A transcriptional enhancer regulates cardiac maturation

Abstract

Cardiomyocyte maturation is crucial for generating adult cardiomyocytes and the application of human pluripotent stem cell derived-cardiomyocytes (hPSC-CMs). However, regulation at the cis-regulatory element level, and its role in heart disease remain unclear. Alpha-actinin 2 (ACTN2) levels increase during CM maturation. Here, we investigate a clinically relevant, conserved ACTN2 enhancer’s effects on CM maturation using hPSC and mouse models. Heterozygous ACTN2 enhancer deletion led to abnormal CM morphology, reduced function, and mitochondrial respiration. Transcriptomic analyses in vitro and in vivo showed disrupted CM maturation and upregulated anabolic mammalian target for rapamycin (mTOR) signaling promoting senescence and hindering maturation. As confirmation, ACTN2 enhancer deletion induced heat shock protein 90A expression, a chaperone mediating mTOR activation. Conversely, targeting the ACTN2 enhancer via enhancer CRISPR activation (enCRISPRa) promoted hPSC-CM maturation. Our studies reveal the transcriptional enhancer’s role in cardiac maturation and disease, offering insights into potentially fine-tuning gene expression to modulate cardiomyocyte physiology.

Introduction

Cardiomyocytes (CMs) undergo various morphological, functional, and metabolic changes as they continue to grow during embryonic and postnatal cardiac development^1-3. These adjustments are essential for cardiac maturation and heart’s ability to pump efficiently to accommodate the metabolic needs of the growing body. The process of CM maturation is regulated by several unspecified local and systemic factors that function in a coordinated and complex manner to promote the development of adult CMs⁴. Consequently, in vitro differentiated human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs), are lacking essential elements that promote maturation and thus they demonstrate persistent embryonic properties^5,6. This has become a major hurdle for the application of hPSC-CMs for disease modeling, screening assays and heart regeneration strategies, and major efforts are currently underway to generate myocytes with adult-like features. Furthermore, whether and how disruption of CM maturation contributes to adult cardiac disease remains largely understudied.

Alpha-actinin 2 (ACTN2) is a major sarcomeric protein that is critical for the structural and functional integrity of the Z-disc where it anchors and crosslinks actin thin filaments to sarcomeres^7-9. Importantly, ACTN2 is also a mediator of mechanical stress, it directly interacts with several cardiac ion channels^10-14, and together with other sarcomeric proteins such as Titin and Myosin Heavy Chain, it plays a critical role in early sarcomerogenesis¹⁵. We recently performed a large scale GWAS metanalysis in patients with heart failure and healthy controls and identified 2 genetic variants in proximity with the ACTN2 promoter that were strongly associated with heart failure independent of established cardiovascular risk factors¹⁶. We then utilized published ATAC-seq, Histone 3 Lysine 4 trimethylation (H3K4me3), and Histone 3 Lysine 27 acetylation (H3K27ac) peaks in a hPSC-CM differentiation model, to demonstrate that one of the variants is within a transcriptionally active region, and we performed HiC analysis to determine possible contacts between the identified peaks and nearby genes. Our HiC analysis suggested that this region is likely a transcriptional enhancer that is predicted to interact with the ACTN2 promoter¹⁶. Moreover, the same variant is within an evolutionary conserved region (Extended Data Fig. 1). Since ACTN2 levels increase in hPSC-CMs maintained in culture for prolonged periods¹⁷, and ACTN2 truncations can impair sarcomerogenesis and metabolic maturation of CMs in vivo¹⁸, we hypothesized that disruptions within ACTN2 enhancer could result in cardiomyopathy by altering CM maturation. Conversely, we reasoned that leveraging this enhancer to fine tune ACTN2 levels could promote myocyte maturation.

In the present work, we generated hPSC-CMs with ACTN2 enhancer deletion, which suppressed myocyte maturation. We also found that transcription factors bind a conserved enhancer region to regulate ACTN2 in a consecutive manner. Furthermore, mice with orthologous Actn2 enhancer deletion recapitulated the in vitro phenotype and developed reduced cardiac function, linking cardiac maturation and disease. We further found that in myocytes with ACTN2 enhancer deletion, the Heat Shock Protein 90A (HSP90A) chaperone up-regulates the mammalian target for rapamycin (mTOR) pathway to disrupt maturation. Finally, utilizing an enhancer-based CRISPR-activation system in hPSCs, we leveraged the ACTN2 enhancer to upregulate ACTN2 and promote hPSC-CM maturation. Collectively, our results support the role of an evolutionary conserved cardiac enhancer as a major regulator of CM maturation.

Results

ACTN2 enhancer del suppresses hPSC-CMs functional maturation

To study the role of ACTN2 enhancer in human CMs, we generated human pluripotent stem cells carrying a 1.5Kb heterozygous enhancer deletion using CRISPR-Cas9 (Fig. 1a). First, we examined whether deletion of ACTN2 enhancer altered the differentiation efficiency of hPSC-CMs (ACTN2 enh del hPSC-CMs) compared to isogenic controls, and we found no difference in the percentage of differentiated hPSC-CMs (Extended Data Fig. 2a-b). We then confirmed that our enhancer deletion resulted in a moderate reduction of ACTN2 without affecting any of the alternatively spliced gene variants (Fig. 1b, Extended Data Fig. 2c-d). ACTN2 protein expression was similarly reduced in differentiated hPSC-CMs (Day 25) (Fig. 1c). To analyze the effect of reduced ACTN2 on hPSC-CM morphology we performed immunostaining for ACTN2 and cardiac Troponin T (cTnT) and found that hPSC-CMs were hypertrophic (Fig. 1d), and a higher percentage developed sarcomere thinning and disarray (Fig 1e). Thus, moderate reduction of ACTN2 levels due to a heterozygous ACTN2 enh del results in hPSC-CMs with abnormal morphology.

Figure 1. ACTN2 enhancer deletion impairs the morphological and functional maturation of hPSC-CMs.

a. Generation of hPSCs with ACTN2 enhancer deletion using CRISPR-Cas9. b-c. ACTN2 gene (n=4) and protein expression (n=5) compared to control are reduced in hPSC-derived cardiomyocytes (ACTN2 enh del hPSC-CMs). d. ACTN2 enh del hPSC-CMs develop hypertrophy (>60 hPSC-CMs per group from 4 independent batches were utilized). e. A higher percentage of ACTN2 enh del hPSC-CMs shows myofibrillar thinning and disarray. (at least 80 cells from 4 independent batches were analyzed) (bar graph: 25μm) f. ACTN2 enh del hPSC-CMs demonstrate slower beating rates and suppressed single cell calcium transients measured by IonOptix (4 independent samples/group, 5 different batches). Time to peak and time to 50% to baseline are both prolonged in ACTN2 enh del hPSC-CMs. Peak height is also reduced. g. Force generation of engineered heart tissues (EHTs) derived from ACTN2 enh del hPSC-CMs is suppressed. Baseline force, force time interval (FTI) and relaxation time 50% are all reduced. Additionally, peak systolic force is profoundly suppressed in ACTN2 enh del hPSC-CMs. (5 independent batches were utilized). Data are presented as mean values +/− SEM. All replicates are biological as specified. The Shapiro-Wilk test was performed to assess normal distribution, and student parametric t-test (two-tailed) or Mann-Whitney (two-tailed) non-parametric tests were used as appropriate for all comparisons. For peak systolic force a Bonferroni’s and two-way ANOVA for multiple comparisons tests were used. Only P values <0.1 are reported.

To examine the effect of ACTN2 enh on hPSC-CMs function, we recorded single cell calcium transients, and in accordance with prior reports regarding ACTN2 co-localization and interaction with L-type calcium channels^14,19,20, we found that ACTN2 enh del hPSC-CMs have reduced calcium transients (Fig. 1f). Consistently, average CM beating rates were also decreased (Fig. 1f). Similarly, to test whether the disruption of sarcomeres and calcium handling properties affects force generation in ACTN2 enh del hPSC-CMs, we generated engineered heart tissues and measured force generation using the Myopod system^21,22. Not surprisingly heart tissues from ACTN2 enh del hPSC-CMs demonstrated suppressed force and kinetics (Fig. 1g).

Next, we sought to examine the transcriptomic changes in ACTN2 enh del hPSC-CMs. To focus exclusively on differentiated CMs and avoid potential batch effects, we performed single cell RNA-sequencing. Consistent with our previous analysis, ACTN2 enh del hPSC-CMs showed reduced expression of sarcomeric and calcium handling genes regulating myocyte contraction (Fig 2a-b). Moreover, several metabolic pathways involving mitochondria electron transport, aerobic metabolism and ATP synthesis were downregulated (Fig 2a-b) suggestive of impaired metabolism and energy production. Contrarily, p53-mediated apoptosis, and protein synthesis pathways were upregulated in ACTN2 enh del hPSC-CMs (Fig 2a), likely reflecting pathogenic responses to disrupted sarcomerogenesis, persistent senescence and anabolic pathway activation. Finally, we utilized a previously developed, single cell RNA-Seq-based algorithm to calculate an entropy score²³. Our analysis showed that ACTN2 enh del hPSC-CMs had a higher entropy score consistent with impaired CM maturation (Extended Data Fig. 3).

Figure 2. Single cell transcriptomics of ACTN2 enh del hPSC-CMs reveals disrupted sarcomeric and metabolic pathways.

a. Single cell transcriptomic analysis of ACTN2 enh del vs isogenic control hPSC-CMs. Gene ontology showed reduction of cardiac muscle and muscle contraction genes (black boxes) as well as calcium handling genes (blue boxes). Additionally, mitochondria aerobic electron transport chain and ATP production were also reduced (red boxes), while intrinsic p53-mediated apoptosis and protein synthesis pathways were upregulated in ACTN2 enh del hPSC-CMs. b. Violin plots of representative sarcomeric genes (black), mitochondria DNA (red), and calcium handling genes (blue) that were significantly down-regulated in ACTN2 enh del hPSC-CMs. Only statistically significant pathways and genes are reported.

ACTN2 enhancer del suppresses hPSC-CMs metabolic maturation

During CM maturation, mitochondria increase in size, and they are distributed more uniformly in the cytoplasm. In addition, metabolism switches from glycolysis to oxidative phosphorylation and CMs change their metabolic substrate from glucose to free fatty acids^24,25. To determine the metabolic effects of ACTN2 enh del, we performed Seahorse analysis in hPSC-CMs. Interestingly, ACTN2 enh del hPSC-CMs demonstrated reduced maximum oxygen consumption rates (max OCR) and ATP production, however extracellular acidification rate (ECAR) and ECAR to OCR ratio (marker of glycolysis) were increased in ACTN2 enh del hPSC-CMs (Fig. 3a-e). These data suggests that disruption of ACTN2 enh results in reduced oxidative phosphorylation and increased glycolysis consistent with impaired CM metabolic maturation. Consistently, genes of mitochondria biogenesis, mitochondria DNA synthesis and oxidative phosphorylation, as tested by qPCR, were all suppressed consistent with dysregulated mitochondria (Fig 3f). To analyze the specific effect on mitochondria morphology, we performed mitotracker staining. In agreement with our data, ACTN2 enh del hPSC-CMs formed smaller mitochondria with more irregular networks (Fig 3g). Finally, to examine the ability of hPSC-CMs to metabolize free fatty acids, we cultured hPSC-CMs in media with palmitic acid. ACTN2 enh del hPSC-CMs demonstrated decreased survival, likely due to increased lipotoxicity from impaired oxidative phosphorylation (Fig 3h). Our data thus far supports a striking effect of a cardiac enhancer on several aspects of hPSC-CM maturation.

Figure 3. ACTN2 enh del disrupts mitochondria morphology and function.

a. Seahorse analysis of cultured hPSC-CMs show reduced maximum oxygen consumption rate (OCR) for ACTN2 enh del hPSC-CMs compared to isogenic controls. (5 independent batches per group) b. Consistent with the transcriptomic analysis, ATP production is reduced. c-d. Maximum extracellular acidification rate (ECAR) and ECAR to OCR ratio are consistent with increased glycolysis in ACTN2 enh del hPSC-CMs. e. Recording of OCR over time comparing isogenic control with ACTN2 enh del hPSC-CMs. f. Heatmap of representative mitochondria genes that are down-regulated in ACTN2 enh del hPSC-CMs. Mitochondria biogenesis, mitochondria DNA, fatty acid oxidation and oxidative metabolism genes were characteristically reduced. g. Live hPSC-CMs mitochondria size and morphology analysis using MitoTracker staining. ACTN2 enh del hPSC-CMs have smaller mitochondria with reduced signal intensity consistent with lower mitochondria content and reduced respiration. All comparisons are with control cells. (36 cells per group, 4 independent batches) (bar graph: 50μm) h. Treatment of ACTN2 enh del hPSC-CMs with free fatty acid (FFA) resulted in increased cellular toxicity as quantified by MTT cell viability assay. (6 independent samples per group were analyzed) (bar graph: 200μm). Data are presented as mean values +/− SEM. All replicates are biological as specified. The Shapiro-Wilk test was performed to assess normal distribution, and student parametric t-test (two-tailed) or Mann-Whitney (two-tailed) non-parametric tests were used as appropriate for all comparisons. Given the high sensitivity of Seahorse experiments to cardiomyocyte cell number and batch to batch effects between differentiations, paired student parametric t-test was used for all Seahorse comparisons. Only P values <0.1 are reported.

Temporal regulation of the conserved ACTN2 enhancer

To investigate the mechanism of ACTN2 regulation through this enhancer locus, we focused on a highly evolutionary conserved 258bp region, which also contained an ACTN2 variant associated with heart failure (Fig. 4a, Extended Data Fig. 1)¹⁶. Next, we used CRISPR interference (CRISPRi) in hPSC-CMs and designed several gRNAs that target an intronic region (negative control), proximal and distal regions of ACTN2 promoter (positive controls) and gRNAs against the rs535411 variant in the conserved enhancer region and a variant linked with heart failure (rs663798), located proximal to the conserved region (Fig 4a). gRNAs targeting the ACTN2 promoter, and the evolutionary conserved variant reduced the expression of ACTN2, while gRNAs against the rs663798 variant and intronic region had no effect on ACTN2 (Fig 4a). Similarly, we generated a luciferase reporter system and found that the presence of the evolutionary conserved ACTN2 enhancer region increases luciferase activity in hPSC-CMs, while the same region containing the rs535411 variant did not increase luciferase activity (Extended Data Fig 4a-b). Therefore, the conserved ACTN2 enhancer region that contains a genetic variant associated with heart failure, regulates ACTN2 expression.

Figure 4. Mechanistic analysis of ACTN2 enhancer transcription regulation.

a. CRISPR-interference (CRISPRi) in hPSC-CMs using gRNAs targeting the ACTN2 enhancer, ACTN2 promoter (positive control) and an intergenic region (negative control). CRISPRi targeting ACTN2 promoter strongly suppressed ACTN2 expression, while gRNAs for the intergenic region had no effect on ACTN2. Interestingly, gRNAs complementary to the evolutionary conserved ACTN2 enhancer region containing one of the heart failure-associated variants (rs535411), resulted in reduced ACTN2 expression confirming the regulatory role of this genomic region. gRNAs targeting the other heart failure-associated variant (rs663798) ~500bp upstream of the rs535411, had no significant effect on ACTN2. (5-7 independent experiments per group) b. Chromatin immunoprecipitation (ChIP) qPCR in hPSC-CMs (depicted) showed significant enrichment for the transcription factor MEF2C (M2C), which binds within the evolutionary conserved ACTN2 enhancer region. Similarly, GATA4 (G4) was also enriched. MEF2A (M2A) and TEAD1 (T1) transcription factors, do not appear to bind to this enhancer. The various primer sets utilized for ChIP-qPCR and their respective hybridization sites within the ACTN2 enhancer locus are also shown. (3-4 independent experiments/group) c-e. Like hPSC-CMs, Mef2c and GATA4 transcription factors bind the evolutionary conserved ACTN2 enhancer region in embryonic mouse hearts (embryonic day 15.5). However, Mef2a and GATA4, and not Mef2c, are enriched in adult (P40) mouse hearts supporting temporal regulation of ACTN2 expression through this enhancer region. For all ChIP-qPCR experiments, primers for intergenic region amplification were used as negative control to correct for non-specific antibody binding. Data are presented as mean values +/− SEM. All replicates are biological. The Shapiro-Wilk test was performed to assess normal distribution, and one-way ANOVA with Bonferroni’s multiple comparisons test was used for the CRISPRi analyses, while student parametric t-test (two-tailed) or Mann-Whitney (two-tailed) non-parametric tests were used as appropriate for all ChIP-qPCR comparisons. Only P values <0.1 are reported.

To determine the transcription factors that bind the conserved ACTN2 enhancer region, we utilized JASPAR²⁶ and ENCODE²⁷ databases to identify transcription factor binding motifs within this region. We subsequently relied on published literature^28-31, and focused on four (Myocyte enhancer factor 2A (MEF2A), Myocyte enhancer factor 2C (MEF2C), GATA binding protein 4 (GATA4) and TEA domain transcription factor 1 (TEAD1)) transcription factors that have been reported to regulate gene expression during cardiac development (Fig 4b)²⁷. We subsequently, performed chromatin immunoprecipitation (ChIP) qPCR in hPSC-CMs and found that MEF2C and GATA4 bind the evolutionary conserved enhancer region (Fig 4b). Interestingly, the MEF2C binding site is 30bp proximal to the rs535411 variant. Furthermore, after overexpressing MEF2A and MEF2C modified mRNA in hPSC-CMs transfected with the conserved ACTN2 enhancer-luciferase reporter, we witnessed markedly increased luciferase activity (Extended Data Fig. 4c-d). However, while the same transcription factors increased luciferase activity in hPSC-CMs transfected with the ACTN2 enhancer variant, this effect was significantly lower compared to the wild-type enhancer (Extended Data Fig. 4c-d).

Transcription factors that regulate cardiac development, operate in a dynamic and temporal manner by occupying different chromatin accessible sites from embryonic to adult stages^28,32. Based on ChIP-sequencing data, longer culture times of hPSC-CMs progressively increase the peak height of open chromatin sites within the conserved ACTN2 enhancer locus¹⁶. Furthermore, analysis of available ChIP-seq datasets for histone modifications H3K27ac and Histone 3 Lysine 4 monomethylation (H3K4me1) from mouse hearts³¹, where overlapping peaks mark active enhancers, verified the temporal activation of the Actn2 enhancer locus in vivo (Extended Data Fig. 5a). Also, surveying a recent CM-specific P300 in vivo mouse ChIP-seq dataset³³, supported a time sensitive activation of the Actn2 enhancer region (Extended Data Fig. 5b). We then sought to determine whether there is temporal ACTN2 regulation through different transcription factor occupancy in embryonic compared to adult hearts. First, we analyzed Mef2a ChIP-Seq data²⁸ and found that Mef2a binds the Actn2 enhancer region in adult mouse hearts (Extended Data Fig. 5c). We subsequently validated whether the transcription factors that regulate ACTN2 in hPSC-CMs could also bind the enhancer in vivo. Accordingly, we isolated embryonic mouse hearts (embryonic day 15.5), performed ChIP-qPCR, and found that Mef2c and GATA4 bind this enhancer region (Fig. 4c-d). However, in adult mouse hearts Mef2a instead of Mef2c binds the same region supporting the differential regulation of ACTN2 through this enhancer (Fig. 4e).

We next examined whether an enhancer-mediated temporal regulation of alpha-actinin is preserved across species and specifically whether this phenomenon is also observed in invertebrates. The fruit fly D. melanogaster has one actinin (Actn) gene, which gives rise to three main transcript variants through alternative splicing (flybase.org) (Extended Data Fig. 6)³⁴. To examine whether different regions within the Drosophila Actn locus are transcriptionally active from early to late developmental stages, we analyzed available ChIP-sequencing data³⁵. We found that an extra chromatin accessible region upstream of Actn promoter appears exclusively in Drosophila flies and not in larvae. The transcriptionally active region that appears in adult fly is evolutionary conserved and includes several transcription factors binding sites (Flybase.org)³⁶. We then used qPCR to quantify the expression of the three Actn Drosophila transcript variants and found that their levels vary greatly at different developmental stages, reflecting potential temporal transcription regulation through enhancer regions (Extended Data Fig. 6). Therefore, presumably different combinations and interactions of transcription factors including Mef2, as shown previously³⁷, could temporarily regulate the various Actn transcripts.

Actn2 enhancer del suppresses CM maturation in vivo

To study the role of ACTN2 enhancer during in vivo cardiac development, we searched for the orthologous enhancer in the mouse, and aligned the conserved human enhancer sequence to the mouse genome. Remarkably, the human enhancer is 90% orthologous with a DNA region 17Kb upstream of the mouse Actn2 gene, which based on prior ChIP-sequencing analysis, remains transcriptionally active in postnatal mouse hearts (Extended Data Fig. 5)^27,38. We subsequently, generated a mouse line by deleting this Actn2 enhancer region (Actn2 enh del) (Fig. 5a, Extended Data Fig. 7). We first analyzed the cardiac morphology and observed that the hearts of both heterozygote and homozygote mutant adult mice were enlarged (Fig. 5b). Also, we confirmed that CMs also developed hypertrophy (Fig. 5c). We then, measured the cardiac function by echocardiography and found that Actn2 enh del mice developed reduced left ventricular function and mild left ventricular dilation (Fig. 5d). To describe the transcriptomic changes resulting from the disruption of Actn2 enhancer, we performed RNA-Seq analysis (Fig. 5e). Interestingly, and similar to hPSC-CMs, deletion of the Actn2 enhancer in vivo resulted in mild reduction of Actn2 and down-regulation of sarcomeric and cytoskeletal genes, as well as mitochondria and oxidative phosphorylation genes (Fig. 5f). However, genes of glucose/insulin and the mTOR pathway were upregulated in mutant hearts (Fig. 5f-g). Therefore, deletion of a single ACTN2 enhancer copy in vivo disrupts cardiac function and alters the CM transcriptome, recapitulating the phenotype observed in hPSC-CMs.

Figure 5. Actn2 enhancer deletion in vivo impairs cardiac maturation and function in adult mice.

a. Generation of Actn2 enh del mouse using CRISPR/Cas9. b. WT, heterozygote (Het) and homozygote (Homo) Actn2 enh del mouse hearts presented on the left and heart weight to body weight ratio (HW/BW) quantification on the right. Mutant mice develop cardiac hypertrophy. (n=9-16/group) c. Mouse heart sections stained for WGA to quantify cell size, showed mutant cardiomyocyte hypertrophy. (n=331-358, 5 mouse hearts/group) (bar graph: 50μm) d. The left ventricular ejection fraction (LVEF) is reduced, and the left ventricular internal diameter in diastole (LVIDd) is mildly dilated in mutant hearts. (n=6-11/ group) e. Principal component analysis of bulk RNA-Sequencing data from adult (P40) hearts. f. Gene ontology analysis revealed that sarcomere/myofiber organization and muscle contraction, as well as mitochondria electron transport, ATP production and oxidative phosphorylation genes are downregulated in mutant strains. Contrarily, genes of glucose and insulin signaling, and mTOR and cell growth are upregulated in the mutant hearts. g. Heatmap of representative downregulated genes in mutant mice related to: (1) oxidative metabolism (blue) and (2) sarcomeric organization and function (red). Down-regulated Actn2 expression is highlighted in blue. h. Sarcomeric length analysis of isolated adult cardiomyocytes. Mutant mice developed reduced sarcomeric length consistent with disrupted cardiomyocyte maturation. (Average measurements from 10 sarcomeres/cell, and n=18 cells from 4 mouse hearts/group) (bar graph: 20μm). i. Representative adult cardiomyocytes stained with TOM20. Mutant cardiomyocytes have less organized mitochondria structure, associated with reduced TOM20 signal intensity. Green boxes depicting the zoom in area of CMs (n=15, 3 mouse hearts/group) j. Seahorse analysis confirmed reduced baseline and max oxygen consumption rate (OCR) and increased extracellular acidification rate (ECAR) to OCR ratio suggestive of reduced mitochondria function and increased glycolysis in mutant cardiomyocytes. (n=25-32 samples from 3 animals/group) (bar graph: 50μm). Data are presented as mean values +/− SEM. All replicates are biological as specified. The Shapiro-Wilk test was performed to assess normal distribution, and one-way ANOVA with Bonferroni’s multiple comparisons test or Kruskal-Wallis with Dunn’s multiple comparisons test were used as appropriate. Only P values <0.1 are reported.

To analyze CM sarcomeric structure, we isolated single cells from mouse hearts and performed immunofluorescent staining for ACTN2. Sarcomere length was reduced in mutant mice consistent with impaired functional maturation (Fig. 5h)^39,40. Additionally, immunostaining for mitochondria translocase for outer membrane 20 (TOM20) subunit showed reduced signal intensity in mutant CMs. (Fig. 5i). To measure mitochondria function, we successfully isolated and cultured single adult CMs ex vivo and performed Seahorse analysis. OCR was downregulated, and ECAR/OCR ratio was upregulated in mutant CMs, consistent with reduced oxidative phosphorylation and increased glycolysis respectively (Fig. 5j). Collectively, our analyses support disrupted morphologic, functional, and metabolic maturation of Actn2 enh del CMs in vivo that result in reduced cardiac function.

Next, to avoid potential secondary effects observed in adult Actn2 enh del mice and more accurately capture the effects of impaired CM maturation, we studied early postnatal mouse hearts. During the first postnatal week, CMs cease to proliferate and become more hypertrophic^2,41,42. Therefore, we elected to analyze myocyte proliferation by performing phospho-histone 3 (pH3, mitosis marker) immunostaining using postnatal day 3 (P3) mouse hearts and found no difference in proliferation (Fig. 6a). Interestingly, and in contrast to adult cells, neonatal heterozygous and homozygous Actn2 enh del mouse hearts developed small CM size compared to WT (Fig. 6b), consistent with disrupted maturation^2,42. Then, we analyzed calcium transients, and found that homozygous Actn2 enh del CMs had impaired calcium handling properties with reduced peak height and prolonged time to baseline (Fig. 6c). More importantly, all contractility measurements such as sarcomere fractional shortening, peak height and tau were all disrupted in both heterozygous and homozygous Actn2 enh del CMs (Fig. 6d). In addition, we also examined the metabolic properties of mutant CMs using Seahorse and found that like hPSC-CMs, OCR was suppressed, and ECAR/OCR ratio was increased consistent with reduced oxidative phosphorylation and increased glycolysis (Fig. 6e). Thus, analyses of early postnatal hearts further supported that Actn2 enh del CMs demonstrate impaired morphologic, functional and metabolic maturation.

Figure 6. Actn2 enhancer deletion in vivo disrupts early postnatal cardiomyocyte maturation.

a. Immunofluorescence staining for the proliferation marker phospho-histone 3 (pH3) showed no difference in the percentage of proliferating cardiomyocytes isolated from WT, het and homo Actn2 enh del mouse hearts (postnatal day 3) (n=3 mice/group) b. Mouse heart sections stained for WGA to quantify cell size. The cell size of neonatal CMs (postnatal day 0, P0) is reduced in both heterozygous and homozygous Actn2 enh del mouse hearts (n=~200 cells/group, 3mice/group). (bar graph: 20μm) c. Calcium handling is disrupted in Actn2 enh del mouse CMs. The time to peak was unchanged, however peak height is reduced in homozygous Actn2 enh del CMs and time to 50% baseline is prolonged in homozygous CMs (n=21-22 per group). d. The contractile properties such as fractional shortening, peak height and tau are all disrupted in both heterozygous and homozygous P14 Actn2 enh del CMs (n=13-21 per group). e. Oxygen consumption rates (OCR) are suppressed, and glycolysis (ECAR/OCR) is increased in heterozygous and homozygous Actn2 enh del CMs (n=23-29 per group). All Seahorse measurements were compared to WT control cells. Data are presented as mean values +/− SEM. All replicates are biological as specified. The Shapiro-Wilk test was performed to assess normal distribution, and one-way ANOVA with Bonferroni’s multiple comparisons test or Kruskal-Wallis with Dunn’s multiple comparisons test were used as appropriate. Only P values <0.1 are reported.

ACTN2 enhancer del activates mTOR through HSP90A chaperone

To investigate the mechanism through which ACTN2 enh del disrupts CM maturation, we focused on genes that are consistently differentially expressed in Actn2 enh del CMs (heterozygotes) in vitro and in vivo. Gene Ontology analysis of overlapping genes showed activation of fatty acid import, developmental growth, protein synthesis, autophagy, and response to growth factor pathways (Fig. 7a). Importantly, several genes of the phosphatidylinositol 3-kinase/Akt Serine Threonine kinase/mTOR (PI3K/Akt/mTOR) pathway were upregulated in hearts of adult mice carrying a heterozygous Actn2 enhancer deletion, accounting for persistent growth and protein synthesis activation (Fig. 7b). The mTOR signaling pathway functions as a nutrient sensor to integrate various signals and mediate cell growth^43-46. However, persistent activation of mTOR has been consistently linked with impaired maturation of hPSC-CMs^45,47, and maladaptive stress-related upregulation of mTOR is associated with heart disease⁴³. Thus, we first performed western blotting in adult hearts to quantify the levels of phosphorylated 70-kDa ribosomal protein S6 kinase (p70S6K) at Threonine 389, a downstream effector of mTOR that is known to mediate increase protein synthesis and cell growth. Consistent with our transcriptomic analysis, phospho-p70S6K levels were upregulated in Actn2 enh del adult mouse hearts (Fig. 7c). Likewise, to examine whether the mTOR signaling pathway remains persistently active from early developmental stages and avoid secondary effects, we performed qPCR and found that several genes of the mTOR pathway are up regulated in neonatal mouse hearts (postnatal day 7) (Fig. 7d). Moreover, we similarly quantified the expression of phospho-p70S6K in neonatal hearts and verified mTOR activation in Actn2 enh del mice during early postnatal development (Fig. 7e). However, we were unable to detect p70S6K in hPSC-CMs, therefore we elected to examine the expression of phospho-eukaryotic translation initiation factor 4 binding protein-1 (phospho-4EBP1) a translation repressor whose phosphorylation downstream of mTORC1 at serine 65, prevents its interaction with the translation initiation factor eIF4E and upregulates protein synthesis^43,46. Phoshprylated-4EBP1 levels were increased in ACTN2 enh del hPSC-CMs consistent with mTOR induction (Fig. 7f). Consistently, in ACTN2 enh del hPSC-CMs, the gene expression levels of regulatory associated protein of mTOR complex 1 (RAPTOR) and 3-phosphoinositidine-dependent protein kinase-1 (PDK1) that are part of mTORC1 were upregulated, while RPTOR independent companion of MTOR complex 2 (RICTOR) levels which is part of the mTORC2 complex remained unchanged (Extended Data Fig. 8a). Moreover, to validate whether suppression of mTORC1 signaling could potentially rescue the impaired maturation of ACTN2 enh del CMs, we treated CMs with everolimus an mTORC1 inhibitor. Everolimus treatment improved ACTN2 enh del CMs sarcomeric structure and normalized their size (Extended Data Fig. 8b-c). Also, calcium transients, mitochondria respiration and maturation-related gene expression patters were not different compared to control CMs (Extended Data Fig. 8c-e), thus suggesting that suppressing mTORC1 could reverse the maturation defects of ACTN2 enh del CMs. Collectively, our data supports that ACTN2 enh del leads to persistent induction of mTORC1 and disrupted CM maturation.

Figure 7. mTOR pathway is activated in Actn2 enhancer deletion hearts and hPSC-CMs.

a. Gene ontology analysis of the 472 overlapping differentially expressed genes in hPSC-CMs and adult mice (Het) with ACTN2 enh deletion. b. Heatmap of the mTOR pathway-related genes, quantified by qPCR that are upregulated in adult (P40) Actn2 enh del mouse hearts consistent with early activation of the pathway. c. Protein expression analysis by western blotting using antibodies against the activated phosphorylated P70S6K and unphosphorylated P70S6K in adult mice verified P70S6K activation consistent with induction of mTOR signaling in mutant Actn2 enh del mice. (n=7-8 per group) d-e. Likewise, in neonatal (P3) mutant mouse hearts, mTOR-related genes were upregulated, and protein expression analysis by western blotting verified the induction of mTOR signaling in mutant Actn2 enh del mice. (n=6 per group) f. Protein expression analysis by western blotting using antibodies against the activated phosphorylated 4EBP1 to total 4EBP1 ratio in hPSC-CMs, showed upregulation of mTOR in ACTN2 enh del hPSC-CMs compared to isogenic controls. (n=4 per group) g. HSP90A inducible chaperone protein levels were increased in ACTN2 enh del hPSC-CMs (n=5 per group). h. siRNA mediated suppression of HSP90A expression normalized phosphorylated 4EBP1 to total 4EBP1 ratio in ACTN2 enh del hPSC-CMs. (n=4 per group). All protein measurements were compared to WT hearts. Data are presented as mean values +/− SEM. All replicates are biological. The Shapiro-Wilk test was performed to assess normal distribution, and student parametric t-test (two-tailed) or ANOVA with Bonferroni’s multiple comparisons test were used as appropriate. Only P values <0.1 are reported.

To investigate the upstream mediators of mTOR activation in ACTN2 enh del CMs, we hypothesized that disruption of ACTN2 levels could upregulate an mTOR activator through direct interaction. Therefore, we first analyzed a previously published proteomics dataset of ACTN2 proximity labeling biotinylation (BioID) in hPSC-CMs⁴⁸. We then overlap this dataset with genes that are upregulated in ACTN2 enh del CMs both in vitro and in vivo. Interestingly, of all the overlapping proteins the stress-induced heat shock protein 90-A (HSP90A), is a chaperone that could interact with ACTN2 and mediate mTOR activation^49-52. We subsequently, performed co-immunoprecipitation experiments to verify that ACTN2 can indeed interact with HSP90A in hPSC-CMs, and this effect was more pronounced in ACTN2 enh del hPSC-CMs (Extended Data Fig. 9a). Then we quantified the expression of HSP90A and found it is increased in ACTN2 enh del hPSC-CMs (Fig. 7g). To further investigate whether HSP90A can mediate mTOR activation in ACTN2 enh del hPSC-CMs, we used siRNA to suppress HSP90AA1. siRNA transfection reduced HSP90A protein levels, and suppressed the phoshprylated-4EBP1/4EBP1 ratio, preventing mTORC1 upregulation in ACTN2 enh del hPSC-CMs (Extended Data Fig. 9b-c, Fig. 7h). Thus, the inducible chaperone HSP90A can act as an upstream mTOR signaling activator in ACTN2 enh del CMs.

Leveraging enCRISPRa to promote hPSC-CMs maturation

Considering the effects of ACTN2 on CM maturation, we reasoned that ACTN2 gain of function could promote the generation of more mature hPSC-CMs. To fine tune ACTN2 levels and avoid potential proteotoxicity from markedly disrupted sarcomeric protein stoichiometry^53-55, we decided to leverage our enhancer locus using targeted CRISPR-mediated transcription activation⁵⁶. To specifically target transcriptional enhancers, we utilized a deactivated Cas9 (dCas9) fused with the catalytic core of the human acetyltransferase P300 (dCas9-P300), which catalyzes the acetylation of Histone 3 Lysine 27 (a marker of active transcriptional enhancers)^56-58. We subsequently used CRISPR/Cas9 to knock in dCas9-P300 and engineered a stable hPSC line to develop an enhancer CRISPR activation (enCRISPRa) system. Then, we generated sgRNAs in a scaffold that recruits the transcription activator VP64 co-expressed with mCherry (Fig. 8a). We subsequently, transfected enCRISPRa hPSC-CMs with three different sgRNAs targeting the ACTN2 enhancer conserved region and quantified gene expression by qPCR 48hours later. Transfection with sgRNAs targeting the 5’ ACTN2 conserved enhancer region (gRNA2) and ~350bp upstream of this site (gRNA3), resulted in significant upregulation of ACTN2 (Fig. 8b, Extended Data Fig. 10), and this increased the expression of myosin heavy chain 7 (MYH7) and Troponin I3 cardiac type (TNNI3) consistent with improved functional maturation (Fig. 8c). Additionally, most calcium handling genes (calcium voltage-gated channel subunit 1c (CACNA1C), ryanodine receptor 2 (RYR2), sodium calcium exchanger (NCX), ATPase sarcoplasmic/endoplasmic reticulum Ca²⁺ transporting 2 (SERCA2A)) were also upregulated, suggestive of more mature calcium handling properties (Fig. 8d). Importantly, genes regulating mitochondria biogenesis and oxidative phosphorylation in cardiomyocytes were also upregulated post sgRNA2 and sgRNA3 transfections (Fig. 8e). Together, our gene expression analysis shows that enCRISPRa can be applied to hPSC-CMs and could be leveraged to promote the generation of more mature hPSC-CMs. More importantly, our enCRISPRa system could also be leveraged to interrogate and study enhancers in any human stem cell-derived tissue.

Figure 8. Leveraging the ACTN2 conserved enhancer region to promote the maturation of hPSC-CMs.

a. Illustration of the enhancer CRISPR activation (enCRISPRa) system. A stable hPSC line expressing deactivated Cas9 (dCas9) fused with the catalytic core of the human acetyltransferase P300, which catalyzes the acetylation of Histone 3 Lysine 27 (H3K27ac) under a constitutively active promoter was generated. sgRNAs in a scaffold that recruits the transcription activator VP64 were transfected in hPSC-CMs expressing dCas9-P300. Three different sgRNAs targeting areas within the conserved enhancer region and immediately upstream were utilized. b. Transfection of hPSC-CMs with sgRNAs targeting the 5’ border of the conserved ACTN2 enhancer region (sgRNA2) and ~350bp upstream (sgRNA3) resulted in upregulation of ACTN2. c. Sarcomeric genes, MYH7 (beta-myosin heavy chain) and TNNI3 (cardiac Troponin T) that commonly increase in more mature cardiomyocytes were upregulated after targeting the ACTN2 enhancer with sgRNA2. d. Most calcium handling genes were similarly upregulated after enCRISPRa upregulation of ACTN2. e. Genes regulating mitochondria biogenesis and oxidative phosphorylation in cardiomyocytes were upregulated after using enCRISPRa to target the ACTN2 enhancer. (3 independent experiments were analyzed). Data are presented as mean values +/− SEM. All replicates are biological. The Shapiro-Wilk test was performed to assess normal distribution, and one-way ANOVA with Bonferroni’s multiple comparisons test or Kruskal-Wallis with Dunn’s multiple comparisons test were used as appropriate. Only P values <0.1 are reported.

Discussion

CM maturation occurs from early embryonic to adult stages and is characterized by coordinated changes in cell morphology, function, and metabolism¹. Maturation is commonly disrupted in hPSC-CMs, which remain embryonic-like and fail to develop most properties of adult CMs. This severely impedes the broader application of hPSC-CMs for disease modeling, drug development, and heart regeneration therapies. However, the exact mechanisms of CM maturation remain incompletely understood, and efforts to overcome this critical limitation are needed. Moreover, even though impaired CM maturation is hypothesized to contribute to cardiac disease^2,59, supporting evidence is currently missing.

In our present work we demonstrated that ACTN2 plays a key role on CM maturation, and removal of an evolutionary conserved ACTN2 enhancer, previously linked with heart failure¹⁶, disrupts CM maturation resulting in abnormal cell structure, function and metabolism. We also showed that transcription factors MEF2A, MEF2C and GATA4 bind this enhancer in a temporal fashion, and ACTN2 directly interacts with the inducible chaperone HSP90A, and disruption of ACTN2 levels up-regulate HSP90A, which activates mTOR signaling and can impair the maturation of ACTN2 enh del CMs. Importantly, we demonstrated that an enhancer CRISPRa approach with a gRNA targeting the conserved ACTN2 enhancer region upregulated ACTN2 and promoted hPSC-CM maturation.

ACTN2 is a major sarcomeric protein in the Z-disk and autosomal dominant mutations in ACTN2 are linked to dilated and hypertrophic cardiomyopathy, left ventricular non-compaction and cardiac arrhythmias^24-26. We have previously shown that two variants within a conserved ACTN2 enhancer region are associated with heart failure¹⁶ and different studies have modeled the mechanistic contribution of ACTN2 coding mutations to cardiomyopathy^53,60. ACTN2 mutations can affect sarcomerogenesis by interacting with membrane anchoring proteins⁶¹ and can increase protein toxicity and activate autophagy⁵³. Moreover, ACTN2 is part of a sarcomeric protein complex initiating sarcomerogenesis in CMs¹⁵, and ACTN2 is the main sarcomeric protein which continues to increase in hPSC-CMs²⁷ after prolonged culture. Additionally, a recent in vivo study showed that an ACTN2 truncation can disrupt CM metabolic maturation through myocardin related transcription factor-serum response factor (MRTF-SRF) signaling²⁸. However, we did not detect any change in SRF expression and its downstream pathways in ACTN2 enh del CMs. This is likely due to the fact that an enhancer deletion solely disrupted ACTN2 levels and had no effect on protein structure. In a separate study, Ladha et al. used proximity labeling to show ACTN2 interactions with the insulin like growth factor-2 mRNA binding protein 2 (IGF2BP2) that regulates oxidative phosphorylation in hPSC-CMs⁴⁸. Therefore, ACTN2 appears to be a key regulator of cardiac maturation by coordinating CM sarcomerogenesis, metabolism and function, thus even mild disruptions of ACTN2 levels can have striking effects on CM.

CM sarcomeres are tightly linked to mitochondria function. During CM maturation, mitochondria surround myofibers and develop extensive networks for efficient energy delivery to the sarcomere^62,63. More recently, genes of the peroxisome proliferator activated receptor (PPAR) family, appear to increase mitochondria biogenesis and alter CM metabolism which further improves their structural and functional maturation^25,64. In addition, disruption of mitochondria function, either through mutations in genes regulating fusion/fission or metabolism, results in cardiac dysfunction and myofibrillar disarray^63,65,66. However, several reported mutations that directly affect mitochondria function do not interfere with the sarcomere^67,68, while mutations of myofibrillar genes consistently suppress mitochondria maturation^18,67, supporting a critical role for sarcomeres during CM maturation⁶². Consistent with this paradigm, we showed that disruption of the ACTN2 enhancer has profound effects on CM metabolism by reducing mitochondria size, promoting glycolysis and suppressing oxidative phosphorylation.

Our work also supports the temporal regulation of cardiac gene expression through transcriptional enhancers^28,32,69. Using ChIP analyses, we showed dynamic regulation of ACTN2 expression by an evolutionary conserved enhancer, where GATA4 and MEF2C are the main transcription factors in embryonic hearts, while GATA4 and MEF2A bind with greater affinity in adult stages. Interestingly, ChIP-Seq analysis of the homologous Actn locus in Drosophila showed separate chromatin active regions at different developmental stages, while we found that distinct Actn transcripts are expressed in larvae vs flies, supporting a more universal, cross-species, temporal regulation of actinin genes by cis-regulatory elements.

Interestingly, the rs535411 enhancer variant that we previously linked with heart failure¹⁶ is within an evolutionary conserved region proximal to a MEF2A/C transcription factor binding site and it appears to potentially affect the transcriptional regulation of ACTN2. Given the proximity of the ACTN2 enhancer with RYR2, it is possible that this enhancer could also regulate RYR2 and other regulators of CM maturation. Moreover, though the variant does not directly affect a DNA binding motif, such effects of genetic variants adjacent to transcription factor binding sites have been reported and validated in rheumatologic diseases, cancer etc^70-73. There are several potential mechanisms with which the rs535411 variant can indirectly affect the binding of MEF2A/C, such as altering chromatin accessibility, changing the DNA shape, disrupting the recruitment of MEF2A/C by interfering with its cooperative interaction with other proteins and other transcription factors such as GATA4^70,72,73. Thus, it is not inconceivable that genetic variants within a conserved enhancer could indirectly affect the regulation of critical genes.

mTOR is a nutrient and energy sensor, which acts as a critical regulator of anabolic and catabolic pathways, such as protein synthesis and autophagy respectively, that ultimately regulate cell growth^43,44,46. One major mechanism that specifically activates mTOR is the insulin/PI3K/Akt^74-76 signaling pathway. Interestingly, ACTN2 enh del CMs both in vitro and in vivo, developed upregulation of insulin signaling. Furthermore, mTOR signaling remained increased in ACTN2 enh del CMs and this resulted in upregulation of anabolic protein synthesis, increased hypoxic response and p53 mediated senescence pathways. Also, upregulation of hypoxia pathways by mTOR, commonly shifts metabolism towards persistent glycolysis^77,78 and glucose utilization, which likely disrupted the metabolic maturation of ACTN2 enh del CMs⁷⁹. Moreover, upregulation of mTOR, p53 and cyclin dependent kinase inhibitor 1A (p21) promotes CM senescence that is linked to disrupted hPSC-CM maturation^45,47. Similarly, in ACTN2 enh del hPSC-CMs P21 remained upregulated, promoting senescence, hindering further CM maturation, and suppression of mTORC1 signaling in ACTN2 enh del hPSC-CMs improved their maturation status by normalizing their morphology, calcium transients and metabolism. Finally, persistent mTOR activation could also likely account for the hypertrophic phenotype noted in adult CMs in vivo ⁴³ and hPSC-CMs, while myocyte size in early postnatal heterozygous and homozygous mice was reduced.

The exact mechanism of mTOR upregulation in ACTN2 enh del CMs remains unclear, however based on proximity labeling, ACTN2 interacts with the inducible HSP90A chaperone⁴⁸, which localizes at the Z-disc⁸⁰ and participates in myosin protein folding and sarcomeric assembly^81,82. Importantly, several proteins of the mTOR pathway (RAPTOR, mTOR, S6K etc.) directly interact with HSP90 proteins^49,83,84, HSP90A and the PI3K/Akt remain elevated in hypoxic CMs⁸⁵, and HSP90A induces PI3K/Akt signaling⁸⁶. We found that HSP90AA1 is upregulated in ACTN2 enh del CMs both in vitro and in vivo. Moreover, co-IP in hPSC-CMs showed that ACTN2 directly interacts with HSP90A, and this effect is likely more pronounced in ACTN2 enh del CMs. To test whether HSP90A activates mTOR in ACTN2 enh del CMs, we suppressed HSP90A in hPSC-CMs and found that it prevented mTOR activation. Therefore, it is highly likely that reduction of ACTN2 and subsequent disruption of sarcomeric protein stoichiometry at the Z-disc, upregulates the inducible chaperone HSP90A that directly interacts with ACTN2 and other mTOR pathway proteins, ultimately activating mTOR signaling in ACTN2 enh del CMs. We believe that examining the specific mechanism through which HSP90A regulates mTOR activation in CMs is an interesting topic that needs further investigation. However, this topic is beyond the scope of the present study, which was designed specifically to address the role of a transcriptional enhancer in CM maturation.

To test whether manipulations of the ACTN2 enhancer can be leveraged to fine tune ACTN2 expression and promote hPSC-CM maturation, we utilized an enhancer-CRISPRa system⁵⁶. We first generated a stable hPSC line that constitutively expresses dCas9-P300 and designed three sgRNAs to interrogate different regions of the ACTN2 enhancer in hPSC-CMs and fine tune ACTN2 levels. sgRNAs targeting regions ~200bp and 600bp upstream of the MEF2A/C binding site of the evolutionary conserved enhancer region, resulted in upregulation of ACTN2. A sgRNA targeting a sequence next to the MEF2A/C binding site did not increase ACTN2. This is similar to prior CRISPR activation reports, where sgRNAs targeting regions 200bp upstream or downstream of transcription start sites⁸⁷ or enhancer elements⁵⁶ were successfully utilized. Excitingly, increased ACTN2 levels through enCRISPRa, also upregulated several sarcomeric, calcium handling and mitochondria biogenesis genes consistent with a more mature phenotype. To our knowledge this is the first study where targeted manipulation of a single enhancer element in hPSC-CMs can effectively induce transcriptomic changes consistent with improved functional maturation. Moreover, we engineered an enCRISPRa hPSC line that could be leveraged to interrogate enhancers and fine tune gene expression at physiological relevant levels in any stem cell derived tissue.

In summary, our study provides in vitro and in vivo validation of a conserved ACTN2 transcriptional enhancer on cardiac maturation. It further supports the role of sarcomeric proteins as key regulators of CM maturation and demonstrates a previously unrecognized mechanism by which disruption of ACTN2 can induce a protein chaperone to activate mTOR signaling in CMs. Lastly, we provided further insights on how enCRISPRa strategies can be applied to CMs to manipulate cardiac cis-regulatory elements and study their effects on myocyte physiology and disease.

Methods

Inclusion and ethics

All protocols involving animals were approved by the animal and care use committee of the Johns Hopkins School Medical Institution's Animal Care and Use Committee (protocol ID MO20M343) and are in line with NIH's guidelines.

The H9 cell line (obtained by the WiCell Institute) and the WTC cell line carrying a constitutively expressed deactivated Cas9-KRAB (obtained by the CORIEL institute) were used for all experiments. Use of these lines and these studies were approved by the Johns Hopkins University Institutional Stem Cell Research Oversight Committee (AM00000385).

Human stem cell derived cardiomyocytes

Human embryonic stem cells (hESC line H9, NIHhESC-10-0062, WiCell Institute) were used to generate all lines utilized in the study apart from the CRISPRi line carrying a constitutively expressed deactivated Cas9-KRAB, which was obtained by the CORIEL institute (AICS-0090-391, parent line WTC)⁸⁸. All hPSCs with ACTN2 enhancer deletion were generated using CRISPR/Cas9, using two gRNAs (Supplementary Table 1) targeting a 1.3Kb region. All hPSCs were maintained in essential 8 medium (Gibco) and they were treated with 6 μM of CHIR99021 (Tocris) for 2 days followed by 5 μM of endo-IWR-1 (Tocris) in RPMI media supplemented with B27-minus insulin (Gibco) for 2 days⁸⁹. CMs were then selected using 4mM sodium lactate (Sigma) for three days and then they were replated in 0.1% gelatin (Millipore) coated plates for further experiments. hPSC-CMs were also cultured in RPMI no glucose media (Gibco) supplemented with B27 (Gibco) and 50μM palmitate (Sigma).

Animals

Actn2 enhancer knock out B6SJLF2 mice were generated using CRISPR/Cas9 with two gRNAs (Supplementary Table 1) targeting a 1.3Kb region, which is orthologous to the evolutionary conserved 258bp human ACTN2 enhancer region at the Johns Hopkins transgenic mouse core. For chromatin immunoprecipitation experiments C57BL6 WT mice were used. All animals were housed at the animal facilities in the Johns Hopkins Medical Institutions. The animals were randomly allocated to experimental groups and both male and female mice were equally used in all experimental assays. All mouse hearts and CMs were harvested at random times during the day. yw wild-type Drosophila melanogaster larvae and adult flies were used for our fruit fly experiment.

Mouse cardiomyocyte ex vivo culture

CMs were isolated from P40 mouse hearts using a highly reproducible Langendorff-free method⁹⁰. Briefly, the mice were euthanized by isoflurane inhalation and chest wall was cut open and heart exposed. The right ventricle was injected with 7 ml EDTA buffer (NaCl 130mmol/L, KCl 5mmol/L, NaH2PO4 0.5mmol/L, HEPES 10mmol/L, Glucose 10mmol/L, BDM 10mmol/L, Taurine 10mmol/L, EDTA 5mmol/L) and descending aorta was cut. Following this, the heart was transferred to a Petri dish containing EDTA buffer and the ascending aorta was clamped with Reynolds forceps. Next, digestion was performed by sequential injection of 10 ml EDTA buffer, 3 ml perfusion buffer (NaCl 130mmol/L, KCl 5mmol/L, NaH₂PO₄ 0.5mmol/L, HEPES 10mmol/L, Glucose 10mmol/L, BDM 10mmol/L, Taurine 10mmol/L, MgCl₂ 1mmol/L) and 30 ml collagenase buffer (Collagenase 2 0.5mg/ml, Collagenase 4 0.5mg/ml, Protease XIV 0.05mg/ml) through the left ventricle. Next, the digested heart was separated and pulled into 1 mm pieces using forceps. Single cell CMs were dissociated by gentle trituration and enzyme activity was stopped by adding 5 ml stop buffer (perfusion buffer containing 5% FBS). After the isolation, CMs were first maintained in M199 media (Gibco) for 24 hours prior to immunofluorescent staining, or Seahorse analysis.

P14 CMs were isolated from hearts that were perfused through a Langendorff system and treated with Type II Collagenase (Worthington) and Protease (Sigma) digestion buffer⁸⁹. Then digested hearts were triturated and filtered. After isolation, CMs were cultured in M199 media (Gibco) and used for: (1) calcium transient analyses using the IonOptix system, (2) Seahorse experiments and (3) immunofluorescent staining.

IonOptix Calcium transient measurements

CMs were seeded in 0.1% gelatin coated 5mm cover slips (Warner Instruments). For mTORC1 inhibition, hPSC-CMs were treated for 96 hours with Everolimus (Tocris) (25μM). Two days later CMs were loaded with 1μM Fura2AM Ca²⁺ dye (Invitrogen) in 1x Tyrode’s solution (NaCl 0.14M, KCl 5mM, HEPES 10mM, MgCl₂ 1M, and D-Glucose 5.5mM). Cells were paced at 0.5 Hz with 20 V. Ca²⁺ transient in individual CM was analyzed using the IonOptix imaging system and IonWizard software⁶.

Engineered Heart Tissues (EHT) measurements

EHTs were generated by seeding ~1 million CMs together with ~50,000 human cardiac fibroblasts into scaffolds made from laser cut decellularized porcine myocardium and maintained in PDMS seeding baths²². Two days after the casting, the EHTs were removed from PDMS seeding baths and transferred into cell culture plates and maintained in RPMI media supplemented with B27 plus insulin and the media was changed every other day. Contraction of the EHTs began between day 3 to 5 after the casting. The EHTs were allowed to stabilize for 1 week before force generation was recorded. The EHTs were equilibrated, and baseline contraction force was recorded. After the baseline recording, the EHTs were subject to different percentages (1 to 5%) of stretch for length dependent activation. Data was recorded and analyzed using the MyoLab Software (Propria).

Flow cytometry

Single hPSC-CMs were fixed with 4% PFA for one hour and washed with PBS. Cells were then permeabilized with 0.75% saponin (Sigma), stained with anti-cardiac Troponin T (Abcam, Ab8259) followed by incubation with Alexa fluor secondary antibody (488) (Abcam, Ab150105). Flow cytometry was performed using SH800 Sony. Data was analyzed using FlowJo (version 10.5).

Luciferase assay

24hours after replating hPSC-CMs, cells were transfected with the respective vectors with and without modified mRNAs using the Lipofectamine Stem reagent (ThermoFisher). The dual luciferase reporter assay system (Promega) was used, and CMs were transfected with both the modified luciferase vector pGL4.10 (Promega) and the Renilla luciferase vector pGL4.70 (Promega) (internal control). CMs were lysed two days later and firefly luciferase activity was measured using the Glomax luminescence plate reader (Promega).

Synthesis of modified RNA (modRNA)

DNA was amplified from plasmid templates using custom primers targeted the open reading frames of the genes of interest using Q5 mater mix (NEB M0492S). The custom primers were designed to add a T7 promoter and a Kozak sequence upstream of the start codon and a FLAG protein sequence and a stop codon. The DNA template sizes were confirmed using DNA gel electrophoresis. The templates were then purified using a DNA purification kit (Zymo Research). ModRNAs were synthesized from DNA templates using a custom blend of m7G(5')ppp(5')G RNA cap structure analog (NEB), adenosine triphosphate (Thermo Fisher), guanosine triphosphate(Thermo Fisher), 5-methylcytidine triphosphate (TriLink Biotechnologies), pseudouridine triphosphate (TriLink Biotechnologies), and T7 enzyme mix (Thermo Fisher), by incubating for 6 hours at 37°C. After incubation, 1 μL TURBO DNase (Thermo Fisher) was added to the reaction mixture for 15-minute DNA digestion at 37°C. Subsequently, Poly(A) tailing reagents (Thermo Fisher) were added for a one-hour incubation at 37°C as instructed in the manufacturer’s protocol. The Poly(A) tailing was confirmed by RNA gel electrophoresis (Lonza). The modRNAs were purified using LiCl precipitation (Thermo Fisher), resuspended in nuclease-free water and stored in −80°C.

Immunofluorescence

Hearts were fixed in 4% paraformaldehyde (Thermo Fisher) overnight, then placed in 30% sucrose (Sigma) followed by OCT and sectioning. Isolated mouse CMs and hPSC-CMs were fixed in 4% paraformaldehyde. For immunofluorescent staining all cells/tissues were blocked for 1 hour with 1% BSA and incubated overnight with the following primary antibodies at 1:500 dilutions: α-actinin (Abcam, ab68167), cardiac Troponin-T (Abcam, Ab8259), Tom20 (Proteintech, 1D6F5). Alexa fluor secondary antibodies (488, 594, 647) (Abcam) were used for secondary detection and DAPI was added for nuclei staining. To assess CM size, heart sections were stained with WGA-Alexa fluor 647nm (1:500) (Thermo Fisher, W32466). Cells were imaged using a Leica SP8 confocal microscope. All imaging analysis was performed by blinded investigators using Image J (Version 1.52q)⁸⁹.

RNA-Sequencing

Day 25 hPSC-CMs were dissociated using TrypLE (Gibco), single cells were harvested and sorted using the droplet based 10x Genomics Chromium system, and RNA-seq cDNA libraries were generated using the Chromium Next Gem Single Cell Kit (10x Genomics). cDNA libraries were submitted to the JHU Sequencing Core. Readings were mapped with STAR and featureCounts-based zUMIs ⁹¹. Differential expression analysis on raw counts was performed, and cell type-specific marker genes were used ^25,92. Hearts from the same mouse liter at P40 were harvested, and homogenized in Trizol (Thermo Fisher) and RNA was isolated following the manufacturer’s instructions. cDNA libraries for bulk-RNA sequencing were prepared using the TruSeq kit (Illumina) and sequenced using HiSeq 2500 at the Johns Hopkins University Sequencing Core. Raw sequencing reads were trimmed using Trimmomatic (0.36)^89,93. Processed reads were mapped to the mm10 reference genome using HISAT2 (2.0.4)⁹⁴. Counts were assembled using Subread featureCounts (1.5.2)⁹⁵. Gene expression analyses were done using the DESeq2 (R package)⁹⁶. Gene ontology (GO) analyses were performed using geneontology.org.

Real time quantitative PCR (qPCR)

RNA was isolated from hPSC-CMs and mouse hearts using Trizol, and cDNA was generated using the high-capacity cDNA reverse transcription kit (Applied Biosystems). qPCR reactions were performed using the Sybr Select mix (ThermoFisher) with indicated primers (Supplementary Tables 2 & 3)⁸⁹. Gene expression levels were normalized to Beta 2 microglobulin (B2M) in mouse and hPSC-CM samples, whereas ribosomal protein L32, (Rpl32) was used to normalized gene expression levels in Drosophila samples.

Western blotting

Protein samples were processed from mouse ventricles homogenized in RIPA buffer (Cell Signaling) with 0.1% PMSF (Sigma) and PhosStop (Roche). Protein concentration was determined by Pierce bicinchoninic acid (BCA) assay kit (Thermo Fisher). For ACTN2 protein measurements, protein samples were isolated and prepared using NuPAGE LDS4X sample buffer (Thermo Fisher) with 1% DTT (Thermo Fisher) and boiled for 5 minutes and protein concentration was subsequently quantified using EZQ protein quantification kit (Invitrogen). Electrophoresis for all samples was performed in TGX gels (Bio-Rad) and proteins were transferred onto nitrocellulose membranes⁸⁹. The following primary antibodies were used for overnight immunoblotting at 4°C: α-actinin (Abcam, Ab68167), p70S6K (Cell Signaling, 2708), Phospho-p70S6K (Cell Signaling, 9205), Phospho-4EBP1 (Cell Signaling, 9451), 4EBP1 (Cell Signaling, 53H11), HSP90AA1 (ABC, A13501) (Supplementary Table 5). The membranes were probed with appropriate IRDye secondary-fluorescent conjugated antibodies (Li-Cor, 926-32213, 926-32212, 926-68073, 926-68072) (Supplementary Table 5) at room temperature. Total protein staining (Li-Cor) was used for sample normalization. Membranes were visualized with an infrared imaging system (Odyssey, Li-Cor) and bands were quantified with Image Studio 5.2.5 (Li-Cor)⁸⁹.

Co-Immunoprecipitation

300μg of protein extracted from hPSC-CMs cultures supplemented with protease inhibitors, was mixed with 1.5μg of α-actinin antibody (Abcam, Ab68167) followed by the addition of protein G conjugated to Dynabeads (ThermoFisher Scientific) to isolate antigen-antibody complexes per manufacturer’s instructions. Western blot analysis was performed using antibodies for α-actinin (Abcam, Ab68167) and HSP90AA1 (ABC, A13501) ⁹⁷.

Oxygen consumption measurement

Respiration rates were measured with Seahorse Xfe96 Analyzer. Day 25 hPSC-CMs were plated at ~60,000 cells per well of a 96 well XF96 Cell Culture Microplate (Agilent Technologies) and cultured for ~1h in Seahorse assay medium (0.55 mg/ml pyruvate in base medium, pH7.4). For mTORC1 inhibition, hPSC-CMs were treated with 10μM of everolimus (Tocris) for 96 hours prior to Seahorse measurements. CMs isolated from P14 and P40 mice were similarly plated at 5,000 cells/well and cultured for 24 h in M199 media. The media was switched to phenol free Seahorse media prior to the measurements. Cells were sequentially treated with oligomycin A (1 μM), FCCP (1 μM), and rotenone (1 μM) with antimycin A (1 μM) and oxygen consumption rate (OCR) and extracellular acidification rate (ECR) were recorded. All data was analyzed using the Seahorse Wave Desktop software (Agilent).

Chromatin immunoprecipitation (ChIP) analysis

DNA isolation for ChIP qPCR analysis was performed using a chromatin immunoprecipitation kit (ab500, Abcam). Briefly, day 20 hPSC-CMs or mouse hearts from E15.5 and P40 WT C57BL/6 mice, were finely minced in the presence of protease inhibitor cocktail (Sigma) and fixed in 1% paraformaldehyde followed by quenching with 1x glycine. Cells and heart tissues were subsequently homogenized and lysed using RIPA Lysis buffer supplemented with 0.1% PMSF (Sigma) and PhosStop (Roche). Chromatin bound DNA was fragmented at 15 sec pulses at 50% amplitudes output with 30 sec rest and repeated for six times. DNA shearing was confirmed by DNA electrophoresis. Samples were incubated with MEF2C (Cell Signaling, 5030S), MEF2A (Invitrogen, PA5-27380), TEAD1 (Cell Signaling, 12292S) and GATA4 (Santa Cruz, Sc-25310) ChIP-grade antibodies at 4°C overnight (Supplementary Table 5). Normal mouse IgG antibody (Millipore, 12-371) was used as negative control (Supplementary Table 5). For immunoprecipitation, samples were subsequently incubated with protein A beads for 1h and were reverse cross-linked with proteinase K at 55°C for 1h. Following this, DNA was purified using DNA purifying slurry and nuclease free water. qPCR was performed using Sybr Select qPCR mix with specific primers (Supplementary Table 4). Intergenic control primers were used as negative controls to correct for background. Fold enrichment of DNA fragments was determined by normalizing cycle thresholds of IgG immunoprecipitated samples.

siRNA, CRISPRi, and enhancer CRISPRa experiments

For small interfering RNA (siRNA), predesigned control (catalog number: 51-01-14-03) and HSP90AA1 (catalog number: hs.Ri.HSP90AA1.13.1 and hs.Ri.HSP90AA1.13.3) dicer-substrate short interfering RNAs (DsiRNAs) were purchased from Integrated DNA Technologies (IDT). For CRISPR interference (CRISPRi), a hPSC line carrying a constitutively expressed deactivated Cas9-KRAB, was obtained from Coriell Institute (AICS-0090-391). sgRNAs for CRISPRi were custom designed and purchased from Synthego. Day 20 hPSC-CMs for both siRNA and CRISPRi were transfected with respective sgRNAs (Supplementary Table 1) using Lipofectamine RNAimax (ThermoFisher) following the manufacturer’s instructions.

For enhancer CRISPR activation (enCRISPRa) experiments, a hPSC line was generated from the parent hESC line (H9). Specifically, the deactivated Cas9 fused with P300 Core (cloned from pcDNA-dCas9-P300 Core plasmid, Addgene #61357) under the constitutive expression of a CAG promoter, was knocked in the AAVS locus using CRISPR/Cas9. CRISPick⁹⁸ was used to select three sgRNAs that were then cloned in scaffolds (Supplementary Table 1) containing the transcriptional activator VP64 and mCherry from Lenti sgRNA (MS2) Zero backbone (Addgene #61427) and pJZC34 (Addgene #62331) plasmids. Three different gRNA-VP64-mCherry plasmids targeting the conserved ACTN2 enhancer region were used. Lipofectamine Stem reagent (ThermoFisher) was used to transfect day 20 hPSC-CMs according to the manufacturer’s instructions. Cells were isolated two days post-transfection.

Statistical analyses

All studies in cultured cells were performed using at least 3 sets of independent experiments. For in vivo studies at least 5 animals in each group were analyzed. For all experiments, the numbers of samples utilized are specified in the respective figure legends. Shapiro-Wilk test was performed to assess normal distribution, and parametric or non-parametric tests were used as appropriate for two or more group analyses as reported in the respective figure legends. P value < 0.05 was considered significant. Data were presented as mean±SEM. Graphs and statistical analysis were performed using Graphpad Prism V10. Scientific illustrations were created using BioRender. For RNA-seq analysis, Benjamini-Hochberg correction was used to adjust for multiple testing, with threshold of adjusted p-value < 0.05 (i.e. false discovery rate < 10%) considered significant⁸⁹. For mouse Gene Ontology analysis only pathways with p-value < 10⁻⁵ were reported. For human single cell analyses, genes with p-value < 0.05 were analyzed.

Extended Data

Extended Data Fig. 1.

A genetic variant linked with heart failure (rs535411) is within a highly evolutionary conserved region of the ACTN2 transcriptional enhancer. A second variant (rs663798) also associated with heart failure is upstream of the conserved region.

Extended Data Fig. 2.

a. Gating strategy used to detect differentiated hPSC-CMs. b. Flow cytometry analysis of ACTN2 enh del hPSC-CMs vs isogenic controls after immunostaining for cardiac Troponin T. The differentiation efficiency of both hPSC lines is similar. A sample without primary Troponin-T antibody was used as a negative control. c. Map illustrating the alternatively spliced ACTN2 variants, and the RTPCR products based on the different primer sets. d. RT-PCR using cDNA generated from ACTN2 enh del hPSC-CMs vs isogenic controls showing no difference in the size of alternatively spliced ACTN2 transcripts. The same results were obtained after repeating the experiment 4 independent times.

Extended Data Fig. 3.

Entropy score calculated from hPSC-CM our single cell RNA-sequencing data (one differentiation batch) showed increased entropy in ACTN2 enh del cardiomyocytes consistent with disrupted maturation. The center line of the box corresponds to the 50th percentile. The lower and upper bounds of the boxes correspond to the 25th and the 75th percentiles. The upper whisker marks the largest value within 1.5 times interquartile range above the 75th percentile. The lower whisker marks the smallest value within 1.5 times interquartile range below the 25th percentile. The interquartile range is defined as the distance between the first and the third quartiles.

Extended Data Fig. 4.

a. Illustration of the different luciferase vectors tested. b. Luciferase analysis of hPSC-CMs transfected with minimal promoter vs the conserved ACTN2 enhancer sequence with and without the rs535411 variant, showing increased luciferase activity in the presence of ACTN2 enhancer. This effect was not observed in cells transfected with the mutant variant. c-d. Overexpression of MEF2A and MEF2C transcription factors through modified mRNA transfection in hPSC-CMs resulted in markedly increased luciferase activity in both enhancer vectors. However, this induction was significantly lower in cells with ACTN2 rs535411 enhancer variant. All graphs report Luciferase compared to minimal promoter.(4 independent experiments were analyzed). Data are presented as mean values +/− SEM. The Shapiro-Wilk test was performed to assess normal distribution, and one-way ANOVA with Bonferroni’s multiple comparisons test or Kruskal-Wallis with Dunn’s multiple comparisons test were used as appropriate. Only P values <0.1 are reported.

Extended Data Fig. 5.

a. Analysis of ENCODE ChIP-Seq datasets for H3K27Ac and H3K4me1 from mouse hearts at different developmental stages showed increased ACTN2 enhancer activation in older mice supporting a role in the temporal regulation of ACTN2 gene expression. b. Analysis of P300 occupancy data from mouse hearts at different developmental stages, similarly, showed an increased in P300 peaks in postnatal hearts. c. Mef2a ChIP-Seq data showing binding of Mef2a in the ACTN2 enhancer region in adult mouse hearts.

Extended Data Fig. 6.

a. H3K27Ac ChIP-Sequencing analysis in Drosophila melanogaster reveals an extra active enhancer region upstream of the actinin (Actn) gene promoter that is only present in adult flies. b. Alternatively spliced Actn gene variants (Actn-RB, RC, RD) are expressed at different developmental stages (larva vs fly), likely temporarily regulated by different transcriptional enhancers such as the one revealed by H3K27Ac ChIP-Seq. 5 independent experiments were analyzed.

Extended Data Fig. 7.

Actn2 protein is decreased in both heterozygous and homozygous Actn2 enh del mouse hearts. 4 samples per group were analyzed. The Shapiro-Wilk test was performed to assess normal distribution. Student t-test was used to compare the different groups.

Extended Data Fig. 8.

a. Gene expression of RAPTOR and 3-phosphoinositidine-dependent protein kinase-1 (PDK1) as part of mTORC1 pathway were upregulated, while RICTOR levels which is part of the mTORC2 complex remained unchanged. (3 independent samples per group). b. Both ACTN2 enh del hPSC-CMs and isogenic controls showed very similar cell size after treatment with everolimus (total ~147 cells/group, 4 indipendent treatments). c. The morphology of ACTN2 enh del hPSC-CMs did not show any significant difference as compared to controls after treatment with everolimus. (4 independent samples per group) (bar graph: 25μm) d. ACTN2 enh del hPSC-CMs treated with everolimus (25μM, mTORC1 inhibitor) normalized their calcium handling properties as compared to treated control hPSC-CMs. Time to peak, peak height and time to 50% to baseline are not different between control and ACTN2 enh del hPSC-CMs.(n=8-10 batches per group). e. Seahorse analysis of cultured hPSC-CMs show reduced maximum oxygen consumption rate (OCR) and ECAR to OCR ratio (consistent with increased glycolysis) for ACTN2 enh del hPSC-CMs compared to isogenic controls. Treatment with everolimus, improved oxygen consumption for both for ACTN2 enh del hPSC-CMs and isogenic controls and normalized glycolysis for ACTN2 enh del hPSC-CMs. (4 independent samples per group). f. Both ACTN2 enh del hPSC-CMs and isogenic controls developed very similar gene expression of maturation related genes after everolimus treatment (3 independent samples per group). The Shapiro-Wilk test was performed to assess normal distribution, and the parametric student t-test or ANOVA with Bonferroni’s multiple comparisons test or the non-parametric Mann-Whitney or Kruskal Willis tests were used as appropriate. Only P values <0.1 are reported.

Extended Data Fig. 9.

a. Co-immunoprecipitation of ACTN2 with HSPA90A in both control and ACTN2 enh del hPSC-CMs, supporting the direct interaction between the two proteins. Despite the reduced ACTN2 levels in ACTN2 enh del hPSC-CMs, the ACTN2 immunoprecipitate appears more enriched with the inducible chaperone HSPA90A. b. Transfection of hPSC-CMs with a mix of HSP90AA1 siRNAs suppressed HSP90A protein levels in control hPSC-CMs. (3 independent experiments were analyzed) c. siRNA downregulation of HSP90A in ACTN2 enh del hPSC-CMs. (3 independent experiments were analyzed). The Shapiro-Wilk test was performed to assess normal distribution, and the paramentric student t-test was used for all comparisons. Only P values <0.1 are reported.

Extended Data Fig. 10.

Transfection of hPSC-CMs with sgRNAs targeting the 5’ border of the conserved ACTN2 enhancer region, upregulated ACTN2 protein. Relative protein expression compared to sgRNA1 is presented. (3 independent experiments were analyzed). Data are presented as mean values +/− SEM. The Shapiro-Wilk test was performed to assess normal distribution, and paired student parametric t-test (two-tailed) was used for all comparisons.

Data availability

The bulk RNA-Seq data and single cell RNA-Seq data are available at the Gene Expression Omnibus . Source data are also provided with this manuscript.