IGFBP2 Mediates Human iPSC-Cardiomyocyte Proliferation in a Cellular Contact-Dependent Manner

Abstract

Background:

Induction of cardiomyocyte proliferation in situ represents a promising strategy for myocardial regeneration following injury. However, cardiomyocytes possess intrinsic inhibitory mechanisms that attenuate pro-proliferative signaling and constrain their expansion. We hypothesized that cell–cell contact is a key suppressor of cardiomyocyte proliferation. We aimed to delineate the underlying molecular pathways to enable sustained proliferation in 3D contexts.

Methods:

Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) were cultured at varying plating densities to examine the impact of cell-cell contact on cell cycle activity. Phosphoproteomic profiling was performed in sparse versus dense cultures to identify signaling alterations. Conditioned media from sparse cultures were interrogated using a human growth factor array to identify secreted pro-proliferative factors.

Results:

hiPSC-CM proliferation increased proportionally with plating density until intercellular contacts were established, at which point proliferation was suppressed. Dense cultures exhibited enhanced adherens junction assembly, sarcomeric organization, and contractile function. Increased cell-cell contact in dense conditions attenuated nuclear translocation of β-catenin and reduced TCF/LEF transcriptional activity, providing a mechanistic basis for the reduced hiPSC-CM proliferation. Disruption of adherens junctions or sarcomere assembly via siRNA-mediated knockdown of N-cadherin or α-actinin, respectively, resulted in increased cell cycle activation of hiPSC-CMs, but this was not sufficient to drive division of hiPSC-CMs. Additional screening for putative secreted growth factors in the conditioned media from sparsely plated hiPSC-CMs revealed the enrichment of IGFBP2, which was sufficient to drive hiPSC-CM division in the presence of cell-cell contact in 3D constructs.

Conclusions:

Our findings demonstrate that cell-cell contact inhibits hiPSC-CM proliferation through adherens junction formation, sarcomeric assembly, and reduced IGFBP2 secretion. Importantly, exogenous supplementation of IGFBP2 can overcome cell contact-mediated inhibition of hiPSC-CM proliferation and facilitate the growth of 3D cardiac tissue. These insights provide valuable implications for advancing cardiac tissue engineering and regenerative therapies.

INTRODUCTION

A healthy myocardium is essential for the heart’s function and blood circulation. In mammals, damaged myocardium is often non-regenerative and can be fatal. This is partly due to the limited proliferative capacity of mammalian cardiomyocytes after terminal differentiation¹. For example, it has been demonstrated that the proliferation capacity of murine cardiomyocytes rapidly declines during embryonic development and becomes negligible a week after birth^2–4.

Essential molecular players and responsible signaling pathways have been identified that are involved in the regulation of cardiomyocyte proliferation. Direct modulation of cell cycle positive regulators such as cyclins and cyclin-dependent kinases, and negative regulators such as Rb and p130, drives cell cycle re-entry and induce stable cell division of terminally differentiated cardiomyocytes^5–10. Genetic or pharmacological modulation of signaling pathways involved in heart development such as Hippo, Wnt, Notch, Neuregulin-1/ErbB, and Insulin Growth Factor (IGF) has also been shown to be effective in stimulating cardiomyocyte proliferation^11–19. Recently, transient reprogramming of adult cardiomyocytes using Yamanaka factors (i.e., Oct4, Sox2, Klf4, c-Myc) was demonstrated to be sufficient to facilitate mammalian heart regeneration²⁰. While activation of fetal gene programs was sufficient to overcome cell cycle arrest signals in fully differentiated cardiomyocytes, activation of pro-proliferative signals such as Wnt has shown insufficient to drive cardiomyocyte proliferation in previous studies^21–25. This raises the question of whether there are additional signals in fully differentiated cardiomyocytes that induce cell cycle arrest. In our present study, we hypothesize that there is a repression mechanism regulating the proliferation of fully differentiated cardiomyocytes.

Contact inhibition plays an important role in regulating cell proliferation and controlling organ size. Cell-cell contact effectively suppresses cell proliferation. Most epithelial cells, for example, exit the cell cycle and cease cell proliferation when they come into contact with neighboring cells in vitro or in vivo to form a confluent cell monolayer^26–28. While the role of cell-cell contacts in regulating cardiomyocyte proliferation and the underlying mechanisms remains largely unknown, there is growing evidence that embryonic cardiomyocyte proliferation may be regulated by cell-cell contact²⁹. In a previous genetic ablation study, when 50–60% of E9.0 embryonic cardiomyocytes were ablated, the residual cardiomyocytes compensated and repopulated the heart by increasing their cell cycle activity²⁹. This suggests that the loss of cell-cell contact may serve as a trigger for stimulating cardiomyocyte proliferation. Our previous study also demonstrated that removal of cell-cell contact induces massive expansion of human iPSC-derived cardiomyocytes (hiPSC-CMs), which are considered equivalent to embryonic cardiomyocytes³⁰. Based on these findings, we hypothesized that cell-cell contact would serve as an essential upstream regulator of cardiomyocyte proliferation. In this study, we 1) examined the effect of modulating cell-cell contact on hiPSC-CM proliferation, 2) investigated molecular players involved in cell-contact-mediated inhibition of hiPSC-CM proliferation, and 3) performed growth factor screening assay to identify putative growth factor candidates that are preferentially secreted in the absence of cell-cell contact and have the potential to overcome contact-mediated inhibitory mechanisms.

METHODS

Detailed methods including phosphoproteomics, siRNA-mediated gene knockdown study, conditioned media study, growth factor screening assay, and 3D cardiac spheroid studies can be found in the Supplemental Materials.

Availability of Data

Description	Source / Repository	Persistent ID / URL
Single cell RNA sequencing	Gene Expression Omnibus (GEO)	GSE301852
Phosphoproteomics Analysis	PRIDE proteomeXchange	PXD061405

Cell Culture

Three previously established hiPSC lines (GM25256 (WTC), peripheral blood mononuclear cells (PBMCs), male; SCVI-111, PBMCs, male; SCVI-273, Sendai virus reprogrammed, PBMCs, female) were maintained in DMEM/F12 (ThermoFisher, cat no. 11330057) supplemented with the essential eight (E8) (R&D) growth factors on a Matrigel (Corning, cat no. 356231)-coated (1:400 for 24h) polystyrene 2D culture system. When undifferentiated hiPSC lines reached 80%-90% confluency, hiPSCs were dissociated in PBS with 0.5mM EDTA for 5–10 minutes at 37°C in the CO₂ Incubator (Galaxy 170S Eppendorf). Dissociation was performed through gentle trituration to produce small aggregates of undifferentiated hiPSCs. Passaging was performed in 1:12 splitting ratios or 15,000 cells per cm² to achieve low density and reach full confluency within 4–5 days. For the first 24h after replating, 10 μM of ROCK inhibitor Y-27632 (Selleckchem, cat no. S1049) was included in the E8 hiPSC maintenance media.

Directed Cardiac Differentiation of hiPSCs

The previously described canonical Wnt modulation protocol in RPMI 1640-based differentiation media supplemented with B27 minus insulin (Gibco, cat no. A1895601) was used for hiPSC-CM differentiation³¹. In brief, during days 0–2, a gradient of CHIR99021 (Selleckchem, cat no. S2924) concentrations (4.0, 5.0, 6.0, 7.0 μM) in B27 minus insulin differentiation media were administered. During days 2–4, after the aspiration of the previous media, Wnt-C59 (2.0 μM) (Selleckchem, cat no. S7037) was added to the B27 minus insulin differentiation media and administered. During days 4–6, after the aspiration of the previous media, B27 minus insulin was administered. During days 6–8, after the aspiration of the previous media, B27 with insulin (Gibco, cat no. 17504–44) was administered. During days 8–10, after the aspiration of the previous media, glucose starvation media (RPMI 1640 without glucose, Gibco, cat no. 11879020) supplemented with B27 with insulin), was administered to metabolically select for cardiomyocytes. At day 10, regular CM maintenance media (RPMI 1640 + B27 with insulin) was administered for hiPSC-CM recovery from glucose starvation.

At day 11, wells that contain more than 90% beating cardiomyocytes were treated with TrypLE Select Enzyme 10X, no phenol red (Gibco, cat no. A1217702) and incubated at 37°C for 10–30 minutes in the CO₂ incubator. Gentle rocking of the plates containing differentiated cardiomyocytes and TrypLE Select was performed every 10 minutes during the 10–30-minute incubation. Single cell suspensions of cardiomyocytes were ensured through gentle trituration and transferred to a 15 mL conical tube containing cardiomyocyte replating media (RPMI 1640 + B27 1X with 10% Knock Out Serum Replacement (Gibco, cat no. 10828028) and Thiazovivin 1.0 μM (Sigma, SML1045)). Cells were gently centrifuged at 200 rcf for 3 minutes and frozen in cell freezing media, Bambanker solution (Bulldog Bio, BB01).

Contact Inhibition Study

Three previously established hiPSC lines (WTC, SCVI-111, SCVI-273) were differentiated and stored in liquid nitrogen as day 11 hiPSC-CMs. To examine the effect of contact inhibition and the addition of CHIR99021 on day 20 CM proliferation, day 11 CMs from all three hiPSC lines were thawed and expanded in 6 well plates using either RPMI 1640 + B27 1X differentiation media supplemented with 2.0 μM CHIR99021 (Selleckchem) or RPMI 1640 + B27 1X differentiation media alone without CHIR99021 until day 27. The cells were then passaged into 48 well plates in varying cell densities ranging from sparse to dense conditions (i.e. 10, 20, 40, 80, 160, 320 × 10³ cells/cm²) to examine the direct cell-cell contact effect. Cells were then expanded using RPMI 1640 + B27 1X differentiation media supplemented with or without 2.0 μM CHIR99021 (Selleckchem) until day 30.

On day 30, cells were fixed at 4% paraformaldehyde for 15 min at room temperature, washed three times with washing buffer (0.1% Tween-20/PBS), permeabilized and blocked with blocking buffer (3% bovine serum albumin/2% goat serum/0.01% saponin/PBS) for 30 min at room temperature. Corresponding primary antibodies were diluted at desired ratios in the blocking buffer (cTnT (Thermo Fisher, MS295P): 1:250, Ki67 (Thermo Fisher, PA5–19462): 1:200) and were incubated with cells placed on a shaker overnight at 4°C. After 24h, the cells underwent three 5 min washes using washing buffer at room temperature. Appropriate Alexa-conjugated secondary antibodies were spun down, diluted in blocking buffer at a 1:300 ratio, and incubated with the cells for 2 hours at room temperature on a shaker. After incubation, the samples were washed three times using the appropriate wash buffer to remove any unbound secondary antibodies. Nuclei were counterstained using Dapi (Invitrogen, R37505) prior to imaging.

Statistical Analysis

We first assessed the normality of the data to determine the appropriate statistical test. When the sample size was sufficient for normality testing and the data were normally distributed, we applied an unpaired t-test or one‑way ANOVA, as appropriate. When the data were not normally distributed, or when the sample size was too low to reliably assess normality, we used a Mann–Whitney test or Kruskal–Wallis test. When post hoc comparisons were performed, the specific method used, as well as both raw and adjusted P‑values, are reported. Multiple testing corrections were applied within each independent analysis when necessary (e.g., Figure 5), but no global, experiment-wide corrections were applied across all analyses. For quantitative analysis of immunostained images, multiple locations were imaged from biologically independent samples (i.e., different differentiation batches). A linear mixed-effects model (LMM) was fitted using restricted maximum likelihood (REML) via the lmerTest package in R (v4.1.0). The model included treatment as a fixed effect and sample identity as a random intercept. Residual normality was assessed using the Shapiro–Wilk test. All details related to the statistical analyses—including normality testing, data normalization, sample sizes, statistical test names, post hoc corrections, and both raw and adjusted P‑values—are summarized in Table S1. To ensure objectivity in the visual representation of results, representative images were selected based on the sample with a value closest to the calculated average for that group.

Figure 5. IGFBP2 can drive in situ proliferation of hiPSC-CMs in 3D environment.

(a) Schematic of cardiac spheroid generation and IGFBP2 treatment study. (b) 2D bright-field images of cardiac spheroids throughout culture period; Scale bar = 200 μm. (c) Quantification of 2D spheroid projected area over time (6 biological replicates). (d) Fold-change in the projected area of cardiac spheroids after IGFBP2 treatment (normalized to D0). (e) 3D volumetric imaging of cardiac spheroids after optical clearing and staining; Scale bar = 150 μm. (f) Volume quantification of cardiac spheroids using IMARIS analysis software. (g) Quantification of nuclei (propidium iodide, PI+) per cardiac spheroid and (h) the fraction of Ki67 positive nuclei. Data are presented as mean ± sem. *p<0.05, **p<0.01, ns = not significant (p>0.05) based on One-way ANOVA (d) and Mann-Whitney test (f-h). Cell line: SCVI-111 data are shown here. SCVI-273, WTC line data are in supplemental information (Supplementary Figures 14,15).

RESULTS

Cell cycle activity of hiPSC-CM depends on cell plating density biphasically

We previously showed that cell-cell contact inhibits Wnt-stimulated hiPSC-CM proliferation³⁰. Here we examined the mechanism of cell contact-mediated inhibition of proliferation by assessing the effect of cell plating density on cell cycle activity of hiPSC-CMs (Figure 1a). Interestingly, hiPSC-CMs responded biphasically to cell plating density. At cell densities below cell-cell contact (i.e. 20,000 – 80,000 cells/cm² depending on iPSC line), the expression of Ki67 increased proportionally to cell density. However, at densities where hiPSC-CMs can easily form cell-cell contact (i.e. at or greater than 20,000 – 80,000 cells/cm²), hiPSC-CM expression of Ki67 is either unchanged or decrease with cell density (Figure 1b,c, S1,2). This biphasic response was observed in cells treated regardless of Wnt activation (Figure 1d,e, S1,2). The density-dependent changes in hiPSC-CM cell cycle activity were confirmed in three independent hiPSC lines (SCVI-111, 273, WTC) (Figure 1, S1,2), supporting the consistencies of this density-dependent effect on hiPSC-CM proliferation.

Figure 1. Cell-cell contact reduces proliferation in human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) regardless of Wnt activation.

(a) Schematic of the experimental procedure for examining the effect of cell-cell contact on cell cycle activity. (b,c) Effect of varying cell plating density on the fraction of proliferating day 30 hiPSC-CMs upon Wnt activation. The proliferating fraction of hiPSC-CMs increases as cell density increases in sparse cell condition (i.e. 10,000 – 40,000 cells/cm²). As cellular contacts are made (arrow), hiPSC-CM cycling is saturated or even repressed at high cell density (>80,000 cells/cm²). Data are plotted as mean ± sem (n=3). (d,e) Effect of varying cell plating density on the fraction of proliferating day 30 hiPSC-CMs in the absence of Wnt activation. Despite an overall decrease in the absolute proliferation rate, the biphasic trend of cell-density-dependent proliferation rate was seen regardless of Wnt activation. Data are presented as mean ± sem (n=3). The arrows indicate when cells start forming cell-cell contacts. (b,d) Representative images are from WTC line. Data from SCVI-273,−111 lines are presented in Figure S1,2, respectively. Representative images were selected based on values closest to the group mean. Scale bar = 50 μm.

Cell-cell contact facilitated adherens junction formation, improved sarcomeric alignment, and increased contraction in hiPSC-CM

To gain insights into molecular mechanisms underlying sparse-passaging-induced hiPSC-CM proliferation, we conducted phosphoproteomics analysis on day 30 hiPSC-CMs in sparse vs dense conditions (Figure S3). We used day 60 hiPSC-CMs, which do not respond to the expansion protocol (i.e., CHIR treatment and sparse passaging) as an internal control. Our analysis revealed significant changes in phosphorylation patterns, with 2858 upregulated phosphopeptides and 3047 downregulated phosphopeptides in sparsely cultured hiPSC-CMs compared to their densely cultured counterparts, observed exclusively in day 30 cells (Figure S3b,c). Among the top 50 significantly altered phosphopeptides, many were associated with biological processes related to sarcomere and junction organization, as indicated by Gene Ontology (GO) terms such as ‘Sarcomere Organization’ and ‘Cell-Cell Junction’ (Figure S3d). Additionally, a number of altered phosphopeptides were linked to the AKT, Wnt signaling pathway (Figure S3e). Notably, AKT was identified as one of the top upregulated phosphopeptides in day 30 sparsely plated hiPSC-CMs (Figure S3e).

First, based on GO term analysis, we explore the role of intercellular junction and sarcomere organization in regulating cell cycle in hiPSC-CMs. N-cadherin (CDH2) is an essential transmembrane protein at the cardiac adherens junctions that extracellularly forms homophilic adhesions with another N-cadherin from adjacent cardiomyocytes and intracellularly connects to actin-based cytoskeleton. N-cadherin plays a vital role in maintaining the structural integrity of cardiac tissues and facilitating the mechanical coupling of cardiomyocytes for the transmission of contractile forces. We first investigated the effect of cell-cell contact on the expression of a major adherens junction protein, N-cadherin. We found no significant change in CDH2 transcript or protein levels between the densely and sparsely cultured hiPSC-CMs (Figure 2a,b; CDH2, p = 0.790; N-cad, p = 0.139). However, N-cadherin was preferentially localized at the intercellular junction in densely plated cells, while in sparely plated cells it was diffusely distributed throughout the cell (Figure 2c,d). Distinct from N-cadherin, we found that sarcomeric alignment assessed by ɑ-actinin staining was reduced by 1.94-fold in the sparsely plated cells (Figure 2e–i). Furthermore, video motion-based contractility analysis revealed that sparsely plated hiPSC-CMs exhibited a 3.57-fold decrease in contraction velocity and a 2.49-fold decrease in contraction deformation distance (Figure 2j, Video S1–2).

Figure 2. Cell-contact removal associates with N-cadherin junction disruption, sarcomere disorganization, and reduced contractility in hiPSC-CMs.

(a-d) Effect of cell-cell contact on (a) CDH2 gene expression (Dense n=5, Sparse n=6), (b) N-cadherin protein expression (n=3), (c,d) the distribution of N-cadherin (n=3; Dense: 3 different locations per sample, Sparse: 5 different locations per sample), (e-i) sarcomeric alignment analysis (n=3, 11 different cells analyzed per sample), and (j) cardiomyocyte contractility (n=6). Data are presented as mean ± sem. **p<0.01, ***p<0.001, ****p<0.0001, ns = not significant (p>0.05) based on Mann-Whitney test (a,b,j), and rank-based linear mixed model (d,i). Scale bar = 50 μm (c), 100 μm (e), 25 μm (f). Cell line: SCVI-111

N-cadherin-based intercellular adherens junction formation and ɑ-actinin-based sarcomere organization participate in contact-mediated repression of proliferation

We then asked whether N-cad is required for contacs9t-mediated inhibition of cell cycle activity (Figure 3a). siRNA-mediated knockdown of N-cadherin (CDH2) led to a ~70% reduction in CDH2 transcript level and ~50% reduction in N-cadherin protein expression (Figure S4a,b). Three days after siRNA treatment of day 30 densely plated hiPSC-CMs, there was a 1.41-fold increase in the fraction of Ki67+ hiPSC-CMs (Figure 3b). Despite a significant stimulatory effect of CDH2-knockdown in the densely cultured hiPSC-CMs, the stimulation was absent when CDH2 was knocked down in sparsely plated hiPSC-CMs (Figure 3b). This CDH2-knockdown-mediated stimulation of hiPSC-CM cell cycle activity in dense culture condition was also observed in more mature day 60 hiPSC-CMs (Figure S4c,d). CDH2-siRNA treatment did not significantly increase the fraction of phospho-histone 3+ (phH3+) hiPSC-CMs as well as the number of cTnT+ nuclei (Figure 3c,d).

Figure 3. Contact inhibition of cardiomyocyte proliferation requires junctional localization of N-cadherin and sarcomere organization and it accompanies repression of Wnt signaling via junctional sequestration of Wnt nuclear effector β-catenin.

(a) Schematic of siRNA-mediated CDH2 and ACTN2 knockdown experiments. (b-d) Effect of siRNA-mediated CDH2 knock-down on day 30 hiPSC-CM proliferation. (e-g) Effect of siRNA-mediated sarcomeric ɑ-actinin knock down on hiPSC-CM proliferation. (h) Schematic of Wnt signaling activation analysis using β-catenin antibodies and TCF/LEF reporter system. (i,j) Effect of removing cell-cell contacts on β-catenin localization and the quantification. (k) Effects of Wnt activation and concomitant cell contact removal on TCF/LEF transcriptional activity. Data are presented as mean ± sem. *p<0.05, **p<0.01, ****p<0.0001, ns = not significant (p>0.05) based on linear mixed effects model (b-g), nested t-test (j), and Kruskal-Wallis test with Dunn’s multiple comparisons (k). Scale bar = 100 μm (b,c,e,f), 25 μm (i). Cell line: SCVI-111.

Next, we questioned whether inducing sarcomeric disassembly via ɑ-actinin knockdown is sufficient to drive cell cycle activation. ɑ-actinin (ACTN2) is a Z-disc protein responsible for crosslinking sarcomeric actin filaments. During cardiomyocyte mitosis, ɑ-actinin is one of the first sarcomeric proteins to be disassembled and its knockdown has shown to impair sarcomeric assembly and organization^32–35. Knockdown of ɑ-actinin resulted in a 92.8% reduction in ACTN2 gene expression (Figure S5a) and a 23.2% decrease in sarcomeric ⍺-actinin alignment (Figure S5b,c). ɑ-actinin knockdown led to a 1.61-fold increase in the fraction of Ki67+ cycling hiPSC-CMs only in densely cultured hiPSC-CMs, while cell cycle activity of sparsely plated hiPSC-CMs remained unchanged (Figure 3e). While ACTN2-knockdown significantly stimulated cell cycle activity of hiPSC-CMs, it did not increase the fraction of mitotic hiPSC-CMs indicated by phH3+ hiPSC-CMs (Figure 3f) nor the number of hiPSC-CMs (Figure 3g).

Because intercellular junction proteins are known to form complexes with β-catenin and regulate β-catenin bioavailability³⁶, we asked if cell-cell contact could directly influence canonical Wnt signaling activity (Figure 3h). Immunostaining revealed that β-catenin preferentially localizes at cell-cell junctions in densely cultured hiPSC-CMs. In contrast, it trans-locates into the nucleus in sparsely cultured hiPSC-CMs where the cell contacts are absent (Figure 3i,j). Given the role of β-catenin as a transcriptional activator of TCF/LEF, we further investigated TCF/LEF activity using a luciferase-based reporter system³⁷. We found that contact removal can induce a 2-fold increase in TCF/LEF activity (Figure 3k). In addition, GSK3β inhibition in densely cultured CMs using CHIR also resulted in a 2-fold increase in TCF/LEF activity (Figure 3k). Notably, GSK3β inhibition together with contact removal resulted in a 7-fold increase in TCF/LEF activity (Figure 3k), indicating synergy between Wnt activation and cell contact removal.

IGFBP2 is a novel pro-proliferative factor involved in contact-mediated repression of hiPSC-CM proliferation

Our data indicate that knockdown of CDH2 or ACTN2 alone can activate the cell cycle but is insufficient to drive complete cell cycle progression (Figure 3, S4, S5). Phosphoproteomics analysis suggests that CDH2 KD in day 30 densely plated hiPSC-CMs does not activate cell cycle regulators such as cyclin D and E as much to the extent observed with sparse passaging (Figure S6). To seek additional molecular mechanisms underlying sparse-passaging-mediated cardiomyocyte proliferation, we further hypothesized that sparsely plated hiPSC-CMs secrete distinct, pro-proliferative factors in a paracrine fashion, given the increase in cell cycle activity in sparsely plated hiPSC-CMs in cell-density-dependent manner (Figure 1). To examine this hypothesis, we collected conditioned media from hiPSC-CM cultures with different cell density culture conditions (i.e., 10,000 – 320,000 cells/cm²) and applied the conditioned media to a lowly proliferating cell culture condition (i.e., 20,000 cells/cm²) (Figure S7a). We found that the fraction of cycling cells showed an increasing trend and peaked at optimal sparse conditions (i.e., 40,000 – 80,000 cells/cm²; 1.7-fold increase compared to 20K C/M condition; P-value = 0.0152) (Figure S7b,c). Notably, the conditioned media from densely plated hiPSC-CMs, did not cause a significant increase in proliferation rates (Figure S7b,c). To identify potential pro-proliferative factor candidates, the conditioned media was collected from the sparsely cultured day 30 hiPSC-CMs (i.e., 40,000 cells/cm²) and screened for any enriched growth factors using an antibody-based human growth factor array (Figure 4a). Significantly upregulated growth factors were identified (log₂Fold-Change > 0.5) and validated for their pro-proliferative effect (Figure 4a). Screening assay identified three candidate proteins, IGFBP2, IGFBP6, and PDGF-AA, expression of which was 2.01, 1.79, and 1.51-fold higher compared to the unconditioned fresh media, respectively (Figure 4b, Figure S8). We then tested the individual or synergistic pro-proliferative effect of the 3 identified candidates on sparsely plated hiPSC-CMs (Figure 4c,d). We found that only IGFBP2 supplementation led to a 2.05-fold increase in cell cycle activity of hiPSC-CMs (Figure 4c,d). Next, we examined the effect of cell-cell contact on the expression of IGFBP2. We confirmed that cell contact removal leads to a 2- and 1.5-fold increase in IGFBP2 expression at transcript and protein levels, respectively (Figure 4e–g). Additionally, we found that sparsely cultured hiPSC-CMs increase IGFBP2 secretion (Figure S9).

Figure 4. Contact-inhibited hiPSC-CMs reduced expression of IGFBP2 and exogeneous treatment of IGFBP2 can stimulate cardiomyocyte proliferation in dosage-dependent manner.

(a) Schematic of identification and validation processes of enriched pro-proliferative secretory factors from sparsely cultured hiPSC-CMs. (b) Bar graph of the relative expression level of the factors secreted from sparsely plated, proliferating hiPSC-CMs. Data is normalized to the fresh expansion media. (n=3). (c,d) Quantification and the representative images demonstrating pro-proliferative effect of the identified factors on stimulating hiPSC-CM proliferation (n=3). (e,f,g) Effect of cell-cell contact on IGFBP2 expression (e) at transcript level (scRNAseq data), (f,g) protein level. (h,i) Dosage-dependent effect of IGFBP2 on proliferation of hiPSC-CMs (n=3). Data are presented as mean ± sem. *p<0.05, ****p<0.0001 based on Mann-Whitney test (e), unpaired t-test (g). (c, h) scale bar = 50 μm. Cell line: SCVI-111.

We then examined the requirement of IGFBP2 for proliferation of sparsely plated hiPSC-CMs. We found that siRNA-mediated IGFBP2 knockdown led to a 29.2% lower expansion of hiPSC-CMs compared to the control group after four days of culture in sparse condition, as indicated by the reduced number of TNNT2+ nuclei (Figure S10). Furthermore, conditioned media from IGFBP2 knockdown experiments confirmed that the proliferation-stimulatory effect of sparse passaging is mediated by secreted IGFBP2 (Figure S11).

We next investigated whether exogenous IGFBP2 supplementation is sufficient to overcome contact inhibition of proliferation in densely plated hiPSC-CMs. We found that IGFBP2 treatment can stimulate hiPSC-CM cell cycle activity in a dosage-dependent manner. Remarkably, densely cultured hiPSC-CMs treated with 3000 nM of IGFBP2, showed a 3.39-fold increase in the Ki67+/cTnT+ fraction compared to the untreated, densely plated counterpart (p = 0.02), exhibiting comparable cell cycle activity to sparsely cultured cells (Figure 4h,i).

Serial IGFBP2 treatment is sufficient to drive in situ expansion of hiPSC-CMs in 3D

To examine whether IGFBP2 treatment can continuously overcome the inhibitory effects of cell-cell contacts on hiPSC-CM proliferation and promote complete cell cycle progression and division in 3D, we treated day 20 hiPSC-CM-derived cardiac spheroids with IGFBP2 for additional twelve days in the presence of GSK3β inhibitor CHIR (Figure 5a). Serial imaging with bright-field microscopy revealed a continuous increase in area size in the groups treated with B27+CHIR (67.6 ± 4.6%) and B27+CHIR+IGFBP2 (216.7 ± 4.6%) compared with B27 treatment alone (Figure 5b–d). For precise analysis of spheroid growth and cell cycle activity, we applied iDISCO-clearing technology and light sheet microscopy³⁸ to allow volumetric measurement of whole, intact spheroids. After twelve days of treatment, three-dimensional imaging demonstrated a significant enhancement in spheroid volume. Specifically, the average spheroid volume increased by 4.67-fold when treated solely with CHIR (from 0.89 × 10⁷ μm³ in the control group to 4.14 × 10⁷ μm³ in the CHIR-only group, p = 4.13E-9). Moreover, when treated with B27+CHIR+IGFBP2, the spheroid volumes exhibited a remarkable 8.12-fold increase (reaching 7.2 × 10⁷ μm³, p = 7.20E-12) (Figure 5e,f). Consistently, quantification of nuclei number in spheroids, measured by propidium-iodide staining, showed a 2.79-fold increase in the B27+CHIR group (CTRL: 9.54 × 10⁴ nuclei per spheroid vs. B27+CHIR: 26.61 × 10⁴ nuclei per spheroid, p = 1.45E-9), and a 3.84-fold increase in B27+CHIR+IGFBP2 group (B27+CHIR+IGFBP2: 36.67 × 10⁴ nuclei per spheroid, p = 3.73E-11), indicating an increase in karyokinesis (Figure 5e,g). This was further corroborated by increased expression of cell cycle markers, Ki67 and phH3, which were significantly increased in the B27+CHIR+IGFBP2 group compared to the B27+CHIR only group and the B27 CM maintenance group (Figure 5h, Figure S12). We found that this IGFBP2-mediated 3D in situ expansion was reproducible in all three iPSC-line-derived cardiac spheroids tested (Figure S13–15).

DISCUSSION

In this study, we show that cell-cell contact induces cell cycle exit of hiPSC-CMs, with contact inhibition in part mediated by repressed Wnt/β-catenin signaling (Figure 1,3). N-cadherin-based adherens junction formation was a hallmark of densely plated hiPSC-CMs and was required for contact inhibition of proliferation in hiPSC-CMs (Figure 2,3). However, while N-cadherin knockdown was sufficient to activate the cell cycle, it was not sufficient to drive complete cell division (Figure 3, S4). This is in part due to insufficient activation of key cell cycle regulatory proteins such as cyclin D, cyclin E, which are critical for the G1-to-S phase transition (Figure S6).

During embryonic heart development, Wnt signaling activation is required for cardiomyocyte proliferation and myocardial wall thickening^15,16,39. Moreover, the canonical Wnt signaling nuclear effector β-catenin was shown to interact with YAP, Hippo nuclear effector, to function as transcriptional co-factors to their counterparts, TCF/LEF and TEAD respectively, to activate cardiomyocyte proliferation genes¹⁶. Previously, Wnt activation via the treatment of Wnt ligands or inhibitors of the β-catenin destruction complex was shown to increase cell cycle activity of embryonic cardiomyocytes or ESC/iPSC-derived cardiomyocytes^{21,22,39–41}. In this study, we expand the Wnt signaling regulators to N-cadherin, an important adherens junction protein in cardiomyocytes, as a cell-surface receptor regulating Wnt signaling outside in. A recent study reported that N-cadherin knockdown reduces the proliferation activity of neonatal mouse cardiomyocytes, which may initially seem to contradict our findings⁴². However, a key distinction exists between this study and ours. Tsai et al. demonstrated a non-canonical role of non-junctional N-cadherin as a binder and stabilizer of the pro-mitotic transcriptional regulator β-catenin, showing that N-cadherin remained primarily localized at cell junctions even in their N-cadherin haploinsufficiency model. In contrast, our study focuses on the role of junctional N-cadherin in contact inhibition mechanisms in densely plated hiPSC-CMs. These differences in biological context likely influence N-cadherin protein distribution and localization, which in turn may account for the differential effects on cardiomyocyte proliferation observed between the two studies.

We further demonstrate that additional growth signaling via IGFBP2 treatment synergizes with Wnt signaling to drive complete hiPSC-CM division (Figure 5, S10–15). While the role of IGFBP2 has not been well understood in a cardiac context to date, the pro-proliferative role of IGFBP2 has been reported in embryonic development, hematopoietic stem cells, and cancers^43–48. Our findings indicate that IGFBP2 functions independently of the YAP and Wnt/β-catenin signaling (Figure S16). Additionally, we show that AKT signaling plays a role in both sparse passaging and IGFBP2 treatment (Figure S3e, S17). Based on these findings, we propose the following model (Figure S18): N-cadherin junction disruption and sarcomeric disassembly serve as critical upstream events that enable cardiomyocytes to overcome contact inhibition and enter the cell cycle. However, successful completion of the cell cycle and cell division requires additional molecular signaling. IGFBP2 plays a key role in this process by promoting proliferation through AKT activation while functioning independently of the canonical Hippo-YAP and Wnt/β-catenin pathways. Detailed molecular mechanisms underlying IGFBP2-driven overcoming of contact inhibition of proliferation in cardiomyocytes warrant further studies.

There are a few limitations to this study. First, the pro-proliferative effect of cell-cell contact is evaluated only in vitro. Also, since iPSC-derived cardiomyocytes are shown to be equivalent to embryonic cardiomyocytes, an important remaining question is whether cell contact removal can induce cell cycle activation of postnatal or even adult cardiomyocytes. A recent clinical study reports that N-cadherin disorganization is associated with increased cell cycle activation even in postnatal cardiomyocytes⁴⁹, reinforcing the relevance of our in vitro findings to human biology. Lastly, future work needs to address the molecular mechanisms underlying IGFBP2-driven cell cycle stimulation and cell division.

In conclusion, this study demonstrates that cell-cell contact plays an important role in regulating proliferation of hiPSC-CMs. We show that exogeneous IGFBP2 supplementation can drive complete hiPSC-CM division, overcoming the contact-mediated inhibition of hiPSC-CM proliferation that was limited by CHIR treatment alone. We believe the findings from this study provide new insights into cardiomyocyte cell cycle regulation, which can be harnessed to develop new therapies for myocardial regeneration.

Novelty and Significance.

What Is Known?

• Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) are a promising cell source for cardiac tissue engineering

• Terminally differentiated cardiomyocytes exhibit limited proliferative capacity

• Activation of pro-proliferative signaling pathways such as Wnt alone has proven insufficient to expand hiPSC-CMs

What New Information Does This Article Contribute?

• Proliferation of hiPSC-CMs is cell density-dependent

• Cell-cell contact inhibits the proliferation of hiPSC-CMs and suppresses mitogenic Wnt signaling

• IGFBP2 can drive hiPSC-CM division and is secreted by sparsely but not densely plated hiPSC-CMs in a paracrine fashion

• Exogeneous supplementation of IGFBP2 restores proliferation and drives hiPSC-CM division in 3D environments despite cell-cell contacts

hiPSC-CMs are widely recognized as a valuable resource for cardiac tissue engineering, yet their limited proliferative capacity upon terminal differentiation remains a major barrier. Although activation of pro-proliferative pathways such as Wnt has been attempted, these strategies alone have proven insufficient to drive robust expansion. This study shows that contact inhibition restricts hiPSC-CM proliferation via two independent mechanisms: suppression of mitogenic Wnt signaling and modulation of the IGFBP2–AKT axis. Importantly, we demonstrate that exogenous IGFBP2 supplementation can overcome these inhibitory cues, enabling cardiomyocyte division even in the presence of cell–cell contacts within a 3D environment. These findings highlight a dual requirement for promoting proliferation: not only activating growth pathways (“accelerator”) but also relieving inhibitory signals (“brake”). This paradigm offers new mechanistic insights with direct implications for advancing cardiac tissue engineering and regenerative medicine.