Innate Mechanisms of Heart Regeneration

Abstract

Heart regeneration is a remarkable process whereby regrowth of damaged cardiac tissue rehabilitates organ anatomy and function. Unfortunately, the human heart is highly resistant to regeneration, which creates a shortage of cardiomyocytes in the wake of ischemic injury, and explains, in part, why coronary artery disease remains a leading cause of death worldwide. Luckily, a detailed blueprint for achieving therapeutic heart regeneration already exists in nature because several lower vertebrate species successfully regenerate amputated or damaged heart muscle through robust cardiomyocyte proliferation. A growing number of species are being interrogated for cardiac regenerative potential, and several commonalities have emerged between those animals showing high or low innate capabilities. In this review, we provide a historical perspective on the field, discuss how regenerative potential is influenced by cardiomyocyte properties, mitogenic signals, and chromatin accessibility, and highlight unanswered questions under active investigation. Ultimately, delineating why heart regeneration occurs preferentially in some organisms, but not in others, will uncover novel therapeutic inroads for achieving cardiac restoration in humans.

A myocardial infarction (MI) occurs when a downstream segment of cardiac muscle becomes deprived of oxygen due to coronary artery obstruction (Laflamme and Murry 2011). In the absence of early intervention, billions of cardiomyocytes are lost during these ischemic events, which are a leading cause of death worldwide. In those individuals fortunate enough to survive the acute phase, any damaged muscle is replaced with a noncontractile scar. In the most severe cases, congestive heart failure often ensues due to pathologic remodeling of the left ventricular wall. As such, the root cause of MI-mediated heart failure and mortality is a significant shortage of cardiomyocytes that cannot be resolved naturally. Three major therapeutic strategies are actively being pursued to reconstitute damaged cardiac muscle (Tzahor and Poss 2017). These include transplantation of human iPS-derived cardiomyocytes or bioengineered muscle patches, reprogramming of resident fibroblasts to cardiomyocytes, and stimulation of endogenous cardiomyocyte proliferation, the last of which will be the subject of this review.

Although the heart has long been considered one of the least regenerative organs in the human body, any propensity for myocardial regeneration, even on a microscopic scale, could be exploited and bolstered for therapeutic benefit. While there is currently no evidence to support the existence of a cardiac stem cell population in the adult mammalian heart (Berlo and Molkentin 2014; Berlo et al. 2014), mounting data suggest that cardiomyocyte turnover continues into adulthood, albeit at a very low level (Senyo et al. 2013; Ali et al. 2014; Kimura et al. 2015; Yester and Kühn 2017). During the first decade of life, cardiomyocytes in the human heart remain largely diploid (Bergmann et al. 2015), exhibit cell-cycle activity, and express markers of cytokinesis, the latter of which become undetectable after 20 yr of age (Mollova et al. 2013). In addition, a radiocarbon-based cardiomyocyte birth dating study documented cardiomyocyte turnover at ∼1% per year in the uninjured hearts of 20 yr olds, which gradually decreased to ∼0.3% per year by 75 yr of age (Bergmann et al. 2009, 2015). Importantly, these data demonstrate that the molecular and cellular machinery needed for cardiomyocyte cell division remain minimally but measurably intact throughout life and could be targeted clinically.

An intricate roadmap for achieving restorative levels of cardiomyocyte proliferation already exists in nature. Cold-blooded vertebrates including zebrafish, newts, and axolotls achieve near-perfect organ renewal through robust cardiomyocyte proliferation following injury (Oberpriller and Oberpriller 1971, 1974; Flink 2002; Poss et al. 2002; Witman et al. 2011; Piatkowski et al. 2013). Uncovering the variables that account for the vastly different magnitudes of cardiac regenerative capacity between highly regenerative and poorly regenerative species is arguably the most pressing goal driving the field. While regenerating the human heart will require replenishing a variety of essential cell types, including cardiomyocytes, endocardium, epicardium, and fibroblasts as well as the coronary and lymphatic vasculature, we will focus this review on the fundamental mechanisms regulating regenerative cardiomyocyte proliferation.

A BRIEF HISTORY

Partial myocardial regrowth was first observed following mechanical crushing of the ventricle in the common frog (Rumyantsev 1966). Not a decade later, heart regeneration was investigated in the adult newt where substantial organ renewal was also seen (Oberpriller and Oberpriller 1974). In these amphibian models, heart regeneration was associated with mitotic figures within wound edge cardiomyocytes that contained disorganized myofibers (Oberpriller and Oberpriller 1971; Rumyantsev 1972, 1973, 1977). This pioneering work established the propensity of the amphibian heart to regenerate and linked cardiomyocyte proliferation and sarcomere disassembly, both hallmarks of dedifferentiated cardiomyocytes, to natural organ renewal. While the field was founded on these discoveries, the catalyst for its re-emergence was a landmark report in 2002 demonstrating that adult zebrafish, which are genetically tractable, regenerate their hearts following ventricular injury (Poss et al. 2002), the magnitude of which approaches 100% (Bertozzi et al. 2021). Since this discovery, the zebrafish model of cardiac regeneration has served as the foundation of many research programs worldwide, which have uncovered much of what is currently known about innate mechanisms guiding heart regrowth.

Following this initial finding, the quest to uncover the cellular origins of the regenerated zebrafish myocardium began. Although an early study suggested that new cardiomyocytes derive from undifferentiated progenitors (Lepilina et al. 2006), unequivocal evidence for their myocardial origin came from the application of Cre/loxP lineage-tracing technology. Specifically, two laboratories documented that preexisting cardiac myosin light chain 2 (cmlc2)-expressing cardiomyocytes give rise to new myocardium (Jopling et al. 2010; Kikuchi et al. 2010). More recently, the majority if not all cardiomyocytes were shown to be capable of renewing lost muscle (Gupta and Poss 2012; Sánchez-Iranzo et al. 2018).

The original conclusion from amphibian models that preexisting cardiomyocytes dedifferentiate following injury has gained further support in zebrafish. Specifically, transmission electron micrographs (TEMs) and immunofluorescence studies have documented sarcomere disassembly in wound edge cardiomyocytes (Jopling et al. 2010; Kikuchi et al. 2010; Ben-Yair et al. 2019). Accordingly, differential gene expression analysis identified down-regulation of transcripts encoding sarcomere and cytoskeletal components in the same population (Wu et al. 2016; Ben-Yair et al. 2019). Moreover, immunofluorescence and in situ hybridization staining showed reexpression of an embryonic isoform of myosin and transcripts encoding developmental cardiac transcription factors like gata4: nkx2.5, tbx5a, and hand2 (Lepilina et al. 2006; Sallin et al. 2015; Sánchez-Iranzo et al. 2018).

These collective discoveries have led to a consensus in the field that cardiac injury in highly regenerative species induces preexisting wound edge cardiomyocytes to dedifferentiate and proliferate to create new heart muscle. Therefore, uncovering fundamental mechanisms promoting this injury-induced response holds the promise of promoting human heart restoration instead of scarring.

WHAT ACCOUNTS FOR DIFFERENCES IN INNATE CARDIAC REGENERATIVE CAPACITY?

Since the discovery of heart regeneration in zebrafish, alternative animal models have emerged that provide the opportunity to compare and contrast cardiac adaptations that influence regenerative outcomes (Potts et al. 2021). Examples of new models include the neonatal mouse (Porrello et al. 2011), the Mexican surface and cave-dwelling fish, Astyanax mexicanus (Stockdale et al. 2018), and the Japanese medaka, Oryzias latipes (Ito et al. 2014).

In a landmark publication, Porrello et al. (2011) demonstrated that P1 neonatal mouse hearts fully regenerate following injury through robust cardiomyocyte proliferation, while hearts from P7 pups permanently scar. Importantly, the scarring response in P7 hearts is associated with failed cardiomyocyte proliferation, which creates an opportunity to identify dynamic variables regulating differential cardiomyocyte proliferation in the same species at different ages.

More recently, heart regeneration was assessed in two morphotypes of the single species A. mexicanus (Stockdale et al. 2018), which diverged when a subset of the surface-dwelling population became trapped in underground caves lacking sunlight (Gross et al. 2015). Much like the disparate regenerative outcomes of P1 and P7 neonatal mouse hearts, hearts from the surface dwellers were found to regenerate while those from the Pachón cave scar. In contrast to the P7 neonatal mouse heart, Pachón hearts exhibit cardiomyocyte cell-cycle reentry, suggesting that other morphotype-specific barriers exist that negatively impact regenerative capacity. One such barrier could be impaired cytokinesis; however, this hypothesis has yet to be tested.

The medaka, which is a teleost like the zebrafish, has also proven to be an interesting player in the field due to its surprising lack of cardiac regenerative capacity (Ito et al. 2014; Lai et al. 2017). Following amputation injury, cardiomyocytes in the medaka heart fail to proliferate, which was attributed to an unfavorable immune response (Lai et al. 2017). Specifically, the number of macrophages infiltrating the wound is significantly less in medaka hearts, and their arrival is delayed when compared to zebrafish hearts. Moreover, neutrophil clearance from the wound is also hindered in medaka. However, the link between the differential immune responses and cardiomyocyte proliferation remains unknown.

What features are common between hearts that regenerate and those that scar? Because cardiomyocyte proliferative capacity is a defining feature of regenerating hearts, we will describe hypothesized or well-established determinants of differential cardiomyocyte cell division that fall into three categories: (1) the cellular features of cardiomyocytes, (2) the availability of mitogenic signals, and (3) the chromatin landscape in cardiomyocytes. We will outline known determinants in each category and highlight unanswered questions for further investigation.

CELLULAR FEATURES OF CARDIOMYOCYTES

During postnatal life, mammalian cardiomyocytes acquire new cellular properties while simultaneously losing their proliferative capacity in a process termed maturation (Fig. 1). The temporal overlap of these events is suggestive of causality. Specific features of mature cardiomyocytes that are inversely correlated with regenerative potential include polyploidy, fatty acid oxidative (FAO) metabolism, reduced telomere length, and increased sarcomere complexity (Vivien et al. 2016). How these features affect regenerative capacity, how they are interconnected, and how they are controlled genetically represent active areas of investigation that have already yielded interesting insights.

Figure 1.

Cellular features of cardiomyocytes that influence proliferation. Schematic diagram summarizing intrinsic properties of cardiomyocytes that correlate with their proliferative capacity. The poorly regenerative adult mammalian heart is composed of cardiomyocytes with limited proliferative potential that have increased nucleation/ploidy, metabolism, and sarcomere complexity as well as shortened telomeres (top). By contrast, highly regenerative zebrafish and neonatal mouse hearts are composed of cardiomyocytes with robust proliferative potential that have decreased nucleation/ploidy, metabolism, and sarcomere complexity, as well as longer telomeres (bottom).

Nucleation and Ploidy

Diploid cells contain two complete sets of chromosomes, while polyploid cells contain greater than two sets (Orr-Weaver 2015). Highly regenerative hearts, such as those in adult zebrafish, newts, and P1 neonatal mice, are almost exclusively composed of mononucleated diploid cardiomyocytes (Oberpriller et al. 1988; Soonpaa et al. 1996; Wills et al. 2008; González-Rosa et al. 2018), while poorly regenerative hearts, such as those found in adult mice and humans, consist predominantly of polyploid cardiomyocytes containing one (humans) or two (mice) nuclei (Brodsky et al. 1980, 1985, 1994; Li et al. 1996; Soonpaa et al. 1996; Mollova et al. 2013; Alkass et al. 2015). Although the transition from a diploid to a polyploid state coincides temporally with the loss of cardiac regenerative potential in neonatal mice (Porrello et al. 2011), a causal relationship between these two traits was only recently established.

González-Rosa et al. (2018) engineered adult zebrafish with hearts composed of varying percentages of indelibly labeled polyploid cardiomyocytes to test whether increased cardiomyocyte ploidy would negatively impact cardiac regenerative outcomes (González-Rosa et al. 2018). Cardiomyocyte polyploidy was induced by transient expression of a dominant-negative version of the essential cytokinesis component Ect2 (dnEct2) during hyperplastic heart growth. Importantly, ectopic dnEct2 expression was silenced prior to injury so that elevated cardiomyocyte ploidy would be the only modified variable in an otherwise regenerative environment. Injured hearts composed of ∼50% polyploid cardiomyocytes exhibited significant decreases in cardiomyocyte proliferation and permanent scarring. When the percentage of polyploid cardiomyocytes was reduced to ∼25%, injured hearts were able to regenerate, but the new myocardium derived preferentially from the diploid fraction. Consistent with the zebrafish study, knocking out Ect2 in cardiomyocytes during mouse embryonic development caused premature cardiomyocyte polyploidization and impaired neonatal regeneration (Windmueller et al. 2020). In the same study, transcriptional profiling efforts revealed that mononucleated cardiomyocytes are transcriptionally distinct from binucleated ones, which down-regulate E2F target genes that otherwise promote DNA synthesis.

Patterson et al. reported that hearts in adult inbred mouse strains comprise variable percentages of mononucleated diploid cardiomyocytes ranging from 2% to 10% (Patterson et al. 2017). Those with higher percentages display increased functional recovery following experimental MI, which correlates with increased cardiomyocyte cell-cycle activity and markers of cell division, suggesting that diploid cardiomyocytes have a higher propensity for regeneration. A genome-wide association study identified deleterious variants in TNNI3 interacting kinase (Tnni3k) linked to increased diploid composition, suggesting that Tnni3k activity promotes cardiomyocyte polyploidization. Accordingly, a loss-of-function Tnni3k allele decreases the percentage of polyploid cardiomyocytes in mice (Patterson et al. 2017; Gan et al. 2019), while constitutive Tnni3k overexpression in zebrafish increases polyploid percentages (Patterson et al. 2017). Although the mechanism by which Tnni3k influences cardiomyocyte ploidy is still under investigation, it remains the only known gene with naturally occurring variants in a single species that influences this trait (Gan et al. 2019).

By analyzing multiple species, Hirose et al. established that thyroid hormone levels also impact cardiomyocyte ploidy (Hirose et al. 2019; Amram et al. 2020). Specifically, the authors quantified the percentage of diploid cardiomyocytes in hearts of more than 41 vertebrate species and found that their abundance inversely correlates with the transition from ectothermy (cold-bloodedness) to endothermy (warm-bloodedness). This evolutionary shift toward endothermy coincides with increased thyroid hormone levels, which increases cardiomyocyte polyploidy and decreases regenerative capacity. The authors suggest that this reduction in cardiac regenerative capacity might have been sacrificed in adult mammals for the acquisition of endothermy.

Although there is much speculation about what competitive advantages might have been gained during evolution by postnatal cardiomyocyte polyploidization (Gan et al. 2020), an equally interesting question is how cardiomyocyte polyploidy is induced from a molecular perspective. A recent study demonstrated that the nuclear lamina filament protein Lamin B2 (Lmnb2), which is essential for nuclear envelope breakdown, promotes the diploid state in mammalian cardiomyocytes (Han et al. 2020). Lmnb2 protein levels are high during fetal life, but rapidly decline after birth. Loss of Lmnb2 increases the percentage of polyploid cardiomyocytes in P1 neonatal mouse hearts and decreases their regenerative capacity, while ectopic expression of Lmnb2 reduces the percentage of polyploid cells at P21. Inactivating Lmnb2 in human iPS-derived cardiomyocytes and in primary cardiomyocytes harvested from infant human hearts also increases the percentage of polyploid nuclei. Interestingly, Han et al. also found that zebrafish cardiomyocytes, which remain diploid and proliferative throughout life, naturally maintain high levels of Lmnb2.

While one path to mammalian heart regeneration might be through stimulation of diploid cardiomyocyte division (Kühn et al. 2007; Bersell et al. 2009; Senyo et al. 2013), another path might capitalize on a recently reported time-lapse imaging study of cultured rat binucleated cardiomyocytes, which captured their ability to divide and complete cytokinesis (Leone and Engel 2019a,b). This in vitro phenomenon has also been observed in binucleated cardiomyocytes from newts (Bettencourt-Dias et al. 2003) and mice (D'Uva et al. 2015; Wang et al. 2017). However, a new study using a transgenic reporter system to highlight cardiomyocytes expressing aurora kinase B, a marker of cytokinesis when localized to the contractile ring, found no evidence of reporter activity following MI in adult mice (Fu et al. 2020). Coupling this finding with the lack of macroscopic regeneration in adult mammalian hearts suggests that most polyploid cardiomyocytes do not naturally divide in vivo. With this said, their potential for division indicates that a better understanding of this process could be beneficial for therapeutic intervention. How efficiently binucleated cardiomyocytes can be stimulated to divide in vivo is unknown. Moreover, whether polyploid mononucleated cardiomyocytes, which comprise the majority of the human heart, can be coaxed to divide in vitro or in vivo also remains to be determined.

Metabolism

Prior studies in pig and rabbit hearts have shown that embryonic/neonatal cardiomyocytes rely more heavily on glycolysis, while postnatal/adult cardiomyocytes meet their higher energy demands by primarily utilizing FAO (Ascuitto et al. 1989; Lopaschuk and Spafford 1990; Lopaschuk et al. 1991, 1992). In mice, the transition from glycolysis to FAO occurs during the first week of life (Kimura et al. 2017) when the nutrient supply changes from glucose and lactate supplied by the placenta to fatty acids found in the mother's milk (Maroli and Braun 2021). This metabolic transition has also been observed in human cardiac organoids where increased FAO is associated with cardiomyocyte maturation and cell-cycle arrest (Mills et al. 2017).

In addition to a nutrient supply change during the first week of life, environmental oxygen, which is essential for the final step of the electron transport chain, also increases and supports augmented FAO during cardiomyocyte maturation (Puente et al. 2014). As a result of increased mitochondrial respiration, reactive oxygen species (ROS) accumulate and lead to nuclear and mitochondrial DNA damage, which are associated with cell-cycle arrest and polyploidy (Pohjoismäki et al. 2012; Puente et al. 2014). Potential exceptions to this metabolic switch are cardiomyocytes residing in hypoxic regions of the adult mouse heart that are small in size, diploid, and proliferative (Kimura et al. 2015). However, it remains unclear whether cellular respiration occurs predominantly by glycolysis in this adult population.

Recent evidence suggests that polyploid and diploid cardiomyocytes isolated from adult mouse hearts differentially utilize glycolysis and FAO to generate ATP (Windmueller et al. 2020). Specifically, the polyploid fraction is enriched in transcripts encoding components essential for FAO. Moreover, electron micrographs revealed an accumulation of stored glucose in the form of glycogen granules specifically in the polyploid population. This observation is consistent with polyploid cardiomyocytes reserving excess glucose as glycogen because they are primarily consuming fatty acids for energy production.

Although polyploidization is tightly associated with FAO metabolism in mammals, zebrafish cardiomyocytes, which are almost exclusively diploid throughout life, also use FAO (Honkoop et al. 2019). Interestingly, in an injury setting, highly proliferative border zone cardiomyocytes up-regulate genes required for glycolysis, whereas those located in a remote region of the heart express genes required for FAO. Chemical inhibition of glycolysis following injury impairs cardiomyocyte proliferation, revealing a strict requirement for this localized metabolic switch during heart regeneration. To determine whether forcing this metabolic switch would be sufficient to stimulate cardiomyocyte proliferation in adult mice, investigators utilized extreme hypoxia, genetic manipulation, and pharmacologic strategies to limit FAO and enhance glycolytic metabolism following MI (Nakada et al. 2017; Cardoso et al. 2020; Bae et al. 2021). Remarkably, these interventions induce cardiomyocyte proliferation, suppress pathologic remodeling, and improve cardiac function. Why this metabolic switch is crucial for cardiomyocyte proliferation remains unclear. Informative insights will likely stem from advanced metabolic profiling and follow-up studies to identify glycolytic metabolites that are crucial for cardiomyocyte division. Furthermore, whether experimental induction of glycolysis or inhibition of FAO can induce human polyploid cardiomyocytes to divide remains to be determined.

Telomere Length

The length of the protective caps on the ends of chromosomes known as telomeres also inversely correlates with cardiac regenerative capacity (Aix et al. 2018). Shortly after birth in humans, the enzyme responsible for maintaining telomere length, telomerase, is naturally down-regulated (Wright et al. 1996). However, telomerase expression in zebrafish is maintained throughout life and even up-regulated following injury (Bednarek et al. 2015). Bednarek et al. found that zebrafish lacking telomerase grow to adulthood but fail to regenerate their hearts following cryoinjury. In these animals, cardiomyocytes contain shortened telomeres, elevated DNA damage, and never enter the cell cycle. Following MI in adult mice, ectopic expression of telomerase limits scar formation and improves ventricular function (Bär et al. 2014). In these telomerase-expressing murine hearts, cardiomyocytes exhibit longer telomeres and elevated DNA synthesis, suggesting that augmentation of cardiomyocyte proliferation contributes to improved outcomes. Overall, the absence of telomerase in the adult human heart represents another possible barrier to natural organ renewal.

Sarcomere Complexity

During postnatal maturation, mammalian cardiomyocytes must dramatically increase their contractility to meet the increased hemodynamic demands of extra-uterine life. To that end, sarcomeres grow in width and length. They also become more highly ordered and structurally complex by switching out fetal components for adult-specific isoforms (Guo and Pu 2020). While these features are readily evident in transmission electron micrographs, the molecular mechanisms driving isoform switching and increased sarcomere size and complexity remain relatively unknown. Because successful cardiomyocyte proliferation is strongly associated with sarcomere disassembly, the increasingly intricate architecture that emerges during cardiomyocyte maturation presents a potential physical barrier to cell division and regeneration. However, it remains unclear whether genetic strategies that decrease sarcomere complexity, without compromising function, can be used to study this variable in the context of heart regeneration.

SIGNALING PATHWAYS THAT REGULATE CARDIOMYOCYTE PROLIFERATION

Several sources of mitogenic signals that induce cardiomyocyte proliferation have been identified in the heart (Fig. 2) including the endocardium and coronary vessels (Fernandez et al. 2018), epicardium and epicardial-derived tissues (Cao and Poss 2018), nerves (Garikipati et al. 2015; Mahmoud et al. 2015), and immune cells such as macrophages (Han et al. 2015; Lai et al. 2019) and regulatory T cells (Hui et al. 2017). In addition, mechanical cues created by the changing composition of the extracellular matrix (ECM) shortly after birth are also known to alter cardiomyocyte proliferative capacity (Bassat et al. 2017; Wu et al. 2020). Here, we highlight select signaling programs that influence cardiomyocyte proliferation following insult.

Figure 2.

Noncardiomyocyte sources of mitogenic signals. Several sources of signals that influence cardiomyocyte proliferation have been identified in the injured heart including the endocardium and coronary vessels, epicardium, and epicardial-derived tissues, nerves, immune cells including macrophages and T-regulatory cells, and the extracellular matrix.

The Hippo (Wang et al. 2018) and Neuregulin (NRG1)-ErbB (Tzahor and Poss 2017) pathways represent arguably the most well-characterized signaling programs known to instruct cardiomyocyte proliferation during heart regeneration (Fig. 3). Specifically, Hippo pathway inhibition and NRG1-ErbB pathway activation both stimulate cardiomyocyte dedifferentiation and proliferation to boost regenerative capacity.

Figure 3.

Components of the Hippo-Yap and NRG1-Erbb signaling pathways that regulate cardiomyocyte proliferation. (A) With the DNA-binding transcription factor TEAD, Yap functions in cardiomyocyte nuclei as a cotranscriptional activator of downstream target genes that drive proliferation. Translocation of YAP into the nucleus is prevented by at least two mechanisms: (1) Yap is bound to the transmembrane dystrophin glycoprotein complex (DGC) and sequestered in the cytoplasm, and (2) Yap is phosphorylated and degraded in the cytoplasm when the Hippo signaling kinase cascade is active. The proteoglycan agrin is a fetal-specific extracellular matrix (ECM) component that binds to DGC. This interaction prevents Yap from binding to DGC, which allows for its nuclear translocation. Hippo signaling is activated by mechanosensing through increased ECM rigidity and cytoskeletal tension, which leads to Yap phosphorylation and degradation. (B) NRG1 binds to the tyrosine kinase receptors, ERBB4 and ERBB2, which lead to phosphorylation of ERK/MAPK and PI3K/AKT and inhibition of GSK3β/β-catenin. ERK/MAPK phosphorylation results in Yap phosphorylation on serine residues distinct from those modified by the Hippo pathway. This event promotes Yap nuclear translocation and activation of a pro-proliferative program. (MST1/2) Mammalian STE20-like protein kinase 1/2, (SAV1) Salvador homolog 1, (Lats1/2) large tumor suppressor homolog 1/2, (MOB1) Monopolar Spindle One Binder.

Hippo signaling activates a kinase cascade that impedes cardiomyocyte proliferation by phosphorylating the promitogenic cotranscriptional effector Yap on specific serine residues, which results in its degradation (Wang et al. 2018). During fetal life, when cardiomyocyte proliferation is high, Hippo signaling is off. However, Hippo becomes activated by P7 when cardiomyocytes fail cytokinesis and become polyploid. Genetic inhibition of Hippo signaling, which heightens nuclear Yap, extends the regenerative window in neonatal mice (Heallen et al. 2013), promotes adult heart regeneration following coronary artery ligation (Heallen et al. 2013), and restores pump function in mice with systolic heart failure (Leach et al. 2017). Moreover, inhibition of Hippo through overexpression of Yap5SA, a gain-of-function version of the protein that cannot be phosphorylated, drives uncontrolled cardiomyocyte proliferation in adult mouse hearts by preferentially expanding the diploid fraction (Monroe et al. 2019).

Activation of the Hippo pathway is known to occur through mechanosensing of increased ECM rigidity and cytoskeletal tension (Bassat et al. 2017; Wang et al. 2018; Ooki et al. 2019). After P7 when the heart loses regenerative potential, the transmembrane dystrophin glycoprotein complex (DGC) binds directly to Yap in cardiomyocytes and inhibits its translocation to the nucleus (Morikawa et al. 2017). Prior to P7 when the heart is capable of regenerating, the ECM proteoglycan agrin is highly expressed and binds to DGC, which disrupts its intracellular interaction with Yap (Bassat et al. 2017). As such, the agrin–DGC interaction during fetal and early neonatal life allows Yap to translocate to the nucleus, where it drives cardiomyocyte proliferation. Genetic deletion of agrin inhibits heart regeneration in neonates, while administering agrin to adult mice or pigs after MI reduces scar formation, which is attributed in part to enhanced cardiomyocyte proliferation (Bassat et al. 2017; Baehr et al. 2020). These promising observations suggest that modulation of the agrin–DGC–Yap signaling axis holds genuine therapeutic promise.

NRG1 signaling through the ErbB2 and 4 receptor tyrosine kinases is necessary and sufficient for inducing cardiomyocyte proliferation (Tzahor and Poss 2017). Although an initial study documented a mitogenic effect of recombinant NRG1 on rat neonatal ventricular cardiomyocytes in vitro (Zhao et al. 1998), it would take ∼10 yr before an in vivo role was established (Bersell et al. 2009). Bersell et al. reported that injection of NRG1 into adult mice a week following MI leads to improved cardiac function and increased cardiomyocyte cell-cycle activity, although this latter finding has been contested (Reuter et al. 2014). In the highly regenerative zebrafish heart, NRG1 expression is naturally induced in epicardial-derived perivascular cells following injury (Gemberling et al. 2015). Inhibition of ErbB2 in mice and zebrafish disrupts cardiomyocyte proliferation in response to insult, whereas constitutive Nrg1-ErbB2 overexpression enhances it resulting in cardiomegaly (D'Uva et al. 2015; Gemberling et al. 2015; Polizzotti et al. 2015). Interestingly, the Nrg1-ErbB2 proliferative response appears to rely on nuclear translocation of Yap, which is triggered by cytoskeletal remodeling (Aharonov et al. 2020). In addition, induction of the proliferative phenotype by Nrg1-ErbB2 in both zebrafish and mice is coupled with metabolic reprogramming from FAO to glycolysis (Honkoop et al. 2019). Taken together, these data suggest that supplying activated versions of ErbB2 to cardiomyocytes post-MI could improve clinical outcomes.

The endocardium supplies a rich array of mitogens that stimulate cardiomyocyte proliferation following insult (Fig. 4). Retinoic acid (RA) was among the first injury-induced endocardial-derived signals to be identified in zebrafish (Lepilina et al. 2006; Kikuchi et al. 2011). Genetic inhibition of RA signaling dampens cardiomyocyte proliferation and blunts the regenerative response in zebrafish (Kikuchi et al. 2011). Interestingly, endocardial RA production fails to become induced in adult mouse hearts after MI (Kikuchi et al. 2011). Why RA signaling is not naturally activated by cardiac injury in endocardial cells of mammals remains unknown, as does whether forced expression of endocardial RA signaling is sufficient to instruct cardiomyocyte division.

Figure 4.

Endocardial mitogens that stimulate cardiomyocyte proliferation following injury. Retinaldehyde dehydrogenase 2 (Raldh2), an enzyme that oxidizes retinaldehyde to make retinoic acid (RA), is expressed in the endocardium. RA signaling promotes cardiomyocyte proliferation after injury in zebrafish. Notch signaling is activated in the endocardium after amputation injury where it induces expression of the secreted Wnt antagonists Wif1 and Notum1b, which temper Wnt signaling to promote cardiomyocyte proliferation. After cryoinjury, Notch signaling down-regulates expression of the Serpine1, a secreted factor that inhibits cardiomyocyte proliferation. Secretion of IGF2 from endocardial cells induces phosphorylation of the insulin receptor (INSR) in cardiomyocytes to promote proliferation. Secretion of VEGFA promotes cardiomyocyte proliferation through increased angiogenesis.

Insulin-like growth factor 2 (IGF2) expression is also stimulated in endocardial cells during zebrafish heart regeneration (Choi et al. 2013; Huang et al. 2013a). Chemical or genetic inhibition of IGF signaling dampens cardiomyocyte proliferation and induces scarring (Choi et al. 2013; Huang et al. 2013a). A conserved role for endocardial IGF signaling has also been observed in neonatal mice (Shen et al. 2020). Specifically, endocardial IGF2 expression becomes induced by amputation injury and, in turn, stimulates phosphorylation of the insulin receptor (INSR) in cardiomyocytes, which is required for myocardial proliferation and heart regeneration. However, this sequence of events does not occur in injured hearts of adult mice, even though endocardial IGF2 induction is conserved. These observations underscore a potential requirement for proliferation-competent cardiomyocytes to respond to IGF2 in the infarcted region.

Transcripts encoding the VEGFA ligand, Vegfaa, appear in the endocardium of zebrafish hearts following many forms of cardiac injury (Marín-Juez et al. 2016; Karra et al. 2018). Ubiquitous expression of a dominant-negative form of Vegfaa blocks coronary artery revascularization of the damaged region, which in turn leads to cardiomyocyte proliferation deficiencies and scarring (Marín-Juez et al. 2016). By contrast, ubiquitous overexpression of wild-type Vegfaa stimulates new coronary vessel growth and cardiomyocyte hyperplasia even in the absence of injury (Karra et al. 2018). It is likely that increased angiogenesis is primarily responsible for Vegfaa-induced cardiomyocyte proliferation. Intramyocardial injection of a synthetic, modified RNA (modRNA) encoding Vegfa into infarcted hearts of mice improves cardiac function and long-term survival by increasing angiogenesis (Zangi et al. 2013), suggesting that stimulation of Vegf signaling might be useful clinically for enhancing regeneration.

In addition to secreted mitogens, receptors for the Notch signaling pathway also become induced in the endocardium following cardiac injury in zebrafish (Raya et al. 2003; Zhao et al. 2014, 2019; Münch et al. 2017; MacGrogan et al. 2018). Global suppression of Notch activity dampens cardiomyocyte proliferation and leads to scarring after either apex amputation or cryoinjury (Zhao et al. 2014; Münch et al. 2017). In the amputation model, endocardial-specific Notch inhibition suppresses myocardial proliferation by tempering Wnt signaling (Zhao et al. 2019). In the cryoinjury setting where inflammation is more pronounced, decreased cardiomyocyte proliferation following ubiquitous Notch inhibition was attributed in part to sustained expression of Serpine1 (Münch et al. 2017). In cultured neonatal rat cardiomyocytes, constitutive Notch activation boosts proliferative capacity through transcriptional activation of Notch-responsive genes, which retain an accessible chromatin conformation at this developmental stage (Felician et al. 2014). By contrast, constitutive Notch activation in adult mouse hearts fails to induce a regenerative response following MI, which is attributed to a repressive epigenetic signature at Notch target gene promoters. These findings demonstrate that while activation of pro-regenerative signaling programs, such as the Notch pathway, are critical for inducing cardiomyocyte proliferation, a permissive chromatin landscape is equally vital.

Hormonal control of cardiomyocyte proliferation has recently gained attention for its role in regulating heart regeneration. Throughout evolution, elevations in thyroid hormone (TH) levels, which accompanied the transition from ectothermy to endothermy, also coincided with an increased prevalence of polyploid cardiomyocytes (Hirose et al. 2019), which decreases regenerative capacity (Patterson et al. 2017; González-Rosa et al. 2018). During the first week of murine postnatal life, a rise in TH levels correlates temporally with cardiomyocyte maturation and cell-cycle exit (Hirose et al. 2019). To test for causality, Hirose et al. manipulated TH signaling experimentally in mice and zebrafish. Inhibition of TH signaling decreased cardiomyocyte polyploidization and enhanced proliferation at P14 (Hirose et al. 2019). Inactivation of TH signaling in adult mouse hearts following injury also increased cardiomyocyte proliferation, improved function, and decreased fibrosis. Last, exposing adult zebrafish to exogenous TH moderately increased the prevalence of binucleated cardiomyocytes, reduced cardiomyocyte proliferation, and supplanted regeneration with scarring. The percentage of polyploid cardiomyocytes induced by TH in this study (∼6%) was well below the threshold (25%–50%) required to induce scarring in González-Rosa et al. (2018), suggesting that other variables contribute to TH-mediated regenerative failures. Interestingly, in the frog species Xenopus laevis, larval-to-adult metamorphosis is also controlled by rising levels of TH (Beck et al. 2009). Moreover, a link between high TH and reduced cardiac regenerative capacity was recently established in this species (Marshall et al. 2019). Specifically, tadpoles with low TH efficiently regenerate their hearts, while adults with high TH scar instead. In contrast to the observations made by Hirose et al., defects in cardiomyocyte proliferation were not evident in adult frog hearts after injury, but alterations in ECM deposition and remodeling were reported that ostensibly promote scar formation.

Emerging data suggest that endocrine control of heart regeneration extends beyond TH to include other hormones such as vitamin D (Han et al. 2019; Cutie et al. 2020), oxytocin (Jankowski et al. 2012), glucocorticoids (Huang et al. 2013b; Cutie et al. 2020), and estrogen (Xu et al. 2020). In a small molecule screen carried out in zebrafish embryos, vitamin D signaling was found to augment cardiomyocyte proliferation during heart development (Han et al. 2019). In adult zebrafish, chemical activation of vitamin D signaling increased cardiomyocyte proliferation following injury, whereas inducible expression of a dominant-negative isoform of the vitamin D receptor, Vdra, dampened cardiomyocyte proliferation and blocked heart regeneration. This stimulatory role for vitamin D relies on Erbb2 signaling. In the mouse, discordant observations were reported following stimulation by vitamin D signaling in cultured neonatal cardiomyocytes with one group reporting increased proliferation (Han et al. 2019) while the other observed decreased cell division (Cutie et al. 2020). The reasons for this discrepancy remain unclear.

CHROMATIN DYNAMICS THAT FACILITATE CARDIOMYOCYTE PROLIFERATION

A permissive three-dimensional chromatin confirmation is essential for cardiomyocytes to respond to proliferation cues that promote heart regeneration. During cardiomyocyte maturation, chromatin architecture changes drastically in that embryonic-specific regions condense, while those associated with differentiated muscle properties become accessible (Quaife-Ryan et al. 2016). In general, changes to the chromatin landscape are implemented by pioneer transcription factors, ATP-dependent chromatin remodeling complexes, histone code alterations, and DNA methylation (Guo and Morris 2017; Chiarella et al. 2020). All of these mechanisms are employed by cardiomyocytes to change their cellular properties during postnatal maturation or regeneration (Fig. 5). In settings where robust regeneration does not naturally occur, reprogramming the chromatin landscape to a more fetal-like confirmation will be an essential feature of any therapeutic strategy. As such, additional research in this understudied field is required.

Figure 5.

Chromatin dynamics that facilitate cardiomyocyte proliferation. (A) The AP-1 transcriptional heterodimer composed of Fos and Jun components mediates chromatin remodeling either alone or in conjunction with the YAP–TEAD complex. (B) The SWI/SNF ATP-dependent chromatin remodeling complex recruits the Dnmt3ab DNA methylase to the cdkn1c promoter to induce transcriptional repression through hypermethylation. (C) Alterations to the epigenome by histone replacement with histone variants or posttranslational modifications modulate gene expression, which influences cardiomyocyte proliferation.

To begin addressing where the chromatin landscape changes during postnatal cardiomyocyte maturation, investigators employed assay for transposase-accessible chromatin coupled with high-throughput sequencing (ATAC-seq) to compare chromatin accessibility between proliferative (P1) and nonproliferative (P14 and P56) cardiomyocytes isolated from mouse hearts. This approach uncovered progressive chromatin remodeling at specific loci during cardiomyocyte maturation (Quaife-Ryan et al. 2017). Specifically, genomic regions regulating cell-cycle progression and ECM deposition are accessible at P1 but progressively condense by P14 and P56. By contrast, promoters associated with FAO metabolism, muscle development, and muscle contraction, which are closed at P1, become increasingly open by P14 and P56, reflecting the observed changes in gene expression that are associated with maturation.

Using a similar approach, investigators characterized chromatin accessibility dynamics in border zone and remote cardiomyocytes following experimental MI in adult mice (Duijvenboden et al. 2019). In border zone cardiomyocytes, thousands of muscle-specific regions become newly condensed, consistent with cardiomyocyte dedifferentiation. In addition, enhancers enriched in AP-1 motifs, which are bound by heterodimeric transcriptional complexes of Fos and Jun family members, become newly accessible. Moreover, nuclear localization of Jun was observed specifically in border zone cardiomyocytes, highlighting regional activation of AP-1 following MI.

Similar findings were obtained by studying proliferative cardiomyocytes in the context of natural heart regeneration. In zebrafish, ATAC-seq was used to compare chromatin states between proliferative and nonproliferative cardiomyocytes following injury (Beisaw et al. 2020). While a large number of genomic regions become differentially remodeled, the majority that became newly accessible in proliferative cardiomyocytes contain AP-1 motifs (Fig. 5A). Accordingly, fos and jun are naturally up-regulated in wound edge cardiomyocytes following injury. Of the 573 newly accessible AP-1 motifs, only 22 motifs have orthologous counterparts that behave similarly in mouse border zone cardiomyocytes. This observation suggests that largely dissimilar AP-1-containing genomic regions become accessible in cardiomyocytes of regenerative versus nonregenerative organisms following injury. In a separate study, AP-1 motifs were identified as a conserved feature of injury-responsive enhancers in zebrafish and a related teleost, the African killifish (Wang et al. 2020). Although some of these AP-1-containing enhancers are also conserved in humans, Wang et al. (2020) speculate that they were repurposed for biologic functions other than regeneration, which might explain why some organisms have high regenerative faculties, while others do not.

To learn whether AP-1 transcription factors serve a critical function during zebrafish heart regeneration, Beisaw et al. (2020) blocked all AP-1 transcription factor activity by expressing a dominant-negative Fos protein (specifically in cardiomyocytes following injury). Loss of AP-1 function dampened cardiomyocyte proliferation and resulted in permanent scarring. Moreover, deleting a highly conserved AP-1-containing enhancer in killifish also derailed natural cardiac regeneration (Wang et al. 2020). In the absence of cardiac injury, ectopic expression of Jun is sufficient to open chromatin at roughly one-third of the regions naturally remodeled in regenerating zebrafish cardiomyocytes (Beisaw et al. 2020). This latter observation is consistent with AP-1 having pioneer factor activity where engagement of the AP-1 complex with nucleosomal DNA initiates unwinding of heterochromatin (Iwafuchi-Doi and Zaret 2014).

In adult mouse cardiomyocytes, ectopic expression of Yap5SA, which completely silences Hippo signaling, can also initiate chromatin remodeling to achieve a more fetal-like architecture (Monroe et al. 2019). On a cellular level, Yap5SA expression leads to extensive cardiomyocyte hyperplasia and heart failure. Newly accessible chromatin in Yap5SA-expressing cardiomyocytes was found to be highly enriched for AP-1 and TEAD motifs, the latter of which are essential for Yap-mediated transcription. The opening of genomic regions containing TEAD motifs was also observed in border zone cardiomyocytes after experimental MI in mice (Duijvenboden et al. 2019). Much like overexpression of Jun (Beisaw et al. 2020), Yap5SA also appears to have pioneer activity because its forced expression reprograms chromatin to a more embryonic state through a TEAD-Yap-AP-1 network.

Nucleosomes are the fundamental unit of chromatin that consist of ∼150 base pairs of DNA wrapped around a histone octomer (Gangaraju and Bartholomew 2007). ATP-dependent chromatin remodeling complexes hydrolyze ATP to slide nucleosomes along DNA to alter chromatin accessibility and gene expression (Hota and Bruneau 2016). The SWI/SNF complex contains greater than ten components of which BRG1 is a central catalytic ATPase. Following amputation injury in zebrafish, Brg1 and several other SWI/SNF subunits including Baf60c and Baf180 are induced in multiple cell types including cardiomyocytes (Xiao et al. 2016). Inhibition of Brg1 activity using a dominant-negative isoform dampened cardiomyocyte proliferation and induced scarring following injury (Fig. 5B). Mechanistically, Brg1 represses expression of the cyclin-dependent kinase (CDK) inhibitor Cdkn1c by recruiting the DNA methyltransferase Dnmt3ab to its promoter, which induces hypermethylation-mediated silencing. The SWI/SNF component Baf60c is also up-regulated in the regenerating neonatal mouse heart where it is required for cardiomyocyte proliferation and regeneration (Nakamura et al. 2016).

Modulation of the histone code can also reconfigure chromatin through the addition of histone variants that replace a core histone or through changes in posttranslational modifications including acetyl (Ac) or methyl (me) groups on lysine (K) residues of histone tails (Fig. 5C). Transcriptionally active chromatin is thought to be enriched in the histone 3.3 (H3.3) variant, which is deposited de novo after nucleosomes are displaced (Henikoff and Smith 2015). To identify changes in chromatin accessibility, Goldman et al. (2017) compared H3.3 occupancy between cardiomyocytes from regenerating and uninjured zebrafish hearts. Promoter regions with increased H3.3 occupancy were found to be associated with enhanced gene expression during heart regeneration.

Following cardiac injury, the epigenetic modifications H3K27Ac, H3K4me3, and H3K27me3, which mark active enhancers, accessible chromatin, and condensed chromatin, respectively, are dynamically regulated and modulate transcriptional output (Goldman et al. 2017; Ben-Yair et al. 2019; Wang et al. 2019). This was revealed by comparing H3K27ac dynamics and gene expression profiles in P1 and P8 neonatal mouse hearts following sham surgery or experimental MI (Wang et al. 2019). Interestingly, the up-regulated gene sets acquiring new H3K27ac marks were distinct between the regenerative (P1) and nonregenerative (P8) states. Specifically, P1 hearts gained H3K27ac peaks over loci involved in vascular development indicative of regenerative angiogenesis, while P8 hearts accumulated peaks over genes associated with fibrotic remodeling indicative of a pronounced immune response and scarring.

In a separate study, integrated differential expression and chromatin immunoprecipitation analyses of wound edge cardiomyocytes from injured zebrafish hearts revealed enrichment of the H3K4me3 activating mark on transcriptionally up-regulated loci that encode proteins involved in cell cycle, DNA replication, and ECM remodeling (Ben-Yair et al. 2019). In addition, accumulation of H3K27me3 repressive marks were also observed on transcriptionally down-regulated loci-encoding genes associated with sarcomere formation and cytoskeletal organization. Importantly, genetic inhibition of H3K27me3 deposition, through cardiomyocyte-specific expression of a dominant-negative histone, compromised heart regeneration and stimulated scarring. On a cellular level, cardiomyocytes were found to initiate the cell cycle but failed cytokinesis. They were also hindered in wound invasion. These data provide evidence that modulation of the histone code is required for heart regeneration in zebrafish. Moreover, this work highlights epigenetic reprogramming as an essential component of any pro-regenerative therapies.

CONCLUDING REMARKS

Restoring the proliferative capacity of adult cardiomyocytes following MI in humans is the holy grail of cardiac regenerative medicine. From comparisons of model organisms that are either highly or poorly regenerative, new concepts have emerged to explain how cellular traits, signaling pathways, and chromatin structure might be advantageous or detrimental for cardiomyocyte proliferation. Looking forward, additional insights into heart regeneration will be achieved by answering the following questions. Specifically, how is cardiomyocyte polyploidy established in mammals, and how do increases in the diploid population affect heart function in adults? Are the additional sets of chromosomes in polyploid cardiomyocytes exact replicas of those in diploid cells? Why is a metabolic switch from FAO to glycolysis necessary to promote proliferation? How does the composition of the ECM alter interactions with the cardiomyocyte cytoskeleton to influence cell division? Of particular promise is controlled activation of Yap, which appears to override all barriers to cell division and induce a robust myocardial proliferative response in adult mice (Monroe et al. 2019). Ultimately, ongoing and future studies into innate mechanisms of heart regeneration will improve our understanding of this remarkable biologic process and inform the development of novel therapeutic approaches for regenerating the human heart.