Adult stem cells and their cardiac potential


  • Leonard M. Eisenberg,

    Corresponding author
    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    • Department of Cell Biology and Anatomy, Medical University of South Carolina, BSB Rm 654, 171 Ashley Ave., Charleston, SC 29425
    Search for more papers by this author
    • Fax: 843-792-0664

  • Carol A. Eisenberg

    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author


Adult cardiac muscle is unable to repair itself following severe disease or injury. Because of this fundamental property of the myocardium, it was long believed that the adult myocardium is a postmitotic tissue. Yet, recent studies have indicated that new cardiac myocytes are generated throughout the life span of an adult and that extracardiac cells can contribute to the renewal of individual cells within the myocardium. In addition, investigations of the phenotypic capacity of adult stem cells have suggested that their potential is not solely restricted to the differentiated cell phenotypes of the source tissue. These observations have great implications for cardiac biology, as stem cells obtained from the bone marrow and other readily accessible adult tissues may serve as a source of replacement cardiac myocytes. In this review, we describe the evidence for these new findings and discuss their implications in context of the continuing controversy over stem cell plasticity. Anat Rec Part A 276A:103–112, 2004. © 2004 Wiley-Liss, Inc.

Research on adult stem cells offers great promise for cardiovascular medicine. Although this opening sentence may seem rather bland and noncontroversial, five years ago statements like this were dismissed out of hand. Even today, there is a surprising amount of skepticism expressed concerning the use of adult cells for tissue regeneration. Yet, studies over the past few years have challenged two long-held views about cell biology in the adult, namely, that 1) the adult myocardium is a postmitotic tissue and 2) the potential of adult stem cells is restricted to the differentiated cell phenotypes of the source tissue. In this review, we will provide an overview of the evidence showing the replication and/or regeneration of cardiac myocytes in the adult heart, and that multiple stem cell populations in the adult have myocardial potential. These findings will be discussed in context of the controversy over adult stem cell plasticity. Accordingly, we will present the current interpretation of stem cell plasticity, both by scientists who embrace this concept and by those who advocate a more restricted view of cell lineage diversification. Finally, we offer a different and admittedly speculative view on cell plasticity, which may have relevance for understanding how new cardiac myocytes are generated as a function of an immune cell response.


The adult heart was said for many years to be a postmitotic organ (Soonpaa and Field, 1998; MacLellan and Schneider, 2000; Chien and Olson, 2002). Of course, it was known that the endothelial, smooth muscle, and fibroblast cells of the heart do proliferate. But the cells that make up the meat of the heart—the myocardium—were thought to be terminally differentiated and therefore had lost their proliferative capacity (Claycomb, 1992; Soonpaa and Field, 1998; MacLellan and Schneider, 2000; Chien and Olson, 2002). A major implication of this belief is that the myocytes comprising the adult myocardium are present at birth and are not replenished during an individual's life span. This traditional viewpoint was supported by several lines of evidence. First, studies examining mitosis in the heart showed that the number of myocytes undergoing proliferation was either very low or nonexistent in the adult myocardium. When comparing the heart to other tissues that have obvious high levels of cellular regeneration (e.g., the bone marrow, liver, etc.), it was reasonable to conclude that the myocardium was postmitotic. Second, it has been well established that the myocardium responds to increased load, due to physiological and pathological stress, principally by cellular hypertrophy (increase in cell size) without obvious cellular hyperplasia (increase in cell number). Third, in response to severe injury of the myocardium, as occurs in an infarct, the myocardium is unable to restore functional cardiac muscle within the damaged area (Claycomb, 1992; MacLellan and Schneider, 2000; Frey and Olson, 2003).

Throughout the years, there were several reports suggesting that cells within the myocardium undergo cell division (Linzbach, 1960; Rumyantsev, 1964), thus challenging the prevailing concept that the adult myocardium is postmitotic. However, the idea that the adult heart contains at least some newly regenerated myocytes has only been taken seriously during the past five years (Anversa and Kajstura, 1998; Orlic et al., 2002; Nadal-Ginard et al., 2003). Although this issue still engenders much debate, it is an issue that is now openly debated and no longer thought of as heresy. In large part, this is due to the use of improved methods for detecting dividing cells in the heart, which has yielded data that are more definitive (Kajstura et al., 1998; Beltrami et al., 2001; Quaini et al., 2002). Equally important has been the impact that stem cell biology has had on cardiovascular biology, specifically in regard to adult stem cell plasticity, which has led researchers to consider the idea that new myocardial cells may arise in the adult from stem cells (Hughes, 2002; Orlic et al., 2002; Anversa et al., 2003; Caplice and Gersh, 2003).

Classically, the common method for examining myocardial tissue was by standard histological preparations. Mitotic figures are difficult to observe on these sections, which makes screening for low numbers of cells undergoing mitosis an arduous task. Thus, the unambiguous identification of dividing myocytes in the heart required more modern imaging methods (Kajstura et al., 1998; Beltrami et al., 2001; Quaini et al., 2002). These latter studies employed confocal microscopy to examine heart tissue, which was fluorescently labeled for both muscle proteins and chromatin markers (e.g., propidium iodide). In an initial study, control human hearts (from patients that died from non-cardiovascular-related causes) showed on average 14 myocytes per million that were undergoing mitosis (Kajstura et al., 1998). Among patients who died from either end-stage heart disease or a myocardial infarction, the numbers of cardiac myocytes going through mitosis was 10- to 60-fold higher. In these hearts, the proportion of myocytes that were mitotic ranged from 0.015–0.08% (Kajstura et al., 1998; Beltrami et al., 2001). Since mitosis in most cell types typically transpires within an hour, it is probable that these numbers greatly underestimated the number of newly regenerated myocytes. This was borne out in part by immunohistochemical analysis of the sectioned tissue from these patients with a nuclear marker of cell division (Ki-67), which suggested that the actual frequency of cardiac myocytes that had recently undergone proliferation may have approached 4%.

Further support for regeneration of cardiac myocytes in the adult heart came from multiple studies examining cellular infiltration of donor-transplanted hearts in sex-mismatched recipients (Fig. 1). In these investigations, cardiac biopsies obtained from male patients who had received a female heart were analyzed for the presence of myocytes that possessed a Y-chromosome (Laflamme et al., 2002; Müller et al., 2002; Quaini et al., 2002). Remarkably, the majority of the female donor hearts that were examined contained Y-chromosome-positive myocytes, although the prevalence of these cells within individual hearts differed greatly. In addition to providing further proof that the myocardium contains newly generated cells, these studies indicated that cardiac muscle may be replenished by an extracardiac source of cells (e.g., stem cells). In other words, newly generated cardiac myocytes in the adult heart may be generated de novo, in addition to, or instead of, the replication of preexisting cardiac myocytes. It has not yet been established what nonmyocyte cell populations contribute to replenishing the myocardium in the adult, although evidence suggests circulating hematopoietic stem cells (HSCs) as a potential regenerative source for the myocardium (Jackson et al., 2001; Orlic et al., 2001b).

Figure 1.

Detection of extra-cardiac-derived myocytes in heart transplant patients.


Until recently, it was believed that adult stem cells were restricted in their ability to generate only the differentiated cell phenotypes of the host tissue. Over the past several years, many laboratories have demonstrated that some stem cell populations in the adult possess a phenotypic potential that extends beyond the cell types of their resident tissue (Krause, 2002; Poulsom et al., 2002; Verfaillie et al., 2002). For example, circulating HSCs appear to contribute to the liver, lung, gastrointestinal tract, and blood vessels (Harraz et al., 2001; Krause et al., 2001; Bonnet, 2002). Mesenchymal stem cells (MSCs), the presumptive multipotential progenitors of the bone marrow stroma, have the capacity to supply the blood, lung, liver, and intestine (Pittenger et al., 1999; Jiang et al., 2002). Stem cell populations found in the brain, skin, and adipose tissue also display a previously unsuspected multipotency (Bjornson et al., 1999; Clarke et al., 2000; Zuk et al., 2001; Liang and Bickenbach, 2002). In accordance with these findings, it should not be surprising that extracardiac stem cells can generate cardiac myocytes.

Several adult stem cell populations have demonstrated myocardial potential when injected directly into the adult heart (Fig. 2). Multiple studies have indicated that enriched populations of MSCs when introduced into adult mouse or rat hearts will integrate into the myocardium and exhibit a myocyte phenotype (Tomita et al., 1999; Wang et al., 2000; Toma et al., 2002). Although the overall integration of these cells was low, MSC-derived cells within the myocardium exhibited sarcomeric myosin heavy chain, cardiac troponin I, desmin, and α-actinin protein expression, as well as connexin43 gap junctions. Another study reported that liver-derived stem cells will also exhibit myocardial potential when transplanted into an adult mouse heart (Malouf et al., 2001). Analysis by electron microscopy demonstrated that these cells integrated into the myocardium, as indicated by their formation of gap junctions and intercalated discs with adjacent endogenous myocytes, and displayed well-formed sarcomeres and myofibrils. However, the most impressive result reported to date employed bone marrow-derived multipotent HSCs to restore cardiac function of injured adult mouse hearts (Orlic et al., 2001a). The donor cells, obtained from green-fluorescent protein (GFP)-expressing transgenic mice, were purified by their absence of lineage-specific markers and expression of c-kit (Linneg, c-kitpos)—a procedure that is thought to yield highly enriched populations of multipotent HSCs. The HSCs were then injected directly into the myocardium of host mice that suffered myocardial infarcts by coronary ligation. In this study, significant repair of the damaged myocardium was observed in 12 of 30 mice that were stem cell injected. Moreover, among the hearts that demonstrated a positive response, on average, 53% of the newly regenerated cardiac myocytes in the infarcted area appeared to be derived from the injected HSCs.

Figure 2.

Stem cells obtained from adult bone marrow exhibit myocardial potential. Bone marrow-derived stem cells will contribute to cardiac myocyte regeneration when either injected directly into the heart (Tomita et al., 1999; Wang et al., 2000; Orlic et al., 2001a; Toma et al., 2002) or exposed to myocardium via the circulation (Bittner et al., 1999; Jackson et al., 2001).

One implication of the cardiac competency of HSCs is that these cells may be a normal source of newly generated cardiac myocytes in the adult, by virtue of their exposure to myocardial tissue during circulation. A similar phenomenon appears to occur for other tissues, and recent data seem to suggest that circulating progenitors may also contribute cells to the adult heart. Experimental support for this phenomenon has been indicated by studies introducing labeled cells into the circulation and examining their subsequent integration into heart tissue (Fig. 2). In one study (Jackson et al., 2001), lethally irradiated mice were provided with HSCs selected by their exclusion of the fluorescent dye Hoechst 33342, the so-called side population (SP) cells. After verifying positive engraftment of the injected cells in the circulating blood, based on their expression of a β-galactosidase marker, the mice were rendered ischemic by coronary occlusion. Thereafter, donor cell incorporation into the myocardium was examined for α-actinin expression. Based on data obtained from five mice that survived this procedure, approximately 0.2% of the myocytes that surrounded infarcted areas were β-galactosidase positive and thus were derived from cells that had been implanted into the circulation. Similar results were observed following bone marrow transplantations into dystrophic mdx mice, which suffer from both skeletal and cardiac muscle degeneration, with donor cells showing incorporation into the myocardium (Bittner et al., 1999). As these data suggested that circulating stem cells can give rise to new cardiac myocytes, it was postulated that administering cytokines to increase the number of stem cells in the circulation may promote the healing of damaged myocardium (Orlic et al., 2001b). To test that hypothesis, both stem cell factor (SCF) and granulocyte-colony-stimulating factor (G-CSF) were injected into adult mice over an eight-day period, during which a myocardial infarct was induced. After four weeks, these mice showed a marked decrease in infarct size and increased numbers of replicating cardiac myocytes, with a significant enhancement in both survival rates and cardiac function, compared to control mice that were not provided with SCF and G-CSF. Whether the increased myocyte proliferation provoked by the cytokine treatment resulted directly from the recruitment of circulating stem cells to the myocardial lineage was not addressed by this study. Since others have shown that circulating stem cells can contribute to vascular structures, it is conceivable that increased blood vessel formation may have enhanced the capacity of cells resident to the myocardium to regenerate the tissue, and thus facilitate the improvement in cardiac health. Although that determination will have to await further investigation, this latter study does provide further support for the contribution of circulating cells to the maintenance of the adult heart.


The observation that extracardiac progenitors (e.g., HSCs) possess myocardial potential is in accordance with many published reports indicating that adult stem cells have a very broad phenotypic potential. Yet, the notion of adult stem cell plasticity, often referred to as transdifferentiation, still engenders great skepticism (Hawley and Sobieski, 2002; Wagers et al., 2002; Wells, 2002; Wurmser and Gage, 2002; Medvinsky and Smith, 2003). Despite the multitude of reports supporting a broad potentiality of adult stem cells, stem cell plasticity is still often thought to be an unproven phenomenon requiring more rigorous scientific proof. One of the suggested criteria to rigorously establish stem cell plasticity is that “transplanted stem cells should give rise to robust and sustained regeneration of the target tissues” (Anderson et al., 2001). In the adult, the normal rate of cardiac myocyte replacement is far from being robust, although presumably high enough to maintain the normal homeostatis of the myocardium throughout an individual's life span. While not proven, good evidence has been reported suggesting that circulating HSCs may generate replacement myocytes in the adult heart. Thus, the contribution of HSCs to myocyte replacement may be a physiological reality, but would not meet the criteria of robustness that is inappropriate to the adult myocardium.

Another point of contention concerns recent observations that tissue-derived stem cells can undergo fusion with other cell types (Terada et al., 2002; Ying et al., 2002; Vassilopoulos et al., 2003; Wang et al., 2003), which has been used as evidence that cell fusion artifacts may account for previous results purporting to show transdifferentiation of adult progenitors (Hawley and Sobieski, 2002; Wurmser and Gage, 2002; Medvinsky and Smith, 2003). Because this argument is often cited to question reports on adult stem cell plasticity, it is perhaps warranted to address this issue in detail.

The connection between cell fusion and stem cell plasticity was first proposed in simultaneously published reports of hybrid cell formation from co-cultures of tissue-derived stem cells and embryonic stem (ES) cells (Terada et al., 2002; Ying et al., 2002). The tissue-derived stem cells were obtained from neural or bone marrow tissue of mice carrying both GFP and antibiotic resistance (puromycin or G418) genes. Following their mixture with ES cells, the cultures were treated with the appropriate antibiotic, under standard conditions developed to maintain ES cells in a nondifferentiating state (Fig. 3A). A few weeks later, small numbers of individual GFP-positive clones were generated, which exhibited the potentiality of ES cells when placed under differentiation conditions. However, these GFP-positive ES-like cells were multinucleated and contained markers from both starting cell populations (i.e., the tissue-derived stem cells and the ES cell line), which could only occur if the resulting clones were hybrids of the two cell types. More recently, two studies published in tandem reported that cell fusion may account for the contribution of circulating cells to the repair of damaged liver tissue (Vassilopoulos et al., 2003; Wang et al., 2003). In these studies (Fig. 3B), mutant male mice deficient for the liver enzyme fumarylacetoacetate hydrolase (FAHneg) were irradiated to deplete endogenous hematopoietic cells and then injected with HSCs from FAHpos female mice. After engraftment of the donor hematopoietic cells, liver damage was promoted in the chimeric mice by withdrawal of the drug 2-(2-nitro-4-trifluoro-methyl-benzyol)-1,3-cyclohexanedione (NTBC), which is normally provided to FAHneg mice for preventing fatal metabolic liver disease. Several weeks later, cytogenetic, genomic blot, and immunohistochemical analyses revealed that the apparent fusion of donor FAHpos cells and FAHneg host hepatocytes appeared to have played a significant role in the production of newly generated liver tissue.

Figure 3.

Experimental protocols employed for producing cell fusion of tissue-derived stem cells. A: Mixed cultures of ES cells and antibiotic-resistant tissue-derived stem cells produced colonies of hybrid cells in the presence of antibiotic (Terada et al., 2002; Ying et al., 2002). B: The contribution of HSCs to the repair of diseased liver. For these experiments (Vassilopoulos et al., 2003; Wang et al., 2003), the blood system was reconstituted in lethally irradiated host FAHneg mice using wild-type FAHpos donor HSCs. Following hematopoietic cell engraftment, liver failure in the chimeric mice was promoted by withdrawal of the drug NTBC. Subsequently, liver tissue was regenerated and examined for the formation of polyploid cells that exhibited markers of both donor and recipient origin.

Based on the cell fusion results, the inference has been made that previous reports of adult stem cell plasticity were instead observing cell fusion phenomenology. Moreover, evidence of cell fusion has been portrayed as an invalidation of the concept of stem cell transdifferentiation (Wurmser and Gage, 2002; Medvinsky and Smith, 2003). However, there is more than enough ambiguity in the experimental models employed to give one pause before jumping to any premature conclusions. For example, the tissue-derived stem cell/ES cell co-cultures employed severe selection conditions that appear to have allowed only for the survival of fused cells (Fig. 3A). Obviously, the antibiotic-sensitive ES cells would not survive in the presence of either puromycin or G418, and therefore, their survival was dependent on the acquisition of the antibiotic resistance gene from the tissue-derived stem cells. In addition, control data were not provided in either study showing that the tissue-derived stem cells cultured by themselves would thrive under the conditions used in these studies (which had been optimized to maintain ES cells in a nondifferentiating state). Since both reports described the emergence of individual clones after a few weeks of co-culture (Terada et al., 2002; Ying et al., 2002), it may be surmised that the fusion with ES cells was also necessary for the survival of the tissue-derived stem cells. Yet, even under these severe selection conditions, the resulting cell fusion event was rare, with the frequency of cell fusion among bone marrow stem cells ranging from 2–11 events per million cells (Terada et al., 2002). Although it may be argued that the conditions used in these studies were required to reveal a previously unsuspected property of certain cell types, why is it presupposed that the formation of hybrid cells observed in these studies was due to abnormalities of tissue stem cells and not of the ES cells? After all, the maintenance of ES cells in a nondifferentiated state for extended periods in culture is not physiological, and may possibly introduce abnormal properties.

Although the studies with the FAH mutant mice have provided compelling evidence for the importance of cell fusion in liver repair (Vassilopoulos et al., 2003; Wang et al., 2003), there are issues peculiar to liver homeostasis and the FAH mutant mouse model that should preclude a hasty judgment on its relevance to stem cell differentiation. It is well recognized that high numbers of polyploid cells are often generated in the adult liver in response to injury and disease (Brodsky and Uryvaeva, 1977; Sigal et al., 1999; Gorla et al., 2001). While this is usually thought to involve DNA synthesis without cytokinesis, the recent demonstration of cell fusion in a mouse model of pervasive liver disease warrants a reexamination of the mechanism of polyploid cell formation in the liver. Moreover, an unusually high percentage of liver cells in FAHneg mice exhibit an abnormal karyotype (Wang et al., 2003)—with 34% of hepatocytes in the mutant mice being aneuploid, as compared to 2% aneuploidy in wild-type strains. Thus, it appears that the hepatocytes in FAHneg mice exhibit abnormal properties, which may make them more prone to cell fusion. Interestingly, cytogenetic analysis indicated that 6–15% of the hepatocytes within the chimeric mice following the engraftment of female FAHpos HSCs had a normal female karyotype. In other words, there was evidence that hematopoietic cells transdifferentiated to hepatocytes, though this data did not draw any comment in the text. This positive evidence of transdifferentiation suggests the intriguing possibility, which is fully consistent with the data, that donor-derived cells underwent cell fusion after undergoing transdifferentiation to hepatocytes.

These new studies on hybrid cell formation suggest that cell fusion may play a greater biological role than previously presumed. The zygote, osteoclasts, and skeletal muscle are examples of cells whose derivation has been definitively established to involve cell fusion. The new data indicate that hepatocytes, stem cells, and possibly other cell types may also exhibit phenotypes that result from cell fusion. Although the cell fusion findings may have revealed the capabilities of tissue-derived stem cells to form hybrid cells, there is little justification to presume that this attribute is only relevant to transdifferentiation. Rather, the ability to undergo fusion may be a physiological property of stem cells that also comes into play when they give rise to their normal progeny. Moreover, there is no reason to presume that cell fusion and transdifferentiation of stem cells are contrary phenomena. After all, the formation of skeletal muscle involves both cell differentiation and fusion.

In regards to the multiple studies showing the myocardial potential of adult stem cells, there is no evidence that would suggest cell fusion as the mechanism allowing these cells to generate cardiac myocytes when introduced into the heart directly or via the circulation. Since some of these studies counterstained their tissue sections with propidium iodide to stain nuclear chromatin, evidence of heterokaryon formation should have been obvious (Orlic et al., 2001a). In a recent study, we investigated the capability of HSCs to directly generate cardiac myocytes in culture (Eisenberg et al., 2003). Our data indicate that in response to growth factor treatment, HSCs obtained from chick bone marrow will undergo cardiac differentiation, as indicated by their expression of sarcomeric myosin (Fig. 4A and B), desmin, α-actinin, smooth muscle actin, and the cardiac-associated transcription factors Nkx2.5, GATA4, GATA5, and eHAND (Fig. 4C). Since the cultures contained only cells harvested from bone marrow, and thus cardiac myocytes were not present in the starting cell population, cell fusion could not account for the experimental outcome. Thus, our culture data unambiguously show that under the appropriate conditions, hematopoietic cells can convert to a cardiac cell fate by a normal pathway of cell differentiation.

Figure 4.

Myocardial differentiation of bone marrow cells in culture. HSCs obtained from chick bone marrow were treated for four days with retinoic acid (RA), dexamethasone (dex), prostaglandin E2 (PGE2), interleukin-2 (IL2), fibroblast growth factor-4 (FGF4), and bone morphogenetic protein-4 (BMP4), and immunostained for sarcomeric myosin (A and B) or harvested for RNA and examined for cardiac-associated gene expression by reverse transcriptase-polymerase chain reaction (RT-PCR) amplification (C) (Eisenberg et al., 2003). A and B: Fluorescent images of representative cultures of treated bone marrow HSCs, which display clusters of sarcomeric myosin-positive cells exhibiting a morphology characteristic of cultured cardiomyocytes. C: RNA was harvested from bone marrow-derived HSCs that were cultured in the absence of growth factors (GF) (lane 1); presence of RA, FGF4, and BMP4 (lane 2); or presence of RA, Dex, PGE2, IL2, FGF4, and BMP4 (lane 3). After four days, RNA was RT-PCR amplified for the cardiac-associated transcription factors Nkx2.5, GATA4, GATA5, and eHAND. Note that the complete GF cocktail (lane 3) promoted cardiac differentiation of HPCs, as shown by positive expression of the cardiac genes. In contrast, neither Nkx2.5, GATA4, GATA5, nor eHAND was exhibited in cultures that were treated either in the absence (lane 1) or with a subset (lane 2) of the GF cocktail. Equivalent expression of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase demonstrates that each sample contained equal concentrations of template RNA.


Why would cells derived from noncardiac tissues have the potential to generate cardiac myocytes? To begin to answer this question, let us first consider the standard view of stem cells and cell lineage diversification (Anderson et al., 2001; Orkin and Zon, 2002; Verfaillie, 2002). As defined, stem cells are self-renewing, nondifferentiated cells that have the ability to produce large numbers of descendant differentiated cells. What distinguishes different populations of stem cells is their potential to generate multiple types of specialized cells, in other words, whether stem cells are totipotent, pluripotent, multipotent, bipotent, or unipotent. According to this model, individual cell lineages arise by pathways of progressive restriction of cell potential, with the originating stem cells undergoing a series of commitment steps that irreversibly generate in sequence pluripotent, multipotent, unipotent, and differentiated progeny (Fig. 5A).

Figure 5.

Schematic diagrams depicting the diversification of cell lineages from stem cells. A: Traditional concept of cell phenotypic diversification showing the hierarchy of stem cell subsets, with individual cell lineages arising by progressive restriction of stem cell potential. According to this view, these differentiation pathways are unidirectional and irreversible. This diagram displays a compressed version of the traditional model of stem cell biology, as it depicts in sequence the generation of pluripotent, multipotent, unipotent, and differentiated progeny. B: Plasticity model of stem cell differentiation. According to this hypothesis, stem cell transdifferentiation is a real, but infrequent phenomenon that occurs when changes in the surrounding tissue environment redirect the phenotypic potential of multipotent stem cells. By limiting both the scope and prevalence of transdifferentiation, this model of limited plasticity attempts to preserve the hierarchy of stem cell subsets of the traditional model. Moreover, by defining plasticity as a reprogramming event, this hypothesis maintains the distinction between normal differentiation and transdifferentiation.

In response to the recent findings concerning the broad phenotypic potential of adult stem cells, variations on the traditional model of stem cell diversification have been proposed (Orkin and Zon, 2002; Verfaillie, 2002). In the plasticity model (Fig. 5B), stem cells possess great flexibility in their phenotypic potential, with their eventual cell fate controlled by extracellular signals. Thus, changes in the surrounding tissue environment redirect stem cell differentiation. For example, multipotent stem cells that normally give rise to the blood lineages (i.e., multipotent HSCs) may be reprogrammed to multipotent neural stem cells when exposed to a different in vivo environment. This plasticity can be represented schematically as lateral movement among individual lineage differentiation pathways (Fig. 5B). An alternative viewpoint suggests that stem cell plasticity may be an illusion reflecting the high degree of heterogeneity among stem cell populations within adult tissues. This hypothesis preserves the hierarchical structure of the traditional model of stem cell diversification (Fig. 5A), but adds the corollary that embryonic-derived pluripotent and multipotent stem cell populations persist, and subsequently undergo self-renewal, within the adult tissue microenvironment. According to this developmental heterogeneity hypothesis, reports on the capacity of HSCs to give rise to nonblood phenotypes (e.g., cardiomyocytes) are due to contaminating pan-mesodermal stem cells within preparations of HSC-enriched bone marrow cells.

In deciding which model may be more relevant for understanding cell potential in the adult, it is important to understand the biological implications of each model. Both the traditional and developmental heterogeneity models (Fig. 5A) require total fidelity to a hierarchical, unidirectional, and irreversible pathway of cell differentiation. If instead phenotypic commitment of stem cells is not fixed, then the distinction between stem cell subsets (e.g., pan-mesodermal and HSCs) becomes blurred, which relieves the necessity of invoking developmental heterogeneity to explain the plasticity of adult stem cell potential. By representing a shift in stem cell fate as a reprogramming event, the plasticity model (Fig. 5B) attempts to retain the distinction among stem cell subsets (i.e., pluripotent, multipotent, bipotent, and unipotent). However, the flexibility allowed under this model has its limits since either shifting of phenotype among highly committed stem cell subsets (i.e., unipotent) or a reversibility in the directionality of cell commitment (i.e., dedifferentiation) would compromise the concept that stem cell subsets can be arranged in a hierarchical pathway. Yet, as discussed below, individual examples have been reported where the acquisition of cell phenotype is generated by nonhierarchical pathways—suggesting that neither the traditional/developmental heterogeneity nor plasticity models can satisfactorily explain how differentiated tissues arise in the adult.


In current scientific discussions, the term transdifferentiation is usually employed to refer to stem cell plasticity. For example, HSCs that produced cardiac myocytes would be considered to have undergone transdifferentiation. Yet, this definition of transdifferentiation has been co-opted from an older meaning of the term that was used to describe the conversion from one differentiated cell type to another. Transdifferentiation of differentiated cells has been demonstrated definitively during limb regeneration in the amphibian, and the conversion of pigment epithelia into lens and neural retinal cells (Tsonis, 2000; Del Rio-Tsonis and Tsonis, 2003; Nye et al., 2003). In these instances, differentiated tissue dedifferentiates into cells with an apparent stem cell phenotype, prior to their metamorphosis into other differentiated cell phenotypes. Transdifferentiation has also been reported in mammals, such as the conversion of pancreatic cells to hepatocytes, hepatocytes to pancreatic cells, and vascular endothelium to smooth muscle (Frid et al., 2002; Shen et al., 2003). Additionally, we have shown (Eisenberg et al., 2003) that macrophages derived from mouse bone marrow can transdifferentiate to a cardiac myocyte phenotype when cultured in the presence of myocardial tissue (Fig. 6).

Figure 6.

Transdifferentiation of macrophages into cardiac myocytes. Macrophages were obtained from adult mouse bone marrow, labeled with the green fluorescent marker carboxyfluorescein diacetate succinimidyl ester (CFDA-SE), and cultured with fragments of contractile embryonic chick myocardium (Eisenberg et al., 2003). Six days later, cultures were immunostained for cardiac muscle proteins (red) and examined for integration of macrophage-derived cells into the myocardial tissue. A: Low-magnification view showing distribution of contractile sarcomeric myosin-positive heart fragments (red; arrowheads) within the co-cultures. B: Individual cardiac fragment that integrated sarcomeric myosin-positive mouse macrophage-derived cells (arrow) within the tissue. C: A heart fragment that displayed macrophage-derived cells (arrows) or cell clusters (asterisk) that were smooth muscle α-actin positive. D: An individual explant stained for desmin and shown at low magnification, with the box indicating the position of the higher magnification view shown in panel E. E: Arrows indicate desmin-positive macrophage-derived cells (arrow) within the wall of the embryonic heart fragment. F: Heart explant that exhibited macrophage-derived cells that have integrated into the myocardium and express sarcomeric α-actin (arrows). Scale bars: A, 160 μM; B, 16 μM; C, 20 μM; D, 50 μM; E, 15 μM; F, 18 μM.

The generation of newly differentiated cells in the adult is significantly enhanced during the wound healing response. It is postulated that in response to tissue damage, stem cells are specifically recruited to the wound site (Theise et al., 2000; Harraz et al., 2001; Mahmood et al., 2001). However, it has been observed that cellular regeneration is enhanced within tissues when the immune response is activated, as occurs when tissue or cells are transplanted into a host (Krause et al., 2001; Laflamme et al., 2002; Quaini et al., 2002). For example, in a study examining cardiac tissue from male patients who received a heart transplant from female donors, the greatest amount of Y-positive cardiac myocytes was observed in regions of acute rejection (Laflamme et al., 2002). These results are intriguing since immune response cells comprise the initial wave of cells recruited to the wound site, as they are required as a first step in halting more serious trauma to the injured area. If immune cells can serve as a source of new cells for repairing damaged tissue, than their transdifferentiation would mitigate the need to mobilize a second cell population (i.e., stem cells) to the wound. This hypothesis is consistent with our published results (Eisenberg et al., 2003), which indicate that macrophages invading myocardial tissue can contribute to the formation of new myocytes (Fig. 6).

Further evidence supporting the transdifferentiation of immune response cells has been shown in studies of monocytes. Although monocytes are commonly thought to be the immediate progenitors of the macrophage, several studies have demonstrated that these cells are able to transdifferentiate to endothelial cells (Fernandez Pujol et al., 2000; Harraz et al., 2001; Schmeisser et al., 2001). For example, monocytes will contribute endothelial cells to new blood vessels that form in limbs recovering from injury (Harraz et al., 2001). A recent study has further suggested that the phenotypic potential of monocytes may extend to other nonblood lineages (Zhao et al., 2003). These findings on the broad potential of monocytes, which is generally considered to be a cell that is fully committed to myeloid blood lineages, has significant implications for stem cell biology. According to standard hierarchical models of stem cells, lineage diversification occurs at the level of multipotent stem cells. Yet, if unipotent progenitors can be redirected into multiple cell fates, then what is the distinction between unipotent and multipotent stem cells?

The ability of both differentiated cells and cells characterized as highly committed progenitors (e.g., monocytes) to transdifferentiate into other cell phenotypes suggests that current models of cell diversification may not adequately represent cell phenotypic acquisition. Although transdifferentiation of differentiated cells has been shown conclusively in only a limited number of instances (Tsonis, 2000; Del Rio-Tsonis and Tsonis, 2003; Nye et al., 2003), the definitive evidence in these cases demonstrates that regenerated tissue does not always arise from hierarchical pathways of differentiation from stem cells. Yet, these examples of transdifferentiation are often considered special cases with little relevance to a discussion of stem cell biology and tissue regeneration. However, new evidence (Frid et al., 2002; Shen et al., 2003) suggests that transdifferentiation has a larger relevance for understanding mammalian biology then is now currently thought. Thus, models of cell phenotypic diversification that incorporate transdifferentiation and dedifferentiation, by including both lateral and bidirectional arrows between cell phenotypes and lineages (Fig. 7), may better reflect the acquisition of cell phenotype. Although rates of differentiation may exceed those of transdifferentiation and dedifferentiation between phenotypes, thereby providing directionality to lineage acquisition, the implication of a transdifferentiation model of cell diversification is that the totality of cell phenotype within an organism is a part of a continuum.

Figure 7.

Transdifferentiation model of cell phenotypic diversification. In this model, we incorporate transdifferentiation of both stem cells and differentiated cells (bidirectional arrows), as well as dedifferentiation of cell phenotypes (reverse arrows). As indicated by the gray arrows, the rates of dedifferentiation or transdifferentiation are in many instances much less than the forward differentiation rates (black arrows)—thereby providing a semblance of phenotypic directionality. However, events have been characterized where the incidence of dedifferentiation and transdifferentiation is very high, such as during limb regeneration in the amphibian, and the conversion of pigment epithelia into lens and neural retinal cells (Tsonis, 2000; Del Rio-Tsonis and Tsonis, 2003; Nye et al., 2003). During those events, cell phenotypic diversification may be represented as a cycle or continuum, instead of discrete lineage pathways.


In this review, we have described the evidence that new cardiac myocytes are generated throughout the life span of an adult and that extracardiac cells can contribute to the renewal of individual cells within this tissue. Yet, as is well known, new myocytes are not normally generated in large enough numbers to fully repair a severely damaged myocardium. Thus, there is a need to uncover cellular supplies that can be used for producing replacement myocardium. Unfortunately, attempts to identify sources of adult cells that could generate new cardiac myocytes inevitably get caught up in the overall controversy on whether stem cell plasticity is a reality. However, as discussed in this essay, a broader consideration of the scientific literature reveals that plasticity of cell potential not only exists, but may even extend beyond classically defined multipotential and pluripotent stem cells.

In previous publications (Eisenberg et al., 2003; Eisenberg and Eisenberg, 2003), we have hypothesized that the plasticity of cell potential may be related to how new cell phenotypes may have been generated throughout evolution. Accompanying the development of more complex organisms during evolution, more distinct types of specialized cells were also produced.

This spawning of new specialized cell types may have resulted by selective mechanisms acting on either stem cells or cells exhibiting differentiated phenotypes. Among these two possibilities, the emergence of new differentiated cell types by variation from existing differentiated phenotypes would appear to be a more direct and less complicated mechanism for promoting evolutionary diversification. If variation of existing specialized cells played a role in the emergence of phenotypic diversification during at least the earliest stages of multicellular evolution, then remnants of transdifferentiation may have persisted in higher species. Thus, a division among differentiated cell types may be apparent according to their early or late appearance during evolution. Accordingly, the more primitive, early-evolved phenotypes would possess a greater capacity to transdifferentiate into other specialized cell phenotypes.

The most primitive vertebrate muscle phenotype is the nonstriated smooth muscle cell, which is also the predominant myocyte contained in invertebrate species (Elphick and Melarange, 2001). In the vertebrate embryo, cardiac myocytes of the tubular heart initially express several molecules that are hallmarks of differentiated smooth muscle cells, including smooth muscle α-actin, smoothelin, and SM22α (Ruzicka and Schwartz, 1988; DeRuiter et al., 1993; Li et al., 1996). Expression of these smooth muscle proteins is then turned off as cardiac muscle undergoes further differentiation. This transient display of a smooth muscle phenotype by myocardial cells may be evidence of an evolutionary linkage among these cell types, with the cardiac myocyte being derived as a variation on the earlier evolved smooth muscle cell.

Evolutionary relationships among differentiated phenotypes may not necessarily be restricted to cells possessing similar functional attributes. In adult tissues, the greatest amount of newly generated differentiated cells occurs during the wound healing response. Thus, the immune response cells, which are normally recruited to damaged tissue, may themselves be a contributing cellular source for repairing the wound site. In addition, immune response cells play a role in normal tissue homeostasis by their removal of dead and diseased cells (Mevorach, 1999). The earliest evolved immune response cell is the macrophage, which appears in many primitive organisms (Gross et al., 1999; Iwanaga, 2002). It was for this reason that the focus of our recent work has been on the transdifferentiation potential of macrophages (Eisenberg et al., 2003). Based on the prevalence of macrophage infiltration into adult cardiac tissue, we believe that the macrophage would be a reasonable candidate as a cellular contributor to cardiac myocyte renewal in the adult.