Heart failure is a major cause of mortality worldwide with a steady increase in prevalence. There is currently no available cure beyond orthotopic heart transplantation, which for a number of reasons is an option only for a small fraction of all patients. Considerable hope has therefore been placed on the possibility of treating a failing heart by replacing lost cardiomyocytes, either through transplantation of various types of stem cells or by boosting endogenous regenerative mechanisms in the heart. Here, we review the current status of stem and progenitor cell-based therapies for heart disease. We discuss the pros and cons of different stem and progenitor cell types that can be considered for transplantation and describe recent advances in the understanding of how cardiomyocytes normally differentiate and how these cells can be generated from more immature cells ex vivo. Finally, we consider the possibility of activation of endogenous stem and progenitor cells to treat heart failure.
Heart disease is the leading cause of mortality in the industrialized world . The prevalence of myocardial infarction (MI) is steadily increasing because of an ageing population and lifestyle changes leading to increased obesity. Improved treatments for acute MI have resulted in more survivors, but with compromised heart function. Consequently, the prevalence of heart failure, a condition characterized by a decrease in contractile capacity below a critical threshold , is increasing steadily, with 23 million new patients diagnosed worldwide every year . There is currently no curative treatment for heart failure, with the exception of heart transplantation, which for many reasons will be applicable for only a very limited proportion of all patients. In addition, there are a number of congenital malformations of the heart (e.g. hypoplastic left heart syndrome, characterized by a critical reduction in myocardial mass) for which novel therapeutic options are needed. A major underlying problem is that the cells in the heart of humans and other mammals have very limited capacity for self-renewal in response to injury, which is in contrast to the more widespread regenerative capacity in lower vertebrates, such as zebrafish (recently reviewed in ). Therefore, new modes of therapy are warranted. The potential of using various types of stem cells for heart repair is an attractive prospect.
Cell types for transplantation
The prospect of repairing an injured heart with cells that can be cultured and expanded ex vivo and then functionally integrated upon transplantation is appealing. A number of different cell types for transplantation have been considered. For a successful outcome, there are several factors that need to be taken into account, including the survival, engraftment and functional integration of transplanted cells within the injured myocardium, and the ability to avoid rejection of the transplant by the immune system of the recipient. In addition, during a myocardial infarct, up to 109 cardiomyocytes, corresponding to 25% of the total number in the heart, are lost due to ischaemia. Thus, a large number of cells need to be replaced, raising the issue of scalability of any cell population used for transplantation . Two fundamentally different sources of cells, non-cardiac and cardiac cells, that may be suitable for transplantation will be discussed below.
Non-cardiac cells for cardiac therapy
Skeletal muscle cells
Because of the relative abundance of satellite cells (i.e. the stem cell population in skeletal muscle) as well as the functional and histological similarities between cardiac and skeletal muscle, animal models were initially used to investigate grafted satellite cells to experimentally injured hearts in rabbit  and sheep . These and other early studies showed improved cardiac function following transplantation, although the mechanism(s) underlying this effect were not completely understood. These initial positive results led to early clinical trials, including the randomized, multicentre MAGIC study. The results of this clinical study revealed that treatment with autologous myoblasts failed to improve cardiac function, but did increase the incidence of cardiac arrhythmias . Data from experiments in various animal models showed that grafted myoblasts fail to adopt the cardiomyocyte fate; rather, they follow their normal developmental route and differentiate to skeletal muscle cells [9, 10]. Thus, it appears from the findings of satellite cell transplantations and the MAGIC trial that cells lacking a documented cardiac potential cannot be expected to differentiate to cardiomyocytes.
Bone marrow-derived cells
It has also been suggested that various cell populations from the bone marrow may be suitable for cardiac transplantation. Such cells demonstrated considerably broader developmental potential than originally recognized, including the ability to differentiate to cardiomyocytes [11, 12]. Coupled with the capability to expand such cells in vitro, and the proven safety of infusing bone marrow-derived cells, such cell populations emerged as attractive candidates for regenerative cardiac therapy. Many studies were rapidly initiated to examine the efficacy of such treatment. To date, more than 1000 patients have been treated with non-cardiac cells, particularly bone marrow-derived cells, for various cardiac conditions (for a recent meta-analysis of more than 20 such trials see ). Table 1 shows a selection of clinical trials in which cellular therapies have been evaluated as treatment for different forms of cardiac disease. The results have been variable, ranging from no beneficial effect  to a small but statistically significant effect on left ventricular ejection fraction . Two of the most comprehensive studies to date, REPAIR-AMI and STAR-heart, demonstrated an approximately 3% increase in left ventricular ejection fraction in the acute  and chronic  setting respectively. The STAR-heart study further showed a small decrease in mortality . Although such an effect on mortality can also be seen after administration of various drugs that do not improve ventricular function, it is clear that bone marrow stem cell infusion, as opposed to satellite cell transplantation, does have a small but beneficial effect on heart function. However, additional experiments in mice enabling genetic tracing of the transplanted bone marrow cells and their progeny have revealed that bone marrow stem cells appear to lack the capacity to transdifferentiate to cardiomyocytes [18, 19]. Consequently, the mechanism by which infusion of bone marrow stem cells exerts a positive effect on cardiac function remains unknown. It is currently believed that upon transplantation these cells secrete paracrine factors that stimulate cells present in the cardiac tissue (i.e. either existing cardiomyocytes or resident cardiac stem cells). Defining such paracrine factors and identifying resident stem cell populations in the adult heart have become areas of intense research.
Table 1. Selected randomized clinical trials to evaluate stem cell therapy in cardiac disease
No. of patients
BMC, bone marrow stem cells; SMB, skeletal myoblasts; MI, myocardial infarction; PCI, percutaneous coronary intervention; LV, left ventricular; CABG, coronary-artery bypass graft; LVEF% Δ, left ventricular ejection fraction, absolute% difference compared to control; mon, months of follow-up after treatment; N.S., not significant.
Mobilization of endogenous progenitor cells in the heart
Although initially considered to lack regenerative potential, there is now growing consensus that the mammalian heart does exhibit a limited capacity to produce new cardiomyocytes. There is still debate regarding the scale of this regenerative response. Myocyte turnover rate has been reported to vary from 40% per year at 100 years of age  to between 1% in young adults and 0.5% in the elderly . Irrespective of the precise rate of cardiomyocyte turnover, it is theoretically important that new cardiomyocytes are formed in the adult human heart. Identifying the molecular mechanisms that control cardiomyocyte renewal would be of great interest within the field of regenerative cardiology. In principle, replenishment of cardiomyocytes can be due to cell cycle re-entry by existing cardiomyocytes or activation of endogenous cardiac stem cells. There is growing evidence that both mechanisms operate in the adult mammalian heart. Some extracellular signalling molecules [22, 23] as well as transcription factors  that control the capacity of cardiomyocytes to undergo mitosis have been identified, and experiments using inducible genetic labelling of cardiomyocytes in mice have indicated that activation and cardiac differentiation of endogenous cardiac stem cells underlie cardiomyocyte renewal following cardiac injury in mice . Efforts have focused on identifying such resident cardiac stem cells. For example, identification of stem cells has been based on distinct surface marker expression, including c-kit  and Sca-1 . Cells expressing these markers are enriched in cardiospheres, i.e. cellular spheres grown as suspension cultures obtained from cardiac biopsies. Cardiospheres can adopt the cardiomyocyte fate in tissue culture and promote cardiac regeneration after introduction into the infarcted heart . Several clinical trials have been initiated to explore the potential benefit of transplanting putative adult cardiac stem cells to the failing heart, including the SCIPIO  and CADUCEUS  studies in which patients with a recent myocardial infarction were treated with autologous biopsy-derived c-kit-positive cells or cardiospheres respectively. Preliminary results from both these trials were recently published. Both studies demonstrated a decrease in scar mass in the injured myocardium in the treatment group, but no statistically significant increase in the risk of adverse effects was seen. In addition, an increase in left ventricular ejection fraction was observed in the SCIPIO trial. Given that both studies were small and lacked a placebo group due to the invasive nature of the procedure, larger studies are required to confirm beneficial clinical effects.
The importance of paracrine effects
The mixed results from the clinical trials with different types of transplanted cells discussed above may to some extent be explained by the fact that transplanted cells function in part by secreting paracrine factors that modulate the function of the host myocardium. Experimental studies in rodents have provided several examples of paracrine signals that modulate the response to cardiac injury; signals secreted from hypoxic myocardium regulate homing of bone marrow-derived cells to the infarcted heart  and activation of epicardial cells following infarction with subsequent secretion of molecules promoting angiogenesis . It is conceivable that transplanted cells, following engraftment in the heart of the recipient, exert their positive effect by secreting factors that are beneficial for the remaining cells in the heart rather than by differentiation to cardiomyocytes and a role in the functional circuitry of the heart. Several modes of action have been proposed for paracrine factors derived from transplanted cells, including direct protective effects on existing cardiomyocytes, mobilization and activation of endogenous cardiac progenitor cells and modulation of the injured cardiac tissue (e.g. creating a favourable extracellular milieu for cardiomyocyte survival by reducing scar formation and increasing neovascularization) .
If paracrine factors mediate the positive effects of transplanted cells, it would be beneficial to simply administer recombinant forms of these factors rather than transplanting cells to the injured heart. However, identifying the bioactive factors from the secretome of the transplanted cells is a major challenge. Directed efforts have focused on different molecules with proven angiogenic effects (VEGF, SDF1), or ability to mobilize cardiac progenitor cells (G-CSF) or protect against cardiomyocyte apoptosis (PDGF) [34-36] (for recent review see ). In addition, proteins that increase cardiomyocyte proliferation have been identified (neuregulin  and periostin ). The concept of complex paracrine effects may support a ‘whatever works, works’ philosophy, or in other words that transplantation studies should proceed although the underlying molecular principles are not fully understood; alternatively, perhaps a better molecular understanding should be obtained first.
Endogenous cardiac progenitor cells and molecular programmes underlying cardiomyocyte differentiation
As mentioned above, although mobilization of endogenous cardiac progenitor cells in the adult heart is an attractive therapeutic concept, this area of research has been hampered by several factors. First and foremost, adult cardiac progenitor cells have not been well defined. Several markers of such cells have been suggested, but the relationship between the cell populations identified by expression of the genes encoding such markers in the cardiac lineage is poorly understood. Furthermore, we currently lack sufficient understanding of the developmental origin of these cell populations and the molecular mechanisms that control their self-renewal and differentiation to cardiomyocytes to utilize them in a rational manner. Likewise, paracrine factors that control endogenous regenerative mechanisms in the heart is an exciting area of research, but presently several key issues remain to be addressed, including the most suitable factors for therapeutic use and their cellular targets.
Consequently, studies have focused on understanding the development of the heart during embryogenesis and identifying the molecular circuitry controlling this complex process. Cardiac progenitor cells are relatively well defined during embryogenesis compared to their adult counterparts. In mammals, the heart is primarily derived from two distinct populations of mesodermal cells known as heart fields (Fig. 1) (see  and  for review). The first heart field is located in the anterior splanchnic mesoderm and initially gives rise to the cardiac crescent, the first morphologically identifiable structure associated with the future heart, and later the primitive linear heart tube. Multipotent cells of the second heart field derived largely from pharyngeal mesoderm are identified by the expression of Isl1, and migrate into the looping heart tube and differentiate into cardiomyocytes, smooth muscle cells and endothelial cells [40, 41]. These and subsequent complex morphological processes eventually result in the formation of the four-chamber mammalian heart, where the first heart field gives rise to the left ventricle, the second heart field to the right ventricle, and both contribute to the atria. In addition, epicardial cells give rise to scattered cardiomyocytes , and the cardiac neural crest  forms the outflow tract together with the second heart field.
A core set of transcription factors that is critical for proper formation of the heart during embryogenesis has been identified from animal models [43, 44]. Interestingly, studies of the human genetics of congenital heart disease have shown that mutations in several of the human homologues of this core set of cardiogenic transcription factors cause various forms of congenital heart disease [39, 45, 46]. Defining the transcriptional output of these transcription factors, as well as exploring whether concepts postulated from experimental models can be translated to the human setting, will be crucial for understanding human cardiogenesis. This will have important implications not only for use in regenerative medicine, but also for understanding the pathogenesis of congenital heart disease, the most common type of human birth defect.
Pluripotent and reprogrammed cells as a source of cardiomyocytes
Embryonic and induced pluripotent stem cells
Generation of cardiomyocytes from undifferentiated pluripotent cells represents an alternative approach; recent advances in the understanding of the molecular circuitry controlling cardiomyocyte differentiation have led to more refined protocols for steering the undifferentiated cells towards a cardiac fate. Embryonic stem (ES) and induced pluripotent stem (iPS) cells represent two pluripotent cell populations that have been extensively investigated as potential starting material for cardiomyocyte generation. ES cells are derived from the inner cell mass of the pre-implantation embryo that can differentiate to almost any cell type of the body. Since the landmark report by Thomson et al. describing the first isolation of human ES cells , the ability of these cells to differentiate to cardiomyocytes has been proven  and increasingly efficient protocols are being established for generation of cardiomyocytes [49, 50].
An important part of improved cardiomyocyte differentiation is a better understanding of the molecular programmes that control the various differentiation steps from ES and iPS cells to differentiated cardiomyocytes via several intermediary progenitor cell stages. As mentioned above, new insights have recently been gained into the core set of important transcription factors for cardiomyocyte differentiation. Definition of extracellular signalling molecules that can induce expression of the intrinsic transcriptional programme necessary for cardiomyocyte differentiation to occur is essential for directed differentiation of cardiomyocytes from human ES cells. A recurrent issue in developmental biology is the reiterative use of a limited set of extracellular signalling molecules for the specification of distinct cellular fates, and much effort has been directed towards defining the optimal culture conditions to induce cardiomyocyte differentiation. There is growing consensus that protocols based on application of molecules of the TGF-β/BMP and Wnt signalling pathways are successful in terms of generating cardiomyocytes from ES cells. However, it is clear that a more detailed characterization of the different cellular intermediates between pluripotency and mature cardiomyocytes, and a better definition of how these extracellular signalling molecules act at these different developmental stages, is required for the establishment of highly efficient differentiation protocols. Furthermore, it is becoming increasingly clear that there is a large variability in cardiogenic potential between different human pluripotent stem cell lines. There have been efforts to further advance differentiation protocols from ES and iPS cells to obtain cells with the precise differentiation state and to increase yields of the correct cell type.
ES cell-derived cardiomyocytes have been characterized extensively in vitro and it has been demonstrated that they share molecular markers and electrophysiological, mechanical and ultrastructural properties with primary cardiomyocytes (for review see ). However, ES cell-derived cardiomyocytes exhibit characteristics of foetal rather than adult cardiomyocytes [48, 51], and to date it has been difficult to define the culture conditions that most favour the specific generation of ventricular cardiomyocytes. Both these issues are of potential importance for transplantation and illustrate that, despite the increasing efficiency of in vitro differentiation protocols, more studies are needed before cells with characteristics of the cardiomyocytes lost after a myocardial infarct can be generated from ES cells. Furthermore, the fact that human ES cells are derived from early human embryos will pose ethical issues, and unless a large number of human ES cell panels with matching HLA haplotypes for a wide range of patients can be generated, a lack of immunocompatibility may also be a problem.
The iPS cells are generated by reprogramming to pluripotency of differentiated cells (e.g. fibroblasts). Takahashi and Yamanaka showed that pluripotency could be induced by introduction of a small number of transcription factor genes  and the iPS technology is now used to revert many different differentiated cell types to a pluripotent ES cell-like state. The iPS cells alleviate the problem of using embryo material for cell generation and should in principle not cause immunoincompatibility as cells from the patient can serve as the starting material. As with ES cells, iPS cells can be induced to various differentiation fates, including cardiomyocytes  (for review see [49, 50]). However, it was reported that transplanted iPS cells were immunologically rejected in mice . It should also be noted that although iPS cells are pluripotent, they may not be identical to ES cells. It has been reported that the transcriptomes of ES and iPS cells differ to some extent, and that iPS cells may retain an epigenetic memory of the cell type from which they were derived [54, 55]. It has also been observed that iPS cells may carry a higher mutational load than ES cells [56-58]. It largely remains to be determined whether the transcriptomal differences, such as the possibility of a higher mutational load and an ‘epigenetic memory’, will impact on cell types differentiated from iPS cells. It has been shown that cardiomyocyte-derived iPS cells exhibit an increased capacity to differentiate into ventricular cardiomyocytes . However, the fact that patient-specific iPS cells can be differentiated into cardiomyocytes is important in its own right, as in vitro differentiated cardiomyocytes may provide new insights into cardiac disorders such as the long QT syndrome [60, 61]. Thus, considerable work remains to be done before ES and iPS cells are likely to be used clinically in regenerative cardiology.
Reprogramming and transdifferentiation
An obvious risk of using pluripotent cells as a starting material for generation of cardiomyocytes for transplantation is the possibility of contamination of the transplant with residual pluripotent cells, which when introduced into the host can form tumours known as teratomas. This has prompted modifications to the original iPS cell reprogramming protocol to try to achieve only a partial dedifferentiation followed by cardiomyocyte differentiation or direct transdifferentiation to the cardiomyocyte state. By overexpression of the same genes as when deriving iPS cells, but complementing this with a chemical cocktail and modifications of the cell culture conditions, it has been shown that it is possible to accomplish a partial dedifferentiation and then direct such partially reprogrammed cells to cardiomyocytes . Another approach has been to try to identify a master cocktail of cardiac genes that can reprogramme fibroblasts directly to cardiomyocytes without dedifferentiation to a progenitor stage. The feasibility of this approach is illustrated by early examples of cellular reprogramming; initially fibroblasts were converted to myogenic cells through overexpression of the myogenic transcription factor MyoD . Similar experiments demonstrating reprogramming into other cells were then performed, including transdifferentiation in the haematopoietic system , reprogramming from exocrine to endocrine cells in the pancreas [65, 66] and reprogramming of fibroblasts to several distinct cell types including neurons  (for a recent excellent review see ). Similarly, studies in which cardiac fibroblasts were converted to cardiomyocytes by addition of a ‘cardiac master transcription factor cocktail’ consisting of three (Gata4, Mef2C and Tbx5 ) or four (Gata4, Mef2C, Tbx5 and Hand2 ) transcription factors have been recently reported. Of interest, ectopic expression of these genes in fibroblasts of the mouse heart following MI resulted in transdifferentiation of fibroblasts to cardiomyocytes with far greater efficiency than has been achieved in tissue culture, and led to improved heart function [70, 71]. It will be important to determine whether this is due to increased contractile capacity because of an increase in the number of cardiomyocytes, or whether it can be explained by other factors, such as paracrine molecules released by the reprogrammed fibroblasts, or a reduction in scar tissue formation. Moreover, comparisons with endogenous cardiomyocytes are required to assess the maturity, function and durability of cardiomyocytes derived through reprogramming events.
Optimal transplantation strategies
The choice of an optimal cell type and how to influence such cells to become cardiomyocytes are important; however, the mode of introducing the cells, the cell differentiation stage and how cells are structurally and functionally organized before transplantation are also likely to considerably affect the outcome of cardiac therapy.
How to deliver cells?
The heart loses a large number of cells following MI, therefore it is important to optimize all aspects of the transplantation process in order to obtain a sufficiently large number of cells for the transplant. As discussed above, several different cell types have been considered, and a number of different modes of delivery of cells to the heart have been explored, including intramyocardial, intravenous and intracoronary injection (for review see ). Initial experiments in animal models focused on delivery of cells to the myocardium by direct injection through the epicardium after opening the rib cage, thereby allowing visualization of the injured area. However, from a clinical perspective, open-chest surgery is associated with a risk of morbidity and is not justifiable unless performed simultaneously with other surgical procedures such as coronary-artery bypass grafting. Alternative less-invasive techniques for administering regenerative therapies have included percutaneous injections, both intravenous and directly targeting coronary arteries. However, because it is not clear how to influence peripherally injected cells to home to the injured myocardium, the most common experimental approach has been direct injection into the injured area. The ideal clinical delivery approach would combine the benefits of intramyocardial delivery and visual access to the injured area with the safety of percutaneous procedures to avoid the morbidity associated with open-chest surgery. It is possible that this can be achieved by future developments in videoscopic catheter technology.
Irrespective of how cells are delivered to the myocardium, the heart as an organ poses a number of significant challenges for transplantation-based approaches to therapy. At the cellular level, cardiomyocytes are coupled to each other by gap junctions, thereby facilitating rapid spreading of electrical impulses from one cardiomyocyte to the next. Arrays of coupled cardiomyocytes form muscle fibres, which are subsequently organized into layers. Anisotropically oriented layers of cardiomyocyte fibres are required for optimal contraction of the myocardium. Therefore, at the cellular level, transplanted cells must integrate properly with cardiomyocytes in the recipient heart to augment contractile function at the organ level. The contraction of the heart is another complicating factor as transplanted cells are at risk of being ejected from the site of transplantation; it has been estimated that close to 90% of the cells in a cell suspension may be lost after transplantation. Although approaches such as embedding cells in a viscous polymer solution prior to transplantation have yielded positive results to some extent , low rates of engraftment due to leakage of cells from the site of transplantation remain a complication.
Findings from the trial in which satellite cells from skeletal muscle were transplanted into the failing heart illustrate many of the disadvantages associated with transplantation of cells to the heart. It was found that engrafted satellite cells did not differentiate into cardiomyocytes, did not couple electrically with the cardiomyocytes of the recipient and, furthermore, did not contract in concert with the host myocardium. The net effect was not only no significant effect on heart function, but an increased risk of arrhythmias .
Introduction of biomaterials
The problems associated with transplantation of cellular suspensions to the heart have resulted in an increased interest in bioengineering approaches to generate functional cardiac tissue ex vivo. Theoretically, by combining high-efficiency differentiation protocols to drive progenitor cells to the cardiac lineage with tissue engineering technology, cardiac tissue engineered to mimic the cellular organization of the endogenous myocardium could be generated in the laboratory and thereafter transplanted to the patient as a single patch of cardiac tissue. To achieve a higher degree of structural organization, mimicking some of the characteristics of the heart prior to transplantation is fundamentally appealing, and thus there have been several attempts to structure the cells into more functional units before integration. The most common approach has been to seed cells onto a previously assembled scaffolding structure to serve as a three-dimensional matrix. In an elegant proof-of-principle study, it was demonstrated that this method could be used to engineer patches of rat cardiomyocytes which improved the function of the heart when transplanted into rats with compromised cardiac function . Of note, blood vessels from the recipient grew into these transplanted patches of ex vivo-generated cardiac tissue. Advances in biomaterials and nanotechnology have yielded scaffolds with interesting properties custom-made to facilitate transplantation to and integration with the host myocardium. One example is the utilization of temperature-sensitive polymers to generate cardiac sheaths that can be assembled as multilayered cardiac grafts [74, 75].
A more radical approach to tissue engineering has been to attempt to generate an entire organ ex vivo. Remarkable results were achieved using a decellularized rat heart (consisting only of extracellular material) as a scaffold for seeded rat heart cells. This resulted in a recellularized heart that, in a bioreactor mimicking normal physiological conditions for a heart, contracted with sufficient force to generate pump function . It will be crucial to modify this technique to allow the use of an alternative to rat cardiomyocytes for the recellularization procedure; a scalable immunologically matched cell population such as cardiogenic cells derived from iPS cells would be ideal. Furthermore, if the complex biological signals by which the decellularized scaffold controls the recellularization process can be elucidated, the use of a synthetic scaffold would further advance this technology towards therapeutic application.
In conclusion, there has been rapid progress in several areas, which may justify cautious optimism that regenerative approaches will eventually be successful in cardiology; there are, however, several obstacles to overcome before this can be achieved. Here, we have outlined some of these difficulties and discussed how they may be overcome in the future (Fig. 2).
Conflict of interest statement
No conflict of interest was declared.
We apologize to our colleagues whose work could not be cited due to limitations to the length of this review. EMH is a Wenner-Gren Foundation fellow and is supported by Hjärt-lungfonden. UL is supported by the Swedish Research Council (DBRM; Linneus Center in Developmental Biology and Regenerative Medicine), the Theme Center for Regenerative Medicine, the Swedish Cancer Society, and Knut och Alice Wallenbergs Stiftelse (WIRM; the Wallenberg Institute for Regenerative Medicine).