Aim and Design
This study aimed to find an optimal way of producing atrial cardiomyocytes derived from iPSCs, to this end two existing methods from the literature were compared to a novel protocol described within. Utilizing two iPSC lines, 15C1 and 100C1, the three methods were compared to each other and the ventricular base protocol for, optical action potentials, field potential durations, and response to Ikach activator carbamylcholine. Further characterization was performing on the protocol described within, including an expanded set of iPSC lines, and the ability to form arrhythmic rotor activity.
iPSC Cell Culture
hiPSC lines 15-C1 (16 y/o Male), 100-C1 (unknown aged Female), 8-C1 (unknown aged Male), 477-C1 (unknown aged Male), 273C1 (42 y/o Female) with no observed karyotypic abnormalities, were generated by the Stanford Cardiovascular Institute Biobank from healthy patients as previously described (Burridge et al. 2016). Line VG1-C as characterised in Holliday et al., 2018, was a generous gift from Christopher Semsarian, Centenary Institute. Cell lines were maintained with ReLeSR™ (05872, STEMCELL Technologies) using mTeSR Plus (100–0276, STEMCELL Technologies) on hESC pre-screened Matrigel (354277, Corning).
Cardiomyocyte Differentiation & Dissociation
Cardiomyocytes were produced using published protocols (29, 30). In short, hiPSCs were TryPLE (12605010, ThermoFisher) dissociated and seeded at 105k cells / cm2 in 10 µM ROCK inhibitor Y-27632 (72304, STEMCELL Technologies). Upon reaching 60–80% confluence, hiPSCs were preconditioned at day − 1 in RPMI B27 minus insulin (A1895601, ThermoFisher) with 2 ng/mL BMP4 (PHC9531, ThermoFisher), 1% GlutaMAX (35050079, ThermoFisher), 200 µM L-ascorbic acid (A15613.22, ThermoFisher) and 1:100 Matrigel for 16 hours. Subsequently at day 0, hiPSCs were further induced to mesoderm in RPMI B27 minus insulin supplemented with 8 ng/mL Activin A (PHC9564, ThermoFisher) and 10 ng/mL BMP4 with 1% GlutaMAX. Day 2 cardiac mesoderm specification is induced with RPMI B27 minus insulin, 200 µM L-ascorbic acid, 10 µM KY02111 (4731, TOCRIS) and 10 µM XAV939 (X3004, Sigma) for 48 hours before switching to the same formulation in RPMI B27 plus insulin for a further 48 hours. Day 6 onwards, differentiations were medium exchanged every 48 hours for RPMI B27 plus insulin (1504001, ThermoFisher) supplemented with 200 µM L-ascorbic acid, until ready for dissociation day 12–15.
For atrial conditions the above protocol was adjusted as follows. Atrial (D1RA) differentiation included the addition of 1 µM retinoic acid (R2625, Sigma) at day 1 of differentiation (midway through mesoderm formation), followed by the addition of 1 µM retinoic acid at day 2 of differentiation (cardiac mesoderm formation). The Atrial Standard condition had the addition of 1 µM retinoic acid at day 2 of differentiation (cardiac mesoderm formation). Atrial (Low GFs) had reduced concentrations of Activin A and BMP4 to 2 / 3 ng /mL respectively during mesoderm formation, and 1 µM retinoic acid at day 2 of differentiation.
Cardiac dissociation was carried with 0.2% collagenase type I (17018029, ThermoFisher), in PBS supplemented with 20% fetal bovine serum (FBS) (SH30084.04, Cytiva Life Sciences), for 45 minutes at 37°C, followed by centrifugation at 300 g for 3 minutes. Cardiomyocytes were resuspended in 0.25% Trypsin with EDTA (25200056, ThermoFisher) for 10 minutes at 37°C, neutralized in FBS containing medium and filtered through 70 µm cell strainer (352350, Corning) centrifuged at 300 g for 3 minutes, and resuspended in B27 plus insulin supplemented with 10% FBS. After 48 hours of seeding, medium was exchanged for B27 plus insulin, without FBS.
Immunocytochemistry
Cells were fixed in 4% paraformaldehyde (PFA) (C004, ProSciTech) for 15 minutes at room temperature before washing x3 in PBS, and permeabilizing/blocking in blocking buffer (PBS with 0.1% Triton-X100 (X100, Sigma) and 4% goat serum (G9023, Sigma)) for 1 hour, incubation with primary antibodies diluted in blocking buffer was carried out overnight at 4°C. Primary stained samples were washed in PBS x3 before addition of secondary antibodies in blocking buffer for 1 hour at room temperature, exchanged to PBS containing DAPI (D9542, Sigma) for 10 minutes, washed in PBS x3 and stored in PBS. Antibodies and concentrations provided in Supplementary Table 1.
Imaging was carried out using the either Opera Phenix (PerkinElmer) spinning disk confocal microscope (equipped with 2x 16-bit sCMOS cameras), or LSM900 (Zeiss). Identification of objects, fluorescent intensity calculations and percentage positivity was carried out using Harmony software (PerkinElmer).
Cardiomyocyte And Bead Size Flow Cytometry
Live hiPSC-derived cardiomyocytes were determined by staining the cells with a mix (Live/Dead mix) of Zombie Yellow™ Fixable Viability Kit (1:1000, 423103, BioLegend), Propidium iodide (1µg/mL, P4170-10MG, Sigma-Aldrich), Calcein AM (1µM, C3100MP, LifeTechnologies), DAPI (10µg/mL, D9542-10MG, Sigma-Aldrich) in PBS. The stained samples were analyzed using a CytoFLEX Flow Cytometer (see below for details).
Cardiomyocytes were fixed with 2% PFA for 10 min at room temperature. Cells were washed once with 500 µl of BD perm/wash buffer, incubated for 20 min and then, centrifuged (500 ×g for 5 min). Cardiomyocytes were incubated with Troponin T (1:500 dilution, MS295P1, ThermoFisher) for 1 h at room temperature in BD perm/wash buffer. Cardiomyocytes were then washed three times with PBS then incubated with a secondary antibody for 1 hr (1:500 dilution, A-21037, ThermoFisher). Subsequently washed twice with PBS, cells were stained Hoechst 33342 (Hoechst 33342, bisBenzimide H 33342 trihydrochloride, B2261-25MG, Sigma-Aldrich) prepared in perm/wash buffer.
Samples were analyzed using a CytoFLEX Flow Cytometer (Beckman Coulter). 20,000 events were collected for each experiment. Data was collected using CytExpert Software. Flow cytometry grade microbeads were obtained from Spherotech (PPS-6K). 1000–2000 microbeads were recorded for each bead size using identical parameters used for cardiomyocytes. All flow cytometry data were analyzed using FlowJo software (Becton Dickinson & Company (BD)). Cell size measurements were obtained using the linear equation and formula obtained from plotting bead size against FSC-A. Supplementary Fig. 2 bead size (µm) vs FSC-A was plotted using Prism (Version 9.0.0, GraphPad Software, LCC).
Action Potential And Calcium Transient Recordings
To record the action potentials of cardiomyocytes, cells were loaded with 1X FluoVolt dye as per manufacturer's instructions (F10488, ThermoFisher) and exchanged into phenol red free RPMI (11835030, ThermoFisher), supplemented to 1 mM calcium chloride. For calcium transient recordings, Cal 520 AM dye (21130, AAT Bioquest) was incubated at 2.5 µM for 30 minutes at 37 degrees. Cardiomyocytes were placed in the Nikon Eclipse Ti2-E Inverted Microscope, fitted with an Andor Zyla sCMOS (Oxford Instruments) high speed camera, with a Nikon Plan Fluor 10X objective (NA, 0.3) used for these experiments. Data was collected at 5 ms temporal resolution from regions of 512x512 pixels. Before recording, cells were placed into the live chamber (37ºC & 5% CO2) and left to equilibrate for 30 minutes. Data was processed utilising an in-house Matlab script.
Electrical Field Potential Measurements (Mea)
Electrical recordings of field potentials from multi-layered sheets of cardiomyocytes were performed using a Maestro-APEX multi-electrode array (MEA) system (Axion Biosystems). Cultures were seeded on CytoView (M768-tMEA-48B, M384-tMEA06W, Axion Biosystems) or E-stim 48 well MEA plates (Axion Biosystems). Spontaneous recordings were taken using AxIS v2.5.1.10 software (Axion Biosystems) before processing with the CiPA tool. For drug experiments, identical golden electrodes were selected across recordings. All experiments were performed at 37°C and 5% CO2.
Reverse Transcriptase (Rt) Quantitative Pcr (Rt-qpcr)
Total mRNA was isolated using TRIzol (15596026, ThermoFisher). RNA samples were processed with miRNeasy kit (217084, Qiagen), cDNA was generated using SuperScript™ IV reverse transcriptase including ezDNase (11766050, ThermoFisher), for digestion of genomic DNA, both following manufacturer’s instructions. RT-qPCRs were performed in triplicate using Lightcycler 480 SYBR Green master mix I (04707516001, Roche) in a CFX384 Optical Reaction Module on C1000 Touch Thermal Cycler (Bio-Rad). Expression data was analyzed using 2^(-ΔΔCT) method relative to expression level of GAPDH housekeeper gene. RT-qPCR analysis was performed on age matched differentiations produced simultaneously, across two different iPSC lines. Primers sequences contained in Supplementary Table 2.
Phase Mapping And Phase Singularity Detection
Optical recordings were first processed using the Sliding Window Normalisation technique as described in (31), using a window size of 100 frames, to amplify optical signals. Further normalisation methods outlined in (32) were used for spatial and temporal de-noising. Frames were passed through a gaussian kernel filter before optical signals were filtered using a 4th order Butterworth filter with a 1 to 30Hz bandpass applied in forward and reverse mode. Signal edge tapering was achieved using an Hanning window, and signal smoothing was accomplished with sinusoidal wavelet recomposition.
Phase mapping and phase singularity detection was conducted following the transforms and calculations laid out by (32) and (33). The Hilbert transform was utilised to obtain the instantaneous phase for the optical data. Phase singularities were detected using the Double Ring Method proposed by (32) where a singularity is present when phase difference around a point of interest is greater than pi.
Statistical Analysis
For comparison of differentiation methods, data were fitted with a mixed effect model using the glmfit function in Matlab, with differentiation method assigned as a fixed categorical variable and differentiations (N) and technical replicates (n) as random variables. Estimated marginal means were derived from the generalized linear mixed effect models using the emmeans package (Hartman, 2022) with comparison between differentiation undertaken using a Wald test in input contrasts. Violin plots of data were created using Violin SuperPlots in Matlab (Kenny & Schoen, 2021). Error bars on violin plots represent estimated marginal means from the generalised linear mixed model and their standard errors. Datasets and code for statistical analysis and data visualisation are available for download from Zenodo (doi tbc). Values presented as estimated marginal means obtained by generalised linear mixed model and their standard errors.