The pCAG-Fucci2a plasmid (the RIKEN Center for Life Science Technologies, RDB13080) was obtained through RIKEN (19, 20). mCherry was kindly provided by Shaner et al. (21). The Fucci2a portion (mCherry-hCdt1-T2A-mVenus-hGem) was PCR amplified and cloned in frame downstream of the TNNT2 start codon into a custom synthesized (Thermo Fisher Scientific) plasmid backbone containing the ColE1 origin of replication, ampicillin resistance, 1,000 bp 5′ and 980 bp 3′ homology arms to the hTNNT2 (NM_000364.4) transcription start site, IRES-Puro and an FRT-flanked PGK-hygromycin selection cassette to generate the targeting vector using the NEBuilder High-Fidelity DNA Assembly Cloning Kit (NEB, E5520). Twenty-one bp downstream of the start codon, including the sgRNA target sites, was excluded from the targeting vector to protect it from CRISPR/Cas9-induced double-strand breaks. The resulting targeting vector was verified by Sanger sequencing, linearized by restriction with SspI and SfiI enzymes and phenol–chloroform-extracted. Two different sgRNAs (spacer sequences below), each targeting the region behind the start codon of hTNNT2 that was excluded from the targeting vector, were transcribed in vitro using an EnGen® sgRNA Synthesis Kit (NEB, E3322S). The generation and characterization of CRTD5 hiPSCs (hPSCreg, CRTDi005-B) was previously described (22). CRTD5 hiPSCs (800,000) were electroporated with 10 μg linearized targeting vector, 1 μg of each sgRNA and 60 pmol Cas9-NLS protein (EnGen® Spy Cas9 NLS, NEB, M0646 M) with the Lonza 4D X-unit, pulse CB-150 and Primary Cell 4D-Nucleofector Kit L (V4XP-3024, Lonza). Transfected cells were seeded at clonal density in dishes coated with hESC-qualified Matrigel (Corning Life Sciences, 354277) in mTeSR1 medium (StemCell Technologies, 85850) supplemented for 3 days with 10 μM Rock inhibitor Y-27632 (StemCell Technologies, 78003). Cells were selected with 50 μg/ml hygromycin B (Thermo Fisher Scientific, 10687010) starting on Day 3 after nucleofection for 7 days, and resistant colonies were manually selected, clonally expanded and screened for correctly targeted clones by colony PCRs amplifying the 5′ and 3′ junction of the targeted alleles from outside of the homology arms into the insert as well as the presence of an intact second allele. Heterozygously targeted clones were selected, and the complete insert was analyzed by Sanger sequencing. The selected complete clone CRTD5-TNNT2-FUCCI #19 was karyotyped by Giemsa banding and showed an intact chromosome set 46, XY[cp20], similar to the parental line (Supplementary Figure 1A).

hTNNT2-sgRNA#3: 5′- GACCATGTCTGACATAGAAG-A3′

hTNNT2-sgRNA#4: 5′- GGTGGTGGAAGAGTACGAGG-3′

The hiPSC line (CRTD5) generated from human fibroblasts was used in this study as a control and was obtained from the Stem Cell Engineering facility of the Center for Molecular and Cellular Bioengineering (CMCB), TU Dresden. Research with CRTD5 hiPSC was approved by the Ethikkommission an der Technischen Universität Dresden (BO-EK-38012020). Cells were propagated using ReLeSR (StemCell Technologies, 05873) and maintained in mTESR1^TM (StemCell Technologies, 85850) on Matrigel^TM-coated plates (Corning Life Sciences, 354234) under standard culturing conditions (37 °C, 5% CO₂). Cell cultures were routinely checked for mycoplasma using the LookOut® Mycoplasma PCR Detection Kit (Sigma Aldrich, D9307).

Differentiation of CRTD5 and FUCCI-CRTD5 lines was induced by adaptation of a previously described protocol (17, 23). Briefly, undifferentiated hiPSCs were passaged into 12-well plates using Versene (Thermo Fisher Scientific, 15040066). When hiPSC culture reached 90–95% confluency, differentiation was induced using CDM3 medium (17) supplemented with the GSK3β inhibitor CHIR99021 (4 μM, Sigma, SML1046) for 48 h followed by treatment with IWP2 (5 μM, Tocris, 3533) for an additional 48 h. After, cells were cultured in CDM3 medium. At Day 15 of differentiation, hiPSC-derived CMs were gently detached from the plate by incubating with 1 mg/ml collagenase B dissolved in CDM3 medium for 30 min at 37°C. hiPSC-derived CMs were further dissociated using 0.25% trypsin-EDTA for 5 min at 37°C. The reaction was stopped by adding a double volume of stop medium (80% CDM3, 20% FBS, ROCK inhibitor). Cells were plated at a density of 1,000,000 cells/well into Matrigel-coated 12-well plates. CDM3 was exchanged every second day until further analysis.

hiPSC-derived CMs were detached from the plate by incubating with 1 ml of TrypLE (Thermo Fisher Scientific, 12604013) for 5 min at 37°C. A double volume of PBS was added to stop the reaction, and the cells were passed through a 100 μm cell strainer. hiPSC-derived CMs were centrifuged at 200 g for 5 min, fixed with 1% PFA for 20 min and washed 3x with PBS. hiPSC-derived CMs were incubated with primary antibodies against mVenus (Biorbyt, orb334993, 1:800) and mCherry (Abcam, ab125096, 1:250) and anti-cardiac troponin T-APC (Miltenyi Biotec, 130-120-403, 1:50) in blocking buffer (PBS, 4% donkey serum, 0.1% Triton X-100 in PBS, 2 mM EDTA) for 2 h at RT. After washing, the cells were incubated with the secondary antibodies anti-goat Alexa Fluor® 488 (Jackson ImmunoResearch, 705-546-147, 1:500), anti-rabbit Alexa Fluor® 555 (Abcam, ab150062, 1:500) and Hoechst 33342 (Thermo Fisher Scientific, H21492) for 1 h at 4°C. Cells were then washed 3x with PBS and centrifuged at 200 g for 5 min. Finally, 5 million cells were resuspended in 500 μl of FACS buffer (PBS, 2% FBS, EDTA) and kept on ice until further analysis. Cells were analyzed on Amnis ImageStream-X MkII (Luminex, United States).

Cells were fixed with 4% formaldehyde solution in PBS for 10 min and stained overnight at 4°C with primary antibodies against mVenus (1:800, Biorbyt, orb334993) and mCherry (1:250, Abcam, ab125096) and anti-cardiac troponin T (1:250, Thermo Fisher Scientific, MA5-12960) in blocking buffer (PBS, 4% donkey serum, 0.1% Triton X-100 in PBS, 2 mM EDTA). Cells were then washed 3x with PBS and incubated with the following secondary antibodies in PBS: anti-goat Alexa Fluor® 488 (1:500, Jackson ImmunoResearch, 705-546-147), anti-rabbit Alexa Fluor® 555 (1:500, Abcam, ab150062), anti-mouse Alexa Fluor® 647 (1:500, Jackson ImmunoResearch, 715-606-151) and Hoechst 33342 (Thermo Fisher Scientific, H3570).

CRTD5 and FUCCI-CRTD5 cardiomyocytes at Day 25 of differentiation were first costained with antibodies against TNNT2 (mouse, Thermo Fisher Scientific, MA5-12960) and α-actinin (rabbit, ThermoFisher Scientific, 701914). Imaging of single cardiomyocytes was performed using a Zeiss LSM 700. For cell area analysis, using ImageJ-Fiji software, a defined region of interest (ROI) was defined to outline the outer edges of the cell, and the cell area was measured for each cardiomyocyte. For sarcomere spacing measurements, an ROI with at least 10 sarcomeres was defined. The intensity or the ROI shows a series of peaks that correspond to the spatial frequency of the sarcomeric pattern, and the amplitude between the peaks was determined to assess the sarcomere spacing.

Whole litters of C57BL/6JRj mice (P0) were used for isolation of mNCMs. C57BL/6JRj mice were originally obtained from Janvier labs and bred internally in CRTD animal facilities. All procedures were approved by the local ethics committee, Landesdirektion Sachsen (TVT-1/2017). P0 mNCMs were isolated using the Neonatal Heart Dissociation kit (Miltenyi Biotec, 130-098-373) following the manufacturer's instructions. Cells were seeded in plating medium (20% M199 (Thermo Fisher Scientific, 12340030), 65% DMEM (Thermo Fisher Scientific, 31966021), 5% FBS (fetal bovine serum), and 10% HS (horse serum) at a seeding density of 35,000 cells/well in a 96-well plate coated with Matrigel (Corning Life Sciences, 354234). Cells recovered for 1 day in an incubator (37°C, 5% CO₂).

HiPSC-derived CMs at different time points post cardiomyocyte induction (0, 6, and 30 days) were imaged using a Keyence BZ-X800E microscope (Keyence, Japan). Images were acquired using brightfield, YFP and Cy3 filter sets. For time-lapse imaging, FUCCI hiPSC-derived CMs at Day 30 of differentiation were seeded in CDM3 medium at a density of 150,000 cells per well in a 24-well plate (Cellvis, P24-0-N) coated with Matrigel (Corning Life Sciences, 354234). Time-lapse imaging was performed as previously described (14). Briefly, FUCCI hiPSC-derived CMs were imaged every 20 min for 72 h, and the quantification of the mCherry and mVenus fluorescent intensities was performed using Keyence image measurement and analysis software (Keyence, Japan) and ImageJ-Fiji software. A region around the cell was first defined as the background region, and the cell nucleus was segmented using the brightfield signal. The mCherry and mVenus intensities were detected in both the background region and the nucleus. The background signal was subtracted from the nuclear intensity. Single-cell intensity data were aligned based on the peak mVenus intensity and plotted over a period of 40 h using GraphPad Prism.

Culture hiPSC-derived CMs or mNCMs were fixed with 4% formaldehyde solution in PBS for 10 min and stained overnight at 4 °C with primary antibodies against cardiac troponin I (1:500, Abcam, ab56357) and Ki-67 (CellSignaling, 9449T) in blocking buffer (PBS, 4% donkey serum, 0.1% Triton X-100 in PBS, 2 mM EDTA). Cells were then washed 3x with PBS and stained with the secondary antibodies anti-goat Alexa Fluor® 488 (Jackson ImmunoResearch, 705-546-147, 1:500), anti-rabbit Alexa Fluor® 555 (1:500, Abcam, ab150062), and Vybrant DyeCycle Violet Stain (Invitrogen, V35003) in PBS. Images were acquired using a Keyence BZ X800 fluorescence microscope (Keyence, Japan) equipped with an imaging cytometer (BZ-H4XI). Image analysis was performed in the open source software CellProfiler 4.2.1 (24). Nuclei segmentation was performed using the identify primary objects module with an adaptive thresholding method to account for background variances. Cardiac troponin I and Ki-67 intensities were measured in segmented nuclei, and thresholds were determined according to their histograms. DNA staining intensities of non-cycling (Ki-67⁻) cardiomyocyte nuclei (cardiac troponin I⁺) were measured, and ploidy levels were plotted as histograms, from which ploidy thresholds were determined (Supplementary Figure 3). The number of nuclei per cardiomyocyte was determined manually analyzing a minimum of 15 field of views for each biological replicate.

TNNT2-FUCCI hiPSC-CMs were differentiated as described above. hiPSC-CMs were used 30 days after starting the differentiation. hiPSC-CMs were plated on 96-well glass bottom plates (Cellvis, P96-1-N) coated with Matrigel (Corning Life Sciences, 354277) at a seeding density of 4000 cells per well. Cells were seeded in a volume of 50 μl RPMI20 medium (RPMI 1640 with 20% FBS). Outer wells were left unused and filled with PBS to exclude well plate edge effects. The medium was changed the next day for RPMI 1640 (Thermo Fisher Scientific, 32404014) + B27 (Thermo Fisher Scientific, 17504044) + 0.1% FBS (Thermo Fisher Scientific, 10500064). Subsequently, hiPSC-CMs were left for 3 days to reduce the baseline level of proliferation induced by the plating medium before starting the screen.

Four days after seeding the cells, the autophagy library (ENZO, BML-2837-0100) was added to wells by total medium exchange at an end concentration of 25 μM. Then, the medium was supplemented with nontoxic concentrations of Hoechst 33342 (10 ng/ml, Thermo Fisher Scientific, H3570) and EdU (5 μM, Thermo Fisher Scientific, C10340). The compound library and controls were divided over two template plates, from which the compounds in the medium were transferred into triplicate screening plates. All pipetting was performed with automated channel pipets to avoid interwell variation. Medium with compounds was not exchanged throughout the 72 h duration of the screen.

All images were acquired using a Keyence BZ-X800E compact fluorescence microscope equipped with live imaging cytometer (BZ-H4XI) and CO₂ control (BZ-H4XT) modules. Live images were acquired using the 10x objective and DAPI, YFP and Cy3 filter sets. Binning was set to 3x3, the gain to 6 dB and exposure times were 28 ms for DAPI, 666 ms for YFP and 167 ms for Cy3 channels. The resulting images were 640x480 pixels. Five non-overlapping sites were imaged per well. LIVE imaging was performed at Day 0 directly after adding the compound library and every 24 h until 72 h. After the last LIVE acquisition, hiPSC-CMs were fixed by incubation with 4% formaldehyde solution in PBS for 15 min. Subsequently, an EdU click-it reaction (Thermo Fisher Scientific, C10340) was performed according to the manufacturer's protocol to visualize EdU incorporation. Images of fixed cells were acquired using DAPI and Cy5 channels, binning was set to 3x3, gain to 6 dB and exposure times were 100 ms for DAPI and 10 ms for Cy3.

Automated image analysis was performed in the open source software CellProfiler 4.2.1 (24). Briefly, single channel images from DAPI, YFP, and Cy3 channels were imported into the program. In the first module, the YFP images were enhanced to reduce uneven backgrounds. Subsequently, the DAPI channel was used to identify all nuclei and assign them as primary objects. In the next modules, the intensity of mVenus and mCherry from the YFP and Cy3 channel images within the nuclei was measured, and a threshold was set to categorize the nuclei as positive (+) or negative (–) for these channels. In the following modules, we assigned all nuclei to a single category as follows: mVenus- and mCherry- nuclei: BLUE; mCherry⁺ and mVenus⁻ nuclei: RED; mCherry⁺ and mVenus⁺ nuclei: YELLOW; mVenus⁺ and mCherry⁻ nuclei: GREEN. In the last module, counts were exported into spreadsheets for further data analysis.

Data analysis was performed using the open-source data analytics software KNIME (Konstanz Information Miner) version 4.3.1. The general approach for screen analysis was followed as described by Stöter et al. (25). Briefly, CellProfiler output files were imported, and the following steps were performed: quality control, filtering, grouping, normalization, and statistics. As a first step, data from individual sites were excluded from analysis if the number of nuclei was far below (>2 STDEV) mean levels (indicating problems with focus, e.g.,). In the second step, the results from imaging sites were grouped by wells, giving mean values per well from the five sites imaged. In the next steps, the data from wells were matched with locations on the plates and the compound information. The percentages per compound were normalized to the control (100%). Next, wells were grouped by treatment, and if parameters showed STDEVs larger than half the value of the parameter, they were excluded as quality controls. Data were exported into Excel, and the main parameter (% of mVenus⁺ in all FUCCI⁺) was plotted using GraphPad Prism software.

Animals were housed in the Comparative Medicine Biomedicum (Karolinska Institutet, Stockholm) animal facility on a 12-h light/dark cycle and were provided food and water ad libitum. All breeding and organ collection protocols were performed in accordance with the Swedish and European Union guidelines and approved by the institutional ethics committee (Stockholms Norra Djurförsöksetiska Nämnd). C57BL/6N neonatal mice were injected subcutaneously with a volume of 0.1 ml of PBS with EdU (20 mg/kg, Invitrogen, E10187) and clonidine (60 ng/pup, Sigma, C7897). Clonidine was given for an estimated weight of 1.5 g/pup throughout the experiment; therefore, a fixed dose of 60 ng/pup was used. Neonatal mice were sacrificed at P7 by decapitation, and hearts were dissected and collected in PBS, cryoprotected in 30% sucrose and flash-frozen in isopentane.

Frozen hearts were sectioned into 10 μm thick sections at the cryostat. After washing with PBS sections were fixed by incubating in 2% formaldehyde solution for 10 min. Primary antibodies against rabbit PCM-1 (1:100, Santa Cruz, sc-67204), mouse SMA-Cy3 (1:1000, SigmaAldrich, C6198), and biotinylated isolectin B4 (Vector labs, B-1205, 1:500) were diluted in blocking buffer (PBS, 4% donkey serum, 0.1% Triton X-100 in PBS, 2 mM EDTA) and sections were incubated overnight at RT. Sections were then washed 3x with PBS (15 min) and stained with matching secondary antibody anti-rabbit Alexa Fluor® 488 (1:500, Jackson ImmunoResearch, 711-546-152) and Streptavidin-Alexa Fluor 647® (1:1000, ThermoFisher, S21374) in PBS. Subsequently, Click-iT™ EdU Cell Proliferation Kit for Imaging, Alexa Fluor™ 647 or 488 dyes (ThermoFisher, C10340 and C10637) was used to detect EdU. Slides were then washed and mounted using ProLong Gold Antifade Mountant with DAPI (ThermoFisher, P36931). Images were acquired using a Keyence BZ X800 fluorescence microscope (Keyence, Japan) and Zeiss LSM 750 confocal microscopy from at least three regions of the left ventricle per heart.

Frozen hearts were placed on ice, cut into small pieces, fixed in four percent paraformaldehyde for 1 h, and digested for 2 h at 37°C (3.6 mg/ml collagenase B, 4.8 mg/ml collagenase D, in PBS). Cells were incubated with primary antibody against α-Actinin (Thermofisher, A7811, 1:500) and Connexin-43 (SigmaAldrich, C6219, 1:1000) for 30 min to label cardiomyocyte cytoplasm and borders. After washing with PBS, cells were resuspended with matched secondary antibodies [AF488-coupled (Abcam, ab150110, 1:1000) and AF555-coupled (Jackson Immuno Research, 711-546-152, 1:500)] for 30 minutes. Cardiomyocytes were placed on a slide with a mounting medium (Invitrogen™ ProLong™ Gold Antifade Mountant with DAPI). Images were acquired using a Keyence BZ X800 fluorescence microscope (Keyence, Japan) equipped with an imaging cytometer (BZ-H4XI). Image analysis was performed in the open source software CellProfiler 4.2.1 (24). Nuclei segmentation was performed using the identify primary objects module with an adaptive thresholding method to account for background variances. α-Actinin intensities were measured in segmented nuclei, and thresholds were determined according to their histograms. DNA staining intensities of cardiomyocyte nuclei were measured, and ploidy levels were plotted as histograms, from which ploidy thresholds were determined (Supplementary Figure 4). The number of nuclei per cardiomyocyte was determined manually analyzing a minimum of 15 field of views for each biological replicate.

We generated a TNNT2-FUCCI hiPSC line using a CRISPR–Cas9 approach (see Methods). TNNT2-FUCCI hiPSCs expressed the FUCCI construct under the control of the cardiomyocyte-specific TNNT2 promoter (Figure 1A and Supplementary Figure 1A). The generated TNNT2-FUCCI hiPSC line showed a normal karyotype, and pluripotency characterization showed high expression of bona fide pluripotency markers and could generate three germ layers (Supplementary Figures 1B–D). FUCCI fluorescence (mCherry/mVenus) became visible from Day 6 post differentiation into cardiomyocytes (Figure 1A). At Day 30, in most cells, FUCCI signal was detected (Figure 1A), and all FUCCI-expressing cells were TNNT2⁺ (Figure 1B). To further confirm the cardiomyocyte-specific expression of TNNT2-FUCCI fluorescence, we used imaging flow cytometry. We analyzed 50,000 cells defined by their TNNT2 expression and found that 85.4% were mCherry-Cdt1⁺ (G0/G1 phase), 5.1% were mVenus-geminin⁺ (G2/M phase) and 1.6% of the cardioymocytes were both mCherry-Cdt1 and mVenus-geminin⁺ (G1/S phase) (Figure 1C). Importantly, we found no FUCCI expression in the TNNT2⁻ cell fraction. However, some signal in the TNNT2⁻ fraction detected in the green and red fluorescence channels could be assigned to autofluorescence (Supplementary Figure 2A).

To demonstrate that the TNNT2-FUCCI signal reliably indicates the cell cycle status of cardiomyocytes, we costained FUCCI cardiomyocytes with cyclin-dependent kinase-1 (CDK1) and found a complete overlap of mVenus⁺ and CDK1⁺ nuclei, supporting that mVenus⁺ cardiomyocytes indeed represent cells in the S/G2/M phase of the cell cycle (Supplementary Figure 2B).

TNNT2-FUCCI cardiomyocytes showed a similar sarcomere spacing pattern (1.97 μm ± 0.06 μm SEM) and cell size (2525.4 μm² ± 259.2 μm² SEM) as controls (Supplementary Figures 2C,D), suggesting that the integration of the TNNT2-FUCCI knock-in did not compromise the functionality of the TNNT2-FUCCI cardiomyocytes.

In summary, we showed that TNNT2-FUCCI expression is limited to cardiomyocytes and reliably detects cell cycle progression.

We tracked single TNNT2-FUCCI cardiomyocytes over 72 h (Figure 2) Among 570 cardiomyocytes analyzed, 90.4% ± 10.6% SEM did not show any cell cycle activity, 5.1 % ± 1.5% SEM underwent cytokinesis and cell division, 3.2% ± 1.5% SEM became binucleated, and 1.40% ± 0.8% SEM underwent polyploidization (Figure 2A). We could not detect any binucleated or polyploid cardiomyocyte undergoing cell division, supporting the notion that mainly diploid mononucleated cardiomyocytes show proliferative capacity (9). Next, we plotted the FUCCI oscillation pattern for cardiomyocytes undergoing cell division, multinucleation and polyploidization (Figures 2B–D and Supplementary Movies 1–3). In all three cycling populations, mVenus-geminin fluorescence started to increase concomitant with the reduction of the mCherry-Cdt1 signal, marking the end of the G1 phase and the beginning of the S/G2 phase. In early mitosis, mVenus fluorescence dropped sharply concomitant with nuclear envelope breakdown in dividing and multinucleating TNNT2-FUCCI cardiomyocytes (Figures 2B,C). The average duration of the S/G2/M phase was 16.38 h ± 0.84 SEM h in dividing cells and 17.29 h ± 0.55 SEM h in multinucleating cells, with no significant difference between the two populations (p > 0.05, Figure 2E). In contrast, polyploid cells showed a different FUCCI oscillation pattern. An increase in the mCherry signal was detected before the loss of mVenus, which dropped slowly, suggesting G2 phase arrest without activation of the anaphase-promoting complex (26). Accordingly, no mitotic features, such as cell rounding or cytoplasmic localization of mVenus, were detected (Figure 2D). Moreover, mVenus expression was detected over a significantly longer period in polyploid cells compared to dividing (p = 0.0009) and multinucleating (p = 0.0029) cells, with an average duration of the S/G2 phase of 24.5 h ± 2.77 h SEM (Figure 2E). These data suggest that hiPSC-derived CMs that become polyploid do not enter mitosis and remain arrested in theG2 phase.

Next, we devised a live screening platform in which we probed TNNT2-FUCCI cardiomyocytes for cell cycle entry using a library of 94 autophagy-related compounds (Figure 3A, and Methods). TNNT2-FUCCI live cardiomyocytes were imaged, and mVenus and mCherry nuclear fluorescence was documented at 24 h, 48 h, and 72 h (Figure 3A). Proproliferative activity was determined as the percentage of mVenus⁺ (S/G2/M phase) cardiomyocyte nuclei and normalized to the control group at all three time points (Figure 3B and Supplementary Figures 3A,B). The acquisition time point of 48 h with the smallest coefficient of variance (24 h, 48 h, 72 h; 13.1%, 12.1%; 28.0%, respectively) was chosen to select six pro-proliferative candidates. We selected the six pro-proliferative candidate compounds based on their increase in the percentage of mVenus⁺ nuclei and their biological significance (see Methods and Supplementary Figures 3A,B).

The six selected compounds were tested at three different concentrations to establish potential concentration dependencies (2.5 nM, 250 nM, 25 μM) in mNCMs. Cell cycle activation was validated by EdU incorporation after 72 h which can detect DNA synthesis during S-phase (Figure 3C). We found two compounds that significantly increased cell cycle activity at all tested concentrations. Of these, the alpha-adrenergic and imidazoline receptor agonist clonidine showed the most pronounced concentration-dependent effect (2.78-fold increase compared to untreated, p < 0.0001) on cell cycle activity in mNCMs (Figures 3C,D and Supplementary Figure 3C). Hence, we continued to explore the pro-proliferative potential of clonidine (Supplementary Figure 3D).

We assessed whether clonidine-induced cell cycle activity results in proliferation or polyploidy in hiPSC-derived CMs (Figures 4A–E). Cell cycle activity was first determined by Ki-67 expression. A cell cycle related gene, which expression is gradually increased during cell cycle progression, reaching a maximum at G2/S (27). Although there are reports that Ki-67 expression is linked to DNA damage response (27, 28), Ki-67 is widely used to demonstrate cell cycle activity (28). We could show a high fidelity of Ki-67 in detecting cell cycle progression (S/G2/M) in neonatal mice (14). We measured an increase in cell cycle activity after clonidine treatment through Ki-67 expression from 9.0% ± 1.23% SEM to 22.0% ± 4.51% SEM at 72 h after treatment in hiPSC-CMs (p = 0.03, Figure 4A). Clonidine did not cause any changes in polyploidy (p > 0.05, Figure 4B, and Supplementary Figure 4A) or in binucleation (p > 0.05, Figure 4C), suggesting that most clonidine-induced cell cycle activity results in proliferation. Consistent with these findings, we found an increase in aurora B kinase (AurKB)-positive midbodies, indicative of late phase cytokinesis, from 0.83% ± 0.17% SEM in untreated hiPSC-derived CMs to 1.69% ± 0.24% SEM in clonidine-treated hiPSC-derived CMs (p = 0.04, Figures 4D,E).

Next, we investigated whether clonidine elicits similar effects in mNCMs that are more restricted in their capacity to proliferate. We found an increase in cell cycle activity, measured through Ki-67 expression, from 10.97% ± 1.68% SEM in untreated mNCMs to 23.23% ± 2.63% SEM in clonidine-treated mNCMs (p = 0.002, Figure 4F). The ratio of binucleated cardiomyocytes was not altered by clonidine treatment (p > 0.05, Figure 4H), but we found an increase in nuclear ploidy with an increase in the tetraploid fraction from 26.52% ± 0.54% SEM in untreated mNCMs to 38.93% ± 2.32% SEM in clonidine-treated mNCMs (p = 0.0004, Figure 4G, Supplementary Figure 4B), suggesting that a substantial proportion of clonidine-induced cell cycle activity can be attributed to nuclear polyploidy. Additionally, we performed AurKB staining on clonidine-treated and control mNCMs. Although we observed AurKB⁺ cardiomyocyte nuclei in all cultures (Supplementary Figure 4C), we could not detect any positive midbodies in these cultures (more than 10,000 cardiomyocytes analyzed), suggesting that clonidine does not stimulate cytokinesis in cell cycle-active mNCMs. In agreement with this, cardiomyocyte cell count did not show any significant increase (p > 0.05, Supplementary Figure 5D) in the number of cardiomyocytes after clonidine treatment.

To explore whether clonidine triggers cell cycle activity in the mouse heart, similar to what we found in vitro, we administered clonidine to neonatal mice. Clonidine was given along with EdU daily from P1 to P5 (60 ng/day), and the hearts were collected for analysis at P7 (Figure 5A). The number of EdU⁺ cardiomyocyte nuclei (EdU⁺/PCM-1⁺) was significantly increased in the clonidine-treated group (36.15% ± 2.72% SEM) compared to control animals (25.54% ± 1.80% SEM) (p = 0.02, Figures 5B,C), demonstrating that clonidine triggers cardiomyocyte cell cycle activity as more cells have entered S-Phase in both in vitro and in vivo mouse neonatal hearts. In contrast to neonatal cardiomyocyte in vitro, nuclear ploidy levels remained constant (Figures 5D,F and Supplementary Figure 5C), whereas the fraction of mononucleated cardiomyocytes was reduced in clonidine treated animals from 14.8% ± 1.3 SEM%, compared to 22.3% ± 1.7% SEM in controls (p = 0.01, Figure 5E), demonstrating mitotic activity without cytokinesis.

As alpha adrenergic signaling is not restricted to cardiomyocytes (29), we explored the possibility that cell cycle activity is altered in other cell types of the heart. Whereas, we did not see any changes in fraction of EdU⁺ endothelial cells (control: 29.4% ± 2.8% SEM vs. clonidine: 30.1% ± 1.7% SEM, p = 0.85), we found an increase in smooth muscle cell cycle activity from 44.2% (29.4–46.9%, interquartile range) in controls to 59.3% (51.9–61.7%, interquartile range) in clonidine treated animals (p = 0.04, Supplementary Figures 5D–F).

The identification of cycling cardiomyocytes is critical for studying cardiomyocyte proliferation (10, 14, 31–34). Here, we generated the TNNT2-FUCCI hiPSC reporter line and differentiated it into cardiomyocytes. Consistent with previous data, we showed that inactivation of one TNNT2 locus by FUCCI knock-in did not cause structural impairments in cardiomyocytes (35). After 6 days in culture, we found the first cells expressing TNNT2-FUCCI along with the appearance of beating cardiomyocytes at Day 8 of differentiation (17). We found a complete overlap of cyclin-dependent kinase 1 (CDK1), which oscillates in the cell cycle and shows activity in S/G2/M phases (36), with mVenus-geminin⁺ cardiomyocytes, verifying the fidelity of geminin expression and thereby with the TNNT2-FUCCI system.

Video time-lapse microscopy is considered the gold standard to unequivocally determine the outcome of the cardiomyocyte cell cycle as division, multinucleation or polyploidization (10, 14, 37). Here, we combined time-lapse microscopy with our TNNT2-FUCCI sensor to determine the length and distinct phases of the cardiomyocyte cell cycle. We found that the S/G2/M phase took approximately 18 h for both dividing and binucleating cells with similar TNNT2-FUCCI fluorescent oscillation patterns, documenting a slightly longer cell cycle length than in mouse embryonic E14.5 cardiomyocytes (10.4 h) (15) and in mNCMs at P0 (15.1 h) (14). In contrast, human cardiomyocytes undergoing polyploidy showed a longer S/G2/M phase duration of ~25 h and prominent changes in the TNNT2-FUCCI fluorescence intensity pattern, exemplifying the difficulty in estimating annual cell cycle activity rates based on a fixed duration of cell cycle length (38, 39). Furthermore, our data suggest that polyploidy is not only a result of karyokinesis failure (40) because we did not observe nuclear envelope breakdown in cardiomyocytes but can also be a result of G2-phase arrested cycling cardiomyocytes.

Here, we developed a fluorescence-based live imaging screening platform based on the TNNT2-FUCCI system. This screening platform does not require immunofluorescence staining and eliminates the need for total cell counts and the incorporation of nucleotide analogs (e.g., BrdU and EdU). Live imaging of cycling cardiomyocytes allows for multiple imaging time points to establish the efficacy time course of pro-proliferative compounds. Due to the cardiomyocyte specificity of the TNNT2-FUCCI system, we eliminated false-positive detection of proliferative events originating from non-depleted cycling non-cardiomyocytes. Moreover, TNNT-FUCCI cardiomyocyte specificity allows sophisticated cocultures of cardiomyocytes with other cell types and could even be combined with recently developed cardiac organoids (41, 42). Our platform can be used with any type of library to identify pro-proliferative candidates. In this study, we subjected TNNT2-FUCCI hiPSC-derived CMs to a library of compounds regulating autophagy. Autophagy has been implicated in heart regeneration (43) and exerts numerous roles in myocardial stress responses (44). Autophagy reduces cellular stress caused by reactive oxygen species, which in turn plays a role in cardiomyocyte cell cycle exit (45).

We selected six compounds for further validation in mNCMs (Supplementary Table 1), which are considered more mature than iPSC-derived CMs (18), with a partial loss of their capacity to proliferate. We found that only two of the six compounds significantly enhanced cell cycle activity, potentially due to higher levels of maturation in mNCMs compared to hiPSC-derived CMs and cross-species variation in cardiomyocyte physiology. One of these two compounds was dihydrocapsaicin, an analog of the active component of chili pepper with documented effects on cardiomyocytes, including autophagy induction (46) and attenuation of mitochondrial function (47).

The most robust and dose-dependent effect on cardiomyocyte cell cycle entry was elicited by the alpha-adrenergic and imidazoline agonist clonidine. Although clonidine is an alpha2-adrenergic agonist, it also shows binding to alpha1-adrenergic receptors (48). Furthermore, clonidine activates the imidazoline receptor, leading to a decrease in the level of cAMP in cells, resulting in downstream autophagy activation (49). Alpha-adrenergic receptors are predominantly found on smooth muscle cells of blood vessels, mediating vasoconstriction (29), but can also be detected on cardiomyocytes (50). Accordingly, we found an increase in cell cycle activity of smooth muscle cells in clonidine treated neonatal hearts. Single cell RNA sequencing data, re-analyzed from our previous study (14), showed expression of alpha1B and beta1 adrenergic receptors in P0 and P7 mNCMs. Apart from adrenergic receptor expression, we also found nischarin (Nisch), the mouse homolog of human imidazoline receptor I1 candidate (51), robustly expressed in mNCMs (Supplementary Figures 5A,B). We hypothesized that clonidine mainly elicits its pro-proliferative effects via direct alpha1 adrenergic receptor (50), and imidazoline receptor interaction on cardiomyocytes (52). Alpha1 adrenergic receptors are members of the G protein-coupled receptor superfamily. They activate mitogenic responses and regulate growth and proliferation of many cell types through several downstream signaling cascades such as the c-Jun NH2-terminal kinase (JNK) and the mitogen activated protein kinase (MAPK) pathways (53). Thus, we chose to further investigate whether clonidine treatment in addition to cell cycle entry also results in successful mitosis and cytokinesis.

Based on detected AurKB⁺ midbodies, clonidine promotes cytokinesis in hiPSC-derived CMs. While the increase in cell cycle activity did not alter the ratio between diploid and polyploid cells in hiPSC-derived CMs, in mNCMs, we did observe an increase in polyploid cardiomyocytes and no AurKB⁺ midbody formation after clonidine treatment. The latter indicates that clonidine-induced cell cycle activity was not productive and that for successful completion of mitosis and cytokinesis in mNCMs, additional stimuli are required. Cardiomyocyte maturation is linked to a number of processes including changes in structure, metabolism and gene expression, and plays a major role in cardiomyocyte cell cycle exit (54). Whereas, immature hiPSC-derived CMs still have the capacity to divide given optimal condition (55), mNCMs gradually lose their proliferative capacity during the first postnatal days (1, 2, 45). Silencing of cell cycle and cytokinesis related genes such as AURKB and ECT2 along with an activation of mitochondrial biogenesis and a metabolic switch to oxidative phosphorylation has been attributed to a proliferation-to-hypertrophy transition in the developing and neonatal heart (45, 56). These changes could explain why clonidine acts differently on hiPSC-derived CMs and mNCMs in this study.

In conclusion, we generated TNNT2-FUCCI hiPSCs and demonstrated that TNNT2-FUCCI hiPSC-derived CMs enable analysis of cell cycle entry and progression, providing a powerful platform for screening and validation of pro-proliferative candidates in human cardiomyocytes.