Maturation of iPSC derived cardiomyocytes in a heart-on-a-chip device enables modeling of dilated cardiomyopathy caused by R222Q-SCN5A mutation

Abstract

To better understand sodium channel (SCN5A)-related cardiomyopathies, we generated ventricular cardiomyocytes from induced pluripotent stem cells obtained from a dilated cardiomyopathy patient harbouring the R222Q mutation, which is only expressed in adult SCN5A isoforms. Because the adult SCN5A isoform was poorly expressed, without functional differences between R222Q and control in both embryoid bodies and cell sheet preparations (cultured for 29-35 days), we created heart-on-a-chip biowires which promote myocardial maturation. Indeed, biowires expressed primarily adult SCN5A with R222Q preparations displaying (arrhythmogenic) short action potentials, altered Na⁺ channel biophysical properties and lower contractility compared to corrected controls. Comprehensive RNA sequencing revealed differential gene regulation between R222Q and control biowires in cellular pathways related to sarcoplasmic reticulum and dystroglycan complex as well as biological processes related to calcium ion regulation and action potential. Additionally, R222Q biowires had marked reductions in actin expression accompanied by profound sarcoplasmic disarray, without differences in cell composition (fibroblast, endothelial cells, and cardiomyocytes) compared to corrected biowires. In conclusion, we demonstrate that in addition to altering cardiac electrophysiology and Na⁺ current, the R222Q mutation also causes profound sarcomere disruptions and mechanical destabilization. Possible mechanisms for these observations are discussed.

Introduction

Dilated cardiomyopathy (DCM) is characterized by left ventricular (LV) dilation and reduced contractile function with eventual progression to systolic heart failure (HF)^1,2. Although mutations in the pore-forming α-subunit (SCN5A) of cardiac voltage-gated sodium channels (Nav1.5) are most frequently associated with cardiac arrhythmias (LQT3 and Brugada syndrome), a subset of SCN5A mutations, particularly those involving the voltage-sensor domains (VSDs), are linked to familial DCM often in association with arrhythmias^3–12. Several distinct, but potentially interdependent, mechanisms have been proposed to explain SCN5A-dependent DCM³ including 1) reduced pumping efficiency due to defective conduction related to loss of Nav1.5 function^4,13,14, 2) excessive ectopic triggered activity and impaired pumping efficiency due to altered Nav1.5 gating^3,9–11 or 3) disruption of ion homeostasis leading to metabolic stress of CMs arising from either persistent (non-inactivating) Na⁺ entry^15,16 or the presence of a non-selective pore (the Ω-pore) through the voltage-sensor domain (VSD).^5–7,17 as seen in R222Q as well as other mutations^5,8–10,17. It has also been proposed that SCN5A mutations can disrupt established interactions between Nav1.5 channels and various structural proteins, particularly those in cytoskeleton and intercalated discs¹⁸. In this regard, mutations leading to Arrhythmogenic Right Ventricular Dysplasia as well as some mutations causing Brugada Syndrome occur in structural proteins, in desmosomes and intercalated discs as well as other membrane regions.^19–30 Such conditions have been linked frequently to intercalated disc disruption and impaired conduction, often in association with reduced Nav channels in the intercalated disc and perinexus regions which may disrupt ephaptic coupling. Collectively, these observations suggest that some SCN5A mutations could disrupt critical interactions between Nav1.5 channels and structural proteins that are needed for maintaining the structural integrity of CMs³¹.

Consistent with the diverse potential mechanisms for DCM with SCN5A mutations^1,19–30, the treatment of DCM in patients with SCN5A mutations varies. Currently, best practice guidelines for patients SCN5A-linked DCM recommend standard treatments for heart failure patients i.e. β-blockers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, diuretics, etc ^3,32. However, in some SCN5A patients, for example those with R222Q-SCN5A, standard heart failure medications are ineffective¹⁰ while antiarrhythmic therapies improve LV function^9,10,33, despite generally being contraindicated for patients with heart diseases^34,35. The basis for the variable responses of DCM patients with SCN5A mutations to various heart therapies remains unclear but likely reflects differences in the mechanisms underlying DCM in these patients.

Since the basis for the DCM in patients with VSD mutations and their response to treatment remains unclear, we examined the effects of an autosomal dominant missense mutation (665G>A or R222Q) in the VSD of the first internal repeat of Nav1.5 channels^8–10 in cardiac preparations generated from patient-derived iPSCs. This mutation was chosen because it is associated with a plethora of cardiac arrhythmias in most patients (93-100%) as well as DCM in some patients^8–10, which can be reversed with antiarrhythmic therapy^9,10,33. An important and clinically relevant feature of this mutation is that it is exclusively found in adult splice variants of the SCN5A gene (containing exon 6 versus 6A) ^36,37. As a result, investigating the basis creates a major challenge of iPSC-based cardiac studies, specifically tissue maturation. We utilized our 3D biowires tissue model³⁸ to overcome the challenges associated with development regulation of alternative splicing in the SCN5A expression while simultaneously allowing routine extensive assessment of the effects of the R222Q mutation on cardiac function. Our findings establish that when expression of the R222Q mutation is low (in cultured isolated cells and cell sheets (CSs) or in biowires cultures for short period), no functional differences were uncovered between tissues created from iPSCs with the R222Q mutations versus tissues created from isogenic corrected iPSCs. By contrast, electrically stimulated biowires (cultured for 8 weeks) resulted in 100% conversion to the adult isoform of SCN5a and induced action potential (AP) abbreviation along with profound contractile dysfunction and sarcomere disruption which were associated with differential regulation of gene sets and cellular processes related to sarcoplasmic reticulum, dystroglycan complex; calcium ion regulation and APs. Our studies support the conclusion that, the R222Q mutation can affect the integrity of contractile machinery resulting in impaired contractile function, independent of arrhythmias, which can explain the DCM seen in these patients.

Results

Directed differentiation of ventricular CMs

Employing a directed differentiation method we have used extensively in the previous studies³⁹, cardiac embryoid bodies (EBs) ^24,25 were generated from patient-derived iPSCs with the R222Q-SCN5A mutation as well as iPSCs with the R222Q-SCN5A mutation corrected (Supplemental Fig. 1 and Fig. 1). After 20 days in culture (T20) both R222Q EBs and corrected EBs showed similar beating rates and morphometry (data not shown) with indistinguishable numbers of cells that were positive for TnT (83-84%) and the ventricular marker, MLC2v (55-59%) (Fig. 1 B–E).

Figure 1: Generation of iPSC-derived ventricular constructs.

A) Schematic of generation of R222Q-SCN5A iPSCs, CRISPR correction, iPSC-derived ventricular differentiations and the three systems used to study R222Q-SCN5A. B) Representative flow cytometry plots of mesoderm (PDGFRα and CD56) and ventricular mesoderm (CD235a) marker expression in cells dissociated from T3 EBs. C) Quantification of T3 mesoderm marker expression. D) Representative flow cytometry plots of CM (TnT) and ventricular CM (MLC2v) marker expression in cells dissociated from T20 EBs. E) Quantification of T20 CM marker expression. Plots show mean±SEM, *p<0.05 using unpaired student’s t-test.

Properties of iPSC-derived EBs and cardiac cell sheets

To assess the electrical properties of EB-derived CMs, T25-32 EBs were dissociated and CMs were plated for 3 days before electrophysiological measurements. As seen in intact EBs, CMs isolated from R222Q and corrected EBs showed similar beating patterns without evidence of arrhythmogenicity, and no differences (P=0.81) in the frequency of spontaneous AP generation during current-clamp recordings (Fig. 2). Moreover, no differences in the AP profile (P>0.18) were observed when APs were generated using anode breaks from a holding potential of −80 to −90 mV (Fig. 2A and C).

Figure 2: Action potentials (APs) and sodium current I_Na in corrected and R222Q iPSC-derived ventricular CMs, from EBs and cell sheets.

A) Typical AP measurements in single CMs isolated from EBs cultured for 28 days using the current-clamp patch-clamp technique. The cardiomyocytes were held at −85mV for 1 second and then allowed to spontaneously depolarize (i.e. anode breaks). B) APs recorded using sharp microelectrodes (filled with 3M KCl) in cell sheets paced at 1 Hz after 28-35 days in culture. C) Summary of AP parameters for APs recorded in single CMs isolated from EBs (with spontaneous beating and following an anode break) as well as in cell sheets paced at 1 Hz.

Representative I_Na recordings from single CMs isolated from E) EBs at day 28-35. These measurements were made using the voltage-clamp patch-clamp technique with the voltage protocol shown in the insert. Peak I_Na versus voltage in cardiomyocytes from corrected and R222Q CM isolated from G) EBs after 28-35 days in culture. Steady-state activation and inactivation of I_Na in corrected and R222Q CMs isolated from H) EBs after 28-35 days in culture. MDP=Minimum diastolic potential; APD90=AP duration at 90% repolarization; V_1/2=voltage of half-maximal activation or inactivation; Data show mean±SEM. (Colour required in print).

Since I_Na is critical for electrical conduction in the heart, we performed optical mapping studies of CSs created from T20 EBs and cultured for an additional 14-15 days. We found that no difference (P>0.51) in either the maximum capture rate (MCR) or conduction velocity (CV) between corrected and R222Q sheets (Supplemental Fig. 2). Moreover, we were not able to induce either re-entrant or triggered arrhythmias in either groups, using overdrive pacing (i.e. 20 Hz for 30 seconds) as we did previously in hESC-derived atrial sheets²². Consistent with the absence of inducible arrhythmias, no differences (P>0.34) in AP profiles were observed between cell lines in CSs paced at 1 Hz (Fig. 2B and C).

Since previous studies have established that the R222Q mutation alters Nav1.5 channel gating properties^5,8–11,40, we assessed Na⁺ currents (I_Na) using patch-clamp voltage-clamp recordings of isolated CMs dissociated from EBs and plated 3 days prior to recording. I_Na density measured following depolarization steps to −30 mV from holding potentials of −100mV were not detectably different (P=0.35) between R222Q and corrected CMs, although there was marked cell-to-cell variability in both groups (Fig. 2D–F). Moreover, Boltzmann fits of the I_Na conductance (See Methods) revealed small (~+3 mV) rightward shifts (P=0.004) in V_1/2 for channel steady-state activation, without shifts (P=0.89) in steady-state inactivation (Fig. 2D–F) between R222Q and corrected cells.

Since some studies previously concluded that R222Q-Nav1.5 channels can display Ω-pore currents through VSDs^5,40, although this conclusion is controversial^8–10, we used protocols and ionic condition (i.e. high Cs⁺) to identify non-selective Ω pore currents, as previously described⁵ (See Methods). The currents measured under these conditions revealed no difference (P=0.89) between the groups (Supplemental Fig. 3). We noted however, that NMDG-sensitive inward currents were observed in both groups at the end of 160 ms depolarizing steps (consistent with Ω-pore currents⁵) and these appeared to be related to cesium conduction through hERG channels³⁰, because they were blocked by the hERG blocker³⁴, dofetilide (0.5 μM), as previously reported³⁰ (Supplemental Fig. 3C).

To examine possible links between R222Q, sarcomeric proteins and DCM, we performed immunohistological evaluations of TnT expression in EBs digested at T25 and plated for 5 day to reveal far lower (P<0.0001) TnT levels in R222Q versus corrected CMs (Fig. 3A to C). Blind analysis of sarcomere organization in cells plated for 5 days revealed that sarcomeres were less organized (P=0.02) in R222Q compared to corrected cells. (Fig. 3A and D to F). It is possible that the observed structural changes in the absence of electrical changes may reflect differences in the gene dosage necessary to elicit these aspects of the R222Q phenotype.

Figure 3: Cardiac Troponin T expression patterns in isolated corrected and R222Q iPSC-derived ventricular CMs.

A) Representative images of TnT staining in isolated CMs dissociated from EBs at day 20 of the differentiation and plated at a low density for 5 days before imaging. B) Percentage of the area of CMs positive for TnT from day 21-25. Plots show mean±SEM. *p<0.05 using unpaired student’s t-test. (Colour required in print) C) Percentage of the area of CMs positive for TnT at day 25. D) Sarcomere disarray in CMs derived from corrected and R222Q iPSCs at 25. E) The alignment of the sarcomere proteins (quantified by eccentricity) and F) the amount of protein staining (quantified by density analysis) of cardiac TNT. Data show mean ± SEM. *p<0.05 using unpaired student’s t-test.

Our I_Na results in CMs from T29-35 EBs are inconsistent with results in heterologous expression systems and a murine model of R222Q-SCN5A^5,8–11,40. To explore whether the absence of electrical differences between R222Q and corrected CMs could arise from incomplete maturation of our iPSC-derived cardiac tissues, we measured the expression levels of the adult versus fetal SCN5A variant (exon 6A in the fetal form is replaced with exon 6 which harbours the R222Q mutation) ^15,16 as a function of time in culture (Fig. 4A,B). We found that the ratio of fetal to adult isoforms in corrected EBs decreased (P=0.02) from T20-22 to T26-28 with trends (P=0.19) toward isoform switching in R222Q EBs (Fig. 4C&D). Based on these measurements, we estimated that, at T26-28, the expression level of SCN5A with the R222Q mutation was only 36.6% of the total RNA, meaning that the adult SCN5A variants are expressed 73.3% (i.e. twice of 36.6%). Additionally, measurements in CSs over similar culture periods revealed comparable SCN5A maturation to that seen in EBs (Fig. 4D). These results establish that SCN5A maturation was incomplete in EBs and CSs, which may underlie the absence of measureable differences in I_Na between corrected and R222Q CMs obtained from EBs. To explore whether longer times in culture might increase I_Na maturation, we also examined CMs from EBs cultured for 64-67 days. However, even with these long culture periods, only small (~3 mV) differences (P=0.03) were seen in the V_1/2 for I_Na activation between corrected and R222Q CMs unlike what has been reported previously for heterologously expressed R222Q channels¹⁷. (Supplemental Figure 4).

Figure 4: SCN5a isoform expression in iPSC-derived ventricular EBs, cell sheets and biowires.

A) A schematic illustrating a portion of the genomic structure of the SCN5A. In development, the fetal SCN5A mRNA variant incorporates exon 6A which is replaced by exon 6 in the adult SCN5A transcript. It is the adult SCN5A transcript, not the fetal form, that expresses the R222Q mutation found in patients presenting with arrhythmias and cardiomyopathy. B) Sequencing comparison of the qPCR products using primers to identify the fetal exon 6A versus the adult exon 6 results in cardiac tissues derived from iPSCs in our studies. C) Comparisons of the relative levels of adult versus fetal in embryoid bodies (at 2 time points) generated from R222Q and corrected iPSCs. Graphs show mean±SEM, *p<0.05 using paired student’s t-test. D) Results showing the time course of changes in the relative expression of the adult SCN5A isoform in cultured embryoid bodies (EBs) and cell sheets (CSs). P values for slopes (slope ≠ 0) were estimated from linear regression to assess the levels of the adult transcript changes with time in culture. E) A graph was generated using the percentage of spliced SCN5a transcript reads relative to the first exon (mutation carrying exon) in biowires. F) Principal component analysis of RNAseq measurements in Mutant and Control biowires after 3 days and 8 weeks of cultivation.

Modeling the R222Q-SCN5A mutation in biowires

Based on our studies in EBs and CSs showing incomplete SCN5A maturation, we focused our efforts on electrically stimulated 3D biowire preparations, which have been shown to promote myocardial maturation^41,42. Consistent with enhanced maturation of biowires, our RNAseq measurements revealed that, after culturing for 8 weeks in the presence of rapid electrical stimulation (See Methods), both R222Q and corrected biowires expressed almost entirely the adult SCN5A variants, whereas maturation was incomplete in biowires after only 3-days in culture without electrical stimulation (Fig. 4E). Additionally, principal component analysis of our biowires revealed clearly distinct expression patterns as a function of time in culture as well as between R222Q and corrected preparations (Fig. 4F).

After 8 weeks in culture, electrically stimulated biowires showed abbreviated (P<0.05) action potentials (APs) compared to corrected biowires, as seen in a mouse model of R222Q⁴⁰, without detectable differences in the minimum diastolic potentials (P=0.17), upstroke velocities (P=0.06) or amplitudes (P=0.29) (Fig. 5A and B). MCRs and CVs (measured using optical mapping) were also similar (P>0.14) between corrected and R222Q biowires (Fig. 5C and D). While these observations are consistent with enhanced SCN5A maturation in these biowires, neither re-entrant nor triggered arrhythmias were inducible (Data not shown) using overdrive pacing (i.e. 20 Hz for 30 seconds), despite differences in AP profiles.

Figure 5: Electrophysiological properties of corrected and R222Q iPSC-derived ventricular biowires.

A) Representative intracellular AP recordings from biowires paced at 1 Hz. B) Quantification of actional potential (AP) parameters indicated for biowires paced at 1 Hz. MDP=Minimum diastolic potential. C) representative activation maps of conduction across corrected and R222Q biowires. D) Conduction velocity (CV) and maximum capture rate (MCR) across corrected and R222Q biowires. APD90=AP duration at 90% repolarization; Plots show mean±SEM, *p<0.05 using unpaired student’s t-test. Representative I_Na recordings from single CMs isolated from E) Biowires after 4 weeks in culture. These measurements were made using the voltage-clamp patch-clamp technique with the voltage protocol shown in the insert. Peak I_Na versus voltage in cardiomyocytes from corrected and R222Q CM isolated from F) Biowires after 4 weeks in culture. Steady-state activation and inactivation of I_Na in corrected and R222Q CMs isolated from G) Biowires after 4 weeks in culture. V_1/2=voltage of half-maximal activation or inactivation; Data show mean±SEM. (Colour required in print).

Consistent with the improved maturation and the high levels of expression of the adult SCN5A variant in the cultured biowires (conditioned with rapid electrical pacing), we observed a large hyperpolarization (leftward) shift in the voltage required for 50% activation (i.e. V_1/2) of I_Na for CMs isolated from R222Q biowires compared to corrected CMs (i.e −51.02±0.81 in R222Q versus −37.78±0.64 mV in corrected) (Figure 2 D–F and Figure 5 D–F). On the other hand, no differences were observed in either the voltage-dependence of inactivation or the kinetics of I_Na during voltage steps between the groups. This leftward shift in activation of R222Q biowires versus corrected biowires, with minimal changes in inactivation and I_Na kinetics is very similar to the shifts reported previously (by us and others^8,10) using heterologous expression systems expressing the adult splice variants. Interestingly, we also observed higher I_Na densities (pA/pF) in the R222Q CMs compared to corrected CMs as quantified from either maximal conductance estimates (see Methods) or comparisons of peak I_Na magnitudes in response to voltage steps to −20mV (92.86±38.80 pA/pF in R222Q versus 51.87±20 pA/pF in corrected) (Figure 2 D–F and Figure 5 D–F). The basis for these differences in I_Na amplitude is unclear. It should be mentioned that these I_Na measurements were made in biowires after only 4 weeks in culture, because it was not possible to isolate healthy CMs following digestion of biowires at the 8-week time point (Supplemental Figure 8). Nevertheless, these observations support the conclusion that the adult Nav isoform, harboring the R222Q mutation, is expressed at higher levels in electrically stimulated biowires after only 4 weeks in culture.

Consistent with a DCM seen in R222Q patients^9,10, biowires cultured for 8 weeks with electrical stimulation generated far less (P<0.0001) force than corrected biowires, whether quantified as the absolute force or force per cross-sectional area and regardless of the pacing frequency (Fig. 6). These reductions in force were not associated with difference (P>0.14) in either the amplitudes or the kinetics of calcium transients between R222Q and corrected biowires, regardless of whether the calcium transients were assessed using Fluo-4 (non-ratiometric Ca²⁺ dye) or Fura-2 (ratiometric Ca²⁺ dye), (Supplemental Fig. 5). Additionally, no differences (P=0.88) in diastolic Ca²⁺ levels (measured with fura-2) were observed between the groups (Supplemental Fig. 5D).

Figure 6: Active force generation in corrected and R222Q biowires.

A) Representative force tracings in corrected and R222Q biowires paced from 1-3 Hz after 8 weeks of cultivation. B) and C) Absolute force and force relative to cross-sectional area paced from 1-3 Hz demonstrating contractile dysfunction in R222Q biowires. Plots show mean±SEM, p<0.0001 using a one-way ANOVA with Sidak’s multiple comparison test.

The impaired contractility of R222Q biowires in the absence of changes in Ca²⁺ suggests possible defects at the level of contractile proteins. To explore further the basis for the reduced force generation, we performed immunohistochemical studies to examine sarcomeric structure in biowire tissues. Similar to single cells, reductions in TnT staining and sarcomere disarray in matures biowires were observed in R222Q compared to corrected controls (Fig 7A and B). R222Q biowires consistently showed 2.7-fold lower (P=0.001) TnT staining per cross-sectional area and 2.2-fold less total TnT staining than corrected biowires (Fig. 7C–E). These changes in TnT expression were accompanied by greater (P=0.02) cross sectional areas and lower (P=0.02) compaction of the R222Q biowires compared to corrected biowires (Fig. 7C), possibly indicative of a dilated cardiomyopathy^1,2,32. Consistent with a DCM phenotype, there was a strong correlation (P<0.006) between absolute force and TnT expression for both corrected and R222Q biowires (Fig. 7C and D). Associated with these changes in TnT, we observed a profound (~3.5 fold) reduction (P=0.0004) in the slope of the relationship between force and TnT expression in R222Q versus corrected biowires (Fig. 7E), suggesting that marked reductions in the efficiency of force generation per amount of TnT staining in mutant biowires (i.e. 0.95 mN/pixel in R222Q biowires versus 3.29 mN/pixel in corrected biowires). This reduced efficiency might be explained by the marked sarcomeric disarray seen in R222Q compared to corrected biowires (8 weeks in culture) (Fig. 7A and B) which was also accompanied by reduced expression of MLC2v (p=0.0265) and α-sarcomeric actin (p=0.0316), as well as f-actin (p=0.0241).

Figure 7: Structural protein expression in corrected and R222Q iPSC-derived ventricular biowire tissues after 8 weeks of culture.

A) Representative images of TnT staining in cross-sections of biowire tissue (first column); as well as the representative images of α-sarcomeric actin, f-actin, MLC2V, and cTNT in longitudinal view of biowire tissues, scale bars= 100μm B) The alignment of the sarcomere proteins (quantified by eccentricity) and amount of protein staining (quantified by density analysis) of each structural protein staining from left to right: α-sarcomeric actin, f-actin, myosin light chain 2V (MLC2V), and cTNT. Data show mean ± SEM. *p<0.05 using unpaired student’s t-test.

C) Total TnT staining in the tissue per field of view (left) and TnT staining per cross-sectional area (right). D) Cross-sectional area (left) and percent compaction (right) in biowire tissue. E) Relationships between absolute force generation and TnT expression with slopes of linear regression lines in brackets (left) and mean absolute force and mean TnT expression (right). P values in E) compare the slopes of linear regression lines to zero (in brackets) and between cell lines (next to the plot). (Colour required in print).

To explore further the basis for the profound effects of the R222Q mutation on the myocardium, we performed RNA sequencing to compare either biowires cultured for either 3 days without stimulation (immature 3D tissues) or cultured for 8 weeks with electrical stimulation (mature 3D tissues) (Supplemental Fig. 1 B). Importantly, PCA analyses of RNAseq results (Fig. 3F) showed clear separation between R222Q and corrected biowires at both time points. Moreover, heatmap representations of the RNAseq results (Fig. 8A) revealed systematic differences between R222Q and corrected biowires in the expression levels of genes that interact directly (primary interactome, labeled red) and indirectly (secondary interactome, labeled black) with Nav1.5 channel proteins, at both immature 3-day and mature 8-week biowires. Moreover, differences were also observed in a number of primary and secondary interactome genes between corrected and R222Q biowires in both 3-day and 8-week biowires (e.g. DMD, DLG1 and CDH2, Fig. 8A).

Figure 8: Gene expression in Mutant and Control biowire cardiac tissues.

A) A heat map representation of the expression of genes that interact directly with Nav1.5 channels (primary interactome, genes listed in red) or interact indirectly with Nav1.5 channels (secondary interactome genes, black labels).

Gene Ontology (GO) term enrichment were performed for biowires from Mutant and Control groups after 8 weeks of maturation, to identify sets of genes that were significantly upregulated in Mutant group compared to Control, including B) dystrophin-associated glycoproteins genes and dense body (desosome) genes, and C-E) gene sets associated with cytosolic calcium ion regulation and action potentials.

In contrast, sarcomere structural protein expressions F) did not stand out in the GO term enrichment analysis.

Consistent with the sarcomeric disarray (Fig. 7), Gene Ontology (GO) term analyses of the RNAseq results revealed that 8-week R222Q biowires showed differential regulation of pathways related to dystroglycan (Fig. 8B) and the sarcomplasmic reticulum (Fig. 8D), as well as calcium ion regulation and APs (Fig. 8C). Interestingly, dystrophin-associated glycoprotein complex gene sets were differentially regulated in R222Q compared to the control biowires along with reduced expressions of genes associated with dense bodies and desmosomes, which are central to intercalated disc structures, along with severely reduced expression of actin molecules (ACTB, ACTG1) (Fig. 8 B). These results are consistent with effects of the R222Q mutation on Nav channel assembly and mechanical stability in intercalated discs and the perinexus regions. On the other hand, sarcomeric proteins were not differentially related between the four groups (Fig. 8F). Together, these findings indicated that the elevation of cardiac-related genes between Control and Mutant groups at each timepoint (Supplemental Fig. 6) was due to the increased AP and calcium regulation-related genes (updated Fig. 2E, F and Fig. 8), which were highly relevant to the disease phenotype. These findings also demonstrate that the functional changes in biowire cardiac tissues could be related to the changes in CMs themselves, rather than the cell population shift (Supplemental Fig. 8).

To assess possible cell composition shifts over time in culture which may complicate the interpretation of the Gene Ontology (GO) term analyses of the RNA seq results, we assessed the expression patterns of different cell-specific markers associated with the major cell types in our biowire tissues, which consist of iPSC-derived CMs and primary fibroblasts as well as some endothelial cells created by our cardiac directed differentiation protocols. We found that genesets of CMs were significantly upregulated in the Mutant group compared to the Control group at 3 days and 8 wks respectively (Supplemental Fig. 6), while the fibroblast and endothelial cell signatures were unchanged among the R222Q and corrected controls at both time points (Supplemental Fig. 7).

Discussion

Previous heterologous expression studies demonstrated that R222Q-SCN5A channels show voltage-dependant gating shifts^5,8–11,17, possibly with non-selective Ω-pore currents^5,17, when compared to wild-type channels. In contrast we observed no EP differences (i.e. AP morphology, spontaneous firing rates, Na⁺ channel properties, CVs, MCRs or arrhythmia susceptibility) in isolated CMs or CSs generated from R222Q EBs compared to corrected EBs after 29-35 days in culture. While the absence of EP differences between these R222Q and corrected preparations could arise for several reasons, incomplete CM maturation under our experimental conditions is likely to be a major contributor. Indeed, the SCN5A gene undergoes alternative splicing during development and the R222Q mutation is only expressed in adult isoforms (containing exon 6) ^36,37. Consistent with immaturity, we observed incomplete fetal-to-adult isoform switching for SCN5A in these preparations (i.e. adult SCN5A mRNA in R222Q EBs at day 26-28 contributed only 37% (of a possible 50% given heterozygousity).

Incomplete tissue maturation is a well-recognized limitation that continues to plague studies of CMs derived for pluripotent stem cells^43–47 which has been associated with deficiencies in tissue architecture, the supporting extracellular matrix, necessary humoral factors and an appropriate complement of cell types^48–51, to name a few. The incomplete SCN5A maturation seen in our studies using isolated cells and CSs derived from EBs as well as the absence of altered tissue properties in R222Q tissues are similar to findings from a previous study of iPSC-derived CMs homozygous for the I230T-SCN5A mutation (in exon 6)⁵². Taken together, it appears that the level of mutant R222Q-Nav1.5 expression was insufficient to display a disease phenotype in isolated CMs and CSs after 26-28 days in culture. Incomplete CM immaturity might also have limited the manifestation of a phenotype in our R222Q tissues as a result of the relatively depolarized MDPs observed routinely in iPSC-derived cardiac preparations^43,53 which may limit the contribution of Nav1.5 channels to EP properties by promoting inactivation.

To overcome the potential issues with tissue immaturity, we generated biowires that underwent chronic electrical stimulation, which promotes CM maturation^3,54. Consistent with maturation, R222Q and corrected biowires almost exclusively expressed the adult variant of SCN5A (31% to 89% in corrected and 45% to 90% in R222Q, p-values <0.05; Figure 4 E) in association with AP abbreviation similar to that seen in transgenic mice expressing human R222Q-SCN5A¹⁷. Moreover, CMs isolated from biowires after 4 weeks in culture display changes to the biophysical properties of I_Na (i.e. a leftward shift in the activation of R222Q biowires versus corrected biowires in the absence of other changes) consistent with previous reports using heterologous expression systems ^5,8–11,40

Interestingly, R222Q biowires also displayed profound reductions in force generation and reduced compaction (i.e. dilation) compared to corrected biowires, which aligns with a DCM phenotype^1,2. The differences in contractile function between corrected and R222Q biowires occurred in the absence of differences in either calcium handling, arrhythmia susceptibility or shifts in cell composition (as assessed via cell specific markers), suggesting that the contractile defect in R222Q biowires occurs primarily downstream of calcium handling at the level of the contractile apparatus in CMs. Consistent with this, immunohistochemical measurements in both biowires and isolated CMs revealed large reductions in TnT staining associated with a ~4-fold reduction in force generation per unit of TnT expression in R222Q compared to the corrected biowires. This reduced ratio of force to TnT level is consistent with the sarcomere disarray observed in isolated R222Q CMs, which may contribute to the reduced contractile efficiency seen clinically in the R222Q-SCN5A myocardium.

While our R222Q biowires revealed dilation combined with profound reductions in force generation, matching the DCM phenotype seen in some R222Q-SCN5A patients^8–10 and AP abbreviation seen in the mouse model¹⁷, they failed to display spontaneous or inducible arrhythmias as commonly observed in these patients^8–10. This absence of arrhythmias in biowires might (at first glance) appear to be inconsistent with the relatively high arrhythmia incidence in R222Q-SCN5A patients (i.e. 93-100%)^8–10; however biowires have relatively small dimensions (typically 4 mm x 500 μm x 300 μm), which minimize the appearance of complex re-entrant-type events. Nevertheless, although the penetrance of DCM in R222Q-SCN5A patients is relatively low (i.e. 20-53%)^8–10, it is conceivable that chronic stimulation over the 8 week incubation period, which is used to drive tissue maturation, could simulate chronic cardiac stress and unmask this aspect of the disease phenotype. Indeed, in a previous study we found that following up to 8 months of rapid stimulation, which involved ramping up from 1 to 6 Hz (as in the studies presented here) followed by maintenance at 3 Hz, biowires generated from patients with hypertension reproduced the contractile properties observed clinically in response to chronic increases in afterload³⁸. In this regard, the LV dysfunction seen in the patient who provided the iPSCs for our studies was exacerbated by dual right atrial and ventricular pacing at 70 beats per minute, which was reversed upon cessation of pacing (unpublished observations), suggesting that external stimulation might promote a DCM phenotype.

Taken together, our results suggest that the R222Q mutation alters the molecular structure and disrupt interactions with various proteins leading to reduced myofibril density, sarcomeric disarray, tissue dilation and impaired contractile performance. This is consistent with mechanisms previously proposed to explain DCM in some patients with SCN5A mutations by Asatryan, et al⁴². Although the underlying basis for the impaired contractile function and the R222Q mutation is unclear, it is known that Nav1.5 channel function is altered by interacting directly with sarcomeric proteins including α-actinin-2 via the cytoplasmic loop connecting domains III and IV and indirectly through PDZ motif gene (ZASP) (PSD-95/Discs-large/ZO-1, PDZ)^19,21,28. Possibly more relevant is the observation that a distinct pool of Nav1.5 channels are localized to the intercalated discs, which are structures that are the centres of cell adhesions and desmosomes needed for mechanical transmission^25,27, as well as connexin-dependent cell-cell communication and ephatic conduction. Disruptions in interactions between several structural proteins within intercalated discs are known to impact Nav1.5 channels expression levels thereby contributing to disease conditions such as arrhythmogenic right ventricular cardiomyopathy^{22–24,52,55–57}. Additionally, DCM leads to reduced expression of proteins in the intercalated disc and a loss of intercalated discs integrity⁵⁸ resulting in Nav1.5 channel lateralization and associations with the syntrophin-dystrophin complex⁵⁹, which is critical for mechanical coupling as well as mechanotransduction of CMs.⁶⁰ Disruption of Nav1.5 dystrophin complex have also been shown to impair Nav1.5 function^25,59,61.

Consistent with the interaction of Nav1.5 channels with multiple structural and contractile proteins, we observed marked reductions in the levels of sarcomeric proteins (TnT, MLC2V, f-actin) combined with severe sarcomeric disarray in R222Q tissues compared to the corrected biowire tissues. We suggest that post-translational changes in contractile proteins might arise from structural changes in Nav1.5. Changes induced by the R222Q mutation may prevent appropriate co-assembly of Nav1.5 channels with structural proteins in the membrane (Cav3⁶², ankyrin⁶³, dystrophin⁶⁴, syntrophin⁵⁹) as well as with proteins at the intercalated disc and perinexus regions such as Cx43 and α-actinin 2 (which may or may not involve the Nav channel subunit Navß) ^19,65. This disruption may interfere with the transmission of mechanical forces between CMs and lead to increased sarcomeric degradation and sarcomeric disarray, a common feature of many conditions leading to biomechanical stress. Note that this observation is not inconsistent with our gene sets enrichment analysis (GSEA) which revealed that dystrophin-associated glycoprotein complex gene sets were differentially regulated between R222Q and control biowires as well as the reduced expression of the actin proteins (Dense Body). Thus, we suggest that the R222Q mutation may interfere with the assembly of Nav channels in the intercalated disc leading to mechanical destabilization that radiates to the sarcomere. Alternatively, disturbances of the ionic environment in localized regions surrounding Nav1.5 channels may differentially impact the integrity of sarcomeric and structural proteins, which would be consistent with the mechanism proposed previously Moreau, et al⁵.

Regardless of which mechanisms underlies the DCM phenotype in R222Q-SCN5A patients, it is clear from our studies that the biowire platform was able to recapitulate the DCM aspect of the clinical phenotype and to demonstrate that the reduced contractility was associated with disruption of the sarcomeric structure. Moreover, our experimental approach may provide a convenient strategy for future studies of other Nav1.5 mutations associated with a DCM phenotype.

Limitations

It is important to acknowledge that despite efforts to model adult human ventricular CMs and tissue in vitro, our study has notable limitations. The biowires presented in this study did not display the R222Q-associated arrhythmias seen clinically, potentially resulting from the small tissue dimensions. The cylindrical 3D cardiac tissues lack the native chamber shape; therefore, heart dilation, observed in the patients with R222Q mutation, cannot be measured. Moreover, the current Biowire chip design does not allow for tissue stretching. Therefore, it remains unknown how the mutation affects the force-length relationship. Additionally, single cells and CSs did not present the EP phenotypes due to the lack of maturation. Since we are unable to isolate cardiomyocytes after 8 weeks of culture, EP in isolated cells is performed at 4 weeks, i.e., the time point when tissues are slightly less mature. Finally, these studies utilize R222Q iPSCs from a single cell line isolated from one patient, given the variable and relatively low penetrance of the DCM phenotype, which limits the ability to generalize our findings.

Future studies can include more in-depth disease phenotype characterization. Specifically, incorporating additional isogenic cardiac cell types, such as endothelial cells and resident macrophages, along with a more precise control over cellular compositions, are pivotal in yielding further insight into the disease mechanisms, as we only have CMs differentiation with 65% population expressing MLC2v. Furthermore, the use of atrial cardiomyocytes to determine the impact of the SCN5A mutation in atrial cardiac tissues is essential, as it is unclear to what extent the atrial effect drives the phenotype in vivo. Moreover, further investigation is needed to understand the mechanotransduction mechanism of the R222Q mutation on the molecular structure of SCN5a, intercalated discs, and the assembly of actin cytoskeleton and sarcomeres.

Summary and Implications

In this study we use several different cardiac preparations created from iPSCs to investigate the effects of the R222Q-SCN5A mutation. Single cells and CSs failed to display the EP properties associated with R222Q-SCN5A, which we linked to tissue immaturity (i.e. <37% adult SCN5A containing R222Q). By contrast, long-term culturing of biowires improved CM maturation and displayed EP changes combined with severe impairment of contractility, which was associated with increased tissue dimensions, reduced TnT expression levels and sarcomeric disarray evidenced by MLCV2v and f-actin staining, in the absence of altered calcium handling. Since we did not observe arrhythmias in our R222Q biowires, our findings suggest that the DCM phenotype seen in R222Q-SCN5A patients does not require the presence of (gain-of-function) ectopies or tachyarrhythmias^3,9–11. Rather, our findings support the mechanisms wherein DCM can arise, independent of arrhythmias, either from the disruption of (direct or indirect) interactions between Nav1.5 channels and various structural proteins or from localized changes to ionic homeostasis that in turn interfere with the integrity of key structural proteins^{5–7,17,19,20,24,25,27,42}. Future studies aimed at understanding the basis of these disruptions may lay the foundation for developing novel approaches for more effective treatment of patients with R222Q as well as other DCM causing SCN5A mutations. In addition, our studies identify an experimental approach, which may prove useful for screening and identifying mechanisms underlying DCM induced by other SCN5A mutations, especially those requiring tissue maturation.

Methods and materials

Corrected and R222Q iPSCs generation

These studies were approved through the coordinated approval process for clinical research through the research ethics board (CAPCR-REB: 13-7063.12) at the University Health Network. Informed consent was obtained from all participants. PBMCs from a 32-year old female patient heterozygous for the R222Q-SCN5A mutation were reprogrammed to iPSCs using Sendai viral-driven expression of Oct4, Sox2, Klf4, and Myc, to induce pluripotency. Isogenic controls were generated using CRISPR/Cas9 gene editing to correct the c.665G>A missense mutation (R222Q). Reprogramming to iPSCs, CRISPR correction and characterization of both corrected and R222Q iPSCs was performed by the Center for Commercialization of Regenerative Medicine (CCRM).

iPSC-derived ventricular CM differentiation

Corrected (T447 CCRM47G C7H) and R222Q (CCRM47 G) iPSCs were cultured on irradiated (30 Gy) MEFs in iPSC culture media containing Dulbecco’s modified eagle media (DMEM)/F12 (Cellgro) with penicillin/streptomycin (1%, Gibco), L-glutamine (1%, Gibco), non-essential amino acids (1%, ThermoFisher), β-Mercaptoethanol (0.1%, Gibco), KnockOut^™ serum replacement (20%, Gibco) and basic fibroblast growth factor (bFGF) (15 ng/ml, Bio-Techne) under normoxic conditions (5% CO2) with daily media changes.

Ventricular CMs were differentiated from iPSCs using a modified version of a EB CM differentiation protocol previously documented by Lee et al and Kattman, et al. (2011)^39,66 (Supplemental Fig. 1) with the concentrations and timing of cytokines for mesoderm induction and Wnt inhibition optimized for our cell lines. Once iPSCs reached 80-90% confluence, EB were generated by dissociating iPSCs into single cells using TrypLE^™ Express (1x), Phenol Red (Gibco), filtering cells through 70 μm cell strainers and culturing at 500,000 cells/ml of StemPro-34 media (Gibco) containing penicillin/streptomycin (1%, Gibco), L-glutamine (1%, Gibco), ascorbic acid (50 μg/ml, Sigma), monothioglycerol (MTG) (50 μg/ml, Sigma), transferrin (150 μg/ml, Sigma/Roche), ROCK inhibitor Y-27632 (10 μM, Bio-Techne) and bone morphogenic protein 4 (BMP4) (1 ng/ml, Bio-Techne) in 6 cm petri dishes (4 ml/dish) on an orbital shaker rotating at 70 RPM (revolutions per minute) (Day 0 of the differentiation). After 18-20 hours on the orbital shaker Day 1 of the differentiation), EBs were filtered through 100 μm cell strainers, transferred to mesoderm induction media containing StemPro-34 with the day 0 supplements excluding ROCK inhibitor Y-27632, with the addition of bFGF (5 ng/ml, Bio-Techne) and with the optimized concentrations of BMP4, (9 ng/ml, Bio-Techne) and Activin A (9 ng/ml, Bio-Techne) and plated on poly (2-hydroxyethyl methacrylate) (5%, Sigma) coated plates to prevent attachment. After 2 days (at day 3 of the differentiation), EBs were washed with Iscove’s modified Dulbecco’s media (IMDM) (Gibco) and transferred to cardiac mesoderm induction media containing StemPro-34 with the day 1 supplements excluding bFGF, BMP4 and Activin A, with the addition of the Wnt inhibitor IWP2 (2 μM, Bio-Techne) and vascular endothelial growth factor (VEGF) (10 ng/mL, Bio-Techne). After 2 days (at day 5 of the differentiation), EBs were transferred to StemPro-34 with the day 3 supplements excluding IWP2 and with reduced VEGF (5 ng/ml, Bio-Techne). After another 7 days (at day 12 of the differentiation), EBs were transferred to StemPro-34 with the day 5 supplements excluding VEGF and transferrin and were maintained in this media for another 8 days (until day 20 of the differentiation). Media was changed every 2-3 days. Cells and EBs were incubated under hypoxic conditions (5% CO₂, 5% O₂, 90% N₂) for from day 0 to day 12 of the differentiation and then transferred to a normoxic environment (5% CO₂) from day 12 to day 20 of the differentiation.

Single and multicellular construct plating

Single cells

For studies of isolated iPSC-derived ventricular CMs, EBs were dissociated to single cells by incubating in Collagenase Type II (250 units/ml, Worthingtons) in HANK’s buffer for 1.5 hours at 37 degrees Celsius followed by treatment with TrypLE^™ Express (1x), Phenol Red (Gibco) for 2-4 minutes at room temperature (RT).

For patch-clamp recordings, EBs were dissociated at day 25-32 as described above and plated at densities of 1,000-3,000 cells per coverslip in 40 μl droplets of StemPro-34 media (Gibco) containing penicillin/streptomycin (1%, Gibco), L-glutamine (1%, Gibco), ascorbic acid (50 μg/ml, Sigma), MTG (50 μg/ml, Sigma) and ROCK inhibitor Y-27632 (10 μM, Bio-Techne) on fragments of coverslips cut to fit the perfusion chamber (~5 mm²) and pre-coated with growth factor reduced Matrigel (25%, Fisher/Corning) in IMDM (Gibco). 8-12 hours after plating, dishes were flooded with the media above excluding ROCK inhibitor Y-27632. Recordings were initiated 3 days after plating at day 29-35.

For histology, EBs were dissociated at day 20 of the differentiation as described above and plated at densities of 3,000-5,000 cells per coverslip in 200 μl droplets of StemPro-34 media (Gibco) supplemented as detailed above for patch-clamp studies on 35 mm glass bottom dishes with a No. 1.5 coverslip and a 10 mm glass diameter (MatTek) pre-coated with growth factor reduced Matrigel (25%, Fisher/Corning) in IMDM (Gibco). 8-12 hours after plating, dishes were flooded with the media above excluding ROCK inhibitor Y-27632. Cells were fixed for immunohistochemistry 1, 3 or 5 days after plating at day 21, 23 or 25.

Cell sheets

CSs were generated by dissociation day 20 EBs to single cells as described above. 1.5 million cells were plated in 200 μl droplets of StemPro-34 media (Gibco) containing penicillin/streptomycin (1%, Gibco), L-glutamine (1%, Gibco), ascorbic acid (50 μg/ml, Sigma), MTG (50 μg/ml, Sigma) and ROCK inhibitor Y-27632 (10 μM, BioTechne) on 35 mm glass bottom dishes with a No. 1.5 coverslip and a 10 mm glass diameter (MatTek) pre-coated with growth factor reduced Matrigel (25%, Fisher/Corning) in IMDM (Gibco). 8-12 hours after plating, dishes were flooded with the media above excluding ROCK inhibitor Y-27632. CSs were cultured for 14-15 days before preforming additional studies at day 34-35.

Biowires

Biowires were seeded in polystyrene strips with 5 mm long x 1 mm wide x 300 μm deep microwells with poly (octamethylene maleate (anhydride) citrate) (POMaC) wires 3 mm apart in microgrooves and running perpendicular to the longitudinal axis of each microwell to allow for tissue attachment. PoMAC wires were fixed in place using polyurethane 2-part adhesive (GS Polymers). Polystyrene sheets and PoMAC wires were generated as described previously^38,67. Prior to seeding, the polystyrene sheets and POMaC wires were soaked in supplemented StemPro-34 media (Gibco) containing penicillin/streptomycin (1%, Gibco), glutamax (1%, Gibco), L-ascorbic acid (13 mM, Sigma), transferrin (150 μg/ml, Sigma/Roche) and HEPES (10 mM) overnight and then percolating them with pluronic F127 (5%, Sigma) and allowing them to air dry before seeding.

At day 20, EBs were dissociated to single cells as described above and seeded as described previously³⁸. Briefly, following dissociation CMs were combined with human cardiac fibroblasts (PromoCell). Fibroblasts were prepared by culturing to 75% confluence and were dissociated with TrypLE^™ Express (1x), Phenol Red (Gibco) for 5 minutes at 37 degrees Celsius. CMs and fibroblasts were combined in collagen hydrogel containing combining rat-tail collagen (3 mg/ml, Corning), 1X Medium 199 (Sigma), NaHCO₃ (2.38 mM), NaOH (10 mM), growth factor reduced Matrigel (15%, Fisher/Corning) and sterile H₂O. The ratios of CMs to fibroblasts were adjusted for each batch based on day 20 flow cytometry to maintain a final percent of non-cardiac cells of 30% (i.e. for a differentiation with 80% TnT⁺ cells, an additional 10% fibroblasts were added). 150,000 cells were seeded in 2.5 μl droplets per well. Following seeding, the hydrogel/cell mixture was incubated for 15-20 minutes at 37 degrees before flooding with the supplemented Stempro-34 media used to soak the polystyrene sheets and POMaC wires detailed above. Following seeding, cells were cultured for 7 days to allow for tissue formation and attachment to the POMaC wires before initiating stimulation. Stimulation was performed by placing the polystyrene strips of biowires between two ¹/₈ inchdiameter carbon rods (Ladd Research Industries) affixed to the bottom of 10 cm tissue culture dishes using a polyurethane 2-part adhesive (GS Polymers) 1 cm apart, running perpendicular to the long axis of the tissue as described previously³⁸. These rods were connected through platinum wires (Ladd Research Industries) to electrical stimulators (Grass Technology S88X Square Pulse Stimulator). Biowires were continuously field stimulated with 1 ms bipolar pulses at 1 Hz beginning at the second week in culture and increasing by 1 Hz per week up to 6 Hz during the seventh week in culture and then decreasing to 1 Hz for the eighth week in culture. Stimulation pulse amplitudes were 1.5x the excitation threshold of the tissue. For all biowires, the supplemented Stempro-34 media was changed weekly up to 4 Hz and biweekly there after. Subsets of biowires were either not stimulated beginning 1 week after seeding and continuing for 7 weeks. For all biowires, the supplemented Stempro-34 media was changed weekly up to 4 Hz and biweekly there after.

RNA expression studies

iPSC-derived ventricular EBs and CSs intended for RNA isolation were harvested, pelleted and frozen at −80 degrees Celsius. Total RNA was isolated using RNAqueous^™-Micro total RNA isolation kits including DNAse treatment (Ambion). Only RNA with a 260/280 ratio greater than 1.8 was used for downstream qPCR as recommended by Fleige and Pfaffl (2006)⁶⁸.

Reverse transcription of RNA to cDNA was performed using 500 ng to 1.5 μg of RNA with SuperScript^™ III reverse transcriptase (Invitrogen) or iScript^™ reverse transcription supermix (Bio-Rad). qPCR was performed in triplicate for each sample using QuantiFast SYBR green PCR kit (Quiagen) on a MasterCycler ep Realplex (Eppendorf) with reverse transcriptase⁻ and water negative controls. Fetal human SCN5A primers were CCTTCACCGCCATTTACACC (forward) and GAAGAGCCGACAAATTGCCT (reverse) and adult human SCN5A primers were CCTTCACCGCCATTTACACC (forward) and GCCCAGGTCCACAAATTCAG (reverse) resulting in 168 and 152 base pair amplicons respectively. Primers were confirmed to have efficiencies between 0.90 and 1.10. Ct values greater than 32 were discarded to avoid amplification of genomic contamination. Data was acquired and analyzed using MasterCycler ep Realplex software (Eppendorf).

Electrophysiology

Single cell patch-clamping

EBs were dissociated and plated as single cells 3 days prior to recording as described above. Pipettes were pulled from filamented borosilicate glass (1.5 mm OD, 1.12 mm ID, World Precision Instruments) using a Flaming/Brown pipette puller with a box filament (p-87, Sutter Instrument Company) and were heat polished to between 3 and 6 MΩ resistance. Pipettes were filled with intracellular solution, inserted into a pipette holder with a platinum wire and mounted on a headstage (CV 203BU, Axon Instruments) on the stage of an inverted microscope (Olympus IX70). The position of the pipette was controlled by a micromanipulator (Burleigh PCS-250 system). An agar-KCl salt bridge was used as the reference electrode. Membrane potential was controlled through an Axopatch 200B voltage-clamp amplifier (Axon Instruments). Data was digitized (Axon Digidata 1322A), acquired using Clampex software (pClamp 8/9, Molecular Devices) and subsequently analyzed using Clampfit software (pClamp 10, Molecular Devices).

All whole-cell patch-clamp recordings of single cells were performed at room temperature. For all recordings, myocytes were initially perfused with tyrodes (Supplemental Table 1, Extracellular solution 1). APs were recorded in tyrodes. For I_Na recordings, following gigaseal formation and membrane rupture in tyrodes, cells were locally perfused with extracellular solution 2 (Supplemental Table 1) through PE90 tubing mounted on a micromanipulator and brought close to the cell of interest. For Ω-pore current recordings, following membrane rupture, cells were locally perfused with extracellular solution 3 (Supplemental Table 1) with cesium as the primary charge carrier followed by extracellular solution 4 (Supplemental Table 1) to block Ω-pore currents with NMDG. Intracellular solutions for AP, I_Na and Ω-pore current recordings were solutions 1, 2 and 3 respectively (Supplemental Table 1). Nifedipine containing solutions were protected from light.

Current- and voltage-clamp configurations were used to record APs and ionic currents respectively. Considering that our cells display immature properties including relatively depolarized MDPs, APs were recorded both following 1 second long anode breaks, which involve the injection of negative current to hold cells at −80 to −90 mV (the RMP of mature CMs), to allow all channels to recover from inactivation and during spontaneous firing. Na⁺ current-voltage curves and the voltage-dependence of activation were investigated using 20 ms steps from a holding potential of −100 mV to voltages between −100 and +20 mV in 5 mV increments. The voltage-dependence of inactivation was measured using 500 ms pre-pulses from a holding potential of −100 mV to voltages between −100 and −10 mV in 5 mV increments followed by a 50 ms test-pulse to −10 mV. Ω-pore currents were recorded using 160 ms steps from a holding potential of −120 mV to voltages between −100 and +40 mV in 5 mV increments. Whole-cell capacitance was measured by integrating the area under the capacitance transient in response to a depolarizing step from a holding potential of −50 mV to −40 mV and dividing by the amplitude of the step (10 mV). Following capacitance recordings and prior to ionic current recordings, cell capacitance and series resistance were compensated by 60–80% for all voltage-clamp recordings. Acquisition rates were 10 kHz and 50 kHz for current- and voltage-clamp studies respectively. Current density was calculated by dividing the current amplitude by whole-cell capacitance. P/4 leak subtraction was performed for Ωpore current analysis. Conductance (G) was calculated using: G=I/(V_m-E_rev)) where G is the conductance in Siemens (S), I is the current measured during pulses for activation curves and test-pulses for inactivation curves, V_m is the voltage of the pulse or test-pulse and E_rev is the reversal potential. Theoretically, the reversal potential should be 0 mV ([Na⁺]_i=[Na⁺]_o=10 mM), however linear regression analysis was preformed on the linear portion of current-voltage curves to determine the actual E_rev for each cell. Conductance was normalized to G_max with G_max determined by fitting the non-normalized conductance with a Boltzmann function for each cell.

Disassociating 4 months-old biowires

To disassociating 4 month-old biowires, each tissue was incubated at 37C with rocking motion for 20 minutes in 100ul of the digestion solution, which was made with 2000 unit/ml hyaluronidase (sigma) , 0.25 unit/ml of dispase (stem cell technologies), 100unit /mL of collagenase type II (Worthington), 1μM rock inhibitor (Selleck Chemicals) ,and 10μg/ml DNase I (Calbiochem) in HBSS (14Invitrogen). At the end of incubation, mixture was pipetted up and down for 5-10 times and was added with equal volume of DMEM supplemented with 10% FBS containing 1uM of rock inhibitor and 10μg/ml DNase I. Cell mixture was then centrifuged at 400g for 5 minutes and was resuspended in 50 μL StemPro-34 media (Gibco) containing 10μg/ml rock inhibitor (Selleck Chemicals) and 10% FBS. The cells were then plated on Matrigel coated cover slides. The next day cover slide was submerged in supplemented StemPro-34 media (Gibco).

Intracellular AP recording

Intracellular AP recordings of CSs 14-15 days after plating (AT day 34-35) and biowires 8 weeks after seeding were conducted in StemPro-34 media (Gibco) containing penicillin/streptomycin (1%, Gibco), L-glutamine (1%, Gibco), ascorbic acid (50 μg/ml, Sigma), MTG (50 μg/ml, Sigma). Pipettes were pulled from filamented borosilicate glass (1.5 mm OD, 0.75 mm ID, Sutter Instrument Company) using a Flaming/Brown pipette puller with a box filament (p-87, Sutter Instrument Company) to a resistance between 60 and 90 MΩ. Pipettes were filled with 3M KCl and mounted on a headstage connected to a Duo 773 Electrometer amplifier (World Precision Instruments). To minimize biowire movement, constructs were incubated with blebbistatin (10μM, Toronto Research Chemicals) for 20 minutes prior to AP recordings. Intracellular APs were acquired by placing the microelectrodes into the cell sheet or biowire while point stimulating using a FHC 6i stimulator (FHC- Neural microTargeting Worldwide). Data was digitized (Axon Digidata 1322A) and acquired using Clampex software (pClamp 8/9, Molecular Devices) at a rate of 10 kHz of using the gap-free configuration. Data was analyzed using Clampfit software (pClamp 10, Molecular Devices).

Optical mapping

Optical mapping studies were performed on CSs 14-15 days after plating (AT day 34-35) and biowires 8 weeks after seeding by loading these constructs with Fluo-4 AM (10 μM, Molecular Probes) for a minimum of 15 minutes and mapping in StemPro-34 media (Gibco) containing penicillin/streptomycin (1%, Gibco), L-glutamine (1%, Gibco), ascorbic acid (50 μg/ml, Sigma), MTG (50 μg/ml, Sigma) at 37 degrees Celsius. A complementary metal oxide semiconductor (CMOS) BrainVision Ultima camera system (SciMedia) was used to acquire data using BrainVision software. The frame rates were 200-500 Hz for CSs and 5 kHz for biowires. The camera was installed on an Olympus MVX10 microscope. Optical zoom was set at 2X for CSs and 4X for biowires, giving a field of view of 5x5 mm (50 μm resolution) and 2.5x2.5 mm (25 μm resolution), respectively. Preparations were point stimulated with 10 ms, 4V bipolar pulses through a pair of twister AWG#30 silver wires mounted on a micromanipulator and connected to a S88X Square Pulse Stimulator (Grass Technology). CVs were measured during 1 Hz pacing and MCRs were measured by increasing the stimulation rate during optical mapping until capture was lost. Arrhythmia induction was attempted 3 times per preparation using overdrive pacing (12V, 20 Hz for 30 seconds). CV was calculated using CV=(FPS x Distance)/#Frames with FPS being the frames acquired per second, distance being the distance between 2 points and #Frames being the number of frames between the take-off of calcium transients at point 1 to that at point 2.

MCR was measured as the highest rate at which 1:1 capture was maintained without alternans. Analysis was performed using BrainVision and a custom software written in Matlab (Mathworks). Activation maps were generated using custom software written in IDL (Harris Geospatial).

Calcium transient recordings

Biowires

Initial characterization of calcium handling in biowires 8 weeks after seeding was preformed using a single wavelength dye by loading constructs with Fluo-4 NW according to the manufacturer’s protocol (Molecular probes) and incubating at 37°C for 30 min. Recordings were generated in a custom environmental chamber set to 37 degrees Celsius and 5% CO₂ enclosing a CKX41 inverted microscope (Olympus). Tissue was paced at 1 Hz through carbon rods, as described above and calcium transient recordings were generated using the green light channel (Excitation/Emission=490/525 nm) at 4X magnification as described previously³⁸. The Stacks plugin in ImageJ software (National Institute of Health (NIH)) was used to determine the average intensity of a region of interest (ROI) and data was acquired using CellSens software (Olympus). To avoid movement artefacts, the chosen ROI was relatively distant from the POMaC wires. Transient amplitudes and kinetics following normalization while pacing at 1 Hz were analyzed using Clampfit software (pClamp 10, Molecular Devices).

Subsequent measurements of calcium handling were performed using Fura-2 AM to quantify diastolic calcium levels in our biowires. The Fura-2 AM solution (2 mM, Molecular probes) was prepared in DMSO with pluronic F127 (20%, Sigma) heated to 40 degrees Celsius. This solution was diluted to 0.05% pluronic F127 with 5 μM Fura-2 AM in StemPro-34 media (Gibco) containing penicillin/streptomycin (1%, Gibco), Lglutamine (1%, Gibco), ascorbic acid (50 μg/ml, Sigma), MTG (50 μg/ml, Sigma) for 30 min at 37 degrees Celsius, which was sonicated for 10 minutes prior to staining. Biowires were incubated with the staining solution for 1.5 hours at 37 degrees Celsius and washed 3 times with the Stempro media above before transferring to a glass bottom dish and placing a coverslip on top to prevent movement. Biowires were perfused with Krebs solution containing (in mM) 120 NaCl, 5.4 KCl, 1 MgCl₂, 1.2 CaCl₂, 19 NaHCO₃, 10 HEPES, 10 glucose, 1 NaPyruvate, pH=7.3 with NaOH heated to 37 degrees Celsius. Following transient recording, biowires were perfused with 2 mM Mn²⁺ in calcium free Krebs solution to quench the Fura-2 AM signal and measure background fluorescence. Biowires were field stimulated at 1 Hz through 2 platinum electrodes connected to a MyoPacer (IonOptix) with 5 ms bipolar pulses with an amplitude 1.5x the excitation threshold (typically 30V). Fura-2 AM fluorescence was measured by illuminating CMs via the rear light port of an Olympus IX70 microscope, alternating every 2 ms between 340 nm and 380 nm light (10 nm band-pass) originating from a 75 W xenon arc lamp. The light was projected to the CMs by a 40X objective (UApo/340, 40X/0.9NA, Olympus). The emitted light at 510 nm (±20 nm) from the CMs was projected onto a Photomultiplier tube400 (PMT400, IonOptix) and acquired using IonWizard software (IonOptix). Background subtraction was performed using: R_340/380=(F₃₄₀ – F₃₄₀background)/(F₃₈₀ – F₃₈₀background) where F₃₄₀/F₃₄₀background and F₃₈₀/F₃₈₀background are the fluorescence intensities at 510 nm when illuminated with 340 nm and 380 nm wavelengths respectively prior to/after quenching measured in the same part of the cell. Background subtraction was performed and transient amplitudes analyzed using IonWizard software (IonOptix) and kinetics of normalized transients were analyzed using Clampfit software (pClamp 10, Molecular Devices).

Force measurements

To characterize the functional properties of our biowires, 8 weeks after seeding we measured active force generation as described previously³⁸. Briefly, force measurements were generated in a custom environmental chamber set to 37 degrees Celsius and 5% CO₂ enclosing a CKX41 inverted microscope (Olympus). 4X bright field videos were recorded to measure the tissue dimensions necessary for the force calculation, normalization and tissue compaction. Force was measured by recording fluorescent videos (100 FPS, 10x magnification) of the POMaC wire (which have intrinsic fluorescent properties, Excitation/Emission=350/470 nm) movement while pacing from 1-3 Hz as described above. Custom MatLab code allowed for accurate tracing of the deflection of the POMaC wires during systole and diastole. Total (systolic) and passive (diastolic) wire deflections were converted to force measurements using the force calibration curves described previously³⁸ with the active force being the difference between the total and passive forces.

RNA sequencing and gene set enrichment analysis

RNA was isolated using a commercially available kit: PicoPure^™ RNA Isolation Kit (Thermo Fisher, KIT0204) and RNase-Free DNase Set (Qiagen #79254).

Isolated RNA sample quality was assessed by High Sensitivity RNA Tapestation (Agilent Technologies Inc., California, USA) and quantified by Qubit 2.0 RNA HS assay (ThermoFisher, Massachusetts, USA). Paramagnetic beads coupled with oligo d(T)25 are combined with total RNA to isolate poly(A)+ transcripts based on NEBNext^® Poly(A) mRNA Magnetic Isolation Module manual (New England BioLabs Inc., Massachusetts, USA). Prior to first strand synthesis, samples are randomly primed (5´ d(N6) 3´ [N=A,C,G,T]) and fragmented based on manufacturer’s recommendations. The first strand is synthesized with the Protoscript II Reverse Transcriptase with a longer extension period, approximately 30 minutes at 42°C. All remaining steps for library construction were used according to the NEBNext^® Ultra^™ II Non-Directional RNA Library Prep Kit for Illumina^® (New England BioLabs Inc., Massachusetts, USA). Final libraries quantity was assessed by Qubit 2.0 (ThermoFisher, Massachusetts, USA) and quality was assessed by TapeStation HSD1000 ScreenTape (Agilent Technologies Inc., California, USA). Final library size was about 400bp with an insert size of about 250bp. Illumina^® 8-nt dual-indices were used. Equimolar pooling of libraries was performed based on QC values and sequenced on an Illumina^® NovaSeq S4 (Illumina, California, USA) with a read length configuration of 150 PE for 40 M PE reads per sample (20M in each direction).

FASTQ files were trimmed of adaptor primer sequences using cutadapt ⁶⁹ and filters for quality. Trimmed and filtered reads were aligned against the human genome build GRC37 using HTSEQ2⁷⁰. SAM files were converted into BAM format, sorted and indexed using samtools⁷¹. Counts were made against exons and genes and splice junctions using the package Rsubread function featureCounts. A raw count matrix was assembled into a DEGList object in R, low read genes were removed and data was normalized using TMM in edgeR⁷². For differential expression a voom transformation was applied to remove heteroscedasity and linear models applied. A principal component was calculated using the prcomp function in R and plotted using the autoplot function from ggfortify package in R.

Gene set enrichment was calculated using the function camera form the limma package against the Gene Ontology Biological Process and Cellular Component gene sets obtained from the Bader lab (download.baderlab.org/EM_Genesets/current_release). Sets of cell-specific genes were generated from the PanglaoDB database (https://panglaodb.se/). These were converted into gene sets and assessed for enrichment in comparisons of time and mutation status samples using the camera function from the package limma⁷³ in R. An FDR corrected p-value of <0.1 was considered significant. Barcode plots were generated using the function barcode plot from the package limma in R. The heatmaps of specific genesets was generated using bespoke R scripts and the pheatmap library. Output from camera gene set enrichment analysis was formatted as a generic table format for graphing and analysis in Cytoscape⁷⁴ using bespoke R scripts. Network graphs of gene set enrichments were generated in Cytoscape using Enrichmentmap⁷⁵. Sub-networks were named using clustermaker and word cloud annotating for enrich words with a bonus for adjacent words.

The set of SCN5A interacting protein was obtained from the STRING database⁷⁶ (https://string-db.org/). Direct and 1^st nearest neighbor interactions were used as the interactome set.

Histology

Fixation, embedding and sectioning

Single cells plated at day 20 as detailed above were fixed 1, 3 or 5 days after plating with paraformaldehyde (PFA) (4%) in phosphate buffered saline (PBS) for 10 minutes at RT.

Biowires were fixed with PFA (0.5%) in PBS overnight at 4 degrees Celsius followed by sectioning and embedding. Preparations were dehydrated by incubating in increasing concentrations of ethanol followed by xylene (70%->80%->95%->100%>100%->1:1 ratios of 100% ethanol to xylene->xylene->xylene, 1 hour each) at RT. Incubations in ethanol were performed on an orbital shaker set at a low speed for gentle agitation. Biowires were then incubated in paraffin wax overnight at 58-60 degrees Celsius (Wax 1), the wax was replaced and biowires were incubated for 1 hour at 58-60 degrees Celsius (Wax 2) followed by embedding in an orientation to generate cross sections. 5 μm thick cross sections 600-800 μm along the longitudinal axis of each biowire were generated using a HM 325 Rotary Microtome (ThermoFisher).

Staining

Single cells and biowire sections prepared as described above were stained to visualise TnT expression and sarcomere alignment (single cells only). For TnT expression, fixed cells were washed with tris buffered saline (TBS) (50 mM Tris, 150 mM NaCl, pH=7.5) (3x5 min on an orbital shaker). Permeabilization was performed by incubating in TBS with TritonX-100 (0.2%) for 30 minutes on an orbital shaker. Blocking was performed by incubating in TBS with TritonX-100 (0.1%) and bovine serum albumin (BSA) (1%, Sigma) for 1 hour on an orbital shaker. Following blocking, cells were incubated with mouse anti-human cTnT primary antibody (1:125, ThermoFisher) in blocking buffer overnight at 4 degrees Celsius. Cells were washed with TBS with TritonX-100 (0.1%) (3x15 min on an orbital shaker) before incubating with goat anti-mouse AlexaFluor 647 secondary antibody (1:125, ThermoFisher) for 1.5 hours at RT. Samples were washed with TBS with TritonX-100 (0.1%) (3x15 min on an orbital shaker) before incubating with DAPI (1:1000, ThermoFisher) for 5 minutes at RT on an orbital shaker. Finally, samples were washed with TBS with TritonX-100 (0.1%) (3x15 min on an orbital shaker) and stored in TBS until imaging.

Biowire sections were deparaffinised by incubating in xylene followed by decreasing concentrations of ethanol (xylene->xylene->1:1 ratios of 100% ethanol to xylene->100% ethanol->100% ethanol->95% ethanol->70% ethanol, 5 minutes each). Antigen retrieval was performed by placing slides in boiling antigen retrieval buffer containing 10 mM sodium citrate and 0.05% tween (pH=6.0) in a pressure cooker and incubating for 3 minutes beginning when the cooker reached full pressure. Biowire sections were washed with TBS with TritonX-100 (0.025%) (3x5 min on an orbital shaker) followed by blocking with TBS containing TritonX-100 (0.025%) and BSA (1%, Sigma) for 1 hour at RT. Biowires were incubated with mouse anti-human cTnT primary antibody (1:100, ThermoFisher) in blocking buffer for 1 hour at RT. Sections were washed with TBS with Triton X-100 (0.025%) (3x5 min on an orbital shaker) followed by incubation with goat anti-mouse AlexaFluor 647 secondary antibody (1:250, Abcam) for 1 hour at RT. Sections were washed with TBS with Triton X-100 (0.025%) (3x5 min on an orbital shaker) followed by mounting with mounting media with DAPI (Abcam).

For α-actinin (Abcam) and MLc2v (Abcam) staining, the whole biowires tissues (fixed with 4% PFA) were blocked and permeabilized with PBS (0.1% Triton X and 5% FBS) for 1 hour. The tissues were then stained with primary antibody (1:200) overnight at 4C, washed with PBS 3 times, followed by secondary antibody, Goat anti mouse 647 (Abcam) at room temperature for 2 hours. F-actin was stained with conjugated antibody (Phalloidin 488, ThermoFisher) for 2 hours. DAPI (Abcam) were used for all counterstaining.

Confocal imaging

Imaging of antibody and PSR stained single cells and biowire sections and SHG of intact biowires was performed using a Zeiss LSM710 Two-Photon/Confocal in the Advanced Optical Microscope Facility (AOMF) with a 20x/1.0 NA Water Immersion.

Zen software (Zeiss) was used for data acquisition. Images of intact antibody stained biowires were taken by Nikon AR1 with 60x oil Immersion.

Quantification of TnT expression deposition and sarcomere alignment

Fiji/Image J were to quantify of TnT expression using thresholding to determine the number of TnT positive pixels and normalize to the total/cross sectional area of the cells/biowires. To quantify sarcomere orientation in cells dissociated from EBs and plated for 5 days before fixation, we performed blinded analysis of 6 differentiations of iPSC derived ventricular CMs from each cell line with 4-11 images per differentiation to characterize sarcomeres as organized or disorganized (disarray) as done previously^77–79.

For sarcomere alignment quantification in biowires IHCs, α-actinin, actin, cTNT and MLC2V image analysis was done by an in-house written Matlab code. Briefly, images were converted to a binary image, followed by the regionprops function, which provides a set of geometry properties of the various structures in the image. Eccentricity was presented in the manuscript as the percentage of elements that reach an eccentricity value of 0.95-1. Density was calculated by the number of pixels divided by the total number of pixels in the image and this number was then normalized to cell number by dividing it by the number of nuclei in the images.

Flow cytometry

At day 3, EBs were dissociated into single cells using TrypLE^™ Express (1x), Phenol Red (Gibco) for 2-5 minutes at RT. Day 20 EBs were dissociated as described above. day 3 cells were stained with anti-PDGFRα-PE (1:20, BD Biosciences) and either CD56-APC (1:100, BD Biosciences) or anti-CD235a-APC (1:100, BD Biosciences) in FACS buffer consisting of PBS with fetal bovine serum (FBS) (5%, Fisher/Hyclone) for 15 minutes at RT or 30 min at 4 degrees Celsius. Day 20 cells were fixed for 10 min at RT with PFA (4%) in PBS followed by permeabilization using methanol (90%) in PBS for 10 min at 4 degrees Celsius. Cells were stained using mouse anti-human cTnT (1:2000, ThermoFisher) and rabbit anti-human MLC2v (1:1000, Abcam) primary antibodies in FACS buffer overnight at 4 degrees Celsius. Cells were then stained with secondary antibodies including goat anti-mouse IgG-APC (1:500, BD Biosciences) and donkey anti-rabbit IgG-PE (1:500, Jackson ImmunoResearch) for 1 hour at RT. Stained cells were analyzed using the LSR II Flow cytometer (BD) at the SickKids/UHN flow cytometry facility/Keller laboratory. Data were acquired using BD FACSDiva^™ (BD Biosciences) and analyzed using FlowJo software (Tree Star).

Statistics

Statistical analysis was performed using GraphPad Prism (7.0, GraphPad Software Inc.). Statistical significance was established using a unpaired or paired student’s t-test (two-tailed) as appropriate, a one-way ANOVA with Sidak’s multiple comparison test or a repeated measure two-way ANOVA with Tukey multiple comparison test. Data are presented as mean±SEM. P<0.05 was considered significant for all statistical tests.

Key resource table : RNAseq resource tables

Reagent or resource	Source	Catalogue #
Antibodies
Donkey anti-rabbit IgG, PE conjugated	Jackson ImmunoReserch	711-116-152
Donkey anti-rabbit IgG, AlexaFluor 647	Thermo Fisher Scientific	A32795
Goat anti-mouse, AlexaFluor 647 conjugated	Abcam	A21235
Goat anti-mouse, Alexa 488 conjugated	Thermo Fisher Scientific	A11029
Goat anti-mouse, APC conjugated	BD Biosciences	550826
Moues anti-PDGFRα, PE conjugated	BD Biosciences	556002
Mouse anti-CD235a, APC conjugated	BD Biosciences	551336
Mouse anti-CD56, APC conjugated	BD Biosciences	555518
Mouse anti-human cTnT	Thermo Fisher Scientific	MS-295-P1
Mouse anti-cardiac troponin T isoform Ab1	Thermo Fisher Scientific	MA512960
Rabbit anti-human MLC2v	Abcam	ab79935
Phalloidin 488	Thermo Fisher Scientific	A12379
Mouse anti-α-actinin	Abcam	Ab9465
Biological samples
Human cardiac fibroblasts	PromoCell	C-12375
Commercial assays
iScript™ reverse transcription supermix	Bio-Rad	1708841
SuperScript™ III reverse transcriptase	Invitrogen	18080085
QuantiFast SYBR green PCR kit	Qiagen	204057
RNAqueous™-Micro total RNA isolation kit	Ambion	AM1931
Chemicals, peptides and proteins
1x Medium 199	Sigma	M0650
Activin A	Bio-Techne	338-AC-01M/CF
Ascorbic acid	Sigma	A4544-100G
Basic fibroblast growth factor	Bio-Techne	233-FB
Bone morphogenic protein 4	Bio-Techne	314-BP
Bovine serum albumin	Sigma	A1470
Collagenase Type II	Worthington	LS004176
DAPI	ThermoFisher	62248
Dofetilide	Sigma	PZ0016
Dulbecco’s modified eagle media/F12	Cellgro	MT10092CV
Fetal bovine serum	Fisher/HyClone	SH3039603Q
Flecainide	Sigma	F6777
Fluo-4 NW	Molecular Probes	F36205
Fura-2 AM	Molecular Probes	F14201
Gelatin	Sigma	G1890
Glutamax	Gibco	35050061
Growth factor reduced Matrigel	Fisher/Corning	354230
Iscove’s modified Dulbecco’s media	Gibco	12200069
IWP2	Bio-Techne	3533/10
KnockOut™ serum replacement	Gibco	10828028
L-ascorbic acid	Sigma	A8960
L-glutamine	Gibco	25030081
Monothioglycerol	Sigma	M6145-25ML
Mounting media with DAPI	Abcam	ab104139
Non-essential amino acids	ThermoFisher	11140-050
Penicillin/streptomycin	Gibco	15070063
Pluronic F127	Sigma	P2443
Poly(2-hydroxyethyl methacrylate)	Sigma	P3932-25G
Polyurethane 2-part adhesive	GS Polymers, Inc.	GSP 1552-2
Collagen I, Rat tail, High concentration	Corning	354249
Rho-associated protein kinase inhibitor Y-27632	Bio-Techne	1254/50
(S)-(−)-Blebbistatin	Toronto Research Chemicals	B592500
StemPro-34 media	Gibco	10639011
Transferrin	Sigma/Roche	10652202001
TrypLE™ Express (1x), Phenol Red	Gibco	12605010
Vascular endothelial growth factor	Bio-Techne	293-VE
β-Mercaptoethanol	Gibco	21985-023
Hyaluronidase	Sigma	H3506
Dispase	Stem Cell Technologies	07913
HBSS	Invitrogen	14175-103
DNase I	Calbiochem	260913
Software
BD FACSDiva™	BD Biosciences	https://www.bdbiosciences.com/en-us/instruments/research-instruments/research-software/flow-cytometry-acquisition/facsdiva-software
BrainVision software	SciMedia	http://www.scimedia.com
CellSens software	Olympus	https://www.olympus-lifescience.com/en/software/cellsens/
FlowJo	Tree Star	https://www.flowjo.com
GraphPad Prism 7.0	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
IDL	Harris Geospatial	https://www.harrisgeospatial.com/Software-Technology/IDL
Matlab	Mathworks	https://www.mathworks.com/products/matlab.html?s_tid=hp_products_matlab
MasterCycler ep Realplex software	Eppendorf	https://www.qiagen.com/ca/resources/resourcedetail?id=adcdeee3-98c7-43e8-987f-a84e4cfe01d2&lang=en
pClamp 8, 9 and 10	Molecular Devices	https://www.moleculardevices.com/products/axon-patch-clamp-system/acquisition-and-analysis-software/pclamp-software-suite
Zen software	Zeiss	https://www.zeiss.com/microscopy/int/products/microscope-software/zen.html

Data Availability

The raw/processed data required to reproduce these findings cannot be shared at this time due to technical or time limitations.