Derivation of Induced Pluripotent Stem Cells for Human Disease Modeling

Abstract

The successful derivation of human induced pluripotent stem cells (hiPSCs) by de-differentiation of somatic cells offers significant potential to overcome obstacles in the field of cardiovascular disease. hiPSC derivatives offer incredible potential for new disease models and regenerative medicine therapies. However, many questions remain regarding the optimal starting materials and methods to enable safe, efficient derivation of hiPSCs suitable for clinical applications. Initial reprogramming experiments were carried out using lentiviral or retroviral gene delivery methods. More recently, various non-viral methods that avoid permanent and random transgene insertion have emerged as alternatives. These include transient DNA transfection approaches using transposons or minicircle plasmids, protein transduction approaches, and RNA transfection approaches. In addition, several small molecules have been found to significantly augment iPSC derivation efficiency, allowing the use of a fewer number of genes during pluripotency induction. Here, we review these various methods for the derivation of hiPSCs, focusing on their ultimate clinical applicability, with an emphasis on their potential for use as cardiovascular therapies and disease modeling platforms.

BACKGROUND

Human embryonic stem cells (hESCs) initially generated much enthusiasm because of their self-renewing and pluripotent properties¹. hESCs theoretically can generate an unlimited number of any somatic cell, given the proper culture conditions. Directed differentiation can be promoted by varying the concentrations of various growth factors². Recently, the Food & Drug Administration (FDA) approved two hESC-based clinical trials for treatment of acute spinal cord injury (Geron; http://www.geron.com) and Stargardt’s macular dystrophy (Advanced Cell Technology; http://www.advancedcell.com).

Although hESCs have been established as a renewable source of definitive cardiac tissue cells, no clinical cardiovascular application of hESCs has yet been realized. hESC-derived cardiomyocytes (hESC-CMs) can be exploited as a human in vitro modeling system, in which the study of cardiogenesis, myocardial-related pathology, drug targets, drug screening, and tissue engineering can be easily conducted³, paving the way for their future application in vivo as bona fide transplantation cells. Of note, the hESC-CMs derived to date are more akin to fetal/neonatal rather than adult cardiomyocytes^{4, 5}, and enhancing their in vitro maturation using electric and mechanical cues is an ongoing area of investigation⁶. Despite this, the advantages of using the human-based modeling system provided by hESCs remain appealing, given that non-human transgenic models may not accurately reflect all aspects of the human disease phenotype. However, in the United States, hESC research funding can be subject to the vagaries of the courts and Congress, as evidenced by the recently lifted ban on federal funding for hESC research⁷. In addition, the issues of potential immunological rejection^{8, 9} and tumorigenicity¹⁰ continue to be challenges to clinical cell transplantation approaches. The recent discovery of human induced pluripotent stem cells (hiPSCs) has mitigated some of these concerns, as hiPSCs can be generated autologously and do not require the destruction of ex utero embryos. hiPSCs are comparable to hESCs in morphology, feeder dependency, surface markers, gene expression, promoter methylation status, telomerase activities, in vitro differentiation potential, and in vivo teratoma forming capacity¹¹. Hence, hiPSCs have great potential to replace hESCs as disease models or prospective treatment options.

Although the results of these initial comparisons are promising, the functional equivalence and safety of hESCs and hiPSCs remain contested¹². For instance, hESCs display chromosomal instability with extended in vitro culture¹³, while hiPSC genomes appear to remain intact¹⁴. Also, reactivation of reprogramming transgenes in iPSCs may enable malignant transformation of the cells post-transplantation¹⁵. Most recently, variations in gene expression profiles between different hESC and hiPSC lines have been investigated, with mounting evidence indicating that cell-line specific differences in epigenetic landscape underlie differences in differentiation propensity^{12, 16–21}. In light of these variations in performance, it is becoming clear that choice of reprogramming technique plays a crucial role in the quality and ultimate utility of the resulting hiPSC lines.

Since the ability to directly reprogram somatic cells to pluripotency has been well-established in adult mouse fibroblasts^22–27, attempts to generate hiPSCs have become more pronounced, and the logistics involved in appropriately generating and utilizing hiPSCs is of significant current interest. Several techniques have evolved for derivation of hiPSCs suitable for cellular regenerative medicine and disease modeling. This review discusses the methods, benefits, and drawbacks of several techniques, with an emphasis on their potential for future cardiovascular clinical applications.

INTEGRATING METHODS

Efforts to reprogram human somatic differentiated cell types to a state that resembles hESCs began with the pioneering work of Takahashi and Yamanaka¹¹. Their methods included retroviral integration of four vital reprogramming factors - OCT3/4, SOX2, KLF4, and c-MYC -into adult human dermal fibroblasts. These four transcription factors would later become known as the “Yamanaka factors”, and their roles in reprogramming are now known to be significant²⁸ but not collectively necessary^29–36. Often the omission of one or more of these reprogramming genes was contingent upon the endogenous network of the donor cell-type. For example, one study found that hiPSC derivation from keratinocytes required only ten days, while neonatal skin fibroblasts required ~30 days³⁷. It was postulated that perhaps the keratinocytes’ higher endogenous expression levels of c-MYC and KLF4 predispose them to quicker reprogramming³⁸. Starting cell-type is thus an important consideration in the derivation process, and a topic that is more thoroughly discussed elsewhere³⁹. Two other transcription factors, namely NANOG and LIN28, were initially shown to be able to substitute for c-MYC and KLF4, although a number of other different factor combinations have been subsequently demonstrated^{26, 29, 36, 40, 41}. In any event, several cocktails comprising any number of these six reprogramming factors, and in some cases, additional supplements such as small molecules and enzymes, have been shown to be capable of reprogramming cells to pluripotency.

A chief aim of clinical hiPSC researchers is to achieve a high efficiency of derivation of hiPSCs, as current yields of bona fide hiPSCs can be as low as 0.001-0.1% of the starting cell population^{42, 43}. Even in so-called “secondary” reprogramming systems, in which all of the somatic cells homogeneously express the reprogramming factors, the efficiency of inducing pluripotency remains low at 1–5%. Two mutually non-exclusive models have been proposed to explain the apparent resistance to pluripotency induction, termed the “elite” and “stochastic” models⁴⁴. The elite model proposes that only a small percentage of somatic cells, presumably resident tissue progenitor cells, are amenable to reprogramming. In support of this notion is evidence that hematopoietic stem cells undergo more efficient reprogramming than their differentiated progeny⁴⁵. However, reports of successful reprogramming of terminally differentiated cells such B lymphocytes⁴⁶ and pancreatic β islets⁴⁷ favor a stochastic model of reprogramming, in which successive cell divisions allow rare cells to acquire the stochastic changes necessary for conversion to full pluripotency⁴⁸. Perhaps these seemingly contradictory hypotheses can be reconciled by a model in which adult stem/progenitor cells require fewer stochastic changes to undergo reprogramming than more differentiated cells. Further investigation of the reprogramming process using single-cell resolution imaging and other techniques will undoubtedly help yield further insight into these reprogramming roadblocks.

Clearly, the choice of gene delivery vector can change reprogramming efficiency by directly affecting the degree of expression of the reprogramming genes. Retroviral/lentiviral infection provides the benefit of high transgene expression levels in primary cells as compared to nonviral methods of reprogramming. However, retroviral/lentiviral methods for hiPSC generation have come under scrutiny due to concerns regarding their ultimate clinical safety. In particular, the random integration of transgenes into the human genome can potentially cause insertional mutagenesis, leading to malignant transformation of a clonal cell population and disastrous consequences⁴⁹. Second, leaky expression due to ineffective silencing of the transgenes may interfere with the physiological expression of the factors endogenously present within the cell, thereby potentially restricting differentiation propensity⁵⁰. This residual expression may hamper the validity of in vitro hiPSC uses, such as in disease modeling, drug screening, and toxicology tests. Third, reactivation of OCT4 and c-MYC has been shown to promote tumor formation in chimeric mice^{26, 51, 52}, prompting legitimate concern over post-transplantation tumorigenic risk if such methods were employed in human patients.

In one of the initial forays into generating safer hiPSCs, Maherali et al. created a doxycycline-inducible lentiviral system, attempting to maintain the silencing of transcription factors and thus reduce the tumorigenicity of the cells post-differentiation³⁷. Although this system is a step towards safer hiPSCs, the leakiness of the doxycycline-inducibile promoter and the permanent incorporation of oncogenes into the host genome still warrant concern.

Viral integration followed by excision: Cre-loxP

Soldner et al. generated viable hiPSCs free of exogenous reprogramming factors using doxycycline-inducible lentiviral vectors that integrated into the host genome, but were subsequently excised by Cre recombinase⁵³. Fibroblasts were obtained via skin biopsies from five patients exhibiting sporadic Parkinson’s Disease, and transduced using three or all of the Yamanaka factors. The reported reprogramming efficiency was 0.005% after transduction with the three-factor combination and 0.01% with the four-factor combination. Furthermore, the three-factor transduced cells required twelve days of DOX-exposure, as opposed to the four-factor cells, which required only eight days. Despite its lower reprogramming efficiency and temporal requirements, the three-factor may be preferred over the four-factor combination since the transduced cells are not overgrown by granulate colonies^{11, 31}. Southern blot analysis demonstrated successful excision of the transgenes, and the resulting hiPSC lines maintained pluripotency independent of residual exogenous transcription factor expression.

Since the isolated hiPSC lines are patient-specific, they provide a system for investigating the proposed molecular and cellular mechanisms of the disease. Soldner et al. demonstrated successful derivation of dopaminergic neurons from Parkinson’s Disease patients’ cells, indicating that the underlying age or disease of the donor most likely does not affect the ability of their cells to produce hiPSC-derived replacements ex vivo⁵³. Interestingly, in this study, factor-free hiPSCs were found to be more closely related to embryo-derived hESCs than provirus-carrying parental hiPSCs based on gene expression analysis. Hence, basal expression of proviruses carried in conventional hiPSCs can affect the molecular characteristics of the hiPSCs. Although transgenes are expected to be completely silenced in bona fide hiPSC lines, residual sequences and chromosomal disruptions during and after viral integration may still result in harmful alterations that could pose clinical risks. As previously mentioned, reactivation of reprogramming transgenes post-transplant can result in malignant transformation of the cells and formation of a tumor¹⁵. Such unpredictable effects of incomplete transgene silencing on downstream hiPSC phenotype further highlight the need for transgene-free hiPSC derivation methods.

Nonviral integration followed by excision: piggyBac Transposition

Although Cre recombinase-driven excision utilizes a highly efficient and widely-used system, small residual vector backbone sequences remain at the site of integration and may engender unpredictable downstream effects. Woltjen et al. and Kaji et al. demonstrated successful reprogramming of human embryonic fibroblasts using doxcycline-inducible reprogramming factors that were delivered as plasmids, stably integrated into the host genome, and subsequently excised using piggyBac transposition^{54, 55}. Woltjen et al. noted that successful transposon-based nonviral reprogramming has several advantages over traditional lentiviral integration-based reprogramming: (1) improved accessibility of reprogramming techniques through the use of plasmid DNA preparations and commercial transfection products, thereby eliminating the need for specialized biohazard containment facilities; (2) increased variety of reprogrammable donor cell-types because susceptibility to viral infection is no longer an important consideration; (3) feasibility of xeno-free production of hiPSCs; and (4) most importantly, near complete elimination of the expression of reprogramming factors after establishment of hiPSC lines by piggyBac transposase-mediated excision. However, successful excision of the reprogramming cassette was only achieved in approximately 2% of the bona fide hiPSCs exposed to piggyBac transposase, limiting the amount of vector-free hiPSCs that could potentially be produced. Of note, Mali et al. demonstrated the use of butyrate to enhance reprogramming efficiency 15- to 51-fold when used in conjunction with piggyBac transposase-driven integration and excision of the Yamanaka reprogramming factors⁵⁶. Genome-wide analysis of the effects of butyrate exposure at days 6–12 demonstrated significant changes in H3 acetylation and promoter methylation status in a variety of pluripotency-related genes, including DPPA2.

NONVIRAL NONINTEGRATING METHODS

Despite the aforementioned progress towards safe and efficient reprogramming, the ideal method for hiPSC derivation would avoid temporary or permanent genomic modification. However, nonviral, nonintegrating reprogramming techniques typically have significantly lower reprogramming efficiencies than the aforementioned retroviral and lentiviral methods. Nonviral, nonintegrating methods for hiPSC generation include episomal plasmid DNA delivery, minicircle DNA vector delivery, protein delivery, and synthetic RNA delivery. DNA- and RNA-based methods accomplish hiPSC induction by the transient expression of reprogramming factors, often through repetitive transfection protocols.

Episomal Plasmid

Yu et al. derived hiPSCs from human foreskin fibroblasts using non-integrating episomal plasmids vectors⁵⁷. Repeated transient transfection of three plasmids expressing seven reprogramming factors resulted in hiPSCs that were completely transgene-free (as confirmed by RT-PCR) but maintained the proliferative and developmental potential of hESCs. The seven factors include OCT4, SOX2, c-MYC, KLF4, NANOG, LIN28, SV40 large T antigen (SVLT), and Epstein-Barr nuclear antigen-1 (EBNA1). Of note, stable extrachromosomal replication of these plasmid vectors is enabled by the presence of the cis-acting oriP sequence and the trans-acting EBNA1 protein^58–60. Interestingly, co-expression of the six reprogramming factors (OCT4, SOX2, NANOG, LIN28, KLF4, and c-MYC) in initial experiments resulted in substantial cell toxicity and produced no hiPSC colonies. Only the inclusion of the SVLT and modification of the reprogramming factor ratio allowed for isolation of bona fide hiPSCs⁵⁷. These results emphasize the importance of balanced absolute expression levels during the reprogramming process. Subsequently, the progressive loss of the episomal vectors enables isolation of hiPSCs devoid of vector or transgene sequences. However, reprogramming efficiency remained low at approximately 3–6 colonies per 10⁶ input cells.

Minicircle Vectors

Jia et al. constructed a minicircle vector for reprogramming consisting of a cassette of the reprogramming factors OCT4, SOX2, LIN28, and NANOG, and a GFP reporter gene, each separated by sequences encoding the 2A ribosomal slippage site⁶¹. Though minicircles are also supercoiled derivatives of plasmids, they are unique in that they primarily consist of eukaryotic expression cassettes, which lack both a bacterial origin of replication and antibiotic resistance gene. These vectors are preferred to standard plasmids since they have shown comparatively enhanced and more persistent transgene expression both in vivo and in vitro, perhaps on account of their smaller sizes and minimal expression cassettes^{62, 63}. Indeed, Jia et al. reported that a plasmid vector-based protocol failed to generate hiPSCs in their system⁶¹. In their minicircle protocol, the vectors were introduced into human adipose stromal cells three times: first via nucleofection, a proprietary electroporation system of Lonza®, then twice via Lipofectamine®. On days 14–16, colonies that were morphologically similar to hESC colonies were produced from human adipose stem cells and neonatal fibroblasts, whose reprogramming efficiency was reported at ~0.005% and ~0.0005%, respectively⁶⁴. Southern blot analysis confirmed the absence of genomic integration of the minicircle transgene in the subclones.

Protein Delivery

Direct delivery of reprogramming factor proteins into the cell is one way to entirely eliminate risks associated with viruses, genome manipulation, and DNA transfection. Kim et al. reported successful generation of hiPSCs from human newborn fibroblasts by direct delivery of the Yamanaka factor proteins fused with a cell-penetrating polyarginine peptide⁶⁵. After six protein treatment cycles over 8 weeks, five alkaline phosphatase positive hiPSC-like colonies were derived from a starting cell population of 5 × 10⁵ cells. Expansion of these hiPSC lines allowed confirmation of pluripotency by in vitro RT-PCR and in vivo teratotoma formation assays. Despite the hampered reprogramming process created by the need for repeated treatments and relative subdued reprogramming efficiency (~0.001%), reprogramming via protein transduction creates an attractive alternative by allowing generation of hiPSCs free of vectors or limitations caused by viral delivery. In conjunction with other methods such as use of purified recombinant mammalian proteins or concomitant use of small molecules, one could potentially further improve the reprogramming efficiency.

RNA Delivery

Nonintegrating DNA- and protein-based reprogramming strategies still have certain disadvantages. Even episomal DNA vectors used for hiPSC derivation entail a small risk of unintended genetic integration. Protein-based reprogramming requires relatively challenging production and purification of recombinant proteins. Recently, Warren et al. demonstrated the successful production of hiPSCs via repeated transfection of modified synthetic mRNAs⁶⁶. These modified mRNAs potentially bypass any innate immune response to foreign DNA elements. Tra-1–60⁺ hiPSCs were derived from BJ fibroblasts by repeated transfection of a combination of four modified synthetic mRNAs encoding the Yamanaka factors at an efficiency of 1.4%. Retrovirally-mediated reprogramming experiments conducted in parallel achieved an efficiency of 0.04%. Therefore, RNA-mediated reprogramming represented a 36-fold improvement in reprogramming efficiency. The kinetics of hiPSC generation using modified mRNAs was also noted to be accelerated approximately two-fold. Addition of LIN28 to the four-factor cocktail and subjection of the cells to low oxygen culture conditions (5% O₂) generated a striking reprogramming efficiency of 4.4%. The significantly enhanced reprogramming efficiency of Warren et al. may be attributable to several unique experimental advantages. First and foremost, transfected mRNA is translated into functional protein within several hours, whereas several days may elapse after lentiviral transduction before expression of functional protein. The ability to achieve stable intracellular stoichiometric concentrations of mRNA using transfection is presumably preferable to the somewhat variable expression patterns of transfected or transduced DNA constructs. Also, the immunogenicity of the foreign RNA was reduced by substitution of 5-methylcytidine and pseudouridine bases for cytidine and uridine, respectively, as well as media supplementation with the Vaccinia virus decoy protein B18R to allay any interferon-mediated immune response. These changes allowed for repetitive daily transfections for a period of ~17 days without significant cytotoxicity. Such results represent a significant advance in our ability to derive transgene-free hiPSCs. In agreement with previous reports, the authors observed that transgene-free hiPSCs derived by RNA transfection more faithfully recapitulated the global gene expression profile of hESCs than retrovirally-derived hiPSCs.

SMALL MOLECULES AUGMENT REPROGRAMMING EFFICIENCY

Small molecules may improve the reprogramming efficiency or completely replace one or more reprogramming factors during the hiPSC derivation process. Proponents of small molecules also assert that they are more timely in reaching their target, easier to control via concentration variations, simpler to utilize, and cheaper than the aforementioned reprogramming agents⁶⁷. Chemical approaches to screening combinatorial small molecule libraries have successfully identified several molecules that can augment pluripotency induction. O’Malley et al. separated these small molecules into two groups: 1) affectors of chromatin modification and 2) affectors of cell signaling pathways⁶⁸. This general classification system helps to organize and differentiate the targets of small molecules, allowing prediction of their effects and potential synergism. Detailed and comprehensive knowledge of the involved pathways is especially important since the effects of cell exposure to drugs may entail non-specific and wide-ranging changes that unpredictably affect the safety and phenotype of the resulting cells. In addition, an understanding of the effects of small molecules on intracellular signaling mechanisms, developmental pathways, and cell fate may provide important insight into the molecular mechanisms of reprogramming.

Valproic Acid

Valproic acid (VPA) is a small molecule histone deacetylase inhibitor that is already FDA-approved to treat epilepsy and bipolar disorder^{69, 70}. VPA has been used to successfully reprogram neonatal foreskin fibroblasts when used in conjunction with OCT4 and SOX2. Thus, it can effectively replace the oncogenic c-MYC and KLF4 factors, albeit at the cost of a significantly reduced reprogramming efficiency (<0.005%)³². When VPA was used with OCT4, SOX2, and KLF4, a reprogramming efficiency of 1% was reported. Notably, reprogramming was accomplished in dermal fibroblasts, a differentiated cell type that does not endogenously express any of the reprogramming factors. Using VPA in conjunction with a cell type that endogenously expresses high levels of c-MYC or KLF4, such as keratinocytes or adipose stromal cells, might therefore significantly improve reprogramming efficiency^{37, 39}.

SB431542, PD0325901, and Thiazovivin

Lin et al. tested a combination of three molecules in an effort to precipitate speedier production of hiPSCs (created by retroviral transduction of fibrobasts (CRL2097 or BJ) with cDNAs encoding the Yamanaka factors)⁷¹. They concluded that dual inhibition of the TGF-β and MEK pathways in a dose-dependent, temporal manner may guide and perhaps accelerate the kinetics of partially reprogrammed colonies to a fully pluripotent state. This is based on a ~100-fold improvement in reprogramming efficiency that occurred when TGF-β inhibitor SB431542 and MEK inhibitor PD0325901 were added to the cultures, promoting improved induction at earlier stages than expected, and a decrease in the number of granulate non-iPSC colonies produced. These results highlight the importance of signal transduction pathways in the reprogramming process.

For example, TGF-β signaling can be active in tumorigenic suppression and immune response and cell migration regulation, as members of this family primarily promote cellular senescence, differentiation, and apoptosis⁴². TGF-β antagonists likely benefit reprogramming in other ways as well, including inhibition of the aforementioned cellular processes and promotion of the mesenchymal-to-epithelial transition by upregulating genes such as E-cadherin⁷², a regulator of NANOG expression⁷³.

An additional impediment to high reprogramming efficiency is the poor survival of hiPSCs after dissociation into single cells by trypsinization. Addition of thiazovivin to hiPSC reprogramming culture media has been found to improve cell survival during protocols involving splitting via trypsinization⁷¹. Remarkably, by 30 days post-transduction, ~900 hESC-like colonies were produced from 10,000 seeded cells after a single 1:4 split on day 14 by using thiazovivin in conjunction with SB431542 and PD0325901. Independent of its demonstrated ability to promote single-cell survival, thiazovivin appears able to enhance reprogramming efficiency by approximately two-fold when used in conjunction with SB431542 and PD0325901.

CHIR99021 and Parnate

CHIR99021 and Parnate are two other small molecules that have been used to successfully optimize reprogramming efficiency. Li et al. reported successful reprogramming of neonatal epidermal keratinocytes with only two transcription factors (OCT4 and KLF4) in conjunction with CHIR99021 and Parnate³⁰. On average, a reprogramming efficiency of 0.002% was achieved. CHIR99021 is a GSK-3 inhibitor that activates the Wnt signaling pathway, and Wnt3a-conditioned media is known to promote reprogramming in mouse embryonic fibroblasts⁷⁴. Parnate is a monoamine oxidase inhibitor used as a second-line treatment for depression. It is also classified as an epigenetic modifier due to its inhibition of lysine-specific demethylase 1⁷⁵. It is thereby thought to enhance reprogramming efficiency by inhibiting H3K4 demethylation. Interestingly, the use of Parnate and CHIR99021 in addition to SB431542 and PD0325901 did not result in any significant improvement in reprogramming efficiency, as only 5–10 hiPSC colonies could be identified from 10⁵ transduced cells. The pathways targeted by these additional molecules and their lack of synergy with other known small molecule enhancers of reprogramming remain under investigation.

Vitamin C

Of particular interest is the effect of vitamin C on the reprogramming efficiency of hiPSCs. Esteban et al. found that vitamin C and other antioxidants had no effect on reprogramming efficiency when added to cell culture media supplemented with knockout serum replacement⁷⁶. However, when added to cell culture media supplemented with Dubelco’s modified fetal bovine serum (DFBS), vitamin C used in conjunction with VPA produced roughly three times as many alkaline phosphatase positive colonies as VPA alone (7.06% vs 2.10% efficiency). Improved reprogramming efficiency (maximum efficiency of 0.40%) using vitamin C was also demonstrated by Cai et al using umbilical cord-derived cells⁷⁷. These cells were transduced with retroviruses using the Yamanaka factors and a chemical concoction of vitamin C, DFBS-based medium, and VPA. Given the known role of cellular senescence in impeding reprogramming, it is perhaps expected that vitamin C, a natural antioxidant, would enhance cellular reprogramming by reducing intracellular levels of reactive oxygen species⁷⁸. However, alternative mechanisms for vitamin C’s effects could include promotion of epigenetic modifications, as vitamin C is a cofactor for many significant enzymes.

Butyrate, A-83-01, and PS48

Since keratinocytes express KLF4 and c-MYC endogenously, it was believed that full reprogramming was possible with the sole transduction of OCT4 in the presence of small molecules³⁰. Only recently was this hypothesis definitively confirmed in a study by Zhu and colleagues, who used a mixture of sodium butyrate (NaB, a histone deacetylase inhibitor), PS48 (an inhibitor of phosphoinositide-dependent kinase-1), A-83-01 (TGFβ kinase/activin receptor-like kinase (ALK 5) inhibitor), PD0325901, CHIR99021, and Parnate to reprogram not only adult keratinocytes, but umbilical vein endothelial and amniotic fluid derived cells at an average of 0.004% reprogramming efficiency⁷⁹. Interestingly, butyrate is believed to assist the reprogramming process epigenetically in mouse embryonic fibroblasts by modulating the flexibility of chromatin structure and repressing cancerous cell growth, but only in the presence of c-MYC⁸⁰. PS48 was found to upregulate glycolytic gene expression⁷⁹, promoting a switch from mitochondrial oxidation, a metabolism employed by somatic cells when they are no longer proliferating⁸¹. The respective pathways affected by these molecules are critical in the reprogramming process, underscoring the need for further investigation into the mechanisms of de-differentiation. Ideally, a chemical concoction without transcription factors would be sufficient to reprogram somatic cells to pluripotency, and progress towards this goal is readily apparent in recent work. Additional research into reprogramming enhancement using these and other small molecules is imperative given the possibility that off-target effects of these small molecules may be detrimental to cell phenotype.

CARDIOVASCULAR APPLICATIONS OF iPSC-DERIVED CELLS

Once derived, hiPSCs can serve as an inexhaustible source of more differentiated cardiovascular cells, including the cardiomyocyte, endothelial cell, and smooth muscle lineages. The adult heart has limited regenerative capacity, and pluripotent stem cell-derived lineages are ideal candidates for replacement therapies. Transplantation treatment of myocardial infarction may demand replacement of up to 1 billion damaged cardiomyocytes⁸². Aside from the potential benefits that hiPSC-derived cardiomyocytes (hiPSC-CMs) can provide following myocardial infarction, endothelial cells generated from hiPSCs may help repair heart valves, vessels, and ischemic tissue⁸³, perhaps by preventing cell death and providing vascular support in grafts. Obstacles to overcome before applying hiPSC derivatives clinically include (1) standardization of lineage specification protocols to produce large quantities of pure, quality-controlled cells of the desired cell-types, (2) methods to ensure their safe delivery, and (3) consideration of all possible adverse effects post-treatment. As mentioned above, however, the hESC-CMs generated to date phenotypically resemble fetal/neonatal cardiomyocytes more closely than adult cardiomyocytes^{4, 5}, and due care should be exercised in extrapolating results using hESC-CMs or hiPSC-CMs to the adult heart.

Generation of cardiovascular cells

Pioneering work by Gordon Keller’s group has definitively demonstrated the cardiovascular potential of a specific population of KDR^low/C-KIT^neg cells derived from hESCs². Specifically, by mirroring developmental cues, they successfully induced generation of cardiovascular colony forming cells from hESCs after temporal exposure to a combination of cytokines and signaling molecules, including BMP4, Activin A, bFGF, VEGF, and DKK1. The KDR^low/C-KIT^neg cells were able to differentiate into all three cardiovascular lineages, cardiomyocytes, smooth muscle cells, and endothelial cells. Cardiovascular progenitors were identified by gene expression profiling showing upregulation of cardiac genes (including NKX2.5, ISL1, TBX5, and TBX20), expression of cardiac troponin T (CTNT), and contractile ability. The cells were also found to be clonal, and therefore could be sources of cardiovascular regeneration. Chemical augmentation of the Wnt signaling pathway was explored, as DKK1, a Wnt pathway inhibitor, increased production of CTNT⁺ cells, while Wnt3a suppressed development. The largest amount of CTNT⁺ cells (~45% of the developing embryoid bodies (EBs)) was produced with a combination of BMP4, bFGF, and Activin A. KDR^low/C-KIT^neg/GFP cells transplanted into the hearts of non-obese diabetic/severe combined immunodeficient mice improved ejection fraction by 31%, and no further complications were reported. Other small animal studies using transplanted hESC-CMs have revealed more modest short-term benefits^{4, 5}, although grafted tissue is susceptible to acute donor cell death, tumorogenesis, and arrhythmogenesis⁸².

More recently, Zhang et al. investigated the cardiovascular differentiation potential of hiPSC lines⁸⁴, potentially bypassing the aforementioned obstacles associated with hESC use. The cardiomyocytes derived from the EBs of hiPSCs lentivirally transduced with OCT4, SOX2, NANOG, and LIN28 were comparable to cardiomyocytes derived from the EBs of hESCs. hiPSC-CMs and hESC-CMs both displayed relative downregulation of OCT4 and NANOG, upregulation of cardiac genes, myofilament protein expression, and sarcomeric organization. hiPSC-CMs also proliferated robustly, generated atrial, nodal, and ventricular action potentials, and responded to electrical and chemical stimulation of the β-adrenergic signaling pathway. There were somewhat fewer hiPSC-CMs that exhibited contractile ability than hESC-CMs, and ineffective silencing of the OCT4 and NANOG transgenes was demonstrated in RT-PCR analyses of the hiPSC-EBs. However, Zhang et al. note that the differences between hiPSC-CMs and hESC-CMs are comparable to the differences already observed between hESC lines. Moreover, the aforementioned nonviral nonintegrating techniques to generate hiPSCs can bypass the problem presented by the transgene insertions in this study, thus showing promise for hiPSC lines as an alternative to hESC lines for a variety of applications.

Although they do not involve pluripotent stem cells, two recent studies warrant mention due to their novelty. Ieda et al. demonstrated direct transdifferentiation of murine fibroblasts into cardiomyocytes, entirely bypassing the pluripotent stage, using a three-factor combination of developmental transcription factors⁸⁵. In vitro, 30% of cells exhibited CTNT expression one week after viral transduction with the GATA4, MEF2C, and TBX5 factors. Immunocytochemistry confirmed the presence of sarcomeric α-actinin and atrial natriuretic factor in some of these cells. Additionally, transduction of these factors into murine hearts induced cardiomyocyte differentiation after a single day, demonstrating a potential proof-of-concept therapeutic application. Although the induced cardiomyocytes were found to be epigenetically and electrophysiologically similar to wild-type cardiomyocytes, the cardiomyocyte-specific genes ACTC1, MYH6, RYR2, and GJA1 were not detected.

Direct reprogramming of murine fibroblasts to cardiomyocytes was subsequently been demonstrated using a slightly different transdifferentiation strategy⁸⁶. By briefly overexpressing the Yamanaka factors and carefully controlling cardiogenic media supplementation, Efe et al. successfully transdifferentiated up to 40% of mouse fibroblasts into functional CTNT⁺ cardiomyocytes within 18 days. Importantly, small molecule signaling played a pivotal role in modulating cell fate, for example by inhibiting JAK/STAT-driven pluripotency induction. These recently developed protocols provide an important alternate platform for the production of patient-specific cardiomyocytes while entirely avoiding the pluripotent state and its concomitant tumorigenic risk.

Disease modeling using hiPSC-CMs

Several exciting demonstrations of the disease modeling capability of hiPSC-CMs have recently been published (Table 3). In all studies to date, the resulting patient-specific hiPSC-CMs have been found to at least partially exhibit the phenotype of the diseases under investigation: long QT syndrome, Timothy syndrome, and LEAPORD syndrome.

Table 3.

Studies utilizing hiPSC-CMs for cardiac disease modeling.

Disease	Reprogramming Method	Methods for evaluation of phenotype	Reference
LEOPARD syndrome	retroviral integration	microscopic morphometry immunocytochemistry phosphoproteomic analysis using antibody arrays and Western blots	89
type 1 long QT syndrome	retroviral integration	single cell patch clamping immunocytochemistry adrenergic response	87
type 2 long QT syndrome	retroviral integration	single cell patch clamping microelectrode arrays pharmacological response	88
Timothy syndrome	retroviral integration	single cell patch clamping calcium transient imaging response to roscovitine	90

Abbreviations: hiPSC-CM = human induced pluripotent stem cell-derived cardiomyocyte

A seminal study evaluated the use of hiPSCs in modeling LEOPARD syndrome, an autosomal-dominant developmental disorder of multiple organ systems resulting from a missense mutation in the PTPN11 gene⁸⁹. Compared to control hESC-CMs, diseased hiPSC-CMs were noted to have a higher mean cell surface area as well as nuclear translocation of the NFATC4 transcription factor, perhaps representing in vitro molecular surrogates of the disease’s cardiac hypertrophy phenotype. In addition, phosphoproteomic comparison of the diseased and wild-type hiPSCs revealed that diseased hiPSCs did not respond to attempted MAPK activation by bFGF. Although defects in MAPK activation are an expected result of mutation in the PTPN11 gene encoding SHP2 tyrosine phosphatase, it is worthwhile to note that the standard procedures for analysis of cardiomyocyte hypertrophy, such as protein synthesis rate and activation of the fetal gene program, could not be reliably assessed in the mixed population of cells resulting from attempted cardiomyocyte differentiation.

A subsequent study evaluated type 1 long-QT syndrome in hiPSCs by comparing wild-type cells to patient-specific cells containing a [569G→A] missense mutation in the KCNQ1 gene⁸⁷. Once the patient-specific hiPSCs had been directed to the cardiac lineage, the ventricular and atrial myocytes’ action potentials had significantly longer QT-intervals and slower repolarization velocity as compared to wild-type cardiomyocytes. Type 1 long-QT syndrome is attributed to a reduction in the I_KS (slow outward potassium current) responsible for mediating action potential repolarization, due to the defective KCNQ1 channel. In agreement with this, single-cell electrophysiological analysis on the ventricular patient-specific hiPSC-CMs revealed a reduction in this current, confirming that the KCNQ1 mutant interferes with the function of the wild-type subunit. Immunocytochemical tests of both populations of cardiomyocytes suggested that the phenotype is the result of a trafficking defect, in which the mutated KCNQ1 protein fails to achieve membrane targeting. Lastly, stimulation of the ventricular patient-specific hiPSC-CMs using isoproterenol had little to no effect on the patient-specific hiPSC-CM’s repolarization and I_KS currents, whereas a significant reduction of both was produced in the wild-type cells. Additionally, the patient-specific hiPSC-CMs had reduced action-potential duration:action-potential interval ratios. These adrenergically stimulating tests cumulatively suggest that long-QT syndrome hiPSC-CMs are predisposed to arrhythmic events.

Another study has extended the above-mentioned findings by modeling type 2 long QT syndrome using similar methods⁸⁸. These hiPSCs were derived from type 2 long QT syndrome patients and contained a missense mutation in the KCNH2 gene. The resulting hiPSC-CMs exhibited the electrophysiological hallmarks of the disease, including prolonged action potential duration and early after depolarizations in patch clamping studies, as well as prolonged field potential duration in microelectrode array studies. Diseased hiPSC-CMs displayed the expected defect in I_Kr (delayed-rectifier potassium current), as well as increased susceptibility to pharmacologically-induced arrhythmogenesis. The authors also demonstrate an important proof-of-concept drug screening experiment to evaluate the effects of nifedipine, pinacidil, and ranolazine on the electrophysiological properties of the diseased hiPSC-CMs.

Yazawa et al. derived hiPSCs from patients with Timothy syndrome, a disorder in which patients suffer from long QT syndrome, autism, immune deficiency, and syndactyly due to a mutation in the CACNA1C gene encoding the Ca_v1.2 L-type channel⁹⁰. Beating hiPSC-derived EBs displayed irregular contraction rates, while single hiPSC-CMs displayed increased action potential duration as well as reduced voltage-dependent inactivation of the L-type calcium channel current. Interestingly, ventricular cardiomyocytes, but not atrial cardiomyocytes, displayed the prolonged action potential phenotype indicative of long QT syndrome, in contrast to the results of Moretti et al.

CONCLUSION

The ability to safely and efficiently derive hiPSCs may be of decisive importance to the future of regenerative medicine, and it depends on researchers’ eventual ability to generate hiPSCs free from foreign chemical or genomic elements while maintaining a stable cell line. Many techniques for hiPSC derivation have been developed in recent years, utilizing different starting cell types, vector delivery systems, and culture conditions. A refined or perfected combination of these techniques might prove to be the key to generating clinically applicable hiPSCs. Although this review is by no means meant to be exhaustive, we have attempted to highlight the significant developments in clinical translation of hiPSCs and their relevance to cardiovascular disease. Among these developments, the most exciting techniques involve enhancement of reprogramming by small molecules (Table 1), nonviral nonintegrating methods for hiPSC derivation (Table 2), and their subsequent application to disease modeling platforms (Table 3). Although outside the scope of this review, it bears noting the critical need for standardization of the parameters that define a cell as a hiPSC³⁹. Without consensus on the criteria for identification of hiPSCs, comparison between the results of various investigators is difficult, as the guidelines by which various investigators select their fully reprogrammed hiPSCs vary. Future endeavors to advance hiPSCs as clinically relevant cardiovascular therapies may utilize an exciting combination of the methods reviewed here, with additional improvements sure to follow.

Table 1.

Summary of different techniques for generating hiPSCs by integration of reprogramming factors into the genome. Retroviral and lentiviral methods are shown with associated enhancements due to addition of small molecules to reprogramming cell culture media.

	Vector	Factors (+ conditions)	Human Starting Cell Type	Approximate Reprogramming Efficiency	Approximate Speed	Reference
VIRAL	retroviral	OSKM	fibroblasts	0.02%	25 days	11
				<0.01%	31 days	91
			keratinocytes	0.8%	10 days
		OSK		0.05%	20 days
	lentiviral	OSKMN	fibroblasts	0.001%	31 days	37
			keratinocytes	0.002%	10 days
		OSKM	adipose-derived	0.2%	18 days	92
		OSK		0.002% - 0.02%	29 – 42 days	93
		OSNL	hematopoetic ESC-derived	0.02%	20 days	41
		OSN	amnion-derived	0.1%	>14 days	94
	retroviral	OSK + VPA	neonatal fibroblasts	1%	30 days	32
		OS + VPA		0.002%
		OSKM + VPA + vitamin C	adipose-derived	7.06%	30 – 34 days	76
			fibroblasts	6.2%
			umbilical cord matrix-derived	0.4%	13 – 20 days	77
			placental amniotic membrane-derived	0.1%
		OSKM + butyrate	fetal fibroblasts	16.3%	18 days	56
	lentiviral	OSNL + hLIF + PD0325901 + A-83-01 + CHIR99021		0.0025%	28 days	95, 96
		OK + Parnate + CHIR99021	neonatal keratinocytes	0.002%	49 days	30
		OK + Parnate + CHIR99021 + PD0325901 + SB431542		0.0075%
		O + PD0325901 + butyrate + A-83-01 + PS48		0.00017%	56 days	79
		O + Parnate + CHIR99021 + PD0325901 + butyrate + A-83-01 + PS48	keratinocytes	0.00025%
			umbilical vein endothelial-derived	0.01%	42 days
			amniotic fluid-derived	0.004%

Abbreviations: O = OCT4; S = SOX2; K = KLF4; M = c-MYC; N = NANOG; L = LIN28; VPA = valproic acid; hLIF = human leukemia inhibiting factor.

Table 2.

Summary of different nonviral, nonintegrating techniques for generating hiPSCs. Episomal, excisional, protein- and RNA-mediated methods are shown with associated enhancements due to addition of small molecules to reprogramming cell culture media.

	Vector	Factors (+conditions)	Starting Cell Type	Approximate Reprogramming Efficiency	Approximate Speed	Reference(s)
EXCISIONAL	lentiviral	OSKM	neonatal keratinocytes	0.00001%	22 – 35 days	97
			fibroblasts	0.0024%	21 – 28 days	53
	piggyBac transposition	OSKM	embryonic fibroblasts	0.008%	14 – 28 days	55
				0.00625%	> 14 days	54
		OSKML + butyrate	bone-marrow-derived	0.16%	24 days	56
EPISOMAL	plasmid DNA	OSKMNL, SVLT, EBNA1	fibroblasts	0.00045%	17 days	57
	minicircle DNA	OSNL	adipose-derived	0.005%	26 – 28 days	61
			fetal fibroblasts	0.0005%
PROTEIN	polyarginine tract tagged polypeptides	OSKM	neonatal fibroblasts	0.001%	56 days	65
RNA	modified synthetic mRNAs	OSKM	fibroblasts	1.4%	17 days	66
		OKSML + ↓[O2]		4.4%

Abbreviations: O = OCT4; S = SOX2; K = KLF4; M = c-MYC; N = NANOG; L = LIN28; SVLT = SV40 large T gene; EBNA1 = Epstein-Barr nuclear antigen-1.

Non-standard Abbreviations and Acronyms

hiPSCs

human induced pluripotent stem cells

hESCs

human embryonic stem cells

hESC-CMs

hESC-derived cardiomyocytes

hiPSC-CMs

hiPSC-derived cardiomyocytes

DOX

doxycycline

RT-PCR

reverse-transcription polymerase chain reaction

TGF-β

transforming growth factor beta

VPA

valproic acid

GSK-3

glycogen synthase kinase 3

ALK 5

activin receptor-like kinase 5

NaB

sodium butyrate

MEK

mitogen activated protein kinase kinase

BMP4

bone-associated morphogenic protein 4

bFGF

basic fibroblast growth factor

VEGF

vascular endothelial growth factor

DKK1

Dickkopf-related protein 1

CTNT

cardiac troponin T

KDR

kinase insert domain receptor

GFP

green fluorescent protein

EBs

embryoid bodies

YFP

yellow fluorescent protein