Transient Notch Activation Converts Pluripotent Stem Cell-Derived Cardiomyocytes Towards a Purkinje Fiber Fate

Abstract

Cardiac Purkinje fibers form the most distal part of the ventricular conduction system. They coordinate contraction and play a key role in ventricular arrhythmias. While many cardiac cell types can be generated from human pluripotent stem cells, methods to generate Purkinje fiber cells remain limited, hampering our understanding of Purkinje fiber biology and conduction system defects. To identify signaling pathways involved in Purkinje fiber formation, we analyzed single cell data from murine embryonic hearts and compared Purkinje fiber cells to trabecular cardiomyocytes. This identified several genes, processes, and signaling pathways putatively involved in cardiac conduction, including Notch signaling. We next tested whether Notch activation could convert human pluripotent stem cell-derived cardiomyocytes to Purkinje fiber cells. Following Notch activation, cardiomyocytes adopted an elongated morphology and displayed altered electrophysiological properties including increases in conduction velocity, spike slope, and action potential duration, all characteristic features of Purkinje fiber cells. RNA-sequencing demonstrated that Notch-activated cardiomyocytes undergo a sequential transcriptome shift, which included upregulation of key Purkinje fiber marker genes involved in fast conduction such as SCN5A, HCN4 and ID2, and downregulation of genes involved in contractile maturation. Correspondingly, we demonstrate that Notch-induced cardiomyocytes have decreased contractile force in bioengineered tissues compared to control cardiomyocytes. We next modified existing in silico models of human pluripotent stem cell-derived cardiomyocytes using our transcriptomic data and modeled the effect of several anti-arrhythmogenic drugs on action potential and calcium transient waveforms. Our models predicted that Purkinje fiber cells respond more strongly to dofetilide and amiodarone, while cardiomyocytes are more sensitive to treatment with nifedipine. We validated these findings in vitro, demonstrating that our new cell-specific in vitro model can be utilized to better understand human Purkinje fiber physiology and its relevance to disease.

Figure 1.. Analysis of murine embryonic cardiomyocyte and Purkinje fiber single-cell signature reveals differential expression of Notch signaling.

A) Top: Schematic of experiment performed by (Goodyer et al., 2019) to obtain embryonic ventricular tissue dissected to enrich for bundle branch and Purkinje fiber cell types. B) UMAP projection of single cell sequencing data from Goodyer et al 2019 obtained from E16.5 embryonic mouse ventricular tissue. C) Feature plots for representative genes of major populations identified, including CMs (Tnnt2+/Acta2+), endothelial cell types (Pecam1+) and fibroblast populations (Col1a1+). D) UMAP projection following subclustering of CMs (top left of Figure 32B) identifies transcriptionally distinct PF cluster as well as CMs of trabecular and compact identity. E/F) Feature plots (top) and violin plots (bottom) for (E) compact/trabecular cardiomyocyte markers and (F) Purkinje fiber markers. G) Violin plot showing differentially expressed genes between PFs and compact myocardium. H) Violin plot showing differentially expressed genes between PFs and trabecular myocardium. I) GO (top), Wiki Pathway (middle) and KEGG Pathway (bottom) enrichment analysis for DE comparison between Purkinje fiber and trabecular myocardial cells demonstrating enriched terms and pathways.

Figure 2.. Transient Notch Upregulation in Pluripotent Stem Cell-Derived Cardiomyocytes Leads to Morphology Changes and Increased Expression of Purkinje Fiber Markers.

A) Schematic demonstrating the strategy used to perturb Notch signaling in hPSC-CMs through addition of 4OHT or DAPT. B) Quantification of cell length between control and Notch perturbed hPSC-CMs. C) Treatment with 4OHT leads to development of elongated morphology of hPSC-CMs by day 28. D) RNA sequencing data demonstrating increased expression of canonical Notch targets 48 hours after induction/inhibition of Notch signaling in hPSC-CMs. E) RNA sequencing data demonstrating increased expression of PF markers in 4OHT induced hPSC-CMs 28 days after Notch perturbation. F) Immunofluorescence for alpha actinin (AACT - green) co-stained with DAPI (blue) and various PF marker genes (red) 28 days after Notch perturbation demonstrates increased expression of PF markers in 4OHT treated hPSC-CMs. Data is shown as fold change relative to time-matched replicate control. * = p <0.05 according to students t-test.

Figure 3.. Transient Notch Activation in Pluripotent Stem Cell-Derived Cardiomyocytes Causes Global Transcriptomic Changes

A) Principal component analysis of bulk RNAseq data from control and Notch perturbed hPSC-CMs at d28. B) Volcano plot demonstrating results of C) Clustered heatmap demonstrating top 100 differentially expressed genes between control and 4OHT-treated hPSC-CMs. D) Chord plot demonstrating gene set enrichment analysis and GO processes that are associated with 4OHT-treated hPSC-CMs at d28. D) GO and KEGG/Wiki Pathway enrichment analysis on 4OHT-treated PSC-CMs at d28 demonstrating enrichment of processes involved in conduction and heart rhythm regulation. F) Heatmap demonstrating log2-fold change expression changes of selected genes at d28.

Figure 4.. Conversion Toward Purkinje-like Lineage Following Notch Activation Occurs Through Gradual Expression of Stage Specific Genes and Processes

A) Schematic demonstrating strategy for converting hPSC-CM toward hPSC-PF lineage and time points collected for bulk RNAseq analysis. B) UpsetR plot demonstrating shared and exclusive differentially expressed genes between time points, demonstrating that most differentially expressed genes are exclusive to particular stages during conversion. C) Venn diagram demonstrating shared and exclusive GO terms between time points analyzed through bulk RNA sequencing. D) Bubble plot demonstrating expression of GO term enrichment at particular timepoints during conversion. E) Bar plots demonstrating log2-fold change expression of selected genes across conversion. Fold changes are calculated by normalizing gene expression in 4OHT treated cells to their experimental and time-matched control.

Figure 5.. Transient Notch Activation in Pluripotent Stem Cell-Derived Cardiomyocytes Leads to Action Potential Morphology Changes

A) Quantification of changes in beat rate (manually counted) 28 days after removal of 4OHT or DAPT, as indicated. B) Representative traces of MEA recordings for hPSC-CMs in the indicated conditions. C) Quantification of beat period, conduction velocity, spike slope and spike amplitude in d28 hPSC-CMs corresponding to the indicated conditions. D) Representative traces of AP morphology (left) and of CaT morphology (right) analyzed using fluorescent dye for control and Notch perturbed hPSC-CMs. E) Quantification of APD90 and APD50 at d28, demonstrating increase in APD90 in Notch activated hPSC-CMs (left) and quantification of CaD90 and APD50 (right) at d28. F) Still images of N=2 representative dynEHT from hPSC-CMs and hPSC-PFs 4 weeks after casting. G) Force traces of hPSC-CM and hPSC-PF dynEHTs 4 weeks after casting (left) and analysis of beat per minute (BPM), diastolic force and systolic force (right) for N=5 replicates/differentiations. * = p <0.05 according to students t-test. Data consists of n=5 biological replicates, with 2–3 technical replicates per condition across each biological replicate.

Figure 6.. Computational Modeling of Pluripotent Stem Cell-Derived Cardiomyocytes and Purkinje Fibers Demonstrates Cell-Specific Sensitivity to Drug Perturbations

A) Simulated AP profile for hPSC-CMs (blue) and hPSC-PFs (orange) using a scaled computational model. B) Predicted spontaneous beating rate for hPSC-CMs (blue) and hPSC-PFs (orange). C) Predicted calcium transient profile for hPSC-CMs (blue) and hPSC-PFs (orange). D) Partial least squares regression analysis comparing the sensitivity of response to inhibition of commonly expressed channel proteins involved in CM physiology. E) Model simulation of the effect of dofetilide on the hPSC-CM/PF AP profile. F) Model simulation of the effect of dofetilide on the hPSC-CM/PF CaT profile. G) Quantification of the change in AP features across hPSC-CMs (blue) and hPSC-PFs (orange) as a function of dofetilide treatment. H) Quantification of the change in CaT features across hPSC-CMs (blue) and hPSC-PFs (orange) as a function of nifedipine treatment. I) Model simulation of the effect of nifedipine on the hPSC-CM/PF AP profile. J) Model simulation of the effect of nifedipine on hPSC-CM/PF CaT profile. K) Quantification of change in AP features across hPSC-CMs (blue) and hPSC-PFs (orange) as a function of nifedipine treatment. L) Quantification of the change in CaT features across hPSC-CMs (blue) and hPSC-PFs (orange) as a function of nifedipine treatment. M) Model simulation of the effect of modifying extracellular sodium concentration on hPSC-CM/PF AP profile. N) Model simulation of modifying extracellular sodium concentration on hPSC-CM/PF CaT profile. O) Quantification of the change in AP features across hPSC-CMs (blue) and hPSC-PFs (orange) as a function of sodium concentration changes. P) Quantification of the change in CaT features across hPSC-CMs (blue) and hPSC-PFs (orange) as a function of sodium concentration perturbation.

Figure 7.. Pluripotent Stem Cell-Derived Purkinje Fibers Demonstrate Differential Sensitivity to Dofetilide and Nifedipine Treatment

A/B) Representative (A) action potential and (B) CaT waveforms for hPSC-CMs and hPSC-PFs treated with varying concentrations of dofetilide. C/D) Quantification of changes in (C) APD50 and (D) APD90 in hPSC-CMs (blue) and hPSC-PFs (teal) compared to untreated controls. E/F) Quantification of changes in (E) CaD50 and (F) CaD90 in hPSC-CMs (blue) and hPSC-PFs (teal) compared to untreated controls. G/J) Representative (G) action potential and (J) CaT waveforms for hPSC-CMs and hPSC-PFs treated with varying concentrations of nifedipine. H/I) Quantification of changes due to nifedipine treatment in (H) APD50 and (I) APD90 in hPSC-CMs (blue) and hPSC-PFs (teal) compared to untreated controls. K/L) Quantification of changes in (K) CaD50 and (L) CaD90 in hPSC-CMs (blue) and hPSC-PFs (teal) compared to untreated controls. * = p <0.05 according to pairwise students t-test. Dots represent average of (n=10–15) technical replicates for each biological replicates for a total of (n=5) biological replicates.