What’s in a cardiomyocyte – And how do we make one through reprogramming?

Abstract

Substantial progress is being made in the field cardiac reprogramming, and those in the field are hopeful that the technology will be formulated for therapeutic use. Beyond the excitement around generating a revolutionary new approach for treating ischemic heart diseases, cardiac reprogramming has delivered provocative findings that challenge common notions of cell fate and cell identity. Have we really made de novo cardiomyocytes? To answer this question, the essential characteristics of this unique and important cell type must first be defined. In this review, we walk through the history of scientific inquiry into cardiomyocytes, and then we examine the core features of cardiomyocytes as detailed in modern definitions. Informed by this, we turn to cardiac reprogramming to analyze the various screening approaches and ultimate factor combinations used in each study. We follow this with a dissection of the evidence used to support the authors’ claims of successfully creating cardiomyocytes, and we end by discussing what is known about the molecular mechanisms of cardiac reprogramming. Through this analysis, we find interesting differences between the study designs and their results, but it becomes clear that the field at large is generating cells that closely match the textbook definition cardiomyocyte. However, the differences noted between the results of each study are largely unexplained, reflecting the need for further research in both cardiac reprogramming and in native cardiomyocyte biology. Knowledge gained from future research will help move the field towards better reprogramming techniques and technologies.

1. Introduction

Examining its etymology, the definition of the term cardiomyocyte is clear: a muscle (-myo-) cell (-cyte-) of the heart (cardio-). Cardiomyocyte is often used in a clinical or research setting without a second thought being given to the accuracy of one’s use of the term, and in most cases, it is not necessary to think too carefully before calling a cell a cardiomyocyte. However, modern molecular techniques have enabled unprecedented insights into cells’ molecular constituents [1–3], prompting fundamental questions about how we ascribe cells a given identity. Furthermore, there is an ever-increasing number of cell types that have been transdifferentiated through advances in reprogramming and stem cell biology [4–9]. These simultaneous developments present a challenge: those seeking to study reprogramming in a particular cell type must grapple with a shifting cell type definition [10]. In this review, we seek to partially reconcile this challenge by briefly evaluating the scientific history of the cardiomyocyte, by condensing the fundamental characteristics of cardiomyocytes from the bookshelf of biomedicine, and finally by analyzing the primary studies that have formulated approaches to cardiac reprogramming.

2. The cardiomyocyte: then and now

2.1. The heart, the cell, and the heart cell

Antiquity contains many studies of the heart, with various anatomical descriptions originating from ancient Egypt, Greece, India, and China [11,12]. In general, the heart was regarded as a vital organ that functioned as pump. There was disagreement about its chief importance as the vital organ – a “cardiocentric” view of the body – and there were varying descriptions regarding what the heart pumped and how it pumped it. Hippocrates described the heart as a strong muscle, and later the Alexandrian physicians Herophilus and Erasistratus would provide very accurate descriptions of how it moved blood through the body [12]. In the second century, Galen formulated his own descriptions of the heart that differed in important ways from his predecessors [11]. Notably, he did not think that the heart was composed of muscle – a view that prevailed for the next millennium.

In the 15th century, Leonardo da Vinci had the unique opportunity to dissect the heart, and he came to the same conclusion as the ancient physicians before him: the heart is a muscle [13]. While an important milestone, this re-revelation that the heart is made of muscle was only one piece of many needed for William Harvey in the 16th century to paint an accurate picture of the heart’s role in circulating blood [13]. Even then, Harvey’s description of the heart was challenged by contemporaries like Descartes [14], further complicating and side tracking the evolution of our modern understanding of this vital organ.

While a more recognizable picture of the heart was developing in the 16th century, the technology necessary to study the myocardium’s smallest constituents – cardiomyocytes – was just being born. In the early 17th century, the first compound microscopes were invented, and the entire field of histology was born. Who invented the compound microscope is the subject of debate [15], but agreed upon is the importance of one individual’s application of it – Robert Hooke. In his book Micrographia, Hooke produced then unparalleled descriptions and drawings of the microscopic world. He also termed the word that today we use to describe single, indivisible units of life: cell.

From these earliest discoveries developed the fields of anatomy and histology, which form the foundation of our understanding of tissues, cells, and organs – including cardiomyocytes. However, without the ability to manipulate individual cells directly, further progress was limited. Pioneering work done in the 1910s enabled scientists to isolate single cells from tissues and culture them in vitro, allowing for experimentation and single-cell studies [16]. Working with embryonic chick tissue, Burrows observed cells arising from the heart and beating rhythmically – the first instance of cardiomyocytes in culture. Decades passed until isolation and study of cardiomyocytes in culture became more feasible, marked first by the successful isolation of rat cardiomyocytes [17].

With the ability to study cardiomyocytes in culture – and with the development of model transgenic organisms – our knowledge of cardiomyocytes grew exponentially in the second half of the 20th century. Advances in electron microscopy [18], electrophysiology [19], metabolism [20], and developmental biology [21] have vastly improved our ability investigate cardiomyocyte biology. Today, our accumulated knowledge of cardiomyocytes is at a level of detail that would have been unimaginable fifty years ago.

2.2. Current definition

Many fields have contributed to this knowledge and have helped to formulate a set of defining characteristics for this unique cell type; considered here are textbook descriptions of the cardiomyocyte from a few of these fields. From the perspective of histology [22], cardiomyocytes are the main cellular component of the myocardium, exhibiting centrally located nuclei, regular cross striations, and – due to their high oxygen demand – close proximity of an abundant capillary network. From the perspective of cell biology [23], a cardiomyocyte is a contractile, excitable heart cell that has a central nucleus, that has specific sarcomeric protein isoforms that distinguish it from other muscle cells, and that contracts rhythmically without rest. From the perspective of cardiology [24], a cardiomyocyte is the cell responsible for the contraction of the heart – utilizing an intricate network of contractile proteins and ion transporters for this work – with the main purpose of effectively executing the contraction-relaxation cycle. Looking beyond textbooks and instead to the latest curated ontologies of cardiomyocyte and cardiac muscle cell [25], a cardiomyocyte is generally defined as a myocyte that 1) resides in the heart 2) is responsible for the heart’s contraction 3) develops from myoblasts 4) has a central nucleus 5) is smaller than skeletal myocytes and 6) has abundant sarcoplasm.

These being the defining features of a cardiomyocyte, they serve as specific criteria to fulfill for the field of cardiac reprogramming. Success in the field can be measured by how closely we match these features; the following is a discussion of the various strategies that have been used to achieve cardiac reprogramming and an analysis of the evidence provided by each study to support its claim of making de novo cardiomyocytes.

3. Reprogramming approaches

3.1. A brief history of the field

After it was demonstrated that a handful of transcription factors could reinstitute a cell’s pluripotency [26], some groups suspected that unique cocktails of transcription factors could give rise to other cell types. Cardiac reprogramming was first accomplished in cultured mouse fibroblasts [4] and soon after in the ischemic mouse myocardium [27]. Viral vectors driving overexpression of three transcription factors – Mef2c, Gata4, and Tbx5 (MGT) – induced the formation of cells with structural and functional features similar to cardiomyocytes. Other groups soon began developing their own methods and added other transcription factors (TFs) to the mix [28–31]. As expertise in the field was born, debate about the most effective combination of transcription factors ensued [30,32]. Nonetheless, the most commonly used technique for creating induced cardiomyocytes (iCMs) remains transcription factor overexpression, and almost all cocktails include all three of MGT.

Nearly in concert with experiments utilizing TFs, work using micro RNAs (miRs) for cardiac reprogramming began [33,34]. miRs 1, 133, 208, and 499 form a cocktail capable of creating iCMs that are very similar to the iCMs borne of TFs. Now, miRs have been added to the formulae of many cocktails – notably, some groups have found them necessary for any appreciable amount of human cardiac reprogramming [31,35]. These constitute the initial approaches developed in the field, and they remain the main strategies for cardiac reprogramming – forced expression of cocktails of TFs, miRs, or both. Another – somewhat unrelated – approach has also developed that instead starts with the original cocktail of TFs used to create iPSCs [36]. Using these factors to initiate cellular reprogramming towards pluripotency, the trajectory is “shortcut” with signaling molecules (JAK-STAT inhibitors, BMP4, and others), chemically-defined medium, specific culture substrate, and cytokines. Other chemical/iPSC-based approaches to cardiac reprogramming have been developed [37,38], some which eliminate transcription factors completely [39].

Cardiac reprogramming in mice is well established. Relatively newer – and different in important ways – are the approaches used to reprogram human fibroblasts into cardiomyocytes. The approaches to date still use TFs and/or miRs to cause sweeping changes in gene expression, but the cocktails are different and larger. The first report in human cells used the original MGT cocktail and added MYOCD, miR-1, and miR-133 [35]. Another approach leaves out one of the core TFs, Mef2c, after adding MYOCD, miR-1, and miR-133 [35]. Like murine iCMs, the human iCMs created using these approaches have cardiac-specific gene expression, cytoskeletal structure, and excitable membrane properties. Compared to cardiac reprogramming in mice, human cardiac reprogramming has fewer reports overall.

3.2. Developing the cocktail

One can appreciate that a variety cocktails exist that can generate cardiomyocytes. Table 1 details the screens used to develop approaches for mouse cardiac reprogramming [4,28–30,33,36,38,39], and Table 2 details the screens performed in human cells [31,35,37,40–42]. By comparing these screens, several interesting observations can be made.

Table 1.

Mouse cardiac reprogramming screens. “Fibroblast Purification Method” is reported for studies in which the initial cell population was purified to enrich its fibroblast component. “Rationale for Screening Pool” is reported in paraphrase of the authors with as much accuracy as possible. Color key for optimal cocktails is as follows: transcription factors, micro RNAs, small molecule inhibitors, protein ligands.

Authors	Year	Fibroblast type	Fibroblast purification method	Screening Variable(s)	Rationale for Screening Pool	Screening Assay	Optimal Cocktail
leda et al.	2010	Neonatal cardiac fibroblasts	FACS for Thy1+/αMHC- GFP- cells	Transcription factors	Factors differentially expressed between cardiac fibroblasts and myocytes at E12.5	FACS for αMHC- GFP	Gata4, Mef2c, Tbx5
Efe et al.	2011	MEF transduced with Oct4, Sox2, and Klf4	(none)	Culture substrate, medium components, small molecule inhibitors	Signaling pathway modifiers: TGFβ, BMP, Hedgehog, Wnt, P(I)3K, JAK-STAT, and Notch pathways	In situ staining for Nebulette-lacZ	LIF removal, Matrigel or Geltrex substrate, knockout serum replacer, JI1, BMP4
Jayawardena et al.	2012	Adult cardiac fibroblasts	Percoll gradient centrifugation	miRs	miRs important in cardiac development	qPCR for CM- specific genes	miR-1, miR-133, miR-208, miR-499, JI1
Song et al.	2012	Tail-tip fibroblasts	(none)	Transcription factors	Evolutionarily-conserved factors important for cardiac development	FACS for aMHC- GFP	Gata4, Mef2c, Hand2, Tbx5
Protze et al.	2012	MEF	(none)	Transcription factors	Factors with cardiac phenotypes in KO mice and/or those capable of produce ectopic cardiac tissue formation	qPCR for CM- specific genes	Tbx5, Mef2c, Myocd
Addis et al.	2013	MEF	(none)	Transcription factors	Factors previously used for cardiac reprogramming	Detection of calcium flux with reporter transgene	Gata4, Mef2c, Hand2, Nkx2.5, Tbx5
Wang et al.	2014	MEF expressing Oct4, Sox2, Klf4	(none)	Small molecule inhibitors	Known enhancers of reprogramming	Number of beating clusters	Oct4, SB431542, CHIR99021, pamate, forskolin
Fu et al.	2015	CiPSCs (as described by Hou ef a/., 2013)	(none)	Culture substrate, medium components, small molecule inhibitors, protein ligands	Modulators of cardiac development or somatic reprogramming	Number of beating clusters	CHIR99021, RepSox, Forskolin, VPA, Pamate, TTNPB, Rolllipram, NRG1, GCSF LIF

Table 2.

Human cardiac reprogramming screens “Fibroblast Purification Method” is reported for studies in which the initial cell population was purified to enrich its fibroblast component. “Rationale for Screening Pool” is reported in paraphrase of the authors with as much accuracy as possible. Color key for optimal cocktails is as follows: transcription factors, micro RNAs, small molecule inhibitors, protein ligands.

Authors	Year	Fibroblast type	Fibroblast purification method	Screening Variable(s)	Rationale for Screening Pool	Screening Assay	Optimal Cocktail
Nam et al.	2013	Foreskin fibroblasts	(none)	Transcription factors and miRs	Factors important in heart development and muscle-specific miRs	FACS for cTnT and Tropomyosin	GATA4, HAND2, TBX5, MYOCD, miR-1, miR133
Wada et al.	2013	Cardiac fibroblasts	FACS for Thy1+/CD31- cells	Transcription factors	Additional factors to complement GATA4, MEF2C, and TBX5 and miRs previously reported in reprogramming cocktails	qPCR for TNNT2, NPPA, and RYR2	GATA4, MEF2C, TBX5, Myocardin, MESP1
Fu et al.	2013	H9-derived fibroblasts	FACS for Thy1+/αMHC- mCherry- cells	Transcription factors	Additional factors to complement GATA4, MEF2C, and TBX5 / Factors highly expressed in H9-derived cardiomyocytes	FACS for αMHC- mCherry	GATA4, MEF2C, TBX5, ESRRG, MESP1, Myocardin, ZFPM2
Muraoka et al.	2014	MEF (results validated in human cells)	(none)	miRs	Combination of previously reported reprogramming cocktails	FACS for αMHC- EGFP	GATA4, MEF2C, TBX5, Myocardin, MESP1, miR-133
Cao et al.	2016	Foreskin fibroblasts	(none)	Small molecule inhibitors	Known enhancers of reprogramming / known cardiogenic signals	Count of Beating clusters	CHIR99021, A83–01, B1X01294, AS8351,SC1, Y27632, OAC2, SU16F,
Christoforou et al.	2017	Dermal fibroblasts	(none)	Transcription factors, small molecule inhibitors, protein ligands	Additional factors to complement GATA4, MEF2C, and TBX5	ICC for ACTN2	GATA4, MEF2C, TBX5, Myocardin, NKX2.5, miR- 1, miR-133, JAK1i, HDACi, NRG1

In the studies that use a cocktail of transcription factors for reprogramming, there is significant overlap in the cocktails’ constituents. Mef2C, Gata4, and Tbx5 appear most often. They were the transcription factors used in the first demonstration of cardiac reprogramming by Ieda et al., and they were again found in later screens to be indispensable. This almost certainly reflects the central importance of these factors in transdifferentiation of cardiomyocytes – an observation that also has support from developmental studies [43–45]. Leaving aside this similarity, the studies’ screens differ in important ways.

The studies’ authors each picked different readouts to indicate that a cocktail might be effective for reprogramming. Five of the fourteen screens utilized some form of reporter gene construct – four of them based on the promoter of αMHC. Three of the screens used qPCR to assay genes of significance in cardiomyocytes – sarcomere components, ion channels, and/or developmentally-important transcription factors. Three of the screens simply counted the number of beating clusters of cells. Two of the screens used immunocytochemistry against structural proteins: cTnT and ACTN2. One of the screens used a calcium-sensitive fluorophore. Each screen’s readout is sensitive to different aspects of cardiomyocytes biology; a screen which assays sarcomere proteins will bias the results towards reprogramming strategies well suited to build sarcomeres, but it will provide no information on how effective a given cocktail is at promoting ion channel expression. The converse is also true – a screen which assays calcium flux will provide no information on the cells’ actin and myosin.

The screens also differed in their choice of initial cell type. Fibroblasts of skin, cardiac, and embryonic origin were all used. It is difficult to argue that one type of fibroblast is “best” as the choice depends on a variety of factors. For a screen, the cells need to be available in sufficient quantity to allow for testing of all the candidate cocktails. Furthermore, the cells should be available for follow-up studies after an effective cocktail has been identified – for the immediate validation of the screening results and for future studies that utilize the cocktail developed. Additionally, because one of cardiac reprogramming’s main goals is application, the cells used should be – depending on the potential clinical application – an adequate model for fibroblasts residing in the human myocardium (though much remains unclear about the exact nature of cardiac fibroblasts) [46,47] or an adequate model for readily-accessible human fibroblasts. It is not known how the initial fibroblast type directly affects the characteristics of the resultant cardiomyocyte.

Regarding the cocktail variables in screen, the most common class of molecules screened is transcription factors, followed by (in order) miRs, small molecule inhibitors, and protein ligands. Most of the studies focus on one class of molecules, testing combinations of only transcription factors, only miRs, or only small molecules. Besides varying the cocktail proper, some studies also investigate various alterations to medium components and/or substrate conditions for their effect on reprogramming. The reasons why each group chose to screen a particular class of molecules are again both based in biology as well as in translation. In the same way that Yamanaka and Takahashi developed their cocktail based on knowledge of early embryonic development, Ieda et al. developed their initial cocktail based on knowledge of cardiac development, and those who entered the field thereafter began with a similar rationale. As more was learned about the importance of miRs in cardiac development [48–54], Jayawardena et al. posited that this class of molecules ought to also be able to produce de novo cardiomyocytes. In terms of clinical relevance, finding non-viral means to accomplish cardiac reprogramming is also translationally attractive – indeed, miRs are becoming a new platform for delivering therapies [55]. Using a quite different approach, Efe et al. conducted the first screen of small molecules by reasoning that cells on the trajectory towards induced pluripotency could be guided towards cardiac reprogramming by modulators of various evolutionarily conserved signaling pathways (Wnt, Hedgehog, Notch, etc.). By using only small molecules, this study also took a significant step towards making transgene-free cardiac reprogramming possible. Finally, in the one of the latest screens, Christoforou et al. reasoned that combining knowledge from all the previously described reprogramming strategies would produce a cocktail of superior efficiency. Theirs is the only study to produce a cocktail containing transcription factors, miRs, small molecule inhibitors, and a protein ligand.

One can see how the contents of a screening pool affect the cocktail found to be optimal – i.e. earlier screens only considered transcription factors, and thus the cocktails formulated can only contain TFs. It is also apparent that the rationale for composing the screening pools has changed overtime. Initially, screens were designed around pools of developmentally-important factors/miRs/signaling pathways. More recently, screens have been based around molecules that have already been shown to contribute to cardiac reprogramming. In this way, screens have moved from less biased to more biased – new cocktails are being formulated by starting with “old” cocktails rather than by starting with a pool of molecules of some biologic significance. This is expected – our past knowledge drives our current hypotheses – but it is possible that by biasing our latest screening pools, the field has left out potentially untapped drivers of cardiac reprogramming.

Beyond the influences discussed above, the exact reasons why different screens produced different cocktails have not been rigorously studied. Furthermore, because each study used a different cocktail, it is likely that the respective cardiomyocytes are also different. Nonetheless, each of the authors presented substantial evidence to claim that they had successfully achieved cardiac reprogramming. In this regard, there are again important similarities and differences between the studies, as discussed in the next section.

3.3. Criteria for reprogramming success

Table 3 and Table 4 detail the results that each study presented as evidence of reprogramming success. (Almost all of results were tabulated – some of the studies presented evidence that is not summarized here.) That these summary tables could be made connotes the broad similarity in the results presented by each study; there is a “formula” to support one’s claim to cardiomyocytes (Fig. 1). Often one of the first assays done is measuring the expression of genes thought to be specific to cardiomyocytes. In general, the studies used very similar sets of genes for their assays. In some cases, the studies assayed the exact same genes, or they otherwise chose genes of the same functional significance: sarcomere genes such as Myh6, Actc1, and Tnnt2; ion channel genes such as Ryr2, Scn5a, and Cacna1c; gap junction proteins such as Cx43 and Cx40; and the endocrine signaling gene Nppa. These genes were selected purposefully to assay hallmarks of cardiomyocyte biology, and thus the similarity in the gene lists is expected and encouraging. It is also worth noting that all of the studies assayed a panel of markers to characterize their cardiomyocytes. This implies that no single gene’s expression can serve as definitive evidence of cardiomyocyte identity, and rather that collective expression data for many genes is what is necessary. A recent report found that Srf is required for cardiomyocyte maturation in neonatal mice [56]; interestingly, in neonatal and adult knockouts of Srf, the downregulated genes overlap significantly with the marker genes used in cardiac reprogramming studies. This concordance demonstrates the relatedness between the process of cardiac reprogramming and cardiomyocyte maturation in vivo.

Table 3.

Evidence of Cardiac Reprogramming Success in Mice. An * in proteins assayed indicates a reporter transgene. Epigenetic assay is defined as ChIP-qPCR or bi-sulfite sequencing. Cell relatedness analysis is defined as hierarchical clustering or principal component analysis of microarray or RNA-seq data from reprogrammed cardiomyocytes and native cardiomyocytes. Calcium flux is defined as assay by fluorometric dye, while depolarization is defined as any plasma membrane potential recording. Pharmacologic challenge is defined as exposure to any agent known to affect cardiomyocyte electrophysiology. In situ reprogramming is defined as induced cardiomyocyte generation within the intact mouse myocardium. Post MI function is defined as assay of cardiac output by echocardiography after a coronary ligation procedure.

Molecular profile							Cell function				In vivo testing
Authors	Year	Proteins assayed (list)	Transcripts assayed (list)	Epigenetic assay (list)	Cell relatedness analysis	Ultrastructure	Beating	Ca2+ flux	Membrane depolarization	Pharmacologic challenge	In situ reprograming	Post-MI function	Scarring
Ieda et al.	2010	αMHC-GFP*, cTnT, α- Actinin, ANF, Isll	Myh6, Actcl, Actc2, Nppa, Ryr2, Gjal	Tnnt2, Ryr2, Actn2, Nppa, Myh6	Y	N	Y	Y	Y	N	Y (performed later by Qian et al. 2012)	(n/a)	(n/a)
Efe et al.	2011	Flkl, Nkx2.5, Gata4, MHC, cTnT, Connexin-43, α- Actinin, MLC-2a	Gata4, Mespl, MHCβ, Mlc2a, Isll	Actn2, Ryr2, Tnnt2	N	N	Y	Y	Y	Y	N	(n/a)	(n/a)
Jayawardena et al.	2012	α-Actinin, Tnni3, αMHC- CFP*	Isll, Nkx2.5, Gata4, Hand2, Mef2c, Tnni3, Cx43, Myh6, Cacnalc, Kcnj2	(None)	N	N	Y	Y	N	Y	Y	N	N
Song et al.	2012	αMHC-GFP*, α-Actinin, cTnT, Cx43	(None)	(None)	N	N	Y	Y	Y	N	Y	Y	Y
Protze et al.	2012	Myh6:tdTomato*, α- Actinin, cTnT,	Myh6, Myl2, Nkx2.5, ScnSa, Actcl	(None)	N	N	N	N	Y	Y	N	(n/a)	(n/a)
Addis et al.	2013	αMHC, α-Actinin, cTnT	Pln, Casq2, Cacnalc, Atp2a2, Jph2, Ryr2	(None)	Y	N	Y	Y	N	N	N	(n/a)	(n/a)
Wang et al.	2014	cMHC, cTnT, ANP, Cx43, α- Actinin	Myh6, Tnnt2, Ryr2, ANF	Tnnt2, Ryr2, Actn2	N	N	Y	Y	Y	N	N	(n/a)	(n/a)
Fu et al.	2015	αMHC, cTnT, ANP, Cx43, α-Actinin, cTnl, HCN4, MLC2v	cTnT, Ryr2, βMHC,	(None)	Y	N	Y	Y	Y	Y	N	(n/a)	(n/a)

Table 4.

Evidence of Cardiac Reprogramming Success in Human Cells. An * in proteins assayed indicates a reporter transgene. Epigenetic assay is defined as ChIP-qPCR or bi-sulfite sequencing. Cell relatedness analysis is defined as hierarchical clustering or principal component analysis of microarray or RNA-seq data from reprogrammed cardiomyocytes and native cardiomyocytes. Calcium flux is defined as assay by fluorometric dye, while depolarization is defined as any plasma membrane potential recording. Pharmacologic challenge is defined as exposure to any agent known to affect cardiomyocyte electrophysiology. In situ reprogramming is defined as induced cardiomyocyte generation within the intact mouse myocardium. Post MI function is defined as assay of cardiac output by echocardiography after a coronary ligation procedure.

Molecular profile						Cell function
Authors	Year	Proteins Assayed (list)	Transcripts assayed (list)	Epigenetic assay (list)	Cell relatedness analysis	Ultrastructure	Beating	Ca2+ flux	Membrane depolarization	Pharmacologic challenge
Nam et al	2013	cTNT, Tropomyosin, α-Actinin	ACTC1, MYH6, MYL7, TNNT2, TPM1, NPPA, NPPB, ATP2A2, GJA1, GJA5	(None)	Y	N	Y	Y	N	N
Wada et al	2013	cTNT, α-Actinin, cTNI, ANP, GJA1	TNNT2, NPPA, RYR2, ACTC1, MYH6, MYL2, PLN, SCN5A, MEF2C, TBX20, MYH11	(None)	N	N	Y	Y	Y	N
Fu et al	2013	cTNT, α-Actinin	ACTC1, ACTN2, MYH6, MYL2, MYL7, TNNT2, NPPA, ATP2A2, PLN, RYR2	MYH6, MYH7, MYL7, ACTC1, RYR2, TNNT2, PLN, ACTN2	Y	Y	Y	Y	Y	Y
Muraoka et al	2014	cTNT	MYH6, ACTN2, NPPA, TTN	(None)	N	N	N	N	N	N
Cao et al	2016	αMHC-GFP*, cTAT, α-Actinin, ANF, cTNI, MLC2v, MLC2a, MHC, GJA1, NKX2–5, GAT A4, MEF2C, ISL1	TNNT2, MYH6, NPPA, HCN4, MYL2, MYL7, NPPB, GJA5, MESP1, GATA4, KDR, ISL1, MEF2C, HAND2	MYH6, MYL7	Y	Y	Y	Y	Y	Y
Christoforou et al	2017	ACTN2, TNNT2, GJA1, MYH6/7, TAGLN, MYH11	TNNT2, MYH6, MYH7, MYL2, MYL7, CASQ2, NPPA, ATP2A2, SLC8A1, RYR2, PLN, MYH11	(None)	Y	N	Y	Y	N	Y

Fig. 1.

Common assays in cardiac reprogramming. By combining 1) multiplex assays of gene expression, 2) assays for membrane excitability like patch clamp or calcium flux, and 3) in vivo tests in the setting of myocardial ischemia, cardiac reprogramming labs demonstrate that their cardiomyocytes are comparable to naturally occurring ones.

There is also wide agreement among the studies on the importance of two functional assays in particular: transmembrane calcium flux and cell contraction (“beating”). All but two of the fourteen studies contain assays for cytoplasmic changes in calcium levels and for spontaneous movement of cells or cell clusters. These assays serve as proof-of-concept tests to show that one’s induced cardiomyocytes 1) have contractile cytoskeletal structure, 2) exhibit dynamic changes in cytoplasmic calcium levels, and 3) have rhythmic, correlative oscillations in these phenotypes. With similar consensus, nine of the fourteen studies recorded the electrophysiologic properties of their cells to demonstrate that they have spontaneous and/or provoked membrane depolarization. Compared to assays of gene expression, functional studies are superior in showing that reprogrammed cardiomyocytes can perform like their natural counterparts.

Only one of the tabulated functional studies was not commonly performed by reprogrammers: pharmacologic testing. One could argue that lack of response to pharmacologic agents does not disprove the creation of cardiomyocytes – and visa versa – but testing pharmaceutical agents in cardiac reprogramming does have important translational implications. If therapeutic cardiac reprogramming cannot create cardiomyocytes that respond to cardiovascular pharmaceuticals, this could have unpredictable and dangerous effects on patients – a serious safety concern and one with analogy in the field of cardiac stem cell therapy [57–59].

Demonstration of cardiac reprogramming in vivo is the key test of a cocktail’s therapeutic potential. In vivo reprogramming answers important questions: can it be used to perform reprogramming in the native context of the heart? Can the fibroblasts generated by ischemic injury serve as the substrate for reprogramming? It also provides answers to fundamental questions about the induced cardiomyocytes. Do the cells look like the other cardiomyocytes in the heart, with appropriate junctions and sarcomeric structure? Do they contribute to the working myocardium and augment cardiac output? As is often the case, so too is it true of cardiac reprogramming: validation of in vitro results through in vivo testing strongly suggests that one’s findings have relevance to whole organisms. Nonetheless, while it is important to perform in vivo testing of cardiac reprogramming, it is technically very challenging to recreate the necessary reprogramming conditions within the heart of a live animal. For this reason, only three of the eight studies in mice showed evidence of cardiac reprogramming in vivo. As for proof of in vivo human cardiac reprogramming, the field must make more progress before the technology is ready for clinical trial [60].

Lastly, ultrastructural studies provide cellular phenotype details that are impossible to measure by any other method. In cardiomyocytes, those details include sarcomere size, mitochondrial morphology, and T-tubule structure among other things. Only two of the fourteen studies examined the ultrastructure of their reprogrammed cardiomyocytes; the exact reasons why the other studies did not use electron microscopy are not immediately clear.

3.4. Molecular mechanisms and follow-up investigations

Cardiac reprogramming is an entirely synthetic process; nowhere in nature do fibroblasts turn into cardiomyocytes without overexpression of genes or microRNAs. For the field – now that we have proved it is possible – the natural question that has followed is how? Investigations into the fundamental biology underlying this cell fate transition have yielded many interesting results. As one might expect based on Waddington’s vision of cell fate [61], epigenetic changes and chromatin remodeling are important steps in the course of cardiac reprogramming. Genes related to cardiomyocyte function lose H3K27me3 marking and gain H3K4me3, while Gata4 and Tbx5 lose H3K27me3 marking without changes in H3K4me3 [62]. Conversely, genes related to fibroblast function gain H3K27me3 and lose H3K4me3, while a number of fibroblast-enriched transcription factors gain H3K27me3 marking without changes in H3K4me3 [62]. This suggests that as a fibroblast transitions, it loses its epigenetic signature and adopts a cardiomyocyte signature. Furthermore, perturbing proteins which mediate epigenetic change affects the probability of reprogramming success. Exposing reprogramming cultures to the histone methyltransferase inhibitors GSK126 and UNC0638 increases reprogramming efficiency [63], and RNA interference of several chromatin-associated proteins – including the polycomb group complex 1 component Bmi1, the transcriptional corepressor Bcor, and the cohesin Stag2 – augments reprogramming [64,65]. Chemical inhibition of the histone methyltransferase Mll1, a protein critical to early development, also increases reprogramming efficiency [66]. These studies suggest that the epigenetic changes that occur in cardiac reprogramming are key to reprogramming success.

The interactions between other cellular processes and cardiac reprogramming are also being elucidated. Alternative splicing of mRNA is a key feature of tissue specification; the same has been found to be true of cardiac reprogramming. RNA interference of Ptbp1 [67], a heterogeneous nuclear ribonucleoprotein, or Zrsr2 [65], a protein involved in intron recognition, results in increased reprogramming efficiency, while interference of Sf3a1 and Sf3b1 [65], both components of the U2 snRNP particle, suppresses reprogramming. Attenuating fibroblast functions can also affect cardiac reprogramming. Inhibition of the TGFβ signaling [68,69] or ROCK signaling [69] pathways, which promote fibrosis and cell motility respectively, and RNA interference of Snai1 [31], a transcriptional corepressor that potentiates mesenchymal gene expression, all enhance reprogramming by suppressing expression of pro-fibrotic genes. In a large chemical screen for compounds that might augment cardiac reprogramming, the combination of a TGFβ inhibitor and a Wnt inhibitor were found to be a potent combination [70]. In vitro and in vivo evidence suggested that these inhibitors increased both the efficiency and the quality of reprogramming. Growth factor signaling pathways also potentiate cardiac reprograming. PI3K signaling is a potentiator of reprogramming – a finding which came after the observation that addition of Akt1 to the reprogramming cocktail increases efficiency [71]. Addition of FGF2, FGF10, and VEGF to the medium of transdifferentiating cells also promotes reprogramming [72]. Lastly, pro-inflammatory pathways are likely detrimental to cardiac reprogramming; Znf281 augments reprogramming by cooperating with Gata4 to bind cardiac promoters and suppress inflammatory genes [73]. Consistent with this observation, treating reprogramming cultures with the potent anti-inflammatory agents dexamethasone or nabumetone results in increased numbers of cardiomyocytes.

While the cited works have undoubtedly produced key findings, a more detailed molecular understanding of cardiac reprogramming is one of several pursuits that will be necessary before translating it into therapy [74]. Further studies are needed to better understand the mechanisms underlying this novel form of cardiomyocyte specification.

4. Conclusion: did we make cardiomyocytes?

After reviewing the evidence presented in the initial studies of cardiac reprogramming, it becomes clear that the properties of the induced cells align very closely with the textbook definition of cardiomyocyte. Induced cardiomyocytes express genes from all critical domains of cardiomyocyte biology: sarcomere components, ion channels, and calcium transport. They express cardiac-specific isotypes of sarcomere proteins, and though they don’t exist natively in the heart, they look and act like resident cardiomyocytes when placed there through in vivo reprogramming. For all intents and purposes, cardiac reprogramming makes cardiomyocytes.

Reprogrammers have checked all the right boxes, but there are differences between their respective cardiomyocytes. Reflecting this are the different terms used to describe the cells they have created: cardiomyocyte, induced cardiomyocyte (iCM), induced cardiac-like myocyte (iCLM), and chemically-induced cardiomyocyte (CiCM). Beyond the terms, biological differences exist between the cardiomyocytes made through different cardiac reprogramming methods. Some cells appear to have action potential traces similar to those of ventricular myocytes, others have traces similar to atrial myocytes, and still others have traces similar to nodal myocytes. Chemically-induced cardiomyocytes appear to pass through a multipotent stage before becoming cardiomyocytes, while cardiomyocytes generated using transcription factors and/or miRNAs undergo direct transdifferentiation. But perhaps the most widely discussed and debated difference between each approach is the efficiency with which cardiomyocytes are generated [32]. The debate is generated in part by the variability in how efficiency is reported. Some studies calculate their reprogramming efficiency based on expression of a cardiomyocyte marker, others calculate it based on presence of cytosolic calcium flux, etc. As is true for any screening assay, no one method for measuring efficiency is “best” – each assay is differentially sensitive and specific for the core properties of a cardiomyocyte.

Moreover, in a pool of reprogrammed cells, most cells will have a subset of the cardiomyocyte features assayed – few have all of them. One explanation for reprogrammed cells lacking “total” cardiomyocyte identity is that they are more similar to immature cardiomyocytes. Few studies to date have specifically examined the maturation of reprogrammed cardiomyocytes – in general, the later stages of cardiac reprogramming are less well characterized [74]. This is not to say that mature cardiomyocytes should necessarily be the ultimate translational goal. Before the advent of direct reprogramming, many studies tested various types of stem cells as therapy for acute myocardial infarction [75]. Their results suggest that repairing the myocardium with terminally differentiated cardiomyocytes may not be the most effective strategy. More recently, multipotent cardiac progenitors were produced using a direct reprogramming strategy, and transplantation of these cells in mice after coronary artery ligation significantly improved their survival [76]. Whether mature or immature cardiomyocytes are better for therapeutic cardiac reprogramming remains to be seen, but what is certain is that more studies into the maturation of reprogrammed cardiomyocytes are needed.

Aiding these and other studies, single-cell sequencing technologies enable researches to analyze the heterogeneity within cardiac reprogramming cell populations [67]. Separate populations of cells exist along the path from fibroblast to cardiomyocyte. And, strikingly, when single cell analyses are applied to the heart, it becomes readily obvious that wildtype cardiomyocytes themselves are a heterogeneous group [77]. Our knowledge of this heterogeneity is rapidly growing, and with this knowledge comes greater imperative to answer the question – what is in a cardiomyocyte? Finally, further understanding of the molecular details of cardiac reprogramming – what causes different cell populations to exist during reprogramming and the molecular obstacles hindering their progression through reprogramming – will provide the field with new avenues for generating higher quality cardiomyocytes with better efficiency.