Concise Review: The Role of C-kit Expressing Cells in Heart Repair at the Neonatal and Adult Stage

Authors

  • Michael Hesse,

    1. Institute of Physiology 1, Life and Brain Center, University of Bonn, Bonn, Germany
    Search for more papers by this author
  • Bernd K. Fleischmann,

    Corresponding author
    1. Institute of Physiology 1, Life and Brain Center, University of Bonn, Bonn, Germany
    • Correspondence: Michael I. Kotlikoff, V.M.D., Ph.D., Department of Biomedical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, New York 14853-6401, USA. Telephone: 607-253-3771; Fax: 607-253-3701; e-mail: mik7@cornell.edu; or Bernd K. Fleischmann, M.D., Institute of Physiology 1, Life and Brain Center, University of Bonn, 53105 Bonn, Germany. Telephone: 492286885200; Fax: 492286885201; e-mail: Bernd.Fleischmann@uni-bonn.de

    Search for more papers by this author
  • Michael I. Kotlikoff

    Corresponding author
    1. Department of Biomedical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, New York, USA
    • Correspondence: Michael I. Kotlikoff, V.M.D., Ph.D., Department of Biomedical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, New York 14853-6401, USA. Telephone: 607-253-3771; Fax: 607-253-3701; e-mail: mik7@cornell.edu; or Bernd K. Fleischmann, M.D., Institute of Physiology 1, Life and Brain Center, University of Bonn, 53105 Bonn, Germany. Telephone: 492286885200; Fax: 492286885201; e-mail: Bernd.Fleischmann@uni-bonn.de

    Search for more papers by this author

Abstract

Ischemic heart disease is the number one cause of morbidity and mortality in the developed world due to the inability of the heart to replace lost myocytes. The cause of postinfarction myogenic failure has been a subject of intense scientific investigation and much controversy. Recent data indicate a brief perinatal developmental window exists during which postinfarction myogenesis, and substantial heart regeneration, occurs. By contrast, repair of an equivalent injury of the adult heart results in prominent revascularization without myogenesis. Here, we review recent experiments on neonatal postinjury myogenesis, examine the mechanistic hypotheses of dedifferentiation and precursor expansion, and discuss experiments indicating that postinfarction revascularization derives primarily from cardiac vascular precursors. These data have profound consequences for the understanding of human heart repair, as they address the long standing question as to whether human postinfarction myogenic failure is due to the loss of precursors existent at the neonatal stage or to a context-dependent inhibition of these precursors within the infarct, and suggest strategies for the recapitulation of neonatal myogenic capacity and the augmentation of revascularization. Stem Cells 2014;32:1701–1712

Introduction

Our understanding of the regenerative capacity of the human heart has undergone antipodal shifts over the past 150 years of medical discovery. In the 19th century two major diverging views emerged regarding the plasticity of the human heart and its reaction to injury: (a) the heart increases the number of muscle fibers (formation of new cardiomyocytes), or (b) it augments the size of muscle fibers (growth of individual cardiomyocytes) [1]. In the early 20th century, Wideröe [2] and Karsner et al. [3] suggested that hyperplasia is unlikely to explain adaptive increases in heart size due to the absence of mitotic figures in cardiomyocytes; rather these changes were solely attributed to hypertrophy and atrophy of cardiomyocytes, whereas neocardiomyogenesis was thought to be negligible. However, several reports claimed that hyperplasia of cardiac myocytes occurs during pathological hypertrophy exceeding the critical heart mass of 500 g [4, 5], and after myocardial infarction in the atria of rats [6] and ventricles of humans [7].

This scientific controversy was informed by the adoption of new techniques such as the identification of stem cell markers, fate mapping and more recently 14C dating, setting off a lively discussion in the field of cardiac regeneration. 14C dating in the laboratory of Jonas Frisen determined the turnover of cardiomyocytes in adult human hearts by isolating cardiomyocyte nuclei to be about 0.45% per year [8], but the experimental data also indicated that the regenerative capacity is higher early in life. The analysis of the 14C data was complicated, because only DNA synthesis could be measured and therefore the turnover rate estimate is based on a mathematical model with several assumptions, including earlier measurements of binucleation in human cardiomyocytes [9]. A similarly low estimate of turnover had previously been reached by Loren Fields' group in mice, by estimating DNA synthesis using radioactive thymidine and assuming that every positive cardiomyocyte will eventually undergo cell division [10].

Another “conventional” approach using pHH3 and MKLP-1 as a marker for cytokinesis determined an annual turnover rate of 1.9% in hearts from 20 years old humans [11]. While in this study, cell cycle activity as well as an increase in nuclear ploidy could be detected in older hearts, cytokinesis and cell division were undetectable. More recently, a more sensitive method for detection of DNA synthesis by incorporation of 15N indicated an annual turnover rate of 0.76% and a DNA synthesis rate of 4.4% per annum in adult mouse cardiomyocytes [12], which could be attributed to the higher sensitivity of the used multi-isotope imaging mass spectrometry. Higher numbers (1.3%–4%) were obtained by BrdU incorporation studies [13], again providing only indirect evidence for cardiomyocyte renewal. However, much higher rates (up to 80% in mice [14] and 23% in humans [15]) have also been reported, using BrdU incorporation [14] and 14C dating [15], and different modeling assumptions. In this context, it is interesting that the turnover of cardiomyocytes in the adult heart has been claimed to be increased by a factor of 4.5 under pathological conditions, such as dilated cardiomyopathy and ischemic cardiomyopathy [15]. An increase in ploidy after cardiac lesion as has been reported for men [16, 17] and mice [12, 18], as well as the mathematical model underlying 14C data analysis could explain the discrepancies in cardiomyocyte renewal rates [19, 20].

One constant in this ongoing controversy has been the agreement that, whether or not the adult mammalian heart retains significant regenerative potential, cardiac infarction results in permanent damage in the infarct area, markedly increasing the risk of contractile and/or electrical failure. In contrast to the adult mammalian heart, a remarkable capacity for postinfarction regenerative neomyogenesis has been reported in the neonatal rat heart, during a period of physiological cardiac expansion [21] and recently in mice [22, 23], and the comparison of postinfarct regenerative capacity between neonates and adults indicates fundamental differences [23]. The identification of the molecular cues underlying these development-dependent differences in regenerative potential of the mammalian heart could be also helpful in manipulating and enhancing the plasticity of the adult heart muscle.

Two mechanisms have been advanced to explain neomyogenesis in the injured neonatal heart: (a) dedifferentiation of resident cardiomyocytes, cell cycle re-entry and cell division [22]; and (b) progenitor cell pool expansion and myogenic differentiation [23]. In this review, we will focus on the existing data in support of each of these mechanisms, contrast repair processes between the neonatal and adult mammalian heart, and highlight areas in which critical gaps remain in our understanding.

Cardiovascular Precursors and c-kit

Multipotent cardiovascular progenitor cells (CPc) have been identified in ESC-directed differentiation studies, and also in the embryonic and neonatal heart [24-29]. While the existence of CPc in the adult heart is controversial, the above cited studies clearly indicate that during heart development a common precursor cell responds to specific cues within the developing heart to form all the cardiovascular lineages in the heart (cardiomyocytes, endothelial cells, and smooth muscle cells). ESCs can be induced to mimic the developmental paradigm of progressively cardiac restricted development from undifferentiated mesoderm, and the subsequent formation of all three cardiovascular lineages (Fig. 1A) [24-26, 30, 31].

Figure 1.

Cardiovascular progenitor cell and c-kit. (A): Directed differentiation of ESCs to cardiovascular fates occurs through several stages. Several studies have shown that c-kit is expressed relatively late in this process and may identify a stage between cardiac mesoderm and individual cardiovascular cell fates. (B): SCF/c-kit signaling underlies stem or precursor cell maintenance, differentiation, and proliferation in a variety of organ contexts. (C): Bright-field and fluorescent images of PN0 c-kitBAC-EGFP pups. Bar is 1 mm. (D): (c-kit)EGFP expression (brown) in the heart wall of PN3 neonates. In the adult heart rare immunostained cells were observed. Bar is 25 µm. (E): (c-kit)EGFP+ cells in the atrioventricular region showing a gradient of EGFP expression, from less differentiated bright cells (arrows), to less fluorescent α-actinin+ cells. Some (c-kit)EGFP+ cells coexpress PECAM1 or Flk-1, indicating endothelial commitment. (F): Representative fluorescence-activated cell sorting from PN c-kitBAC-EGFP and WT littermate hearts. WT and EGFP histograms were normalized to same maximum. tEGFP overlaps with brightly autofluorescent WT cells; sEGFP identifies a pure (c-kit)EGFP+ population of cells five times brighter than background fluorescence. Merged bright-field and fluorescent image show after 1 day in culture some tEGFP cells form clusters. Nodal-like morphology of action potentials was observed in some of those cells. Bar is 50 µm. Pictures (C), (D), (E), and (G) are taken from [28]. (G): Cardiac-resident (c-kit)EGFP+ cells (green) from either neonatal or adult ACT-EGFP hearts were cocultured with fetal cardiomyocytes. Staining for α-actinin (red) revealed cardiomyogenic differentiation of (c-kit)EGFP+ cells from neonatal (left) but not from adult hearts (right). Bars are 20 µm. Pictures are taken from [28, 29] reproduced with permission from Wolters Kluwer Health. Abbreviations: ESC, embryonic stem cell; SCF, stem cell factor.

While no single marker exclusively identifies CPc at all developmental stages, there is strong agreement that expression of the type III tyrosine kinase receptor c-kit (CD117) identifies a stage between formation of cardiac mesoderm and differentiation into the specific cardiovascular lineages [24-26, 30]. Autophosphorylation of the c-kit receptor is triggered by the binding of stem cell factor (SCF), which is expressed as a soluble or membrane bound splice variant. SCF activation of c-kit, which occurs either by the binding of free ligand or by heterotypic cell-cell interactions, underlies progenitor maintenance, differentiation, proliferation, and migration in hematopoietic [32], germ [33], melanocyte [34], and other lineages [35, 36] and c-kit activating mutations promote unregulated cell growth and tumor development [37] (Fig. 1B). Little is known, however, with respect to the precise signals promoting vascular (endothelial and smooth muscle) versus cardiac formation, although lineage-specific transcription factors, such as the myogenic program factors NKX2.5, ISL1, and TBX [25-27], are activated during this process. The transient and overlapping nature of expression of early multilineage markers, as well as their presence in other developmental contexts, limits the usefullness of any single marker as sufficient to establish CPc status. This point is particularly important with respect to the expression of c-kit, the marker most commonly associated with cardiac and cardiovascular precursors. While expression of c-kit had been reported during heart cell specification and appears to be involved in heart repair [38, 39], the receptor is expressed by numerous undifferentiated and terminally differentiated cell types, such as basket neurons in the hippocampus [40], interstitial cells of Cajal in the intestine [41], hematopoietic progenitor cells [42], mast cells [43], and Leydig cells and spermatogonia in the testis [44] and, as indicated above, plays a prominent role in the development of most organ systems, most notably in hematopoiesis, pigmentation, reproduction, and oncogenesis. In an effort to further understand the role of c-kit expression during heart development and repair, we developed a BAC transgenic indicator mouse that expresses Enhanced Green Fluorescence Protein (EGFP) under full transcriptional control of the kit locus (c-kitBAC-EGFP mice) [28]. These mice have enabled classic cell labeling experiments in the heart that track expression during mammalian developmental stages, autologous bone marrow reconstitution experiments to discriminate marrow-derived and intrinsic (i.e., cells within the heart) c-kit expression using c-kitBAC-EGFP mice as marrow donors or acceptors, cardiac tissue transplantation to scid mice, and the fluorescence-activated cell sorting (FACS) purification and in vitro differentiation of cells in which c-kit is transcriptionally active [23, 28].

Cardiac Neomyogenesis in the Injured Neonatal Heart

In c-kitBAC-EGFP mice (Fig. 1C), EGFP is robustly expressed in the embryonic and neonatal heart, but expression rapidly falls after birth such that by postnatal day 14 (PN14) very few (c-kit)EGFP+ cells are found in the myocardium, and expression is virtually absent in the normal adult heart (Fig. 1D). As predicted by the data indicating that c-kit is expressed at various stages during the specification of cardiovascular cells, labeled cells are found within the neonatal heart in multiple states of differentiation (Fig. 1E), but many cells also exist in an undifferentiated state, that is, negative for differentiation markers used to identify the three cardiovascular lineages. Importantly, neonatal (c-kit)EGFP+ cells can be purified by FACS (Fig. 1F) and clonally differentiated in vitro into all cardiovascular lineages, indicating that the population of neonatal (c-kit)EGFP+ cells contains tripotent CPc [28]. Similar results were reported by Zaruba et al. [29] using anti-c-kit antibodies to purify cells by FACS and magnetic-activated cell sorting (MACS). Using both in vitro and transplantation methods, this group found that c-kit+ cells isolated from neonates, but not adults, have cardiomyogenic potential (Fig. 1G, taken from [29]). It should be noted, however, that the gene expression status of neither authentic cardiovascular precursors, nor of nonstriated committed cardiac precursors derived from them, is known, a fact of particular relevance for fate mapping by recombination, using lines expressing Cre recombinase under control of specific promoter constructs, as discussed below.

The existence of undifferentiated, neonatal (c-kit)EGFP+ cells that are capable of cardiac differentiation in vitro enabled us to address a long debated question in the field of cardiac repair: does the failure of postinfarction myogenesis result from a context-dependent inhibition of regeneration within the infarction, or rather is it due to a lack of competent precursors in the adult mammalian heart? To address this question, we developed an infarction model in the neonatal heart beginning in 2008, and initiated a series of studies to compare the cellular and regenerative response to the identical cryolesion in neonatal and adult mice. As shown in Figure 2A, the infarcted neonatal mouse heart undergoes a robust and localized recruitment of c-kit+ cells (recruitment because all cells within the infarcted region are cryo-ablated), followed by myogenesis and revascularization of the infarct by (c-kit)EGFP+ cells. Both c-kit and Nkx2.5 gene transcription are induced within the infarcted area, and increasing numbers of these cells adopt cardiac myocyte fates over the course of a week postinjury [23]. Recruited (c-kit)EGFP+ cells undergo significant expansion, as seen by pulse-chase BrdU labeling and pHH3 staining. Strikingly, the myogenic response can also be observed ex vivo when infarcted cardiac tissue is immediately transplanted to the subcutaneous tissue of scid mice, indicating that the myogenic response does not require external cells (Fig. 2B). By contrast, adults infarcted in an identical fashion undergo far less robust c-kit induction, do not increase Nkx2.5 transcription, or display any evidence of myogenesis within the infarction in vivo or when tissue is explanted to scid mice, although the infarction is revascularized almost completely by c-kit cells derived from the heart, as discussed below.

Figure 2.

Postinjury myogenesis in the neonatal heart. (A): Induction of (c-kit)EGFP+ cells in neonatal and adult heart at 3 days after cryoinfarction (D3). Merged fluorescent/bright-field images show localized fluorescence within the neonatal infarct. Bar is 2 mm. Expression of (c-kit)EGFP in infarcted neonatal heart (upper right). Bar is 100 µm. Striated (c-kit)EGFP+ myocytes within neonatal infarct (arrows) by D3; some cells are also Nkx2.5+ (lower right). Bar is 20 µm. (B): Scheme for transplantation and (c-kit)EGFP induction and myogenic/endothelial differentiation in neonatal c-kitBAC-EGFP infarcted tissue in SCID mouse. (Left) Low magnification of infarcted area. (Right) Higher magnification shows clusters of striated (c-kit)EGFP+ myocytes (arrows) as well as endothelial cells. Bar is 25 µm. (C): Experimental scheme for purification of cardiomyocytes and (c-kit)EGFP+ cells and example of purified (c-kit)EGFP population (red/green fluorescence overlay). Lower picture shows a spontaneous beating cell obtained from FACS-purified α-mCherry/(c-kit)EGFP+ cells which were differentiated in cardiac media for 11 days. (D): Myogenic fate of neonatal c-kitBAC-EGFP/Rosa-β-gal cells transplanted within adult infarct. X-Gal-positive myocytes are shown. Bar is 25 µm. (E): Fate mapping of myocytes (c-kitBAC-EGFP/αMHC-CreER/R26-floxed STOP-βGal) at D5 shows recombination outside of infarct zone. Graph shows recombination rate of (c-kit)EGFP+/lacZ+ cardiomyocytes in the infarcted area normalized to recombination rate of lacZ+ cardiomyocytes in the remote area from three separate experiments. Bar is 200 µm. Asterisks indicate significance as determined by the unpaired t test: p < .001. Pictures are taken from [23]. Abbreviation: FACS, fluorescence-activated cell sorting.

While these studies were underway, Porrello et al. published an example of regenerative myogenesis at this developmental stage [22]. This group followed heart repair in PN1 mice after excision of the heart apex, and concluded that sarcomeric myocytes divide, regenerating the excised heart apex in a process that involves dedifferentiation, sarcomeric disassembly, and re-entry into the cell cycle [22], and mirrors recent findings in zebrafish regeneration [45, 46]. The mechanistic determination of dedifferentiation of what are assumed to be terminally differentiated myocytes depends on the result of fate mapping using an inducible Myh6-MerCreMer mouse [47]. Induction of recombination at PN1 resulted in the tagging of most (approximately 70%) myocytes within the area of regrown myocardium. This group has subsequently shown that ischemic injury results in a similar percentage of fate mapped myocytes in the infarcted region [48].

The pattern of expansion of cells observed through BrdU pulse chase or pHH3 staining in these experiments is also of interest, as injury in the studies by Porrello et al. provokes an equivalent mitotic response at sites remote to the injury. While we also observed BrdU and pHH3-labeled cells diffusely spread throughout the neonatal heart, our studies in which c-kit+ cells could be tracked and separately assessed indicated that the infarction zone contained an almost sevenfold greater density of (c-kit)EGFP+ cells, and that by day 5 following the injury, more than 30% of these cells were BrdU+, with very few (c-kit)EGFP+/BrdU+ cells outside of the infarct [17]. To some extent, the differences between these studies may reflect the type of injury imposed, as tissue excision in the Porello et al. study would minimize the inflammatory response to cell death, and the attendant cellular influx, observed in cryo or anoxia-induced injury models; similar differences of cardiac regeneration in response to different injury types have been also reported in zebrafish [49]. The two mouse studies together suggest, however, that injury provokes a diffuse mitotic response throughout the heart, but that c-kit-expressing cells constitute an expanding pool that is concentrated at the site of injury.

Thus, our data indicate that a subset of c-kit+ cells migrate to the site of injury, undergo mitosis, and form new myocytes in the neonatal heart. Three critical pieces of evidence support the mechanistic basis of neomyogenesis as involving nonstriated c-kit+ CPc: (a) cardiac lin (c-kit)EGFP+ cells form within the infarct and can be differentiated in vitro to beating cardiac muscle cells (Fig. 2C) [23]; (b) cardiac lin (c-kit)EGFP+ cells differentiate to striated myocytes, when transplanted to adult hearts (Fig. 2B) [23]; and (c) the percentage of (c-kit)EGFP+ myocytes within ablative myocytes that are lacZ+ following inducible recombination with a myh6-Cre transgene is significantly lower than the overall recombination rate of myocytes (Fig. 2E) [23]. However, the work of Porrello et al. suggests a more generalized activation of mitosis in previously differentiated cardiac myocytes throughout the neonatal heart following injury, and our results using similar methods were equivalent in the cryoinjury model. A distinct limitation of our work is that the transcriptional activity (c-kit)EGFP marker cannot definitively identify the original lineage of EGFP+ cells. Thus, it is possible that observed undifferentiated (c-kit)EGFP+ cells derive from previously differentiated cardiac myocytes that have been induced to express c-kit, a finding that would be consistent with the induced recombination lineage tracing experiments described above. To address this question, we crossed c-kitBAC-EGFP mice with a line expressing mCherry under control of a cardiac-specific αMHC promoter fragment, thus marking all cardiac myocytes. These mice were infarcted at PN1 and cells from the infarcted region of the heart separated by two color FACS within 6 hours of injury to prevent dedifferentiation of myocytes, allowing us to separately track the fate of green-only (i.e., αMHC) cells and red-only (i.e., c-kit) cells in vitro. In these experiments, green-only cells turned red-only (i.e., become mCherry+ and EGFP) and began to beat in culture (Fig. 2C); conversely, red-only cells displayed no green-only conversion [23]. Although these experiments do not exclude the dedifferentiation and expansion of existing cardiac myocytes, they provide strong evidence of the ability of undifferentiated cardiac precursors to support neomyogenesis at this developmental stage.

Nonetheless, the fate mapping experiments of Porrello et al. and similar experiments in our laboratory indicate that most myocytes found within the regenerated areas of the neonatal heart are derived from αMHC-expressing cells. While the experiments of Porrello et al. did not attempt to address the contribution of undifferentiated CPc to myocyte regeneration, and do not exclude the migration of recombined heart cells to the site of injury, our data indicate that approximately 30% of new cardiomyocytes are not derived from existing myh6+ cardiomyocytes, but most likely from c-kit+ CPc (Fig. 2E). However, several factors limit the degree to which fate mapping can distinguish between precursor and differentiated myocyte expansion in the context of these experiments. First, induced recombination with the αMHC-MerCreMer mouse does not distinguish existing myocytes from those derived from mitosis, complicating the interpretation of events within the area of injury, which is affected by remodeling and cell migration. Second, the specificity of myocyte labeling depends entirely on the restriction of the driver Cre mouse to differentiated cells, and here one must be cautious regarding the interpretation of this result. αMHC transcripts are detected as early as E7.5 in the early heart tube before the formation of mature myocytes [50], as well as in early myocyte developmental stages found in embryonic carcinoma and embryoid body cells [51, 52], and the stage of expression of the Myh6 fragment used to create Myh6-MerCreMer mice [22] has not been shown to be restricted to the sarcomeric stage of heart cells (i.e., not expressed in earlier precursor stages), and may be active in existing precursors. Third, the temporal context of fate mapping complicates these experiments, because the studies of Porrello et al. involved the delivery of subcutaneous tamoxifen at birth, followed by injury 24 hours later. This protocol almost certainly involves some overlap between recombination and repair, raising the possibility that precursor cells responding to the injury were fate mapped following induced differentiation. These concerns highlight the importance of an enhanced understanding of the precursor population at this developmental stage, and the development of genetic reagents (e.g., Cre transgenic lines) that could be used for to fate map these cells.

The fate mapping experiments of Porrello et al. have led to the conclusion that the regenerative capacity of the neonatal heart derives from the ability of mammalian cells to “dedifferentiate” [22], analogous to the process of regeneration within the zebrafish heart, which does not rely on stem cells, but rather on the ability of striated, mononuclear heart cells to dedifferentiate and re-enter the cell cycle [45], which underlies the plasticity and regenerative capacity of multiple zebrafish organs [53]. In mammalian heart cells, this process is reported to involve a phenotypic remodeling of the myocyte, including sarcomere disassembly as well as the re-expression of cell cycle genes. These features are normally observed in dividing mammalian cardiomyocytes during embryonic development [54, 55], however, and the extent to which this reflects a distinct process provoked by injury remains unclear. Considering that physiologic myocyte division is occurring at this developmental stage and that there has been no clear demonstration of developmental reversion, the characterization of neonatal, postinjury myogenesis as identical to chordate regeneration would seem to require a more in depth molecular and cell biological characterization. This is particularly true since pathologies such as dilated cardiomyopathy and myocardial infarction involve the induction of an embryonic gene program, which may or may not indicate full conversion to a dedifferentiated phenotype with the potential of proliferation [28, 56], and as induction of an embryonic gene program and DNA synthesis may occur without increased rates of proliferation [56].

Despite these limitations, the findings outlined above narrow the developmental window of cells supporting neomyogenesis to the developmental states of: (a) multipotent CPc; (b) early committed myogenic precursors; (c) immature cardiac myocytes; and (d) mononuclear sarcomeric heart cells, all of which might be tagged by the αMHC cre-recombinase used to mark differentiated myocytes in our studies and those of Porrello et al. Thus, the conclusion that neonatal sarcomeric heart cells, like those of zebrafish, retain the capacity to dedifferentiate and enter the cell cycle awaits definitive evidence of heart cell dedifferentiation. Likewise, the status of c-kit+ cells as definitive CPc awaits more specific fate-mapping studies.

Postinjury Myogenesis in the Adult Heart

Regeneration of adult heart cells following injury and their potential origin remains controversial. While some groups, including our own, have not found noteworthy neoformation of cardiomyocytes after cardiac lesions in mice [18, 29], others claim regeneration rates of cardiomyocytes as high as 15.3% [13, 57, 58]. Several excellent reviews of these conflicting data have been published over the past several years [59-61]. One consistent feature associated with studies of adult heart repair has been the focus on c-kit expressing cells, and in many cases the assumption that the expression of c-kit indicates heart cell CPc status.

In a recent transcriptome analysis, cardiac c-kit+ cells were identified as the most primitive cell population in the rodent heart compared to Sca-1+ and side population cells [62]. Resident c-kit+ cardiac stem cells that are able to undergo cardiomyogenesis have been claimed to exist in a variety of species, including humans [63] and rats [64]. In support of this notion, it has very recently been reported that isoproterenol-induced heart lesions lead to the formation of new cardiomyocytes by c-kit+ cardiac stem cells in adult rats and mice, resulting in a remarkable cardiac regenerative potential [65]. However, these studies often rely on lineage tracing using limited Cre-expressing mouse lines as discussed above, and numerous conflicting studies have been reported using different methods such as visualization of cell division [18] as well as 15N detection by mass spectrometry [12].

Using c-kitBAC-EGFP transgenic mice, we demonstrated that, while there is an activation of c-kit expression within resident cells of the heart following infarction, this induction is overwhelmingly associated with neovasculogenesis rather than cardiac myogenesis [23]. A small number of differentiated cardiomyocytes in the border zone display low levels of c-kit expression; however, these cells are nonmitotic and do not incorporate BrdU or express cell cycle markers. Unlike neomyogenesis following an ablative neonatal infarct, c-kit+ cells isolated from adult infarctions do not form beating myocytes in vitro and cardiac myogenesis is not observed in tissue explanted to scid mice [17]. These experiments are consistent with studies from Loren Field's group in which transplantation of cardiac c-kit+ cells, isolated by MACS sorting according to the same protocol as in the studies reporting high myogenesis [66], did not undergo myogenesis in vitro nor in vivo (Fig. 1E) [29]. Thus, following infarction of the adult heart there is a marked induction of c-kit in cells intrinsic to the heart, but genetically tagged c-kit+ cells lack significant myogenic potential, while playing a major role in neovasculogenesis. Moreover, the prominent expression of c-kit in vascular endothelial cells in virtually all new vessels within an infarction (Fig. 3A, 3B), clearly invalidates the common experimental assumption that c-kit expression is a marker for cardiac precursor or stem cells following injury.

Figure 3.

Heart-derived (c-kit)EGFP+ cells adopt myogenic and EC fates following neonatal infarction, but only vascular fates following adult infarction. (A): CD45+ cells surround (c-kit)EGFP+ cells within infarct and do not participate in angiogenesis/myogenesis. (Right Up) New vessel within infarct (a) formed by (c-kit)EGFP+ endothelial cells; vessel outside of infarct (b) has no incorporation. (Right down) Myocyte clusters formed by (c-kit)EGFP+ cells, surrounded by EGFP−/CD45+ mononuclear cells. Bar is 200 µm in (A) Left; 50 µm in (A) Center and Right. (B): Infarction of reconstituted ((pCAGGS)dsRed marrow) c-kitBACEGFP mice indicates (c-kit)EGFP+ cells adopt endothelial (PECAM-1) fates; bone marrow-derived cells are mainly CD45+. Bar is 500 µm in (B) Left; 100 µm in (B) Right. (C): Re-expression of c-kit-EGFP after injury in the adult heart (left). (Right) Fluorescent and nonfluorescent striated cardiomyocytes. (D): EGFP cells within infarct. (Left) Fluorescence wide-field image of adult c-kitBAC-EGFP heart 14 days postcryoablation. (E): Adult c-kitBAC-EGFP infarcted tissue in SCID mouse undergoes (c-kit)EGFP induction, but adopt strictly vascular fates (Right). Bar is 50 µm. Pictures taken from [23, 28].

While it is difficult to explain discrepant reports regarding neomyogenesis by c-kit+ cells in the adult mammalian heart, a key factor appears to be the methods used to identify proliferating and dividing cardiomyocytes. The expression of cell cycle markers within injured tissue is commonly used to identify proliferating cells, but this method is complicated by two factors. First, as these markers are localized in the nucleus, they must be unequivocally determined to be within myocytes, rather than in proximity to cardiomyocyte cytoplasm, as has been reported [60, 67, 68]. The use of transgenic [69] or immunochemical [70] cardiac-specific markers appears to be a more robust experimental approach. Second, cardiomyocytes are known to undergo variations of the cell cycle, such as endoreduplication (no cytokinesis and no karyokinesis, resulting in polyploid cells) and acytokinetic mitosis (no cytokinesis, resulting in binucleated cells) [71, 72]. To distinguish these processes from completed cell division, cells in late M-phase must be unequivocally demonstrated, whereas many studies use incorporation of thymidine analogs or expression of the early cell cycle marker Ki-67, which are not cytokinesis specific [68, 73]. Similarly, while pHH3 is a more specific M-phase marker, it does not discriminate between acytokinetic mitosis and cell division [18]. Detection of cytokinetic structures, such as the use of antibodies against Aurora B kinase to identify the contractile ring and midbody, is a more specific detector of cell division; however, this approach is known to be technically challenging. The generation of transgenic mice overexpressing anillin, a scaffolding protein of the contractile ring [18], fused with EGFP provides a simple method to identify authentic dividing cardiomyocytes.

Neovasculogenesis in the Neonatal and Adult Heart

As mentioned above, both neonatal and adult hearts respond to injury by extensively revascularizing the area of infarction. In the adult, this response is coincident with the induction of c-kit expression [28, 29]. When identical infarctions of neonatal and adult hearts are examined, one finds that c-kit is induced far less robustly in the adult than in the neonate, however, perhaps reflecting a regenerative response that is restricted to neovascularization (Fig. 2A). In contrast to the response in the neonatal myocardium, however, the developmental fate of (c-kit)EGFP+ cells is exclusively vascular in the infarcted adult heart. That is, while c-kit-tagged cells participate in the formation of new heart cells and blood vessels in the neonate, only vascular formation by these cells is observed in the adult following the identical cryoinjury [23] (Fig. 3B–3E).

As (c-kit)EGFP+ cells are not found in the adult, one possible explanation for these results is that revascularization occurs through the recruitment of cells that are distinct from endogenous CPc, such as bone marrow-derived, c-kit+ mesenchymal stem cells that have previously been reported to track to the injured myocardium [38]. However, full thickness cryo-ablation of adult myocardium in c-kitBAC-EGFP transgenic mice resulted in a prominent pattern of EGFP-labeled new vessels surrounded by CD45+, unlabeled cells. As shown in Figure 3A, essentially all new vessels are lined with EGFP+ endothelial cells, and less than 3% of these were CD45 positive; similarly, CD45+ cells are not observed within vessels, but rather surround areas of neovascularization. To definitively determine the source of (c-kit)EGFP+ endothelial cells, we reconstituted irradiated c-kitBAC-EGFP mice with dsRed bone marrow. Infarction of chimeric (dsRedmarrow/c-kitBAC-EGFP) adult mice indicated that c-kit+ cells adopt predominately vascular fates in vivo, and that marrow-derived cells invade the infarct, but do not form cardiovascular lineages (Fig. 3B). Conversely, adoptive transfer of c-kit(EGFP)+ bone marrow cells does not result in c-kit(EGFP)+ labeled cells adopting vascular fates. Finally, immediate implantation of infarcted adult tissue from c-kitBAC-EGFP mice into scid mice resulted in extensive new vessel formation by (c-kit)EGFP+ endothelial cells (Fig. 3C). In contrast to the transplantation of neonatal infarcted c-kitBAC-EGFP myocardium, no cardiac myogenesis is observed in adult mice.

These experiments raise important questions regarding the cellular basis of postinfarction revascularization. For example, is revascularization-dependent on undifferentiated precursors or an angiogenic process associated with sprouting from existing vessels through mitosis of differentiated endothelial cells (and an attendant induction of c-kit in those cells), or is the process the result of induction of latent vascular precursors? If c-kit+ vascular progenitors drive this process, are these cells more lineage-restricted descendants of neonatal CPc, and thus might they be ideal targets for directed myodifferentiation? Are endothelial and vascular smooth muscle cells derived from the same precursor population, or does revascularization occur by activation of separate precursors, such as resident Sca1+ smooth muscle progenitor cells [74]? Do local vascular precursors derive from proepicardial cells (see also below) that reside as undifferentiated, perivascular cells [75]? Finally, these results should emphasize that while the isolation of lineage negative (lin), c-kit+ cells often assumes that these cells have authentic CPc status (i.e., myogenic potential), it is clear that such cells can have markedly different lineage potential and that the developmental outcome of these distinct pools must be separately determined.

As mentioned above, also stem cells/progenitors originating from the epicardium are thought to be of relevance for repair upon injury. In fact, neovascularization after myocardial infarction was also observed to originate from stem cells residing in the epicardium, which are defined by expression of Wilms' tumor protein 1 (Wt1) [76]. After myocardial infarction, the epicardial layer expands and gives rise to endothelial cells and vascular smooth muscle cells [77]. These cells are distinct from c-kit+ stem/progenitor cells, although a c-kit+ subpopulation has been identified among the Wt1-positive cells [78], which needs further verification [79]. Vascularization after lesion is greatly enhanced by stimulation with the actin sequestering protein thymosin β4 (Tβ4) [76, 80]. Lineage tracing using Wt1-Cre expressing mice has shown that Wt1+ cells also contribute to cardiomyogenesis using Tβ4 stimulation before induction of myocardial infarction [79]. However, the extent of cardiomyogenesis induced by this approach is very limited and a therapeutic approach would need an additional external source of cardiomyocytes [81].

Discussion

While the extent of cardiomyocyte renewal in adult mammalian hearts remains an area of remarkable controversy, our results using a BAC transgenic c-kit indicator mouse demonstrate that the response to heart injury is myogenic during the first week of life, but that the capacity for neomyogenesis is limited to early neonatal stages [22, 23]. The difference in myogenic potential is most clearly shown in studies in which an identical lesion is imposed, and an identical analysis undertaken, at neonatal and adult time points [23]. As shown in Figure 4, the neonatal response to infarction is marked by balanced myogenesis and vasculogenesis, whereas adult injury provokes a virtually exclusive vasculogenic response, despite the induction of c-kit expression within endogenous heart cells. The vascular repair response within the adult heart occurs by the activation of endogenous vascular precursors, or by the angiogenic sprouting of existing vessels, whereas circulating or marrow-derived precursors do not appear to play a prominent role in the formation of new endothelial cells. These findings directly address the long standing question as to whether the lack of substantial myogenesis following adult myocardial infarction relates to a context-dependent inhibition of endogenous myogenesis within the infarction, as proposed by those who estimate high turnover in the adult heart, or an intrinsic loss of myogenic capability in the adult. As an identical infarction results in myogenesis in the neonate, but myogenic failure in the adult, the condition of infarction is clearly not the limiting factor. The finding of myogenic repair of the neonatal heart has appropriately focused attention on the identification of the cellular source(s) of neomyogenesis, and whether this process is confined to the expansion of existing cardiomyocytes [12], induced formation of new cardiomyocytes from CPc [57, 64], or a contribution of both sources [13]. Three factors both inform and complicate this analysis. First, the neonatal heart is undergoing a basal level of new heart cell formation in the absence of injury, and increases in myocyte formation may simply reflect the expansion of an existing process in immature myocytes, rather than the assumption of a more unique dedifferentiation of mature, quiescent myocytes [82, 83]. Second, authentic c-kit+ CPc are forming myocytes and vascular cells at this developmental stage [23, 28, 29], complicating the determination of their role in infarction repair, and the lack of an inducible c-kit/mer-cre-mer mouse prevents definitive determination of progeny of these cells. Third, given the numerous roles of c-kit in biological processes, expression of this receptor is not sufficient to identify precursor, much less heart cell-specific precursor, status. Despite these confounding factors, the enrichment of c-kit+ cells at the site of infarction, the high percentage of these cells that incorporate BrdU, and the high proportion of c-kit+ cells that are striated myocytes within, but not outside, the infarction, argues for a role for these cells in injury provoked myogenesis. Because of the local accumulation of c-kit+ cells at the site of the neonatal infarct, it would be interesting to identify and characterize the cells expressing SCF, the c-kit-ligand, and determine the functional importance of c-kit expression. Other organs such as the bone marrow or testis exhibit an interaction of SCF with c-kit, which is of crucial importance for either stem cell maintenance or differentiation, and for maintenance of differentiated progeny. In the bone marrow, maintenance of c-kit+ Hematopoetic Stem Cell (HSC) is mediated by SCF, which is expressed by perivascular cells of the stem cell niche [84]. In contrast, in the testis, the SCF/c-kit interaction is crucial for differentiation and maintenance of spermatogonial cells, but not undifferentiated spermatogonial stem cells [85]. In this setting SCF expressed in Sertoli cells has an antiapoptotic function, inducing c-kit expression in differentiated type A spermatogonia [86] thus enabling spermatogenesis.

Figure 4.

Schematic depiction of the role of c-kit+ cardiovascular progenitor cells during regeneration after cardiac lesion. Neonatal hearts response to infarction by induction of (c-kit)EGFP+ cells which are capable of differentiation into cardiomyocytes, endothelial cells, and smooth muscle cells, thereby contributing to myogenesis and vasculogenesis. In adult hearts, injury leads to a weaker induction of (c-kit)EGFP+ cells, which differentiate into endothelial and smooth muscle cells, but not cardiomyocytes, resulting in a virtually exclusive vascular repair response. The loss of myogenic capability in the adult is an intrinsic phenomenon and not due to an inhibition of endogenous myogenesis caused by the infarction. Images modified from Patrick J. Lynch, medical illustrator; C. Carl Jaffe, MD, cardiologist; http://patricklynch.net Creative Commons Attribution 2.5 License 2006.

Although SCF expression has been reported in cardiac tissue [87], neither the functional role nor the identity of cells in which c-kit or SCF is induced has been determined. This is particularly true for the adult heart in which c-kit expression is low relative to the neonatal infarction, but markedly induced in vascular cells following injury. By analogy to processes of renewal in bone marrow and testis, SCF/c-kit homotypic or heterotypic interactions could play a role either in the maintenance of a pool of undifferentiated CPc, or rather underlie the induction of differentiation of endothelial, smooth muscle, and cardiac cells derived from CPc after lesion in neonates as well as the induction of endothelial and smooth muscle cells from vascular precursors after cardiac lesion in adults. Studies examining the precise functional role c-kit expression in the neonatal and adult context may have important relevance for therapeutic strategies seeking to reactivate cardiomyocyte proliferation or induce the differentiation of resident cardiac precursors. Such reactivation could be a more promising strategy to promote heart repair than cell replacement due to the known difficulties associated with the long term survival, functional integration, and immunological tolerance of engrafted cells [61, 88].

Finally, these studies shine some light on the somewhat neglected process of revascularization of damaged heart tissue. Vascular repair in postnatal as well as adult infarcted mouse hearts occurs by activation of endogenous vascular precursors, and/or angiogenic sprouting of existing vessels, associated with the expression of c-kit. The therapeutic importance of this vascular response may be significant. A recent phase I clinical study using c-kit+ cells derived from the hearts of patients suffering from postinfarction left ventricular dysfunction was encouraging [89, 90]; however, such clinical trials need to be performed in a randomized, double-blind, and placebo controlled fashion, in order to provide definitive information on their therapeutic potential. In analogy to this phase I trial, the same group transplanted c-kit+ cells into infarcted pig hearts as a large animal model and observed an improvement in cardiac function in the treated group, mimicking the results obtained in humans [91]. Consistent with a functional vascular fate of c-kit+ cells, this study identified vascular structures formed by engrafted cells. The mechanism underlying the improvement in cardiac function associated with c-kit+ transplantation in these studies has not been established, but may well relate to enhanced vascularization of the injured heart.

Precursor or stem cells derived directly from the bone marrow do not form new vessels and are therefore excluded as a source for cardiac plasticity [23], although these cells may play an important role in vascular induction. Bone marrow-derived c-kit+VEGFR+ cells have been transiently observed in mouse hearts after infarction [38, 92] and are suggested to modulate the vascularization response upon myocardial infarction by release of Vascular Endothelial Groeth Factor (VEGF) [38, 92], but not via the formation of new endothelial or smooth muscle cells. The stage appears set for a significant advance in our understanding of vascular formation within the damaged heart, which may have important therapeutic consequences [93-95].

Acknowledgments

This work was supported by NIH 00536501 and 11270781 (to M.I.K.), and EU FP7 consortium CardioCell Grant No223372 (to B.K.F.), and the Deutsche Forschungsgemeinschaft FOR 1352 (to B.K.F.).

Author Contributions

M.H. and B.K.F.: conception and design and manuscript writing; M.I.K.: conception and design, manuscript writing, and final approval of the manuscript.

Disclosure of Potential Conflicts of Interest

The authors declare no conflict of interest.

Ancillary