Regenerative loss in the animal kingdom as viewed from the mouse digit tip and heart

Abstract

The myriad regenerative abilities across the animal kingdom have fascinated us for centuries. Recent advances in developmental, molecular, and cellular biology have allowed us to unearth a surprising diversity of mechanisms through which these processes occur. Developing an all-encompassing theory of animal regeneration has thus proved a complex endeavor. In this chapter, we frame the evolution and loss of animal regeneration within the broad developmental constraints that may physiologically inhibit regenerative ability across animal phylogeny. We then examine the mouse as a model of regeneration loss, specifically the experimental systems of the digit tip and heart. We discuss the digit tip and heart as a positionally-limited system of regeneration and a temporally-limited system of regeneration, respectively. We delve into the physiological processes involved in both forms of regeneration, and how each phase of the healing and regenerative process may be affected by various molecular signals, systemic changes, or microenvironmental cues. Lastly, we also discuss the various approaches and interventions used to induce or improve the regenerative response in both contexts, and the implications they have for our understanding regenerative ability more broadly.

Graphical Abstract

INTRODUCTION

The historical fascination with animal regeneration is centuries-old (Goss, 1969; Dinsmore, 1991). The first formal experimental studies date back to the 18th century, including René-Antoine Réaumur’s studies of the spontaneous regeneration of crayfish claws (Réaumur, 1712) and Abraham Tremblay’s discovery of hydra whole-body regeneration (Tremblay, 1744). From these studies arose a growing awareness of the extensive capacity of animal bodies to regrow and remake themselves, and the myriad diverse animal and organ systems it takes place in. This fascination persists into modern day, with recent advances in developmental, molecular, and cellular biology allowing us to uncover the mechanisms through which these processes occur (Stocum, 2012). The closer we examine these phenomena, however, the more different they appear. The regeneration of the same structure, even between two relatively closely related animals, can be a heterogeneous process. For example, salamander limb regeneration involves the dedifferentiation of postmitotic, multinucleated muscle cells, which re-enter the cell cycle and contribute to the new limb (Wang and Simon, 2016). In contrast, no muscle dedifferentiation is observed in axolotls, as muscle in the regenerated limb is derived from a resident population of satellite stem cells (Fei et al., 2017).

For this reason, forming a unifying theory for animal regeneration is complex—why does it occur in some animals but not others? To what extent does regeneration share or co-opt mechanisms from physiological and developmental processes? Might there be a way to enable non-regenerative species to regenerate? These questions remain relevant even as we gain more fine-grained insights into regeneration across animal phylogeny.

Regeneration in an evolutionary-developmental framework

In 1893, August Weismann put forth the earliest attempt to create a conceptual framework for the emergence of animal regeneration. According to Weismann, regeneration could be conceived of as an adaptive trait—that there was a causal relationship between the frequency or likelihood of damage and the ability to regenerate (Weissman, 1893).

One contemporary critic of Weismann’s theory was Thomas Hunt Morgan, who spent much of his early career studying a diverse array of regenerating animals. In one of his classical experiments, Morgan observed that hermit crab hind limbs, often hidden away and protected under their shell, regenerated just as frequently and readily as their less well-protected forelimbs (Morgan, 1902). He would synthesize his own findings and theories in a 1901 book, Regeneration, in which he proposed that regeneration was an intrinsic property of an organism, and therefore should be studied like a developmental process (Morgan, 1901).

Much of Morgan’s theoretical and experimental framework is still relevant to the study of regeneration today. Similar to a developmental process, Morgan considered regeneration to refer broadly to a heterogenous set of organized spatiotemporal processes, including the renewal of an organ, the replacement of entire body parts or even asexual reproduction (Sinigaglia et al., 2022). The broadness of this definition may appear to confuse the study of regeneration, but casting a relatively wide net may help draw parallels between different levels of biological similarity—specific processes, molecular pathways or morphological structure.

Elaborating on Morgan’s concept of regeneration as a developmental process, contemporary work in evolutionary-developmental biology has created frameworks through which to draw these parallels in a more consistent and rigorous manner. First, the concept of modularity, which divides complex phenomena such as regeneration into distinct processes, capturing interactions within and between multiple levels of organismal organization (Wagner et al., 2007). This includes regulatory phenomena at the molecular level, or structural entities such as cells or tissues that interact in order to produce a complex structure (Atchley and Hall, 1991). In a regeneration context, limb regeneration can be separated into distinct but recognizably similar stages across taxa. In both the arthropod Parhyale and the vertebrate axolotl, amputation of the limb leads to wound healing and closure, followed by the formation of a blastema with the activation of Pax7-expressing satellite progenitor cells, which then undergo proliferation, growth, and morphogenesis to form a new limb (Konstantinides and Averof, 2014; Sandoval-Guzmán et al., 2014) (Figure 1).

Figure 1:

Comparison of limb regeneration in the axolotl and the arthropod Parhyale showing distinct and similar stages. Limb amputation is followed by wound healing and the formation of the blastema. Pax7+ activated satellite progenitor cells are recruited to the wound site and proliferate, differentiating to undergo growth and morphogenesis to form the regenerated limb.

While the arthropod and vertebrate limb are not homologous in the typical sense—i.e., sharing a common evolutionary origin— they are analogous structures in that they perform similar functions. Evolutionary-developmental biology proposes an extended definition of homology that also includes the regulatory networks underlying modular processes, even if their ultimate anatomical or physiological outcomes differ (Gilbert and Bolker, 2001). In the case of Parhyale, the discovery of Pax7 satellite cells was interpreted as a case of deep homology with vertebrates, suggesting that this pool of progenitor cells share an evolutionary origin (Konstantinides and Averof, 2014). A similar population of satellite cells was also discovered recently in adult Drosophila flight muscles, further suggesting that these progenitors have conserved regulatory mechanisms across all arthropods (Chaturvedi et al., 2017).

Given that Pax7 often labels mesodermal precursors during development across animal phylogeny, it is possible that these arthropod and vertebrate satellite cell populations arose independently. Thus, there may have been multiple independent co-option events of the same ancient gene regulatory kernel into muscle stem cell lineages. However, interpreting these cellular, molecular, and regulatory similarities as convergent occurrences may also highlight generalizable properties that facilitate or hinder regeneration. From a developmental perspective, we could view the presence of Pax7+ cells as an embryonic cellular state persisting through adulthood—another potentially generalizable framework through which to view regenerative capacity (Lai and Aboobaker, 2018). Whether homologous or convergent, these comparisons open salient avenues of exploration within each module of the regenerative process. Are there, for example, shared physiological or regulatory changes that allow them to remain quiescent or re-access developmental pathways through adulthood? Are they re-activated with the same signals? Do they have a similar developmental origin? The heterogeneity of regeneration across animals, combined with the difficulty in ascertaining shared evolutionary origin, highlights the need for comparisons that examine all modules of the regenerative process, and investigate how they interact with each other across multiple levels of organization.

Developmental constraints and decline in regenerative ability

The evolutionary-developmental framework of regeneration allows us to return to our initial question—why some animals are able to undergo extensive regeneration while others cannot—with greater clarity about the comparisons we are making. It also highlights how we might be able to investigate an understudied aspect of the field: why regeneration does not occur. Examining the loss or restriction of regenerative ability across animal phylogeny reveals a widespread distribution, with occurrences in most phyla (Bely and Nyberg, 2010). Comparisons between relatively closely related regenerating and non-regenerating members of a clade may be one avenue through which the mechanisms of regeneration loss may be better understood. For example, while planarians are thought of as capable of extensive regeneration, this is not true of all species, with the distribution of regenerative ability among planarians suggesting multiple gains and losses (Maden, 2018). However, in a planarian species (Dendrocoelum lacteum) that is unable to regenerate its head, regeneration was restored by simply inhibiting Wnt/β-catenin signaling (Liu et al., 2013). In regenerating planarians, the head and tail are typically specified by a gradient of Wnt gene expression along the anteroposterior axis (Gurley et al., 2008)—the successful restoration of head regeneration thus indicated that despite the loss of the ability, the underlying molecular machinery was still present and functional in this non-regenerative species. The failure of a process to occur is not usually perceived to be as relevant as success, but if regeneration is an organismal property, understanding the specific properties of failure is at least half of the biological picture. This understanding proves important to biomedical interests as well: if we want to improve human regenerative ability, better insight into failure might produce more targeted avenues for success.

The question, then, is if this principle might apply to other contexts, both regarding the regeneration of other structures as well as within different groups of related animals. Mammals are commonly thought to have generally lost the ability to regrow complex tissues, but certain examples of mammalian tissue regeneration question this assumption, and demonstrate the insight gained from broader phylogenetic comparisons (Maden, 2018). Examples of mammalian regeneration are often seen as exceptional—for instance, unlike mice, the spiny mouse Acomys is seen as highly regenerative due to its ability to regenerate large sections of its skin, which it can shed to escape predators (Seifert et al., 2012a). The spiny mouse also carries out blastema-mediated regeneration of ear tissues after an ear punch injury, including cartilage, dermis, hair, skeletal muscle, and sebaceous glands (Seifert et al., 2012a). But this ability, not shared by mice, was first observed in rabbits (Vorontsova and Liosner, 1960; Joseph and Dyson, 1966; Goss and Grimes, 1972), as well as an array of other mammals such as chinchillas, cows, pigs (Williams-Boyce and Daniel, 1986), pikas and bats (Goss, 1987). Placed in this context, the existence of dormant and potentially still functional molecular machinery that might allow the mouse to regenerate ear tissue seems highly probable—particularly when certain mutant mouse strains such as the Murphy’s Roth Large (MRL) mouse are able to do so (Clark et al., 1998).

Following the modular processes within a non-regenerative response might assist us in assessing why, how, and when regeneration is no longer feasible. Examining different examples of non-regenerative outcomes in different phyla shows that the regenerative response fails at different parts of the process (Bely and Nyberg, 2009). Studies of annelids, which typically can regenerate body segments, show that non-regenerative annelids wound heal after segment amputation without even forming a blastema (Bely, 1999). In contrast, the African clawed frog Xenopus laevis, which loses limb regenerative ability post-metamorphosis, forms a proliferative blastema which abortively forms a cartilage spike that lacks both the morphology and function of the original limb (Simon and Tanaka, 2013). These examples raise two pertinent questions— firstly, in comparing different non-regenerative outcomes, what factors cause these regenerative responses to fail at those specific parts of the process? And secondly, in comparing non-regenerative and regenerative outcomes, what allows the wound healing process to lead to successful regeneration in one case but not the other? To invoke another concept from evolutionary-developmental biology, we could hypothesize that the wound healing and regeneration modules are developmentally constrained in non-regenerative cases, preventing the processes governing one module from interacting with the other (Richardson and Chipman, 2003).

Metabolism and energy expenditure

Energetic allocation and tradeoff can also arise indirectly. Another trend is the generally negative correlation between endothermy and regeneration particularly in vertebrates (Cutie and Huang, 2021). Thermogenic regulation is energetically costly, and poikilothermic animals such as zebrafish and salamanders tend to possess higher regenerative ability compared to endothermic mammals. Likewise, neonatal mammals are not as capable of efficient thermoregulation compared to their adult counterparts and tend to display increased regenerative potential (Tourneaux et al., 2009). Vertebrates with mostly diploid cardiomyocytes (a proxy for cardiac regeneration) and high cardiac regeneration ability are known to have metabolisms an order of magnitude lower than their endothermic mammalian counterparts (Makarieva et al., 2008). A phylogenetic association between diploid cardiomyocytes and standard metabolic rates (factoring out organism size) in 41 species of animals showed that diploid cardiomyocytes negatively correlated with metabolic rates, and that in mammals, body temperature also negatively correlates with the percentage of myocardium composed of diploid cardiomyocytes (Hirose et al., 2019).

Cellular plasticity

A key factor in regenerative success is having an abundant cellular source from which regenerated tissue originates from. Thus, cellular plasticity within an organism correlates with regenerative capacity. Methods of yielding new cells can include drawing on a pool of progenitor cells, dedifferentiation of differentiated cells to a progenitor state, or transdifferentiation of one cell type into another (Jopling et al., 2011). Diploblastic animals, or animals with bodies derived entirely from ectoderm and endoderm, tend to possess reservoirs of stem cells from which new tissue can be formed (Elchaninov et al., 2021). For example, the incredibly regenerative Hydra has epithelial stem cells and interstitial stem cells, with epithelial stem cells contributing to epidermally-derived cell types such as neurons and secretory cells, and interstitial stem cells contributing to all other remaining cell types (Wittlieb et al., 2006; Hemmrich et al., 2012). The earliest branching triploblastic animal, with tissues derived from ectoderm, endoderm, and mesoderm, are planaria. While similarly possessing whole-body regenerative ability, planarians have a single pool of pluripotent progenitors, the neoblasts, which proliferate and form blastemas to restore missing body parts (Reddien et al., 2005). In vertebrates, where regeneration is typically limited to specific parts of the body, blastema cells appear relatively fate-restricted. In both axolotl limb regeneration and mouse digit tip regeneration, de-differentiated blastema cells only form tissues deriving from the same lineage as the tissue of their origin (Kragl et al., 2009; Lehoczky et al., 2011). For instance, regenerated bone in the mouse digit tip derives from pre-amputation osteoblasts (Lehoczky et al., 2011), and in the axolotl limb, dermal cells will form cartilage but not muscle as limb dermis and cartilage originate from lateral plate mesoderm (Kragl et al., 2009) It has thus been hypothesized that the evolution of more specialized tissue types within a single organism resulted in more complex regulation of the cell cycle and differentiated cells’ ability to re-enter it (Galliot and Ghila, 2010), thereby reducing their plasticity and decreasing regenerative capacity.

Age

One association that relates to cellular plasticity is the consideration of age and regenerative ability. Broadly speaking, regenerative potential generally appears to be higher during early life: embryos, larval stages and juvenile animals tend to have greater ability to regenerate than their adult or aged counterparts (Seifert and Voss, 2013). Fetal mammals can regenerate skin, leading to scar-free healing, while adults typically form fibrotic scars when wound healing (Bullard et al., 2003), and even regenerative animals such as zebrafish show a decline in the ability to regenerate pectoral fins in older animals (Itou et al., 2012). This decline has been associated with cell cycle regulation, as age-related stressors such as DNA damage and oxidative stress may impair cellular ability to proliferate and self-renew (Signer and Morrison, 2013). Dysregulated tissue homeostasis in major signaling pathways due to aging can also lead to increased chronic inflammation and fibrosis, which are generally not conducive for regeneration (Singh et al., 2011; Beggs et al., 2004). While aging and dysregulation are linked, it does not appear to be an inevitable consequence of aging, as restoring homeostatic signaling seems to ameliorate detrimental effects. For example, exogenous testosterone treatment prevented sarcopenia in aged mice by suppressing age-specific increases in JNK signaling and decreases in Akt signaling, which led to decreased oxidative stress and cell death and increased muscle hypertrophy respectively (Kovacheva et al., 2010).

Mechanisms of age-related loss of plasticity and proliferation may occur through changing epigenetic state of cells with age, as changes in DNA methylation and histone markers influenced genes related to proliferation, pluripotency, and differentiation state (Sousounis et al., 2014). They may also be affected by systemic factors, as demonstrated in parabiosis experiments—serum from younger mice stimulates muscle and neuronal renewal in older mice, while serum from older mice conversely decreased neurogenesis and myogenesis (Conboy et al., 2005; Villeda et al., 2011). Recreating a “younger” systemic microenvironment also encouraged oligodendrocyte renewal and promoted axonal remyelination in older mice (Ruckh et al., 2012). Thus, while aging correlates with the loss of regenerative ability, not all changes are due to the inherent properties of aging tissue. Changing molecular signals and the surrounding systemic milieu also influence how and if cells respond to regenerative cues.

Immune system

The immune system is obviously instrumental in both wound healing and regeneration. A wide body of work across animals and organs implicates the immune system in regenerative responses, explicitly in creating a permissive environment for regeneration to take place (Godwin et al., 2013; Fukuzawa et al., 2009; Aurora et al., 2014; Thomas and Puleo, 2011).

It is also a factor that seems to broadly correlate with regenerative ability on a phylogenetic level. The emergence of a complex immune system, and specifically a specialized adaptive immune response generally appears to have a negative correlation with regenerative ability. Frogs and mice, for example, possess rapid and diverse adaptive immune responses upon injury (Mescher and Neff, 2005). In comparison, salamanders lack some adaptive immune responses, such as reactivity to soluble antigens (Charlemagne, 1979) and an acute rejection of xenografts (Godwin and Rosenthal, 2014), relying primarily on an innate immune response. Other studies point to the maturity of the immune system as a relevant factor. Xenopus tadpoles lose regenerative ability with age (Dent, 1962), and mice only maintain heart regeneration in early neonatal life (Porello et al., 2011) and scarless skin regeneration in embryonic stages (Ferguson and O’Kane, 2004). It is hypothesized that the mature immune system mounts a strong inflammatory response that is not conducive to regeneration, as regenerative failure in Xenopus tadpoles can be partially restored with anti-inflammatory molecules (King et al., 2012)

Immunomodulatory strategies for cardiac regeneration have thus focused on blunting the adaptive immune response (Sattler et al., 2017). CD4+ T cells, which secrete various cytokines to stimulate immune responses, have proved of particular interest—regulatory T-cells (Tregs) specifically are known to suppress inflammatory responses and promote macrophage polarization toward an M2-like anti-inflammatory phenotype (Weirather et al., 2014). Comparisons of CD4+ T-cells in neonatal and adult mice show that neonatal T-cells default to becoming Tregs, though this tendency diminishes within the first 2 weeks of life (Wang et al., 2010). Tregs have also been shown to be essential in zebrafish and neonatal mouse regeneration, due to their immunomodulatory effect and secretion of the CM mitogen Neuregulin (Nrg) (Hui et al., 2017; Li et al., 2019). Human fetal immune systems also generate more Tregs than adult immune systems, which suggests that Treg function might be a conserved therapeutic target to improve heart regeneration.

However, due to the complexity of immune responses, it is unlikely that a single factor is the determinant of regenerative success. There are important caveats to these broad trends. For instance, the spiny mouse Acomys, able to regenerate epidermal tissue, has a longer and more marked adaptive immune response than the standard lab mouse (Gawriluk et al., 2020). An inflammatory response has also proved essential in other contexts such as the mouse digit tip (Simkin et al., 2017a) and amphibian limb (Godwin et al., 2013). The apparent pleiotropy of immune-related factors in different contexts points to the fact that a “pro-regenerative” immune response is a multiphasic process that is sensitive to the nature but also the duration of each phase (Aztekin and Storer, 2022). Thus, studying how immune responses are regulated will be crucial to understanding how the immune response lays the groundwork for regeneration.

The mouse as a model of regeneration loss

These physiological correlates demonstrate how regeneration is not simply an intrinsic property that a given organism or body part does or does not possess. Rather, a regenerative response is the result of cellular interactions and states, the local milieu of microenvironmental and molecular cues, and the influence of organismal-level physiology. Understanding how each of these factors function and intersect to help or hinder a regenerative response is crucial to identify key points in the process that play a significant role in determining the ultimate outcome. These critical junctures could provide us with optimal targets for intervention—how best to stimulate a regenerative response where one would not typically occur. To do so, we use the mouse as a model of regeneration loss. Mice possess limited but reliable regenerative ability, in the very distalmost portion of the digit tip (Neufeld and Zhao, 1995), and in the heart as neonates within the first week of life (Porello et al., 2011). Removing anything beyond the digit tip, or damaging the heart in an older mouse instead results in regenerative failure, which provides us with the opportunity to investigate why and how these regenerative responses are so specifically spatially or temporally limited. In the rest of this chapter, I will discuss how these regenerative processes occur endogenously, the physiological and molecular factors that play a significant role in their success or failure, and the attempts at inducing or improving further regenerative responses in each context.

Mouse digit tip

The mouse digit tip is an important model in the study of mammalian regeneration. It possesses some endogenous regeneration ability throughout the entire life of the animal, as mice are able to regenerate the distal tip of the terminal phalanx from neonatal stages to adulthood (P3). This ability mirrors documented cases in clinical literature where young children (often ages 6 or younger) have been able to regenerate severed fingertips (Illingworth, 1974), highlighting its potential relevance to improving human regenerative ability.

Mouse digit anatomy

The mouse digit consists of three phalanges, referred to as P1, P2, and P3 proximally to distally. The regenerative portion of the digit is found in P3, a structurally distinct bone that has a wide, flared base with a large bone marrow cavity that tapers into a pointed distal tip (Seifert and Muneoka, 2018). The distal end of P3 is encased by the nail organ, while the proximal end articulates with P2 to form the P2/P3 joint. Distal amputations, or amputations that removed up to the first third of P3 successfully regenerate, although other components such as the bone marrow, fat pad, or nail matrix should also remain undamaged to ensure regeneration (Dolan et al., 2018)(Figure 2). In contrast, amputation that removes over 50% of P3 generally results in wound healing without regeneration (Neufeld and Zhao, 1995).

Figure 2.

(A) Second and terminal phalanges of the mouse digit. Amputations made within the distalmost third of the terminal phalanx (P3) is regenerative (area in green). Amputations that remove over 60% of P3 are non-regenerative (area in red). (B) A regenerative amputation through P3 (dotted line) at 0 days post amputation (0 dpa) and 28 dpa. By 28 dpa, the regenerated digit tip recapitulates the tapered morphology of P3, though it is structurally distinct as it is more porous than the original bone. (C) A non-regenerative amputation through P3 (dotted line). The post amputation response concludes at wound healing, with no regenerative response seen at 28 dpa. (D) Steps of mouse digit regeneration. Amputation is followed by a rapid immune response and histolysis of the bone prior to wound re-epithelialization. Blastema cells are recruited to the wound site and proliferate, then re-differentiate into specific cell types. Revascularization and reinnervation of the regenerating digit occur as it undergoes morphogenesis. Finally, the regenerated digit tip is formed through intramembranous ossification, preserving the shape but not the structure of the original digit tip.

Digit regeneration as blastema-mediated

Digit regeneration in mice is unusual in two respects: firstly, that it is a mammalian example of blastema-mediated regeneration, and second, that it occurs in a markedly different manner from digit development. Unlike other repair processes such as healing a fracture or muscle injury, digit regeneration involves the formation of a transient blastema, a mass of proliferative, pluripotent cells that differentiate and undergo morphogenesis to pattern and regenerate the lost structure (McCusker et al., 2015). Other known instances of mammalian blastema mediated regeneration are the regeneration of ear punch wounds in rabbits (Vorontsova and Liosner, 1960) or certain mouse models such as the spiny mouse Acomys (Seifert et al., 2012a) or the Murphy’s Roth Large (MRL) mouse (Clark et al., 1998), and the annual regeneration of antlers in deer (Kierdorf et al., 2007).

The process of mouse digit regeneration

During the endogenous regeneration process, the digit undergoes sequential phases of wound healing, blastema formation, and differentiation of regenerating tissues (Fernando et al., 2011; Dolan et al., 2018). The initial inflammatory response post digit amputation is crucial to regeneration success—a physiological component that is also true in other systems and organs, such as the zebrafish tail fin, axolotl limb, and ear tissue of the spiny mouse (Petrie et al., 2014; Godwin et al., 2013; Simkin et al., 2017b). Depleting macrophages in the digit tip at an early stage of regeneration leads to the subsequent inhibition of bone histolysis and wound closure, and the failure of the blastema to form (Simkin et al., 2017a). Complete depletion of macrophages in other systems such as axolotl limb and neonatal mouse heart also results in the inhibition of regeneration (Godwin et al., 2013; Aurora et al., 2014). Due to the functional diversity of macrophage subtypes, it is impossible to say if macrophages in general are inhibitory or essential to regeneration (Novak and Koh, 2013). It is likely that a specific subtype is involved in blastema formation, howeverthe exact immunoregulatory dynamics have not been studied at that resolution.

Concurrently with the inflammatory phase of wound healing, the digit bone undergoes histolysis, the degradative loss of organized bone tissue. In regenerative amputations, the digit can lose up to half its original volume in bone (Simkin et al., 2015a). While the exact role of histolysis in mouse digit tip regeneration is not entirely clear, it appears to facilitate the regenerative response (Dawson et al., 2016). This is supported by the observation that blastema size correlates positively with the amount of histolysis observed, and that degradation of the extracellular matrix may release pro-regenerative factors and cells that encourage new bone deposition (Simkin et al., 2015a; Dawson et al., 2016).

The wound epithelium forms after histolysis occurs, which is a relatively delayed timeframe compared to most rapid wound healing responses. Wound closure occurs in the digit tip when the injured tissue retracts proximally, ensuring the periosteum of P3 is not exposed. Following histolysis, the wound epithelium migrates distally, forming over the digit bone where it will act as a crucial signaling center in the formation of the blastema (Takeo et al., 2013). Secreting chemoattractants such as Sdf-1, the wound epithelium regulates the migration of mesenchymal blastema cells to the wound site (Lee et al., 2013). The activity of the wound epithelium is another response that plays a vital role in other systems, as inhibiting its formation negatively affects the regenerative response in salamanders and humans as well (Thornton, 1957; Illingworth, 1974).

The origin of blastema cells is highly heterogeneous, though their exact cellular composition is complicated due to the histolytic events that precede it (Simkin et al., 2015a). Lineage tracing studies show that they are lineage-restricted, though they derive from a wide variety of tissues, including the epidermis, bone, connective tissue, and vasculature (Lehoczky et al., 2011; Rinkevich et al., 2011; Takeo et al., 2013; Wu et al., 2013). No transdifferentiation occurs between ectodermal and mesodermal lineages, as all regenerative epithelium is derived from keratinocytes and bone and periosteum are derived from osteoblast (Lehoczky et al., 2011; Rinkevich et al., 2011). However, a potentially understudied source of progenitor cells may also originate from fibroblastic cells of the connective tissue, as fibroblasts are known to be phenotypically plastic and multipotent in the context of tissue repair (Plikus et al., 2021).

The complete formation of the wound epithelium marks the end of the wound healing phase, and the initiation of blastema formation. The digit tip blastema is a transient aggregation of undifferentiated cells, with their physiology broadly characterized as proliferative, avascular, and hypoxic. Blastema cells are recruited to the wound site through signaling from the wound epithelium, and receive signals that are crucial to maintain high proliferative ability. Broad signals include bone morphogenetic proteins (BMPs), as inhibiting BMPs with noggin in amputated digits results in their failure to regenerate (Han et al., 2003). Other signals are regulated by digital nerves, which play a crucial role in creating a proliferative environment. Denervation of the digit tip reduces blastema proliferation, suggesting that neurotrophic factors released by paracrine signaling contribute to digit regeneration (Mohammad and Neufeld, 2000; Rinkevich et al., 2014). Surrounding tissues such as the nail bed also regulate the secretion of FGFs from digital nerves through Wnt signaling (Takeo et al., 2013). Recent work also suggests that these neurotrophic factors are not dependent on axonal innervation, but secreted by nerve-associated Schwann cells precursors that dedifferentiate and secrete growth factors such as oncostatin M (OSM) or platelet-derived growth factor AA upon amputation (PDGF-AA) (Johnston et al., 2016). The rapid proliferation of blastema cells may be facilitated by the maintenance of their undifferentiated state. A genetic marker of blastema cells is the transcription factor Msx1, which is a transcriptional repressor necessary for digit regeneration (Han et al., 2003; Lehoczky et al., 2011). Msx1 is also expressed in the clot that forms over the wound site, and removal of the clot attenuates regeneration, suggesting that the clot may play a functional role in regeneration beyond wound healing (Lehoczky et al., 2011).

The avascular nature of the blastema is also important in maintaining its regenerative ability. Regenerating mouse digits were found to have lower numbers of endothelial cells compared to non-regenerative amputations, which suggests the reduced presence of vasculature (Said et al., 2004). Blastema expression of the anti-angiogenic gene Pedf correlated with successful regeneration (Yu et al., 2010), while BMP9, which induces the expression of the angiogenic factor Vegfa, inhibited regeneration (Yu et al., 2014). Pedf expression was also present in regenerative digit amputation wounds, while it was absent in proximal non-regenerative amputations, confirming that vasculature formation is inhibited in regenerative responses (Muneoka et al., 2008; Yu et al., 2010). The delayed formation of vasculature in regenerative responses contrasts with the rapid angiogenic response of non-regenerative wound healing, where the granulation tissue that eventually becomes scar tissue is highly vascularized (Kawasumi et al., 2013). Additionally, granulation tissue expresses high levels of Vegfa (Semenza, 2010), supporting the hypothesis that Pedf expression during the early stages of wound healing and blastema formation suppresses Vegfa to maintain a regeneration-permissive wound environment.

The blastema’s lack of vasculature may be permissive to regeneration by forming a hypoxic microenvironment. Reactive oxygen species have been demonstrated to be essential to regeneration in other systems such as Xenopus and zebrafish (Love et al., 2013; Gauron et al., 2013), and the changing vascular profile of regenerating mouse digits are likewise suggestive of a fluctuating oxygen microenvironment. Studying the oxygen profile of regenerating mouse digits showed that the hypoxic areas of the amputation site significantly increase during blastema formation, and that oxygen profiles vary temporally over the course of regeneration (Sammarco et al., 2014). Spatially, hypoxic cells are first restricted to the bone marrow of the amputated digit, before becoming associated with the forming blastema itself, and prematurely ending those hypoxic phases attenuates the regenerative response (Sammarco et al., 2014). It is hypothesized that bone marrow specific hypoxia may activate and encourage the proliferation and delay differentiation of stem cells such as osteoblast progenitors (Tuncay et al., 1994; Zahm et al., 2008), which are known to contribute to the regenerated digit (Lehoczky et al., 2011). Furthermore, hypoxia-inducible factor (HIF-1), which is the primary mediator of cell survival under hypoxia, upregulates the blastema markers Sdf-1 and Cxcr4 under hypoxic conditions (Ceradini et al., 2004; Staller et al., 2003). It is therefore likely that the hypoxic microenvironment of the regenerating digit is essential to activate Sdf-1/Cxcr4 signaling, which in turn increases blastema cell recruitment and retention at the wound site.

Oxygen dynamics may also play a role in shifting regeneration from the blastema formation stage to the re-differentiation stage. The following phase is marked by a shift in oxygen dynamics, as a release from hypoxia is necessary for re-differentiation (Sammarco et al., 2014). Physiologically, certain oxygen thresholds need to be met for osteoblasts to secrete collagen that forms the mineralized bone matrix (Ramaley and Rosenbloom, 1971), which suggests that oxygen might act as a primary cue for other molecular signals needed for re-differentiation.

The differentiation phase of digit regeneration does not occur in the same manner as digit development: instead, it occurs through a different mechanism that allows for more rapid osteogenesis. In development, the digit bone, as with most long bones, forms through a process of endochondral ossification, with a chondrogenic scaffold later replaced by bone (Han et al., 2008). However, unlike most long bones, the P3 bone possesses only a single growth plate at its proximal end, instead of at both proximal and distal end (Dixey, 1881). Unique to the P3 bone, the length of the phalanx is also increased by an additional ossification center located in the distal end of the digit bone, with up to 55% of postnatal elongation achieved through this process of intramembranous ossification (Han et al., 2008).

Similarly, in regenerated digits, bone forms through direct intramembranous ossification as re-differentiating blastema cells do not form chondrogenic cells during regeneration (Sensiate and Marquez-Souza, 2019). New bone is simply built directly onto the digit stump. This process creates a regenerated digit that is histologically distinct from the original bone. In the original digit bone, collagen fibers are arranged in parallel with one another, while regenerated bone fibers are “crosshatched,” producing a structure often termed woven bone (Simkin et al., 2015b). Regenerated bone often has numerous trabecular spaces, resulting in a more porous and likely weaker structure, but tends to be up to 50% larger in volume than the original digit (Fernando et al., 2011). While the bone increases in density over time, it has been hypothesized that mouse digit regeneration thus represents an evolutionary outcome that traded anatomical fidelity for a relatively rapid and functional regenerative structure (Simkin et al., 2015a).

However, despite the differing architectures of original and regenerated digit, the distinct shape of the P3 bone allows us to conclude that P3-specific patterning does occur during regeneration. Most obviously, the tapered shape of the digit tip is recapitulated in the regenerate (Fernando et al., 2011), as well as the dorsal curvature of the nail (Rinkevich et al., 2011). Patterning genes involved in embryonic limb patterning are also re-expressed in digit regeneration, such as engrailed-1, which patterns the dorsoventral axis of developing limbs (Rinkevich et al., 2011).

Positionally-determined regeneration ability

The proximity of regeneration-competent and regeneration-incompetent zones in the mouse digit also makes it an ideal model to study regeneration failure. If the digit can form a blastema at one level of amputation, why does it fail to do so at another? It is hypothesized that proximal amputations of P3 remove crucial signaling centers and sources of progenitor cells, namely the nail epithelium and the periosteum (Sensiate and Marquez-Souza, 2019). The nail organ appears to have a determinate role in digit regeneration, as Wnt signaling from the nail bed to surrounding mesenchymal cells is both necessary and sufficient to encourage distal growth of the phalanx (Takeo et al., 2013; Lehoczky et al., 2015). In parallel, the periosteum is a mesenchymal tissue that is a source of bone progenitor cells generally (Murao et al., 2013), and is known to drive antler regeneration in deer (Li et al., 2014). As bone progenitor cells are the primary actors in digit regeneration (Lehoczky et al., 2011), it has been hypothesized that the periosteum acts as a source of progenitor cells while the nail bed acts as a source of osteogenic signaling (Sensiate and Marquez-Souza, 2019). Proximal amputations that remove over 50% of P3 eliminate both the nail bed and the periosteum, thus removing key regeneration-responsive elements.

Another potential contributing factor to the positionally-dependent regenerative response is the role of connective tissue fibroblasts. Fibroblasts regulate wound healing and regeneration in diverse contexts, such as amphibian limb regeneration (Gerber et al., 2018), mammalian wound healing (Bainbridge, 2013) and zebrafish and neonatal mouse heart regeneration (Hu et al., 2022; Wang et al., 2020). They also maintain positional memory in the human body (Chang et al., 2002). Manipulating signaling pathways that affect positional information in regenerative systems have led to ectopic expansions of proximal domains, increased blastema size and lengthened or duplicated anatomical structures (Kujawski et al., 2014; Thoms and Stocum, 1984, Blums and Begemann, 2015; Wang et al., 2019). Thus, positionally dependent differences within certain cell types may contribute to their differing ability to mount a regenerative response.

Positionally-specific cellular properties

At a cellular level, fibroblasts derived from the connective tissue of mouse digits displayed different properties when derived from either P2 or P3 phalanges. When injected into regenerating digits, both retained the ability to form the blastema, but P3 cells were more proliferative, responsive, and were able to integrate into the regenerate more successfully (Wu et al., 2013). Experiments with isolated stromal cells suggest that the difference in rates of proliferation may be attributed to the higher levels of cell cycle inhibitors expressed in P2 cells, as P3 cells displayed much higher proliferative rates in vitro as well (Lynch and Ahsan, 2013). Their differential responsiveness and ability to integrate could be explained by structural and signaling differences that affect the cells’ ability to sense and migrate toward the regenerate, as in vivo studies showed that transplanted P3 cells respond and migrate toward the wound site while this honing behavior is not observed in transplanted P2 cells (Wu et al., 2013).

Structurally, the cytoskeletal properties of P2 cells differ from those in P3, as microfilament and microtubule-related genes were expressed at higher levels, highlighting the potential role of modifying cell migration and cytoskeletal properties in regeneration (Lynch and Ahsan, 2013). P2 and P3 cells also interact differently with other cell types in the wound microenvironment—co-culture with adult and neonatal fibroblasts increased the proliferation and migration of P2 cells, but only neonatal fibroblasts increased the migratory behavior of P3 cells (Lynch and Ahsan, 2014). Moreover, P3 cells may be more responsive to signals from the amputation site itself, as exposing both P2 and P3 cells in vitro to BMP2—a signaling molecule expressed in the marrow of the amputated bone—recapitulated the differential migratory behavior observed between P2 and P3 cells in vivo (Lynch and Ahsan, 2014). Thus, positionally separated cells show intrinsic differences in their responses to other cells and secreted factors encountered in the wound microenvironment or from the wound site itself.

Induced digit regeneration

The mouse digit has also been used to characterize induced regeneration, in which key morphogenetic agents were used to induce regeneration from a typically non-regenerative amputation. Both proximal amputations through P3 (removing 60% or more of the phalanx) and proximal amputations through P2 are used to investigate the possibility of inducing regenerative responses in contexts where they would typically fail.

Similar approaches have also been carried out in other non-regenerative vertebrate contexts, demonstrating the potential viability of modulating an injury response to generate regenerative outcomes. In experiments with adult mice, rats, and pigs, modulating the wound response to full thickness skin wounds to mimic that of fetal skin led to scar-free skin regrowth in adults, an ability that is usually only observed in fetal skin (Ferguson and O’Kane, 2004). In embryonic chicks, amputated limb buds, which do not typically regenerate, could be coaxed to do so by locally stimulating mesodermal cells using FGFs (Taylor et al., 1996; Kostakopoulou et al., 1996). These successful attempts to induce regeneration in multiple organs across different species and life stages have thus motivated the idea that tissues or organs could possess dormant regenerative potential—and identifying the agents that unlock it could be key to understanding how to systematically expand regenerative ability.

Proximal P3 system

Proximal amputation through P3 (removing 60% or more of the terminal phalanx) is an obvious means to study positionally dependent regenerative responses. Recent work differentiates the proximal P3 amputation from P2 amputation due to the observation of blastema-like mesenchymal tissue located in between the amputation site and regenerating dermis (Sensiate and Marquez-Souza, 2019). This is further supported by recent work showing that proximal P3 amputation results in a highly attenuated regenerative response, including blastema formation and partial bone regrowth, rather than complete failure as previously characterized (Dawson et al., 2020). Most models of scar-free healing display a limited fibrotic response (Seifert et al., 2012a; Seifert et al., 2012b), and no fibrosis is observed in distal P3 regeneration. P3 proximal amputation results in the formation of transient fibrosis, followed by a regenerative response limited to the marrow cavity, while the little remaining periosteum remains inert (Dawson et al., 2020).

P2 system

Non-regenerative, proximal amputation through the P2 digit is also a common system to investigate regenerative failure. The standard outcome of a non-regenerative amputation is a truncated skeletal element, and the formation of fibrotic scar tissue over the wound site (Dawson et al., 2016). The P2 bone stump appears inert at first glance, as it is visually similar before and after the injury response, but this seeming similarity obfuscates a dynamic and elaborate tissue repair process that is reminiscent of an initiated, but failed, attempt at regeneration (Dawson et al., 2016). The P2 amputation response resembles skeletal repair after fracture, as an initial inflammatory response is followed by periosteum-derived chondrocyte formation of a cartilage callus, which is then vascularized and organized first into a woven bony callus and then reorganized into lamellar bone (Dolan et al., 2018). The bone also undergoes histolysis and is truncated, and fibrotic scarring also occurs on the soft tissue around the wound site.

Even though cells possess position specific qualities, this limited regenerative response shows that there are indeed regeneration-competent cells located in digit amputations that fail to completely regenerate. Thus, much work in inducing and expanding regenerative capacity focuses on modulating the wound site to create a more supportive environment for regeneration. In neonatal mice, treating proximal P3 amputations with exogenous BMP7 induced regeneration by potentially reactivating the embryonic program active during development (Yu et al., 2010). Instead of intramembranous ossification, blastema formation was followed by endochondral ossification, with endochondral marker genes expressed in a similar pattern as in development along the proximal-distal axis (Yu et al., 2010).

A similar response to exogenous BMP2 was also observed when neonatal digits were amputated at the P2 level, as well as in the long bones of amputated adult hindlimbs (Yu et al., 2012). By forming a new endochondral ossification at the distal end of the skeletal element, amputated elements could regenerate distally patterned structures, but were restricted to regrowing specific segments—i.e., no joints or further distal structures were regenerated (Yu et al., 2012). However, the induced regeneration of bone and joint could be achieved through the sequential treatment of P2 amputations with exogenous BMP2 and then BMP9 (Yu et al., 2019). It was hypothesized that BMP9 treatment reactivated the developmental program for joint formation, allowing amputated digits to form a synovial cavity and an articular cartilage lined structure that articulates with the bone stump (Yu et al., 2019). The ability of transient treatments with growth factors being able to stimulate regenerative responses highlights the potential of extrinsically manipulating typically non-regenerative injuries to engineer a regenerative response.

Mouse heart regeneration

Another organ with limited regenerative ability in the mouse is the heart. Instead of being spatially limited as in the case of the digit tip, cardiac regeneration in the mouse is temporally limited. Neonatal mice possess cardiac regenerative ability up till the first week of postnatal life, after which this transient ability rapidly diminishes (Porello et al., 2011)(Figure 3). In contrast, the adult mammalian heart was long considered a post-mitotic organ, as cardiomyocytes were thought to be terminally differentiated upon maturity (Poolman and Brooks, 1998). However, radiocarbon tracing showed that cardiomyocytes do undergo renewal, albeit at a low rate of about 1%, decreasing throughout adult life in humans (Bergmann et al., 2009). However, this rate of renewal will not successfully replace the significant loss of cardiomyocytes lost to cardiac injury such as myocardial infarction, which can compromise around 25% of cells in the heart (Murry et al., 2006). When myocardial infarction occurs in non-regenerative hearts, the site of injury is typically replaced with a fibrotic scar that affects tensile properties of the heart and decreases cardiac function, potentially leading to heart failure (Leask, 2010).

Figure 3.

Diagram of the heart with structures derived from the cardiac neural crest labelled. Ao: Aorta P: Pulmonary artery.

Upon cardiac injury affecting approximately 15% of the heart, neonatal mice can restore myocardial damage in around the next month with little evidence of fibrosis and the restoration of cardiomyocytes and cardiac contractile function (Porello et al., 2011; Porello et al., 2013). Lineage tracing of cells in the mammalian heart have shown that regeneration occurs through the proliferation of existing cardiomyocytes (Porrello et al., 2011; Zhu et al., 2018). The regenerative response is marked by cardiomyocyte cytokinesis and sarcomere disassembly, as well as a rapid revascularization response that perfuses the injured tissue (Porello et al., 2012). It is thus commonly thought that the failure to regenerate in older life stages is due to the limited potential of mature cardiomyocytes to proliferate and renew themselves (Cardoso et al., 2020).

Physiological and molecular factors involved in heart regeneration

This limitation is sometimes attributed to an intrinsic property of the cardiomyocytes themselves, as regeneration loss coincides with the time point when cardiomyocytes undergo cell cycle arrest and binucleation at postnatal day 7 (Soonpaa et al., 1996). Regenerative species, such as zebrafish, also possess mononucleated, diploid cardiomyocytes (Kikuchi, 2015). Ploidy appears to correlate with cardiomyocyte proliferation and functional recovery after injury (Hirose et al., 2019) and experimental polyploidization of zebrafish cardiomyocytes was sufficient to inhibit regeneration (González-Rosa et al., 2018). Comparative studies between closely related frog species Xenopus tropicalis and Xenopus laevis also reveal that while adult X. tropicalis can completely regenerate myocardium after 10% cardiac apical resection (Liao et al., 2017), X. laevis adults only mount a partial regenerative response to equivalent amounts of damage (Marshall et al., 2017). Notably, X. tropicalis has a diploid genome with mononucleated cardiomyocytes, while X. laevis has a psuedotetraploid genome with majority tetraploid cardiomyocytes (Marshall et al., 2018).

However, studies in neonatal pig regeneration seem to suggest that other factors may be involved. Similarly to mice, pigs also have a short postnatal window during which they are able to regenerate cardiac tissue from pre-existing cardiomyocytes (Zhu et al., 2018). While pig cardiomyocytes also eventually become multinucleated and withdraw from the cell cycle, the majority of them do so after the neonatal period (Adler et al., 1996). Hence it is clear that although retaining cardiomyocyte proliferative ability is a permissive condition for regeneration, it is not the only factor determining regenerative outcome.

Beyond cardiomyocyte-intrinsic factors, other physiological responses are known to influence cardiac regenerative ability. The overall physiology of the heart undergoes a number of changes shortly after birth: a metabolic shift likely in response to changing concentrations of oxygen, and a mechanical shift, as the composition of cardiac extracellular matrix (ECM) changes rapidly in postnatal life. Different molecular regulators, such as the YAP and Hippo signaling pathways, specific transcription factors such as Gata4 or Meis1, or microRNAs, have also been implicated as key regulators of cardiomyocyte physiology. Furthermore, while 90% of myocardium is composed of cardiomyocytes, cardiomyocytes only represent about 30% of the total cells in the heart (Pinto et al., 2016). Other types of cells, aside from cardiomyocytes, immune cells, fibroblasts, and neural cells—may also contribute to the changing physiology of the postnatal heart.

Hippo/YAP signaling

A pertinent physiological factor in cardiac maturation is the switch between primary modes of growth in the fetal and adult mammalian heart. While the main mode of growth during fetal development is cardiomyocyte proliferation, growth after birth is mostly achieved through increase in cardiomyocyte size, which is known as cardiac hypertrophy (Maillet et al., 2013). One pathway of interest for this transition is the Hippo/YAP signaling pathway, which plays a conserved role in organ size control, stem cell fate and differentiation, as well as the regulation of cellular proliferation and apoptosis (Pan, 2010). YAP (Yes-associated protein) mediates Hippo signaling as a transcriptional co-factor—when Hippo signaling is active, YAP is phosphorylated and inactivated, leading to cell apoptosis, while Hippo inactivity leads to YAP activation and translocation into the nucleus where it binds to transcription factors that promote cell survival and proliferation (Ikeda and Sadoshima, 2016). In cardiac development, Hippo limits cardiomyocyte proliferation through inhibiting YAP, and Yap overexpression during development increases cardiomyocyte proliferation and results in hyperplasia of ventricular walls (Heallen et al., 2011). Likewise, conditional ablation of Yap in cardiac cells leads to hypoplasia through reduced cardiomyocyte proliferation (von Gise et al., 2012). In both cases, cardiomyocyte size was unchanged, meaning that Hippo/YAP specifically affects cardiac growth in development through cardiomyocyte proliferation.

As the loss of cardiomyocyte proliferative ability is thought to be the primary reason behind regenerative failure, investigating Hippo/YAP activity in heart regeneration is of much interest. Hippo is generally thought to suppress mitosis in the adult mouse heart, as the amount of YAP protein detected in the heart declines with age (von Gise et al., 2012). This has been studied in a regenerative context in neonatal rats, which also lose the ability to regenerate heart tissue within the first week of life (Zogbi et al., 2014). Regeneration-competent P1 rats displayed reduced YAP phosphorylation after ventricular resection, which was accompanied by an increase in proliferative cardiomyocytes (Bei et al., 2023). In addition, Yap has also been shown to be necessary for neonatal mouse regeneration. Cardiomyocyte-specific knockout of Yap led to extensive fibrosis and a lack of healthy myocardium in P2 mice post-infarction (Xin et al., 2013). Endogenous activation of Yap through cardiomyocyte-specific knockouts of Hippo effectors Lats1/2 and Sav1 conversely was cardioprotective in adult mouse models of infarction and was even able to induce mitosis in adult myocardium (Heallen et al., 2013). As cardiomyocytes expressing Yap were observed at the border of infarcted cardiac tissue, it is hypothesized that Yap prevents cardiomyocyte apoptosis and promotes proliferation in the border zone, which encourages myocardial survival and prevents further infarction (Xin et al., 2013; Del Re et al., 2013). Work in adult pigs suggests that this hypothesis may apply to mammalian hearts more broadly, as gene therapy knockdown of Sav in border zone cardiomyocytes resulted in cardiomyocyte division and improved tissue renewal after infarction (Liu et al., 2021).

The Hippo/YAP pathway is notable for its crosstalk with numerous other physiological factors known to play a role in heart regeneration. One of the closest associations is its interactions with cardiac ECM and regulation through mechanical stress. Changes in tissue stiffness and cytoskeletal rearrangement are known to mediate YAP activation (Halder et al., 2012; Aragona et al., 2013; Vite et al., 2018). Direct interactions between the ECM and the Hippo/YAP pathway have also been established through the dystrophin-glycoprotein complex (DGC), a transmembrane complex that links the actin cytoskeleton to the ECM. Agrin (an ECM protein known to promote cardiomyocyte proliferation) and YAP both interact directly with DGC, with agrin binding resulting in YAP dissociating from the complex and being released into the nucleus, where its transcriptional activity promotes cardiomyocyte proliferation (Bassat et al., 2017). Other work indicates that Hippo signaling is required for YAP binding to the DGC, as phosphorylated, inactive YAP interacts directly with components of the DGC (Morikawa et al., 2017).

Inflammatory and oxidative stress related signaling have also been associated with mediating Hippo/YAP activity. YAP appears to play an immunosuppressive, anti-inflammatory role after infarction, as deletion of YAP from cardiac epicardial cells led to lower levels of anti-inflammatory cytokines, prolonged inflammation, and drastically increased fibrosis—all of which contributed to high incidence of cardiomyopathy and poorer outcomes post-infarction (Ramjee et al., 2017). The Hippo effector Mst1 is also known to be activated by oxidative stress (Odashima et al., 2007), which is greatly increased in post-infarction hearts (Murphy and Steenbergen, 2008). Inhibiting Mst1 in mice reduced fibrosis and apoptosis post-injury (Odashima et al., 2007), but did not stimulate increased cardiomyocyte proliferation, unlike the inhibition of other Hippo effectors (Yamamoto et al., 2003). This may be due to compensatory effects from Mst2, or Mst1 acting to promote cell survival through alternate means, such as the prevention of autophagy (Wang et al., 2018). Mst1 inhibition is thought to mediate cell survival through the YAP-FOXO1 complex, which is inhibited by MST1 and induces the expression of antioxidant genes (Lehtinen et al., 2006; Shao et al., 2014). The Hippo/YAP pathway has an abundance of complex interactions with the physiology and signaling that varies across hearts of different ages and conditions and is a promising avenue for further investigation.

Cardiac metabolism and oxidative stress

A noticeable physiological transition within the first week of postnatal life is metabolic: embryonic hearts typically use glycolysis to generate energy (Lopaschuk et al., 1992), while adult cardiomyocytes use mitochondrially-dependent oxidative phosphorylation (Wisnecki et al., 1985). The postnatal cell-cycle arrest of mouse cardiomyocytes was found to be partly mediated by this metabolic switch, with the shift to oxidative phosphorylation resulting in increased reactive oxygen species (ROS) production (Puente et al., 2014). ROS signaling is thought to be the cue to promote immune cell recruitment into injured tissue (Niethammer et al., 2009), as necrotic cells and neutrophils secrete ROS and activate a local inflammatory response by cardiac fibroblasts and mast cells, leading to the secretion of pro-inflammatory cytokines (Kawaguchi et al., 2011). Excessive ROS production results in chronic inflammation and increased myocardial injury (Muntean et al., 2016). Increased ROS generation may account for the loss of mouse heart regeneration, as ROS in postnatal day 7 mouse hearts was shown to cause DNA damage and cell cycle arrest in cardiomyocytes (Puente et al., 2014). Conversely, exposure to chronic hypoxia post-infarction induced heart regeneration in adult mice (Kimura et al., 1985; Nakada et al., 2017), supporting the hypothesis that reducing oxidative damage from ROS could facilitate cardiomyocyte renewal. However, while their presence in excessively high levels impairs regeneration, ROS signaling appears to have pleiotropic effects in cardiac injury—low levels of ROS are needed for cardioprotective redox signaling (Muntean et al., 2016), while hydrogen peroxide (a potent ROS) in zebrafish has shown to promote leukocyte recruitment and cardiomyocyte proliferation (Yan et al., 2014; Niethammer et al., 2009). Understanding how redox signaling is differentially modulated in regenerative and non-regenerative contexts might therefore give us better insight into the variable physiological effects of ROS.

Additionally, the specific metabolites used in cardiac energetic metabolism also appear to affect cardiomyocyte proliferation. The shift to oxidative phosphorylation in the neonatal-adult also results in a shift in primary energetic substrate from pyruvate in neonates to fatty acids in adults (Lopaschuk et al., 1992). Inhibiting fatty acid metabolism has been shown to decrease DNA damage and promote cardiomyocyte proliferation, demonstrating that the changing metabolic profile of the heart impacts its proliferative potential (Cardoso et al., 2020b). Conversely, inhibiting glycolytic enzymes resulted in increased cell death and fibrosis after cardiac injury in mice and decreased heart regeneration due to low cardiomyocyte proliferation in zebrafish (Wu et al., 2011; Fukuda et al., 2020).

Though the adult mammalian heart typically uses fatty acids as an energy source, it shifts to anaerobic glycolysis during pathological conditions such as ischemia or hypertrophy (Tuomainen and Tavi, 2017). This shift is thought to be cardioprotective, which is supported by the naturally-occurring metabolic reprogramming observed in zebrafish—cardiomyocytes at the border of the injury site upregulated glycolytic genes, while down-regulating genes involved in mitochondrial oxidative phosphorylation (Honkoop et al., 2019). It follows that stimulating glycolysis and glucose metabolism may promote heart regeneration as it returns the heart to a neonatal-like state. The expression of glycolytic enzymes changes over development—pyruvate dehydrogenase kinases (PDK), for example, increase during mammalian heart development, and PDK4 is the most upregulated enzyme in P7 mice hearts, coinciding with the loss of regenerative ability (Sugden et al., 2000). PDKs inhibit pyruvate dehydrogenase (PDH), which is a limiting step of glycolysis, and inhibiting PDKs in turn results in PDH activation (Bae et al., 2021). PDK inhibition or cardiomyocyte-specific deletion has shown to promote cardiomyocyte proliferation and heart regeneration in adult mice (Piao et al., 2017; Cardoso et al., 2020b). Similarly, cardiomyocyte-specific overexpression of rate-limiting glycolytic enzyme pyruvate kinases muscle isoenzyme 2 (PKM2 encouraged cardiomyocyte proliferation and heart regeneration in adult mice (Mangadum et al., 2020).

Neural signaling

The mammalian heart is physiologically regulated by sympathetic and parasympathetic nerves, with cardiac innervation playing a role in heart rate and contraction. Nerves play a role in many regenerative processes, and often secrete pro-regenerative growth factors and mitogens (Kumar and Brockes, 2012). Ablation of either sympathetic or parasympathetic nerves impairs cardiac regeneration in neonatal mouse hearts— chemical ablation of sub-epicardial sympathetic nerves resulted in fibrotic scarring and regenerative failure (White et al., 2015), while vagotomy of the parasympathetic nerve impaired cardiomyocyte proliferation. Cholinergic denervation of the mouse heart through severing the left vagus nervedecreased the level of cell cycle regulators such as Cdk4 and Ccnd2, as well as growth factors Nrg1 and Ngf, with cardiac regeneration partially rescued by exogenously administering NGF and neuregulin1 proteins (Mahmoud et al., 2015).

Extracellular matrix components and properties

The cardiac extracellular matrix (ECM) changes significantly over the course of postnatal maturation and is known to interact with cells directly or contain signaling molecules that regulate migration or pro-regenerative growth factors (Hynes, 2009). The ability of the ECM to mediate the transition from fibrosis to scar resolution may play a significant role in determining regenerative outcome. In adult mammalian hearts, the deposition of fibrous collagenous matrix is permanent, and the myocardium remains fibrotic after injury. However, in regenerative models such as zebrafish and neonatal mice, the scar formation is transient, and is eventually replaced by cardiomyocytes. Scar formation is not itself detrimental to regeneration, as inhibiting scar formation attenuated the regenerative response in zebrafish, decreasing cardiomyocyte proliferation and resulting in an abnormally remodeled ventricular chamber (Chablais and Jazwinska, 2012). This highlights that transient scarring is not inherently inhibitory to regeneration, and that it is specifically the capability to resolve scarring that is crucial to a successful outcome.

The ability to mediate this transition may be due to differing ECM components. Unlike adult ECM, neonatal ECM components have been shown to modulate cardiomyocyte proliferation by stimulating cell cycle reentry in differentiated cardiomyocytes. Levels of agrin, an extracellular proteoglycan, decrease postnatally in mice coincident with the loss of regeneration. Agrin was shown to be necessary for neonatal mouse heart regeneration and administering exogenous agrin in adult mice resulted in cardiomyocyte cell cycle re-entry (Bassat et al., 2017). In pigs, local delivery of exogenous recombinant agrin into the infarcted heart also encouraged cell cycle reentry and resulted in improved heart function and decreased infarct size (Baehr et al., 2020). Agrin is thought to activate cell cycle reentry by promoting sarcomere disassembly and activation of pro-regenerative signaling molecules such as YAP and ERK (Bassat et al., 2017).

In typically regenerative organisms such as zebrafish, specific ECM components mediate different aspects of the regenerative process. Many of these processes are regulated by TGF-β/Activin signaling, as it affects the synthesis of different ECM components with properties crucial to various stages of healing and regeneration. Inhibiting TGF-β/Activin resulted in decreased collagenous scar deposition, as well as decreased fibronectin, an adhesive matrix protein that assists in recruiting and integrating cardiomyocytes in damaged tissue (Chablais and Jazwinska, 2012; Wang et al., 2013). The expression of Tenascin C, an anti-adhesive protein typically expressed at the border of the infarct zone and myocardium (Imanaka-Yoshida et al., 2004) is also completely abolished, which impaired cardiomyocyte ability to migrate into the infarcted area and subsequent tissue remodeling (Chablais and Jazwinska, 2012). The active regulation of ECM composition during specific stages of the injury response might therefore play a significant role in ensuring that a regenerative response supersedes a fibrotic one.

The mechanical properties of ECM, such as its stiffness or flexibility also exert influence over cellular behavior such as migration, adhesion, and proliferation during injury, thereby influencing the success of a regenerative response. From embryonic to postnatal life, cardiac ECM displays a marked and rapid change in both composition and mechanical properties. In neonatal rats, the elastic modulus of ventricular tissue immediately increased upon birth, demonstrating that the transition from a gestational environment has a distinct effect on heart maturation (Jacot et al., 2010). Cardiac stiffness, from the decreased elasticity of the myocardium and cardiac vasculature is associated with aging and the onset of cardiac disease, such as dilated cardiomyopathy and diastolic heart failure (Villalobos Lizardi et al., 2022; Singam et al., 2019). In vitro, substrate stiffness appears to influence mouse cardiomyocyte maturation. Cardiomyocytes grown on stiffer substrates were post-mitotic and binucleated, while those grown on softer substrate began to lose mature contractile gene expression and de-differentiate and continued replicating (Yahalom-Ronen et al., 2015). In neonatal mice, ECM stiffness noticeably increases even within the first two days of postnatal life, and decreasing the stiffness by inhibiting the cross-linking enzyme LOX resulted in an improved regenerative response by postnatal day 3 (P3) mice as far less fibrotic deposition was observed three weeks after apical resection (Notari et al., 2018).

Immune response

The immune response to cardiac injury is highly spatially and temporally regulated. How and when the immune response is activated can result in beneficial or detrimental impacts on heart regeneration. Broadly, the immune response post myocardial infarction can be divided into an inflammatory state triggered by necrotic cells, in which immune cells are recruited to the injury site, followed by an inflammation resolution phase in which reparative cellular responses suppress inflammatory signals and begin to repair tissue (Lai et al., 2019). It is well established in regenerative models that inflammation is essential to cardiac regeneration— suppressing the early immune response by depleting resident macrophages in both zebrafish and neonatal mice resulted in impaired regeneration (Huang et al., 2013; Aurora et al., 2014). Triggering an acute inflammatory response also seems to increase regenerative efficacy, acting as a stimulus for regenerative responses in neonatal mice and zebrafish (Han et al., 2015; de Preux Charles et al., 2016).

However, overactive or chronic inflammation in injury typically leads to increased tissue damage and dysfunction. Prolonged inflammation has been linked to increased apoptosis of cardiomyocytes, fibrogenic signaling in non-infarcted myocardial tissue, and the activation of proteases that further degrade cardiac ECM (Chen and Frangiogiannis, 2012; Frangogiannis and Entman, 2005; Huang and Frangogiannis, 2018). Failure to resolve or contain inflammation has been shown to directly result in adverse cardiac remodeling in mice (Dobaczewski et al., 2010; Cochain et al., 2012), and persistently elevated inflammatory markers in human patients after an acute coronary syndrome was associated with higher mortality (de Lemos et al., 2007). Thus, the poorer prognosis of these patients might be due to increased injury and/or the inability to activate anti-inflammatory pathways to resolve the deleterious effects of prolonged inflammation (Frangogiannis, 2007).

The heterogeneous effects of the immune response on regeneration are likely due to spatiotemporal and phenotypic differences which may be illustrated more clearly at a cellular level. Neutrophils, for example, are rapidly recruited to injured tissue and make up the majority of the immune cells present in the first few days post-injury in both mice and zebrafish (Yan et al., 2013; Lai et al., 2017).

Neutrophil activity is associated with inflammation, as they encourage monocyte recruitment and secrete inflammatory cytokines (Soehnlein and Lindbom, 2010). Prolonged neutrophil retention is generally associated with poorer outcomes after cardiac injury (Mocatta et al., 2007) and impairs regeneration, as it led to excessive fibrosis and unresolved scarring in zebrafish (Lai et al., 2017). However, it would be inaccurate to think of them as purely pro-inflammatory and detrimental. Neutrophils secrete myeloperoxidase, which neutralizes the ROS hydrogen peroxide released after injury (Pase et al., 2012), and facilitate the entry into the reparative phase by shifting macrophages to an M2-like anti-inflammatory phenotype (Horckmans et al., 2017). Blunting the acute inflammatory response might thus have unintended negative downstream consequences, with early attempts at broad inhibition of inflammation post-injury leading to adverse effects (Hartman et al., 2018; Huang and Frangiogiannis, 2018; D’Amario et al., 2021).

The multifaceted and highly interconnected nature of the immune response is perhaps best illustrated by the role of macrophage polarization in heart regeneration. It was observed that distinct subpopulations of macrophages populate the infarct zone of adult mouse hearts over time, with pro-inflammatory M1 macrophages gradually giving way to anti-inflammatory M2 macrophages (Nahrendorf et al., 2007; Yan et al., 2013). Further depletion of M2-like macrophages also led to decreased repair and poorer outcomes post-injury (Shiraishi et al., 2016). However, these dynamics differ between adult and neonatal mice. After injury, the neonatal heart is dominated by M2-like resident macrophages, with the early response coordinated by increasing the numbers of these resident macrophages, instead of recruiting pro-inflammatory monocyte-derived macrophages to infiltrate the wound site like in the adult heart (Lavine et al., 2014). It is unlikely, however, that the M1/M2 paradigm captures the full complexity of macrophage function during heart regeneration as the heterogeneity of immune cell phenotypes and kinetics has only become apparent relatively recently (Martinez and Gordon, 2014). Recent transcriptomic studies of macrophage subpopulations in the adult and neonatal heart show that macrophages possess distinct transcriptional signatures depending on age, and that differences in transcriptionally-regulated macrophage functions might contribute to their differing ability to facilitate regeneration (Simóes et al., 2020).

Elucidating the full dynamics of the immune response during regeneration—including both the kinetics and subpopulations of cells involved will likely be crucial in understanding how it facilitates regeneration. In a comparative study between zebrafish and the non-regenerative medaka, the medaka heart displayed both delayed recruitment of macrophages and neutrophils to the injury site, as well as delayed clearance of neutrophils compared to zebrafish (Lai et al., 2017). Changing the dynamics of this response by administering the inflammatory agonist poly I:C by accelerating immune cell recruitment and subsequent clearance led to an improved regenerative response in medaka hearts, including more rapid revascularization, scar resolution and improved cardiomyocyte proliferation (Lai et al., 2017). This study demonstrates how interventions that affect the timing and duration of specific phases of the immune response might be key to modulating the immune response to produce a pro-regenerative outcome.

Strategies to induce heart regeneration

In addition to understanding how different physiological factors affect regenerative outcomes in the heart, there has also been much research into targeted strategies and therapeutics to improve mammalian heart regeneration. In clinical settings, heart failure after myocardial infarction is often due to a combination of post-injury complications: myocardium is replaced by mechanically inferior fibrotic scarring, with exacerbated hypertension and valvular diseases often following the infarction (Velagaleti et al., 2020). Currently, in the case of severe infarction, whole heart transplants remain the only viable strategy to regain heart function. However, the long waitlist of patients due to insufficient donors and numerous post-transplant complications is obviously limiting (Benjamin et al., 2019). Therefore, using our understanding of basic cardiac physiology to target and promote cardioprotective and regenerative processes, while attenuating damaging ones remains both biologically and clinically relevant to current research.

Strategies to induce heart regeneration have typically involved cellular transplants to generate new myocardium, using exogenous bioactive factors to encourage existing cardiac tissue to become more regenerative, or modifying the structural microenvironment of the heart to promote regeneration. The myriad interactions between cardiac cell types and the increasing appreciation for the role of their microenvironment likely means that successful interventions will make use of a combination of these strategies, or require the coordination of different interventions at different timepoints in the recovery process.

ESCs and iPSCs

Cell-replacement therapies often exogenously introduce new cardiomyocytes generated from either embryonic stem cells (ESCs) or somatic cells via induced pluripotent stem cells (iPSCs) (Shiba et al., 2016). While iPSCs and ESCs were injected directly into the myocardium in earlier attempts, they are now often re-differentiated into CMs before being introduced into the heart, as the pluripotency of iPSCs led to the risk of teratoma formation (Nussbaum et al., 2007). iPSC and ESC-derived CMs have shown beneficial effects in improving cardiac function in mammalian models (Kawamura et al., 2012; Chong et al., 2014). However, cell transplantation with ESCs or iPSCs often have low rates of survival, with only around 10% of cells at best surviving 24 hours after transplantation (Guo et al., 2017).

To improve survival and retention, much research has focused on integrating and recreating cardiac cell-cell and cell-ECM interactions. For example, mouse ESC-CMs cultured in fibroblast medium obtained from neonatal hearts displayed improved contractile ability compared to those cultured in adult fibroblast media (Liau et al., 2017). Furthermore, given the role of cardiac ECM in affecting cellular physiology, migration and organization, much research has investigated scaffold-based approaches to recreate a supportive microenvironment. Hydrogel scaffolds have been shown to increase cell retention as well as induce maturation and integration with host cardiac tissue (Ban et al., 2014), while cellular transplants that utilized three-dimensional cardiac fibroblast ECM as a transfer media were better retained in ischemic cardiac tissue (Schmuck et al., 2014). The question of the optimal combinations of biomechanical and biochemical components to recapitulate a regenerative cardiac microenvironment is an active area of research—the most ambitious goals include cardiac patch designs that aim to develop fully-functioning heart tissue ex vivo, creating a patch of tissue that would be ready to transplant and used to replace large sections of infarcted tissue (Liau et al., 2011; Zhang et al., 2013; Hayoun-Neeman et al., 2019).

Cardiac stem cells

Besides ESCs and iPSCs, the potential use of resident cardiac stem cells (CSCs) in heart regeneration has also been the subject of much research over the last two decades. Though their existence remains deeply contested, the general hypothesis follows that resident progenitor stem cells in the heart could be cardiomyogenic, with different subpopulations of putative stem cells being investigated for their ability to generate new cardiomyocytes after injury (He et al., 2020).

The first putative cardiac stem cell population proposed were cKit+ cells. Initial studies reported that cKit+ CSCs were multipotent in vitro, differentiating into smooth muscle, cardiomyocytes, and endothelial cells (Beltrami et al., 2003). They were also reported to contribute to recovery after myocardial infarction by differentiating into new myocardium (Beltrami et al., 2003; Dawn et al., 2005). However, more recent studies reported that cKit+ CSCs did not differentiate into cardiomyocytes in the infarcted adult mouse heart (Zaruba et al., 2010), and that putative new cKit+ cardiomyocytes in earlier human heart studies were likely mast cells (Pouly et al., 2008). Lineage tracing of cKit+ cells demonstrated that cKit+ cardiomyocytes were likely the result of cell fusion rather than differentiation, and that all cKit+ cells in the adult heart did not have a cardiomyocyte identity (van Berlo et al., 2014; He et al., 2017).

Another such population were Sca1 (stem cell antigen)/Ly6a+ cells, which were shown to be multipotent and possess cardiomyogenic potential in vitro (Oh et al., 2003). While a study reported that endogenous Sca1+ cells in the heart contributed to myocardium after injury (Uchida et al., 2013), this finding was not supported by subsequent work, as transplants of Sca1+ cells did not differentiate into new cardiomyocytes in injured myocardium (Soonpaa et al., 2018), and lineage tracing of Sca1+ cells revealed that they mostly adopted endothelial and not cardiomyocyte fates (Vagnozzi et al., 2018; Zhang et al., 2018). Thus, both Sca1+ and cKit+ cells are generally no longer thought of as endogenous myocardial progenitors.

However, certain subpopulations of these cells have proved beneficial in ways other than their cardiomyogenic potential. A subset of Ly6a+ cells of mesenchymal origin are known as mesenchymal stromal cells (MSCs). MSCs, similarly to other Ly6a+ cells, had some, relatively limited, ability to differentiate into cardiomyocytes in vivo (da Silva Meirelles et al., 2006), but had a beneficial effect when transplanted in injured myocardium. It was later found that MSCs benefit host tissue through paracrine signaling rather than direct differentiation into myocardial tissue (Guo et al., 2020). MSCs secrete a wide variety of signaling molecules, which have been shown to promote cell survival, possess immunomodulatory properties and encourage neovascularization (Pittenger and Martin, 2004; Thakker and Yang, 2014; Wehman et al., 2016).

MSCs attenuate pathological outcomes after infarction directly by interacting with various immune cells involved in the innate and adaptive immune response. Their presence can decrease the number of pro-inflammatory monocytes (Miteva et al., 2017), in addition to promoting M2 macrophage polarization and inhibit T cell proliferation (Chiossone et al., 2016). The introduction of MSCs to injured tissue also has a marked antifibrotic effect, reducing scarring after injury directly by lessening ECM deposition and inhibiting fibroblast activation through matrix metalloproteinase (MMP) regulation (Guo et al., 2020). They may also regulate cardioprotective signaling in a paracrine manner, as MSC transplants to infarcted cardiac tissue displayed lower expression of the inflammatory markers TNF-α, IL-1 and IL-6, and decreased apoptosis in the myocardium (Guo et al., 2007).

Thus, while certain identified subpopulations improve cardiac function after injury, the current research consensus suggests that the adult mammalian heart does not possess a dedicated progenitor population capable of readily differentiating into cardiomyocytes (Maliken and Molkentin, 2018; He et al., 2020). Genetic lineage tracing of cardiomyocyte and non-cardiomyocyte cells in the developing mouse heart shows that lineage segregation occurs by E11.5, as non-myocyte conversions to cardiomyocyte conversions peak at E8.5 and gradually declines (Li et al., 2018). It was also found that non-myocytes do not contribute to the regenerating neonatal mouse heart, in line with results from studies in zebrafish that the primary source of new cardiomyocytes are existing cardiomyocytes (Li et al., 2018; Jopling et al., 2010).

Direct transdifferentiation of non-cardiomyocyte cell types

Another method that bypasses the potential oncogenic or immunological side effects of cell transplants, while avoiding the limitations of targeting either mature cardiomyocytes or low numbers of potential cardiac progenitors is to target a far more abundant and plastic cell type in the heart, the cardiac fibroblast. Cardiac fibroblasts are one of the most numerous cell types in the heart (Baudino et al., 2006) and are amenable to transdifferentiation, having been reprogrammed into pluripotent stem cells, smooth muscle cells and neurons by combinations of lineage specific transcription factors (Takahashi and Yamanaka, 2006; Wang et al., 2003; Vierbuchen et al., 2010). Overexpression of the transcription factors Gata4, Mef2c and Tbx5 through retroviral injection resulted in cardiac fibroblasts expressing cardiomyocyte markers, spontaneous contraction, and cardiomyocyte-like electrophysiology in vivo (Ieda et al., 2010). Subsequent work has improved upon the efficiency of the original protocol through varying proportions and combinations transcription factors used (Song et al., 2012; Wang et al., 2015; Protze et al., 2012). Induced cardiomyocyte conversion yield and function has also been increased with the addition of small molecules. Adding protein kinase B (AKT1) to activate insulin-like growth factor (IGF1) and mammalian target of rapamycin (mTORC1) signaling resulted in more polynucleated, hypertrophic cells with increased spontaneous contraction (Zhou et al., 2015). Inhibiting pro-fibrotic pathways through the TGF-β inhibitor A83–01 also promoted the expression of cardiac contractility genes Actc1, Myh6 and Ryr2, and increased the incidence of spontaneous contraction (Zhao et al., 2015). Converting endogenous cardiac fibroblasts into cardiomyocytes may also alter the biomechanical properties of the heart as fibroblasts typically secrete and maintain cardiac ECM, which is known to affect regenerative outcomes (Notari et al., 2018). The initial study inducing cardiomyocyte conversion was more successful in vivo than predicted based on in vitro observations (Ieda et al., 2010), which suggests that altering the native cell composition and microenvironment of the heart may in turn promote further endogenous changes conducive to regeneration.

The CNCC as a potential internal source of regenerating cells

Much of the work examining potential sources of new cardiomyocytes from non-cardiomyocyte cell types overlook the lineage from which the cells originate from. This has led to cells expressing certain markers being categorized as a single population when more detailed examination has revealed more heterogeneity than previously appreciated. For example, cardiac fibroblasts arise from different lineages, and single cell RNA-sequencing has demonstrated multiple distinct subpopulations (Sadoshima and Weiss, 2014). Sca1+ cells were also able to be separated into 4 populations based on PDGFRa (platelet-derived growth factor receptor α) and CD31 expression, with pro-angiogenic PDGFRa⁺CD31⁻Sca1⁺ cells displaying stemness attributes from multiple cardiac cell types, including cardiomyocytes and endothelial cells (Pfister et al., 2005; Noseda et al., 2015). This raises the possibility that differential myogenic potential of these cell populations may be related to their developmental origin. Proliferating cardiomyocytes in the regenerating heart might thus be heterogeneous (Tang et al., 2019; Sánchez-Iranzo et al., 2018; Kikuchi et al., 2010). Certain lineages of cardiomyocytes might be more amenable to re-enter the cell cycle, or in the case of non-myocyte cell types, perhaps more readily transdifferentiate into cardiomyocytes.

One cell population of interest are cardiac neural crest cells (CNCCs), a multipotent, migratory group of stem cells that contribute to the outflow septum, outflow tract and smooth muscle of the developing heart (Figure 4). In zebrafish, CNCCs are known to contribute to myocardium (Sato and Yost, 2003; Cavanaugh et al., 2015), with recent work extending this finding to amniotic vertebrates, as CNCC-derived cardiomyocytes were found in the chick and mouse (Tang et al., 2019). In the context of heart regeneration, CNCC-derived cells in zebrafish contribute to new proliferative cardiomyocytes both through the proliferation of CNCC-derived cardiomyocytes, but also the de novo cardiomyogenesis of CNCC-derivatives, as they reactivate a molecular signature reminiscent of embryonic neural crest cells (Sande-Melon et al., 2019; Tang et al., 2019). Notably, the most recent claim of a CSC population in the adult mammalian heart is a population of interstitial Twist2+ cells, which appear to contribute to cardiomyocytes, endothelial cells, and fibroblasts (Min et al., 2018). While a relatively small number of Twist2+ cells differentiated into cardiomyocytes in vivo, Twist2+ cells were strongly activated after myocardial infarction, largely at the border of the infarcted zone, contributing to cardiac remodeling by differentiating into endothelial cells and fibroblasts (Min et al., 2018). Twist2 is a member of the Twist-family basic helix-loop-helix (bLHL) transcription factors, which is known to be involved in CNCC migration and cardiac development (VanDusen and Firulli, 2012), raising the possibility that some of these Twist2+ cells may be of CNCC origin. Their plasticity and multipotency both in vitro and in vivo indicate that these cells might be an ideal therapeutic target for direct transdifferentiation.

Figure 4.

Differential heart regeneration ability in P1 and >P7 neonatal mice. At P1, cardiac injury results in regeneration of the injured tissue 21–28 days post injury. However, the same injury at >P7 concludes in a collagenous scar being deposited at the injury site. Some factors that change in the first week of neonatal development include increased cardiomyocyte polyploidy, differential activation of Hippo/YAP signaling, differing cardiac extracellular matrix (ECM) components and mechanical properties, and variation in the timing, duration, magnitude and signaling pathways involved in the immune response to injury.

This notion is supported by other work that identifies Isl1+ CNCCs as a progenitor population in the heart, as Isl1+ CNCCs develop into proliferative cardiomyocytes in the embryonic heart (Hatzistergos et al., 2020). Isl1 is first activated during the specification of precardiac mesoderm, and forms the first and second heart fields, from which most cardiomyocytes develop (Cai et al., 2003; Prall et al., 2007). However, while Isl1+ CNCC derivatives persist in the postnatal heart, the high clonal expansion capacity that they possess in fetal stages rapidly declines, and their numbers and proliferative ability are greatly limited after birth (Hatzistergos et al., 2020). Thus, while the early lineage segregation of nonmyocytes and cardiomyocytes might contribute to the loss of neonatal mouse heart regeneration (Li et al., 2018), the loss of cardiomyocyte proliferative capacity through clonal expansion may also play a role. An open avenue of investigation is why this proliferative capacity diminishes, and how it might be reactivated in the adult mammalian heart.

CONCLUSION

The most stunning exemplars of animal regeneration were once mythologized: the Hydra was named for the many-headed hydra of Greek mythology, and the axolotl’s namesake is the death-defying Aztec deity Xólotl. However, regenerative ability is far from being mysterious and unattainable. The body of research on animal regeneration shows that it is a process orchestrated and modulated by fundamental physiological conditions and cellular interactions, many of which are shared among animals across phylogeny. Though regenerative mechanisms differ across different animals and organ systems, methods of potential intervention—regulating proliferative ability, the immune response, the local microenvironment, and organismal metabolism—arise as broadly similar factors that might play a crucial role in determining regenerative outcome.

To best understand how these interventions would be most effectively deployed, prior work revealing the heterogeneity of regenerative failure is vital. For example, non-regenerative P2 amputation of the mouse digit tip maintains a permissive wound environment and available progenitor cells, but specifically lack the ability to recruit progenitor cells to the wound site (Yu et al., 2010; Yu et al., 2012). In contrast, frog limb blastema cells coalesce at the appropriate location, but are likely epigenetically constrained compared to their axolotl counterparts, failing to fully dedifferentiate and re-express limb bud progenitor genes necessary to form all the tissues in a complete limb (Lin et al., 2021). Careful study of the mechanisms of non-regenerative outcomes in conjunction with regenerative ones provides much needed context to inform where, when, and which interventions might be most successful in changing the nature of a regenerative response.

While specific strategies to modulate these factors are likely to need to be tailored to specific dynamics and cellular processes in each context, previous work has created a robust foundation on which these shared themes can be explored at a finer resolution across diverse systems. Having spent his early career researching animal regeneration, Thomas Hunt Morgan once told the marine biologist Norman Berrill, then a young scientist, that he was being “foolish” for studying regeneration in marine invertebrates, and that “we will never understand the phenomena of development and regeneration” (Berrill, 1983). Perhaps if Morgan could see the body of understanding amassed from the humble capabilities of the common lab mouse today—representing a century’s worth of research and technological advances from when he spoke to Berrill, he might be convinced otherwise.

Highlights.

Many animals can regenerate limbs and organs, an ability lost in most adult birds and mammals

Post-natal mice retain regenerative ability of the heart and digit tip

We discuss possible cellular and molecular signals implicated in regenerative ability