Long-term Contractile Activity and Thyroid Hormone Supplementation Produce Engineered Rat Myocardium with Adult-like Structure and Function

Abstract

The field of cardiac tissue engineering has developed rapidly, but structural and functional immaturity of engineered heart tissues hinder their widespread use. Here, we show that a combination of lowrate (0.2 Hz) contractile activity and thyroid hormone (T3) supplementation significantly promote structural and functional maturation of engineered rat cardiac tissues (“cardiobundles”). The progressive maturation of cardiobundles during first 2 weeks of culture resulted in cell cycle exit and loss of spontaneous activity, which in longer culture yielded decreased contractile function. Maintaining a low level of contractile activity by 0.2 Hz pacing between culture weeks 3 and 5, combined with T3 treatment, yielded significant growth of cardiobundle and myocyte cross-sectional areas (by 68% and 32%, respectively), increased nuclei numbers (by 22%), improved twitch force (by 39%), shortened action potential duration (by 32%), polarized N-cadherin distribution, and switch from immature (slow skeletal) to mature (fast) cardiac troponin I isoform expression. Along with advanced functional output (conduction velocity 53.7 ± 0.8 cm/s, specific force 70.1 ± 5.8 mN/mm²), quantitative ultrastructural analyses revealed similar metrics and abundance of sarcomeres, T-tubules, M-bands, and intercalated disks compared to native age-matched (5-week) and adult (3-month) ventricular myocytes. Unlike 0.2 Hz regime, chronic 1 Hz pacing resulted in significant cardiomyocyte loss and formation of necrotic core despite the use of dynamic culture. Overall, our results demonstrate remarkable ultrastructural and functional maturation of neonatal rat cardiomyocytes in 3D culture and reveal importance of combined biophysical and hormonal inputs for in vitro engineering of adult-like myocardium.

1. Introduction

Engineered heart tissues hold great promise for regenerative therapies and for in vitro studies of cardiac development, disease, and drug response [1-3]. Over the past decades, cardiac development in vivo has been extensively studied in mouse and rat hearts under the premise that developmental principles in rodents apply to humans, albeit at an extended time scale [4]. Engineered heart tissues made from neonatal rat ventricular myocytes (NRVMs) have been a valuable platform for developing methods and providing mechanistic insights for improved cardiomyocyte maturation in vitro, eventually resulting in engineering of functional heart tissues made from human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs) [5-14]. Compared to hPSC-CMs, NRVMs show advanced maturation state which makes them suitable for in vitro studies of postnatal cardiac development, a process much less understood than the development of embryonic heart. Still, despite extensive efforts by numerous groups, the progression from neonatal to adult cardiac phenotype in rodents has not been recapitulated in vitro since engineered tissues constructed from NRVMs exhibit structural and functional properties far less mature than those of adult rat myocardium.

Recently, we developed a method for engineering cylindrical NRVM tissues named “cardiobundles” that after only 2 weeks of culture attained functional properties (e.g. conduction velocity, CV, and contractile stress) approaching those of adult rat myocardium [9]. Still, NRVMs in cardiobundles exhibited important differences from adult cardiomyocytes including longer action potential duration (APD), lower CV, smaller cell size, random vs. polarized membrane distribution of Connexin-43 and N-cadherin junctions, and less mature ultrastructure. Regarding that rat cardiomyocytes in vivo require several weeks post-birth to achieve adult phenotype prompted us to explore if a longer-term culture under appropriate biophysical and biochemical conditions could further advance NRVM maturation in cardiobundles. Of interest, in vivo electrical maturation of late embryonic and early neonatal ventricular myocytes involves hyperpolarization of cell resting membrane potential, leading to loss of spontaneous activity and attainment of a quiescent but highly excitable phenotype [15, 16]. Similarly, our cardiobundles attain both electrical quiescence and high CVs by 2 weeks of culture [9]; however, for a longer-term maturation, the lack of active mechanical loading by tissue contractions in quiescent cells would yield loss of myosin heavy chain and myofibrillar disorganization [17, 18]. Of additional interest, a critical regulator of postnatal rat cardiac tissue maturation in vivo, thyroid hormone (T3, Tri-iodothyronine) [19], has been shown to facilitate several protein isoform switches in cardiomyocytes, including β- to α-myosin heavy chain (MHC) [20], slow skeletal to cardiac TnI (ssTnI to cTnI) [21], and slow to fast transient outward current (I_to) component (K_v1.4 to K_v4.2/4.3) [22-24]. Previous studies have indicated that T3 can improve contractile force of spontaneously active engineered NRVM tissues [20] and accelerate kinetics of calcium transients in hPSC-CM monolayers [25].

Based on these considerations, we assessed if long-term active mechanical loading of cardiobundles induced by chronic electrical stimulation between culture weeks 3 and 5 can additionally advance NRVM maturation beyond an already high level achieved after 2-week culture. We further explored the effects of T3 supplementation on NRVM structure and function in both quiescent and mechanically active cardiobundles. Overall, results of our studies reveal that low-rate sustained contractile activity and T3 treatment synergistically promote ultrastructural and functional maturation of cardiomyocytes to levels characteristic for adolescent and adult rat myocardium.

2. Materials and Methods

2.1. Cell isolation and cardiobundle construction

All animal procedures were performed in compliance with the Institutional Animal Care and Use Committee at Duke University and the NIH Guide for the Care and Use of Laboratory Animals. NRVMs were isolated from 2-day-old Sprague Dawley rats as previously described [9]. Briefly, ventricles were minced and washed in ice-cold buffer (HBSS with 1.2 mM MgCl₂), followed by trypsin incubation (1 mg/mL) for 16-18 hours at4°C with gentle agitation. Then, minced tissues were dissociated by 4 serial digestions with 250 U/mL Collagenase Type 2 (Worthington, 3-4 minutes each, 37 °C, 90 rpm). Cells collected after each digestion were centrifuged, resuspended, passed through a 100 μm cell strainer, centrifuged again, and resuspended in pre-plate media consisting of DMEM / F12, 10% fetal bovine serum, 10% horse serum, penicillin (5 U/ml) and vitamin B12 (2 μg/ml). The cell suspension was preplated on a non-coated tissue culture flask for 45 minutes to enrich the fraction of cardiomyocytes in the cell suspension. Floating cells were collected for cardiobundle construction. Typical yield from 1 litter of 10 neonates was 50-55 × 10⁶ cells.

For each cardiobundle, 3.75 × 10⁵ NRVMs were mixed in a fibrin-based hydrogel (2 mg/ml fibrinogen, 1 U/ml thrombin, 10% v/v/ Matrigel®) and quickly pipetted into a PDMS tissue mold with two 2 mm × 7 mm troughs and a porous nylon frame as previously described [9, 26]. The mixture was incubated for 45 minutes at 37 °C to allow polymerization and attachment to the frame, then submerged in 3D cardiac medium (DMEM, 10% horse serum, 1% chick embryo extract, aminocaproic acid (1 mg/ml), ascorbic acid 2-phosphate sesquimagnesium salt hydrate (50 μg/ml), penicillin (5 U/ml) and vitamin B12 (2 μg/ml)) and subjected to dynamic culture. The next day, cardiobundles with frames were carefully removed from the molds and cultured in free-floating conditions. Full media changes of 1.5 mL per well were completed every other day.

2.2. Design and fabrication of electrical stimulation chambers

Custom electrical stimulation chambers (Fig. S1) were produced by molding polydimethylsiloxane (PDMS) solution in polycarbonate molds drawn in SolidWorks and machined by the Duke University Physics Shop. The mold consisted of: (1) a “base” with a protrusion designed to create a tissue culture trough with 56 mm length, 20 mm width, and 15 mm depth, (2) two removable “walls” to contain the poured PDMS solution, and (3) two “spacers” to create void spaces where carbon electrodes are subsequently inserted and attached to electrical wiring. The mold was assembled by attaching the walls to the base with 8 screws, then inserting the spacers. Degassed PDMS solution was poured into the mold, cured at 80 °C for at least 4 hours, and removed from the mold to generate the tissue culture trough. Two carbon rods (70 mm length, 3 mm width, 0.5 mm thickness, Goodwinds LLC) were pre-washed in deionized water by orbital rotation at 90 rpm for 24 hours, inserted through the trough, and connected to 22-gauge stranded electrical wire by silver conductive epoxy (MG Chemicals). To secure the connection of the electrode and the wire, the trough was then inserted into second polycarbonate assembly with a flat base, and PDMS was cured around electrode-wire junctions. The free ends of the two wires were soldered to 2 mm banana connectors for easy attachment to an electrical stimulator. Before culture, stimulation chambers were sterilized by 70% ethanol and UV exposure overnight.

2.3. Hardware and software to apply custom electrical stimulation patterns

A schematic of the electrical stimulation system is depicted in Fig. S2. Electrical stimuli were delivered to the culture chamber by a 4-channel, current-controlled stimulator (Digitimer D330 MultiStim) that has a maximum output of 500 mA per channel. The stimulator was remotely controlled by a high-speed TTL pulse generator (National Instruments NI-9401, 100 ns update rate), which was connected to the stimulator by customized wiring between a 25-pin and a 15-pin D-sub connector. The timing and duration of stimulus pulses for each of the 4 independent channels were user-specified in a custom LabVIEW program.

2.4. Electrical stimulation experiments

The threshold for inducing action potentials in cardiobundles was determined by setting the pulse width to 2 ms and adjusting the stimulus amplitude (in mA) while visually inspecting cardiobundles for macroscopic contractions. Initial experiments activated cardiobundles at a rate of 1 Hz from day 7 to day 21 of culture, and the days of stimulation were adjusted as described in the results section. Where indicated, cell culture media was supplemented with 5 nM triiodothyronine (T3) during stimulation. Carbon rod electrodes were replaced between each experiment.

2.5. Isometric force testing

Custom force measurement apparatus was used to measure contractile force during field stimulation as previously described [5, 9]. Briefly, cardiobundles immersed in 37 °C Tyrode’s solution with 1.8 mM CaCl₂ were fixed at one end, with the other end connected to a force transducer. Cardiobundles were paced at a basal rate of 2 Hz with 10 s of pre-pacing, followed with 2 s of data collection. Data were recorded and analyzed for active twitch force using a custom MATLAB program. Total twitch duration was measured by defining the beginning and end of a contraction at 10% of peak amplitude; rise time and decay time were measured from 10% - 90% amplitude and from 90% - 10% amplitude, respectively. To evaluate changes in active and passive tension in response to muscle elongation [27], cardiobundles were stretched in increments of 4% elongation, with recordings taken at each length until maximum active force was achieved (typically 4% or 8% elongation). For studies of passive tension, cardiobundles were stretched to a maximum of 12% elongation.

2.6. Optical mapping

Action potential propagation was assessed by optical mapping according to previously described methods [9, 28, 29]. Briefly, live cardiobundles were stained with the potentiometric membrane dye di-4 ANEPPS (10 μM, 5 minutes, room temperature), then immersed in 37 °C Tyrode’s solution with 10 μM blebbistatin to inhibit contractions. Action potentials were initiated at one end of the bundle by a platinum point stimulus electrode using a near-threshold stimulus. After 10 s of prepacing, signals were recorded 2 mm away from the pacing site using a photodiode array at a sampling rate of 2.4 kHz. Activation times of individual channels were calculated as the time of maximum voltage upstroke, and average conduction velocity (CV) was calculated by a linear fit of distance vs. activation time. Action potential duration (APD) was calculated as the difference between activation time and 80% repolarization. To obtain restitution curves, pacing rate was increased from a basal rate of 2 Hz in increments of 1 Hz, with maximum capture rate (MCR) recorded as the highest rate at which each stimulus elicited an action potential for a duration of 2 s (after 10 s of pre-pacing).

2.7. Histology

Intact cardiobundles or 10 μm thick cardiobundles cross-sections were fixed in 2% paraformaldehyde (PFA) and immunostained as previously described [9, 30]. Briefly, intact or sectioned tissue samples were permeabilized with 0.5% Triton-X for 30 min, then blocked with a blocking solution (1% bovine serum albumin and 10% chicken serum in PBS) at 4 °C overnight. Primary antibodies were diluted in blocking solution and incubated with samples overnight at 4 °C, followed by secondary antibody incubation at room temperature for 2 hours. The primary antibodies used included Sarcomeric α-actinin (SAA, Sigma A7811, 1:200), Connexin-43 (Abcam Ab11370, 1:200), N-cadherin (Abcam Ab12221, 1:400), Vimentin (Abcam Ab92547, 1:500), Caveolin 3 (Abcam Ab2912, 1:200), Ki67 (Santa Cruz sc-7844, 1:300), Nkx2.5 (Santa Cruz sc-8697 1:100), and RyR2 (Thermo MA3916, 1:150). Samples were imaged using Zeiss 510 inverted confocal microscope and image analysis was performed using image J software.

To stain cell membranes for visualization of T-tubules, live cardiobundles were washed in Tyrode’s solution, then incubated for 5 min in 10 μM di-4 ANEPPS dye at room temperature, and subsequently rinsed and incubated for additional 5 min in Tyrode’s solution. Di-4 stained cardiobundles were imaged on a Zeiss 510 inverted confocal using a 40× NA/1.3 oil objective.

2.8. EdU incorporation as a marker of DNA synthesis

To cumulatively label all cell nuclei that underwent DNA synthesis during a given period, culture media was supplemented with 10 μM of the thymidine analog 5-ethynyl-2’-deoxyuridine (EdU) at each media change (every 2 days). For a single pulse (specific time point) EdU labeling, cardiobundles were incubated with 10 μM EdU for 24 h, after which tissues were fixed and processed for immunostaining. Nuclei with incorporated EdU were detected according to manufacturer instructions in the Click-iT Plus Alexa Fluor 555 Picolyl Azide Toolkit (Thermo), using a 1:1 mix of copper solution and copper protectant to avoid degradation of other fluorophores. EdU detection was performed after incubation with primary and secondary antibodies to other targets. To quantify number of cardiomyocytes that underwent DNA synthesis, we counted EdU-labeled nuclei in F-actin⁺/Vimentin⁻ or SAA⁺ cells. Similarly, non-cardiomyocytes that underwent DNA synthesis were identified by EdU labeling of nuclei in Vimentin⁺/ F-actin⁻ cells.

2.9. Western blot

Protein was extracted from cardiobundles, separated by SDS-PAGE, and transferred onto nitrocellulose membranes as previously descried [8, 9]. To visualize the relative amounts of cardiac troponin I (cTnI) and slow skeletal troponin I (ssTnI), the membrane was stained using a pan-TnI antibody (Millipore MAB1691) that binds to both isoforms. cTnI and ssTnI were distinguished by the higher molecular weight (and slower migration) of cTnI [21].

2.10. Transmission Electron Microscopy (TEM)

Cardiobundles were washed three times with Cacodylate buffer, and then fixed overnight with 2% buffered gluteraldehyde at 4 °C. After 1% OsO₄ post-fixation, samples were sequentially dehydrated in a series of ascending acetone concentrations (70%, 95%, 100%), then in the 50/50 acetone:epoxy overnight, after which samples were embedded in 100% epoxy and baked at 60 °C overnight. Ultrathin sections were placed on copper grids and stained with uranyl acetate. Images were acquired using Philips CM12 transmission electron microscope operated at 120 kV, with XR60 camera system (Advanced Microscopy Techniques). Native heart tissue samples for TEM imaging and analysis were dissected from left ventricular free wall.

2.11. Statistics

All data were presented as mean ± SEM, unless otherwise noted, and analyzed for statistical significance using one-way ANOVA with Tukey’s post-hoc test. For the Z-band width and sarcomere length analyses, data did not pass D’Agostino-Pearson normality test and were expressed as median ± interquartile range and assessed for statistical significance using one-way ANOVA with Kruskal-Wallis test. P < 0.05 was considered statistically significant in all analyses.

3. Results

3.1. Loss of cell cycle activity in prolonged cardiobundle culture

We previously developed dynamic, free-floating 3D culture conditions that allowed us to engineer 2-week old NRVM cardiobundles with highest reported conduction velocity (CV) and specific force of contraction [9]. Here we assessed if this advanced functional maturation of NRVMs in cardiobundles was associated with progressive loss of cell cycle activity as characteristic of native postnatal heart development [31, 32]. Specifically, we applied 24-hour EdU pulses on different days of cardiobundle culture (d4 – d26, Fig. 1A, B) and quantified fraction of cardiomyocytes (CMs) and non-myocytes (non-CMs) that underwent DNA synthesis (Fig. 1C). The percent of cardiomyocytes undergoing S-phase within 24 hours declined from 10.6 ± 0.6% on culture day 4 to 0.2 ± 0.2% on culture day 26, with the most significant decline occurring between day 6 and day 14 (9.5 ± 2.5% vs. 0.8 ± 0.4%, Fig. 1C). Non-myocyte cell cycle activity also decreased with time of culture but remained ~10-fold higher than that of CMs.

Figure 1: Cell cycle exit in cardiobundles with time of culture.

(A) Representative cross-sections of cardiobundles subjected to 24-hour EdU pulse on indicated days of culture (d4-d26), fixed at the end of the EdU pulse, and stained for EdU, F-actin (F-act), and nuclei (DAPI). (B) Representative longitudinal sections of cardiobundles subjected to 24-hour EdU pulse on day 4 (d4) or 18 (d18), fixed at the end of the EdU pulse, and stained for EdU, sarcomeric α-actinin (SAA), vimentin (Vim), and DAPI. (C) Quantification of the percent of nuclei labeled with EdU for samples described in (A). n = 16 cardiobundles from 3 cell isolations per time point.

3.2. Effects of 1 Hz electrical stimulation on cardiobundle function at 3 weeks.

During the 2-week maturation process in vitro, cardiobundles also exhibit a progressive loss of spontaneous activity [9], which leads to electrical quiescence and decrease in functional output between 2 and 3 weeks of culture (Fig. 2A, B). This functional decline indicates that active contractions and mechanical loading of cardiomyocytes (CMs), which are hallmarks of their physiological environment in vivo, are likely required for their continued in vitro maturation. Therefore, to counteract the observed functional decline and allow long-term studies of NRVM maturation, we applied chronic electrical pacing at a rate of 1 Hz (typically used in literature [33-35]) starting on day 7 of cardiobundle culture, when the spontaneous beating rate typically begins to decline below 1 Hz [9]. During the entire pacing period, we ensured by daily visual inspections that the stimulus pulse duration of 2 ms and amplitude of 70 mA successfully induced continuous contractile activity in cardiobundles.

Figure 2: Continued contractile activity rescues quiescence-induced loss of cardiobundle function at 3 weeks of culture.

(A-B) Twitch force (A) and conduction velocity (CV, B) measured during 2 Hz pacing in cardiobundles cultured for 7, 14, or 21 days. Cardiobundles lose spontaneous activity and become quiescent within the first 2 weeks of culture¹. n = 12-20 cardiobundles per group from 4 cell isolations. * p < 0.05 vs. 7 days; ^# p < 0.05 vs. 14 days. (C) Experimental protocol for a 2-week, 1 Hz chronic pacing of cardiobundles. (D-E) Twitch force (D) and passive tension (E) as a function of elongation. (F-G) Conduction velocity (CV, F) and action potential duration (APD, G) restitution curves as a function of pacing cycle length (CL). For D-G, n = 12 cardiobundles (3-week old) per group from 3 cell isolations. (H) Representative cardiobundle cross-sections stained for F-actin (F-act), vimentin (Vim) and nuclei (DAPI). (I) Representative higher magnification cross-sections stained for wheat germ agglutinin (WGA), F-actin (F-act), and nuclei (DAPI). (J-K) Quantified F-actin⁺ area per nucleus (J) and ratio of F-actin⁺ area per WGA⁺ area (K) in cardiobundles described in H-I. n = 4 cardiobundles per group from 2 cell isolations. * p < 0.05 vs. quiescent. (L) Representative longitudinal sections stained for sarcomeric α-actinin (SAA) and N-cadherin (N-cad). (M) Fraction of N-cadherin plaques oriented perpendicular to longitudinal cardiobundle axis. n = 6 cardiobundles per group from 3 cell isolations. * p < 0.05 vs. quiescent.

After 3-week culture (2 weeks of pacing) quiescent and mechanically active cardiobundles were compared for their structural and functional properties (Fig. 2C). The mechanically active cardiobundles exhibited higher twitch force (Fig. 2D), similar passive tension (Fig. 2E), faster CV (Fig. 2F), and prolonged APD (Fig. 2G) compared to age-matched quiescent cardiobundles. Interestingly, pacing rate of 1 Hz resulted in the formation of cardiomyocyte-free core in the tissue center (Fig. 2H), while closer to the cardiobundle surface cardiomyocytes had larger size (Fig. 2I, J) and less interstitial space (Fig. 2I, K). Furthermore, continued contractile activity yielded shift in membrane distribution of N-cadherin plaques from a spatially uniform to a polarized at end-end junctions of CMs as characteristic for adult rat ventricles (Fig. 2L, M, Fig. S3), while connexin-43 junctions remained uniformly distributed throughout the membrane (Fig. S4). Overall, these results demonstrated that compared to electrical quiescence, continuous contractile activity yielded enhanced function of 3-week cardiobundles, but the resulting functional parameters (twitch force of 2.05 ± 0.21 mN; CV of 47.75 ± 2.88 cm/s) were not improved relative to those at culture week 2 [9]. In addition, the prolonged APD and apparent necrotic core were adverse effects of 1 Hz pacing which we sought to address in the subsequent experiments.

3.3. Thyroid hormone supplementation shortens AP duration of cardiobundles

The longer AP duration of paced cardiobundles, which may result from reduced expression of K_v4.3 [36], contradicts normal postnatal maturation of the rat heart where APD progressively decreases into the adulthood [22, 37]. Thus, we explored known physiological cues leading to the accelerated AP kinetics in cardiac development. Specifically, the thyroid hormone T3 is known to increase expression of fast-repolarizing K⁺ currents in postnatal rat CMs [22-24], which prompted us to apply it from day 7 to day 21 of cardiobundle culture. We applied the dose of 5 nM, consistent with the level of T3 measured in postnatal rodent serum and previously used in vitro [38-41]. The T3 supplementation markedly shortened the APD of both active (1 Hz) and quiescent cardiobundles (Fig. 3A, Fig. S5A). The T3-induced shortening of APD was also manifested as a reduced duration of individual twitches (Fig. 3D, E, Fig. S5C) and increased maximum capture rate (Fig. 3C, Fig. S5D). On the other hand, T3 had no significant effect on CV and twitch force amplitude in either quiescent or active cardiobundles (Fig. 3B, F and Fig. S5B). These results demonstrated the ability of thyroid hormone to induce shortening of cardiac action potential consistent with postnatal maturation in vivo and to counteract the chronic pacing-induced APD prolongation.

Figure 3: Effects of T3 on function of quiescent and mechanically active 3-week old cardiobundles.

1 Hz pacing and 5 ng/ml T3 were continuously applied between 1 and 3 weeks of culture. (A-C) T3-induced changes in APD (A), CV (B), and maximum capture rate (MCR, C) in 3-week quiescent and mechanically active cardiobundles. (D) Representative cardiobundle force traces in response to 2 Hz pacing. (E-F) T3-induced changes in twitch duration and force amplitude in 3-week quiescent and mechanically active cardiobundles. Q: quiescent (non-paced) cardiobundles; Q+T3: quiescent cardiobundles treated with T3; A: active (1 Hz-paced) cardiobundles; A+T3: active cardiobundles treated with T3. n = 8-12 cardiobundles per group from 2-3 cell isolations. * p < 0.05 vs. Q; ^# p < 0.05 vs. A.

3.4. Effects of slower pacing rate on core necrosis in cardiobundles

As shown in Fig. 2H, cardiobundles paced at 1 Hz from culture day 7-21 developed a central necrotic core devoid of F-actin⁺ CMs. This effect could be attributed to increased oxygen and nutrient consumption of CMs during chronic pacing, along with hindered diffusion to the center of the cardiobundle through the reduced interstitial space at cardiobundle periphery (Fig. 2I-K). To address this issue and still maintain long-term contractile activity of NRVMs, we tested lower pacing rates of 0.5 Hz and 0.2 Hz where both T3 and pacing were initiated on day 14 and continuously applied thereafter till 5 weeks of culture (Fig. 4A). The start of pacing was moved from day 7 to day 14 when spontaneous contractions completely ceased yielding the subsequent functional decline (Fig. 2A, B) and enabling the continuous tissue capture at low pacing rates. The end-point of 5 weeks was chosen after observing that both cell cycling activity (Fig. 1) and resulting changes in cardiobundle size (Fig. S6) stabilized by culture week 5. From these experiments, we found that 0.5 Hz pacing still resulted in the cell loss, while 0.2 Hz pacing did not yield formation of necrotic core (Fig. S7A), instead, it increased the F-actin⁺ muscle area in cardiobundle cross-section (Fig. S7B). Moreover, rather than simply maintaining the force amplitude, 0.2 Hz pacing significantly increased the contractile strength of cardiobundles (Fig. S7C). Furthermore, we found no adverse effects of 0.2 Hz pacing on MCR, CV, or APD relative to 2-week non-paced or 5-week 1 Hz-paced cardiobundles (Fig. S7D-F). Based on these findings, the low pacing rate of 0.2 Hz was adopted for all subsequent experiments.

Figure 4: Long-term contractile activity and T3 induce maturation of 5-week cardiobundles.

(A) Experimental protocol for a 3-week chronic pacing and/or T3 supplementation of cardiobundles starting at day 14. (B) Representative optical action potential traces of 2-week control and 5-week cardiobundles. (C-D) APD (C) and CV (D) restitution curves as a function of pacing CL. (E-G) APD (E) and CV (F) in response to 2 Hz pacing (CL = 500 ms) and maximum capture rate (MCR, G) in response to point pacing. (H-I) Twitch duration (H) and twitch force amplitude (I) of cardiobundles measured during 2 Hz pacing. (J) Muscle-only specific force (twitch amplitude normalized by muscle-only, F-actin⁺ area). (K) Representative Western blot using pan-TnI antibody that detects both cardiac troponin I (cTnI, top bands) and slow skeletal troponin I (ssTnI, bottom bands) isoforms. GAPDH is shown as loading control. (L) Representative live membrane staining of cardiobundles in longitudinal view using di-4 ANEPPS. Red arrowheads in rightmost panel denote cross-striated membrane labelling indicative of T-tubules. 2 wk, Ctrl: non-paced cardiobundles cultured for 2 weeks; 5 wk, Q: quiescent (non-paced) cardiobundles cultured for 5 weeks; 5 wk, Q+T3: quiescent cardiobundles treated with T3; 5 wk, A 0.2: active (0.2 Hz-paced) cardiobundles at 5 weeks of culture; 5 wk, A 0.2+T3: active cardiobundles treated with T3. n = 9-12 cardiobundles per group from 3 cell isolations for (C-G); n = 16-20 cardiobundles per group from 4 cell isolations for (H-J); * p < 0.05 vs. 5 wk Q, ^# p < 0.05 vs. 5 wk A 0.2+T3.

3.5. Combined effects of low-rate contractile activity and T3 on functional and structural maturation of cardiobundles

We further examined effects of T3 supplementation in conjunction with 0.2 Hz pacing delivered between 2 and 5 weeks of culture. Similar to what was observed at 3 weeks of culture, T3 significantly shortened the APD of cardiobundles (Fig. 4B, C, E) towards the adult rat CM values [7, 42], without changing CV that was maintained at ~50 cm/s (Fig. 4D, F), still lower than in adult rat ventricles [7, 43]. The T3 treatment also increased the average MCR from ~8.5 Hz to 9.86 ± 0.46 Hz in quiescent and 10.91 ± 0.52 Hz in actively contracting (0.2 Hz-paced) cardiobundles (Fig. 4G) as well as induced a shortening of twitch duration (Fig. 4H). Quiescent cardiobundles displayed a decline in contractile force from 2.43 ± 0.10 mN at 2 weeks to 1.14 ± 0.11 mN or 1.72 ± 0.13 mN at 5 weeks without or with T3, respectively (Fig. 4I). In contrast, mechanically active cardiobundles displayed increased contractile force to 3.19 ± 0.09 mN (non-T3 treated) and 3.38 ± 0.10 mN (T3-treated) (Fig. 4I) with corresponding specific forces showing similar trends and remaining >60 mN/mm² for active cardiobundles (Fig. 4J), i.e., comparable to those of adult rat ventricular myocardium (40-71 mN/mm²) [17, 44, 45]. At the molecular level, T3 induced a switch from the immature troponin I isoform (ssTnI) to the mature cTnI isoform, while pacing alone failed to induce mature cTnI expression (Fig. 4K). After 5 weeks of culture, mechanically active, T3-treated cardiobundles showed evidence for robust T-tubulation and formation of staircase-like cell-cell boundaries characteristic of adult myocardium, which was not observed in 2-week control or 5-week quiescent cardiobundles (Fig. 4L, Fig. S8).

To further explore if the functional and molecular improvements from continuous contractile activity and T3 supplementation coincided with ultrastructural maturation of NRVMs characteristic of postnatal cardiac development, we performed transmission electron microscopy (TEM) in both engineered cardiobundles and 2-day, 5-week, and 3-month old native rat ventricular tissues (Fig. 5, Fig. S9). In low-magnification images, mechanically active, T3-treated 5-week cardiobundles appeared to have the most mature sarcomeric organization and significant presence of T-tubules compared to 2-week or 5-week quiescent cardiobundles (Fig. 5A, Fig. S9). In high-magnification images (Fig. 5B-E), NRVMs in active, T3-treated cardiobundles ubiquitously exhibited highly registered sarcomeres, abundant mitochondria with well-defined cisternae, T-tubular structures abutting sarcoplasmic reticular membrane, and intercalated discs with adherens, gap, and desmosomal junctions. By quantifying the abundance of H-zones, M-lines, and T-tubules from a large number of TEM images, we found that continuous contractile activity and T3 supplementation in 5-week cardiobundles (but no other in vitro conditions) resulted in the spatial densities of H-zones (0.94 ± 0.02 per sarcomere), M-lines (0.91 ± 0.03 per sarcomere), and T-tubules (0.40 ± 0.02 per sarcomere) that matched those quantified in 5-week and 3-month rat ventricular tissues (Fig. 5F). We also assessed Z-band width (Fig. S9A), known to vary with development and muscle type [46, 47], and found that the longer, 5-week culture reduced the width and increased the uniformity of Z-bands (from 115.60 ± 76.64 nm in 2-week to 77.53 ± 35.73 nm in 5-week cardiobundles), while adding T3 and maintaining contractile activity by 0.2 Hz pacing further increased Z-band uniformity (81.10 ± 22.31 nm, Fig. 5G). Additionally, we quantified sarcomere length (Fig. 5H), a measure of the thick and thin filament overlap that regulates actin-myosin interaction and force generation [48]. Consistent with the differences in twitch force (Fig. 4I), NRVMs in the 5-week quiescent cardiobundles exhibited significantly shorter sarcomeres (1.58 ± 0.21 μm) compared to those in 2-week cardiobundles (1.90 ± 0.23 μm), while the 5-week active/ T3 treated cardiobundles maintained the sarcomere length while increasing the length uniformity (1.85 ± 0.13 μm), consistent with the highest twitch force measured in this group.

Figure 5: Ultrastructural maturation of cardiomyocytes in cardiobundles.

(A) Representative low-magnification TEM images of 2-week control and 5-week cardiobundles. Arrowheads indicate T-tubules. (B-E) Representative high-magnification TEM images of 5-week mechanically active (0.2 Hz-paced), T3-treated cardiobundles showing examples of (B) T-t: T-tubule; SR: sarcoplasmic reticulum; Mito: mitochondria, (C) S: sarcomere; I: I-band; Z: Z-line; A: A-band; M: M-line; H: H-zone, (D) Intercalated disc structure with AJ: adhesion junction; GJ: gap junction; DJ: desmosomal junction (arrowheads), and (E) mitochondria with well-developed cisternae. (F-H) Quantification of sarcomeric structures in cardiobundles and native ventricular tissues including NRV: 2-day old neonatal rat ventricle; 5 wk RV: 5-week old rat ventricle; and ARV: 3-month old rat ventricle. (F) Quantification of average number of H-zones, M-lines and T-tubules per sarcomere; n = 1125 (2 wk), 757 (Q), 1240 (A+T3), 911 (NRV), 823 (5 wk RV) and 1220 (ARV) sarcomeres studied. (G) Quantification of Z-band width. Scatter plots show all data points along with medians and interquartile ranges; n = 303 (NRV), 378 (2 wk), 365 (Q), 373 (A+T3), 314 (5 wk RV), and 307 (ARV) sarcomeres studied. (H) Quantification of sarcomere length. Scatter plots show all data points along with medians and interquartile ranges; n = 310 (NRV), 281 (2 wk), 259 (Q), 268 (A+T3) 364 (5 wk RV) and 300 (ARV) sarcomeres studied. 2 wk: control non-paced cardiobundles cultured for 2 weeks; Q: quiescent (non-paced) cardiobundles cultured for 5 weeks; A+T3: active (0.2 Hz-paced) cardiobundles treated with T3. * p < 0.05 vs. 2 wk; # p < 0.05 vs. A+T3.

We also analyzed the cross-sections of 5-week cardiobundles for the total myocardial area, nuclei counts, and cumulative EdU incorporation (starting at 2 weeks of culture) (Fig. 6A, B). Continuous contractile activity and T3 supplementation resulted in significant increase in cardiobundle myocardial mass (Fig. 6C) and nuclei number (Fig. 6D), as well as apparent increase in cardiomyocyte size (assessed as the myocardial area per nucleus, Fig. 6E). The increased number of nuclei (Fig. 6D), and expectedly, cumulative DNA synthesis (assessed via EdU incorporation, Fig. 6F) was observed in continuously active but not quiescent (Fig S10A, B) cardiobundles regardless of T3 supplementation, however, only the T3 induced significant increase in total myocardial mass and cardiomyocyte size (also visible in Fig. 6B, rightmost panel), consistent with previous reports [49]. We further assessed cell cycle activity of NRVMs by co-staining of sarcomeric α-actinin and a cell cycle marker, Ki67 [50], on culture day 25 (e.g., approximately midway between 2 and 5 weeks of culture). Ki67⁺ staining in quiescent cardiobundles was only apparent in peripherally located fibroblasts but not cardiomyocytes (Fig. S10C, left), while continuously active cardiobundles contained Ki67⁺ CMs that exhibited sarcomere disarray characteristic of dividing or binucleating cells (Fig. S10C, right) [51,52]. Taken together, our results showed a clear synergistic effect of T3 and continuous low-rate contractile activity on the in vitro maturation of engineered NRVM tissues.

Figure 6: Structural characterization of 5-week cardiobundles.

(A) Representative cross-sections of cardiobundles cultured for 5 weeks under different conditions, stained for F-actin (F-act), Vimentin (Vim), EdU, and nuclei (DAPI). 0.2 Hz pacing, T3 treatment, and EdU supplementation were all started on culture day 14. (B) Representative high-magnification images of cross-sections stained for F-act, wheat germ agglutinin (WGA), and DAPI. (C-F) F-actin⁺ cross sectional area (C), nuclei within F-actin+ area per section (D), F-actin+ cross sectional area per nucleus within F-actin+ area (E), and percent of nuclei within F-actin+ area labeled with EdU (F) in cardiobundles. Q: quiescent (non-paced) cardiobundles; Q+T3: quiescent cardiobundles treated with T3; A: active (0.2 Hz-paced) cardiobundles; A+T3: active cardiobundles treated with T3. n = 9-10 cardiobundles per group from 4 cell isolations; * p < 0.05 vs. Q; ^# p < 0.05 vs. A.

4. Discussion

In this study, we assessed combined effects of sustained contractile activity and thyroid hormone supplementation on structural and functional maturation of post-mitotic engineered rat cardiac tissues. Similar to native postnatal development, neonatal rat cardiomyocytes in 3D cardiobundles exited cell cycle by 2 weeks of culture, containing <1% of cycling cells despite the presence of 10% horse serum in culture media. Application of low-frequency electrical stimulation and T3 during the subsequent 3 weeks of culture synergistically improved various aspects of cardiobundle maturation, yielding: (1) increased cell and tissue volume, (2) higher amplitude of contractile force, (3) shorter action potential and twitch duration, (4) higher MCR, (5) switch to mature troponin isoforms, (6) robust formation of T-tubules, (7) more organized sarcomere ultrastructure, and (8) polarization of N-cadherin junctions. To the best of our knowledge, the resulting engineered cardiac tissues are the first to exhibit both ultrastructural and functional characteristics approaching or matching those of native adult myocardium.

Late fetal maturation of ventricular myocytes involves gradual hyperpolarization of resting membrane potential leading to loss of spontaneous activity and acquisition of an electrically quiescent phenotype [15, 16]. Still, postnatal ventricular myocytes undergo active mechanical loading every heart cycle as the action potential spreads from sinoatrial node throughout the ventricles and calcium enters the cells to engage into a complex process of excitation-contraction coupling [53]. Mechanical loading is known to regulate important aspects of cardiac tissue development including organization of cytoskeleton, extracellular matrix, and cell-cell junctions [54-56]. Our engineered tissues lost spontaneous activity by 2 weeks of culture which resulted in significant decrease in their functional output by culture week 3. This necessitated the use of chronic electrical pacing to induce periodic contractions and maintain active mechanical loading, which in turn led to continued structural and functional maturation. Notably, the frequency of applied pacing needed to be limited to 0.2 Hz as higher stimulation rates (e.g. 0.5 and 1 Hz) resulted in significant loss of cardiomyocyte viability in the tissue core (Fig. S7A), consistent with Hirt et al. [57] who reported a decline in contractile force for 5 vs. 0.5 Hz pacing. The fact that faster pacing yielded a necrotic core in highly dense and mature cardiobundles was not surprising regarding the lack of perfused capillaries which in native ventricles are critical for sustaining contractile activity at physiological heart rates. Recently, Ronaldson-Bauchson et al. showed that gradually increasing electrical stimulation rate with time of culture (2-6 Hz over 2 weeks) significantly advanced ultrastructural and excitation-contraction coupling maturation of early-stage but not later-stage human iPSC-derived CMs, without any reported loss of cell viability [12]. Still, in these engineered tissues functional maturation lagged behind structural maturation, with the resulting contractile stresses and conduction velocities being more than ten-fold and two-fold lower, respectively, than reported in the current study (Fig. 4F, J). Taken together, it is possible that application of rapid pacing to early embryonic CMs can facilitate certain maturation aspects while cells are still plastic, but it adversely affects cell viability and tissue function if applied at more advanced (postnatal) stages of maturation. Ideally, rapid and simultaneous structural and functional maturation of cardiomyocytes in vitro may require that engineered tissues made of early (plastic) cardiomyocytes undergo gradual increase in both mechanical preload (through passive stretch) and afterload (through increased resistance to active contraction), as characteristic of postnatal left ventricular growth [58, 59].

Notably, while we employed electrical stimulation to sustain active mechanical loading of cardiobundles, it is likely that the application of electric current itself did not directly contribute to the improved tissue maturation. In postnatal hearts in mammals, electrophysiological maturation of CMs is characterized by progressive APD shortening that results from upregulation of various potassium currents including inward rectifier (I_K1) and transient outward (I_to) current, as well as downregulation of Na⁺-Ca²⁺ exchanger (NCX) [22, 60-63]. Contrary to this process, Sathaye et al. showed that chronic electrical stimulation of NRVM monolayers opposed the natural decrease in APD by reducing the expression of Kv4.3 (I_to) and increasing the NCX expression [36]. Similar to these results, we observed that chronic 0.2 Hz and 1 Hz pacing of cardiobundles resulted in significant APD prolongation (Fig. 2G & 4E). Since electrochemical species generated from Faradaic reactions at the electrode/electrolyte interface could contribute to this effect, alternative use of optogenetic pacing [64, 65] may be a means to address this issue in the future. Independent of electrical stimulation, application of thyroid hormone, T3, promoted the maturation of cardiobundles as evidenced from the switch of fetal ssTnI to the adult cTnI isoform (Fig. 4K) and acceleration of action potential repolarization and twitch kinetics (Fig. 3A, E & 4E, H) that were consistent with previous studies [20, 25]. Modulating the APD of engineered cardiac tissues to more closely match that of adult myocardium is especially important for reducing potential arrhythmogenesis upon cardiac implantation [66].

We also found that synergy between T3 supplementation and continuous contractile activity was necessary to induce significant ultrastructural maturation of NRVMs after 5-week of culture that in multiple aspects did not differ from the ultrastructure of age-matched and adult rat ventricular myocytes (Fig. 5F-H). This advanced in vitro maturation was evident from qualitative and quantitative analyses of immunostaining and TEM images that demonstrated the polarization of N-cadherins to cardiomyocyte ends (Fig. 2L, M), formation of well-defined intercalated disks, ubiquitous expression of T-tubular structures, H-zones, and M-lines, and changes in sarcomere length and z-band width (Fig. 5). Specifically, in rodent and human heart development, polarization of mechanical junctions to intercalated discs at the ends of cardiomyocytes precedes that of gap junctions [67, 68]. Partial polarization of N-cadherins (but not C×43, Fig. S4) in 5-week cardiobundles is consistent with the time-course of postnatal cardiac development in rats, where the processes of N-cadherin and C×43 polarization are only completed 3 months after birth [68]. Furthermore, almost 100% of sarcomeres in active, T3-treated cardiobundles exhibited distinct M-lines, a property of highly mature myocardium present also in native 5-week and adult rat CMs (Fig. 5F), but infrequently observed in previous cardiac tissue engineering studies [69, 70]. Our analysis also showed significantly increased abundance of T-tubules that formed dyads with the abutting sarcoplasmic reticulum (Fig. 5B); albeit, Di-4 ANEPPS (Fig. 3L) and Cav3 (Fig. S8) stains at the gross tissue level indicated less developed T-tubular structures compared to those reported for adult rat myocardium [71]. Additionally, compared to 5-week quiescent tissues, active, T3-treated cardiobundles showed increased sarcomere length and more uniform Z-band width, approximating what others [57, 72] and we (Fig. 6G,H) have measured in adult rat cardiomyocytes. While Hirt et al. also found that chronic pacing improved ultrastructure of engineered heart tissues [57], the narrower Z-bands in cardiobundles indicate that synergistic action of T3 and low-rate pacing induce more advanced sarcomere maturation in vitro.

Overall, the 5-week cardiobundles exhibited several key features of the adult ventricular myocardium including: (1) high cell density, alignment, and interconnectivity, (2) high CV (i.e. electrical excitability) without spontaneous automaticity, (3) adult-level contractile stress, (4) highly mature ultrastructure, and (5) low rate of CM proliferation (< 0.2%, in 10% horse serum). However, other tissue properties (e.g. random C×43 distribution, below-adult CV and cell size) remain to be improved, potentially by multiplexing chronic pacing and T3 treatment with other growth factors and cytokines, external mechanical loading, and/or co-culture with non-CMs [1-3]. Besides providing a platform to further optimize methods for rapid CM maturation in vitro, the mature post-mitotic cardiobundles may allow a relatively efficient screening (~15 tissues made per neonatal heart) of candidate mitogens that can stimulate endogenous regeneration of the injured adult heart through CM reentry into a cell cycle. Interestingly, low-rate pacing between 3 and 5 weeks of culture was sufficient to significantly increase cardiomyocyte DNA synthesis and nuclei numbers in cardiobundles (Fig. 6D, F and Fig. S10), while only in the presence of T3, this effect was accompanied with both cardiomyocyte and cardiobundle growth. The simultaneous increase in CM size and number of nuclei indicated that active T3-treated cardiobundles likely underwent increased cell ploidy and multinucleation typical of postnatal heart growth [73], rather than cell division; however, these findings and underlying mechanisms remain to be studied in more detail in the future.

In summary, we have combined T3 treatment with chronic active mechanical loading induced by low-rate electrical stimulation to engineer highly mature rat cardiac tissues with structural and functional properties comparable to those of adolescent or adult rat myocardium. The utility of similar culture strategies combining biophysical and hormonal stimulation to promote in vitro maturation of hPSC-derived cardiac tissues remains to be explored.

Statement of Significance.

Compared to human stem cell-derived cardiomyocytes, neonatal rat ventricular myocytes show advanced maturation state which makes them suitable for in vitro studies of postnatal cardiac development. Still, maturation process from a neonatal to an adult cardiomyocyte has not been recapitulated in rodent cell cultures. Here, we show that low-frequency pacing and thyroid hormone supplementation of 3D engineered neonatal rat cardiac tissues synergistically yield significant increase in cell and tissue volume, robust formation of T-tubules and M-lines, improved sarcomere organization, and faster and more forceful contractions. To the best of our knowledge, 5-week old engineered cardiac tissues described in this study are the first that exhibit both ultrastructural and functional characteristics approaching or matching those of adult ventricular myocardium.