Conventional rigid 2D substrates cause complex contractile signals in monolayers of human induced pluripotent stem cell-derived cardiomyocytes

Abstract

Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CM) in monolayers interact mechanically via cell-cell and cell-substrate adhesion. Spatiotemporal features of contraction were analysed in hiPSC-CM monolayers (1) attached to glass or plastic (Young's modulus (E) >1 GPa), (2) detached (substrate-free) and (3) attached to a flexible collagen hydrogel (E = 22 kPa). The effects of isoprenaline on contraction were compared between rigid and flexible substrates. To clarify the underlying mechanisms, further gene expression and computational studies were performed. HiPSC-CM monolayers exhibited multiphasic contractile profiles on rigid surfaces in contrast to hydrogels, substrate-free cultures or single cells where only simple twitch-like time-courses were observed. Isoprenaline did not change the contraction profile on either surface, but its lusitropic and chronotropic effects were greater in hydrogel compared with glass. There was no significant difference between stiff and flexible substrates in regard to expression of the stress-activated genes NPPA and NPPB. A computational model of cell clusters demonstrated similar complex contractile interactions on stiff substrates as a consequence of cell-to-cell functional heterogeneity. Rigid biomaterial surfaces give rise to unphysiological, multiphasic contractions in hiPSC-CM monolayers. Flexible substrates are necessary for normal twitch-like contractility kinetics and interpretation of inotropic interventions. KEY POINTS: Spatiotemporal contractility analysis of human induced pluripotent stem cell-derived cardiomyocyte (hiPSC-CM) monolayers seeded on conventional, rigid surfaces (glass or plastic) revealed the presence of multiphasic contraction patterns across the monolayer with a high variability, despite action potentials recorded in the same areas being identical. These multiphasic patterns are not present in single cells, in detached monolayers or in monolayers seeded on soft substrates such as a hydrogel, where only 'twitch'-like transients are observed. HiPSC-CM monolayers that display a high percentage of regions with multiphasic contraction have significantly increased contractile duration and a decreased lusotropic drug response. There is no indication that the multiphasic contraction patterns are associated with significant activation of the stress-activated NPPA or NPPB signalling pathways. A computational model of cell clusters supports the biological findings that the rigid surface and the differential cell-substrate adhesion underly multiphasic contractile behaviour of hiPSC-CMs.

Keywords: cardiac physiology; mathematical model; pharmaceutical assay; recombinant collagen polymer; substrate rigidity.

Figure 1. Explanation of physiology parameters for (A) electrophysiology, (B) calcium, and (C) contraction

T_Rise = time from 10 to 90% of amplitude. APD₉₀ = action potential duration at 90% of the amplitude. CaT₅₀ = calcium transient duration at 50% of the amplitude. CaT₉₀ = calcium transient duration at 90% of the amplitude. T_Contraction = contraction time. T_Relaxation = relaxation time. Ampl. = amplitude. CD₅₀ = contraction duration at 50% of the amplitude.

Figure 2. Mathematical model specifics

A, calcium trace from hiPSC‐CMs (Cal520 signal) (continuous line) vs. the calcium profile used in the mathematical modelling (broken line). [Correction made on 20 January 2022, after first online publication: ‘Cal590’ has been corrected to ‘Cal520’ in the preceding sentence.] B, the motion trace from a single contracting unit in the uncoupled system (dotted line) from the model from Rice et al. (2008) compared with the motion traces from a single contracting unit, coupled between two springs with k = 0, from the model directly from Timmerman et al. (2019) (solid blue line). C, a schematic of the force components in the contractile units, based on Fig. 1D from Rice et al (2008). [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 3. Assessment of the morphology of small cardiac patches

A, method of creating 2D cardiac sheets using silicone stencils. (1) Place the stencils of various sizes (1, 2 or 3 mm) inside the well and seed hiPSC‐CMs in the stencil in desired cell density (normal (1×), double (2×) or quadruple (4×) density). (2) Culture the hiPSC‐CM for two days. (3) Remove the stencil, add 200 μL maintenance medium to every well and culture the patches for two more days. Assessment is done on day 4 after seeding. (4) A well of a 96‐well plate (TCP) was seeded with the normal cell density and was used as control (Ctrl). B, F‐actin (green) and DAPI (blue) staining for hiPSC‐CM micro‐patches with different sizes and cell densities. Scale bar represents 1 mm. C, CellProfiler data showing the relative patch size compared with the theoretical value (TV). D, CellProfiler data showing the area coverage by hiPSC‐CM calculated by dividing the cell area by the patch area. E, CellProfiler data showing the remaining percentage of nuclei, which was calculated as the number of cells counted divided by the number of cells seeded. Results from Cor.4U (NCardia) hiPSC‐CM only. Statistical analysis was done using a nested one‐way ANOVA using a Dunnet's post hoc test and groups were tested against Ctrl. n _Plating = 3; n _Recordings = 6 to 10 per plate. Absolute values of the P values for all comparisons (significant and non‐significant) are listed in the online Statistical Supplement. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 4. Differences in electrophysiology and contractility between cell‐seeding densities and commercial hiPSC‐CM cell lines

A, electrophysiology of hiPSC‐CMs showing spontaneous beating frequency (Aa) T_Rise (Ab) and APD₉₀ (Ac) for both NCardia (black) and ICell² (grey) hiPSC‐CM cell lines. B, contractile behaviour of hiPSC‐CMs shown as contraction time (Ba), relaxation time (Bb) and CD₅₀ (Bc) for both Cor.4U (dark grey bars) and ICell² (light grey bars) hiPSC‐CM cell lines. n = 5–19 (ICell²) and 9–30 (Cor.4U), N = 3 experiments for both cell lines. Cell densities were compared with Ctrl for each cell type using mixed effect two‐way ANOVA with Dunnett's post hoc test (statistical significance indicated with ˆ). Differences between NCardia and ICell² were tested between a two‐way ANOVA with Sidak's post hoc test (statistical significance indicated with ^*). Recordings were made on day 2 after stencil removal. Absolute values of the P values for all comparisons (significant and non‐significant) are listed in the online Statistical Supplement.

Figure 5. Significant variation in contractility and negligible variation in voltage across a hiPSC‐CM monolayer seeded on tissue culture plastic

A, example of three different locations within the 2D micro‐patch indicated with blue, red and green squares. Each location is approximately 300 × 300 μm. The white dashed line indicates the outer line of the patch. B, action potentials (AP, top row, panels a–c) and contractility traces (CT, bottom row, panels d–f) recorded on three different locations within one well (blue, green and red inserts in panel A). APs and CTs with the same colour are recorded on the same location. C, overlay of the three APs. D, overlay of the three CTs. E, a 300 × 300 μm field of view, similar to one of the locations in panel A, was subdivided into 3 × 3 grid squares of 100 × 100 μm each. The MM algorithm was applied to every grid square. F, all traces from each grid square (nine in total) were overlapped. G, all traces from each grid square were placed in their corresponding location shown in panel E. H, heatmaps indicating (a) the number of peaks, (b) the start time (T_Start) and (c) the contractile duration at 50% of the amplitude (CD₅₀) (scalebars are 0–5 peaks, 0–40 ms and 0–600 ms, respectively). I, the contractility trace taken from the whole area (300 × 300 μm). [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 6. Cell density does not modulate the complex contractile behaviour on fixed substrates

HiPSC‐CMs seeded in 1× (A), 2× (B) or 4× (C) the cell density suggested by the manufacturer. For each cell density, example traces are shown from three distinct regions of the same micro‐patch. These include (1) the average contractile trace recorded from the whole area (top left panel), (2) a heatmap showing the location of grid squares with either 1 (blue) or >1 peak (light green) (top right), (3) the traces with 1 peak (bottom left) and (4) the traces with >1 peak (bottom left). Recordings are from Cor.4U hiPSC‐CMs (NCardia) on day 2 after stencil removal. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 7. Spatiotemporal analysis of the entire hiPSC‐CM monolayer

A, video frames were subdivided into 30 × 30 grid squares. Each grid square (∼100 × 100 μm) was analysed using the MM algorithm (33), resulting in 900 individual data points. B, various contractile profiles are seen at different locations within one monolayer as represented by the coloured traces (panels A and B). We observed transients containing one peak (Ba and Bb), two peaks (Bc and Bd) and more than two peaks (Be and Bf). C, all traces obtained from one video. D, all traces obtained from one video with one peak. E, all traces obtained from one video with two peaks. F, all traces obtained from one video with more than two peaks. G, values for various measurements are plotted in a heatmap. (a) The number of peaks (scalebar 1–3 peaks), (b) the contraction duration at 50% of the amplitude (CD₅₀) (scalebar 0–1200 ms), and (c) the start times (T_Start) (scalebar 0–600 ms). H, an example of the distribution of 900 data points following spatiotemporal analysis. From this distribution, the 10th–90th percentile difference (IP₉₀) was calculated as illustrated to obtain a more realistic average within one group and thus allowing us to compare different groups. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 8. Contractile behaviour of hiPSC‐CMs before detachment, right after detachment and during reattachment

A, animation explaining the experimental procedure of cell culture on a thermosensitive dish (C), cooling the plate down to 25°C, allowing the cell sheet to detach while maintaining cell–cell adhesions (D), transferring the cell sheet to a new culture dish (T) and allowing the cells to reattach to the new culture dish (R). B, spontaneous beating frequency of hiPSC‐CMs. C, the contractile complexity, expressed as the average percentage of grid squares with either single‐ (black dots) or multiple‐peaked (white dots) transients. D, the IP₉₀ for CD₅₀, describing the variation in contractile behaviour. E, the CD₅₀ taken from the whole area video recording. F, the IP₉₀ for the T_Start, describing the level of synchronized contraction within the cell sheet. G, the average percentage of grid squares containing beating cells, indicating the size of the cell area. ϕ = quiescent patches were excluded from analysis at those time points. Time points were statistically tested against D0 37°C using a nested one‐way ANOVA with Dunnett's post hoc test. H, voltage recordings of reattached (D7) monolayers (Cor.4U (NCardia)) compared with data presented in Fig. 4A as these were from comparable experimental groups. Recordings were made on days 2 (TCP) and 7 after stencil removal. (a) Spontaneous beating frequency. (b) T_Rise. (c) APD₉₀. Groups were compared using a nested, two‐sided t test. n _Experiments = 3, n _Samples = 16. A P value <0.05 is considered significant. ^*= P < 0.05, ^**= P < 0.01, ^***= P < 0.001. Absolute values of the P values for all comparisons (significant and non‐significant) are listed in the online Statistical Supplement. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 9. Investigation of the contractile behaviour of micro‐patches seeded on recombinant collagen‐like peptide (RCP) hydrogels

A, animation describing the methodology of the hiPSC‐CM cell‐seeding procedure on the hydrogel. A glass substrate served as a control. Created with BioRender.com (B) the average beating frequency of hiPSC‐CMs. C, the average percentage of grid squares with a single‐peaked transient. D, the IP₉₀ of CD₅₀ describing the variation in contractile behaviour. E, the CD₅₀ taken from the whole area video recording. F, the IP₉₀ of T_Start, describing the synchronicity of the contractions. G, the percentage of grid squares per recording containing beating cells and thus analysable, indicating the size of the cell area. ICell² (CDI) hiPSC‐CMs were used for these experiments. H, data obtained from voltage recordings made on day 7 after stencil removal. I, data obtained from calcium recordings on day 7 after stencil removal. G = glass. H = hydrogel. n _Experiments = 3, with 2–6 samples per experiment per time point for each experimental group. Groups were compared with the control value at the same time point using a two‐way ANOVA with Sidak's post hoc test (B–G) or a nested, two‐sided t test (H and I), to test statistical significance. A P value < 0.05 was considered significant and is indicated with ^*. Absolute values of the P values for all comparisons (significant and non‐significant) are listed in the online Statistical Supplement. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 10. Single cells only show single‐peaked contraction profiles

Single cell cultures of iPSC‐CMs (ICell² (CDI)) seeded on either glass (A and B) or hydrogel (C and D) substrates. The nine contractility traces with normalized amplitude (A and C) correspond to the bright‐field images of single cells (B and D). E, the spontaneous beating frequency; F, the contraction time; G, the relaxation time; and H, the CD₅₀ of cells seeded on either glass or hydrogel. Recordings taken on day 4 or 5 after stencil removal. Scale bar indicates 50 μm. Statistical analysis was done using a nested unpaired t test. A P value < 0.05 is indicated with ^*. N_glass = 82; N_hydrogel = 48; N_experiments = 3. Absolute values of the P values for all comparisons (significant and non‐significant) are listed in the online Statistical Supplement.

Figure 11. The effect of isoprenaline on the time‐course complexity of hiPSC‐CM seeded on different substrate stiffnesses

HiPSC‐CMs (ICell² (CDI)) were seeded on either a glass substrate or a hydrogel and incubated with either isoprenaline (ISO, 300 nM) or vehicle (culture medium) for 5 min, where after contractility recordings were made. A, the spontaneous beating frequency of hiPSC‐CMs. B and C, contractility traces (900) from the spatial analysis of one representative glass control sample (B) and hydrogel (C) at baseline and after ISO incubation. Classification of the traces is based on the number of peaks (1 vs. >1). D and E, the distribution of grid squares containing 1‐peaked (dark blue) or multiple‐peaked transients (light green) at baseline and after ISO incubation for the glass control (D) and hydrogel substrate (E). F, the change in the percentage of grid squares containing multiple‐peaked transients compared with the baseline. G, the change in average CD₅₀ values as compared with the baseline. H, the change in amplitude as percentage of the baseline values. All recordings were made on day 5 after stencil removal. I, relative gene expression for stress‐activated genes NPPA and NPPB. Endothelin‐1 (ET‐1) treatment served as a positive control. n _Experiments = 3, n _Samples = 5 to 8 per group. Groups were compared with any other group using a nested one‐way ANOVA with Sidak's post hoc test (A, F, G and H) or a one‐way ANOVA with Tukey's post hoc test (I). A P value < 0.05 was considered significant and is indicated with ^*. ^* = P < 0.05, ^** = P < 0.01, ^*** = P < 0.001, ^**** = P < 0.0001. Absolute values of the P values for all comparisons (significant and non‐significant) are listed in the online Statistical Supplement. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 12. The effect of variation in intracellular calcium and number of contractile units on the computed motion by the model presented

A, average motion from 15 sets of simulations with N = 3, 5 and 10 contractile units, with k = 0.01 and randomly assigned maximum calcium concentrations within the ranges: (a) a minimal range of 95–100%, (b) a low level of variation, 75–105%, (c) a medium level of variation, 50–115% and (d) the highest level of variation, 25–125%. The red box indicates the set seen in the main body of the text; specifically, the case with N = 5 and a medium level of variation of the maximum calcium. The average of the 15 sets of averages is plotted with the thick black line and are plotted in B. C, average motion from 15 sets of simulations with N = 3, 5 and 10 contractile units, with k = 1 and randomly assigned maximum calcium concentrations within the ranges: (a) a minimal range of 95–100%, (b) a low level of variation, 75–105%, (c) a medium level of variation, 50–115% and (d) the highest level of variation, 25–125%. The red box indicates the set seen in the main body of the text; specifically, the case with N = 5 and a medium level of variation of the maximum calcium. The average of the 15 sets of averages is plotted with the thick black line and is plotted in D. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 13. Computational model showing the complex contractile behaviour as seen in biological samples

A, diagrams showing the model for the cell arrangement on the stiff (a) and soft (b) substrates. The stiff substrate is represented by a spring at each end with value of the spring constant equal to k = 1 and soft substrate is represented by a pair of springs at each end with value of the effective spring constant equal to k = 0.01. B, graphs showing the output per unit (cell) for intracellular calcium (μM) (a and b), cell length (μm) (c and d) and the displacement of the ends of a cell from their initial positions (motion) (μm) (e and f) of hiPSC‐CMs seeded on stiff (a, c, e) and soft (b, d, f) substrates. Colour schemes of panel (B) correspond to the diagram in panel (A). The black line in panels Be and Bf represents the average position (Av. Pos) of all the positions (1–6) combined. C, the averaged model output per run for five different runs of the same model, but different randomized calcium distribution for each cell. Note that run 2 in both panels Ca and Cb correspond to the average position of panels Be and Bf, respectively. Pos denotes position. L denotes cell length or unit length. [Colour figure can be viewed at wileyonlinelibrary.com]