli necessitating a perpetual adaptation. Polydimethylsiloxane (PDMS) is commonly used in cell culture to study cellular response to changes in the mechanical environment. The aim of this study was to evaluate the impact of using PDMS substrates on the properties of spontaneous activity of cardiomyocyte monolayer cultures. We compared PDMS to the gold standard normally used in culture: a glass substrate. Although mean frequency of spontaneous activity remained unaltered, incidence of reentrant activity was significantly higher in samples cultured on glass compared to PDMS substrates. Higher spatial and temporal instability of the spontaneous rate activation was found when cardiomyocytes were cultured on PDMS, and correlated with decreased connexin-43 and increased CaV3.1 and HCN2 mRNA levels. Compared to cultures on glass, cultures on PDMS were associated with the strongest response to isoproterenol and acetylcholine. These results reveal the importance of carefully selecting the culture substrate for studies involving mechanical stimulation, especially for tissue engineering or pharmacological high-throughput screening of cardiac tissue analog.


Introduction
Cardiomyocytes are central to the electromechanical properties of the heart. Cardiac tissue can exhibit sensitivity to arrhythmias, which are complex spatiotemporal electrical activities. Dynamic electrical behaviors of cardiac tissue include normal propagation from pacemaker sites [1,2], stable rotating spiral waves [3], unstable spiral waves that break up during propagation [4], and bursts of activity often associated with the initiation and termination of spiral waves [5]. In situ, cardiomyocytes are electrically excitable; some exhibit spontaneous activity environment matches that of cardiac cells [43]. Different elastic substrates have been used to study the effect of mechanical properties (microenvironment or deformation) on cells; these materials include polydimethylsiloxane (PDMS), polyacrylamide, alginate, chitosan, agarose, and polyesters [44]. Silicone-based materials offer better tolerance to mechanical deformation compared to hydrogels like polyacrylamide that are weak and brittle; indeed, polyacrylamide is not a suitable material for stretch-related studies [45]. PDMS is hydrophobic and does not allow direct cell attachment and proliferation [46]. This issue can be resolved by coating PDMS with proteins that attach to hydrophobic surfaces, such as gelatin or fibronectin [46]. In addition to surface coating, PDMS can be temporarily rendered hydrophilic by exposing the surface to air plasma [46].
In the present study, we chose to investigate the impact of substrate type (glass vs PDMS) on the frequency, stability, and organization of spontaneous activity, as well as the impact of sympathetic and parasympathetic stimulation, on neonatal rat cardiomyocyte monolayers. We also evaluated the level of different genes implicated in the generation of action potentials and electrical conduction mechanisms. As such, the expression of HCN2 and CaV3.1 mRNA increased on the softer substrate. When focusing on autonomous activity, we observed an increased number of activation sites (meaning changes in the location of activation) on PDMS compared to the glass substrate. This was related to an increased instability of spontaneous activity and a tendency towards a decreased expression of connexin-43 (Cx43) on the PDMS substrate. Finally, cardiomyocytes cultured on PDMS 1:20 exhibited the greatest increase in the frequency of contraction after stimulation with ISO and the greatest decrease with ACh compared to cultures on glass. This study highlights the importance of carefully evaluating the choice of elastic substrate for mechanical stimulation. Silicone-based PDMS can favor unstable spontaneous rhythms and responses to pharmacological stimuli by altering intercellular connectivity and ion channels influencing the voltage clock of cardiomyocytes.

Materials and Methods
All animal-handling procedures were concordant with the Canadian Council on Animal Care guidelines and approved by the institutional Animal Research Ethics Committee (Permit Number: 2011-35-01P). Efforts were made to minimize suffering.

PDMS substrate fabrication
Polydimethylsiloxane (PDMS, SYLGARD 184, Dow Corning) is a binary compound chosen because of its elasticity and optical transparency. Mass ratio of pre-polymer (base) and crosslinker (curing agent) can be varied to get a softer to harder material after 48 hours of curing at room temperature [47][48][49]. De-foaming is performed to remove air bubbles and spin coating (2100 rpm) for 10 seconds is applied to get a thin and uniform surface. The calculated thickness of the PDMS layer is approximately 200 μm. After full curing of the PDMS, but just before coating and seeding of the cells, the dishes were plasma-cleaned (Harrick Plasma, PDC-32G) to increase the wettability of the surface [50]. Different substrates were used: glass, PDMS ratio (curing agent:base) of 1:20, and PDMS ratio of 1:40. These PDMS ratios were chosen because they are close to the 1:10 ratio suggested by the manufacturer while having Young's moduli close to the physiological value. rapidly excised and immediately placed in cold Ca 2+ -and Mg 2+ -free Hank's Balanced Salt Solution. Ventricular muscle was harvested and then minced on ice into 1-3 mm 3 pieces. Purified enzymatic digestion (50 μg/mL trypsin and 136 μg/mL collagenase) was then used to gently dissociate cardiomyocytes. Isolated cells (enriched cardiomyocytes) were counted and seeded at a density of 10 6 cells/mL in 20-mm diameter glass-bottom culture dishes (D29-20-0-N, In Vitro Scientific, Sunnyvale, CA, USA) pre-coated (on glass or on PDMS) with a mixture of 0.2% porcine-derived gelatin (G1890, Sigma-Aldrich, Oakville, Ontario, Canada) and 0.00125% fibronectin solution (F1141, Sigma-Aldrich) [51]. Cells were grown for 24 hours in DMEM (319-050-CL, Wisent Inc., St-Bruno, Canada) with 5% fetal bovine serum (SH30396.03, Fisher Scientific Co. Ltd, Ottawa, Ontario, Canada) and 1% penicillin/streptomycin (450-201-EL, Wisent Inc.). Cardiomyocytes were then starved of fetal bovine serum for another 24 hours in DMEM with 1% penicillin/streptomycin prior to experimentation.
The specific methodology for cell population evaluation based on gene expression can be found in S1 Appendix.

Spontaneous contraction recordings
Phase contrast images of neonatal cardiomyocytes were acquired 48 hours post seeding in vitro with a Dalsa HM640 camera (60 frames by second) coupled to an inverted Nikon optical microscope (10X magnification and a field of view of 0.34 mm × 0.45 mm). Spontaneous activity measurements were performed with an algorithm developed in Matlab. Video analysis was carried out as follows: a frame of reference corresponding to the pre-contraction state of the cells (M rest ) is determined and the corresponding 2D image is then subtracted to all acquired frames (M diff (x,y,t) = M(x,y,t)-M rest ). The sum of the absolute difference M diff (x,y,t) over all pixels (x, y) divided by the number of pixels serves as an aggregate temporal signal. A moving window of 5 samples is then applied to smooth the signal followed by baseline drift removal (subtraction of 40 samples moving window filtered signal). Spontaneous frequency, standard deviation of period (σ noise ), and pauses (corresponding to time interval greater than 3 seconds) were calculated from the video data. Videomicroscopy (30 seconds recordings) was performed at 37°C over a 20 minutes period at the following time points: 0 (pre-drug), 1, 5, 10, 15, and 20 minute (s) for each condition (glass, PDMS 1:20, and PDMS 1:40).

Mapping of calcium dynamics
After 48 hours of culture, the cardiomyocytes were washed once with fresh media and incubated with 10 μM of fluo-4 AM (F-14201, Life technologies, Burlington, Ontario, Canada) and 0.2% Pluronic acid F-127 (P-3000MP, Life Technologies) for 30 minutes at 37°C. Fluo-4loaded cardiomyocytes were then washed 4 times with fresh media, followed by a 15 minutes resting period to allow de-esterification of the dye before starting calcium transient mapping experiments performed in DMEM at 37°C. Fluorescence was recorded for 30 seconds at 125 Hz with a CardioCCD camera (RedShirt Imaging, Decatur, GA, USA). The dye was excited with a quartz tungsten halogen lamp (Oriel Instruments Inc., Stratford, CT). The filters used for excitation and emission were λ excitation 480±20 nm (Chroma Technology Corp, Bellows Falls, VT) and λ emission 535±25 nm (Semrock Inc., Rochester, NY), respectively. The system was set to image with an acquisition frame rate of 125 Hz for all experiments. Signals were filtered and analyzed using a program developed in-house using Matlab software (R2008, MathWorks Inc., Natick, MA). In brief, raw acquisitions were normalized by the minimum fluorescence for each pixel and the dF/dt (the first time derivative of fluorescence) was approximated using a finite difference approach. The resulting signals were then spatially filtered using a Gaussian kernel (5×5 pixels with σ = 1.5) followed by a temporal moving average with a two-sample window.

Quantitative Polymerase Chain Reaction (qPCR)
Briefly, cultured neonatal cardiomyocytes were harvested in RA1 lysis buffer from Nucleospin RNA II kit (Macherey Nagel GmbH & Co. KG, Waltham, MA, USA). mRNA was isolated using the same kit following the manufacturer's instructions, including DNAse treatment to prevent genomic contamination. mRNA was then reverse-transcribed with the High-capacity Reverse Transcription kit (Applied Biosystems, Burlington, Ontario, Canada). qPCR was performed with TaqMan probes from Applied Biosystems for housekeeping genes (hypoxanthine guanine phosphoribosyl transferase (HPRT), Glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and Beta-2 microglobulin (β2M) and with primers designed for SYBR Green experiments to assess the expression of Kir2.1, CaV3.1, CaV3.2, HCN2, HCN4, Cx43, Kir3.1, Kir3.4, and Adrβ1 (S1 Table for sequences). The geometric mean of the expressions of HPRT, β2M and GAPDH was used for normalization. qPCR reactions were performed with TaqMan Gene Expression Master Mix and Power SYBR Green kits from Applied Biosystems. Reactions were run on an Mx3000 qPCR System from Stratagene. Relative gene expression values were calculated by the 2 -ΔCt method.

Sympathetic and parasympathetic stimulation
Acute effects of the sympathetic agonist, isoproterenol (ISO, I6504, Sigma-Aldrich) and the parasympathetic agonist, acetylcholine (ACh, A2661, Sigma-Aldrich) were studied by videomicroscopy and calcium imaging at 48 hours post-culture. Final concentrations of 100 nM ISO and 1 μM ACh were added to the culture media to stimulate sympathetic and parasympathetic pathways, respectively. Drug effects were evaluated by videomicroscopy. In a second set of experiments, changes in spatiotemporal activity by ISO and ACh were studied using calcium imaging with fluo-4.

Statistical analysis
Data were analyzed using Prism 5 by GraphPad. One-way analysis of variance (anova) followed by Tukey multiple comparison test was used to compare groups unless otherwise stated. A p-value below 0.05 was considered statistically significant.

Stiffness of the PDMS substrate
Stiffness measured as a function of the ratio of pre-polymer (base) and cross-linker (curing agent) is presented in supplementary data (S1 Fig). Stiffness is expressed as the modulus of elasticity measured by Young's modulus formula (S2 Appendix). PDMS mixed in ratios of curing to base agent (curing:base) produced substrates with mean moduli of 974±32 kPa (1:5), 293  (1:40). In comparison, stiffness of the glass surface has been shown to be~30 GPa [22].

Spontaneous activity in monolayer cultures
The effect of substrate stiffness on spontaneous activity was first evaluated by videomicroscopy. Softer substrate tends to diminish the autonomous contraction frequency of cultured cardiomyocytes as spontaneous frequency was decreased by 24.4% (PDMS 1:20) and 30.8% (PDMS 1:40) compared to glass (S2A Fig). Interestingly, the variability in the frequency within groups is much greater in the glass group compared to PDMS groups with a subset of higher frequencies.
A mapping study of calcium activity was performed to determine if spatiotemporal activity could explain the presence of high-frequency activity. Although most of the activity was ectopic in nature with spatially restricted origin sites, spontaneous reentry could also be detected in the glass substrate group as illustrated with a typical example in Fig 1. The time course of normalized calcium transients ( Fig 1A) was obtained from the pixel marked by an "x" (in Fig 1B). Snapshots of the dF/dt are depicted in Fig 1B over 1 rotation of reentry. The period of activity in this example was highly stable at 255 ms. Propagation along the circular trajectory (which is not always perpendicular to the activation front) depicted as a white circle ( Fig 1B) shows periodic activity ( Fig 1C). However, it is clear that the local activation delay δ (δ = slope) is not constant with large regions of rapid velocity along the trajectory from 3π/4 to 3π/2, while the region close to 2π illustrates a slow conduction area. Local activation delay was not uniform along the reentrant trajectory and changed from δ 1 = 0.15 second/cm to δ 2 = 0.4 second/cm, δ 3 = 0.18 second/cm, and finally δ 4 = 0.66 second/cm. The activation map ( Fig 1D) was obtained for the time interval between 0.1 and 0.35 second of reentrant activity. The maps shows a central core of~0.31 mm 2 (indicated by the arrow) and highlights the slow region of the reentrant pathway (time near 0.1 to 0.15 second).
The high frequency of reentrant activity found in mapping led us to go back to the videomicroscopy data. Further analysis of the videomicroscopy mean frequency recorded at different time points (0, 1, 5, 10, 15, and 20 minutes) for a duration of 30 seconds shows a significantly higher incidence of high-frequency reentrant activities for cardiomyocyte monolayers cultured on glass compared to PDMS (5% for glass and 0% for both PDMS 1:20 and 1:40, p<0.05 with Fisher's exact test when comparing samples with at least 1 episode of frequency >3 Hz). Highfrequency samples were assumed to be associated with reentrant activity and appeared responsible for the frequency of substrate-dependent activity. Indeed, removal of these reentrant-associated samples (defined as frequencies >3Hz) resulted in a loss of difference, as the mean spontaneous activity was reduced by 1.5% and 9.8% for cardiomyocytes cultured on PDMS (1:20 and 1:40, respectively) compared to glass (S2B Fig). Thus, higher mean frequency of spontaneous activity of monolayers on glass detected by videomicroscopy could be explained by a greater probability of spontaneous reentry.

Stability of the spontaneous activity
Temporal variation of the interbeat interval is an important parameter of spontaneous activity, reflecting the stability of cardiac homeorhesis. A strong oscillator, as desired in tissue engineering aimed at the development of biopacemakers [52] should show less variability in the interbeat interval, while a weak oscillator may be more sensitive to perturbation yielding more variation in spontaneous activity. Thus, the standard variation in the interbeat interval (σ period ) from a 30 seconds acquisition was used as a measure of temporal activity. Typical examples of peak-detection and period signatures are presented in the left column of Fig 2A and the  correlated interbeat intervals (time difference between peaks highlighted by circles in the column on the left) are displayed in the columns on the right. The rate of spontaneousactivation appeared less stable with greater variance of the PDMS groups compared to glass (Fig 2B, Ansari-Bradley test, p<0.05). PDMS 1:20 had the highest variation (median σ period = 0.29 second), followed by PDMS 1:40 (median σ period = 0.23 second), and glass (median σ period = 0.19 second). Evaluation of the number of pauses (absence of spontaneous contraction for longer than 3 seconds) shows that pauses are significantly more frequent in the PDMS 1:20 group compared with the PDMS 1:40 group (presented in Fig 2C). The ratio of samples with at least 1 pause were calculated as 0.1, 0.24, and 0.05 for glass, PDMS 1:20, and PDMS 1:40, respectively. There was a significant difference between the two PDMS groups.

Number of activation sites
Mapping data (30 seconds acquisitions) revealed that the number of activation sites (indicative of spontaneous activity) of the entire seeding area of the petri dishes (20 mm) was generally lower for cardiomyocyte monolayers cultured on glass compared to PDMS. As shown in Fig Fig 3D), the mean interbeat interval was 0.70±0.18 second and activation occurred over 3 sites. Heterogeneity in the spontaneous pacemaking activity of isolated neonatal rat cardiomyocytes is well established [53]. Intercellular communication might therefore affect automaticity of a monolayer consisting of beating and non-beating cells [54,55]. Intercellular communication was then probed through the expression of Cx43, the main protein involved in cardiac electrical coupling. Surprisingly, a lower mRNA expression of Cx43 (p = 0.12) was noted in monolayers cultured on PDMS (1:20 or 1:40) compared with glass ( Fig 4A). Decreased cellular coupling could therefore increase the propensity for the multifocal activation observed on PDMS.

Parasympathetic and sympathetic modulation of spontaneous activity
Relative spontaneous frequency was obtained by videomicroscopy and calculated by taking spontaneous frequency of contraction recorded at t = 1, 5, 10, 15, and 20 minute(s), and by dividing that value by the spontaneous frequency prior to the addition of the drug (pre-drug at  . No significant change in σ period was found over time in cardiomyocytes cultured on glass after the addition of ISO over 20 minutes (Fig 5C).
Statistical differences for σ period in the presence of ISO were achieved on PDMS 1:20 after 10 minutes of ISO addition, indicating a stronger effect on this substrate that also had the highest median of σ period (as shown previously in Fig 2A).  Interestingly, increased σ period could be observed at t = 1 minute after the addition of ACh for glass and PDMS 1:40 substrates, while the effect on PDMS 1:20 appeared later at t = 5 minutes. Overall, this indicates that the substrate influenced drug action on spontaneous activity. In this example, the destabilization effect of ACh seems to be decreased when cardiomyocytes are cultured on PDMS 1:20.

Parasympathetic and sympathetic stimulation: effect on 1 st activation site
As the number of activation sites is significantly higher on PDMS substrates compared to glass substrate, we checked whether the significant decrease in σ period following ISO injection is associated with a diminution in the number of activation sites. Mapping

Stability of cell populations between substrates in culture
Myofibroblasts can be found in cardiomyocyte cultures. These cells can be electrically coupled to cardiomyocytes in culture and possibly in tissue thus affecting depolarization, ectopic activity, electrical conduction, and reentry sensitivity [57][58][59][60]. As such, evaluation of possible changes in myofibroblast population between the groups under study has been done. Expression of alpha-actinin gene in primary cultures was evaluated with qPCR ( S8 Fig). A similar level of alpha-actinin gene expression was found between the glass and PDMS groups. There is no indication of a difference in the number of cardiomyocytes within monolayers between the groups investigated. Results obtained for alpha smooth muscle actin (alpha-SMA), a marker of myofibroblast, are presented in S9 Fig No significant differences in gene expression of alpha-SMA was found between the groups of interest (glass, PDMS 1:20, PDMS 1:40). These results indicate that no important difference in the cell population could behind the difference in dynamics between glass and PDMS substrates.

Discussion
To study stretch-related effects including simulating pathological conditions like hypertrophy, pressure-overload [61], or atrial cardiomyocyte remodeling [62], a flexible cell culture substrate like StageFlexer Membrane with Flexcell systems (Flexcell International Corporation, Hillsborough, NC, USA) or similar silicone-based membrane (e.g. non-reinforced vulcanized silicone membrane) is generally used [63][64][65]. We hypothesized that culturing on such elastomeric substrates affects the spontaneous activity of cultured neonatal rat ventricular cardiomyocyte (NRVM) monolayers, based on published studies illustrating the effect of the substrate properties on cardiomyocyte shape [41,66,67], calcium transient morphology [21], and spontaneous frequency of contraction (using other types of substrates) [35,68]. To the best of our knowledge, this is the first investigation aimed at elucidating the impact of silicone-based substrate properties on the spontaneous activity of NRVM. PDMS has numerous advantages as a material for cell study, including optical clarity, biocompatibility, high tolerance to stretch compared to hydrogels, and possibility to modulate its stiffness to reach physiologically relevant ranges [69]. Control of the final geometry is also possible using a simple molding technique for curing the PDMS. This technical aspect is highly important in tissue engineering to control the seeding area and/or apply patterning to the seeding surface to induce orientation of the cells [3]. Culture of cardiomyocytes on a stiff substrate has been shown to impact their function both for force generation [19,21,23] and spontaneous activation [35,68].
Single [70,71] and multiarmed reentry [72] can be created by point electrical stimulation or spontaneously arise [73] in cardiac monolayers. In the present study, an increased number of spontaneous reentrant events were observed when cardiomyocytes were cultured on a glass substrate compared to a PDMS substrate. This finding increased the intragroup variability. By removing these reentrant events (Fig 1), we obtained a smaller intergroup variability of frequencies. Stiffness differences between glass and PDMS substrates do not significantly affect the spontaneous beating frequency (S2 Fig), but seem to affect the incidence of reentrant activities.
Another factor that could impact spontaneous activity of monolayers is cell density [74]. Density influences cardiomyocyte function because cells react to the substrate and also to their neighboring cells through mechanosensitivity [75]. In our study, we tried to limit the impact of cell density by counting living cells before seeding and by plating the same quantity of cells on each substrate. Thus, the significant increase in the number of nuclei found for PDMS 1:20 (S3 Fig) indicates a possibly greater proliferation on this substrate. In cases of initial low density at seeding, increased cell proliferation could lead to stronger intercellular coupling and favor faster spontaneous frequency [74]; however, we did not observe this effect.

Stability of the spontaneous activity
Our study illustrated an increased temporal variability in the interbeat period (σ period ) on PDMS 1:20 (Fig 2). Lower expression of Cx43 tends to uncouple cardiac cells and could highlight the intrinsic rhythm of each cell (pacemaker activity) [55]. Subsequently, the down regulation of Cx43 mRNA detected in our study could explain why we observed an increased number of activation sites on PDMS compared to the glass substrate (Fig 3). Reentry is more likely to occur in excitable tissue that is partially uncoupled [1]. Probability of reentrant activity has been shown to be increased in low to mildly coupled monolayers [54,72]. Here, Cx43 mRNA has the tendency to be lower on PDMS 1:20 compared to glass while the incidence of reentry is increased on glass compared to PDMS. This result indicates that other factors could protect from spontaneous formation of reentry on PDMS. This might be explained by a non-uniform distribution of Cx as highlighted by the change in local activation delay along the trajectory (more importantly for the region close to 2π illustrated in Fig 1). A non-uniform distribution of Cx can facilitate reentry and non-uniform conduction properties [76]. However, curvature effects can also induce change in conduction velocity [77,78]. The increase in the number of activation sites on PDMS could possibly be linked to the increased median σ period calculated in PDMS compared to glass (52.6% for PDMS 1:20 and 21.1% for PDMS 1:40). These different σ period values are affected by the increased variation between beats, but also by occasional pauses between bursts of faster beats in PDMS 1:20 recordings. These results are in agreement with Ponard et al, who demonstrated that stochastic gating of transmembrane currents and calcium release channels can affect ion channel turnover and thereby induce variability in the spontaneous rate [79]. However, we detected a larger change in spatial position of focal activation than Ponard et al., which we attribute to greater distances (~6 mm) between interacting pacemaker sites compared with smaller distances (1 mm) reported by the authors [79].

Changes in the voltage clock-associated mRNA and effects on spontaneous activity
The voltage clock encompasses membrane currents that play a role in the spontaneous activity of cardiomyocytes, and more importantly, in sinus node cells [10]. Modifications in expression or function of membrane currents could alter the rate of spontaneous activity. In our study, we observed an increased expression of CaV3.1 (I CaT ) and HCN2 (I f ) mRNA. Increased expression of Ca 2+ -permeable (T-and L-type) channels and HCN channels including HCN2 [80] could promote automaticity [8,11,81,82]. Although not a significant finding, the tendency towards decreased Kir2.1 expression (Fig 4) could also favor acceleration of spontaneous activity [83]. However, no significant difference in frequency was detected (S2 Fig). Instead, there was a significant increase in the number of activation sites on PDMS (Fig 3). One possible explanation is that the global change in mRNA expression could lead to protein changes that do not cause a linear shift in the local intrinsic period of individual spontaneous cells. We perceive monolayers to consist of quiescent and spontaneous cells (with a set of individual frequencies) [55]. Cells with higher frequency are thought to become the drivers of monolayers; however, locally connected strong pacemakers could more easily overcome the source/sink ratio and drive spontaneous activation. Therefore, a decrease in cellular coupling (through Cx43 mRNA) and increase in local spontaneous activity (decreased Kir2.1 and increased HCN2) would favor regionalized depolarization. As such, these results suggest that culture on a PDMS substrate may favor the appearance of spontaneous activity in multiple cells and increase the number of regions that can depolarize and generate a propagating response.

Response of different substrates to sympathetic and parasympathetic stimulation
Vagal nerve stimulation slows heart rate, predominantly by the action of the vagal neurotransmitter, acetylcholine, at muscarinic receptors. The binding leads to the dissociation of G proteins and activation of the potassium current, I K,Ach [84]. In our study, ACh (1 μM) significantly decreased the rate of spontaneous activity of monolayers in all groups, with the greatest decrease on PDMS 1:20 compared to the glass substrate ( Fig 5B) and the greatest instability on PDMS (Fig 5D, higher number of time points with significant increase in σ period compared to pre-ACh data). Increased instability is a known hallmark of ACh-induced slowing effects [85]. Interestingly, instability was more rapid on glass and PDMS 1:40, while its effect was more gradual on PDMS 1:20. Increased σ period was only visible after 5 minutes on PDMS 1:20. The mRNA levels for both Kir3.1 and Kir3.4 (S4D and S4F Fig), coding for ACh-activated K+ channels, were not significantly different but showed a tendency towards a slight increase on PDMS. At early developmental stages, I K,ACh is known to be primarily formed by Kir3.1, while in late embryonic and adult cells, Kir3.4 is the predominant subunit [86]. Here, the change in culture substrate did not induce variation in the ratio of Kir3.4/ Kir3.1, indicating that there was no modification in the rectification properties of I K,ACh and therefore function between substrates.
It has recently been shown that substrate stiffness can regulate cellular responses to β-adrenergic receptor agonist (ISO) in human mesenchymal stem cells [87]. It was unknown whether ISO could exert different effects on cardiomyocytes cultured on PDMS. Interestingly, cardiomyocytes cultured on PDMS (1:20) exhibited the greatest increase in frequency of spontaneous activity by 100 nM ISO (relative to pre-drug at t = 0). More precisely, the effect lasted significantly longer on PDMS 1:20 compared to culturing on glass. ISO significantly reduced σ period on the PDMS 1:20 substrate (Fig 5C), which confirms that instabilities are greater for slower activity [85]. However, we found that decrease in instability is also associated with a diminution in the number of activation sites (Fig 6 and S5 Fig). This result is in agreement with a previous study that claims the positive inotropic effect of ISO is associated with a decrease in the relative value of the potentiating effect of the pause (instability) [88].
We hypothesized that expression of beta-1 adrenergic receptor (AdrB1) could be altered by the culture substrate; however, no significant change at the mRNA level was found. It remains to be determined if post-transcriptional regulation process including a change in internalization rate of the receptor [87] could be responsible for the differences. The increased duration of the effect of ISO stimulation may be a consequence of the substrate on phosphodiesterase action. cAMP plays a central role in regulating metabolic and cellular processes through protein kinase A (PKA). Phosphodiesterases are directly involved in the degradation of cAMP controlling the duration and intensity of the response in cAMP signaling [89] and has an impact on sinoatrial node activation frequency [90].

Conclusion
This study is part of the greater project aimed at the development of a system to evaluate the effects of stretch on cardiomyocytes. To apply deformation, a flexible substrate is needed. Due to the brittle characteristics of currently available hydrogels, we opted for the use of a deformable silicone-based substrate to approach in vivo tissue stiffness. According to our results, selecting the appropriate material for the substrate is highly important since it will affect the in vitro spontaneous activity via ionic channel expression modulation. Our study also presents evidence that the substrate mechanical properties can influence the sensitivity to sympathetic (ISO) and parasympathetic (ACh) stimulation, and act as a stabilizer for the propagation of activation through changes in Cx43 mRNA. These results reveal the importance of carefully selecting the culture substrate involving mechanical stimulation, especially for tissue engineering investigations or pharmacological high-throughput screenings of cardiac tissue analog. To determine Young's modulus, the PDMS substrate was molded in a cylindrical mold with defined width and length. Molds were cured at 37°C for 48 hours. Young's modulus was calculated as the slope of the stress/strain curve that was created using a set of weights ranging from 10 to 150 g (except for PDMS 1:40 where the maximum weight was 30 g). Attempts were made to remain within the limits of linear elasticity. PDMS was mixed in ratios of 1:5, 1:10, 1:15, 1:20, 1:25, 1:30, and 1:40 to produce substrates with mean moduli of 974±32 kPa, 293±8 kPa, 112±6 kPa, 87±22 kPa, 42 ±6 kPa, 27±4 kPa, and 16±4 kPa, respectively. Inset: low modulus values for large-mixed ratios. ; n = 11. The rate of spontaneous activity tends to decrease when cardiomyocytes are cultivated on PDMS with a greater effect on softer substrates (p = NS). The large error bar for the mean spontaneous frequency measured on glass can be explained by a group of data with high-frequency rate that match the rate of reentry imaged in the calcium mapping experiments (A). After removing the data with frequency greater than 3 Hz assumed to be reentrant activity, mean spontaneous activity measured by videomicroscopy on glass is 1.