2.1.

Culture of Human iPSC-Cardiomyocytes and Optogenetic Modifications

The human iPSC-CMs were the iCell² (iCell Cardiomyocytes² (Cat. C1016, Donor 01434) from Fujifilm/Cellular Dynamics International, Madison, Wisconsin. The hiPSC-CMs samples were cultured on fibronectin-coated ($50\mu \mathrm{g}/\mathrm{mL}$) 35 mm dishes with 14 mm glass-bottom insert (Cellvis, Mountain View, California) at 270,000 cells per well and were grown in humidified incubator at 37°C and 5% ${\mathrm{CO}}_2$ in manufacturer-provided culture medium and following manufacturer’s instructions. The maintenance medium was changed every other day until 7 days postplating when measurements were conducted. For optogenetic actuation, channelrhodopsin2 (ChR2) was expressed in the cells by adenoviral transduction with Ad-CMV-hChR2(H134R)-eYFP (Vector Biolabs, Malvern, Pennsylvania) at multiplicity of infection (MOI 50) on day 5 postplating, optimized to provide healthy and light-responsive samples.²⁶

2.2.

Functional Experiments with Human iPSC-CMs

On day 7 after plating, samples were prepared for experiments. Comparative experiments were performed either in the phenol-red containing manufacturer-provided culture medium or in Tyrode’s solution (in mM: NaCl, 135; ${\mathrm{MgCl}}_2$, 1; KCl, 5.4; ${\mathrm{CaCl}}_2$, 1.8; ${\mathrm{NaH}}_2{\mathrm{PO}}_4$, 0.33; glucose, 5.1; and HEPES, 5; adjusted to pH 7.4 with NaOH). We observed that the pH value of the Tyrode’s solution varies with temperature, and for stable readings, we chose to pH the solution and to conduct the experiments at room temperature. To mimic operation in an incubator, experiments in culture medium were performed at elevated temperature ($\sim 33°\mathrm{C}$ to 35°C). Prior to optical mapping, cells were dual-labeled with spectrally compatible fluorescent indicators for membrane voltage $V_m$ and intracellular calcium $[{\mathrm{Ca}}^{2+}{]}_i$ in Tyrode’s solution, as described previously.¹⁸ The $[{\mathrm{Ca}}^{2+}{]}_i$ indicator Rhod-4AM (AAT Bioquest, Sunnyvale, California) was used at $10\mu \mathrm{M}$ concentration for 20 min, followed by a wash. Then the near-infrared $V_m$ dye BeRST1,²⁷ courtesy of Evan W. Miller (UC Berkeley), was applied at $1\mu \mathrm{M}$ concentration for 20 min. Final wash was conducted either in culture medium or Tyrode’s solution depending on experimental conditions and cells were prepared for imaging. Electrical stimulation was achieved by a bipolar platinum electrode, connected to a MyoPacer stimulator (IonOptix, Westwood, Massachusetts). Optical stimulation was integrated in the system, as described below.

In some experiments, cells were treated with cell uncoupler 1-Heptanol (Sigma-Aldrich, St. Louis, Missouri) at concentration of 0.5 mM for 30 min. In another subset of experiments, cells were treated with an HDAC inhibitor Trichostatin A (TSA) (Sigma Aldrich)—a drug known to lead to proarrhythmic responses—at 100 nM for 24 h prior to experiments.

2.3.

System Components

The compact macroscopic system for all-optical electrophysiology is structured on a $12\times 18$-inch aluminum breadboard which makes it portable; all mechanical components are standard parts from Thorlabs, Newton, New Jersey. The system is designed for multiparametric optical imaging of $V_m$, $[{\mathrm{Ca}}^{2+}{]}_i$, and dye-free (label-free) imaging of mechanical waves, under electrical and/or optical/optogenetic stimulation. Figures 1(a) and 1(b) show the system, which uses two low-cost cameras, both Basler acA720-520um (Basler AG, Germany), having $720\times 540\mathrm{pixels}$, $6.9\mu \mathrm{m}$ per pixel, capable of running at up to 525 frames per second (fps). Camera 1 collects the emitted fluorescence to image $V_m$ or $[{\mathrm{Ca}}^{2+}{]}_i$, whereas simultaneously camera 2 collects the dye-free signal visualizing the mechanical contraction waves.²⁸ The main imaging lens is a 50-mm camera lens (0.95 F-stop, Navitar Inc., Ottawa, Canada); two additional 25-mm aspherical condenser lenses (focal length 20 mm, NA 0.6, Thorlabs) are placed in front of each camera to focus the image. The system is based on oblique transillumination for the multimodal imaging of cardiac waves. In Fig. 1(a), the bottom F1–F2–F3 cube combines the excitation light for $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ imaging and directs it to the sample via oblique illumination. The top F1–F2–F3 cube represents an identical module that allows for epi-illumination regime, validated here and useful for future imaging of nontransparent samples. For oblique transillumination, the excitation light from the bottom cube is reflected by a mirror oriented at 30 deg from the horizontal plane, so that the angle of incident light related to the sample is 60 deg.

Fig. 1

Portable low-cost all-optical macroscopic system. (a) Schematics of the imaging system generated using SolidWorks. Notations include the sample (S), CMOS cameras Cam1 (for fluorescence voltage/calcium measurements) and Cam2 for simultaneous dye-free contraction measurements. Macroimaging lens is L1 and focusing lenses in front of the cameras are L2 and L3. Blue LED is used for optogenetic stimulation, which is focused on a spot; green LED is used for excitation of calcium-responsive optical or optogenetic sensors (Rhod4, jR-GECO); and red LED for excitation of optical voltage sensors in the NIR range, e.g., BeRST1. Angled mirror (M) is used for oblique transillumination. Filters and dichroic mirrors include: F1: 655/40; F2: 535/30; F3: 640LP dichroic; F4: 470/24; F5: 495LP dichroic; F6: 495LP dichroic (for transillumination) or 473/532/660 dichroic (for epi-illumination); F7: 473/532/660 dichroic (for transillumination) or none (for epi-illumination); F8: 595/40 + 700LP. (b) Picture of the imaging system on a $12\times 18$-inch aluminum breadboard. Total cost was $<\$\mathrm{15,000}$, including two cameras, three light sources and drivers, optical and mechanical components. (c) Examples of $V_m$ activation maps at 0.5 Hz electrical pacing and optical pacing frequency, with a 1-mm scale bar and a color bar indicating the activation time for the wavefront (time is in seconds). (d) Examples of fluorescence transients during dual mapping of $V_m$ and contraction (left) and $[{\mathrm{Ca}}^{2+}{]}_i$ and contraction (right) during 0.5 Hz electrical pacing. For the dye-free transients (red), the higher peak corresponds to contraction followed by a lower peak–relaxation.

The light sources in the system include a red LED (M660L4, Thorlabs) at 660 nm for excitation of the voltage-sensitive dye and a green LED (M530L4, Thorlabs) at 530 nm for excitation of the calcium-sensitive dye. The red and green LEDs are equipped with excitation filters: F1 (FF02-655/40-25, Semrock, Rochester, New York) and F2 (ET535/50m, Chroma, Bellows Falls, Vermont), respectively, and combined by a 640 long pass dichroic mirror F3 (FF640-FDi01-25x36, Semrock). Optogenetic stimulation is done from above the sample via a blue LED (M470L4, Thorlabs) at 470 nm, typical pulse duration 5 to 10 ms, controlled by TTL signals via an LED driver. The blue light is reflected by two 495 long pass dichroic mirrors (F5 and F6) (FF495-Di03-25x36, Semrock) to the sample; the blue LED is slightly tilted off its optical axis and light is focused to a 1.5-mm point off center of the sample.

The emission filters and dichroic mirrors in front of the cameras are as follows: F7—a 473/532/660 multiband dichroic mirror (ZT473/532/660rpc-xt-UF1, Chroma) reflected the dye-free signal to camera 2 and transmitted the fluorescence signal through a multibandpass emission filter F8: 595/40 + 700LP (Semrock) to camera 1. To perform experiments with the epi-illumination module, one can simply remove the kinematic cube which holds F6 and fill the spot with the kinematic cube insert that holds F7. All dichroic mirrors are mounted in the kinematic cubes (DFM1L, Thorlabs), which makes the system modular and switching between transillumination module and epi-illumination module measurements, simple. Measurements of light power were done with an optical power meter (PM100D, Thorlabs).

2.4.

Data Acquisition and Data Analysis

The camera data acquisition can be done through a free Basler acquisition software (Pylon). In this study, we used Norpix software (Norpix, Montreal, Canada) to run the cameras and guarantee synchronization. The reported experiments were done at 100 fps, full frame, and data were stored as sequence (.seq) files (Norpix). At the selected demagnification in this study, the camera-based spatial resolution of the system was $23.6\mu \mathrm{m}$ per pixel and the FOV was $\sim 17\times 13\mathrm{mm}$. The fluorescent records and the dye-free records were co-registered. Here we focus exclusively on the $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ records obtained under various conditions. Custom software^19,28,29 in MATLAB, MathWorks, Natick, Massachusetts, was used to preprocess the data (baseline correction, Bartlett spatial filter, and locally weighted temporal regression filter) before event detection. Activation maps were constructed based on the detected times of activation, and the analysis software quantified the following functional parameters of interest: APD at 80% repolarization (APD80), calcium transient duration (CTD) at 80% recovery (CTD80), and conduction velocity (CV) (cm/s).

Statistical analysis to quantify the effects of various experimental conditions was done using multiway ANOVA with Tukey post hoc correction, and with significance considered at $p<0.05$. The analysis was performed in GraphPad Prism, San Diego, California.

3.1.

Design of a Low-Cost Portable Macroscopic All-Optical Electrophysiology System

The integrated portable low-cost all-optical macroscope is shown in Figs. 1(a) and 1(b). The total cost of the system, including two cameras, all optical components, all light sources and drivers, and the mechanical scaffolding is less than $<\$15,000$, substantially cheaper compared to existing optical mapping solutions. Key enabling technologies for building such a system are as follows: (1) improved low-cost machine-vision CMOS cameras^28,30,31 over the last 5 years, offering sufficient sensitivity, suitable speed, spatial resolution, and spectral response to work in the demanding low-light conditions of fast fluorescence imaging in cell culture; (2) developments in LEDs, that allow easy on-off control and provide a lower-cost, stable alternative to incandescent lamps and lasers, so that they entered the optical mapping field almost 20 years ago;^32,33 (3) improved voltage-sensitive probes in the near-infrared region developed recently, that offer sufficient SNR at high speed;^27,34 and (4) development of optogenetic actuation tools, which offer optical control of cell function.^20,35

The all-optical macroscopic system reported here includes capabilities for optogenetic actuation and multiparametric optical mapping of $V_m$, $[{\mathrm{Ca}}^{2+}{]}_i$, and mechanical waves. The latter is made possible by interferometric dye-free (label-free) imaging with oblique transillumination, as shown by Burton et al.,²¹ expanding upon a slightly different implementations reported earlier by Lee’s group using white light and a pinhole.^36,37 The use of oblique illumination with LEDs or lasers creates local dye-free signal during small mechanical motions. This signal was shown to depend on the light source coherence and likely represents an interference based detection mechanism.²¹

Ongoing work in our group²⁸ provides a rigorous look at the excitation–contraction coupling, made possible by such simultaneous electrical and mechanical mapping. The focus of this study was to demonstrate an all-optical macroscope with low-cost components and to provide detailed quantitative analysis of functional parameters from $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ mapping under various conditions of optogenetic treatment, illumination power, electrical versus optical pacing, and experimental media. Example activation maps under optical and electrical pacing and example traces are shown in Fig. 1(c).

3.2.

Quantification of the Effects of Illumination Power on Key Electrophysiological Parameters in All-Optical Electrophysiology

Spectrally compatible optical actuators and sensors have been combined in various studies of cardiac function over the last decade.^{17,18,20–22,25,38–40} Yet, a detailed study of the effects of irradiance in the presence of multiple optical probes is missing. In Fig. 2(a), we show the spectra for spectrally compatible optogenetic actuator (ChR2) and optical sensors—Rhod-4 (calcium indicator) and Berst1 (voltage indicator). With proper selection of filters and light intensity, these have been demonstrated previously to work well together.¹⁸ However, the ChR2 activation spectrum does extend slightly into the green region, as pointed by an arrow in Fig. 2(a), so ChR2 may potentially be engaged when sufficiently strong green light is delivered.

Fig. 2

Quantifying the effect of illumination power for all-optical electrophysiology experiments in human iPSC-CMs. (a) Excitation/activation spectra of ChR2, Rhod-4AM, and BeRST1 with LED light sources filtered bandwidths shown. ChR2 spectrum has a small overlap with the green LED bandwidth (see arrow). (b) Example voltage $V_m$ (pink) and calcium $[{\mathrm{Ca}}^{2+}{]}_i$ (teal) fluorescence transients 0.5 Hz electrical and optical pacing. Varying illumination power (in $\mathrm{mW}/{\mathrm{mm}}^2$) and keeping 0.5 Hz pacing, the following parameters are shown: (c) APD80 and CTD80 of no-ChR2 samples, $\sim 0.750$ and 1.40 s, respectively. (d) APD80 and CTD80 in response to 0.5 Hz electrical or optical pacing from ChR2 samples show similar behavior to the control no-ChR2 samples. (e) Constant CV from $V_m$ or $[{\mathrm{Ca}}^{2+}{]}_i$ measurements for no-ChR2 samples for the tested illumination powers. (f) The CVs for ChR2 samples in response to electrical and optical pacing (often completely coinciding). The CV obtained from $[{\mathrm{Ca}}^{2+}{]}_i$ mapping decreases significantly as a function of the illumination power, with a slope of $-16.13$, and the CV of $V_m$ shows a very subtle decrease as illumination power increases. (g) Representative activation maps of ChR2-samples as in (f) for $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ under different illumination power, with a 1-mm scale bar and a color bar indicating the wavefront activation times (s). Measurements were done at room temperature in Tyrode’s solution; stimulation pulses were 5 ms; optical stimulation was at $1.8{\mathrm{mW}/\mathrm{mm}}^2$. The number of samples used in each test are indicated as $n$. This applies for the rest of the figures.

Using the designed system, we performed a systematic study to quantify the effects of irradiance (for the excitation light in voltage and calcium imaging) on key electrophysiological parameters—APD, CTD, and CV—in all-optical electrophysiology. Example $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ traces indicate equivalency of electrical and optogenetic stimulation in Fig. 2(b), as suggested in earlier studies. Results in Figs. 2(c) and 2(d) indicate that varying red or green light power has almost no appreciable impact on APD and CTD in control or ChR2-expressing samples, as seen in prior studies.^18,19 However, the new findings here relate to the role of unintended ChR2 engagement during optical mapping and its effects on conduction velocity [Figs. 2(e) and 2(f)]. Although excitation light power did not affect CV in a significant way in control (no ChR2) samples, for optogenetically modified samples, higher light levels significantly decreased CV in green-light obtained calcium maps only, and not in red-light obtained voltage maps [Figs. 2(f) and 2(g)]. The effect was quite strong, such that at reasonable light levels of $0.3{\mathrm{mW}/\mathrm{mm}}^2$ for green light, conduction could be slowed by 50% when ChR2 was present. At higher light levels, engaging ChR2 by green light through the spectral cross-talk can completely block wave propagation. ANOVA analysis (Table 1) corroborates these results. Thus safe optical mapping of calcium using green-light excitable sensors, without side-effects on conduction in the presence of an optogenetic actuator, e.g., ChR2, is only possible at very low irradiance levels, $<0.05\mathrm{mW}/{\mathrm{mm}}^2$. Considering the quality of calcium sensors, these light levels still produce good SNR.

Table 1

P-values with two-way ANOVA analysis on illumination power and ChR2 infection for Figs. 2(c)–2(f).

3.3.

Impact of Experimental Conditions on Restitution and Conduction Properties for V_m and [Ca²⁺]_i

Most functional optical mapping measurements in Langendorff hearts and in cardiac cell culture are done in Tyrode’s solution. For cultured human iPSC-CMs, optical measurements can be performed in culture medium especially when done within the incubator, although that is not the norm. We deployed the all-optical macroscope to quantify differences in electrophysiological parameters when the samples were studied in Tyrode’s solution (at room temperature) versus in culture medium containing phenol-red, serum, and fatty acids (at elevated temperature of about 33°C to 35°C). Figure 3 shows the results when samples were subjected to different pacing frequencies using electrical or optical stimulation. Excitation light was kept low for calcium mapping (green LED at $<0.05{\mathrm{mW}/\mathrm{mm}}^2$) and the red LED for voltage mapping was at $\sim 0.15{\mathrm{mW}/\mathrm{mm}}^2$. In all cases, in Tyrode’s and in culture medium, samples exhibited frequency adaptation (restitution), i.e., shortening in APD80, CTD80 and decrease in CV at higher pacing frequencies. CTD80 values were not influenced by the experimental conditions, whereas APD80 showed higher dispersion at lower pacing rates in culture medium at elevated temperature. The most dramatic effect was on conduction velocity—seen both for voltage and calcium CV, under electrical and optical pacing. Performing the experiments in culture medium at elevated temperature yielded approximately twofold higher CVs compared to experiments in standard Tyrode’s solution at room temperature, especially at lower pacing frequencies [Figs. 3(c) and 3(f)]. Table 2 shows the statistical ANOVA analysis of the data and indicates that in addition to the expected restitution responses, CV is significantly affected by the experimental medium and by the combination of pacing frequency and the experimental medium. Likely explanation for the increase in CV for the warmer culture medium is the strong temperature dependence of gap junctional conductance on temperature.^41,42 The use of phenol-red containing medium required adjustment (increase) in blue light for the optogenetic stimulation due to its delivery from the top of the sample, and it led to some decrease in SNR for voltage imaging, nevertheless, it was possible to obtain maps and quantification for all parameters of interest.

Fig. 3

Quantifying cardiac restitution properties of human iPSC-CMs in different experimental conditions. Top: $V_m$ results and bottom: $[{\mathrm{Ca}}^{2+}{]}_i$ results. (a), (b) APD80 and CTD80 decrease with increasing pacing frequency (in Hz) as expected from adaptive restitution behavior. APD80 has a larger spread of values for lower pacing rates in culture medium at elevated temperature (33°C to 35°C) compared to Tyrode’s solution at room temperature, whereas CTD80 is not very different for the two experimental conditions at any given pacing frequency. (c), (d) The CVs based on $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ optical mapping also show frequency-dependent decrease (restitution). Electrical and optical pacing data show similar responses. The CVs in culture medium at elevated temperature are substantially higher (almost 2× higher) compared to those in standard Tyrode’s solution at room temperature and this is true for voltage and calcium mapping under optical or electrical stimulation. (e), (f) Activation maps of representative maps as in (c) and (d), with a 1-mm scale bar and a color bar showing activation times (s).

Table 2

P-values with two-way ANOVA analysis on pacing frequency and experimental conditions for Figs. 3(a)–3(d).

3.4.

Quantifying the V_m−[Ca²⁺]_i Interrelations under Optical versus Electrical Pacing

Although the equivalency of electrical and optical stimulation has been established when short pulses are used,⁴³ in Fig. 4, we present detailed comparison of APD80, CTD80, and CV in Tyrode’s and in culture medium. The correspondence of responses for all electrophysiological parameters of interest is very high (close to the identity line linking responses to electrical versus optical stimulation). Slight divergence from the identity line is seen for high CVs, both for voltage and calcium, most likely due to the different geometries of the stimulus origin site between electrical and optical stimulation in our setup.

Fig. 4

Comparison of electrophysiological responses in human iPSC-CMs to electrical versus optogenetic pacing. (a) Fluorescence global transients for voltage and calcium at 1.0 Hz electrical or optical pacing and (b) their corresponding activation maps in culture medium or Tyrode’s solution. (c) The APD80 and CTD80s in response to electrical and optical pacing are highly correlated ($R^2=0.97$) and lie close to the identity line (slope of 0.96). (d) CVs for voltage or calcium fall on a line with a slope of 0.95 (electrical versus optical pacing) and there is some divergence from the identity line for faster conduction in culture medium. Lightly dashed line depicts the identity line ($y=x$).

As the better SNR calcium probes are often favored in optical mapping of excitation in cell culture, it is interesting to quantify voltage–calcium relationships. Figure 5 presents APD80 versus CTD80 in different experimental solutions and under electrical and optical pacing. As expected, APD80 is shorter than CTD80, with tendency of APD80 and CTD80 converging at high pacing rates [Fig. 5(a)]. CVs obtained from voltage and calcium optical maps correlate well ($R^2=0.86$) and fall close to the identity line (slope 1.04), regardless of the experimental medium and the mode of pacing (electrical versus optical) [Fig. 5(b)]. Therefore, under controlled paced conditions, voltage, and calcium maps provide similar information.

Fig. 5

Voltage-calcium relationships in human iPSC-CMs. (a) APD80 versus CTD80 in culture medium and in Tyrode’s solution. For shorter APD80 and CTD80 ($<0.6\mathrm{s}$, at higher pacing frequencies), the two values converge; at longer APD80 and CTD80 values, the responses are strictly under the identity line (shorter APD80 values than CTD80 values, where this relationship is exaggerated when measurements are in Tyrode’s solution. (b) CV based on voltage versus calcium optical mapping—the fitted line has a slope of 1.04, close to the identity line ($y=x$, lightly dashed), indicating that CVs extracted from $V_m$ or $[{\mathrm{Ca}}^{2+}{]}_i$ mapping are close to each other; some deviations are seen at high conduction velocities in culture medium.

3.5.

Mapping the iPSC-CMs Response to Cellular Uncoupling Reagents in Different Sample Formats

The utility of the designed system was illustrated in several applications. It was applied to detect changes in cardiac electrophysiology in response to cell uncoupling drugs. In Figs. 6(a)–6(c), the responses to 0.5 mM heptanol are shown (before and after drug application), under different pacing conditions, in Tyrode’s solution. As expected, at this dose, heptanol did not affect APD80 and CTD80 [Fig. 6(a)], however, it decreased CV both obtained through voltage mapping and calcium mapping [Figs. 6(b) and 6(c)]. ANOVA results (Table 3) confirm significant effects of both pacing frequency and heptanol on CV but not on APD80 and CTD80.

Fig. 6

Applications and extensions of the all-optical mapping system. (a)–(c) Mapping changes in electrophysiological parameters (voltage and calcium) in response to treatment with a cellular uncoupling agent (30 min of 0.5 mM heptanol) under different pacing frequencies. (a) Heptanol had no effect on APD80 and CTD80 regardless of pacing frequency under optical or electrical stimulation. (b), (c) CVs for both $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ were significantly reduced upon heptanol treatment at each pacing frequency, for both electrical pacing and optical pacing ($p<0.0001$ for heptanol effects based on ANOVA). (d) Measurements of responses to treatment with TSA, 100 nM for 24 h, in a 96-well plate, where four wells are imaged simultaneously (E3 and F3: control samples and E4 and F4: TSA-treated). Calcium activation maps are shown, where TSA appears to reduce CV; TSA induced spiral waves in a very small spatial scale in E4. (e)–(i) Validation of the extension of the mapping system to an epi-illumination-based imaging. (e) Schematic illustrations of the oblique transillumination versus epi-illumination mode. (f), (g) APD80 and CTD80 results under different pacing frequencies for control no-ChR2 samples and ChR2 samples, respectively. Orange represents transillumination and purple—epi-illumination. Note that there is overlap of data points from the two regimes. (h), (i) CV results for voltage and calcium under different pacing conditions for no-ChR2 and for ChR2 samples, respectively. For each studied condition, restitution properties are visible. The results from transillumination module and epi-illumination module almost completely coincide, indicating the epi-illumination module was functional and can be deployed to measure thicker nontransparent tissue samples.

Table 3

P-values with two-way ANOVA analysis on pacing frequency and drug treatment for Figs. 6(a)–6(c).

Although most experiments were done in glass-bottom 35 mm dishes in Fig. 6(d), we demonstrate that in the current configuration and for the FOV $\sim 19\times 15\mathrm{mm}$ ($26.9\mu \mathrm{m}$ per pixel), the system is compatible with other formats and can map four wells (from a 96-well plate) simultaneously. A different drug with effects on coupling and overall electrophysiology—an HDAC inhibitor, TSA was applied for 24 h at 100 nM concentration. Conduction slowing was documented in the TSA-treated samples, as seen in the activation maps. In one of the shown samples (E4), the conduction was slowed down sufficiently to sustain a rotating wave within a very small space (each well is 7 mm in diameter).

3.6.

Validation of Extensibility to Epi-Illumination

The oblique transillumination mode of imaging in the presented system works well with cultured iPSC-CMs, i.e., transparent monolayer samples. For the study of thicker, nontransparent samples in the future, we explored the possibility to reconfigure the system to an epi-fluorescent mode. Due to the modular design, this reconfiguring was straight-forward. Figure 6(e) illustrates the difference between the oblique transillumination and the epi-illumination configuration. We tested out how these two systems compare in terms of measured electrophysiological parameters from the same transparent iPSC-CMs monolayer. We quantified APD80, CTD80, and CV for both no ChR2 and ChR2 expressing samples at various pacing frequencies, using both modules, and compared the results. Control (no ChR2) results for APD80, CTD80, and CV are shown in Figs. 6(f) and 6(h); and results for ChR2-expressing samples are shown in Figs. 6(g) and 6(i). All measurements in Figs. 6(f)–6(i) show the characteristic restitution for both electrical and optical pacing. Note that in Figs. 6(h) and 6(i), the CVs of $V_m$ and $[{\mathrm{Ca}}^{2+}{]}_i$ are not distinguishable from each other due to them having similar values. Without further optimization, we found that the epi-illumination mode tended to produce noisier $V_m$ signals, which led to the corresponding APD80s to have larger uncertainties, especially at higher pacing frequencies. Overall, the epi-illumination results agree well with the transillumination, which verified that the epi-illumination configuration was functional and can be deployed in the future to study nontransparent samples, such as heart slices or whole hearts. Because of the thicker layer of cells from which photons are emitted, signals from heart slices or a whole heart are expected to be much stronger and easier to measure compared to the monolayer iPSC-CMs culture.