Direct Reprogramming as a Route to Cardiac Repair

Abstract

Ischemic heart disease is the leading cause of morbidity, mortality, and healthcare expenditure worldwide due to an inability of the heart to regenerate following injury. Thus, novel heart failure therapies aimed at promoting cardiomyocyte regeneration are desperately needed. In recent years, direct reprogramming of resident cardiac fibroblasts to induced cardiac-like myocytes (iCMs) has emerged as a promising therapeutic strategy to repurpose the fibrotic response of the injured heart toward a functional myocardium. Direct cardiac reprogramming was initially achieved through the overexpression of the transcription factors (TFs) Gata4, Mef2c, and Tbx5 (GMT). However, this combination of TFs and other subsequent cocktails demonstrated limited success in reprogramming adult human and mouse fibroblasts, constraining the clinical translation of this therapy. Over the past decade, significant effort has been dedicated to optimizing reprogramming cocktails comprised of cardiac TFs, epigenetic factors, microRNAs, or small molecules to yield efficient cardiac cell fate conversion. However, efficient reprogramming of adult human fibroblasts remains a significant challenge, and underlying mechanisms driving efficient conversion have been focused on achieving epigenetic remodeling at cardiac regulatory regions. Further studies to achieve a refined understanding and directed means of overcoming epigenetic barriers is merited to more rapidly translate these promising therapies to the clinic.

1. Introduction

Cardiovascular diseases are the leading cause of death in the US and worldwide, claiming 18 million lives annually [1]. Ischemic heart disease is dominant amongst these, precipitating heart failure, an incurable, costly, and deadly disease. Heart failure impacts 38 million people worldwide and represents the single most significant financial burden to health care systems, costing the United States alone $30 billion annually [2]. Despite remarkable advances in preventive pharmacotherapies and invasive mechanical support devices and techniques, disease burden and mortality remain astronomical, with heart failure bearing a 50% 5 year mortality—a statistic that rivals even the most aggressive neoplasms [3]. Currently, the only cure for heart failure is transplantation [4]. Donor hearts for transplantation are a severely limited resource; ~2,000 heart transplants occur annually despite the >100,000 end-stage heart failure patients who could benefit from this curative therapy [5]. Thus, new therapies and curative approaches for heart failure are desperately needed.

Due to poorly understood evolutionary mechanisms, the adult human heart possesses minimal regenerative capacity [4, 6]. Following myocardial infarction (MI), cardiomyocytes (CMs) undergo massive cell death, and through activation of resident cardiac fibroblasts (CFs), are replaced by noncontractile fibrotic scar. Ultimately, this pathogenic injury response progresses to chronic heart failure and death [7]. Given the inability of CMs to proliferate and reconstitute functional myocardium following injury, numerous cellular approaches have been attempted at both the bench and bedside. These approaches have focused on delivery of pluripotent and somatic stem cells to the injured myocardium, the results of which have been highly variable at best [8]. Alternative approaches to induce repair have focused on harnessing the cells resident in the human heart to induce remuscularization following injury through either stimulated proliferation of existing CMs or forced cell fate conversion of resident fibroblasts into functional CMs [8]. The latter, a process termed direct cardiac reprogramming, is the focus of this review and has emerged as a viable strategy for inducing repair in the adult heart.

In this review, we provide an overview of the efforts, progress, and insights gained from investigating direct cardiac reprogramming over the past decade. Since the advent of the field, its primary challenge and focus has been on achieving efficient conversion of fibroblasts to CMs, particularly starting from adult human CFs. We outline this progress in detail, as well as the mechanisms found to drive efficient cardiac cell fate conversion, with the hope that continued definition and manipulation of the underlying mechanisms will propel reprogramming technologies from the bench to patient care.

1.1. Emergence of Somatic Cell Reprogramming as an Avenue for Cardiac Regeneration

The epigenetic landscape of a cell dictates its identity, gene expression, and fate potential [9]. The relationship between the epigenome and cellular differentiation was initially introduced by Conrad Waddington in 1940. Waddington likened cellular development to a ball rolling down a hill, with the cell beginning in the totipotent state at the top of the hill and rolling downward, achieving differentiation by making a series of fate decisions at inflection points [10]. Upon reaching terminal differentiation, the cell rolls into a valley, with the surrounding hillsides representing epigenetic barriers that maintain the cell in its differentiated steady-state. While many visionary aspects of this model predated our knowledge of cell-cell interactions and gene regulatory networks, the landscape was deemed to be hierarchical, unidirectional, and irreversible, yielding a process such as cell fate reprogramming implausible [10].

1.1.1. Discovery of Transcription Factor Mediated Reprogramming

Incidental discoveries from a screen of chemotherapeutics on fibroblasts by Jones and colleagues in 1979 demonstrated the ability of a small molecule DNA demethylase, 5-azacytidine, to induce differentiation of C3H10T1/2 embryonic fibroblasts into terminally differentiated mesodermal clones [11, 12]. These experiments revealed the importance of epigenetic modification in achieving cell fate conversion, but also introduced the idea that modification of various loci could result in differential fate determination. In 1987, Davis and colleagues isolated myogenic cDNAs common to C2C12 and 5-aza-induced myoblasts, and demonstrated directed conversion of 10T1/2 fibroblasts into myoblasts through overexpression of the cDNA, MyoD [13]. This landmark discovery established MyoD as a master regulator of muscle, and not only marked the first description of transcription factor-mediated transdifferentiation, but also described the first instance of direct somatic cell reprogramming [14].

1.1.2. Discovery of Cardiac Transcription Factors and Master Regulators

Despite this early progress, nineteen years would pass before discoveries by Takahashi and Yamanaka demonstrated the ability of a fibroblast to revert to the pluripotent state and generate induced pluripotent stem cells (iPSCs) through overexpression of lineage defining transcription factors, providing a new understanding of cellular plasticity and rebirth to the reprogramming field [15–17]. Yet in this time, innumerable discoveries in transcription factor biology and cellular development emerged and enabled the successes of reprogramming to follow. Throughout the 1990s and 2000s, master regulators of cardiac development were elucidated and their molecular mechanisms and cooperativity defined. The muscle-specific basic-helix-loop-helix (bHLH) transcription factors eHAND and dHAND (or Hand1 and Hand2) were found to geographically dictate cardiogenesis of the left and right ventricles, respectively, and loss of these factors resulted in arrest of heart development and looping [18–22]. Nkx2–5, a cardiac-specific homeodomain protein, was discovered to be required for heart looping and eHAND expression, with lethality of mutants occurring at day 9–10 postcoitum (p.c.) [23]. A MADS-box transcription factor, Mef2c, was discovered to be essential for cardiogenesis in the mouse, inducing failure of right ventricle (RV) formation and absence of ventricular trabeculation [24]. Tbx5, a T-box transcription factor, was found to be expressed in the cardiac crescent and in the left ventricle with mutants exhibiting lethality at E10.5 due to severe heart defects [25–27]. Gata4, a TF expressed in mesodermal progenitors as well as cardiomyocytes and epi- and endocardium, was found to be necessary for RV and outflow tract development, and cardiomyocyte proliferation, with cardiac-specific deletion resulting in embryonic lethality at E11.5 [28]. Around the same time, Gata4 was described as a prototypical pioneer factor, given its ability to access silent chromatin and bind to nucleosomal DNA, providing greater mechanistic insights to its role in orchestrating heart development [29]. Finally, Myocardin (Myocd) was found to be a cardiac and smooth muscle specific TF that co-activates SRF and CArG box genes, with cardiac-specific deletions resulting in lethality at embryonic day 13.5 due to cardiac hypoplasia [30, 31]. Several other cardiac chromatin factors and complexes including Baf60c/SMARCD3 and Smyd1/BOP were found to be critical mediators of cardiac TF binding and activation of cardiac regulatory regions [32–34]. Many of these factors were shown to mediate differentiation of cardiac progenitor cells (CPCs) into differentiated, beating CMs. Importantly, Gata4, Tbx5, and Baf60c would be shown to mediate differentiation of mesoderm to CMs both in vitro and in vivo [35, 36]. These discoveries would prove essential for discovery and optimization of direct cardiac reprogramming.

1.1.3. Discovery of Direct Cardiac Reprogramming

Riding on the tailwinds of two decades rich in discoveries of TF biology, the discovery of cellular pluripotency unleashed a tidal wave of opportunity in the reprogramming space. Utilizing lineage-defining transcription factors, direct reprogramming converts one somatic cell type to another, entirely bypassing the pluripotent or progenitor cell state. Direct reprogramming of fibroblasts to induced cardiac-like myocytes (iCMs) was first reported by Ieda and colleagues in 2010 through retroviral overexpression of Gata4, Mef2c, and Tbx5 [37]. Selecting factors known to be essential for cardiogenesis with high differential expression in mouse embryonic CMs compared to embryonic CFs, the authors performed a subtractive screen of 14 cardiac epigenetic and transcription factors [38]. By overexpressing these factors in neonatal mouse cardiac fibroblasts (CFs) derived from an α-MHC-GFP reporter line, Ieda and colleagues determined that a combination of three factors, Gata4, Mef2c, and Tbx5 (GMT) were sufficient to induce transdifferentiation to iCMs in vitro. These initial studies demonstrated conversion of fibroblasts to iCMs through co-expression of the cardiac markers α-MHC+ and cardiac troponin T (cTnT) (α-MHC+/cTnT+ iCMs) in ~6.5% of neonatal CFs and ~2.5% tail tip fibroblasts (TTF). However, only ~0.01–0.1% of GMT-transduced iCMs adopted sarcomeric structures, and later studies suggested that reprogramming with GMT factors yielded incomplete or insufficient cellular reprogramming, sparking a decade-long effort in the field to improve reprogramming efficiency [39, 40].

1.1.4. Early Challenges in Reprogramming

Several early challenges manifested for the field, predominantly centered around conversion efficiency. Percent conversion of infected fibroblasts to iCMs in vitro with GMT was low and variable [39, 40]. Divergent protocols across laboratories, sensitivity of the assay itself, and varied approaches to assay readout resulted in reports of high reprogramming variability [41, 42]. Fibroblast type and developmental stage of acquisition resulted in tremendous reprogramming efficiency variability, likely due to substantial differences in epigenetic landscapes impacting fate potential and endogenous gene expression (Figure 1) [43]. For example, Hirai et al showed that embryonic fibroblasts were >100-fold more efficient in making iCMs than neonatal TTFs, and even more resistant were adult TTFs, suggesting that embryonic tissues maintain a greater degree of epigenetic plasticity that diminishes along developmental timelines. Further, even seemingly minute details such as primary fibroblast expansion and time in culture were shown to have a substantial impact on the epigenetic landscape of primary fibroblasts [44, 45]. While in vivo reprogramming with GMT following myocardial infarction showed beneficial results, the effect was again found to be variable with poor efficacy of cellular transplantation studies and modest changes in functional parameters in some studies [39, 40]. Importantly, GMT was not shown to induce reprogramming of human fibroblasts, prompting the search for other factors and delivery modalities [46].

Figure 1.

Advanced Fibroblast Developmental Stage and Species Restrict Cardiac Reprogramming Efficiency. Embryonic and postnatal-derived fibroblasts possess greater global chromatin accessibility and demonstrate higher efficiency in iCM conversion in response to cardiac reprogramming cocktails. Adult murine and human fibroblasts possess a lower degree of accessible chromatin and therefore are more resistant to reprogramming.

2.1. Optimizing Efficiency of Direct Reprogramming

2.1.1. Identification of Transcription Factors to Optimize Reprogramming

Following the initial reports of direct cardiac reprogramming, other groups sought to identify factors that would augment reprogramming of both mouse and human fibroblasts. Results of these studies are summarized in Table 1. In 2012, Li and colleagues demonstrated through lineage tracing studies that GMT could convert ~10% of adult in vivo cardiac fibroblasts to iCMs by local retroviral injection following myocardial infarction [47]. This degree of efficiency was reported to improve cardiac function by 5–10% as measured by ejection fraction (EF) [47]. Simultaneously, our laboratory uncovered that addition of Hand2 to the GMT cocktail (GHMT) significantly improved reprogramming efficiency 3-fold, generating ~7% adult CF iCMs above ~1–2% with GMT alone. Importantly, addition of Hand2 enhanced cardiac function and decreased scar area following myocardial infarction in mice compared to GMT [40]. Hand2 was later found to promote a high degree of plasticity to the reprogramming process, generating a heterogeneous reprogrammed iCM population with atrial, ventricular, and pacemaker phenotypes, as demonstrated by gene expression and patch clamping assays [48]. Further, transient expression of Hand2 utilizing a doxycycline-inducible system was shown to be sufficient to augment GMT reprogramming efficiency and iCM maturation [49].

Table 1.

Efficiency and Mechanisms of Cardiac Reprogramming Cocktails. Summary of published reprogramming cocktails across various fibroblast types and mechanisms of actions for each reprogramming cocktail.

	Cocktail	MEFs	CFs	TTFs	Human	Mechanism of Action
TF/Epigenetic Factors	GATA4, MEF2C, TBX5 (GMT)*		Adult: 1–2% αMHC+/cTnT+40, Neonatal: 0.3–6.5% αMHC +/cTnT+ iCMs37	Adult: 0.01%53,93, Neonatal: 2.5% αMHC+/cTnT+ iCMs37	HFFs: 0.38% cTnT+ iCMs48	TF binding to cardiac loci66
	HAND2 + GMT (GHMT)**	~8.5% αMHC +/cTnT+62, <5% beating iCMs59	Adult: 2–7% αMHC +/cTnT+ iCMs40,59,63	Adult: 1–9% αMHC +/cTnT+ iCMs40,63	HFFs: 1.6% cTnT+ iCMs48	Increased TF binding to cardiac loci66
	NKX2-5 + GHMT (HNGMT)	~1.5% iCMs with GCaMP activity50	Adult: ~4.5% GCaMP activity at day 1450	-	HFFs: 0.32% cTnT+ iCMs48	-
	ESRRG, MESP1 +GMT	-	-	-	H9F: ~15% αMHC + iCMs45	-
	SRF+Myocd(±BAF60C, MESP1) + GHMT	~1.5% αMHC+ oells,+Baf60o/Mesp1: ~2.5% αMHC+ iCMs51	-	-	-	-
	Myocd + GHMT (My- GHMT)	-	-	-	HFFs: 20.7% cTnT+48, ACHFs: ~5% cTnT+ iCMs46	-
	MEF2C-GATA4-TBX5***	4% αMHC+, 0% cTnT+, ~0.1 αMHC+/cTnT+ iCMs93	Adult: ~1.5% αMHC+ iCMs53, 0.2% αMHC+/cTnT+93; Neonatal: 3.8% αMHC+/cTnT+53	Adult: 0.6–2% αMHC+53,93, 0.2% αMHC+/cTnT+93; Neonatal: 1–7% αMHC+53,93, 0.0% αMHC+/cTnT+ iCMs93	-	Polycistronic cassette, increased Mef2c expression37
	MM3, GATA4, HAND2, TBX5	10–20% cTnT+ iCMs52	-	Neonatal: 29.3% cTnT+, 0.025% beating iCMs52	-	Increased accessibility at Mef2c binding sites52
	AKT1 + GHMT (AGHMT)	25% αMHC+/cTnT+, ~50% beating cells at day 2163	Adult: 5% αMHC+/cTnT+ iCMs63	Adult: 2–5% αMHC+/cTnT+ iCMs63,95	-	Increased TF binding and activation of cardiac enhancers66
	ZNF281 + AGHMT	-	-	Adult: 40%cTnT+, 33% αMHC+, ~25% αMHC+/cTnT+ iCMs95	-	Co-factor of Gata4, activation of NURD complex95
	PHF7 + AGHMT	~25–40% αMHC+, 20–35% cTnT+, ~20–30% αMHC+/cTnT+ iCMs (Garry et al, in press)		Adult: ~20% αMHC+, 25% cTnT+, ~20% αMHC+/cTnT+ iCMs (Garry et al, in press)	ACHFs: ~10% α-actinin, ~3–4% cTnT+, ~3% α-actinin+/cTnT+ iCMs (Garry et al, in press)	Reads H3K4me2/3 in fibroblasts; binds cardiac enhancers and cooperates with Smarcd3 to modify chomatin accessibility and TF binding (Garry et al, in press)
Small Molecules	DAPT + GHMT/AGHMT	10% sarcomere+ cells (from 2% with GHMT), 14% Ca2+ fluxing cells (from 6% with GHMT); ~70% cTnT+ or a-actinin+ cells (from ~50% with AGHMT)68	-	2–3 fold increase in αMHC/Actc1 expression68	-	Notch intracellular domain (NICD) inhibition increases binding of Mef2c to cardiac gene promoters68
	Y-27632 + GHMT2m/GHMT	60% cTnT+ or α-actinin+ (48% with GHMT alone), 30% beating at day 1259	-	-	-	ROCK inhibition
	A83–01 + GHMT2m/GHMT	58% cTnT+ iCMs (increase from 48% with GHMT), Increase in beating cells from 10 to 60%59	Adult: 10% +cTnT iCMs (increase from 5% GHMT)59	Adult: 30% cTnT+ iCMs (increase from 10% with GHMT2m)59, 4% beating cells59	-	TGF-β inhibition , decreases phosphorylation of SMAD2/3 and pro-fibrotic signaling59
	GSK126 or UNC0638 +MM3GHT	Both with ~20% beating (~2 fold increase in iCM number). GSK126: ~55% GCaMP oscillating iCMs88	-	-	-	Inhibition of EZH2/PRC2; inhibition of G9a and GLP. Inhibit suppressive histone methylation. 88
	SB431542 + GHNMT	~16% GCaMP activation at day 14 (5 fold increase)58	Adult: ~9% GCaMP activation at day 1458	-	-	TGF-β type1 receptor ALK inhibition58
	SB431542 + XAV939 +GMT*****	-	Neonatal: 35% αMHC+, 23% cTnT+, 17% αMHC+/cTnT+ iCMs60	-	ACHFs: 2 fold increase in TNT-GFP+ iCMs at day 14 from My-GMT60	TGF-β inhibition and Wnt inhibition60
Small Molecules	Diclofenac +GHMT	~5% αMHC+/cTnT+ iCMs (no change from GHMT alone)62		Neonatal: ~5% cTnT+, ~11% αMHC+, ~3% αMHC+/cTnT+ iCMs62		Inhibition of COX-2/PGE2/IL- 1 β signaling62
				Adult: 1.6% αMHC+/cTnT+ iCMs62
	FGF2, FGF10, VEGF + GMT	~13% αMHC+, 3% cTnT+, ~1% αMHC+/cTnT+ iCMs54	-	Neonatal: 9% beating cells with GHMT54	-	Activation of p38MAPK/PI 3K/AKT pathways.
	CHIR99021, RepSox, Forskolin, VPA, Parnate, TTNPB	~9% aMHC+, 14.5% α-actinin+ at day 2473	-	Neonatal: induces beating and cardiac gene expression73	-	-
	CHIR99021, A83–01, BIX01294, AS8351, SC1, Y27632, OAC2, SU16F, JNJ10198409	-	-	-	HFFs: ~7% cTnT+ iCMs72	Increased chromatin accessibility at cardiac regulatory regions72
	Erlotinib, Ruxolitinib +AGHMT	~40% αMHC+, ~40% cTnT+, ~20% αMHC+/cTnT+ iCMs65	Adult: ~40% αMHC+, ~32% cTnT+, ~20% αMHC+/cTnT+ iCMs65	Adult: ~40% αMHC+, 35% cTnT+, ~18% αMHC+/cTnT+ iCMs65		Inhibition of JAK/STAT/EGFR pathways65
	5-azacytidine+ Tbx5	-	-	Neonatal Rat: 31.6% cTnT+ iCMs at day 2192	-	Demethylation, activation of Wnt3a92
	PTC-209 + chemicals	~39% αMHC+ iCMs at day 1694	Adult: 9% αMHC+ iCMs94	-	-	Inhibition of Bmi1/PRC1 complex94
miRNAs	miR-1, miR-133, miR-208, miR499 (miR combo)± Jak inhibitor ****	0% αMHC+/cTnT+ iCMs84	Neonatal: 1.13–5.28% aMHC-CFP+ iCMs, with Jak inhibitor: 13–28% CFP+ iCMs82	Adult: +Jak inhibitor ~28% CFP+ iCMs82	-	Global downregulation of H3K27me3, inhibition of H3K27 methyltransferases89
	miR-133a +GMT	~20% αMHC+, 11% cTnT+, ~8% αMHC+/cTnT+ iCMs84	Adult: ~10% αMHC+, ~4% cTnT+, ~2.5% αMHC+/cTnT+84	-	-	Silencing fibroblast signature/Snai184
	miR-1, miR-133 +GHMT (GHMT2m)	47% cTnT+ cells increased from 31% with GHMT at day 759	Adult: ~10% +cTnT iCMs59	Adult: ~10% cTnT+ iCMs59	-	Inhibition of pro-fibrotic gene expression59
	miR-1, miR-133, My-GHMT	-	-	-	HFFs: 34.1% cTnT+47, ACHFs: 10.4% cTnT+ iCMs48	-
	miR-1, miR-133, My-GHT	-	-	-	ACHFs: 13.9% cTnT+ iCMs48	-
Knockdown of Barriers	shBmi1 + MGT/MT	20% αMHC+, 4% cTnT+, 2% αMHC+/cTnT+93	Adult: 20% αMHC+, 3% αMHC+/cTnT+93	Adult: 15% αMHC+, 0.5% αMHC+/cTnT+, Neonatal: 1.4% αMH+/cTnT+93	-	Inhibition of Polycomb repressive complex 1 subunit Bmi1 at cardiac loci93
	shBeclin1 +MGT******	20% αMHC+, 6% cTnT+ iCMs71	Neonatal: 20% αMHC+, 20% cTnT+, 13.5% αMHC+/cTnT+ iCMs71			Wnt inhibition independent of autophagy regulation71
			Adult: 4% αMHC+, 4% cTnT+71

^*GMT induces 5–10% increase in EF from vehicle alone and decreases fibrotic area.⁴⁷

^**GHMT induces ~20% increase in EF at 12 weeks from vehicle alone and decreases fibrotic area.⁴⁰

^***MGT induces 10% increase in EF from GMT and decrease in fibrotic area.¹⁰²

^****miR combo ~10% increase in fractional shortening at 3 months and decreased fibrosis at 1 month.⁸³

^*****SB431542 + XAV939 +GMT induced ~5% improvement in EF from GMT alone and decreased fibrosis by cMRI at 12 weeks.⁶⁰

^*******Becn1^+/− murine model injected with MGT induced a ~5% increase in FS and ~10% increase in EF compared to WT models.⁷¹

Moving toward functional readouts of iCM conversion, Addis and colleagues sought to identify TF cocktails that activate the genetic calcium indicator GCaMP and found that Hand2, Nkx2–5, Gata4, Mef2c, and Tbx5 (HNGMT) increased GCaMP activation in mouse embryonic fibroblasts (MEFs) roughly 50-fold greater than GMT [50]. Christoforou et al published a similar screen utilizing a GCaMP indicator, and showed that GMT reprogramming was augmented by Myocardin, SRF or in conjunction with Mesp1 and Smarcd3. However, the reprogramming efficiencies described in this study were far less effective than those that other groups reported (0.1% Myh6-GFP+ cells with GMT reprogramming, ~2.5% Myh6-GFP + cells reprogrammed MEF iCMs) [51].

Studies exploring the necessity of GMT factors in mouse fibroblast reprogramming identified Mef2c as a critical factor for iCM maturation and activation of cardiac gene programs [52]. Efforts to generate polycistronic cassettes for single transgene insertion and cocktail delivery demonstrated that order and therefore stoichiometry affected reprogramming efficiency. Using the polycistronic pMXs-MGT vector resulted in highest expression of Mef2c and enhanced reprogramming efficiency both in vitro and in vivo, improving cardiac function ~10% from standard delivery of individual GMT factors [53]. Experiments in the iPSC field found that fusion of a MyoD transactivation domain to reprogramming factors was found to facilitate epigenetic remodeling at reprogramming factor targets [54–56]. Applied to the cardiac reprogramming factors, the MyoD transactivation domain fused to Mef2c (MM₃) was found to significantly increase efficiency and beating cell clusters per well in both embryonic and neonatal fibroblasts [52]. Genome wide analyses performed across numerous cocktails in reprogramming of MEFs at both day 2 and day 7 post-induction similarly identified accession of Mef2c motifs by a core regulatory unit of reprogramming TFs to be critical regions for cardiac reprogramming and iCM maturation [49]. These findings were further supported by single-cell RNA and ATAC sequencing studies confirming the importance of Mef2c and Tbx5 binding sites and consequential alterations in chromatin accessibility at these cardiac loci during GMT reprogramming [57].

2.1.2. Targeting Signaling Pathways with Kinases and Small Molecules

Alteration of intracellular signaling pathways through administration of kinases and small molecules was applied to reprogramming in an effort to broadly impact cell differentiation, growth, and proliferation pathways while minimizing manipulation of the genome with TFs. Several small molecules emerged as meaningful mediators of iPSC reprogramming, and when applied to cardiac reprogramming, were found to markedly accelerate cell fate conversion. Small molecules found to impact iPSC reprogramming were applied to HNGMT reprogramming and the TGF-β inhibitor, SB431542, was found to increase reprogramming of MEFs and neonatal CF five-fold [58]. Further, unbiased transcriptomic analyses of GHMT reprogramming identified activation of pro-fibrotic and TGF- β signaling pathways in the acute phase of reprogramming induction, which limited conversion of fibroblasts to iCMs [59]. Application of small molecule inhibitors of the ROCK signaling pathway (Y-27632) and TGF- β signaling (A83–01) were found to significantly increase reprogramming in MEFs, with A83–01 found to have a marked impact on adult TTFs, generating ~30% cTnT+ TTF iCMs when added to a combination cocktail of GHMT and miRs [59]. Large small molecule screens testing over 5,500 compounds in GMT reprogramming of neonatal CFs similarly identified the TGF-β inhibitor SB431542, as well as the Wnt inhibitor XAV939 [60]. Interestingly, TGF-β is a known activator of canonical p53-mediated Wnt signaling, potentially linking the mechanisms of these two drugs in reprogramming [61]. These small molecules were not only found to augment cardiac function ~5% as measured by EF and decreased fibrosis post-MI, but also were found to increase reprogramming of adult cardiac human fibroblasts (ACHFs) by two-fold [60]. Recently, Muraoka and colleagues undertook a high throughput screen of >8,000 small molecules to identify activating factors in reprogramming of neonatal and adult TTFs, and found the non-steroidal anti-inflammatory drug (NSAID), diclofenac, to be the most potent activator of reprogramming, converting ~5% of neonatal TTFs to α-MHC+/cTnT+ iCMs. Interestingly, the impact of diclofenac on reprogramming was shown to be independent of TGF-β or Wnt inhibition, and attributed to inhibition of the COX-2/PGE2/EP4 axis which is highly active in TTFs [62]. Collectively, these studies underscore the importance of mediating TGF-β, Wnt, and inflammatory signaling pathways to achieve efficient cell fate conversion.

In an effort to identify kinases with potential to mediate regulatory signaling pathways in reprogramming, Zhou et al performed a screen of 192 myristoylated kinases applied to direct GHMT reprogramming of adult CFs and TTFs. From this screen, Akt1 was found to be the most potent activator of reprogramming, generating 2–5% α-MHC+/cTnT+ adult CF or TTF iCMs and ~25% αMHC+/cTnT+ MEF iCMs [63]. Moreover, phosphorylated Akt1 (pAkt1) induced iCM maturation as evidenced by polynucleation, hypertrophy, metabolic reprogramming, and spontaneous beating in >50% iCMs after 3 weeks in culture [63]. Upstream regulators of Akt1, IGF1 and PI3K were shown to activate reprogramming via Akt1 activation. Manipulation of downstream targets through inhibition of mTORC1 with rapamycin and overexpression of Foxo3 were shown to severely inhibit MEF reprogramming with AGHMT [63]. Interestingly, addition of the growth factors FGF2, FGF10, and VEGF to reprogramming media was found to augment GMT reprogramming and induce beating of neonatal TTFs through activation of the p38/MAPK and PI3K/AKT signaling axis, further supporting the involvement of this signaling pathway in achieving efficient and mature iCM reprogramming [63–65]. Hashimoto et al later found through genome wide in silico analyses that Akt1 dramatically increased chromatin accessibility and binding of core cardiac TFs at critical regulatory regions in AGHMT MEF iCM reprogramming [63, 64]. Through construction of a gene regulatory network integrating TF-ChIP, H3K27ac-ChIP, and RNA-Sequencing technologies, Hashimoto and colleagues uncovered downregulation of EGFR and JAK/STAT pathways in day 2 iCMs generated by reprogramming MEFs using AGHMT [66]. Achieving molecular and genetic inhibition of EGFR and JAK/STAT pathways with the drugs erlotinib and ruxolitinib, respectively, the authors discovered robust activation of reprogramming of adult TTFs when added to AGHMT, generating ~20% αMHC+/cTnT+ iCMs. These findings were validated by genetic knockdown of EGFR and Jak2, clearly defining that inhibiting these two arms of the EGF signaling pathway in the context of reinforced p-Akt1 expression promoted reprogramming in diverse fibroblast types [66].

The Notch signaling pathway is a deeply conserved and intricate regulator of developmental patterning and cell fate specification. Nuclear translocation of the Notch intracellular domain (NICD) is one of the end-points of the classical signaling cascade, and inhibition of NICD through the small molecule γ-secretase inhibitor, DAPT, had been previously shown to promote iPSC differentiation [67]. Further, the Notch signaling pathway, known to have a critical role in heart development, was shown by Abad et al to be activated in cardiac reprogramming [68]. By adding DAPT to GHMT and AGHMT reprogramming cocktails, robust reprogramming of MEFs was achieved, generating ~70% cTnT+ or α-actinin+ iCMs [68]. Importantly, these studies demonstrated that NICD overexpression impaired reprogramming. The authors concluded that the impact of NICD inhibition on reprogramming was not through the canonical Notch signaling pathway, but rather by enhancing chromatin accessibility and strengthening binding of Mef2c to its genomic targets [68].

Recently, autophagy was shown to be a critical process in iPSC and somatic cell reprogramming, presumably by regulating clearance of remnants necessary for prior cell fates [69, 70]. Specifically, knockdown of Beclin1 (Becn1), a regulator of autophagy induction, was found to promote iPSC reprogramming. Applied to direct cardiac reprogramming, induction of autophagy was found to augment cardiac cell fate conversion and shBecn1 induced three-fold iCM induction in neonatal CFs [71]. Genetic haploinsufficiency of Becn1 improved heart function in adult mice following injury by MI and reprogramming with MGT [71]. Interestingly, the effect of Becn1 knockdown was found to be independent of autophagy induction and due to activation of the canonical Wnt/β-catenin pathway. Previous studies regarding the Wnt pathway in reprogramming have presented conflicting evidence regarding its benefit in cardiac reprogramming, with Wnt inhibitors and activators both described to augment this process [60, 72]. As such, studies dissecting the specific pathways and timing of Wnt/ β-catenin signaling in reprogramming are merited.

Cocktails comprised entirely of chemicals were found to induce iPSC reprogramming. In an effort to reproduce these findings, Fu and colleagues identified clusters of beating cells at day 6–8 using CRFVPTZ (CHIR99021, RepSox, Forskolin, VPA, Parnate, TTNPB, and DZnep), concluding that this cocktail led to induction of both iPSCs and iCMs [73]. To optimize this strategy for iCM differentiation, induction media was modified with 15% fetal bovine serum and 5% knockout serum replacement, N-2 and B27 supplements, and removed DZnep. This CRFVPT cocktail was found to induce ~15% α-actinin+ iCMs and generate beating in neonatal TTFs by day 24 in culture [73]. Importantly, CRFV reprogramming cocktail was essential for formation of beating iCMs, which included a TGF-β inhibitor (RepSox), a GSK3 inhibitor mimicking activation of the Wnt pathway (CHIR99021), activator of intracellular cAMP (Forskolin), and HDAC inhibitor (VPA). Further, induction efficiency was significantly increased with the addition of Rolipram, a PDE4 inhibitor. Later, this cocktail was shown by the same group to induce cardiac transdifferentiation in vivo post-MI, albeit with low efficiency (1%) [74].

In summary, several high throughput chemical screens in both iPSC and direct cardiac reprogramming assays have identified cocktail additives that not only enhanced efficiency and maturation while minimizing use of transcription factors, but also illuminated critical signaling pathways in direct cardiac reprogramming. Moreover, reprogramming through chemicals alone has been shown to be efficient in vitro, but had a modest impact on cardiac function in vivo post-MI. As such, further studies are warranted to optimize delivery and stoichiometry of chemical cocktails to demonstrate therapeutic benefit in vivo.

2.1.3. miRNAs and Noncoding elements in Reprogramming

miRNAs are single-stranded non-coding RNAs (~22-nucleotides in length) that were discovered to impact broad signaling processes and developmental programs [75]. Use of miRNAs and miRNA inhibitors to achieve cell fate conversion is an attractive therapeutic strategy due to their small size, cytoplasmic activity, and delivery via nanoparticles as opposed to viral vectors [76]. Several conserved heart and muscle-specific miRNAs were found to have broad and critical control of cardiovascular developmental processes. The bicistronic miRs, miR-1 and miR-133, are highly expressed in cardiac muscle and are regulated by Mef2 and SRF [77]. miR-1 was found to be a post-transcriptional regulator of Hand2 and essential for heart development and ventricular septation, with lethality in mutants occurring between E15.5 and birth [77, 78]. Together, miR-1 and miR-133 have been found to direct mesodermal cell fate differentiation in ES cells and inhibit endo- and ectodermal differentiation [79]. miR-208 and miR-499 are among a group of miRNAs encoded by the myosin heavy chain genes (termed MyomiRs) that have been found to control myosin isoform switching, hypertrophy, and stress response pathways in cardiomyocytes [80, 81].

Following the discovery of GMT reprogramming, Jayawardena and colleagues tested 6 miRNAs known to play critical roles in cardiac development and found that miR-1, miR-133, and miR-208 induced reprogramming in the absence of exogenous TFs as measured by α-MHC-CFP reporter activation in 1.79–7.71% of cells and expression of myogenic gene programs in both neonatal and adult cardiac fibroblasts [82]. Addition of miR-499 to this cocktail (termed miR combo), did not increase conversion efficiency (1.13–5.28% α-MHC+ iCMs), but increased cardiac marker expression. Addition of Jak inhibitor to the miR combo cocktail increased reprogramming efficiency 8–10 fold, inducing ~28% α-MHC-CFP+ iCMs and oscillating intracellular calcium flux [82]. Importantly, miR combo delivery post-MI induced a ~10% increase in fractional shortening up to three months following infarct and decreased fibrosis compared to controls [83].

Following the successes of miR cocktails in reprogramming, several groups attempted to harness the benefits of cardiac miRs in conjunction with TFs in reprogramming. Testing all miR combo components individually with GMT, Maraoka and colleagues determined that addition of miR-133 to GMT potently activated reprogramming of MEFs nearly 8-fold, generating ~10% α-MHC+/cTnT+ iCMs [84]. Applied to adult CFs, addition of miR-133 generated ~2.5% α-MHC+/cTnT+ iCMs above 0.1% with GMT alone through repression of Snai1 and fibroblast gene programs [84].

MicroRNAs were also found to play a role in cardiac reprogramming through genome-wide in silico studies. H3K4me2 ChIP performed in GHMT day 7 MEF iCMs revealed increased deposition of H3K4me2 marks at upstream and enhancers and promoter regions of miR-1/miR-133, indicating increased activity of those regulatory elements in the context of GHMT factors [59]. Other muscle-specific miRNAs, including miR-208 and miR-499, were not found to have similar di-methylation patterns, suggesting an important role for miR-1 and miR-133 in cardiac cell fate specification. Added to the GHMT reprogramming cocktail, miR-1 and miR-133 (GHMT2m) increased reprogramming 2-fold, generating ~10% cTnT+ adult CF iCMs [59]. Importantly, addition of GHMT2m to MEFs doubled the number of beating cells, inducing spontaneous beating in ~12% MEF iCMs [59]. Further, addition of ROCK inhibitor Y-27632 and TGF-β inhibitor A83–01 to GHMT2m increased beating from 10% with GHMT2m alone to 35% and ~65% in day 11 MEF iCMs, respectively [59]. These data established the ability of cocktails employing combinatorial modalities including cardiac TFs, miRNAs, and chemicals to exert efficient and mature cardiac cell fate conversion from both embryonic and adult fibroblasts.

2.1.4. Identification of Epigenetic Factors to Promote Chromatin Remodeling

Cellular epigenetic landscapes strongly influence cell fate plasticity and commitment through the regulation of chromatin accessibility and gene transcription programs. In cardiac development, chromatin remodeling complexes critically promote cardiogenesis in a highly temporal and cell-type specific manner [9]. Functioning to either restructure the nucleosome or directly modify histones, these complexes have synergistic roles with TFs to dictate chromatin accessibility and therefore, gene expression. Nearly every cocktail that enhances direct cardiac reprogramming has been shown mechanistically to modify chromatin accessibility of cardiac enhancers and assist loading or strengthen binding of lineage defining transcription factors to cardiac genes and regulatory elements. Many of these mechanistic studies observed demethylation of the repressive H3K27me3 and induction of H3K4 methylation to induce reprogramming with cardiac TFs, miRNAs, or chemicals. However, only recently have genome-wide studies defined the enhancer landscapes of reprogrammed iCMs and the global binding sites of cardiac TFs in reprogramming [57, 66]. While approaches studying known epigenetic factors in reprogramming has been limited, use of these factors to enhance reprogramming of more resistant adult and human fibroblasts has shown great promise.

Trimethylation of H3K27 is a repressive histone modification that labels both silent heterochromatin as well as euchromatin with potentially active or active transcriptional sites [85]. H3K27 methylation is catalyzed by the polycomb repressive complex 2 (PRC2), composed of the critical subunits Ezh1/2 as well as Eed and Suz12. While repressive, PRC2 complex subunits do not prevent reprogramming and H3K27me3-marked regions have been shown to be at least modestly accessible to reprogramming TFs [86, 87]. However, regions marked by H3K27me3 and the heterochromatin-associated mark, H3K9me3, have been considered “reprogramming-resistant”, and several studies have indicated that demethylation of these sites yields significant changes in cell fate conversion [85]. Further, depletion of H3K27me3 was noted at cardiac genes in GMT reprogramming, suggesting that removal of this mark may facilitate transition from fibroblasts to iCMs [37]. Later, genome wide studies demonstrated depletion of H3K27me3 and enrichment of the active modification H3K4me3 at cardiac promoters as early as day 3 with coincident increase in H3K27me3 at fibroblast promoters at day 10 of pMXs-MGT iCM reprogramming [43].

The first studies manipulating methylation in cardiac reprogramming applied inhibitors of H3K27 and H3K9 methylation. Using GSK126, an inhibitor of Ezh2, and UNC0638, an inhibitor of G9a/GLP which mediate H3K27 and H3K9 methylation, respectively, Hirai and colleagues demonstrated that addition of these small molecules to GHMT increased reprogramming efficiency of MEFs to iCMs by 2–3 fold [88]. Recently, studies investigating the mechanisms involved in reprogramming using the miR combo cocktail identified global decrease in H3K27me3 protein (~40%) in miR combo-transfected fibroblasts [89]. The authors discovered an increase in expression of the demethylases Kdm6A/B in miR-combo iCMs and found that knockdown of these factors prevented cardiac TF expression. Further, knockdown of the PRC1 component Eed decreased H3K27 trimethylation by ~60%, and activated cardiac TFs [89]. Lastly, miR combo was found to activate toll-like receptor 3 (TLR3), a protein that globally activates epigenetic modifiers [90]. Interestingly, cardiac maturation achieved through miR combo was shown to be dependent on TLR3 expression, its activation of NF-κB, and consequential inhibition of inflammatory cascades [90].

Modification of other signaling and inflammatory cascades has been shown to alter chromatin accessibility and TF binding in reprogramming. For example, TGF-β inhibition in the context of direct reprogramming with GHMT, miR-1, and miR-133 was shown to facilitate Gata4 interaction with the H3K27me3 demethylase Kdm6b/JMJD3 and the SWI/SNF subunit Brg1, facilitating its accession of Gata4 targets. Further, TGF- β activation was shown to prevent H3K27me3 demethylation and inhibit Gata4 binding to its targets, again linking the repressive modification to resistance to reprogramming [91]. Inhibition of Notch signaling by the small molecule DAPT was shown to increase Mef2c binding, accessibility, and transcription at Mef2c targets [68]. Manipulation of the PI3K/Akt axis through addition of phosphorylated Akt1 was shown to markedly enhance H3K27ac deposition at cardiac enhancers on a genome-wide level and increase reprogramming factor binding to these sites [66]. Recently, the demethylating agent 5-azacytaidine (5-aza) was found to achieve cardiac reprogramming in conjunction with Tbx5 overexpression. Remarkably, 5-aza alone was found to induce ~10% cTnT+ iCMs in day 28 adult CFs, particularly at high (15umol/L) doses, and induce expression of Mef2c while downregulating pluripotency factors [92]. However, infection of 5-aza-treated cells with Tbx5 potently induced reprogramming, generating 31.6% cTnT+ iCMs above 7.1% with 5-aza alone. The authors demonstrated that treatment with 5-aza and Tbx5 coordinately allowed for activation of Wnt pathway signaling, presumably through increased chromatin accessibility at Tbx5 binding sites [92].

To identify chromatin remodeling factors that create and maintain barriers to reprogramming, Zhou and colleagues employed an shRNA-based loss-of-function screen. Through knockdown of 35 chromatin remodeling components in MGT reprogramming, the authors identified both barriers and essential epigenetic factors in reprogramming [93]. Knockdown of 11 of the 35 factors reduced reprogramming efficiency, suggesting the essential nature of these factors to cardiac TF action at cardiac enhancers. Repression of other epigenetic factors including Inhibitor of growth family member 5 (Ing5) and SWI/SNF subunit Hells resulted in a 2–6 fold increase in cTnT+ iCMs. Knockdown of PRC1 subunit Bmi1, however, resulted in the most significant increase in reprogramming efficiency, inducing a 5–10 fold increase in iCMs and ~20% α-MHC+/cTnT+ iCMs when reprogramming neonatal CFs [93]. Applied to adult TTFs, shBmi1 induced a significant α-MHC+ population (~17%), but failed to induce a strong cTnT+ population (1%), yielding ~0.5% α-MHC+/cTnT+ iCMs [93]. Importantly, knockdown of Bmi1 was shown to impact reprogramming by day 3 and induced an α-MHC+ iCM population in the absence of Gata4, with Mef2c and Tbx5 alone. Consistent with known mechanisms of PRC1, the authors identified depletion of the repressive H2AK119ub and a moderate increase in H3K4me3 levels at select cardiogenic loci by ChiP qPCR in Bmi1-depleted iCMs. Notably, increased cardiac gene expression was not detected from loci depleted in H2AK119ub but unchanged H3K4me3, suggesting that removal of the repressive ubiquitination mark alone is not sufficient to induce gene expression [93]. Use of the chemical inhibitor of Bmi1, PTC-209, in addition to the all-chemical cocktail CRFVPT demonstrated the benefit of Bmi1 repression in both MEFs and adult CFs [94]. Consistent with the findings of the early effect of shBmi1 on reprogramming, Testa and colleagues identified that pre-treatment of fibroblasts for 24 hours with PTC-209 prior to chemical induction of reprogramming was sufficient to augment reprogramming of adult CFs, generating ~10% α-MHC+ iCMs from ~7% with CRFVPT alone [94]. These findings further confirmed the role of Bmi1 and the PRC1 complex as a barrier to reprogramming.

Adult TTFs are the most resistant type of murine fibroblast to reprogramming, presumably due to an excess of heterochromatin and “reprogramming-resistant” regions. To identify factors that enable cardiac reprogramming in adult TTFs, Zhou and colleagues performed a high throughput screen overexpressing >1,000 cDNAs encoding TFs, epigenetic factors, and nuclear receptors in reprogramming of TTFs with AGHMT [95]. With α-MHC and cTnT expression as readouts, they identified 25 activators (Z-score>2) and 49 repressors (Z-score<2) of reprogramming, with the histone reader PHF7 (cTnT Z-score = 9.7; α-MHC Z-score=7.4) and ZNF281 (cTnT Z-score = 7.0; α-MHC Z-score=2.9) identified as the two strongest activators from the screen [95]. Markedly, the top hits from this screen were found to be the strongest activators of adult TTF reprogramming to date. Validation and mechanistic investigation of ZNF281 were pursued, and the addition of ZNF281 to AGHMT-treated TTFs generated ~45% cTnT+, ~25% α-MHC+, and ~21% α-MHC+/cTnT+ iCMs [95]. ZNF281 was found to interact with Gata4 and by ChIP-sequencing studies, ZNF281 was recruited to Gata4 binding sites to enhance cardiac gene expression. Furthermore, ZNF281 was found to repress the inflammatory response through recruitment of the NuRD chromatin remodeling complex [95].

The top hit identified from the screen, PHF7, is a histone reader that binds directly to histone H3 as well as H3K4me2/3 modifications [95–99]. Although not expressed in the postnatal heart, PHF7 is strongly enriched in Mesp1+ cardiac progenitors, suggesting an endogenous role in cardiac biology (Garry et al, in press). We validated the findings of the screen in reprogramming of adult TTFs, demonstrating that PHF7 induced ~25–40% α-MHC+, ~30% cTnT+, and ~20% α-MHC+/cTnT+ iCMs, representing a 10-fold increase in reprogramming efficiency above AGHMT alone. Interestingly, PHF7 with Mef2c and Tbx5 alone, induced reprogramming of TTFs to iCMs in the absence of Gata4. We found that PHF7 interacts with cardiac TFs and through ChIP and ATAC sequencing studies, discovered that PHF7 binds to multivalent cardiac enhancers and increases chromatin accessibility and cardiac TF binding at these sites to exact broad changes in cardiac gene expression. PHF7 recognizes cardiac enhancers in fibroblasts in the absence of reprogramming TFs by reading H3K4me2, a mark of poised chromatin and TF binding sites. Additionally, PHF7 interacts with and recruits the cardiac-specific SWI/SNF subunit Smarcd3/BAF60C to execute changes in chromatin accessibility. Through knockdown studies, we established that Smarcd3 expression is necessary for enhanced reprogramming with PHF7, reconciling the long-suspected role for the SWI/SNF complex and Smarcd3/BAF60C in direct reprogramming. These studies emphasize the underexplored potential for histone readers and epigenetic remodelers to recognize barriers and facilitate chromatin remodeling to achieve directed cell fate conversion.

To achieve an understanding of the mechanisms that dictate chromatin accessibility and epigenetic modification is to understand the heart of reprogramming a cell from a fibroblast to cardiac fate. Our understanding is in its infancy. However, rapidly developing next generation sequencing and single-cell omics technologies have only recently been applied to the field and the depth and resolution with which we can begin to understand epigenetic mechanisms of this process merit further exploration.

3.1. Reprogramming of Human Fibroblasts

Achieving reprogramming of human fibroblasts to a cardiac fate has emerged as the single most significant challenge of the field. Like adult murine fibroblasts, human fibroblasts possess staunch epigenetic barriers to maintain differentiated cell fates. Attempts to reprogram human fibroblasts initially dissuaded investigators from the clinically therapeutic potential of direct reprogramming. Several early studies demonstrated that GMT and GHMT cocktails were insufficient to activate cardiac gene expression in human CFs or human foreskin fibroblasts (HFFs) [45, 46]. As such, several laboratories embarked on screens to identify TFs, miRNAs, small molecules, or combination cocktails. Early screens of TFs in human reprogramming identified cocktails that successfully induced cardiac reprogramming, albeit with lower efficiency than murine cell types. Using α-MHC-mCherry transgenic H9 human embryonic stem cell (ESC)-derived fibroblasts (H9Fs), Fu and colleagues conducted a screen of 19 factors (13 TFs and 3 growth factors in addition to GMT) that were enriched in H9-CMs [45]. Starting with a 19 factor pool of factors, the authors conducted a subtractive screen with factor optimization demonstrating that GMT with ESRRG and MESP1 (5F) were sufficient to induce α-MHC+/cTnT+ iCMs. Addition of myocardin (MYOCD) and ZFPM2 (7F) markedly increased reprogramming efficiency, resulting in ~13% α-MHC+/cTNT+ iCMs at day 14 [45]. Interestingly, the authors found that activation of TGF-β1 signaling through direct delivery of TGF- β1 improved human cardiac reprogramming in conjunction with the 5F cocktail and posited that TGF- β1 likely assists in early induction of H9F cardiac reprogramming [45]. Later, PHF7 was found to markedly enhance reprogramming of adult human cardiac fibroblasts, inducing a ~3-fold increase in reprogramming above the MYOCD-containing human reprogramming cocktail, GHMMyT, alone (Garry et al, in press).

Screens involving combinations of TF and miRNAs were conducted in human neonatal and adult cardiac fibroblasts to identify effective human reprogramming cocktails with enhanced therapeutic relevance. Using 14 TFs and 3 miRNAs in addition to GHMT, Nam and colleagues found that addition of MYOCD to GHMT (GHMMyT) reprogramming of HFFs strongly induced cTnT+ iCMs from ~1.5% to ~17% [46]. MYOCD was found to be essential for ACHF reprogramming of HFF, differing from findings in ESC-derived H9Fs [45]. To further optimize human reprogramming of HFFs, the authors tested miRs with the GHMMyT cocktail and found that addition of miR-1 and miR-133 increased cTnT+ iCMs from ~19% to ~34%. Applying this cocktail to AHCFs, GHMyT+miR-1+miR-133 cocktail generated ~14% cTnT+ cells (a ~3 fold increase above GHMyT alone), representing the most optimal human reprogramming cocktail [46]. More dedicated studies of miRs in reprogramming confirmed the inability for miR combo to reprogram human cells. However, utilizing GMT with MESP1 and MYOCD (GMTMM cocktail), addition of miR-133 was found to increase induction rate of ACHFs from 2–8% to 23–27% cTnT+ iCMs and 2% to ~8% α-actinin+ iCMs through repression of Snai1 and fibroblast gene programs [84]. Collectively, these studies suggest an important role for miRs in combination with myocardin-containing TF cocktails in overcoming barriers of human fibroblasts to achieve cardiac cell fate conversion.

Small molecules are more practical for therapeutic application. As such, several studies sought to identify “chemical-only” cocktails to achieve human cardiac reprogramming. In 2014, Wang and colleagues identified a four chemical cocktail, SB431542, CHIR99021, parnate, forskolin, along with overexpression of a single TF, Oct4, that was able to induce human reprogramming in murine cells [100]. Based on these studies, Cao and colleagues conducted a screen of 89 small molecules in reprogramming of HFFs to iCMs. The authors tested small molecules in addition to the previously identified four chemical cocktail and identified a 15 factor combination that successfully generated α-MHC beating clusters at day 30 [72]. Through subtractive optimization, the authors identified seven compounds (7C) that were both necessary and sufficient to achieve reprogramming. Through a screen of 300 additional factors to 7C, the platelet-derived growth factor inhibitors SU16F and JNJ10198409 were found to substantially increase reprogramming of HFFs, generating ~7% cTnT+ cells [72]. Mechanistically, 9C-treated HFFs were found to have overall less heterochromatin and greater chromatin accessibility, with activation of core cardiac promoters, once again implicating epigenetic mechanisms in successful achievement of cell fate conversion [72].

Recently, single cell RNA-sequencing (scRNA-seq) performed throughout reprogramming in H9Fs revealed a decision point or “bifurcation event” that led to cellular regression to a fibroblast fate or differentiation to a cardiac-like cell [101]. From these data, it was determined that inhibition of several miR-133 targets induce immune-response-associated DNA methylation at these bifurcation points, thus promoting human iCM formation. These studies demonstrate the power of application of single-cell and computational modeling analyses in conjunction with functional studies to identify both mechanisms and optimized cocktails to achieve efficient reprogramming.

4.1. Looking to the Future

Despite progress and advances in efficiency of cardiac reprogramming, real challenges remain prior to translation. First, few factor cocktails, as outlined in this review, have been tested in adult human ventricular fibroblasts. These aged fibroblasts, rife with chronically activated inflammatory cascades and staunch epigenetic modifications, present the greatest resistance to clinical translation of reprogramming. As such, optimized cocktails that demonstrate efficient reprogramming of these adult human cell types are desperately needed. Second, delivery and timing of reprogramming factors safely in vivo requires optimization. Ideally, factor delivery would employ non-integrative systems such as Sendai or adenoviral vectors, which provide transient expression of delivered factors. Recently, use of modified mRNAs has emerged as an attractive strategy to deliver reprogramming factors given their recent broad and safe translation to patient care and transient, non-integrative delivery [102, 103]. However, while transient delivery remains ideal for safety purposes, the stability of the phenotype of transient reprogramming factor expression in adult human CFs has yet to be seen. At the bench, in vivo factor delivery has predominantly been through intramyocardial injection following open heart surgery and LAD ligation. While endocardial injection in practice is possible, cases of left ventricular infarct would require deployment of a catheter that either crosses the interatrial septum or the aortic valve, presenting greater risk to the patient. Moreover, endocardial injection through the use of fluoroscopy alone would yield highly variable results given variability in access, scar size, and location across infarcts. While voltage-gated technologies may assist in achieving greater precision of delivery to fibrosed regions, this would require significant planning and delivery of factors in a time window significantly removed from the infarct. Efficient uptake of factors through intracoronary delivery following revascularization is ideal, but may not be attainable with current delivery modalities. Lastly, with concerns with safety and factor delivery, come concerns regarding mosaicism of iCMs and potential for generation of arrhythmogenic foci within the reprogrammed myocardium. We have seen vast heterogeneity of atrial, ventricular, and pacemaker cells induced through variable uptake and stoichiometry of factors. As such, minimal factor cocktails and electrophysiological studies following reprogramming post-MI are warranted.

5.1. Conclusions

Landmark studies demonstrating directed cell fate conversion ignited a hope within the field of regenerative medicine that in vivo application of reprogramming factors could remuscularize the injured heart. Over the past decade, great strides have been made to overcome the barriers of reprogramming and achieve greater cell fate conversion efficiency. Significant effort has been put toward identifying cocktails containing cardiac TFs, miRNAs, or chemicals to achieve greater efficiency of adult and human cell types (Figure 2). While mechanistic studies have consistently pointed to the role of epigenetics and the importance of demethylation of repressive histone modifications in cardiac cell fate conversion, these studies have been largely quite focused, and only recently have broader deep sequencing analyses been applied toward achieving a refined understanding of the dynamic epigenetics of fibroblast to iCM conversion. Application of advanced screening tools and continued use of single-cell transcriptomic and chromatin conformation analyses are merited, and will provide refined and necessary mechanistic understanding for these therapies to advance to patient care.

Figure 2.

Chromatin Accessibility is the Gatekeeper of Efficient Cellular Reprogramming. Reprogramming factors and signaling pathways that achieve cardiac reprogramming (left) have been mechanistically shown to ultimately modify chromatin accessibility at fibroblast and cardiac enhancers. Removal of repressive epigenetic modifications (ie: H3K27me3) and deposition of activating modification (ie: H3K27ac) at cardiac regulatory regions achieves greater chromatin and TF accessibility at these regulatory loci, thereby activating cardiac gene expression and achieving greater reprogramming efficiency.