Advancing physiological maturation in human iPSC-derived cardiac muscle by gene editing an inducible adult troponin isoform switch

Abstract

Advancing maturation of stem cell-derived cardiac muscle represents a major barrier to progress in cardiac regenerative medicine. Cardiac muscle maturation involves a myriad of gene, protein and cell-based transitions, spanning across all aspects of cardiac muscle form and function. We focused here on a key developmentally controlled transition in the cardiac sarcomere, the functional unit of the heart. Using a gene-editing platform, human iPSCs were engineered with a drug-inducible expression cassette driving the adult cardiac troponin I (cTnI) regulatory isoform, a transition shown to be a rate-limiting step in advancing sarcomeric maturation of hiPSC cardiac muscle (hiPSC-CM) toward the adult state. Findings show that induction of the adult cTnI isoform resulted in the physiological acquisition of adult-like cardiac contractile function in hiPSC-CMs in vitro. Specifically, cTnI induction accelerated relaxation kinetics at baseline conditions, a result independent of alterations in the kinetics of the intracellular Ca²⁺ transient. In comparison, isogenic unedited hiPSC-CMs had no cTnI induction and no change in relaxation function. Temporal control of adult cTnI isoform induction did not alter other developmentally regulated sarcomere transitions, including myosin heavy chain isoform expression, nor did it affect expression of SERCA2a or phospholamban. Taken together, precision targeting of sarcomere maturation via inducible TnI isoform switching enables physiologically relevant adult myocardium-like contractile adaptations that are essential for beat-to-beat modulation of adult human heart performance. These findings have relevance to hiPSC-CM structure-function and drug-discovery studies in vitro, as well as for potential future clinical applications of physiologically optimized hiPSC-CM in cardiac regeneration/repair.

Graphical Abstract

hiPSC-CMs are gene edited to contain an inducible cardiac troponin I (cTnI) expression cassette. When these cells are treated with doxycycline, they express cTnI. Expression of cTnI is sufficient to promote a phenotype with physiologic markers of mature myocytes, including faster relaxation and improved adrenergic responsivity, compared to nonedited cells which only express slow skeletal troponin I.

1 |. INTRODUCTION

The dynamic intersection between the fields of cardiac physiology and regenerative biology has been enabled by recent advances in generating pluripotent stem cell-derived cardiac muscle. More specifically, the ability to reprogram human somatic cells into a pluripotent stem cell (hiPSC) state, and then to differentiate hiPSCs into beating cardiac myocytes (hiPSC-CMs) has provided a significant boost to the field [1–3]. As an in vitro cellular platform, hiPSCs, with directed differentiation into hiPSC-CMs, provide a ready and essentially unlimited source of beating cardiac muscle that express human genes and proteins assembled in sarcomeric organization. Thus, hiPSC-CMs are highly useful in a multitude of applications, including studies of basic cardiac muscle development, physiology, drug discovery, and for disease related genotype-phenotype analysis. In addition, although significant obstacles remain [4], hiPSC-CMs may one day have meaningful therapeutic applications in the setting of cardiac regenerative medicine.

One often cited limitation of hiPSC-CMs centers on their immature status [5–7]. Accordingly, advancing the maturation of hiPSC-CMs remains a significant goal for the field. One difficulty in this endeavor centers on defining maturation, including the establishment of high fidelity hiPSC-CM functional maturation “bio-markers.” In mammalian cardiac muscle development, including in humans, there are numerous developmentally programmed transitions central to the functional maturation of cardiac muscle transitioning from the fetal/neonatal state to the mature adult state. These developmental transitions include changes in overall myocyte morphology, the alignment, content and organization of contractile myofilaments in sarcomeres, membrane channels/electrophysiology, cell surface receptors/signaling networks, metabolic pathway maturation, and others [7]. These developmental transitions are governed in part by a programmed genetic program that, in turn, is responsive to both cardiac myocyte intrinsic and cell extrinsic cues. One difficulty in tracking hiPSC-CM maturation is that nearly all these developmentally guided transitions are reversible, making assessment of hiPSC-CM maturation challenging.

Recently, some clarity has emerged in defining hiPSC-CM maturation status by focusing on the developing cardiac sarcomere [6,7]. The sarcomere is the functional unit of cardiac muscle governing overall heart performance [8]. The sarcomere consists of an intricate array of contractile and regulatory proteins aligned in stoichiometric balance, forming a near-liquid crystalline contractile apparatus [8,9]. Thus, while ion channels, Ca²⁺ handling and energetics are vital to myocyte function, the sarcomere is the singular integrated functional unit solely responsible for generating force and power. Based on this premise, targeting mechanistic dissection of the human sarcomere will be essential in forging new opportunities in advancing hiPSC-CM maturation with the ultimate goal of treating the diseased heart.

Potential maturation candidates, such as overall cell morphology, sarcomeric organization, ion channels, cell energetics, and many of the genetically encoded sarcomeric isoform developmental switches, are reversible in conditions of stress and disease. Accordingly, the inherent mutability of these key elements makes their use as physiologically relevant maturation markers difficult. To begin to address this we have recently reported cardiac sarcomere maturation as directed by specific genetic switches governing multiple sarcomeric isoform transitions [6,7]. It has been established that fetal mammalian cardiac myocytes, including in humans, exclusively express the slow skeletal form of the thin myofilament regulatory protein troponin I (ssTnI) [6,7]. Then, shortly after birth, a developmentally directed gene-regulated troponin I isoform switch occurs such that a stoichiometrically conserved transition results from the fetal ssTnI isoform to the adult cardiac TnI (cTnI) isoform in the adult heart. This cardiac maturation switch mechanism has important characteristics, including irreversibility. We recently identified this isoform transition as a key maturation roadblock in the formation of adult cardiac muscle-like hiPSC-CMs, where myocytes are stalled in an immature, ssTnI isoform state [6,7].

The ssTnI to cTnI isoform switch is also well known to confer physiologically relevant alterations in sarcomere mechanics that are essential for overall adult heart pump performance in health and disease [8]. The cTnI isoform plays a critical role in governing cardiac relaxation performance during stress and during β-adrenergic stimulation by decreasing Ca²⁺ sensitivity of the thin filament to facilitate faster mechanical relaxation function [10]. This effect of cTnI is instrumental to conferring the physiological underpinnings of the heart fight or flight response [8].

In this study, we sought to advance physiologically relevant enhancement of hiPSC-CM function by a gene editing-directed inducible TnI isoform switch. Using a TALENs-based gene editing platform, we engineered a doxycycline-inducible expression system into the AAVS1 safe harbor site to induce with temporal control the expression of the adult cTnI isoform in hiPSC-CMs (Figure 1). In principle, this system enables experimenter-directed control of the “turning on” of expression of the adult cTnI isoform during the in vitro development of hiPSC-CMs. We first tested whether this system would be sufficient to cause detectable adult cTnI isoform induction in hiPSC-CMs. Second, we tested the hypothesis that cTnI isoform induction would be sufficient to cause physiologically detectable effects on hiPSC-CMs contractile function. Finally, we tested whether cTnI isoform induction would present a sufficient signal to initiate other maturation transitions, within and outside the context of the sarcomere.

Figure 1.

Genome editing strategy for induction of cTnI. a, Timeline schematic overview for differentiation of hiPSCs to hiPSC-CMs and for the study of hiPSC-CM structure–function before and after cTnI induction by Dox (note: 6 week time point start Dox and at 8 weeks end Dox [2 weeks Dox] and analyze). b, Genome editing strategy for AAVS1 locus targeting (vertical purple bar) of the drug inducible cTnI expression cassette. Vertical green bar in cTnI indicates the C-terminal engineered Flag epitope

2 |. MATERIALS AND METHODS

2.1 |. HiPSC Culture and Differentiation

For the human induced pluripotent stem cell line, we utilized the well characterized DF 19–9–11 line, generously provided to us by Dr. Tim Kamp. This hiPSC line was derived from newborn male foreskin dermal fibroblasts and is transgene and vector free, karyotype normal, pluripotent, and shown to have excellent capacity to differentiate into cardiac myocytes [2,6,11–13]. HiPSCs were cultured as previously described [6,11]; cells were grown in TeSR-E8 media (Stemcell, Vancouver, CA) in plastic dishes coated with Matrigel (Corning, Corning, NY), passaged by 0.5 mM EDTA in PBS when they reached 90% con-fluency, and replated with Rho kinase inhibitor 10 μM Y-27632 (Selleckchem, Houston, TX).

HiPSCs were differentiated when they reached 95–100% con-fluency in 6 well plates according to the small molecule Matrigel sandwich method published previously [2]. At day 0, cells were treated with Matrigel dissolved in RPMI + B27 supplement minus insulin (Thermo Fisher, Waltham, MA) with 6 μM CHIR99021 (Stemgent, Lexington, MA), which is a GSK3 inhibitor. On day 2, media was changed without addition of small molecules. On day 3, cells were treated with 8 μM IWP-4 (Stemgent, Lexington, MA), an inhibitor of Wnt signaling, in RPMI + B27 supplement minus insulin. On day 5 and every 2 days after that, media was changed until cells started to beat vigorously, which is at approximately days 9–13, at which point media was switched to RPMI plus B27, insulin replete. HiPSC-CMs underwent glucose starvation as previously described [14]. Media was replaced every 2 days. All RPMI + B27 was supplemented with antibiotic-antimycotic (Thermo Fisher Scientific, Waltham, MA). For all assays that involved dissociation and replating of cells, cells were dissociated with Accutase (Thermo Fisher, Waltham, MA) for 30 minutes at room temperature, followed by mechanical dissociation with a P1000 pipet and centrifugation for 3 minutes at 800 rpm.

2.2 |. Genome Editing

Genome editing was performed using TALENs designed specifically to the AAVS1 locus on chromosome 19. We designed and implemented a template plasmid containing homology arms specific to the targeted AAVS1 site predicated on extensive past works [15–18], that also included a puromycin resistant gene locus, and a tetracycline-inducible promotor-effector system (Figure 1) pAAVS1-NC-CRISPRi; pAAVS1-NDi-CRISPRi; pAAVS1-NDi-CRISPRi; AAVS1-TALEN-L and AAVS1-TALEN-R. This TALENs AAVS1 targeting platform has been well validated with excellent locus targeting efficiency [19]. As shown in Figure 1, the puromycin marker has no self-promoter; instead, it is in tandem with the endogenous AAVS1 promoter that drives T2A puromycin expression. Thus, expression of puromycin can only occur when the construct integrates correctly and specifically within AAVS1, markedly reducing off-target integration (random integration) in the presence of puromycin selection, and this is well validated in previous works for efficient targeting to the AAVS1 site without off-target effects [20]. Although we did not directly assess editing in terms of homozygosity vs heterozygosity, based on past studies, the heterozygous to homozygous ratio can be expected to be between ~2.5:1 [11] and 8:1 (unpublished). Thus, we can estimate that the line studied was most likely heterozygous. The human cTnI coding sequence was designed with restriction sites to fit into the template plasmid and was constructed by Integrated DNA Technologies (Coralville, IA). It was then cloned into digested template plasmids. In addition, to aid in distinguishing the induced cTnI protein from endogenous cTnI, we engineered a FLAG epitope into the C-terminus of cTnI (Figure 1; green vertical bar). This FLAG epitope was shown previously to have no detectable effects of cardiac muscle structure or function in cardiac myocytes from rodents and from adult human explanted hearts [8,21,22].

All three plasmids (forward TALEN, reverse TALEN, and the cTnI engineered template) were transfected into hiPSCs at a concentration of 15 μg per 400 μL cuvette by electroporation using the Bio-Rad Gene Pulser Xcell (Bio-Rad, Hercules, CA) with an exponential waveform that had parameters of 250 V, 400 μF, and infinite resistance. Cells were replated onto 25 mm wells coated with Matrigel and allowed to recover for 3–4 days. Puromycin selection was done starting with 0.2 μg/mL and increasing to 1 μg/mL over the course of 5–7 days. Individual surviving colonies were identified, dissociated, replated, and continued to be passaged as in the above methods. Based on our experiences, we have confidence that the clones selected are independent. However, confirming this point is technically challenging. Following nucleofection, we immediately performed serial dilutions in 6 well plates and picked clones from independent wells. Specifically, after 7 days of puromycin selection, individual hiPSCs clones were carefully picked under the microscope and expanded. Based on this approach, ten independent clones of hiPSCs were analyzed and we selected a clone with clear cTnI-Flag detection for further structure-function analysis. In selecting clones, we performed QC based on established protocols of characteristic iPSC colony morphology. The second and most critical QC metric for this study was to then assess differentiation efficiency into cardiac myocytes. Based on our experiences, iPSCs clones derived from iPSCs with normal karyotypes (as already demonstrated for the DF-19–9–11 line we are using here [11]) have excellent differentiation into cardiac myocytes [6]. HiPSCs collected for RNA samples were treated with doxycycline on days 1 and 2 after passaging. In order to express the gene of interest in hiPSC-CMs, differentiated cells were treated with doxycycline from day 46 to day 60, so that experiments were carried out on d60 cells after 2 weeks of gene induction.

2.3 |. RNA Isolation and Real-Time PCR

RNA was collected from undifferentiated hiPSCs before doxycycline and after 24 and 48 hours of treatment with doxycycline. HiPSCs were isolated with EDTA as in the passaging protocol mentioned above, then spun at 800 RPM for 3 minutes and treated with the RNEasy Plus kit (Qiagen, Hilden, Germany) to isolate the RNA. RNA was also collected from differentiated 2-month-old untreated hiPSC-CMs and from hiPSC-CMs that went through 2 weeks of treatment with doxycycline. HiPSC-CMs were dissociated with Accutase (Thermo Fisher, Waltham, MA) for 5 minutes at 37° Celsius then spun at 800 RPM for 3 minutes and treated with the RNEasy Plus kit. All samples were quantified by NanoDrop spectrophotometry (NanoDrop, Wilmington, DE). cDNA libraries were created using the Superscript VILO kit (Thermo Fisher, Waltham, MA) and quantitative real-time PCR was performed using the Applied Biosystems PowerUp SYBR Green Master Mix (Applied Biosystems, Foster City, CA) on an Eppendorf Mastercycler machine (Eppendorf, Hamburg, Germany). Three technical replicates were used for each sample. qPCR primers were designed using the Integrated DNA Technology PrimerQuest tool (IDT, Coralville, IA) and synthesized by IDT (Supplemental Table 1). Analysis of gene expression and fold change was done using the ddCt method, with GAPDH used as a housekeeping gene.

2.4 |. RNA Sequencing

RNA sequencing was performed by the Bioinformatics group at the Morgridge Institute for Research at the University of Wisconsin Madison. HiPSC-CMs were cultured and dissociated as above and RNA was extracted from dissociated cells using the RNEasy Plus kit (Qiagen, Hilden, Germany) and quantified by NanoDrop spectrophotometry (NanoDrop, Wilmington, DE). Samples were processed for quality control on an Agilent Bioanalyzer (Agilent, Santa Clara, CA) and a Qubit Fluorometer (Thermo Fisher, Waltham, MA). Libraries were created using Ligation Mediated RNA sequencing (LM-seq) and sequenced on a HiSeq 2500 (Illumina, San Diego, CA). Transcript counts were reported as transcripts per kilobase million (TPM) and mean values were compared between groups for each gene. In order to account for potential differences in heterogeneity of differentiated cell populations (cardiac myocyte vs other cell types) as well as proliferation or death of cardiac myocytes, we normalized TPM values of each gene to TnC TPM values for that group. Bonferroni statistical tests to calculate P values were performed using CLC Sequence Viewer (Qiagen, Hilden, Germany). Gene groups were clustered using DAVID online gene ontology software (https://david.ncifcrf.gov/).

2.5 |. Western Blot, Immunofluorescence and Protein Quantification

Protein was extracted by mechanical dissociation in RIPA buffer containing protease inhibitor. Protein samples were then split off the original tubes and had 100 μL 0.15% DOC along with 100uL of 72% TCA added to them to form a precipitate. Samples containing the precipitate were centrifuged for 30 min at 3000 rpm to form a pellet and then the supernatant was discarded. Quantification was then done with a Pierce modified Lowry protein assay kit (Thermo Fisher, Waltham, MA). Original samples were denatured by boiling. 20 μg of protein in Laemmli buffer with β-mercaptoethanol was loaded into each lane, and samples were run on 12% SDS-PAGE gels at 120 V until the dye front reached the end of the gel. Protein was transferred to PVDF membranes. Membranes were stained for pan-TnI (1E7, Novus, Littleton, CO) at a 1:1000 dilution in TBS-T with 5% milk. Membranes were imaged on the LI-COR Odyssey (LI-COR, Lincoln, NE). In our hands, 1E7 is an excellent pan TnI ssTnI/cTnI antibody in the setting of human iPSC-CMs. In addition, we engineered cTnI with an epitope FLAG tag to uniquely identify this isoform, as the cTnI-FLAG protein migrates more slowly than cTnI.

2.5.1 |. Immunofluorescence

HiPSCs were grown and differentiated on plastic plates into hiPSC-CMs for 9 days, and then fed doxycycline at a concentration of 1ug/mL daily for 3 days. HiPSC-CMs were replated onto glass plates coated with laminin and fixed in a 4% paraformaldehyde solution. Heat induced epitope retrieval was performed with Tris-EDTA buffer (pH 9) at 96°C for half an hour to unmask antigens. Cells were permeabilized for 15 minutes with 1% Tween-20 and 1% BSA in PBS. Prior to incubation with antibodies, hiPSC-CMs were blocked with 5% goat serum, 94.7% PBS, 0.3% triton, and 5% BSA. Primary antibody mouse monoclonal M2-FLAG (Sigma-Aldrich, catalog #F3165) in blocking solution (diluted at 1:500) was incubated overnight at 4oC and then washed off with PBS the next day. Secondary antibody goat anti-mouse AF488 (ThermoFisher Scientific, catalog #A-11001) was incubated for 2 hours at a 1:1000 dilution in blocking solution. Images were taken at 60X magnification. For the inset in Figure 3b, Flag was detected as above plus rabbit α-actinin antibody (Abcam catalog #ab90776) used at a concentration of 1ug/mL together with goat anti-rabbit AF488 (Invitrogen catalog #A-11034) and goat anti-mouse AF568 for flag detection (Invitrogen catalog #A11031) at a concentration of 2ug/mL.

Figure 3.

Documentation of cTnI induction at the mRNA and protein levels in hiPSC-CMs. a, Summary of cTnI mRNA levels in edited hiPSC-CMs after doxycycline induction, normalized to GAPDH. Values are mean +/− SEM, n = 12–25. ***P < 0.0001. b, Western blot detection of TnI isoform content in edited and unedited hiPSC-CMs, +/− Dox. H = explanted human heart sample as positive control for cTnI. M = lane markers. The epitope tagged induced cTnI was only evident in the edited + Dox lanes. c, Immunofluorescence detection (green channel) of sarcomeric localized epitope-tagged cTnI isoform in edited hiPSC-CMs post Dox. Scale bar is 50 um. Inset: In separate dual immuno-labeling experiments of edited hiPSC-CMs post Dox anti-α-actinin (green) and anti-Flag (red) shows cTnI (red) located between the Z-lines demarcated by α-actinin (green). Scale par is 2 um. d, Summary of Western blot date of the cTnI-ssTnI isoform ratio in hiPSC-CMs Values are mean +/− SEM n = 4–6. ** P < 0.01, ns, not significant

2.6 |. Measurement of hiPSC-CM Contractility

For studies on contractile function, 35-day-old differentiated hiPSC-CMs were dissociated and replated at a density of 60 000 cells per well into a 96-well NanIon Sensor Plate (NanIon, München, Germany) that had been coated with Matrigel, and allowed to adhere 18 hours. Media was changed after 18 hours and then every 2 days after that. HiPSC-CMs generally began beating again 7–10 days after replating. HiPSC-CMs were treated with doxycycline (Dox) for 2 weeks, starting at 6 weeks (initial impedance measurements were obtained) and ending Dox at 8 weeks, at which time the post Dox induction impedance measurements were obtained from the same well (Figure 1a). Impedance was measured on the NanIon CardioExcyte 96. Media was changed 3 hours before recording to ensure optimal metabolic conditions. Measurements were taken at 5-minute intervals over the course of 1 hour in an environmentally controlled chamber at 37°C and 5% CO₂. Isoproterenol was used at a concentration of 10 nM. Cells formed syncytia after several days in culture and contracted simultaneously and spontaneously; cells were not electrically stimulated, as attempts to do so interfered with the ability of the machine to collect accurate data. Data was analyzed using a custom MATLAB script (MathWorks, Natick, MA) by normalizing data to a baseline and using LOESS curve fitting on the raw impedance data.

2.7 |. Calcium Imaging

Differentiated hiPSC-CMs grown in a monolayer were loaded with Fura-2 AM (Thermo Fisher, Waltham, MA) at 1 μM in phenol red-free RPMI + B27 for 10 minutes at room temperature, then washed twice and allowed to de-esterify for 10 minutes at 37°C. Isoproterenol was used at a concentration of 10 nM, as previously described [23–25]. Experiments were carried out at 37°C using the IonOptix myocyte calcium and contractility system (IonOptix, Westwood, MA). Ratiometric calcium transient curves were fitted and analyzed using a custom MATLAB script.

2.8 |. Statistical Analysis

All statistical analysis was done using Prism (GraphPad, San Diego, CA). Paired data was analyzed by Student’s t-test. Impedance data and calcium data were analyzed by One-way ANOVA with Geisser-Greenhouse correction and Tukey’s multiple comparisons test, presented as mean ± SEM. For RNA sequencing, Bonferroni statistical tests to calculate P values were performed using CLC Sequence Viewer (Qiagen, Hilden, Germany).

3 |. RESULTS

3.1 |. Genome edited inducible cTnI isoform expression system

HiPSCs were gene edited using a pair of well-validated TALENs plasmids directed to the AAVS1 “safe harbor” locus on chromosome 19 (Figure 1). This system employs a puromycin drug resistance gene that, upon correct targeted insertion into the AAV1 locus, aligns in frame with an endogenous AAVS1 - T2A promoter to confer drug resistance to limit off-target insertion [20]. Here the human cTnI cDNA was cloned into the TALENs template plasmid containing sequences homologous to the AAVS1 site, allowing the human cTnI cDNA to be under temporal regulation via a tetracycline-inducible promoter system. These elements were introduced into hiPSCs and subsequently collected for identifying homology-directed repair (HDR) in selected clones. As above, to aid in the selection for HDR, the targeting construct included a puromycin-resistance cassette to confirm AAVS1 site integration and to limit off-target effects (Figure 1). Multiple independent edited hiPSC lines were identified by antibiotic resistance and subsequent genomic conformation of the targeted locus. For the following structure-function studies, we selected a clone with drug resistance and that displayed the characteristics of normal iPSC colony morphology, excellent cTnI transcriptional induction and with differentiation efficiency into beating cardiac myocytes not different from the isogenic DF-19–9–11 parental line.

To examine the effectiveness of this tetracycline-inducible promoter system, detection of the expression of cTnI mRNA was examined in undifferentiated hiPSCs before doxycycline treatment and after 24 and 48 hours of doxycycline treatment (Dox) (Figure 2a). Here qRT-PCR showed a marked increased cTnI mRNA expression after 24 and 48 hours of Dox induction, with no evidence of transcriptional leakiness in the absence of Dox, indicating high fidelity and rapid responsivity of the system (Figures 2, 3). To elaborate, in edited hiPSCs or hiPSC-CMs there was no significant detection of transgene expression at the mRNA or protein levels in the absence of Dox. In addition, in unedited hiPSCs exposed to Dox, there was little to no evidence of cTnI mRNA or protein induction over the time-course of this study, which is in agreement with our past studies using unedited DF-19–9–11 hiPSCs [6,7].

Figure 2.

Validation of drug-inducible cTnI induction in hiPSCs. a, Experimental time course for inducible expression cassette validation via qRT-PCR in hiPSCs. b, Summary of cTnI mRNA levels in edited, undifferentiated hiPSCs after 24 and 48 hours of doxycycline (Dox), as normalized to GAPDH. Values mean +/− SEM, n = 5 for pre Dox, n = 4 for 24 hours, n = 3 for 48 hours, ****P < 0.0001

Next, hiPSCs were differentiated into hiPSC cardiac myocytes (hiPSC-CMs) using an established small molecule approach [2,12]. At 6-weeks post-differentiation, the spontaneously beating edited (and separately isogenic unedited) hiPSC-CMs were randomized to +/− Dox groups and then studied two weeks later. Prior to Dox treatment, we first determined that the contractile baseline function of edited and unedited hiPSC-CMs were not significantly different, as judged by no difference in FWHM contractile impedance measurement between groups. Thus, because it is known that even within isogenic hiPSC-CMs lines that there is some variability in contractile performance, the edited and unedited hiPSC-CMs were first tested via impedance recordings and then the Dox +/− groups were randomized by ensuring a normal distribution between groups. Stated differently, if by chance +/− Dox groups were sorted with different functional parameters pre-Dox treatment, this would have compromised the accurate assessment of the specific cTnI induction effects on contractile performance. In addition, the rationale for the two week time period for Dox induction was to permit sufficient time for troponin turnover in the sarcomere, which in mammalian cardiac muscle has been estimated at 3.5 days [26].

Edited hiPSC-CMs showed a significant increase in cTnI mRNA following two weeks of Dox, with no Dox induction of cTnI mRNA in unedited hiPSC-CMs or induction in edited hiPSC-CMs in the absence of Dox (Figure 3a). In agreement with past work [6], unedited and cTnI edited hiPSC-CMs had little to no detectable amounts of the endogenous adult isoform cTnI protein expression by Western blot pre-Dox over this time frame (Figure 3b), as revealed by the faint band above ssTnI in edited no Dox, and between the Dox-induced flag tagged cTnI and ssTnI in edited +Dox lanes. This also agrees with the low to non-detectable cTnI mRNA pre-Dox in both groups (Figure 3a). In the Western blot shown in Figure 3, identification of the lower band as ssTnI is based on TnI protein isoform analysis from substantial past works focusing on TnI structure-function in cardiac muscle, including in hiPSC-CMs [6,7,27–29]. These studies provide excellent supportive evidence to substantiate our identification of ssTnI and cTnI. Immunofluorescence-based detection of the induced cTnI isoform (via epitope tagging) showed that the newly expressed cTnI was localized into sarcomeres between the α-actinin labeled Z-lines in hiPSC-CMs (Figure 3c/inset). Importantly, in cTnI edited hiPSC-CMs there was a marked induced expression of cTnI as determined by Western blot, with a 15-fold increase in the cTnI: ssTnI ratio after two weeks of induction (Figure 3b,d). Unedited hiPSC-CMs treated with doxycycline did not express detectable amounts of cTnI over this same time period (Figure 3b,d). Despite the significant cTnI induction in hiPSC-CMs, it is worth noting that the cTnI: ssTnI ratio was below the 100% cTnI isoform content which is a signature of the adult human heart (Figure 3b,d).

3.2 |. HiPSC-CM functional studies: enhanced relaxation performance by induced cTnI expression

We first tested whether the observed induced cTnI isoform expression (Figure 3) would be sufficient to cause any significant effects on contractile performance in hiPSC-CMs. To evaluate hiPSC-CM contractile properties, we used a cellular impedance detection system (Figures 4–6; Supplemental Figure 1). Under controlled temperature and media pH conditions, this system provides data on the kinetics of contraction as reported as left half width (LHW) and 10% to 90% rise time; the speed of relaxation, as reported as right half width (RHW) and 90% to 10% decay time; and the full width half maximum (FWHM) (Figure 4). The amplitude of contraction was recorded; however, it is not useful in comparisons owing to lack of normalization between and within groups.

Figure 4.

Contractility analysis of hiPSC-CMs as measured by impedance. a, Representative impedance traces highlighting relaxation performance in edited hiPSC-CMs before and after Dox, and after Dox + Iso (left panel); definition of impedance trace parameters (right panel). b, Summary of Right Half Width measurements for gene-edited hiPSC-CMs with doxycycline and with isoproterenol. Values mean+/−SEM, n = 17. c, Summary of Decay times for gene-edited hiPSC-CMs with doxycycline and isoproterenol, n = 17. d, Summary of Decay times for isogenic unedited hiPSC-CMs with doxycycline and isoproterenol, n = 7. e, Summary of Right Half Width measurements for isogenic unedited hiPSC-CMs with doxycycline and isoproterenol, n = 7. * = P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Figure 6.

Analysis of the effect of time in culture on the impedance decay and RHW in hiPSC-CMs. a) Right Half Width (RHW) measurements derived from gene-edited hiPSC-CMs at 6 weeks and 8 weeks post-differentiation, and with Iso treatment, n = 9. b) RHW measurements from unedited hiPSC-CMs at 6 weeks and 8 weeks post-differentiation, and with Iso, n = 14. c) Decay times derived from gene-edited hiPSC-CMs at6 weeks and 8 weeks post-differentiation, with Iso, n = 9. d) Decay measurements from unedited hiPSC-CMs at 6 weeks and 8 weeks post-differentiation, with Iso, n = 14. Data calculated as mean ± SEM, One-Way ANOVA with Geisser-Greenhouse correction for analysis. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Contractile effects detected upon cTnI induction were evident in the faster mechanical relaxation rates in edited hiPSC-CMs post Dox, as demonstrated by significantly shortened RHW and faster decay times, as compared to edited hiPSC-CMs pre-Dox (Figures 4, 5). RHW and Decay times were significantly faster post cTnI induction (RHW: 173.4 +/− 5.619, FWHM: 341.079 +/− 8.835, Decay: 232.107 +/− 7.649, n = 17) compared pre-induction (RHW: 203.1 +/− 9.624, n = 17, P < 0.05, FWHM: 389.205 +/− 15.03 n = 17, P < 0.01, Decay: 267.3 +/− 10.48, n = 17 P < 0.05) (Figures 4, 5). This analysis was enabled by employing pair-wise pre-post Dox comparison testing in the same hiPSC-CMs. For comparison, the isogenic unedited hiPSC-CMs, when studied using the same Dox protocol, displayed no significant alterations in RHW or Decay rate relaxation performance (Figures 4,5). In unedited hiPSC-CMs the LHW and Rise times were altered + Dox (Supplemental Figure 1). We do not have a clear understanding of the mechanism of this effect. One possibility relates to potential effects of Dox to modify aspects of sarcomeric performance, as we discuss below.

Figure 5.

Summaries of impedance experiments highlighting beats per minute and full-width half-max times in hiPSC-CMs. a) Impedance trace depiction highlighting the calculation of contractile times used in this study. b) Beats per minute (BPM) of gene-edited cells with doxycycline and isoproterenol. Values +/−SEM, n = 17. c) BPM of unedited hiPSC-CMs with doxycycline and isoproterenol, n = 7. d) BPM of edited hiPSC-CMs from 6 weeks and 8 weeks, with Iso. n = 9 for all 3 groups. e) BPM of unedited hiPSC-CMs tested at 6 weeks and 8 weeks, and with Iso,n = 14. f) Full-width half-max (FWHM) measurements on edited hiPSC-CMs + Dox and Iso, n = 17. g) FWHM for unedited cells + Dox and Iso,n = 7. h) FWHM for edited hiPSC-CMs at 6 weeks and 8 weeks, and with Iso, n = 9. i) FWHM for unedited hiPSC-CMs at 6 weeks and 8 weeks, and with Iso, n = 14. Data shown from b-i was calculated as mean ± SEM, paired One-Way ANOVA with Geisser-Greenhouse correction for analysis. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Because the cTnI isoform is known to be a key sarcomeric target for β-adrenergic PKA-mediated signaling [8,10], we tested whether the contractile function of hiPSC-CMs would be differentially altered by isoproteronol (Iso) in edited cTnI-expressing hiPSC-CMs. The Dox-treated cTnI edited hiPSC-CMs showed faster relaxation rates upon Iso stimulation with significantly shortened RHW and significantly faster decay rates (RHW: 158.3 +/− 4.722, n = 17, P < 0.0001, n = 17, P < 0.01, Decay: 213.3 +/− 8.439, n = 17, P < 0.01, as compared to post cTnI induction, no Iso) (Figure 4a–c). In contrast, there were no significant differences in relaxation kinetics in unedited hiPSC-CMs by Iso (Figure 4d, e). Data further showed the Iso-mediated increase in spontaneous beating frequency was the same in edited and unedited groups (Figure 5b–c). Because the effect of Iso to increase the beating rate was similar in edited and non-edited hiPSC-CMs we take this as physiologically relevant evidence that the adrenergic signaling cascade is intact and similar between groups.

We next examined intracellular Ca²⁺ transients in hiPSC-CMs to test whether altered Ca²⁺ kinetics could account for the faster relaxation in the cTnI-expressing hiPSC-CMs (Supplemental Figure 2). In both edited and in unedited hiPSC-CMs following Dox there were no detected alterations in the Ca²⁺ transient decay kinetics (Supplemental Figure 2a, b). Interestingly, in edited hiPSC-CMs following Dox, Iso had an effect to significantly accelerate the Ca²⁺ transient decay rate (Pre Iso = 0.47 +/− 0.002, post Iso = 0.38 +/− 0.004, n = 12, P < 0.0001); however, this effect was not evident in the unedited hiPSC-CMs (Supplemental Figure 2).

3.3 |. Assessment of Dox effects on hiPSC-CMs function

Mindful of previous studies showing developmental time in culture having effects on hiPSC-CM structure-function, along with studies detailing the potential off-target or toxic effects of chronic application of drugs in gene induction/excision studies [25], we examined both time and Dox effects using unedited hiPSC-CMs. Analysis of the two-weeks Dox exposure protocol on unedited hiPSC-CMs function showed evidence of Dox altering the rise time of contraction and this was accompanied by a significantly faster rise in the Ca²⁺ transient in unedited Dox treated hiPSC-CMs) (Supplemental Figures 1,2). We do not have an explanation for this result. One possibility that could be addressed in future works would be an effect of Dox to alter sarcomere proteins, including effects due cTnI induction (or lack thereof, in unedited cells), or by altered post-translation modifications of sarcomere proteins as possible explanations. Effects on the rate of rise of contraction appeared to be Dox dependent because unedited hiPSC-CMs studied over the two-week period without Dox had no effect on contractile rise.

3.4 |. Gene expression in cTnI isoform induced hiPSC-CMs

We next tested whether inducing expression of the adult cTnI isoform would, in turn, alter the global expression patterns of genes involved in cardiac muscle maturation and function. We hypothesized that by inducing expression of the adult maturation marker cTnI that this would in turn orchestrate, either directly or indirectly, the induction of other adult maturation genes in hiPSC-CMs. We performed RNAseq on hiPSC-CMs that expressed cTnI and compared global gene expression patterns to isogenic hiPSC-CMs that did not express cTnI (Supplemental Figure 3; edited treated with Dox vs unedited treated with Dox). RNAseq analysis revealed a number of gene families with altered expression profiles, including genes clustered in the muscle contraction, cardiac transcription and cell signaling families (Supplemental Figure 3). As derived from the global RNAseq data, we next examined specific genes highly relevant to cardiac development and physiological function that were upregulated or downregulated. Supplemental Figure 3b shows individual genes with a fold change of 2 or more and P value <0.05. Supplemental Figure 3b inset shows upregulation of cTnI in edited cells exposed to doxycycline, as expected, shown as ratio of cTnI to ssTnI mRNA (P < 0.05). Several other genes of potential interest were found to be dysregulated in this RNAseq analysis, including GREM2, CASQ1, and TMEM10 and others. However, in follow up studies, we attempted to verify these findings in parallel independent analysis using qRT-PCR on individual genes (Supplemental Figures 4–7). The qRT-PCR results largely failed to corroborate the RNAseq upregulated/downregulate gene results in the cTnI expressing hiPSC-CMs (Supplemental Figures 4–7). The major exception was that both the RNAseq and qRT-PCR demonstrated a significant increase in cTnI mRNA (and decrease in GREM2) in the edited + Dox hiPSC-CMs. We also tested by qRT-PCR several specific contractile and Ca²⁺ handling genes shown previously to have a significant role on cardiac function with specific focus to the enhanced relaxation kinetics observed in hiPSC-CMs expressing cTnI. Here, no significant change was observed for Ca2+ handling genes PLN, Serca2a, RyR or CALM4 (Supplemental Figure 4), in myofilament genes MLC2v/a, TTN, MYOM3 (Supplemental Figure 5), or MYH6/7 (Supplemental Figure 6). GREM2, a gene that encodes a member of the BMP antagonist family was significantly down regulated by qRT-PCR in cTnI-induced hiPSC-CMs; whereas GATA4, NKX2.5 and HAND2 cardiac differential markers were not changed (Supplemental Figure 7).

4 |. DISCUSSION

Physiological maturation of stem cell-derived cardiac muscle represents a major obstacle toward advancing progress in cardiac regenerative biology and medicine [5–7,30,31]. We sought here to advance physiologically relevant maturation in human iPSC-derived cardiac muscle by targeting the cardiac sarcomere. The sarcomere’s role as the functional unit of heart muscle underscores its significance in advancing physiological performance of hiPSC-CM [8]. Implementing temporal control of a gene edited expression cassette, we induced expression of the adult cardiac troponin I isoform, which is the essential regulatory element of the sarcomere critical for physiological maturation of human cardiac muscle [8,32–34]. Using this experimental approach, the main new findings of this study are: 1) significant induction of the adult cTnI isoform protein, as evident by the increase in the cTnI (adult) / ssTnI (fetal) protein isoform ratio in hiPSC-CMs at the mRNA and protein levels with the induced cTnI correctly localized in the sarcomere; and 2) physiologically relevant acceleration in hiPSC-CM mechanical relaxation by cTnI isoform induction, a result independent of alterations in the intracellular Ca²⁺ transient decay rate. We attribute the enhanced physiological relaxation directly to cTnI induction as study of global gene expression by RNAseq, together with specific gene analysis via qRT-PCR, showed no other consistent changes in gene expression relevant to relaxation, other than cTnI which was induced in the hiPSC-CMs. It is physiologically noteworthy that cTnI induction (edited + Dox) had faster contractile relaxation without any change in the Ca²⁺ transient decay rate. We attribute this to the induction in the cTnI isoform and its incorporation into the sarcomere. Mechanistically, we have demonstrated cTnI directly affects thin filament properties in a manner to facilitate faster relaxation than ssTnI, which is a well-preserved cardiac muscle feature noted during mammalian evolution [8,28,29].

These new findings are significant as TnI isoform-based enhanced cardiac muscle relaxation is an essential physiological signature of mature human myocardium. This point is dramatically demonstrated in infants, in which the normal TnI isoform transition is disabled, resulting in severe cardiomyopathy [34]. In the human heart, TnI isoform switching is stoichiometrically conserved and irreversible, such that the fetal heart exclusively expresses the slow skeletal troponin I isoform, ssTnI, then, shortly after birth, the ssTnI gene is silenced and, in precise synchronization, the mature adult cardiac TnI gene becomes activated to replace the fetal ssTnI isoform [35,36]. Recently, mutations have been identified that prevent cTnI isoform induction in infants, requiring LVAD implantation and then heart transplant before the age of two [34]. Thus, cTnI induction is a required physiological transition in the human sarcomere. Interestingly, we previously showed that hiPSC-CMs (as derived from healthy individuals) are stalled in an immature ssTnI isoform state [6,7]. Accordingly, the present study demonstrating gene editing hiPSC-CM physiological maturation is significant in terms of advancing basic and clinical applications of stem cell-derived cardiac muscle.

Our findings show that inducing expression of cTnI in hiPSC-CMs is sufficient to produce a significant acceleration in relaxation. This is important because faster cardiac relaxation is a well noted physiologically relevant signature of adult cardiac muscle [8]. Faster myocardial relaxation is critical to overall adult heart function, especially during stress. The molecular mechanisms underlying cardiac muscle relaxation performance are complex and have been derived largely from studies on rodent myocardium [8,10,37]. The interplay between intracellular Ca²⁺ handling kinetics and the sarcomere have been well documented in regulating the rate of cardiac muscle relaxation [10]. The present study is the first to demonstrate in hiPSC-CMs that inducing cTnI is sufficient to significantly hasten cardiac muscle relaxation. We attribute the faster relaxation exclusively to cTnI induction, as other factors, including induction of the fast α myosin heavy gene or faster intracellular Ca²⁺ handing which are known to facilitate faster relaxation [38,39], were not altered in this study.

Another main finding of this study is the acceleration in relaxation in cTnI expressing hiPSC-CMs by β-adrenergic stimulation. In comparison, Iso-mediated faster relaxation, which is characteristic of adult human myocardium, was not observed in unedited ssTnI expressing hiPSC-CMs. The effect of β-adrenergic stimulation to accelerate relaxation performance is well documented in adult myocardium, from rodent to human hearts [8]. β-adrenergic stimulation of the heart causes a dramatic increase in heart pump performance [40]. This represents a vital mechanism by which the heart can modify performance during the fight or flight response. During β-adrenergic stimulation, heart rate (chronotropy), pressure development (inotropy) and muscle relaxation rate (lusitropy) are all markedly increased [40]. We previously reported that a PKA-based Ser 23/24 cTnI phosphorylation mimetic significantly accelerated relaxation [10]. Further, in otherwise normal adult myocytes, expression of ssTnI significantly slowed relaxation function and also blunted the effect of Iso to accelerate relaxation [10]. Compared with rodent myocytes, human myocardium has:1) reduced reliance on SR pumps to remove Ca²⁺, 2) slower Ca²⁺ uptake kinetics, 3) larger cardiac reserve capacity, indicating a broader dynamic range in response to β-adrenergic stimulation and thus a correspondingly greater relative role of cTnI-P, and 4) more slowly cycling myosin cross-bridges, β vs α myosin heavy chain isoform. We hypothesized that the presence of the β-MHC isoform in adult human heart [38], as evident in hiPSC-CMs [7], would reveal a greater role of TnI isoforms in determining relaxation kinetics. This is suggested because β-MHC cross-bridges detach more slowly, and through reciprocal interactions with the thin filament regulatory system, would be expected to cause prolonged activation of the thin filament [41]. Overall, our findings in hiPSC-CMs gain support from these studies. This finding is physiologically relevant, as it is well accepted that the adrenergic system changes dramatically throughout the development of a cardiac myocyte, and that fetal cardiac myocytes are less responsive to β-adrenergic stimulation [42]. In myocytes from failing hearts, when immature genes are reactivated and dedifferentiation ensues, β-adrenergic signaling is also downregulated [43]. Several groups working with hiPSC-CMs have noted altered responsiveness to adrenergic agonists [44]. Thus, increased β-adrenergic responsivity by cTnI induction represents a physiologically relevant advance in hiPSC-CMs.

This study also sought to test whether inducing one adult maturation marker [cTnI] would be sufficient to induce the gene expression other components of maturation in hiPSC-CMs. We examined multiple targets relevant to myocyte maturation and found no significant alterations in gene expression, including in myosin heavy chain isoforms or in the expression of Ca²⁺ handing genes SERCA2a/phospholamban. These results suggest that cTnI induction, as shown here, is not sufficient via a “domino effect” to advance more global physiological maturation of hiPSC-CMs. Despite RNAseq analysis finding a large number of genes with altered expression by cTnI induction, confirmation of relevant genes to myocyte structure-function by qRT-PCR were largely not forthcoming, with the notable exception being significant cTnI mRNA induction documented by both methods, as expected. These data also further support that the faster relaxation performance by cTnI induction is specific to this TnI isoform transition. One caveat to consider here relates to the extent of cTnI isoform induction in hiPSC-CMs which, while significant in terms of protein expression and physiological result, did not reach the full adult transition to 100% cTnI isoform content (with no detectable ssTnI), a hallmark of the mature adult myocardium [45]. However, as indicated, the degree of stoichiometric conversion ssTnI to cTnI, while statistically and physiologically significant, still was far below that present in human adult myocardium (100% cTnI). The extent of TnI conversion is estimated in our study to reflect going from an embryonic/fetal signature to that approaching a neonatal signature. This is based on past studies from our group and others that have clearly defined the developmental transitions in TnI isoform composition in the mammalian heart [6,7,46]. We speculate that with further advances and refinement of the gene induction system in potential future works, a more complete transition to the adult human myocardium TnI isoform profile could be achieved in future works.

One potential limitation of inducible gene expression systems is that they typically require chronic drug exposure, raising the potential of off-target effects [25]. The gene induction platform used here required chronic exposure to Dox. To address potential Dox effects, unedited isogenic controls were examined under the same Dox protocol and data showed evidence of chronic Dox altering some aspects of hiPSC-CMs function. This suggests a confounding effect of chronic doxycycline on hiPSC-CMs function. However, as no effects of Dox on relaxation performance were evident, this further supports a direct role of cTnI to hasten hiPSC-CM relaxation. Collectively, these findings necessitate drug dependent control analysis to aid in dissecting gene induction direct effects from off-target effects.

In closing, this study provides new insights towards advancing physiological maturation of hiPSC-CMs. Results show inducible hiPSC-CM maturation, via adult troponin I isoform switching, is sufficient to advance a key physiologically relevant maturation step from the fetal state to the adult state in hiPSC-CMs. Specifically, maturation of the TnI switch enables faster relaxation function in hiPSC-CMs which is an essential component of the physiological adaptations required for beat-to-beat control of the human myocardium. Overall, these findings have relevance to a range of studies employing stem cell-derived cardiac muscle, including in vitro investigations for drug-discovery, as well as for potential clinical applications of physiologically optimized hiPSC-CM in cardiac regeneration/repair.

Significance Statement.

Induced pluripotent stem cell-derived cardiac myocytes (iPSC-CMs) have enormous potential for use as a model of human cardiac function and disease but are limited by physiologic immaturity. Using gene editing, we introduced an adult form of a key contractile protein, cardiac troponin I by drug selectable induction. Expression of cardiac troponin I enabled iPSC-CMs to function more similar to mature cardiac myocytes in terms of relaxation and in the ability to respond to adrenergic stimulation to further accelerate physiological relevant faster relaxation performance. This novel iPSC-CM line is relevant for drug discovery, disease modeling, and for studies understanding basic human cardiac physiology.