Stem cell therapy of myocardial infarction: a promising opportunity in bioengineering

Abstract

Myocardial infarction (MI) is a life-threatening disease resulting from irreversible death of cardiomyocytes (CMs) and weakening of the heart blood-pumping function. Stem cell-based therapies have been studied for MI treatment over the last two decades with promising outcome. In this review, we critically summarize the past work in this field to elucidate the advantages and disadvantages of treating MI using pluripotent stem cells (PSCs) including both embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs), adult stem cells, and cardiac progenitor cells. The main advantage of the latter is their cytokine production capability to modulate immune responses and control the progression of healing. However, human adult stem cells have very limited (if not ‘no’) capacity to differentiate into functional CMs in vitro or in vivo. In contrast, PSCs can be differentiated into functional CMs although the protocols for the cardiac differentiation of PSCs are mainly for adherent cells under 2D culture. Derivation of PSC-CMs in 3D, allowing for large-scale production of CMs via modulation of the Wnt/β-catenin signal pathway with defined chemicals and medium, may be desired for clinical translation. Furthermore, the technology of purification and maturation of the PSC-CMs may need further improvements to eliminate teratoma formation after in vivo implantation of the PSC-CMs for treating MI. In addition, in vitro derived PSC-CMs may have mechanical and electrical mismatch with the patient’s cardiac tissue, which causes arrhythmia. This supports the use of PSC-derived cells committed to cardiac lineage without beating for implantation to treat MI. In this case, the PSC derived cells may utilize the mechanical, electrical, and chemical cues in the heart to further differentiate into mature/functional CMs in situ. Another major challenge facing stem cell therapy of MI is the low retention/survival of stem cells or their derivatives (e.g., PSC-CMs) in the heart for MI treatment after injection in vivo. This may be resolved by using biomaterials to engineer stem cells for reduced immunogenicity, immobilization of the cells in the heart, and increased integration with the host cardiac tissue. Biomaterials have also been applied in the derivation of CMs in vitro to increase the efficiency and maturation of differentiation. Collectively, a lot has been learned from the past failure of simply injecting intact stem cells or their derivatives in vivo for treating MI, and bioengineering stem cells with biomaterials is expected to be a valuable strategy for advancing stem cell therapy towards its widespread application for treating MI in the clinic.

Graphical Abstract

This review outlines the recent advancements in stem cell therapy (SCT) of myocardial infarction (MI), delves into the pros and cons of different stem cells for MI therapy, analyzes cardiomyocyte differentiation from different stem cells (particularly induced pluripotent stem cells), points out major challenges in the field, and elucidates the importance of bioengineering and biomaterials in improving SCT of MI.

1. Introduction

MI continues to be a major health burden in society and a leading cause of morbidity and mortality worldwide. There were ~15.9 million new cases of myocardial infarction in 2015 globally and the annual cost of MI treatments was estimated at $351.2 billion in 2014–2015 according to the report of the American Heart Association^[1]. MI is caused by the death of cardiomyocytes (CMs) in the left ventricle, which usually happens as a result of ischemia (decreased blood supply) after blockage of a branch in the left coronary artery. If the CMs do not undergo reperfusion or other treatments within two hours under ischemia, the progressive death of CMs will ensue. Eventually, the dead cardiac tissue will be replaced first by granulation tissues, and later by fibrosis or scar tissues which do not have the necessary contractile function for pumping blood^[2].

Despite the considerable progress made in MI treatments up to date, many patients still progress to heart failure after an acute MI^[3]. Pharmacological treatments with aspirin, beta blockers, and angiotensin-converting-enzyme inhibitor might help to slow down the progression of CM death^[4]. Percutaneous intervention with drugs is available after MI, but cannot restore functional cardiac tissue^[5]. So far there are no effective therapies which may regenerate the CMs to restore cardiac function. The only viable option available for patients with MI is allogeneic heart transplantation^[6]. In fact, many MI patients progress steadily towards New York Heart Association’s class III-IV heart failure, for which the only possible curative therapy available today is heart transplantation which has extensive costs, risks, and complications^[7],[8]. Therefore, new approaches are needed for myocardial repair after ischemic heart injury^[9].

The discovery of multiple classes of stem cells has generated new hope for curing MI. This may involve the transplantation of auto- and allogenic mesenchymal stem cells (MSCs), cardiac progenitor cells (CPCs), and pluripotent stem cell-derived CMs (PSC-CMs) into the region of the infarcted myocardium to promote regeneration of functional vasculature and/or CMs for restoring the heart function^[10].

2. Cell-Based Therapies for MI

2.1. MSCs for MI Therapy

MSCs originate from the mesoderm lineage and possess superior proliferative capability and good differentiation potential compared to somatic cells^[11]. Previous reports show that MSCs might have the potential to differentiate into non-beating CM-like cells under certain conditions^[12]. MSCs can be derived from multiple tissues like bone marrow, fat, placenta, and umbilical cord blood^[13]. The major difference between MSCs and hematopoietic stem cells (HSCs) is that MSCs attach on the surface of culture dishes, whilst HSCs float in medium. Though MSCs are a heterogeneous cell population, there is a putative standard to identify them with certain surface markers via immunostaining. MSCs derived from different sources should have a positive expression of CD90, CD44, CD105, Stro-1, mesenchymal stem cell antigen (MSCA)-1, vimentin, and CD73 while lacking in hematopoietic markers including CD45, CD34, CD14 or CD11b, CD79a, CD19, and human leukocyte antigen (HLA)-DR (≤2% positive), and being negative for pluripotency markers such as Tra-1–60, Tra-1–81, and SSEA-3 (Table 1)^[14].

Table 1.

A list of positive and negative markers used for verification of MSCs in comparison with some relevant cell types.

	CD13	CD29	CD44	CD49e	CD73	CD90	CD105	CD166	CD200
MSCs	+	+	+	+	+	+	+	+	+
hPSCs	-	-	-	-	-	-	-	-	-
Hematopoietic stem cell	-	-	-	-	-	-	-	-	-
	CD106	CD271	Stro-1	Vimentin	MSCA-1	SSEA-3	SSEA4	Tra-1–60	Tra-1–81
MSCs	+	+	+	+	+	-	+	-	-
hPSCs	-	-	-	-	-	+	+	+	+
Hematopoietic stem cells	-	-	-	-	-	-	-	-	-
	CD11b	CD19	CD31	CD34	CD45	CD79a	CD117	SCA-1	HLA-DR
MSCs	-	-	-	-	-	-	-	-	-
hPSCs	-	-	-	-	-	-	-	-	-
Hematopoietic stem cells	-	-	+	+	+	-	-	+	-

MSCs have good translational potential for MI treatment in the following ways: 1) Chemokine-mediated homing to inflamed lesion of MI: MSCs respond to inflammatory cytokines such as IL-6 and IL-8 and migrate to the inflamed area with up-regulated genes CXCR-4 and CCL2^[15]. 2) Immunomodulatory effect: MSCs can inhibit inflammation via secretion of anti-inflammatory cytokines like TSG6, PTGS2, and IDO1. It also inhibits the maturation of T and B cells via an increase in the T_reg cell number^[16]. 3) Anti-apoptotic effect: MSCs have an anti-apoptotic effect via regulation of BCL2, STAT4, and XRCC family genes which play an important role in mammalian DNA repair processes^[17]. 4) Angiogenic effect: MSCs secrete vascular endothelial growth factor (VEGF), fibroblast growth factor 2 (b-FGF2), and angiogenin which may help with revascularization in the infarcted myocardium. Also, MSCs themselves have the capability of differentiating into vascular cells^[18]. 5) Tropic effect: MSCs may secrete tropic cytokines including epidermal growth factor (EGF), glial-derived neurotrophic factor (GDNF), hepatocyte growth factor (HGF), and Leptin. These factors are produced in MSCs, enveloped in exosomes for secretion, and eventually taken up by nearby cells^[19]. Taking advantage of these merits, MSCs have been widely used in therapeutic trials and demonstrated efficacy in treating lupus, ischemia, stroke, graft versus host disease (GVHD), and colitis^[13]. Considering their convenient source and high isolation efficiency, MSCs derived from the bone marrow, fat, and umbilical cord blood have good potential for treating MI^[20].

MSCs derived from the aforementioned tissues are usually expanded in vitro to produce a large number of cells, followed by quality control with assays including surface marker identification, senescent test, immune inhibition assay, and trilineage differentiation. Finally, the cells can be administered for MI treatment in vivo^[21]. Animal studies show that MSCs injected intravenously localize mainly in the lungs as spheroids and have very low accumulation in the lesion of the MI^[22]. Though the left ventricular function improved after MSC administration, there is no clear evidence suggesting that the MSC’s therapeutic effect is due to their differentiation into CMs^[23]. Three different routes have been commonly used for injecting MSCs including intracoronary injection via the left main coronary artery with a balloon catheter, trans-endocardial or trans-epicardial injection, and intramyocardial injection (Fig. 1)^[24]. More than 2000 MI patients received an administration of MSCs, which showed little side effect and a positive improvement of infarct remodeling in acute MI trials^[25]. Schächinger et al. conducted a multicenter double-blind trial of acute MI, for which bone marrow-derived MSCs were administered by intracoronary infusion. Significant improvement in left ventricular ejection fraction (LVEF) was observed in patients with MSC injection compared to placebo at 4 months^[26]. Lunde et al. reported significantly improved exercise time and heart rate responses to exercise in a clinical trial in which autologous MSCs were administered via intracoronary injection. Improvement in LVEF was reported at 6 months^[27]. Although MSC administration is demonstrated as a possible and safe method for MI treatment in many studies, the outcome of this strategy is not always positive^[28]. For example, Cleland et al. reported a trial of autologous MSCs for acute MI, which showed no significant improvement in LVEF at 6 months in the MSC group as compared to control. Janssens et al. did not detect any improvement in global ventricular function at 4 months in the MSC group compared with the control group, although the infarct size was reduced and regional wall motion was improved in the MSC group^[29]. Similar results were obtained, including modest improvement in LVEF, reduced infarct size, and improved remodeling in animal studies, and the MSCs can be identified for weeks in sacrificed animals and may eventually be eliminated in the following months by the host immune system^[30],[31]. The prospective efficacy of MSC administration is validated by many trials in the timespan of months. However, there are no studies of MSC administration that extend for years. Extending the length of the trials might reveal incidences of death, recurrent MI, and stent thrombosis in treated patients^[32]. There is no clear evidence of MSC differentiation into CMs in vivo. The efficacy is highly attributed to the paracrine signaling from MSCs. Overall, further studies are needed to validate the therapeutic benefit of MSCs and elucidate the mechanisms of MSC-based therapies for MI treatment^[33].

Figure 1.

A schematic illustration of engineering stem cells or stem cell-derived cardiomyocytes with biomaterials and the commonly used methods for implanting the cells and their biomaterials-engineered constructs for treating myocardial infarction (MI). In early studies, stem cells or their derivatives were applied as single cells suspended in saline (carrier), often via peripheral vein injection, endoscopy-mediated injection via ventricle or coronary artery, or intramyocardial injection that can be done through either the minimally invasive thoracoscopy or the conventional thoracotomy (i.e., openchest surgery). More recently, stem cells or their derivatives have been engineered as microscale aggregates, biomaterial-cell constructs, or suspended in solutions that form hydrogel at body temperature for injectable delivery, similar to the injection of single cells. Stem cells or their derivatives have also been seeded in biomaterials scaffolds to form cardiac patches/sheets for delivery via thoracotomy. The schemes of the heart and the blood vessel system in the background are reproduced according to the CC-BY license employed by Smart Servier Medical Art (https://smart.servier.com/).

2.2. CPCs for MI Therapy

CPCs are defined as resident cells which maintain the potential for differentiation into CMs in the heart after birth^[34]. It is a heterogeneous cell population which is mostly quiescent in vivo, has high proliferative capability once activated, and might play a significant role in the regeneration of CMs for cardiac repair after heart attack^[35]. CPCs could be isolated from the cardiac tissue and then expanded for MI treatment. Presently, CPCs are not easily identified and their mechanism in homeostasis and potential for repairing the function of the myocardium are not well understood. CPCs have been identified mainly with five different methods as detailed below together with cardiac fibroblasts. The potential molecular markers for the CPCs together with that of cardiac fibroblasts are summarized in Table 2^[36].

Table 2.

A list of molecular markers of cardiac progenitor cells isolated with five different methods and the markers of cardiac fibroblasts.

	CD29	CD31	CD34	CD44	CD45	CD46	CD90	CD105	CD166	SCA-1	Abcg2	GATA4	NKX2.5	MEF2C	FLK-1	c-Kit	lslet-1	PDGFR-α
c-Kit+ CPCs	N/A	N/A	-	N/A	-	N/A	N/A	+	+	+	+	+	+/−	+	N/A	+	N/A	N/A
SCA-1+ CPCs	N/A	N/A	-	N/A	-	N/A	N/A	+	N/A	+	N/A	+	+/−	+	-	+/−	N/A	N/A
Islet-1+ CPCs	N/A	-	N/A	N/A	N/A	N/A	N/A	N/A	N/A	-	N/A	+	+	N/A	N/A	+/−	+	N/A
Cardiosphere-derived cells/ Epicardium derived cells	N/A	+	+	+	+	+	+	+	N/A	+	+	N/A	N/A	N/A	N/A	+	N/A	N/A
Side population cells	N/A	N/A	+	N/A	+	N/A	N/A	N/A	N/A	+	+	-	-	N/A	N/A	+	N/A	N/A
Cadiac fibroblasts	+	-	N/A	+	-	N/A	+	+	N/A	+	N/A	N/A	N/A	N/A	-	+	N/A	+

2.2.1. The c-Kit Tyrosine Kinase Receptor Positive CPCs for MI Therapy

The c-Kit tyrosine kinase receptor is the first surface marker used for identifying CPCs which have potential efficacy for MI treatment^{[37, 38]}. In one study, c-Kit-positive (+) cells are isolated from the adult heart tissue and expanded in vitro. After demonstrating that the cells expressed cardiac cell markers, they were then injected into the MI lesion of a rodent model^[39]. Though the group show improved function of the heart after c-Kit-positive (c-Kit+) cell administration and other groups reported similar improvements to MI after c-Kit+ cell administration, the identity of c-Kit cells is still controversial^[40]. One concern is that c-Kit is also a widely used marker for hematopoietic stem cells which are not efficient for differentiation into CMs^[41]. The positive effect of c-Kit+ cells in MI treatment are shown to be due to paracrine signaling rather than their differentiation into CMs^[42]. Recently, researchers have reported that only ≈0.027% of c-Kit+ cells may differentiate into CMs with genetic lineage tracing technology^{[37, 43]}.

2.2.2. KDR and PDGFR-α-Positive CPCs for MI Therapy

The kinase insert domain protein receptor (KDR) and the platelet-derived growth factor receptor α (PDGFR-α) have been utilized as important surface markers of CPCs^[44]. KDR and PDGFR-α are broadly expressed in the epicardium, myocardium, and endocardium, in both human fetal and diseased adult hearts. The KDR+/PDGFR-α+ cells could differentiate into CMs, show strong capacity for cardiovascular regeneration, and contribute to artery development after MI^[45]. The purity of the cell population can be enriched by flow cytometry and they can be used to reduce fibrosis via promoting vascularization in the injured heart^[38]. There are reports indicating that PDGFR-α is involved in regulating neural crest cells and maintaining the balance of stromal and adipogenic cells during adipogenesis in embryonic development^[46]. PDGFR-α has also been shown to be important for connective tissue remodeling alongside its role in cardiovascular regeneration^[47].

2.2.3. Stem Cell Antigen 1 Positive CPCs for MI Therapy

Together with the CPC population bearing c-Kit+ and KDR+/PDGFR-α+ markers, stem cell antigen 1-positive (SCA1+) cells isolated from the cell suspension of fetal and adult hearts of murine models showed the expression of early cardiac genes and activation of telomerase demonstrating a self-renewing capability^{[34, 48]}. These SCA1+ cells might differentiate to CMs when treated with 5-azacytidine and transforming growth factor β (TGF-β). The resultant CMs have the regular cardiac gene expression profile and function in vitro. When injected into the lesion of MI, the SCA1+ cell-derived CMs resulted in prospective efficacy on structural and functional cardiac improvements on MI models in murine animals^{[25, 49]}. Nevertheless, the efficiency of cardiogenic differentiation in vivo of the SCA1+ cells is quite low, they are difficult to be identified after sacrificing the animals in these studies and there is a lack of evidence of the SCA1+ cells in human CPCs. There are studies that indicate SCA1+ cells differentiate mostly into cardiac endothelial cells and fibroblasts instead of CMs in an adult mouse heart after MI^[50].

2.2.4. Cardiosphere-Derived Cells for MI Therapy

Cardiosphere-derived cells (CDCs) are often derived from cardiac tissue isolated from patients. They are extracted from multiple rounds of harvesting cells that migrate out of the cardiac tissue, and can be administered as an autologous or allogeneic treatment for MI^[51]. One major advantage of autologous CDCs is the decreased mismatch of the major histocompatibility complex (MHC) and expression of B7 costimulatory molecules, which might provide a favorable hypo-immunogenic profile for transplantation in vivo. There are studies showing that allogeneic CDCs are also safe and may produce sustained functional and structural benefits without immunosuppression after transplantation in immunologically mismatched animals^[52]. Another study shows after administration of allogeneic CDCs, improvements in cardiac structure and function could be observed over 6 months, with only transient mild immune reactions in the lesion and no histological evidence of immune rejection by the host immune system^[53]. However, there is little evidence indicating that CDCs are involved in the regeneration of CMs. In addition, the therapeutic efficacy of CDCs is only speculated to result from the paracrine effects, stimulation of endogenous cardiac regeneration, or recruitment of endogenous cardiac stem cells which positively express c-Kit, NKX2.5, and Islet (ISL)-1^[54]. Meanwhile, the CD34+ hematopoietic stem cells which are also presented in situ with transplantation of CDCs, may help neovascularization and cardiac regeneration after MI^[55]. Specifically, it has been shown that there is increased gene expression of vascular endothelial growth factor, insulin-like growth factor-1, and hepatocyte growth factor in the MI region after transplantation of allogeneic CDCs^[56].

2.2.5. Cardiac Side Population Cells for MI Therapy

The cardiac side population cells (CSPCs) have a unique ability to actively efflux the DNA binding dye (e.g., Hoechst 33342) and have the capability of becoming CM-like cells to varying degrees under controlled conditions in vitro^[57],[58]. They appear as a small “side” population on the flow cytometry plot of the fluorescence-stained cells isolated from the postnatal myocardium^[59]. It is reported that the CSPCs expressing SCA1+/CD31− markers isolated from the mouse (C57BL6/J) hearts might differentiate into CMs and endothelial cells, in addition to enhancing cardiac function within two weeks after intramyocardial injection of the cells in a murine myocardial ischemic model. Up-regulation of CXCR4 is also found in SCA1+/CD31− CMs in the MI lesion. This indicates that SCA1+/CD31− cells migrate from the injected location into the MI lesion, and then differentiate into both CM- and endothelial-like cells, to reduce the damage of acute ischemic injury^[60]. Currently, the DNA binding dye Hoechst 33342 and Rhodamine 123 are used for isolating the CSPCs^[61].

2.2.6. CFs for MI Therapy

CFs are the largest cell population of the heart along with cardiomyocytes, endothelial cells, vascular smooth muscle cells, and neural cells^[62]. The CFs exert important functions in maintaining cardiac functionality, in addition to maintaining its physical structure via the production of extracellular matrix. This supports CM function and cell-cell communication for signaling transduction to keep the normal electrophysiological and physical function of the heart. The CFs secrete endogenous growth factors like IL-1, IL-6, and TNF-α for cardiac remodeling including angiogenesis after MI^[63]. The CFs treated with 5-azacytidine could differentiate into CM-like cells showing typical cardiac phenotypes such as multinucleated myotubes, sarcomeres, cardiac troponin I, β-myosin heavy chains, and the gap junction protein connexin 43^[64]. The CFs improved the function of infarct hearts within 1 month after transplantation into rats via intramyocardial injection^[65]. CD90+ CFs are also reported to have efficacy in reducing fibrosis of acute myocardial injury in rats, in addition to showing recovery of the LVEF and left ventricular fractional shortening (LVFS) on day 14 after injection into the peri-infarct zone compared to controls^[66].

In summary, CPCs are heterogeneous in general and there are no definitive markers for their identification Many of the aforementioned markers together with ABCG2, GATA4, MEF2C, and FLK1 (Table 2) for their identification overlap with that of the mesenchymal and hematopoietic lineages. This makes their identification with specific molecular markers difficult in regard to cellular and physiological characteristics. It is also inconvenient to isolate the CPCs from tissues due to the loss of cells during the processes of dissociation, centrifuging, and sorting. As a result, the eventual amount of the cells for each isolation is limited. Because of their poor proliferation, it requires a long waiting time to expand the cells which inadvertently causes senescence of the cells^[67]. Although the existence of CPCs in an adult heart is controversial, transplanted CPCs might facilitate cardiac regeneration, and lead to improvement of heart function. However, there is no clear evidence that transplanted CPCs directly differentiate into functional CMs. The positive results from the administration of CPCs have been shown to result from cytokine effects to enhance regeneration rather than direct differentiation into CMs.

2.3. PSCs for MI Therapy

PSCs, including ESCs and iPSCs, can be valuable for treating MI. Unlike MSCs or the putative CPCs, PSCs have almost unlimited self-renewal capacity. It has been reported that some PSCs can undergo more than 120 passages, compared with more differentiated normal (i.e., non-cancerous) cells which usually can only continue for up to 7–10 passages^[68]. Their pluripotency means PSCs can differentiate to any somatic cell types derived from the three germ-layer lineages. Protocols have been reported for differentiation of PSCs into various cell types including neural progenitor cells, hepatocytes, CMs, and chondrocytes with varied degrees of sophistication dependent on the specific cell types for tissue engineering and regenerative medicine applications^{[69, 70]}. Human ESCs (hESCs) were first derived by Thomson et al. from the inner cell mass of human blastocysts in 1998^[71]. There are now more than 1000 registered hESC lines, and there might be even more unregistered lines being derived in the assisted reproductive centers of hospitals worldwide^[72]. The sources of hESCs are abundant, but their derivation requires the destruction of human embryos. The ethics of this procedure are continuously questioned even though the embryos have stopped development. To overcome the ethical concern, Yamanaka et al. derived the first iPSCs from fibroblasts via a retrovirus mediated-reprogramming approach using Oct4, Klf4, Sox2, and c-Myc^[73]. The iPSCs have similar characteristics to ESCs, namely self-renewal and pluripotency. Recently, iPSCs have been reported to have variant copies in their genome and an epigenetic memory which can trigger their differentiation into the somatic cells of their origin^[74]. The use of retrovirus for reprogramming is also a concern for their clinical applications due to the potential toxicity of retrovirus in vivo. In addition, c-Myc and Klf-4 are known oncogenes which incur concerns for therapeutic applications. This demonstrates that the reprogramming procedure requires further optimization for the clinical applications of iPSCs and their derivatives. More recently, iPSCs have been derived with a lentivirus-mediated approach and non-integrated Sendai virus vectors with the hope to reduce these concerns^{[75, 76]}. Additional approaches using RNAs, proteins, and chemicals-mediated reprogramming have been developed^[77]. In addition, the original four factors can be reduced to three (Oct4, c-Myc, and Sox2), and only one (Oct4) in embryonic neural stem cells specifically^{[76, 78]}. After 20 years of development, the medium and biomaterials (e.g. Matrigel, vitronectin, and laminin) for PSC maintenance have been optimized to be xeno-free and avoid the use of feeder cells. In the latest E8 system, Beers et al. identified the essential 8 factors in the mTeSR1 (Stemcell Technologies, Vancouver, Canada) PSC medium to greatly improve PSC maintenance, showing less than 10% of spontaneous differentiation after long-term culture^[79].

2.4. Advantages of iPSCs for MI Therapy

Somatic tissue-derived adult stem cells including MSCs and CPCs have been used as therapeutics for decades with promising results as aforementioned. MSCs from the bone marrow, fat, amniotic fluid, and umbilical blood are the most commonly used adult stem cells in procedures such as bone marrow transplantation, plastic surgery, and immune disease treatment^[80]. Generally, stem cell therapy is an efficient and reliable treatment with good efficacy. However, adult stem cells usually face the challenges of limited proliferation, varied quality (dependent on donors), and concerns over safety including exogenous pathogens. Virtually all of the concerns on adult stem cells can be resolved by using PSCs and their derivatives (Fig. 2).

Figure 2.

A schematic illustration of the advantages of PSC over other cell sources for stem cell-based therapies for MI. Cardiac progenitor cells (CPCs), cardiac fibroblast cells (CFs), and mesenchymal stem cells (MSCs) derived from the somatic tissues have been used for MI treatment. Although administration of these cells demonstrated therapeutic effects in MI animal models as a result of their cytokine effect, there are major limitations including the inability of the cells to differentiate into beating cardiomyocytes, limits in cell number, donor-dependent variability, and limited proliferation. Importantly, senescence may occur in the cells which decreases their quality for transplantation. PSCs have the capability of self-renewal and pluripotency, which allows for good propagation with no limit in cell number and the ability to differentiate into all somatic cells including MSCs, CFs, and CMs under controlled conditions. PSCs, particularly induced pluripotent stem cells (iPSCs) that do not have the ethical concerns of embryonic stem cells, are the most promising cell source for stem cell-based MI therapy.

2.4.1. Scalability

All stem cell-based therapies require a large number of cells. Usually one injection of MSCs requires 1×10⁸ cells in rats and that cell number should increase to 1×10⁹⁻¹¹ for non-human primates, which is difficult using MSCs because their proliferation declines after 8 passages^[81]. PSCs may prove to be better since they have excellent proliferation and a shorter doubling time than adult stem cells. Moreover, in a 3D spheroid culture system, the PSCs may be propagated to 1×10⁶ cells/ml in a 200-ml plastic bag within 5 days when ~2 × 10⁸ cells in spheroids could be obtained^{[82, 83]}. Kempf et al. applied a 3D perfusion system to produce high quality PSCs in a 100-ml scale of expanding bioreactors, which satisfies the requirements for differentiation, transplantation, and even tissue/organ engineering^[84].

2.4.2. Pluripotency

Adult stem cells have limited differentiation capacity for MI treatment. PSCs not only can be differentiated into neural stem cells (NPCs), hematopoietic stem cells, and MSCs in a well-defined system within a short time, but may also be differentiable to additional cell types that are difficult to obtain from adult stem cells such as CMs, hepatocytes, natural killer cells, dopaminergic neurons, islet cells, spermatocytes, and oocytes^[85]. This makes PSC-derived cells a therapeutic option for wider range of diseases than adult stem cell-derived cells. For example, iPSC-differentiated spermatocytes and oocytes have been used for assisted fertilization in a mouse model^[86]. In the near future, these technologies could be further translated to a non-human primate model and humans^[87]. For MI treatment specifically, it is now well established that only PSCs could be directly differentiated into functional or beating CMs while adult stem cells facilitate cardiac regeneration mainly via cytokine effects as aforementioned.

2.4.3. Quality Control

Adult stem cells derived from somatic tissues usually consist of a heterogeneous population with many types of cells. For example, MSCs from the bone marrow may contain hematopoietic stem cells if purification is not rigorous^[88]. Extensive separation/purification usually increases the cost and time, in addition to decreasing the cell viability and function. The quality of adult stem cells is dependent on the donors and specific traits such as health, gender, and age. These factors can introduce significant variation. To avoid the possibility of propagating a quiescent virus such as HIV or hepatitis C, a prescreen is usually necessary^[81]. It is easier to ensure better quality control with PSCs than MSCs because comprehensive background information can be obtained^[89]. Moreover, PSCs can be completely manipulated with xeno-free, well-defined conditions in a high grade, GMP-compliant facility with pipeline production for clinical use^[82].

PSCs have enormous value as a tool for drug screening and tissue engineering/regeneration^[83]. Compared with animal models, PSCs can reduce animal usage and can gain more veridical data that better mimics in vivo conditions in the human body. Human iPSCs originating from patients carrying specific diseases not only can be used for drug screening, but may also serve as a model to elucidate disease pathologies^[90]. Wu et al. developed iPSCs from an individual with a myocardial disease^[91]. In the study, they differentiated the patient’s iPSCs to CMs and compared the functionality to normal CMs. They found that the patient had a genomic mutation which prevented sarcomere maturity and limited myocardium function^[92]. Recently, more iPSC lines have been derived specifically for MI research and drug screening^[93].

3. Derivation of CMs from PSCs for MI Therapy

3.1. Signaling Pathways in Heart Development

The heart is the first organ that forms during development^[94]. Mammalian fate-mapping studies show that the embryonic heart is derived from the mesodermal layer produced during gastrulation, when the primitive epiblast layer undergoes an epithelial-to-mesenchymal transition as an early transient structure called the primitive streak^[95]. The pre-cardiac mesoderm migrates to the anterior portion of the embryo, where it forms the cardiac crescent^[96]. At this stage, there are two major cardiac progenitor pools called the first heart field (FHF) and the second heart field (SHF)^[97]. They contribute to the major portions of the ventricle, atrium, and the inflow and outflow tract myocardium. FHF progenitors mainly give rise to the posterior structures of the primitive heart tube including the atrium and left ventricle. By contrast, SHF cells mainly give rise to the right ventricle and inflow and outflow tracts^{[97, 98]}.

To efficiently differentiate hPSCs into CMs, it is necessary to control the signaling pathways that regulate lineage specification. Studies of heart development have demonstrated that the bone morphogenetic proteins (BMPs), Wnt, Activin/Nodal, Fibroblast growth factor (FGF), and retinoic acid (RA) pathways play pivotal roles in establishing the early stages of mesoderm formation and cardiogenesis^[97]. BMP signaling involves a subset of the transforming growth factor (TGF) gene family, which binds to specific cell surface receptors exerting their function via activation of Smad transcription factors. BMP signaling is crucial for the emergence of mesoderm precursors and early FHF cells. BMPs are expressed in the lateral plate endoderm^[99]. After fusion of the heart tube, BMP signaling is activated in the myocardium and endocardium which triggers the SHF progenitors to differentiate. Wnt signaling involves multiple complex signaling cascades^[100]. The β-catenin pathway specifically controls the canonical and noncanonical pathway activation^[101]. During cardiac development, Wnt ligands act as regulators of early cardiogenesis and are involved in the formation of the embryonic axis and mesoderm, and the conversion of the heart tube to the heart chambers^{[102, 103]}. Mesoderm formation via Wnt/β-catenin is an essential step in CM differentiation. High levels of β-catenin in the precardiac mesoderm prevent the fusion of cardiac crescent cells to the linear heart tube, whereas the loss of functional β-catenin prevents cardiac looping^[103]. Suppressing Wnt/β-catenin through Dkk-1 initiates the cardiac mesoderm to undergo CM specification. However, the proliferation, migration, and differentiation of SHF progenitors requires β-catenin signaling^{[104, 105]}. Wnt signaling controls the regional expansion of ISL-1+ progenitors in the SHF-derived right ventricle during the gastrulation process^{[102, 104]}. Nodal/Activin is a member of the TGF-β family, which induces mesoderm differentiation in a concentration-dependent pattern. Activin functions within the Wnt/β-catenin pathway to determine the formation of mesendoderm. Nodal signaling regulates the formation of the anterior mesoderm and endoderm, which plays a direct role in cardiac mesoderm formation. Cardiogenic factors such as BMPs, Activin A, Wnt inhibitors, and FGFs trigger the engagement of Nodal signaling in mesoderm-to-CM differentiation. FGFs are involved in fate decisions of the mesoderm, mesendoderm, and endoderm in humans and mice. In chick embryos, endoderm cells adjacent to the anterior lateral plate mesoderm cells secrete FGF-1 and FGF-4 which induce differentiation of posterior mesoderm cells to cardiac cells. A mutation of the FGF8 gene in zebrafish embryos reduces the number of ventricular CMs. Ectopic FGF8 promotes lateral expansion of the heart field, while ectopic FGF2, FGF4, and BMP4 facilitate cardiac differentiation in posterior non-cardiogenic mesoderm^[106]. RA signaling from the lateral plate mesoderm has been demonstrated to define the posterior of the heart tube^[107]. RA restricts cardiac specification at early stages and regulates the normal patterning of the posterior and anterior heart at later stages.^[107] RA exerts its functions on atrial differentiation and expansion by increasing Tbx5 expression^[108]. In chick and mouse cardiac development, Tbx5 is expressed in the precardiac mesoderm. Later, this expression becomes restricted to the posterior part of the looping heart tube, atrium and the left ventricle^[109].

3.2. CMs Derived from PSCs with Serum and Coculture Method

CMs can be derived from hPSCs using both spontaneous and growth factor-directed approaches. The initial methods for generating CMs from hPSCs was reported in 2001 involving the formation of 3D cell aggregates called embryoid bodies (EBs) for spontaneous differentiation of CMs in serum-containing medium^[110]. The EBs began beating similar to immature CMs within one week after re-plating, but at an efficiency usually less than 1%^[110]. The majority of the cells from these methods were heterogeneous cell types which gave rise to all 3 germ layers^[110–114]. Coculture of hESCs with a mouse visceral endoderm-like cell line (END-2) could result in a beating-phenotype^[113]. END2 cell-conditioned medium can also be utilized to drive hESC cardiac differentiation^{[115, 116]}. The addition of ascorbic acid, PGI2, or SB203580 to hESC-END2 cocultures or END2-conditioned medium increases the hESC-derived CM content (8%−20%)^[115–117]. Based on morphological and electrophysiological parameters, the predominant cell type derived by using these methods were ventricle-like CMs^[118]. Besides the low efficiency (~1–20%), the reproducibility of these protocols is poor for a number of reasons, including undefined components such as serum, coculture with mouse cells, and heterogeneous cell populations differentiated from EBs under undefined conditions^{[117, 119]}. Therefore, subsequent protocols have been focused on removing serum or feeder-based culture and developing fully defined conditions to improve the quality, efficiency, and reproducibility of CM differentiation, as summarized in Table. 3.

Table 3.

A summary of methods for cardiomyocyte differentiation from hPSCs.

hPSC lines	HPSC culture methods	Initial stage of CM differentiation	CM differentiation condition	Contracting initiation	Differential efficiency and subtypes	Length	Reference
iWT.D2.1, iBM76.3, WT1.Bld2	H8, Geltre-coated plates	Monolayer	PRMI1640, B27, CHIR, IWP2, RA, Lactate	Day 8	≈90% atrial cells	141 days	[137]
HES3-NKX2.5gfp/w, MSC-iPSC1, hES2	MEF	EBs	Activin A, BMP4, bFGF, IWP2, VEGF, RA	ND	≈86% atrial cells ≈6% ventricular cells	20 days	[235]
H9–human cardiac troponin T–eGFP, H9, H7, hiPSC-U-Q1	Matrigel in mTeSR1	Monolayer EBs	PRMI/B27, CHIR, Papa, XAV939, KY02111	Day 9	>90% cTnT+ cells	30 days	[131]
H9, H13, hES03, ES03-WT1-eGFP, 19-9-7, 19-9-11	Matrigel or Synthemax in mTeSR1 or E8	Monolayer	PRMI, CHIR, IWP2, GiWi2 LaSR basal medium, bFGF, TGF-β1	Days 7–9	>90% WT1+ cells on D12 Subtypes: Pro-epicardium WT1+ cells, Smooth muscles cells	2 months,	[130]
H7, T16, XVF4, Y7	VTN-coated plate in E8 medium	Monolayer	Chemical defined medium: PRMI1640 insuliin, rhtransferrin, antioxidants L-ascorbic acid, trolox, NAC and sodium pyruvate, sodium selenite, ethanolamine, L-carnitine, RA, BMS493	ND	>76% cTnT positive cells	>100 days	[132]
HES3-NKX2.5gfp/w, hES2, MSC-iPS1	MEF	EBs	BMP4, Activin A, bFGF, IWP2, VEGF, SB431542, RA, PD173074	ND	NKX2.5− cTnT+ cells Sinoatrial node-like pacemarker cells	50 days	[140]
HES3-NKX2.5gfp/w, LQT2-hiPSCsN996I and LQT2-hiPSCsCorr	MEF	EBs	BMP4, Activin A, VEGF, SCF, CHIR, SB431542, IGF-1, SAG, DMH1, SU5402, bFGF, XVA939,	Day 4	>80% Pacemarker-like CMs Ventricular-like CMs SMCs	>40 doublings	[138]
	MEF	EBs d7 replated	Activin A, BMP4, SCF, VEGF, CHIR, RA	Day 10	>85% atrial-like cells <1% nodal-like cells	31dyas	[134]
H9G1, HES3 ISL-1-RFP	Matrigel in mTeSR1	EBs	BMP4, Activin A, WNT3A, EDN1	3–4 weeks	ISL-1+ CVPs Subtypes: CM, endothelial, SMCs		[236]
H9, BHX	Chemically defined condition Brons 2007	Monolayer	CDM-PVA, FGF2, BMP4, WNT3A, RA, IWP2, PDGF-BB, TGFβ1	Day 10	Epicardium SMCs	15 days	[237]
H7, H9 11 hiPSC lines	Matrigel or Synthemax in E8	Monolayer	Chemical-defined condition: PRMI1640, L-ascorbic acid 2-phosphate, rice derived recombinant human albumin	Day 7–9	>95% TNNT2+ cells, 57% immature ventricular-ike cells with atril-like and nodal-like cells at D30–35	60 days	[133]
hES2, H7	MEF	EBs	BMP4, Activin A, bFGF, DKK1, VEGF, SB,	ND	>85% WT1+ P-Epi population on D15 Subtypes: Epicardial lineage, SMCs, and fibroblasts	24 days	[126]
6-6-9, 19-9-11, IMR90C4, H9, H13, H14	Matrigel or Synthemax in mTeSR1	Monolayer	Small molecules method: RPMI/B27-insulin, IWP4, IWP2.	Day 10	Up to 98% cTnT+ cells Subtypes: 91% ventricular-like action potential, 8.5% atrial like action potential, 80–98% ventricular CM	30 days	[100, 101]
hES3, H9, MEL1	Matrigenl in mTeSR1	Monolayer	RPMI1640, B27, BMP4, ActivinA, IWP4, IWP1	Day 16	≈60% MYHlo+ cells, NKX2.5 positive	18 days	[238]
hES-1, hES-3, H1, H9, 253G1, IMR90–1, 90–4, RCHIPC0003, Monkey ESC: CMK6.4	MEF	Monolayer Gelatin or laminin coating	Serum-free, defined condition: BIO, CHIR, KY0211, XAV939	Days 8–10	84–98% cTnT+ cells	30 days	[128]
H7	MEF-CM	Monolayer	RPMI/B27, BMP4, bFGF, Activin A, DKK1, Noggin, RA	Day 10	≈94% atrial-like cells	60 days	[136]
H9, H1,	MEF, Matrigel in MEF-CM prior to plating	Monolayer	Chemical-defined condition: DMEM, N2/B27, CDM, PRMI, B27, IWR-1, SB431542, Purmorphamine	Day 25	70–80% Nkx2.5+ cells >50% cTnT and a-Actinin positive CM	30 days	[239]
H1, H7, fetal lung IMR C1 iPS, neonatal foreskin C1 iPS	MEF	EBs d4 replated	FBS, BMP4, IWR1	Day 12	≈35% beating EBs ≈15% CM	20 days	[125]
H9-MYH6-mCherry hESC	MEF	EBs d4 replated	Activin A, BMP4, IWR, IWP-3, XAV939	Day 7	30% CM	12 days	[127]
H7, cGATA6-EGFP transgene hESCs	MEF-CM	monolayer	RPMI/B27, Activin A, BMP4, neuregulin-1α, neuregulin-1β, anti-neuregulin-1β neutralizing antibody, AG1478, EGF, HB-EGF, betacellulin	Day 12	≈50% nodal like AP cells in H7	60 days	[139]
H1, MSC-iPSC1, Y2–1	MEF, feeder depleted on Matrigel before differentiation	EBs	StemPro-34, BMP4, bFGF, Activin A, DKK1, VEGF,	Day 4	50–70% CTNT+ cells 10–40% SMA+ cells	20 days	[124]
H1, hES2	MEF, feeder-depleted on Matrigel before differentiation	EBs	Serum-free medium BMP4, bFGF, Activin A, VEGF, DKK1	Days 7–10	40–50% CTNT+ cells	14 days	[123]
hES2, HES3-GFP, HES4	Human fibroblasts	EBs	END2-CM, SB20358	Days 9–10	≈20% CM	12 days	[115]
HES2, HES3-GFP, NL-HES1 and 2	Human fibroblasts	EBs	END2-CM Basic SF, PGI2, SB203580	Days 9–12	>40% beating EBs ≈10% CM	12 days	[116]
hES2, hES 3, hES 4	MEF	Monolayer	hESC-END2 coculture, ascorbic acid.	Day 5	20% α-actin+ cells	12 days	[117]
H7	MEF-CM	Monolayer EBs (4d) replated	RPMI-B27, Actin A, BMP4 DMEM, FBS	Day 12	>30% CMs 82.6±6.6% after Percoll enrichment	2 weeks 3 weeks	[121, 240]
H1	MEF	EBs	DMEM, FBS	Day 7 after plating	13% beating EBs	13 days	[122]
H1, H7, H9, H9.1, H9.2	MEF	EBs (4d) replated	DMED, FBS, 5-aza-dC	Day7,	70% beating EBs at D16, 3–5% CMs Percoll density centrifugation (70%)	70 days	[240]
hES2, GCT27X	MEF	Monolayer Coculture with END-2	HES-END-2 coculture DMED, FBS	Day 11 after co-culture with END-2	15–20% of the wells contains one or more areas of beating muscle	2–3 weeks	[112, 118]
H9.2	MEF	EBs (7–10) replated	DMED, FBS	Days 4–22 after plating	8.1% beating EBs	5 weeks	[110, 111, 113, 114]

EB: embryonic body; CM: conditioned medium; SMCs: smooth muscle cells; FBS: fetal bovine serum; MEF: mouse embryonic fibroblast; RA: retinoic acid; cTnT: cardiac troponin T; VEGF: vascular endothelial growth factor; SCF: stem cell factor; HB-EGF: heparin-binding EGF-like growth factor; EGF: epidermal growth factor; ND: not defined

3.3. Defined, Cytokine-Free, and Xeno-Free Conditions for hPSC-CM Differentiation

The advances in understanding the signaling pathways regulating cardiogenesis have facilitated the development of more efficient and controllable protocols for hPSC-CM generation. The early attempts at defined culture conditions use cytokines or growth factors for manipulating multiple signaling pathways to mimic the in vivo embryonic cardiac development. Usually, hESC-derived CMs exhibit a good morphology and performance under culture conditions with serum. To avoid the use of serum, the addition of IGF-1 or IGF-2 to a serum-free medium can maintain CM growth via the PI3-kinase/Akt signaling pathway^[120]. Laflamme et al. reported that combinations of Activin A and BMP-4 induce cardiovascular development in monolayer cultures^[121]. More than 30% of CMs could be generated with the H7 hESC line, which is more efficient than using EBs with serum induction. The engrafted human CMs attenuate ventricular dilation and preserve regional and global contractile function of the hearts with MI in immunodeficient mice^{[121, 122]}. The heart in mammals is embryonically derived from the mesodermal layer, which arises from the primitive streak after gastrulation. Activin A and BMP4 induce a primitive streak-like mesoderm in hPSC differentiation. The combination of Activin A, BMP4, bFGF, VEGF, and DKK1 in a serum-free media induce hESC EBs to generate a KDR low/c-Kit (CD117) negative cell population that displays cardiac, endothelial, and vascular smooth muscle potential^[110]. These KDR low/c-Kit (CD117) negative cells, when plated in a monolayer, differentiate and generate 50% beating CMs^[123]. Activin A and BMP4 induce FLK-1+/PDGFR-α+ cardiac mesoderm in EBs and promote the generation of CMs in mouse and human PSC differentiation. This enriches the cTnT+ CMs to more than 60%^[124]. Furthermore, stage-specific optimization of Activin/Nodal, BMP, and Wnt signaling could promote CM differentiation^[111]. CMs are developed from KDR+/PDGFRA+ mesoderm in the presence of the Wnt inhibitor DKK1 and the Activin/Nodal/TGF-β inhibitor SB431542^[124]. Ren et al. discovered that early treatment of hPSC cells with BMP4 followed by late Wnt inhibition via the small molecules IWR-1 and IWP-1 leads to increased production of CMs^[125]. Stage-specific application of BMP and Wnt ligands to EBs also generate cardiac myocyte lineage cells, following the stimulation of cardiovascular specification via DKK2, VEGF and SB431542^[126].

Use of recombinant cytokines is expensive for large-scale production. Some studies reported the differences in differentiation propensity among cell lines, and the optimal concentrations of growth factors such as Activin and BMP4 for cardiac induction must be optimized for specific hPSC lines^[124]. Small molecules may have great potential for replacing recombinant cytokines and unknown factors in serum to reduce the cost^[116]. They are suitable for making defined media for cost effective and large-scale culture. It has been shown that Wnt inhibition alone is sufficient to derive CMs from hESC-originating mesoderm cells^[127]. Lian et al. reported that inhibitors of Wnt signaling produce up to 98% yield of functional CM from multiple hPSC lines^{[100, 101]}. Activation of the canonical Wnt signaling pathway promotes the generation of primitive streak cells. CHIR99021 treatment at the initial stage leads to greater than 90% primitive streak cells. Following that, the chemical cocktail consisting of IWR-1, SB431542, and purmorphamine produces 70–80% NKX2.5+ cells in 25 days. The Wnt inhibitor, KY02111, promotes the differentiation of hPSCs to CMs at an increased efficiency under serum-free, defined conditions^[128]. The stage-specific application of Gsk3 inhibitors, CHIR99021 and GiWiGi, and the Wnt inhibitor, IWP2, produces CMs with a high yield (> 90%) from multiple hPSC lines^{[84, 129]}. Culture containing the TGF-β inhibitors, A8301 or SB431542, allows long-term maintenance of hPSC-derived CMs for 2 months (≈25 population doublings)^[130]. In addition, coordinating Wnt and other signaling pathways leads to efficient hPSC-CM production. For example, using rapamycin (mTOR inhibitor) and CHIR 99021 during initial stages directs efficient CM differentiation^[131]. A chemical-defined and albumin-free medium (S12) was developed to produce more than 76% cTnT+ cells^[132]. Another chemically defined medium consisting of the basal medium RPMI1640, l-ascorbic acid, 2-phosphate, and rice-derived recombinant human albumin produces up to 95% TNNT2+ CMs. This protocol was effective for 11 PSC lines^[133].

Although many protocols have been developed to efficiently generate hPSC-CMs with high yield and purity, most of them result in heterogeneous CM subtypes consisting predominantly of those with ventricular characteristics with a small amount of atrial-like and nodal-like cells^{[134, 135]}. Substantial evidence indicates that retinoic acid (RA) plays a role in regulating fate specification of atrial versus ventricular CMs, and it can be used to generate 85%−95% hESC-atrial CMs ^{[134, 136, 137]}. Studies on early cardiac developmental stages reveal that the specification of ventricular and atrial CMs is dependent on induction of the appropriate mesoderm^{[137, 138]}. RA specifies that the RAHDL2+ mesoderm produces atrial CMs and the CD235a+ mesoderm gives rise to ventricular CMs^[132]. For differentiation into the nodal subtype, inhibition of NRG-1β/ErbB signaling enhances the proportion of hESC-CMs showing the nodal phenotype^[139]. By enforcement of expression of the c-Myc oncogene, NKX2.5− cells with pacemaker-like characteristics could be generated from hPSC-derived cardiovascular progenitor cells^[138]. Later, stage-specific addition of BMP, SB431542, and RA blocks the development of NKX2.5+ cardiac progenitors and increases the population of the sinoatrial node-like pacemaker cells (SANLPCs) from 26%−55% in hPSC-CMs. After isolation of SIRPA+CD90-cells, the NKX2.5−cTNT+ SANLPCs could be highly enriched to 70%−90%^[140].

3.4. Large-scale Production of hPSC-CMs in 3D Culture

Cardiac tissue engineering/regeneration and innovative drug screening requires a constant supply of functional hPSC-CMs^[135]. Significant progress has been made in cardiac differentiation by modulating multiple signaling pathways as aforementioned. Chemically defined media facilitate human CM production from hPSCs at reduced costs and improved reproducibility. Stage-specific culture protocols can efficiently drive cardiac differentiation to over 80% purity^[141]. However, most of the cardiac differentiation studies were conducted with adherent 2D culture methodology that require large surface area for culturing, multiple enzyme treatments to detach the cells from the 2D substrate, and a large amount of medium supply, which limits scalability to obtain the large number of CMs needed for clinical use. Therefore, a 3D culture system is needed for large-scale production of hPSC-CMs (Fig. 3)^[142].

Figure 3.

A schematic illustration of the advantages of PSC production in 3D over 2D culture. Compared to the conventional 2D culture, it is easy to scale the 3D culture method up for large-scale production of PSC spheroids. In contrast, the monolayer sheet of PSCs cultured on 2D substrate can only be scaled out using a lot of culture flasks or dishes. In addition, there is no need to detach the cells under 3D culture while cell detachment from the substrate is necessary for 2D culture, which may cause significant loss of the extremely stress sensitive human PSCs.

A complex induction procedure on EBs has been reported to be effective in cardiac differentiation in suspension culture systems using growth factors or small molecules^{[142, 143]}. Through the formation of cell aggregates in non-adherent plates, EBs recapitulate many aspects of embryo development and result in differentiation to cells of the three embryonic germ layers similar to gastrulation of an epiblast-stage embryo in vivo ^[144]. Hemmi et al. used suspension dishes to form EBs and induced CM differentiation in a large suspension culture system (125-ml stirred flask)^[145]. Due to the highly variable size and shape of EBs, the portion of α-actinin+ CMs is ~25% after 18–20 days of differentiation. As a result, metabolic purification is needed to enrich the CMs. Thus, for the spontaneous formation of EBs, the differentiation efficiency is highly influenced by undefined EB sizes and morphology. The control of cell number, size, shape, and quality of EB formation are important factors in optimizing the EB-based hPSC-CM differentiation strategies^[94]. To generate EBs, several techniques and tools have been developed including hanging drop, microwells, and micro-patterned ECM surfaces^[143]. Studies have shown that the frequency of endoderm cells decreases when the size of EBs increases^[146]. A total of 1000 cells per aggregate differentiate more efficiently into CMs compared with 100 or 4000 cells per aggregate^[147]. The 400-μm micro-patterned EBs give the highest yield of hPSC-CMs compared to 200-μm and 800-μm EBs. Small sizes (~200 μm) give rise to predominantly primitive endoderm and large sizes (~800 μm) limit the growth and proliferation^[148]. However, Mohr et al. showed that 300-μm EBs produce the highest percentage of contracting EBs^[119]. In addition, a cyclic perfusion-based strategy generated ~85% CMs in 100-ml stirred bioreactors by controlling the initial EB size to be 531.6 ± 35.3 μm^[145].

Despite the high CM differential potential of EBs produced in stirred bioreactors, the initial stage of hPSC expansion for EBs is still based on adherent culture as monolayers. A large-scale hPSC expansion system is required for more efficient and scalable production of hPSC-CMs. Various suspension cultures of hPSCs have been developed with microcarriers, hydrogels, and polymers as supporting matrices that increase the production capacity (Table 4)^[149]. Other approaches for forming hPSC spheroids involve self-aggregation in the absence of supporting materials. For example, dissociated single hPSCs could be cultured in hPSC culture media with ROCK inhibitor to form cell aggregates. The aggregated hPSCs can be expanded and passaged every 7–10 days in a suspension system^[150]. After 5 days of culture, the hPSC aggregates are 175 ± 25μm and CM differentiation is then initiated. At 30 days later, 100% of the EBs are beating and up to 90% cTnT+ CM are produced^[151]. By optimizing the PSC aggregate size, small molecule concentrations, induction timing, and agitation rate, 1.5–2×10⁹ hPSC-CMs are produced in a 1-liter spinner flask with a high purity (> 90%)^[152]. Instead of enzymatic dissociation, a mechanical subculture is performed in which cells are passed through a mesh filter to form aggregates of controlled sizes^[153]. Using this culture system and a small molecule defined medium, CMs with up to 95.7% purity could be differentiated from human and monkey PSCs^[83].

Table 4.

A summary of 3D culture systems used for large-scale hPSC-CM differentiation.

hPSC lines	HPSC culture methods	Initial stage of CM differentiation	CM differentiation condition	Contracting initiation	Differential efficiency and subtypes	Culture vessel	Yield	Length	Reference
H7	Suspension culture [241]	ESC spheroids	RPMI, B27, CHIR, IWP4	Day 8	>90% cTnT + CM	Spinner flask	1.5–2.0×109 cells/L	32 days	[152]
H9, BG02 Monkey cESC3–12, ciPSCJ1	Suspension culture [153]	ESC spheroids	Bio, CHIR, XAV939, IWP2	Days 7–8	≈95% cTnT+ CM	Flask	1×107 cells/20mL	9 months	[83]
hESC lines (RH5, RH6, R725.1, R661.5, and R662.2) hiPSC lines (VC645–9, VC913–5, VC618–3, and VC646–1)	Suspension culture [150]	hPSC spheroids	PRMI, B27, CHIR, IPW2, SB431542, Purmorphamine	Day 7	≈100 beating spheroids ≈90% CM cTnT+ 3–8% αSMA+	Stirred spinner flask	8–9×107 cells/100 mL	30 days	[151]
HES3 NKX2.5eGFP/W, hHSC_F1285T_iPS2	MEF	EBs	PRMI/B27, CHIR, IWP2	Day 6	≈85% ventricular-like CM	Stirred bioreactor	4×106 cells/100 mL	21 days	[84, 145]
253G4, khES-2	MEF	EBs	BMP4, StemPro-34, BMP4, Activin A, IWR-1		≈99% α-actinin postitive	Spinner flasks	1×107 CMs/100 mL	27–30 days	[145]
H9, hES2	MEF	Micropatterned EBs	FBS	Day 3	≈50% contracting EBs at Day16 3–5% CM	Roller-bottle culture	4×105 cells/mL	18 days	[148]

αSMA: α-smooth muscle actin.

4. Application of Biomaterials Engineering in CM Differentiation and MI Therapy

4.1. Biomaterials for CM Derivation and Maturation in Vitro

It is important to recapitulate the native environment of CMs in vivo to improve CM differentiation and maturation. It is particularly important to mimic the material properties of the myocardium which contribute to functions like action potential, contractility, and the electrical profile of CMs^[154]. It is reported that the elastic modulus of the adult mouse myocardium is 5–40 kPa during diastole when measured with atomic force microscopy. It has been reported that CMs may perform best on a substrate with a modulus of 10 kPa^{[155, 156]}. The materials used for cardiac tissue engineering should be low in cytotoxicity, capable of maintaining the physiological characteristics of the extracellular matrix (ECM) in medium, and chemically tunable. The modulus of the materials affects the CM growth, maturation and function^[157]. Hydrogels are the most common materials used as the substrate for CMs. Hydrogels of gelatin-methacrylate, hyaluronan, alginate, and gelatin have hydrophilic surfaces or specific moieties that have affinity to the receptors on the cell membranes^[158]. Hydrogels may be cast into a specific pattern, like paralleled channels, to allow for improved myofiber alignment and sarcomere maturity of CMs along the patterns^[158]. Boothe et al. reported that the modulus of the substrate significantly changed the action potentials of CMs with whole-cell patch clamping^[154]. They cultured CMs isolated from the neonatal rat ventricular heart on a polyacrylamide hydrogel substrate with a tunable modulus ranging over 1 to 25 kPa. They found that the CMs grown on the hydrogels with a 9 kPa elastic modulus have the longest action potential duration, similar to that of CMs in vivo. Meanwhile the dynamic voltage of calcium flux decreases with increasing modulus of the substrate above 9 kPa, which however, has no correlation to the gene expression of the associated L-type calcium channel receptor^[154]. Wheelwright et al. also reported that the substrate modulus significantly changed the contractile force of human iPSC-derived CMs^[159]. They used the iPSC-derived CMs grown on substrates with a micro-pattern for 3 months in vitro and showed that the cells developed a similar phenotype to that of neonatal rat ventricular CMs in terms of size, shape, and force production. By tuning the modulus of the polyacrylamide gel, they changed the modulus to recapitulate the embryonic-like and adult myocardium-like ECM. They found that hiPSC-CMs produce maximal forces on substrates with the lower moduli and significantly less force when assayed on adult myocardium-like substrates with a greater modulus. The strain energy measurements confirmed these observation. This indicates that biomaterials can mimic the physiological conditions of the ECM in vivo and that modulation of the elastic modulus can cause functional differences in CMs.

Both natural and synthetic biomaterials have been used. The most commonly used synthetic materials include the aforementioned polyacrylamide together with polydimethylsiloxane (PDMS)^[160], polyacrylonitrile (PAN)^[161], poly-ε-caprolactone (PCL)^[162], polyethylene glycol ^[69], and carboxylated PCL (cPCL)^[163]. To enhance the adhesion of CMs on the biomaterials substrates, peptides corresponding to the integrin motifs in cells have been widely used to modulate the chemical structure of biomaterials in order to better mimic the native ECM of CMs. Patel et al. reported a synthetic substrate composed of a co-polymer of isobornyl methacrylate and tertbutylaminoethyl methacrylate that supported hESC-derived CM growth for more than 15 days, in addition to showing significantly lengthened sarcomeres compared to the control grown on gelatin^[164]. Chun et al. also reported that iPSC-derived CMs show improved contractility and mitochondrial function when cultured on a PEG-PCL mixed substrate compared to the control without modification. Meanwhile, cardiac myosin light chain-2v, cardiac troponin I, and integrin alpha-7 are also increased on the CMs grown on the modified substrate. Interestingly, the substrate may help the troponin I (TnI) isoform of CMs to switch from fetal slow skeletal TnI (ssTnI) to postnatal cardiac TnI (cTnI). This is strongly suggestive of CM maturation in vitro. Importantly, CMs grown on synthetic biomaterials do not show any evident adverse effect of cell toxicity or change in karyotype^[165]. The arginine-glycine-aspartic acid (RGD) sequence which is found in collagen, fibronectin, fibrins, and laminins, is the most common adhesion peptide used in synthetic myocardial scaffolds^[166]. Recently, the YIGSR sequence, specific to laminins, has been used in synthetic substrates to increase the attachment of CMs to the ECM^[167]. Furthermore, conductive materials such as carbon nanotubes and graphene may assist in electrical conduction between CMs^[168]. Ban et al. reported that rat neonatal CMs can grow on an amphiphilic peptide nanomatrix gel (PA-RGDS) incorporating the cell adhesive ligand Arg-Gly-Asp-Ser (RGDS) over 7 days with 90% survival. They also showed that the CMs injected together with PA-RGDS have a threefold higher engraftment and better echocardiography results than the RGD-free control at 2 weeks post injection. Their results demonstrate that PA-RGDS encapsulation increases the survival of CMs in engraftment and helps to partially recover cardiac function after MI^[169]. The application of biomaterials with CMs may help to elucidate heart development and aid in developing an advanced method for transplanting cells into the ischemic lesion post MI^[170]. Biomaterials research can also help to provide better chemically defined conditions for large-scale culture and maturation of CMs than that contemporarily available.

4.2. Cardiac Patches/Sheets for MI Therapy

PSC-derived CMs are usually dissociated and enriched following cell sorting to obtain a large enough cell number, and may be engineered with biomaterials for transplanting into the lesion of MI. A cardiac patch/sheet is a thin-layer of engineered heart tissue made of functional CMs on a biomaterial matrix that can be placed on top of the infarct area on the left ventricle post MI (Fig. 1)^[171]. The use of a cardiac patch is a good alternative method to heart transplantation since despite both requiring open-chest surgery, a cardiac patch is much simpler to implant than a whole heart that also has a limited supply. PSC-derived CMs can either be seeded on the substrate to form a CM layer, or PSCs can be seeded and then differentiated to functional CMs^[172]. These CMs may have improved alignment when the substrate has an aligned micro-pattern which may help to enhance the contractile force of the cells^[173]. By stacking multiple thin cardiac layers, it is possible to make a cardiac patch as thick as 1 mm which demonstrates strong contractile function^[157]. Gao et al. generated a relatively large cardiac patch with dimensions of 4 cm × 2 cm × 1.25 mm (length × width × thickness)^[174]. The cardiac patch may attach on the surface of the lesion and give physical support to the CMs. The cardiac patch is retained on the lesion by the intrinsically adhesive surface of the biomaterials. Natural materials including matrigel and silk proteins, in addition to synthetic polymers like poly(N-isopropylacrylamide) (PNIPAAm), polyethylene glycol ^[69], and poly(D, L-lactic-co-glycolic acid) (PLGA) have been used for the cardiac patch engineering^[175]. These biomaterials can be tuned to have a modulus similar to the heart tissue which may improve the CM survival and accelerate CM maturation in vivo.

Nanofibers fabricated with electrospinning have also been used for constructing substrates to generate cardiac patches. It may provide a vast surface area for cell attachment, an aligned pattern for enhancing the myofiber formation in CM, and a sponge like structure to facilitate angiogenesis. Li et al. reported a cardiac patch generated from PLGA polymer seeded with high-purity CMs derived from iPSCs^[173]. Their results indicate that human iPSC (hiPSC) derived CMs may grow well on the nanofiber and form multilayered and elongated phenotypes with increased cardiac gene expression after layer stacking. Their cardiac sheet gives promising results after transplantation in vivo, including high CM survival, improved electrical coupling of CMs in the cardiac patch with host cardiac tissue, and suppression of the re-entrant spiral wave after surgery on a nude rat ventricle with MI. Leor et al. reported an alginate sponge (as a substrate) seeded with fetal CMs for myocardial repair^[176]. Once the CMs maintained a high viability in the scaffold and demonstrated spontaneously beating, the cardiac patch was implanted onto the infarcted rat myocardium. Their results show an improvement of the cardiac function and prevention of the imminent deterioration of the lesion in the myocardium post MI. Interestingly, the alginate scaffold eventually degrades and the space is filled with collagen fibers alongside CMs, MSCs, and vascular endothelial cells to help the neovascularization in the grafted cardiac patches on heart. In addition, it was shown that adding VEGF and bFGF in the nanofibers during electrospinning can further help the CMs to survive in the scaffold after implantation and facilitate neovascularization to improve cardiac performance. Chen et al. reported a scaffold composed of poly-L-lactic acid (PLLA) and human endothelial cells for constructing a vascularized cardiac patch^[177]. They first incubate the endothelial cells within the material to form a vasculature and then the pre-vascularized cardiac patch is seeded with a mixed cell population of CMs (derived from hESCs), hESC-derived endothelial cells (hESC-ECs), human umbilical vein endothelial cells (HUVECs), and embryonic fibroblasts (EmFs). After transplantation into an MI rat model, the authors observed a capillary-like network of endothelial cells which was connected to the host blood vasculature and supported CM survival^[178].

Nagase et al. discussed the substrate made from poly(N-isopropylacrylamide) (PNIPAAm), a temperature-responsive polymer for CM attachment and growth^[179]. The CMs can grow on the PNIPAAm substrate at 37 °C and easily detach from the dish as a cell sheet at 32 °C because the PNIPAAm becomes hydrophilic and non-adhesive to the culture dish at this decreased temperature. Using this strategy, it is possible to obtain an intact CM sheet without damaging the cell-cell connections and ECM proteins. This is very important in order to maintain the CM’s electro-signal transduction and contractile function^[179]. The authors also transplanted a cardiac patch of 1 mm thick onto the MI lesion in a rat heart^[180]. They found that the cardiac patch can integrate with the host tissue with identified cell-cell junction formation between transplanted CMs and host CMs. Their electrophysiological data showed a reduced area of left bundle branch blocking compared to controls with no treatment or treated with a fibroblast sheet. Though the fluid in the pericardium contains nutrients which can help the transplanted CMs to survive, the effect is limited because the small volume of the fluid may not be able to provide enough oxygen and nutrients for the large number of CMs on a cardiac patch (2–8×10⁸ cells). To overcome the shortage of oxygen and nutrients in the pericardium, Radisic et al. established a porous solid scaffold made of poly(glycerol-sebacate) (biorubber) which can be preloaded with an oxygen-carrying perfluorocarbon (PFC) emulsion in the packed parallel microchannels with supportive medium containing various growth factors^[181]. They found increased cell survival in the scaffold post transplantation and improved functional properties compared to control conditions. Despite this, there is significant concern over the invasiveness of the cardiac patch transplantation procedure (e.g., open-chest surgery). Due to the limited diffusion length (a few hundred microns) of oxygen and nutrients in highly cellularized system such as the cardiac patch in vivo, the surgical intervention may need to be performed for multiple times as only a thin cardiac patch/sheet can be transplanted during each surgery to ensure a high survival of the CMs in the patch/sheet. This greatly increases the pain and costs for patients and may not be an option for many patients^{[182, 183]}.

4.3. Engineered Stem Cells with Injectable Biomaterials for MI Therapy

Injecting CMs in suspension is one of the most intuitive methods that can be done in multiple ways including the direct intramyocardial injection with minimally invasive thoracoscopy, minimally invasive catheter-mediated injection via the ventricle and coronary artery, and non-invasive injection into the peripheral vein^[184] (Fig. 1). However, this cell suspension injection is sub-optimal because single cells are difficult to retain in the infarcted lesion, and most of them are lost/dead within a few days^[169]. When the dissociated stem cells are embedded in a porous tissue scaffold of hydrogel for injection, retention can be improved^[185]. However, the rate of oxygen diffusion in the scaffold is often insufficient in vivo and the cells eventually undergo apoptosis if the scaffold formed after injection is more than a few hundred microns^[186]. In addition, the dead cells, including the local and transplanted cells in MI, may induce the recruitment of a significant number of immune cells (e.g., macrophages) that may secrete cytokines to compromise the outcome of cell transplantation and reduce cell survival. To overcome this cell loss, using CM spheroids of a few hundred microns may improve the therapeutic efficacy of CM transplantation for MI. The spheroids can be prepared either by hanging drop, derived directly from the PSCs spheroids in 3D, or encapsulated in microcapsules^[187]. Spheroid formation may be desired for multiple aspects such as improving proliferation, differentiation potential, and survival in transplantation^[188]. Drela et al. reported that MSCs grown in spheroids for a short period of time reduce their time of doubling and increase their expression of genes associated with DNA replication^[189]. Yan et al. also reported that spheroid culture of stem cells may increase the expression of anti-apoptotic genes (e.g., Bcl-2), which may activate the hypoxia-inducible factor 1 (HIF-1) signaling pathway by creating a hypoxic environment of MSCs^[190]. Another explanation is that spheroid formation may induce the cell-cell contact-mediated growth inhibition causing cell proliferation and metabolism to be reduced, which may help stem cells to survive in the adverse microenvironment they are in^[191]. Jiang et al. reported that spheroid formation allows stem cells to survive in ambient conditions for over 1 week without losing their immunomodulatory effect and stemness^[192]. Spheroid culture has also been used to isolate cardiac stem cells from the primary fetal murine heart tissue^[51].

Tang et al. reported macroencapsulation of cardiac stem cells using the aforementioned responsive PNIPAAm for treating MI^[193]. The aqueous solution of PNIPAAm at room temperature turns into hydrogel at body temperature, which allows for injectable delivery of the cells. The PNIPAAm hydrogel was found to improve the retention of the injected cells in vivo compared to saline as the carrier for injection. Interestingly, the PNIPAAm hydrogel also help the injected human cardiac stem cells to survive in murine and porcine MI models by avoiding systemic inflammatory and immune reactions. A similar strategy of using thermally responsive hydrogel was employed to improve stem cell therapy of MI in other studies^[194]. A novel inverse scaffold engineering strategy was developed to fabricate a microscale construct of PSC-derived early cardiac cell aggregates for intramyocardial injection with the purpose of enhancing the survival/retention and reduce immune rejection in vivo^[183]. In this work, aggregates of ~130 μm containing ~1500 murine ESCs were formed by culturing the cells in early embryo-like microcapsules with a core-shell structure to minimize spontaneous differentiation of the pluripotent cells during proliferation^[195]. The cells in the aggregates are then pre-differentiated into the early cardiac lineage (NKX2.5+) rather than beating CMs using BMP-4 and bFGF. To mimic the hatching of an early embryo out of its shell (i.e., zona pellucida) and re-encapsulation in trophoblast for implantation into the uterus wall for further differentiation and maturation, the pre-differentiated ESC aggregates are released out of the alginate hydrogel shell of the core-shell microcapsules and re-encapsulated in an alginate-chitosan micromatrix to form cell-biomaterial constructs of ~130 μm for implantation using the inverse scaffold engineering approach^[183]. The constructs are implanted into the peri-infarct region of hearts in both immunocompetent mice with intact immune system and CARD9 knockout mice with deficient macrophage function via intramyocardial injection. Applying the cell aggregates within the micromatrix greatly improves the cell survival and retention to ~50% when compared with the conventional approach of injecting dissociated single cells (less than 10%). Moreover, the pre-differentiated cells are observed to mature in situ into cardiomyocytes with evident striated patterns and cardiac specific markers such as cTnI, α-actinin, and Connexin 43 of the gap junction. Because the pre-differentiated cells were committed to the cardiac lineage without beating and became matured by the biophysical and biochemical cues in situ in the host after transplantation, the resultant mature cardiomyocytes integrated seamlessly with the host tissue with no evidence of arrhythmia. As a result, treatment with the engineered cell constructs significantly improves the heart pumping function and animal survival without abnormal electrophysiological properties. In contrast, without the use of biomaterials, strong immune responses involving both innate (macrophages) and adaptive (T cells) immunity are observed. Teratoma formation is also not observed probably due to the pre-differentiation for cardiac commitment of the PSCs before injection. This novel engineering approach provides a valuable strategy for the development of stem cell therapies with functional biomaterials for MI. These studies demonstrate the great potential of the biomaterials engineering strategy for improving the outcome of stem cell therapy of MI.

5. Remaining Challenges of Using PSCs for MI Therapy

5.1. Hemodynamic Cues for CM Differentiation from PSCs

The heart originates from two endocardial tubes which fuse to from the primitive heart tube that undergoes contraction with blood flow at 21–22 days into embryogenesis for humans^[196]. The tubular heart then loops and separates into different regions that eventually develop into the aortic arch arteries, paired arterial trunks, two atriums, and two ventricles to form a mature heart with the contractile ability to drive the circulatory system^[197]. The human heart starts beating at week 4 when the blood flow is essential for regulating early cardiac morphogenesis and proper heart formation during embryonic development^[198]. Hemodynamic forces (blood pressure and shear stress) exerted on cardiac tissue leads to physical, chemical, and gene regulatory effects during cardiogenesis and maturation due to mechanisms associated with mechanotransduction^[199]. There are studies showing that modification of blood flow with surgical intervention alters the heart development and causes cardiac defects in human congenital heart disease, including a variety of ventricular septal defects (VSDs), spectrum of cardiac defects, and pharyngeal arch arterial malformations^[200]. Midgett et al. also found that altered hemodynamics leads to changes in the myofibrils and reorganization of mitochondria in the cardiac outflow tract of chicken embryos^[201]. Rogers et al. used a biomimetic cardiac tissue model to adapt human iPSCs derived CMs to physiological hemodynamic loads. They found that the CMs adapted to tunable parameters associated with heart function including beating rate, systolic and diastolic pressure, and volume in contraction. There was a significant change of CM viability and remodeling of phenotypes including sarcomere organization, conduction velocity, and action potential^[202].

Despite the aforementioned significant role of hemodynamic loading and blood flow in shaping heart development, current protocols for CM differentiation are based mainly on the events in early stages of forming the cardiogenic crescent and then primitive streak by modulating the signal pathways of BMP, Wnt and TGF-β via the usage of cytokines and chemicals at the cellular level to generate CMs under static culture^[203]. The absence of blood flow diminishes the interplay of important signals associated with blood flow such as Krüppel-like factor 2 (KLF2), VEGF, Notch, endothelin-1 (ET-1), and endothelial nitric oxide synthase (NOS-3), which are associated with the development and differentiation of the pharyngeal endoderm, FHF, SHF and the various components and compartments of the heart^[204]. Obviously, CM generation involves not only modulation at the molecular and cellular levels, but also the hemodynamic cues to promote the differentiation and maturation. There is urgent need to incorporate the hemodynamic cues in cardiac differentiation protocols. This may be achieved by utilizing tissue engineering and microfluidics technologies during the culture process.

5.2. Purification of PSC-derived CMs for MI Therapy

Although PSCs have been successfully differentiated into CMs as aforementioned, the resultant cell population might still contain some undifferentiated hPSCs, fibroblasts, and/or endothelial cells. There remains a need to enrich CMs and ensure a pure population to avoid teratoma formation. The purification methods together with their scalability, need of cell dissociation/detachment, and purity are summarized in Table 5. Generally, the purification methods can be divided into two categories. One is based on CM-specific markers to purify hPSC derived CMs with magnetic-activated cell sorting (MACS) or fluorescence-activated cell sorting (FACS). Dubois et al. reported that using MACS with antibodies against SIRPA and VCAM1 can yield over 95% pure CMs^[205]. FACS separates living cells based on the expression of specific surface markers and fluorescence staining of the markers using antibodies tagged with fluorescence dye. While it is highly efficient in terms of purity and yield, MACS and FACS is costly for scaling up because of the need for antibodies. To eliminate the necessity of antibodies, Hattori et al. discovered that tetramethylrhodamine methyl ester perchlorate, a fluorescence dye that labels mitochondria, could be used to enrich hPSC-derived CMs with 99% purity^[206]. To sustain a high energy quota, CMs contain a larger number of mitochondria than other type of cells. Specifically, mitochondria comprise ~30% of the CM volume. Additionally, co-expressing fluorescence markers with cardiac-specific genes like MYL2 or cTnT may avoid the need to purchase expensive antibodies. Huber et al. established a hESC line which carries an eGFP-expressing reporter sequence downstream of MYL2 and they can obtain the eGFP-hESC-derived CMs of 95% purity with FACS^[207]. However, the aforementioned methods usually require the dissociation of the cells and may cause significant cell loss.

Table 5.

A summary of methods for purification of PSC-derived CMs together with their scalability, need for cell dissociation/detachment, and purity.

Enrichment	Scalability	Cell dissociation	Purity	Reference
Metabolic	Up-scale	No	99%	[133, 242]
Antibiotic	Up-scale	No	99%	[209]
MicroRNA switches	Up-scale	No	95%	[210]
Fluorescent Cell sorting	Low	Yes	95–99%	[243]
Magnetically activated cell sorting	Low	Yes	95%	[244]
Chemical defined differentiation	Up-scale	No	95%	[83]
Enzymatic separation	Low	No	98%	[128]

The other purification method is based on removal of non-CM cells in situ, without the use cell sorting. This approach takes advantage of the difference in gene or protein expression between CMs and other types of cells. Choo et al. developed a negative selection process to remove undifferentiated hPSCs by applying a specific antibody, mAB84, via an oncosis-driven process^[208]. This strategy does not require cell dissociation and can gently remove undifferentiated PSCs to eliminate teratoma formation after transplantation in vivo. Undifferentiated PSCs are known to cause teratoma due to uncontrolled spontaneous differentiation^[158]. Antibiotic resistance genes can be used for purification of hPSC-derived CMs with genetic modification. Klug et al. reported a drug selection system which can stably express aminoglycoside phosphotransferase with the promoter of the cardiac-specific transcriptional regulator MYH6. They could achieve a population of CMs of up to 99% purity, following treatment with G418^[209]. Miki et al. established a microRNA-based switch system to precisely terminate unwanted cell types and gently remove non-CM cells in cardiac patches without harming the structures. When transfected with microRNA 208a, all non-CM cells undergo apoptosis, leaving CMs with more than 95% purity^[210]. The selection is driven by the metabolic difference between CMs and non-CM cells. In addition, since CMs can metabolize lactate while other cells cannot, if the differentiated cells were treated with lactate-containing but glucose-free medium, a population of CMs with 99% purity could be generated within 3 weeks post initiation of differentiation^[133]. This method may be the most promising for use as a simple, defined, and scalable approach to manufacture clinical-grade pure PSC-derived CMs.

Some of the current methods for CM purification still have concerns including inadvertent gene editing, possibly causing misreading and deletions that could further change the phenotype of derived-CMs and increase the heterogeneity of the cell population. By using a precisely inserted locus like AAVS1, a stable cell line can be established after long-term selection^[211]. Then, efficiency of CM differentiation and reproducibility can be increased. Purification can also cause ~10% loss of CMs even with the optimal dosage of agents. Minami et al. optimized CM differentiation using a chemically-induced protocol and a 2D substrate to remove the attached fibroblast cells, followed by enzymatic treatment to create a 3D suspension^[128]. The cells were differentiated into CMs and a 95% pure population could be obtained without sorting. However, this method may generate heterogeneous cardiac spheroids with unintended aggregate formation. Although increasing the efficiency of CM differentiation is the primary goal, there is still a need to develop gentle and efficient methods to purify the CMs from the mixed cell population post differentiation, so that the cells can be used to treat MI in patients without the risk of teratoma formation.

5.3. Maturation of CMs for MI Therapy

Maturity is still a big concern with PSC-derived CMs. Most PSC-derived CMs are immature and representative of the fetal cell type even though they express almost all the cardiac specific markers^[212]. They usually have spontaneous beating which more closely resembles fetal-like CMs with mixed atrial and ventricular phenotypes rather than sole and mature ventricular CMs^[213]. There will be a high chance of arrhythmia in the heart when the immature CMs are transplanted into the myocardium with no cell-cell connection for them to be paced with the host’s heartbeats^[214]. This is undesired for patients. Although the culture of PSC-derived CMs in vitro may generate mature CMs, there are still concerns on their electrical and mechanical mismatch with the host’s heart and arrhythmia post transplantation^[83]. Tohyama et al. tried to use glucose-depleted medium containing abundant lactate to remove non-CM cells via the metabolic difference of CMs versus non-CM cells. They showed an increase in the purity of the PSC-derived CMs, but did not demonstrate any enhanced maturation of the cells^[215].

Arrhythmia may result from the electrical and/or mechanical mis-coupling of transplanted CMs within local tissue. Innervation is also needed between the transplanted CMs and local tissue to help the heart to contract, especially when administering a cardiac sheet with an area greater than a 5×5 centimeter square^[158]. The other concern is the heterogeneity in maturity of PSC-derived CMs. One clear evidence of PSC-CM heterogeneity is the variability of the onset beating time that typically ranges over 7 to 20 days from the initiation of cardiac differentiation. There is an urgent need to synchronize the differentiation stage of PSCs. Although several reports indicate a high efficiency of CM production with EBs and PSCs spheroids of ~250 μm in diameter, it is still difficult to confirm good quality PSCs for starting CM differentiation with a reliable standard^[83]. Jiang et al. generated homogeneous PSC spheroids, via mesh cutting, which could grow to ~250 μm in diameter after 5 days of culture in the mTeSR1 medium. The spheroids remain undifferentiated and could be differentiated into beating cardiac spheroids with 85% efficiency. The onset beating time of the cardiac spheroids ranges over 7 days (from day 7 to day 14), which suggests a heterogeneous population of differentiated CMs in terms of maturity^[83]. It is important to synchronize the onset beating time of PSC-CMs to avoid any possibility of arrhythmia after implantation in vivo. Another concern of CM derivation is regarding the beating rate. Although the beating rate of an adult human is 50–80 beats per minute, the reported beating rate of PSC-derived CMs is usually ~40 beats per minute or less, suggesting an immature state of the CMs. Since the beating rate of a human fetal heart is typically 120–160 times per minute, it is inadequate that there are so few reports showing that human PSC-derived CMs reach a beating rate more than 100 times per minute. Further optimization is necessary to produce PSC-CMs with a fast beating rate for pediatric patients. The same dilemma also exists with PSC-derived neural cells. It is difficult to obtain fully differentiated neurons such as GABA (gamma aminobutyric acid) or dopaminergic neurons. Interestingly, PSC-derived MSCs are also “immature” with better differentiation potential than MSCs derived from adult tissue like bone marrow and fat, particularly with regards to the adipocyte differentiation potential^[216].

The mechanical stimuli produced by heart beating may also help the maturation of CMs^[217]. The anisotropy in structure and modulus of the surrounding ECM in the myocardium dynamically changes with gender, age, and health^[218]. Mechano-signals transduced via the intercalated disc, gap junctions, titin-associated proteins, focal adhesion associated proteins, and stretch-activated ion channels profoundly affect the physiology of CMs^[219]. The structural anisotropy has been shown to influence the CMs maturation and contractile function^[155]. There are many studies using biomaterials and engineering tools like bioprinting, electrospinning, and soft lithography to recapitulate the micro-topographic aspects of the CMs and myocardial ECM for CM maturation^[220]. Furthermore, repeated mechanical activation of the CMs using dynamic micro-topographies could mimic myocardial development or pathological progression of cardiac diseases. CMs are quite sensitive to dynamic and, to a lesser extent, static mechanical stimuli such as change in modulus, viscosity, shear stress, and passive loading force. These stimuli have been demonstrated to influence autocrine or paracrine signaling, while also affecting the cytoskeletal structure in CMs^{[218, 221]}.

5.4. Cell Retention and Survival of Using PSC-derived CMs for MI Therapy

It is difficult to achieve high cell retention and survival of stem cells in vivo, especially in soft tissue like the myocardium. Many reports indicate that adult stem cells derived from somatic tissues (e.g. bone marrow) or PSCs improve the cardiac contractile function and physiology after administration into the MI lesion. However, few of these reports explain the mechanisms for the improvement^[222]. Transplanted hematopoietic stem cells usually have a high survival rate since they are infused into the vascular system where the environment is full of nutrients and oxygen. Stem cells transplanted into soft tissue are usually exposed to reduced levels of oxygen and nutrients. The cells in a normal organ are usually within 200–300 μm away from the nearest blood vessel with sufficient nutrients and oxygen. When several million cells are transplanted, they are usually injected into a space of millimeters in size. If the carrier for cell injection is a solution (e.g., saline), the cells could easily be washed away from the injection site by the tissue fluid as a result of the heart contraction. If the carrier is a gel (e.g., Matrigel) to hold the cells in place after injection and a blood vessel does not come into contact with the cells for more than 2 days, the transplanted cells easily die as a result of low nutrients/oxygen and an accumulation of toxic wastes^{[157, 158, 178]}. The infarcted myocardium has a harsh inflammatory environment with immune cells such as neutrophils/macrophages that may engulf the implanted CMs. The injected cells could be easily repelled out of the myocardium to the pericardium chamber due to strong pressure of the heart contraction. In addition, Matrigel has an undefined composition which is a major concern for biosafety^[223]. More defined materials like alginate, chitosan, and nanomaterials have been used to increase cell retention in the myocardium and protect the cells from immune cells^{[183, 223, 224]}. Ban et al. developed an injectable nanomatrix gel which combines amphiphilic peptide and a cell adhesive ligand Arg-Gly-Asp-Ser (PA-RGDS). They evaluated the therapeutic potential and long-term effect of the suspension of mouse embryonic stem cells (mESCs)-derived CMs for engraftment in a MI rat model^[169]. Their results showed retention of engrafted CMs for up to 3 months and improved function of the heart post administration. Shiba et al. also reported up to 3 months of survival of human ESCs-derived CMs and improved heart function in nonhuman primates (cynomolgus) with MI^[225]. In their study, they used the HLA matched CMs with the host to avoid immune rejection. A mixture with Matrigel was used to reduce the cell death post myocardium injection. Their cells were genetically engineered with a calcium-sensitive fluorescent protein (G-CaMP7.09) to express fluorescence on calcium-driven activities. Surprisingly, this study indicates the iPSC-CM grafts not only re-muscularize the infarct tissue but also exhibit fluorescence transients that occur in a synchronous pattern with the host echocardiogram, suggesting host–graft electromechanical integration in the primate^[225]. Although the use of biomaterials can substantially increase the cell retention of CMs in the myocardium, only 16% of injected cells actually occupy the infarct area and the majority of cells are lost in the process^[226]. Those cells might be pushed out of the myocardium by the heart contraction and/or eliminated by the immune cells in the infarcted tissue.

5.5. Tumorigenesis Associated with PSC-derived CMs for MI Therapy

The concern over tumorigenesis of either PSCs or downstream differentiated cells undermines the safety of cell transplantation in vivo. This is especially concerning for iPSCs made using c-Myc since it is an oncogene^[73]. Early reports show that transplanted PSCs form a teratoma^[227]. There is an urgent need to develop a reliable method to remove the undifferentiated cells after cardiac differentiation of PSCs. Hemmi et al. show that differentiated CMs have a different metabolic pattern compared to PSCs and they use the glucose-depleted and lactate-enriched medium culture to eliminate the undifferentiated PSCs and increase the purity of the differentiated CMs^[228]. Transplantation survival is also a limitation of PSC application in vivo. PSCs are cultivated at 37 °C with a continuous nutrient supply like glucose while also having their byproduct ammonium removed by medium changes. However, the local in vivo environment at a distance greater than 500 μm from a blood vessel has less than 5% oxygen and is short of nutrients compared to culture medium. The difference between the in vivo and in vitro conditions can produce significant stress on the transplanted cells in the first 24 hours, leading to apoptosis. Moreover, angiogenesis in the transplant is important for the transplant’s survival and function. The cells are co-transplanted with MSCs to promote blood vessel formation via MSCs secretion of VEGF^[13]. Shiba et al. also reported encouraging results of CMs transplanted in a cynomolgus MI model. There was no teratoma formation in 5 infarcted monkeys that received 400 million MHC-matched iPSC-CMs each. Their latter results also corroborated earlier work with relevant PSC-CM populations, which emphasize the use of sorted CMs with high-purity before transplantation to reduce teratoma formation^[225]. In addition, tumorigenesis may be avoided by adding inducible suicide cassette PGK-Puro-ΔTK-PA in the genome to induce cell apoptosis when the cells become overly proliferative in vivo^[211]. Wang et al. reported a similar strategy which incorporated a chemically induced iCas9 system in the genome to remove the unwanted cells after transplantation of hESC-derived MSCs in a wound healing mouse model^[229]. There is also concern on whether this strategy can be rapidly translated into clinic, because genome editing may cause unwanted genetic variations including DNA sequence loss and epigenetic changes to the biological phenotypes of the resultant cells. These concerns are a result of the disadvantages of current molecular avenue like the off-target effect of CRISPR-Cas9^[230].

5.6. Immune Rejection of PSC-derived Cells for MI Therapy

Though PSCs have many advantages compared to adult stem cells, one major limitation for PSC-derived somatic cells is the chance for immune rejection in vivo^[231]. Even when derivatives of autologous iPSCs are transplanted into the donor, immune rejection could still occur, possibly due to the contact of xeno-genetic components in culture^[232]. Despite this, there are also reports indicating that PSC-derived MSCs have less immune rejection than BM-MSCs as a result of low expression of IL-6^[81]. They suggest that PSC-derived MSCs may be directly applied for wound healing and osteoarthritis treatment^[81],[158]. Current transplantation work using PSC-derived CMs involves mainly human PSC derivatives in animal models or the allo-transplantation of mouse PSC derivatives in mice. These investigations do not completely recapitulate host immune responses in human patients^[233]. Clinical results of organ transplantation may not represent the outcomes of administering PSC-CMs alone in humans. In the latter case, transplanted cell populations should be devoid of antigen-presenting cells present in whole hearts. Interestingly, there is a report suggesting that PSC-derived CMs typically express low levels of MHC I antigens and even fewer levels of MHC II antigens than other derived cells^[234]. Shiba et al. reported the successful transplantation of allogenic PSC-derived CMs in a non-human primate heart with MI based on the MHC-matching method^[225]. They found that transplanted iPSC-CMs were rejected in the MHC-mismatched recipients within 1 month whereas the sibling control cells survived and integrated into the heart of MHC-matched hosts for up to 3 months. Histological analysis shows no clear immune rejection with the administration of tacrolimus and methylprednisolone, which are traditional immunosuppressant drugs. Though the study was performed on a small number of animals, their results demonstrate the promise of the MHC match system for use in preclinical trials. To overcome immune rejection after transplantation, gene editing has also emerged as a powerful tool to knock out immune receptors such as B2M, CIITA, and MHCII^[211]. It is also suggested that administration of MSCs along with PSC-CMs may confer an immune suppressive effect due to the MSCs that can decrease the T and B cell activation. Beside cell engineering, Zhao et al. also reported that biomaterials including chitosan and alginate can be used to form a protective shield to avoid immune cell attack and increase survival of transplanted CMs in a mouse MI model^[183].

6. Conclusions

Various stem cells have been used for treating MI. However, somatic tissue-derived stem cells like CPCs and CFs for MI therapy are still controversial due to poor reproducibility. MSCs are heterogeneous and difficult to differentiate into functional CMs, although the cytokines released by them could be beneficial for MI therapy. PSCs are the most promising for stem cell-based therapy of MI because they not only can directly and efficiently differentiate into functional CMs, but also have great potential for scale-up compared to somatic tissue-derived or adult stem cells. Methodologies for CM differentiation from PSCs under 2D culture have been well established and can produce functional CMs via the modulation of various signaling pathways including BMP4, Activin A, and Wnt. In spite of this, large-scale production of CMs from 2D culture is difficult, and an efficient and consistent differentiation of PSCs into CMs on a large scale needs to be developed. Moreover, the purification and maturation of PSC-derived CMs need to be further improved to enable their clinical use. Various biomaterials have been developed to mimic the myocardial environment for improving the differentiation and maturation of PSC-derived CMs in vitro and in vivo. In addition to the future propagation of PSCs, combining automated bioreactors with biomaterials may help to more precisely control the biophysical cues and ensure quality control to meet the demands of PSC-derived CMs with large quantity and high quality for clinical applications. Biomaterials have also been developed to enhance CM retention, survival, and escape from the host immune system in vivo. There is a current need to develop novel biomaterials which may aid in CM integration with the host tissue after transplantation via enhancing the revascularization (and possibly innervation) by providing the appropriate physiological and chemical cues. The combination of PSC-derived CMs and biomaterials has greatly advanced stem cell therapy for MI treatment. Biomaterials have been developed to attain biomimetic structures with controlled ECM organization and physical properties, which together with an appropriate combination of cells may significantly increase the survival, maturation, and integration with the host heart of the PSC-derived cells. Engineering stem cell therapy with biomaterials holds great promise to become an effective and safe treatment for MI in the clinic.