ABSTRACT
Hypertrophic cardiomyopathy (HCM) is an inherited heart muscle disease that is characterised by left ventricular wall thickening, cardiomyocyte disarray and fibrosis, and is associated with arrhythmias, heart failure and sudden death. However, it is unclear to what extent the electrophysiological disturbances that lead to sudden death occur secondary to structural changes in the myocardium or as a result of HCM cardiomyocyte electrophysiology. In this study, we used an induced pluripotent stem cell model of the R403Q variant in myosin heavy chain 7 (MYH7) to study the electrophysiology of HCM cardiomyocytes in electrically coupled syncytia, revealing significant conduction slowing and increased spatial dispersion of repolarisation – both well-established substrates for arrhythmia. Analysis of rhythmonome protein expression in MYH7 R403Q cardiomyocytes showed reduced expression of connexin-43 (also known as GJA1), sodium channels and inward rectifier potassium channels – a three-way hit that reduces electrotonic coupling and slows cardiac conduction. Our data represent a previously unreported, biophysical basis for arrhythmia in HCM that is intrinsic to cardiomyocyte electrophysiology. Later in the progression of the disease, these proarrhythmic phenotypes may be accentuated by myocyte disarray and fibrosis to contribute to sudden death.
INTRODUCTION
Hypertrophic cardiomyopathy (HCM) is an autosomal-dominant inherited cardiac disorder with a prevalence of up to one in 200 that can result in arrhythmias, heart failure and sudden death (Ommen et al., 2020; Ommen and Semsarian, 2021). Disease-causing variants associated with HCM most commonly occur in sarcomere genes, including myosin heavy chain 7 (MYH7), myosin binding protein C3 (MYBPC3) and troponin T (TNNT2), with variants in the MYBPC3 gene being the most common cause (Chiswell et al., 2023). HCM is characterised clinically by left ventricular hypertrophy (wall thickness ≥15 mm) in the absence of loading conditions (Elliott et al., 2014; Writing Committee Members et al., 2020) and is also associated with fibrosis, myocyte disarray and altered energy metabolism (Dhalla et al., 2006). Electrical conduction delays and dispersion of repolarisation, both risk predictors for arrhythmias due to an increased risk of re-entry (King et al., 2013), are also clinical features of HCM (Magrì et al., 2017). However, the correlation between these conduction defects and the histopathology of HCM is limited (Aryana et al., 2007; Wolf et al., 2005; Hueneke et al., 2017). As a result, the mechanisms underlying electrical dysfunction in HCM and its role in sudden death in patients is unclear.
The missense arginine-to-glutamine substitution at position 403 in MYH7 (R403Q) is an HCM variant that causes severe disease, characterised by early-onset and progressive myocardial dysfunction, with a high incidence of sudden cardiac death (Geisterfer-Lowrance et al., 1990; Viola and Hool, 2019). Mice homozygous for this variant (αMHC403/403) exhibit neonatal lethality, whereas heterozygous mice (αMHC403/+) also demonstrate cardiac dysfunction, myocyte disarray, hypertrophy and fibrosis (Geisterfer-Lowrance et al., 1996). However, consistent with some observations in patient cohorts, neither the extent nor the location of fibrosis correlated with electrical mapping of conduction properties in this mouse model (Wolf et al., 2005). Furthermore, these histopathological changes also did not correlate with the propensity for arrhythmia. It has also been reported that tachyarrhythmias are observed at a far earlier age than the onset of hypertrophy (Hueneke et al., 2017), and that dysfunction of cardiac calcium current and alteration of mitochondrial metabolism occurs prior to the onset of myopathy (Viola and Hool, 2019). Each of these pieces of evidence suggests an alternative pathway for arrhythmic substrate formation in HCM, such as electrical remodelling, that is at least partly independent of alterations in the structure of the myocardium.
In this study, we used a patient-derived induced pluripotent stem cell (iPSC) model of the MYH7 R403Q variant (MYH7403/+) (Holliday et al., 2018) to investigate whether there are intrinsic electrical properties of HCM cardiomyocyte electrophysiology that provide a biophysical basis for arrhythmia in the absence of structural alteration of the myocardium. Our data show for the first time that in MYH7403/+ cardiomyocytes, a dramatic reduction in the expression of connexin-43 (also known as GJA1) and sodium channel proteins results in reduced conduction velocity. Accompanying this is an increase in spatial dispersion of repolarisation that establishes potential proarrhythmic substrates and may provide a biophysical basis that contributes to sudden arrhythmic death in patients with HCM.
RESULTS
Generation of iPSC lines for MYH7403/+ and isogenic control
An iPSC line was reprogrammed from a 38-year-old female patient with HCM (II:1, Fig. 1A) with the pathogenic p.Arg403Gln variant in myosin heavy chain 7 (MYH7403/+) as previously described (Holliday et al., 2018; includes quality control assessment of karyotype, pluripotency and trilineage differentiation capacity). For this study, we generated an isogenic control line by correcting the pathogenic variant in MYH7 using CRISPR-Cas9 genome editing (MYH7+/+) (Fig. 1B). Sanger sequencing confirmed the A>G correction at c.1280G>A (Fig. 1C) and showed that no unwanted edits were present in the top ten predicted guide RNA off-target sites. MYH7+/+ iPSC lines successfully generated colonies of iPSCs with classical morphology of tightly packed cells with a high cell-to-nucleus ratio (Fig. 1D). In all subsequent experiments, this CRISPR-corrected line was used as the comparator for functional and molecular assessments.
MYH7403/+ modifies electrophysiology of iPSC-derived cardiomyocytes
Cardiomyocytes derived from iPSCs (iPSC-CMs) for both MYH7403/+ and MYH7+/+ were seeded onto microelectrode arrays and extracellular field potentials – in vitro surrogates of electrocardiograms – were recorded from spontaneously beating monolayers of cells (Fig. 2A). To rule out any differences in cardiomyocyte purity between MYH7+/+ and MYH7403/+ differentiations, flow cytometry was performed to measure ACTN1- and CD90 (also known as THY1)-positive populations, reflecting cardiomyocytes and stromal cells, respectively (Fig. S1). Both MYH7+/+ and MYH7403/+ iPSC-CMs showed ∼90% expression of ACTN1 [MYH7+/+, 95.47±1.05%; MYH7403/+, 90.40±3.15%; N=3 differentiations], with no significant difference between lines. The R403Q variant affected both depolarisation and repolarisation properties of the cardiomyocytes. Specifically, the slope of the depolarisation spike of the field potential (Fig. 2B), a measure of the propagating action potential upstroke, was reduced from −0.14±0.02 V/s in MYH7+/+ iPSC-CMs to −0.04±0.02 V/s in MYH7403/+ iPSC-CMs (Fig. 2C), and the Fridericia-corrected field potential duration (FPDc; Fig. 2D) was increased from 281.0±21.6 ms in MYH7+/+ iPSC-CMs to 318±21.5 ms in MYH7403/+ iPSC-CMs, reflecting slowed repolarisation of the cardiomyocytes (consistent with our previous studies showing prolonged action potential duration in MYH7403/+ cardiomyocytes; Cserne Szappanos et al., 2023). In addition, MYH7403/+ iPSC-CMs displayed more irregular/arrhythmic beating (Fig. 3), with a significant increase in the coefficient of variation of beat rate to 17.8±5.2% from 7.0±5.2% in MYH7+/+ iPSC-CMs (Fig. 3C). For MYH7403/+ iPSC-CMs, a variety of arrhythmic phenotypes was observed, ranging from more subtle beat-to-beat variability in cycle length (Fig. 3Aii,Bii) to the presence of ectopic depolarisations, often tightly coupled to regular/sinus beats (Fig. 3Aiii,Biii), compared to those seen in MYH7+/+ iPSC-CMs (Fig. 3Ai,Bi). It should, however, be noted that some ectopic beats observed at slower spontaneous beating rates may be suppressed if monolayers were paced at faster heart rates. β-adrenergic stimulation with 100 nM isoproterenol did not increase beat rate variability for either MYH7+/+ or MYH7403/+ iPSC-CMs (Fig. S2).
Reduced electrical coupling in MYH7403/+ iPSC-CMs slows conduction and induces spatial dispersion of repolarisation
A reduced slope of the depolarisation spike (Fig. 2C) might reflect multiple electrophysiological phenomena, including cellular factors, such as a reduced density of sodium currents, or macroscopic electrophysiological properties, such as altered cell–cell coupling, leading to slowed propagation of the activating wavefront. We therefore next measured conduction velocity in monolayers of iPSC-CMs. Conduction maps for MYH7+/+ and MYH7403/+ iPSC-CMs, coloured according to activation time measured at individual electrodes across the microelectrode array, are shown in Fig. 4A. In these examples, the activation wavefront takes longer to propagate from the bottom right to the top left of the array in MYH7403/+ iPSC-CMs (25±1 ms) compared to in MYH7+/+ iPSC-CMs (13±1 ms) (Fig. 4B), reflecting slower conduction velocity. This slowed conduction velocity suggests that syncytia of MYH7403/+ iPSC-CMs are less tightly coupled than those of MYH7+/+ iPSC-CMs. In normal hearts, the level of electric coupling between cells attenuates differences in electrical properties between individual cells. In the context of reduced electrical coupling in MYH7403/+ iPSC-CMs, we measured the spatial dispersion of repolarization to examine whether intrinsic differences in the repolarization properties of cardiomyocytes in the syncytium are manifested, potentially establishing spatial voltage gradients that might act as a substrate for re-entrant arrhythmia. Field potential durations (FPDs) were measured at each individual electrode in the array (with an interelectrode distance of approximately 300 µm) (Fig. 5Ai), and a greater range of repolarization times was measured in monolayers of MYH7403/+ iPSC-CMs compared to that in MYH7+/+ iPSC-CMs (Fig. 5Aii). The spread of FPDs across the 16 electrodes in each array, measured from ten individual electrode arrays each for MYH7403/+ and MYH7+/+ iPSC-CMs is shown in Fig. 5B, demonstrating a consistently greater global dispersion of field potentials in MYH7403/+ monolayers. Overall, the standard deviation of FPDs within each electrode array increased from 11.4±1.4 ms to 22.5±1.3 ms for MYH7+/+ and MYH7403/+ iPSC-CMs, respectively (Fig. 5C). Finally, FPDs were mapped relative to the electrode position in the array, showing clear local dispersion of repolarization times of up to 40-50 ms in MYH7403/+ (Fig. 5Dii) but not MYH7+/+ iPSC-CMs (Fig. 5Di), potentially resulting in steep voltage gradients that could facilitate re-entrant arrhythmia.
Molecular basis of electrophysiological phenotype in MYH7403/+ iPSC-CMs
To investigate the molecular basis of the observed differences in electrical phenotype, we measured changes in mRNA and protein expression. First, for mRNA, we used a curated panel (nanoString nCounter) targeting cardiac ion channels, calcium-handling proteins and transcription factors. Our data show transcriptional downregulation of rhythmonome genes involved in calcium handling (PLN, RYR2 and SLC8A1) and cardiac repolarisation (KCNH2); sarcomere genes (MYL7, TTN, TNNI1 and TNNI3), and genes involved in ventricular cell fate (NKX2-5 and CORIN) (Fig. 6A,B). To complement this, we measured expression of rhythmonome proteins using western blotting (Fig. 6C,D), including connexin-43, Kv2.1 (also known as KCNB1), Kv1.5 (or KCNA5), Cav1.2 (CACNA1C), Kir6.2 (KCNJ11), Kv1.4 (KCNA4), Kir2.1 (KCNJ2), Kv4.2 (KCND2), SAP97 (DLG1), Nav1.5 (SCN5A), Kv7.1 (KCNQ1), KCNE1, Kv11.1 (KCNH2) and TASK1 (KCNK3). Western blotting showed reduced expression of key proteins involved in cardiac excitation and conduction, including 80±12%, 75±5%, and 40±10% reduction in Nav1.5 (the cardiac sodium channel), connexin-43 and Kir2.1 (inward rectifier potassium channel), respectively (Fig. 6D). This reduced expression of connexin-43 observed by western blotting was also confirmed by immunohistochemical analysis, in which fluorescence intensity per cell associated with connexin-43 was reduced 3-fold (Fig. 6E,F).
DISCUSSION
HCM is an inherited cardiac disorder that results in hypertrophy, fibrosis, myofibre disarray, arrhythmias and sudden cardiac death (Semsarian et al., 2015). Conduction delays and dispersion of repolarisation are clinically associated with HCM (Magrì et al., 2017; Cortez et al., 2017). However, the correlation between conduction defects and the histopathology of HCM is poor, meaning that the mechanism underlying these electrical phenotypes is unclear. In this study, we used an iPSC model of the MYH7 R403Q variant to show that reduced conduction velocity is associated with dramatic reduction in expression of connexin-43 and sodium channel proteins in cardiomyocytes – both key molecular mediators of conduction in the myocardium. Furthermore, we show that this reduced electrical coupling results in significant spatial dispersion of repolarisation – a well-established proarrhythmic substrate (Antzelevitch, 2007; Zahid et al., 2016) – which may provide a biophysical basis that contributes to sudden arrhythmic death in patients with HCM.
MYH7403/+ slows conduction in iPSC-CM monolayers
HCM is a disease that has been modelled in a range of systems including intact heart muscle strips, isolated cardiomyocytes, myofibrils, purified actomyosin and animal models (Mosqueira et al., 2019). More recently, iPSC-CM models of HCM related to variants in MYH7 (Lan et al., 2013; Dainis et al., 2020) as well as other genes (Cohn et al., 2019; Wu et al., 2019) have been shown to reproduce key elements of the HCM phenotype in vitro, including cellular morphology and/or hypertrophy (Cohn et al., 2019), elevated metabolism (Toepfer et al., 2019), disrupted calcium handling and hypercontractility (Lan et al., 2013; Wu et al., 2019). In relation to electrophysiology, data from iPSC models are less extensive and inconsistent. Studies of MYH7-related HCM models in iPSC-CMs report variable electrical phenotypes, with both unchanged or prolonged action potential duration reported, as well as irregular beating (Lan et al., 2013; Han et al., 2014). However, there are no in vitro studies using iPSC models that examine macroscopic electrophysiological properties that might provide a biophysical explanation for the conduction abnormalities seen in patients.
In this study, we used microelectrode arrays to record electrical signals from monolayers of iPSC-CMs, allowing us to study properties such as conduction velocity and repolarisation dispersion. Our data showed, for the first time, significantly slowed conduction velocity (approximately 50% reduction; Fig. 4) in an in vitro model of HCM. Reduced conduction velocity increases the chance of arrhythmia by reducing the spatial scale over which re-entry is possible (Janse and Wit, 1989). In the clinical setting, slow conduction has been reported in patients with HCM (Schumacher et al., 2005; Saumarez, 1994). Given the histopathology of the HCM myocardium, particularly regarding the presence of interstitial fibrosis, it is generally assumed that disruption of normal electrical propagation is a function of this fibrosis and the associated reduction in coupling between cardiomyocytes (Lyon et al., 2018; Varnava et al., 2001). However, several studies have suggested that this is not necessarily always the case. For example, in the clinical setting, Aryana et al. (2007) reported a lack of concordance between imaging of intramural fibrosis and low voltage on endocardial or epicardial mapping. Similarly, in αMHC403/+ mice, Wolf et al. (2005) demonstrated that neither the extent nor the location of fibrosis correlated with electrical mapping of conduction properties, nor did this correlate with the propensity for arrhythmia. Furthermore, in a separate mouse study, Hueneke et al. (2017) showed that tachyarrhythmias are observed at a far earlier age than the onset of hypertrophy, suggesting an alternative pathway for arrhythmogenesis in HCM that is independent of alterations to the structure of the myocardium. Our data are consistent with these observations and support a biophysical basis for conduction slowing in MYH7403/+ iPSC-CMs intrinsic to cardiomyocyte electrophysiology, which may be accentuated by myocyte disarray in addition to fibrosis and later in the progression of disease.
Reduced electrical coupling between MYH7403/+ iPSC-CMs results in increased spatial dispersion of repolarisation
Spatial dispersion or regional differences in repolarisation times facilitate the initiation and maintenance of re-entrant arrhythmias (Mines, 1913). This view is supported by studies showing that regional differences in repolarisation time lead to an increased susceptibility to arrhythmias in response to premature stimulation (Han and Moe, 1964; Clayton and Holden, 2005; Boersma et al., 2002). Similarly, previous work has shown a tight correlation between the range and variability in action potential duration across the tissue and the duration of the vulnerable window for initiation of re-entry (Clayton and Holden, 2005), whereas on patient electrocardiograms, measurements of T wave area, a metric that reflects dispersion of repolarisation, are predictive of ventricular arrhythmias (Nearing et al., 2012). Such spatial dispersion of repolarisation is thought to arise because of variation in ion channel gene expression or ion channel function in different regions of the heart. However, in normal healthy tissue, tight electrotonic coupling ensures that repolarisation of individual cells or different regions of the tissue is synchronised and, hence, is less vulnerable to arrhythmia (Lesh et al., 2018; Joyner, 1986).
In our iPSC model of MYH7 R403Q, we hypothesised that the observed reduction in conduction velocity reflected reduced electrical coupling, and that this might alter the degree to which repolarisation was synchronised between cells, manifesting in increased spatial dispersion of repolarisation. This was indeed the case, with significantly greater variability and range of FPDs (reflective of repolarisation time) measured across MYH7403/+ monolayers compared to their CRISPR-corrected controls (Fig. 5). Moreover, these voltage gradients occurred over small, local spatial scales. Differences of up to 150 ms in FPDs were recorded across electrode arrays spaced over areas just 1-2 mm across, with steep voltage gradients occurring even between neighbouring electrodes (see example in Fig. 5Dii). In relation to arrhythmogenesis, this is important as it is thought that local repolarisation dispersion (as opposed to global differences) often serves as a source for the ectopic beats that trigger arrhythmias (Dunnink et al., 2015; Tan et al., 2017). Such dispersion of repolarisation has been linked to arrhythmogenesis in a range of inherited and acquired disorders including long QT syndrome (Antzelevitch, 2007; Lachaud et al., 2022), Brugada syndrome (Antzelevitch, 2007), heart failure (McIntosh et al., 2000) and post-ischaemia (Gough et al., 2018), in which voltage gradients occur variously due to factors including regional differences in ion channel expression as well as electrical uncoupling following gap junction downregulation. Specifically, in relation to HCM, Saumarez et al. (2018) reported fractionated conduction related to myocardial disarray and fibrosis on patient electrograms, whereas Hueneke et al. (2017) reported dispersion of repolarisation on a global scale, related to differential expression of potassium channel genes in different regions of the mouse heart. Our observations of highly discordant local repolarisation in MYH7403/+ monolayers are therefore consistent with a mechanism that could contribute to arrhythmic susceptibility in patients. To our knowledge, this is the first report of such pronounced local spatial dispersion of repolarisation in a model of HCM that results from the biophysical properties of the HCM cardiomyocyte.
Molecular basis of electrophysiological changes in MYH7403/+ iPSC-CMs
Conduction of electrical signals in the ventricular myocardium is regulated primarily by voltage-gated sodium channels (e.g. Nav1.5), which are the main contributors to cardiomyocyte depolarisation, and gap junctions (primarily connexin-43 in the ventricle) that allow the flow of current between coupled cells. Loss of either of these critical contributors to cardiac conduction can disrupt regular electrical propagation and increase the risk of arrhythmias. Our analysis of protein expression showed a dramatic (∼80%) reduction in the expression of both connexin-43 and Nav1.5 proteins in MYH7403/+ cardiomyocytes compared to that in their CRISPR-corrected controls. In addition to this, we also measured a ∼50% reduction in the levels of the Kir2.1 protein (an inward rectifier potassium channel). Functionally, this reduction in Kir2.1 levels would serve to depolarize the resting membrane potential, leading to functional inactivation of the remaining Nav1.5 population, thus acting as a potential ‘third hit’ on electrical conduction. Although we cannot be certain as to why a pathological variant in MYH7 might lead to these electrophysiological changes, one potential candidate is an effect of altered calcium cycling in MYH7403/+ cardiomyocytes on signalling pathways and gene expression, called excitation transcription coupling (Dewenter et al., 2017). Previous studies of this variant in mice (Semsarian et al., 2002) have reported altered expression of calcium-handling proteins, which was corrected by pharmacotherapy targeted at normalisation of calcium handling, pointing to a role of calcium homeostasis in regulation of gene expression in this MYH7 variant. In a similar manner, it is possible that abnormal calcium handling also contributes to reduced expression of connexins and the sodium channels observed here.
The degree of reduction in conduction velocity observed (Fig. 4) is consistent with previous studies that have examined a similar loss of connexin expression in other systems. In a mouse model of conditional deletion of connexin-43, van Rijen et al. (2004) showed that a 70-95% conditional deletion of connexin-43 was necessary to reduce conduction velocity or increase the dispersion of conduction, whereas a 50% loss of function had no effect. Similarly, an earlier study by Reaume et al. (1995) showed conduction slowing of between 42% and 56% associated with a 95% reduction in connexin-43 protein expression. As a result of these observations, although reduced expression and/or localisation of connexin-43 has been observed in several pathological states, including post-infarction (Zhang et al., 2009) and in HCM (Kostin et al., 2004), it has been considered unlikely that the degree of reduced expression reported for these diseases was sufficient to directly affect conduction and hence promote arrhythmogenesis. Rather, a second factor such as fibrosis or collagen deposition was likely responsible for the increased propensity for arrhythmias in these patients. The mechanism of conduction slowing reported here for MYH7403/+ cardiomyocytes is therefore unique in that it is the first time that reduced expression of connexin protein has been measured in a model of an HCM gene variant of a sufficient magnitude to directly impact cardiac conduction. Moreover, this effect is likely amplified by a simultaneous reduction in the expression of sodium channels and inward rectifier channel protein in a triumvirate of altered rhythmonome protein expression that coalesce to slow cardiac conduction.
Transcriptomic analysis of MYH7403/+ iPSC-CMs
Transcriptome data showed that genes related to calcium handling, cellular electrical repolarisation, sarcomere structure and cardiac cell fate are differentially expressed between MYH7+/+ and MYH7403/+ cardiomyocytes. First, expression of NKX2-5 was 2.3-fold higher in normal cardiomyocytes compared to that in MYH7403/+ cardiomyocytes. NKX2-5 is a key transcription factor involved in cardiac development and cell fate across species (Elliott et al., 2010). Of relevance to this study, NKX2-5 dysregulation is associated with defects in cardiac conduction and electrophysiology (Ellesøe et al., 2016), including reduced expression of connexins (Nakashima et al., 2009; Briggs et al., 2008) as well as Nav1.5, RyR2 and Kv11.1 (Briggs et al., 2008), consistent with our transcriptomic and protein data (Fig. 6). Similarly, a mouse model employing an inducible system for disruption of notch signalling, which resulted in downregulation of NKX2-5, displayed a strikingly similar electrical phenotype to that seen here, including slow conduction velocity and irregular beating related to reduced expression of gap junctions and sodium channels (Qiao et al., 2017). Another gene that was downregulated, which is a target for NKX2-5, was CORIN. Corin is a serine protease highly expressed in the heart that has conserved binding sequences for NKX2.5 in its 5′ flanking regions (Pan et al., 2002). Corin has previously been identified as a cell surface marker for ventricular cell populations (Zhang et al., 2019) and has been reported as being dysregulated (both downregulated and upregulated) in hypertrophy and heart failure (Khoury et al., 2021; Tran et al., 2004; Ichiki et al., 2013; Langenickel et al., 2004). Furthermore, reactome pathway analysis also associates corin with cardiac conduction (Fabregat et al., 2018). Taken together with the altered expression of NKX2-5 in this model of MYH7403/+ cardiomyocytes, our transcriptome data are therefore consistent with the in vitro electrical phenotypes reported here. Furthermore, if maintained during development and into adulthood in patients, these changes would likely contribute to increased risk of arrhythmias and sudden death.
In relation to calcium handling, we observed downregulation of a cluster of genes including RYR2 (encoding the ryanodine receptor), SLC8A1 (encoding the sodium-calcium exchanger) and PLN (encoding phospholamban). A wealth of previous studies has reported altered calcium homeostasis in models of HCM (Viola and Hool, 2019; Lan et al., 2013; Robinson et al., 2018; Helms et al., 2016; Semsarian et al., 2001) and indeed have identified disrupted calcium handling as central to driving HCM pathology (Lan et al., 2013; Semsarian et al., 2001). In particular, there was altered expression of key calcium-handling proteins, including RyR2 (Semsarian et al., 2001), in agreement with the transcriptome data we report here. The third cluster of genes downregulated in MYH7403/+ cardiomyocytes was related to sarcomere structure – specifically troponin I (both TNNI1 and TNNI3 isoforms), myosin light chain 7 (MYL7) and titin (TNN). Although there were no changes in TNNI1:TNNI3 or MYL2:MYL7 ratios, that might reflect a less mature or less ventricular cell identity of MYH7403/+ cardiomyocytes (Fig. S3), and the broad downregulation of sarcomere genes observed here might reflect overall reduced structure and order of the sarcomere associated with this variant in a thick filament protein (myosin heavy chain).
Finally, related to cardiac repolarisation, expression of the KCNH2 gene, encoding the pore-forming subunit Kv11.1 of the rapid delayed rectifier potassium current, one of the main drivers of action potential repolarisation in human ventricles, was reduced. This is consistent with studies in mice, from our group and others, that have shown downregulation of potassium channel gene expression associated with HCM variants in both MYH7 (Hueneke et al., 2017; Cserne Szappanos et al., 2023) and MYBPC3 (Flenner et al., 2021). However, in these studies, a reduction in protein expression or ion channel current density was also reported in association with the reduced gene expression. Conversely, we saw no significant change in the protein levels of the ion channel Kv11.1 encoded by this gene in MYH7403/+ versus CRISPR-corrected controls (as well as only modest changes in repolarisation time). Similar results were also reported by Flenner et al. (2021), where data from either human iPSCs or from human ventricular samples showed no reduction in K+ currents, in contrast to their mouse data. Such discordance between mRNA and protein levels is common (Liu et al., 2016), with specific examples reported regarding potassium channels and arrhythmias (Brundel et al., 2001). This most likely reflects that protein abundance is in large part controlled at the post-transcriptional and translational levels, and that these processes can adapt to compensate for changes in mRNA transcription to keep cellular protein levels at appropriate levels for normal function.
A similar analysis of transcriptome changes associated with the same R403Q variant was previously undertaken by Cohn et al. (2019) using RNA sequencing. Although the methods used in their study (RNA sequencing) compared to ours (nanoString) are not directly comparable, and a higher threshold for false discovery was used here, there are similarities between the results. For example, some of the genes discussed above associated with calcium handling (PLN) and sarcomere structure (MYL7) were differentially expressed in both studies. Similarly, expression of the SCN5A gene, encoding the cardiac sodium channel Nav1.5 was reduced in Cohn et al. (2019). Although this gene did not reach the threshold for significance for differential gene expression in our study, we do report lowered expression of the Nav1.5 protein here, which we propose, together with altered expression of connexin-43, to be the driver of the reduced conduction velocity reported in this study.
Conclusions
In this study, we have shown that the R403Q pathogenic variant in MYH7, a common and severe cause of HCM, results in changes to cardiomyocyte electrophysiology that may contribute to proarrhythmia and increased risk of sudden death in patients. Specifically, in monolayers of MYH7403/+cardiomyocytes, reduced electrical coupling resulted in significantly slowed conduction velocity and an accompanying increase in spatial dispersion of repolarization that established steep voltage gradients in monolayers of iPSC-derived ‘pseudo-tissue’. Analysis of rhythmonome proteins revealed that reduced electrical coupling resulted from lowered expression of key proteins – specifically connexin-43, Nav1.5 and Kir2.1 – that support electrical conduction between cardiac cells. This is the first report of a cardiomyocyte-intrinsic mechanism of disrupted electrical coupling and conduction in a model of HCM that may represent a new focus for targeted antiarrhythmics in patients with HCM. Furthermore, this biophysical basis for proarrhythmia in HCM may also be accentuated by structural changes in the myocardium later in the progression of disease to contribute to sudden arrhythmic death in these patients.
Limitations
Although we observed proarrhythmic electrical phenotypes (slow conduction, irregular beating and increased dispersion of repolarisation) in our iPSC model of R403Q, we did not observe overt re-entrant activity in our experiments. A likely explanation for this is the small spatial scale of the monolayers studied (microelectrode array geometry is on the order of 2 mm) that, even in the context of slow conduction and discordant repolarisation, was not sufficiently large to support re-entry. It may also be the case that the changes in the expression of rhythmonome proteins observed here and their associated phenotypes may be accentuated by the relatively immature properties of iPSC-CMs and may be less prominent in more mature cells. In our study, we used one iPSC clone to model the effects of the R403Q variant; if performed on multiple clones, there may have been a difference in arrhythmia vulnerability. Finally, data from our study were acquired from one clone from a single patient (and its CRISPR-corrected isogenic control). Further studies including additional clones from cell lines from multiple patients with the same variant in MYH7 would add further strength to our conclusions and help quantify the role of the genetic background in fine-tuning the emergent electrical phenotype of the R403Q variant.
MATERIALS AND METHODS
Generation of patient-specific human iPSCs and general cell culture maintenance
Human iPSCs were derived from a patient carrying the HCM-causing variant p.Arg403Gln in MYH7. Peripheral blood mononuclear cells were isolated and reprogrammed iPSCs were generated as previously described (Holliday et al., 2018). iPSC lines were routinely screened for mycoplasma infection every 12 months. Patient informed consent, use and generation of patient-derived iPSCs complied with national guidelines with oversight by the Sydney Local Health District Committee (protocol X19-0108 and ETH00461). iPSC colonies were maintained in a defined, feeder cell-free medium, mTeSR1 PLUS (StemCell Technologies, BC, Canada) and the extracellular matrix, Matrigel hESC-qualified matrix (Corning, NY, USA), and passaged as aggregates using ReLeSR passaging reagent (StemCell Technologies). Brightfield images were captured on the Zeiss Primo Vert inverted microscope and processed on the Zeiss ZEN Lite 3.4 software (Zeiss, Baden-Württemberg, Germany).
CRISPR-Cas9 gene editing of iPSCs
The guide RNAs (5ʹ-CATTGCCCACTTTCACCTGA-3ʹ) for generation of isogenic control lines were designed using a gRNA design tool (Massachusetts Institute of Technology, MA, USA) and ordered as oligonucleotides to be cloned into pSpCas9(BB)-2A-Puro (PX459) V2.0 (Addgene, MA, USA; #62988). Homology-directed repair donor oligonucleotides (5ʹ-C*C*TCATGGGGCTGAACTCAGCCGACCTGCTCAAGGGGCTGTGCCATCCTCGGGTGAAAGTGGGCAATGAGTACGTCACCAAGGGGCAGAATGTCCAGC*A*G-3ʹ) were designed to flank the Cas9 cut site and ordered as ultramers that contained two phosphorothioate bonds on each end of the oligonucleotide (asterisks) (Integrated DNA Technologies, IA, USA).
iPSCs were plated as single cells and transfected 24 h later with 1 µg plasmid and 5 µl of 100 µM donor oligonucleotide using Lipofectamine Stem (Thermo Fisher Scientific, MA, USA) according to the manufacturer's protocol. After transfection for 24 h, cells were selected with 0.5 µg/ml puromycin for a further 24 h. Single cells were selected into a 96-well plate for Sanger sequencing to determine successfully edited clones, which were further expanded into established cell lines. Off-target analysis involved Sanger sequencing of ten potential guide RNA off-target sites (http://www.rgenome.net/cas-offinder/), Sanger sequencing of all TP53 exons and molecular karyotyping (Victorian Clinical Genetics Services, VIC, Australia) to ensure genomic integrity.
Cardiomyocyte differentiation
Cultures of iPSCs at 70-80% confluency were dissociated into single-cell suspensions with TryPLE (Thermo Fisher Scientific) for 7 min at 37°C and 5% CO2. Cells were then plated between 450,000 and 750,000 cells per well of a Matrigel-coated 12-well tissue culture plate in mTeSR1 PLUS supplemented with 10 µM ROCK inhibitor (Y-27632) (Reprocell, MD, USA) and differentiated into cardiomyocytes using the StemDiff Ventricular Cardiomyocyte Differentiation Kit (StemCell Technologies) as per the manufacturer's protocol. For downstream experiments, iPSC-CMs that showed consistent beating across the entire culture area were used.
Multi-electrode array recordings
Western blotting
At day 45 (±2 days) post differentiation, protein was extracted from beating iPSC-CMs in protein lysis buffer [1× Pierce RIPA buffer (Thermo Fisher Scientific), 1× PhosSTOP cocktail (Sigma-Aldrich), 1× cOmplete protease inhibitor (Sigma-Aldrich)]. Protein concentration was quantified using the Bradford assay using bovine serum albumin (BSA) as a standard. 25 µg of total protein was loaded into precast 10% Mini-PROTEAN TGX Stain-Free SDS-polyacrylamide gels (Bio-Rad), then electrophoretically transferred to 0.2 µm nitrocellulose membrane (Trans-Blot Turbo Transfer Pack, Bio-Rad) using the Bio-Rad Trans-Blot Turbo Transfer System. Blots were probed with the following primary antibodies: rabbit polyclonal anti-Kv1.4 (Alomone, APC-167, 1:500), rabbit polyclonal anti-Kv1.5 (Alomone, APC-004, 1:500), guinea pig polyclonal anti-Kv2.1 (Alomone, AGP-109, 1:1000), rabbit polyclonal anti-Kv4.2 (Alomone, APC-023, 1:500), guinea pig polyclonal anti-Kir2.1 (Alomone, AGP-044, 1:500), rabbit polyclonal anti-K2P3.1 (TASK1) (Alomone, APC-024, 1:500), rabbit polyclonal anti-Kir6.2 (Alomone, APC-020, 1:500), rabbit polyclonal anti-Kv11.1 (HERG) (Alomone, APC-109-F, 1:500), rabbit polyclonal anti-KCNE1 (IsK, MinK) (Alomone, APC-163, 1:500), rabbit polyclonal anti-KCNQ1 (Kv7.1) (Alomone, APC-168, 1:500), rabbit polyclonal anti-SAP97 (Thermo Fisher Scientific, PA1-741, 1:1000), rabbit polyclonal anti-Nav1.5 (Alomone, ASC-005, 1:500), rabbit polyclonal anti-Cav1.2 (Alomone, ACC-003, 1:500), rabbit polyclonal anti-connexin 43 (Cell Signalling Technology, 3512, 1:1000) and rabbit monoclonal anti-GAPDH (Cell Signalling Technology, #2118, 1:2000). Blots were subsequently with the pre-absorbed, polyclonal goat anti-rabbit IgG (H&L) HRP (Abcam, ab97040, 1:10,000) or polyclonal goat anti-guinea pig IgG (H&L) HRP (Abcam, ab97155, 1:10,000) secondary antibody. Whenever it was possible, we probed the membranes in a sequential manner with antibodies from different origins. When blots were used for re-probing with multiple antibodies from the same origin, blots were stripped between probing using stripping buffer (consisting of 62.5 mM Tris HCl, pH 6.8, 100 mM 2-mercaptoethanol and 2% SDS) for 30 min at 50°C. Chemiluminescence signal was detected after incubation of the blots with Immobilon Crescendo Western HRP substrate detection reagent (Millipore, WBLUR0500) following the manufacturer's instructions, by using the Bio-Rad ChemiDoc MP Imaging System. Densitometry was performed for each antibody using ImageJ software (Schneider et al., 2012). Background-subtracted intensity values were normalized to loading control GAPDH signal intensity detected on the same blot. All immunoblot experiments were run as triplicate. Representative images are shown in Fig. 6C and uncropped blots are provided in Fig. S4.
nanoString analysis
RNA was harvested with 100 µl of QIAzol lysis reagent (QIAGEN) per well of an MEA plate at the completion of functional assays at ∼day 45. RNA was pooled from three to four wells of an MEA plate and then extracted using the miRNeasy kit (QIAGEN). RNA levels were measured using the nanoString nCounter PlexSet (nanoString, WA, USA) according to the manufacturer's instructions and gene expression analysed using nSolver software (nanoString). A two-stage Benjamini, Krieger and Yekutieli procedure was implemented for controlling the false discovery rate (Benjamini et al., 2006). A false discovery rate of 5% and a fold change of 2 in gene expression were considered significant.
Immunohistochemistry
A monolayer of iPSC-CMs was fixed with 4% paraformaldehyde for 15 min, incubated in permeabilization solution (1% saponin and 0.05% sodium azide in PBS) for 15 min, and blocked with blocking solution (3% BSA and 0.05% sodium azide in PBS) for 30 min at room temperature. Cells were incubated in primary antibodies against connexin-43 (Abcam, ab11370, 1:1000) and α-actinin (Sigma-Aldrich, A7811, 1:800) for 1 h at room temperature, then washed three times for 2 min with PBS. Secondary antibodies [goat anti-mouse IgG (H+L) cross-adsorbed Alexa Fluor 488 (Thermo Fisher Scientific, A11001) and goat anti-rabbit IgG (H+L) cross-adsorbed Alexa Fluor 594 (Thermo Fisher Scientific, A11012)] were diluted 1:250 in blocking solution and added to the cells to incubate for 1 h at room temperature. Cells were washed with 1× PBS twice for 2 min and incubated with NucBlue Fixed Cell ReadyProbes Reagent (DAPI) (Thermo Fisher Scientific) in PBS for 5 min at room temperature. Coverslips were mounted with ProLong Gold Antifade mountant (Thermo Fisher Scientific) and imaged with a Leica SP8 confocal microscope (Leica, Wetzlar, Germany). Fluorescence intensity was compared using Fiji (ImageJ) software (National Institutes of Health, Bethesda, MD, USA). Cell areas were isolated from the background and mean corrected fluorescence was calculated per cell in each image.
Flow cytometry
The purity of 45-day-old iPSC-CMs was determined by analysis of cardiomyocyte and fibroblast marker expression by flow cytometry. Single cells were fixed with 2% paraformaldehyde in PBS for 10 min at room temperature; then, 250,000 cells (per parameter tested) were resuspended in 1% BSA in PBS. After fixation, cells were centrifuged at 500 g for 5 min and resuspended in 100 µl of FACS buffer (0.1% Triton X-100, 1% BSA and 4% goat serum in PBS) supplemented with antibodies and incubated for 45 min at room temperature. The antibodies used are as follows: FITC anti-human CD90 (BioLegend, 328107, 1:20), anti-human α-actinin conjugated with Vio R667 (Miltenyi Biotec, 130-128-591, 1:100) and FITC mouse IgG1 κ Isotype Ctrl (FC) (BioLegend, 400109, 1:20). 900 µl of FACS buffer was added and the cells were centrifuged at 750 g for 5 min. The pellet was resuspended in 300 µl of 1% BSA in PBS. 1 µl DAPI (Sigma-Aldrich, D9542) was added to cell suspensions. Cells were analysed using a BD LSR II SORP flow cytometer (Becton Dickinson, NJ, USA). A minimum of 10,000 events were recorded and analysed on FlowJo (Becton Dickinson).
Data analysis and statistical tests
Data analysis, plotting and statistical tests were performed using Microsoft Excel (Microsoft Office 2016, Microsoft, WA, USA), and GraphPad Prism v9 (GraphPad Software, CA, USA). Data were analysed using a mixed effect model using the ‘glmfit’ function in MATLAB, with cell line identity as a fixed categorical variable, and individual differentiations (N) and technical replicates (n) as random variables. Estimated marginal means were derived from the generalised linear mixed-effect models using the ‘emmeans’ package (https://github.com/jackatta/estimated-marginal-means), with comparison between differentiations undertaken using a Wald test on input contrasts. In all cases, significance was determined with P<0.05. Violin SuperPlots of data were created using Violin SuperPlots in MATLAB (Kenny and Schoen, 2021); error bars on violin plots represent estimated marginal means of N from the generalised linear model and their standard errors.
Acknowledgements
We thank the Victor Chang Innovation Centre, funded by the New South Wales Government.
Footnotes
Author contributions
Conceptualization: A.P.H.; Methodology: S. Lim, S.B.-R., S. Li, J.T., A.P.H.; Software: A.P.H.; Formal analysis: S. Lim, A.P.H.; Investigation: S. Lim, M.M.M., M.H., H.C.S., S. Li, J.T., W.L., G.N.R.; Resources: A.P.H.; Data curation: S. Lim, A.P.H.; Writing - original draft: S. Lim, H.C.S., C.S., A.P.H.; Writing - review & editing: S. Lim, M.M.M., M.H., H.C.S., S.B.-R., S. Li, J.T., W.L., G.N.R., J.I.V., C.S., A.P.H., L.C.H.; Visualization: S. Lim; Supervision: J.I.V., C.S., A.P.H., L.C.H.; Project administration: C.S., L.C.H.; Funding acquisition: C.S., A.P.H., L.C.H.
Funding
This project was funded by the National Health and Medical Research Council (APP1143321). L.C.H. is supported by a National Health and Medical Research Council Senior Research Fellowship (APP1117366) and a grant from Woodside Energy. L.C.H. is the Wesfarmers, University of Western Australia (UWA)-Victor Chang Cardiac Research Institute (VCCRI) Chair in Cardiovascular Research. C.S. is supported by a National Health and Medical Research Council Investigator Grant (APP2016822) and an NSW Health Cardiovascular Disease Clinician Scientist Grant. Open Access funding provided by UNSW Sydney. Deposited in PMC for immediate release.
Data availability
Raw data and code for statistical analysis and data visualisation are available for download from Zenodo (https://doi.org/10.5281/zenodo.12747473).
References
Competing interests
The authors declare no competing or financial interests.