Pluripotent stem cells: the last 10 years

The evolution of technologies for generating pluripotent stem cells, 2006–2016

At the start of the decade, it had been 8 years since the initial isolation of human embryonic stem cells (hESCs) [1] and incremental scientific progress was being made. However, ethical dilemmas regarding the use and/or destruction of human embryos (which occurs during the most commonly used hESC derivation method) as well as legislative barriers in several countries hindered hESC research endeavors [2]. Moreover, the need to source several hundred embryos for the creation of hESC lines to cover the diversity of HLA phenotypes made clinical translation of embryonic stem cell (ESC) based therapies seem difficult [3]. This situation precipitated major initiatives to find alternatives (Figure 1). Single blastomere technology is one such alternative; it was developed in 2006 as a nondestructive ESC derivation method and was first demonstrated for mouse ESCs [4], then adapted for human ESCs in the same year [5]. With this technique, a single cell or ‘blastomere’ is isolated from a morula (8-cell) stage embryo and, after culture and expansion, can give rise to an ESC line. Removal of a single cell has been shown not to interfere with the ability of the remaining embryo to grow and divide normally [4,5]; it was adapted from a single blastomere biopsy process that had been used by in vitro fertilization clinics for pre-implantation genetic diagnostics since the 1990s [6]. Today, pre-implantation genetic diagnostics is routinely performed using later stage embryos as it poses less risk to the developing embryo than biopsying the 8-cell stage [7]. As such, the availability of pronuclear and multicell stage embryos for nondestructive hESC derivation is rather low and despite the success of this technique by other groups [8], single blastomere technology is not widely used in research today (although it worth noting that six of 14 hESC-based clinical trials currently listed on clinicaltrials.gov involves the use of a single-blastomere-derived line, MA09).

Somatic cell nuclear transfer (SCNT) is another alternative for generating hESCs without the destruction of naturally made embryos. This technique has been used successfully in other species such as calves, pigs and mice since the late 1990s and early 2000s [9–11], yet for various reasons including the availability of federal funding, institutional review board (IRB) requirements and public sentiment, it took until 2013 for it to be successfully applied to humans [12]. In SCNT, the nucleus of an unfertilized egg is removed and replaced with the nucleus from a somatic cell. Precise culture conditions coupled with maternal factors within the egg promote the reprogramming of the somatic cell nucleus back to a pluripotent state and can give rise to an ESC line. The first report of human SCNT hESCs used fetal and infant somatic cells as nuclear donors, while a second report used adult cells from 35- and 75-year-old males to successfully derive karyotypically normal SCNT hESC lines [13], thus demonstrating that reprogramming is possible irrespective of the age of the somatic cell donor. Despite these successes, SCNT has not been widely used for ESC derivation due to the need for high-quality eggs and precise microsurgical techniques. Moreover, the requirement for egg donation is a significant barrier to its widespread use. While only a few labs have been able to successfully generate karyotypically normal human SCNT hESCs to date [12–14], further attempts to derive SCNT hESCs are still underway. Just recently, in mid-2016, the Korean Government granted CHA University the right to use 600 cryopreserved eggs in SCNT research in order to generate hESCs that can be used to help find cures for incurable diseases [15]. As researchers are finding ways to improve the efficiency of SCNT [16], it may become a preferred method for the generation of hESCs in the future. Indeed, as far back as the 1960s, SCNT in other species was shown to completely erase lineage-specific signatures in somatic nuclei and reprogram them to a totipotent state [17,18]. SCNT could be used to create banks of HLA-matched hESCs to cover the diversity of HLA types in the human population, especially in countries such as Korea or Japan, where this could be achieved for a significant proportion of the population using a small bank of HLA-homozygous cell lines.

Emergence & optimization of induced PSC technology

Arguably the most important alternative to conventional methods for hESC generation was the invention of induced PSC (iPSC) technology in 2006 [19] and its application to human cells in 2007 [20,21]. iPSC technology avoids the use and destruction of human eggs and/or embryos altogether, thereby largely circumventing ethical controversy. iPSCs are generated through the reprogramming of somatic cells back to an embryonic-like state; the addition of exogenous reprogramming factors triggers this reprogramming process. iPSC technology revolutionized the field of PSC research and led to a 2012 joint Nobel Prize for iPSC pioneer, Shinya Yamanaka, and John Gurdon, for his previous groundwork in exploring the biological principle of reprogramming [22].

Today, generating iPSCs takes many shapes and forms, with different reprogramming factors, different methods for introducing factors to cells, different starting cell types, among others. The technology has undergone a fascinating evolution from its first report in 2006 to the present day and it will continue to evolve in years to come. The first reports of human iPSC (hiPSC) generation by Yamanaka and colleagues used Oct4, Sox2, Klf4 and the proto-oncogene, c-myc to reprogram human dermal fibroblasts using retorviruses [20]. In considering the application of iPSCs for clinical use, these and other early iPSC studies highlighted two important safety issues that would steer iPSC research in the years that followed: any cocktail of reprogramming factors should avoid the use of a proto-oncogene such as c-myc, since it confers a risk of developing tumors if its expression is re-activated and nonintegrating reprogramming methods should be developed to avoid the mutagenic risks associated with viral insertion into the genome. A month after the first human iPSC paper was published, another group showed that indeed, the use of c-myc could be avoided when generating human iPSCs. This group still used Oct4 and Sox2, but replaced c-myc as well as Klf4 with Nanog and Lin28, thus removing the risk of using a proto-oncogene for reprogramming [23]. Several other groups followed suit and began experimenting with different combinations of reprogramming factors and different types of starting somatic cells [24–28]; it soon appeared that various combinations of factors and various starting somatic cell types could be used to generate iPSCs.

Within a couple of years, nonintegrating reprogramming methods were being reported as well. Nowadays, mRNA [29], recombinant proteins [30], episomes [31,32], mini-circles [33], PiggyBac transposons [34] and Sendai virus [35] have all been used to generate so called second-generation iPSCs. In addition, small molecules, such as methyltransferase inhibitors such as 50-azacytidine and RG108 and/or histone deacetylase inhibitors such as valproic acid have been found to enhance reprogramming efficiency when used in combination with the typical cocktails of genetic factors [36,37]. A mixture of seven small molecules alone (without any genetic factors) has also been reported to reprogram mouse somatic cells [38] suggesting that a chemical approach may also work for generating human iPSCs. These second-generation reprogramming methods not only avoid the risk of tumor formation associated with their integrating virus-based predecessors but they have also helped improve reprogramming efficiency [39].

In 2009–2011, right around the same time that various second-generation reprogramming methods were being developed, reports were starting to emerge that iPSCs were not equivalent to ESCs and that differentiation potential of iPSCs was either impaired or skewed based on the starting somatic cell type [40–43]. Epigenetic [42,44–46] and genetic [47–52] analyses showed that iPSCs display different DNA modification, histone modification and gene expression patterns than ESCs and that different iPSC lines also varied from one another in this manner. Differences in the somatic cell type used for reprogramming, the specific reprogramming method employed, as well as the extent of culturing are thought to influence the degree of disparity between various iPSC lines and/or ESCs [42–43,45,53]. Yet, in some instances, epigenetic memory can be reduced or even eliminated through subsequent passaging of iPSC clones, or alternatively by differentiation and secondary reprogramming, whereas errors that arise during reprogramming may be corrected through the addition of chromatin modifying drugs to the culture media [45,54]. Improvements and modifications made to reprogramming methods over the past decade have helped improve the safety and quality of iPSCs such that the development of iPSC-based therapies is moving forward rapidly. In years to come, the development of iPSC-based therapies may overtake conventional hESC-based ones since their generation does not involve the destruction of embryos or even the use of any unfertilized eggs. This is particularly appealing for the long-discussed generation of banks of HLA-matched PSCs to cover patient diversity on a larger scale and reduce or avoid the need for concomitant immunosuppression.

The start of clinical trials for PSC-derived therapies

PSCs may be useful for treating a wide variety of diseases given their ability to differentiate, theoretically, into every cell type in the body. The last 5–6 years have seen the PSC field begin to deliver on this promise, with a handful of clinical trials being approved in spinal cord injury, macular degeneration, diabetes and heart disease. Starting it off in 2009, Geron received investigational new drug (IND) approval to begin testing its hESC-derived oligodendrocyte precursors, GRNOPC1 (location) in a Phase I trial for spinal cord injury. This was the first trial aimed at testing the safety and potential efficacy of a PSC-derived therapy. In October 2010, Geron began transplanting GRNOPC1 into spinal cord patients and the following year, presented safety data at the American Congress of Rehabilitation Medicine suggesting that GRNOPC1 was well-tolerated and caused no serious adverse effects. However, a month later, Geron unexpectedly announced that they were stopping the study to focus on oncology drug-based therapies [55]. A total of five patients received injections of GRNOPC1 in the short-lived trial. In October 2013, Asterias Biotherapeutics, a subsidiary of Biotime, Inc. acquired Geron's hESC assets, and stated its intention to resurrect the defunct trial by rebranding the experimental therapy as AST-OPC1. Within 7 months, Asterias secured US$14.3 million in funding from the California Institute of Regenerative Medicine (CIRM) [56]. A few months later, in August 2014, the US FDA cleared Asterias for a new Phase I/IIa clinical trial (NCT02302157) to transplant AST-OPC1 into spinal cord injury patients. As of July 2016, Asterias has dosed a total of eight patients in this trial: the first three patients received a low dose of 2 million cells each (cohort 1) and the other five received 10 million cells each (cohort 2), which is predicted to be within the efficacious range [57]. Also in mid-2016, Asterias announced results of the 4–5 year follow-up of the original five patients from the Geron trial. The data suggest long-term safety of the therapy as well as reduced spinal cord cavitation (deterioration) in four of the five patients [58].

In 2010, a few months before Geron transplanted GRNOPC1 into its first patient, Advanced Cell Technology (ACT; MA, USA) received IND approval to begin testing hESC-derived retinal pigment epithelium (RPE) for age-related macular degeneration and Stargardt disease, a juvenile form of macular degeneration. ACT transplanted hESC-RPE into their first age-related macular degeneration (AMD) patient in July 2011 and continued to enroll patients over the next few years without incident, despite Geron's abandonment of its trial during this time. In 2012 and 2014, ACT published preliminary (4 months) [59] and medium to long-term (average follow-up period of 22 months) [60] safety data from its trials, which showed that subretinal injection of hESC-RPE was well-tolerated. There were no reported serious adverse events and no evidence of abnormal cell growth or tumor formation from the transplanted cells. Optical coherence tomography imaging showed the existence of subretinal pigmented patches in the eyes receiving hESC-RPE, suggesting that the cells engrafted and could survive even after perioperative immunosuppression was stopped. Vision-related quality of life was reported to improve in many patients who received the therapy. In addition, out of 18 patients (nine with AMD, nine with Stargardt), eight improved on their visual acuity tests by 15 letters or more. Three patients improved by 5–15 letters, six remained stable and one decreased. Currently, ACT (briefly called Ocata Therapeutics, but renamed the Astellas Institute for Regenerative Medicine after being acquired by Astellas Pharma Inc. in early 2016) has six Phase I and Phase II clinical trials registered on ClinicalTrials.gov evaluating the use of its hESC-derived RPE for treating dry AMD and Stargardt in the USA and UK. In addition, RPE derived from their hESCs re being used in two additional Phase I/II clinical trials in South Korea for dry AMD and Stargardt; these trials have similarly shown positive safety data, with no adverse proliferation, tumorigenicity or ectopic tissue from the treatment [61].

Around the same time that ACT's 2014 safety data were being published, Japan's RIKEN Institute successfully transplanted the world's first iPSC-derived therapy into humans. They too chose the eye and (wet) AMD as a first indication but decided to transplant autologous iPSC-derived RPE into patients instead of using an off-the-shelf allogeneic cellular product. The use of autologous cells is thought to avoid the risks of immune rejection [62] and has therefore been an attractive option, although it necessitates more time and labor since custom-made, individualized lots of iPSCs need to be generated for each patient. In the RIKEN trial, transplantation of autologous iPSC-RPE into the first patient went well and there were no adverse events, yet the trial was suspended after just this one patient. A genetic mutation, potentially in a known oncogene, was found in the autologous iPSCs generated for the trial's second patient. Investigators involved in this trial have since tested the concept of using HLA-matched allogeneic iPSC-RPE in nonhuman primates and observed no rejection in the absence of immunosuppression [63]; however, this concept has not yet been tested empirically in humans. The lead investigator, Masayo Takahashi, has said that the trial will likely resume with the use of allogeneic iPSC-derived RPE presumably since a single lot of quality control (QC)-validated RPE can be used for many patients, yet so far the trial remains suspended [64]. Since 2012, a handful of other groups have received IND approval to test their own PSC-derived RPE for AMD (Figure 2). Given the risks of first-in-human PSC-based therapies, the eye is considered a logical place to begin developing therapies. First, the eye is a locally contained environment, providing a natural barrier to any potentially deleterious cells spreading systemically. Second, its immune-privileged nature may make it more accepting of transplanted allogeneic cells in the long-term. Third, the lens provides a way to noninvasively image the transplantation site over time and functional readouts such as visual acuity are easy to obtain. Indeed, CellCure Neuroscience, Pfizer, Regenerative Patch Technologies (RPT)/California Project to Cure Blindness (CPCB) and most recently two groups in China all have active trials listed on clinicaltrials.gov for evaluating hESC-derived RPE as a therapy for AMD. Pfizer and RPT/CPCB are using an immobilized membrane approach while CellCure Neuroscience is testing a cell suspension. Pfizer's trial, in collaboration with the London Project to Cure Blindness, Moorsfield Eye Hospital, the University College London Institute of Ophthalmology and the National Institute for Health Research in the UK successfully transplanted cells into its first patient in the fall of 2015 [65]. RPT utilizes a hESC-derived RPE monolayer developed by the California Project to Cure Blindness with a grant from the CIRM [66] and they are actively recruiting patients for their trial.

More than a decade of PSC research and development has also led to clinical trials for PSC-derived therapies in other disease areas (Figure 3). In 2014, Viacyte received IND approval to begin a Phase I/II trial to treat Type 1 diabetes. Their product, VC-01, consists of hESC-derived pancreatic endoderm cells encapsulated in a biocompatible drug delivery device that can be implanted under the skin. The semipermeable device permits the release of metabolically active factors while allowing nutrients and oxygen into the device and protecting the encapsulated cells from immune-rejection. Preclinical work has shown that once implanted, the cells differentiate and produce insulin, which is released from the device in sufficient quantities to regulate blood glucose levels in a mouse model of diabetes [68]. Viacyte's trial, NCT02239354, is now recruiting patients.

In addition to the above trials, a PSC-derived therapy was approved for an ischemic heart disease Phase I clinical trial in 2013. The Assistance Publique-Hopitaux de Paris is testing the feasibility and safety of CD15⁺ ISL1⁺ hESC-derived cardiac progenitors for improving heart function in patients with severe left ventricular systolic dysfunction. The hESC-derived cardiac progenitors are embedded in a fibrin patch and engrafted onto an area of epicardium during a scheduled coronary artery bypass or mitral valve surgery. An autologous flap of pericardium placed over the patch is designed to provide nutrient support to the embedded progenitors. Preclinical evidence shows the cells engrafted and differentiated into cardiomyocytes in a nonhuman primate myocardial infarction model [69]. In addition, these cells were shown to improve cardiac function in rodents even though the engrafted cells were no longer found 4 months after the surgery [70]. Transient survival of engrafted progenitors may provide paracrine signaling to recruit endogenous progenitors and/or accelerate endogenous repair mechanisms and potentially explain the sustained functional improvement despite disappearance of the engrafted cells. This trial, NCT02057900 is also recruiting patients, with six enrolled thus far.

The next wave of PSC-derived therapies destined for clinical trials

The last decade has also seen incredible progress on the development of other PSC-based therapies, some very close to beginning clinical trials. Several groups including the New York Stem Cell Consortium and Jun Takahashi's group at Kyoto University have made great progress in generating PSC-derived dopaminergic (DA) neurons for the treatment of Parkinson's disease (PD). Preclinical work has shown that both hESC- [71,72] and iPSC-derived [73] DA neurons rescue motor function in a 6-OHDA rat model of Parkinson's disease. Another study showed that transplantation of autologous iPSC-DA neurons into the putamen of cynomolgus monkeys resulted in long-term (up to 2 years) survival of the engrafted cells and improvements in motor neuron function [74]. A consortium of researchers developing stem cell-based therapies for PD called G-force PD was established in 2014 as a forum to discuss their collective progress and challenges [75]. Universal challenges include uncertainties regarding the body of preclinical evidence that regulatory agencies will require in order to demonstrate safety and efficacy of PD cell-based therapies, GMP manufacturing and scale-up issues, clinical trial design, ethics and commercialization. Despite these potential hurdles, it is likely that one or more of these groups will be able to start clinical trials in the next few years.

A long-standing goal for PSC research has been the in vitro generation of glucose-responsive, insulin-producing mature pancreatic β cells to treat diabetes. Many ESC/iPSC differentiation protocols have been developed for β-cell generation, yet it has been consistently challenging to fully mature them in vitro [76]. In 2014, a protocol developed in Doug Melton's lab was finally able to overcome this challenge and resulted in the in vitro generation of β cells expressing mature pancreatic β cell markers such as Pdx1 and Nkx6.1. Importantly, these cells were shown to secrete insulin in a glucose-responsive manner and be capable of regulating hyperglycemia in preclinical models [77]. A follow-up study has shown that encapsulation of these β cells within an alginate matrix protects them from rejection in an immune-competent streptozotocin-induced diabetic mouse model without compromising their ability to reverse hyperglycemia [78]. This work is now part of the recently established Semma Therapeutics, which joins Viacyte in the quest to develop a PSC-based therapy for Type 1 diabetes.

PSCs are being developed for therapeutic use in various other diseases as well. For example, autologous iPSCs are being generated for patients with the blistering skin disorder, epidermolysis bullosa as part of a cell replacement strategy. Patches of skin with spontaneous revertant mosaicism, in which the disease-causing gene has spontaneously corrected itself [79] or diseased skin samples that have undergone gene editing [80] are being used to produce iPSCs, which in turn are differentiated into normal keratinocytes to use in skin grafts for these patients. In the eye, retinal progenitors are being developed from both ESCs [81–83] and iPSCs [81,84] to use as a cell replacement therapy for retinal degenerative diseases, such as retinitis pigmentosa (RP), whereby transplantation of the progenitors would lead to in vivo differentiation and functional engraftment by mature photoreceptors. PSCs are also being developed to provide trophic support and/or to maintain the health of endogenous cells at risk for degeneration in various diseases. For example, iPSC-derived macrophages are being manipulated for therapeutic use in Alzheimer's disease (AD) patients. These macrophages have been engineered to express high levels of the β-amyloid-degrading enzyme, neprilysin 2, in an effort to reduce the burden of disease-associated plaques and spare the health of existing neurons in AD [85]. Similarly, in amyotrophic lateral sclerosis (ALS), iPSC-derived neural stem cells may provide therapeutically useful trophic support to endogenous neurons, as shown in a SOD1G93A ALS mouse model [86].

Researchers have also made some progress in combining PSCs and tissue engineering for transplantation. For example, iPSC-derived endodermal progenitors were combined with human endothelial cells and mesenchymal cells to generate 3D liver buds. Upon transplantation, these buds established vascular connections within host animals, differentiated into mature liver cell types and rescued chemically induced lethal liver failure [87]. PSC-based 3D tissues are under development for other organs such as eye, heart, lung, kidney and brain. These may be used in the future as a source of cells/tissue for transplantation, or as discussed below, for disease modeling and/or drug screening efforts.

PSCs as tools for ‘disease-in-a-dish’ models and drug screening platforms

In addition to direct therapeutic uses for PSC-derived cell types, both hESCs and iPSCs have been used in nonclinical applications (Table 1). This includes the establishment of ‘disease-in-a-dish’ models for various ailments, although the ease of generating iPSCs has made them a more attractive option than sourcing suitable ESCs for this purpose [88]. Major initiatives have been developed to provide central resources for disease-specific iPSCs, including the Coriell CIRM iPSC Biorepository. As of June 2016, this searchable collection had roughly 3000 iPSC lines for diseases ranging from childhood neurodevelopmental disorders to hepatic conditions, eye disorders and respiratory diseases [89]. Similarly, the NYSCF has a repository of over a thousand iPSC lines covering a broad range of indications and a searchable database to help researchers identify relevant disease-specific iPSC lines for their specific interests [90].

Over the past several years, numerous publications have described the differentiation of disease-specific PSCs into relevant cell types for disease modeling purposes (reviewed in [91]). Monogenic diseases such as familial hypercholesterolemia, spinal muscular atrophy, Huntington's disease or the pulmonary disorders α-1 antitrypsin deficiency and cystic fibrosis are among the most straightforward to model since alterations in a single gene are largely responsible for the disease phenotype. For example, the Huntington's disease consortium established a small collection of iPSC lines from Huntington's disease patients and nondiseased controls; differentiation of these lines toward a neuronal phenotype has allowed consortium investigators to examine HD-specific alterations in cell metabolism, stress responses, adhesion properties, among others. [92]. In another example, over 100 iPSC lines from patients with various lung diseases were including α-1 antitrypsin deficiency, cystic fibrosis and emphysema were generated to study their specific disease phenotypes and are also being used in the development of gene-correction strategies for many inherited pulmonary disorders [96]. In addition, iPSCs from patients with NKX2-1 haploinsufficiency have been used to model hypothyroidism and to determine the signaling pathways governing thyroid lineage differentiation and maturation with the goal of developing cell replacement strategies to treat this disease [100].

Complex diseases, which may involve multiple genes, environmental influence, or interplay between multiple cell types are more challenging to model in vitro but progress has been made in using PSCs for this purpose as well. As an example, iPSCs have been generated from both familial and sporadic AD patients to examine differential stress responses from AD iPSCs upon their differentiation into neurons [114]. Sporadic AD iPSC-derived neurons have also been used to show that variants in the SORL-1 gene can lead to increased risk of developing AD due to reduced responsiveness to BDNF and its resulting effects on apolipoprotein processing [93]. In many other examples, the modeling of complex diseases is facilitated by the development of 3D organoid systems. 3D cultures involving stratified layers of the retina have been created to study retinal degenerative diseases [94], while intestinal organoids complete with epithelial derived villus-like structures, Goblet and paneth cells have been created to study intestinal development and diseases such as enteric anendocrinosis [98]. iPSC-based kidney organoids are also under development. In a recent study, kidney organoids were shown to contain nephrons which descend into distal and proximal tubules, an early loop of Henle and vascularized glomeruli, similar to kidneys during the first trimester of embryonic development [99]. Such organoids can be used to study nephrogenesis and various kidney diseases. 3D iPSC-based lung organoids have been created to study diseases such as idiopathic pulmonary fibrosis and may also be applied to development of organ transplant strategies [97]. Lastly, self-organizing midbrain organoids derived from iPSCs are being used to generate DA neurons and neuromelanin-producing cells for the study of Parkinson's disease and other neurologic diseases [95].

Over the last decade, there has been great interest in using PSCs for drug discovery, drug screening and evaluation of potential drug toxicities. Differentiation protocols have been improved in terms of efficiency, maturation and yield such that now, a variety of different PSC-based platforms are being used. For example, neural crest progenitors differentiated from familial dysautonomia patient-specific iPSCs were used to screen an approximately 7000 small molecule library for those that could rescue expression of IKBKAP, the inadequate transcription of which causes the fatal neurological disease. The screen led to the discovery that α2 adrenergic receptor activity can regulate IKBKAP expression; drugs that increase this receptor's activity may be useful therapeutic agents for familial dysautonomia [103]. In another study, cardiac progenitor cells derived from iPSCs were used in a screen for compounds that would enhance the proliferation and differentiation of the progenitors, which could help facilitate cardiac tissue repair. The screen led to the discovery that inhibitors of TGF-β type 1 receptor kinase stimulate cardiac progenitor differentiation into cardiomyocytes [115]. In another proof of concept experiment, ALS patient-specific iPSC-derived motor neurons contained insoluble protein aggregates in the cytosol and had short neurites, mirroring the phenotype of ALS patient motor neurons. Gene expression analysis of these cells led to proof of concept testing of chemical compounds for correcting these abnormalities. The histone acetyltransferase inhibitor, anacardic acid, was found to reverse the deleterious ALS motor neuron phenotype. The authors suggest that this iPSC motor neuron based system should therefore be a useful drug screening platform to develop novel drugs for treating ALS [102].

Cardio- and hepatotoxicity are major safety concerns for any new drug. It is estimated that the development of one third of drugs has been discontinued because of suspected cardiotoxic effects [104]. Many of these failures come late in drug development; therefore implementing toxicity screening early in development could save considerable time and costs. Over the past 10 years, researchers have begun to use PSC derivatives to screen for potentially toxic effects of drugs earlier in their development (reviewed in [101]). For example, a panel of iPSC-derived cardiomyocytes was generated from patients with inherited cardiac diseases such as long QT syndrome, familial hypertrophic or familial dilated cardiomyopathy as well as from normal healthy controls. These were used to model disease-specific cardiotoxicity profiles and to evaluate differences in susceptibility to known cardiotoxic drugs [105]. In another study, iPSC-derived cardiomyocytes were used in a system that was designed to mimic the organization and physiological responsiveness of cardiac tissue better than individual cardiomyocytes. The system can evaluate electrophysiological, physiological and biological properties of cardiac tissue in response to different pharmacological agents and therefore may be quite useful in evaluating potential cardiotoxicities of new drug candidates [104].

Hepatotoxicity is another major safety concern for drugs in development and PSC-based systems are also being used to evaluate the potentially toxic effects of novel drugs to the liver. For example, in a proof of principle experiment, iPSC-derived hepatocytes from a variety of individuals were used to screen a library of 240 heterogeneous compounds with known hepatotoxic effects. The iPSC-hepatocytes were assessed for drug-induced effects on viability, apoptosis, mitochondrial membrane potential, phospholipid accumulation, cytoskeletal alterations and other properties [106]. Such a multiparametric system can provide information on potential mechanisms of toxicity through a comparison to compounds with well-characterized toxicities and also be used to address potential hepatotoxic effects of drugs on specific patient populations, including the elderly or those with a specific type of disease. Pushing PSC-derivatives toward mature, adult cellular phenotypes should help improve the accuracy and reliable of PSC-based toxicity screening platforms. Toward this end, 3D scaffolds supplemented with primary cardiomyocytes are being used to drive the maturation of iPSC-cardiomyocytes [116] and 3D liver buds are being used to functionally mature iPSC-derived hepatocytes [117].

Gene editing & PSCs: approaches

Gene editing technologies have been developed to correct disease-causing genetic mutations, functionally replace and/or knock-out expression of dysfunctional genes. Nuclease-based methodologies for editing the genome dominate the field of gene editing and major classes include natural homing endonucleases-zinc finger nucleases (ZFNs) as well as transcription activator-like effector DNA-binding domain nucleases (TALENs), and clustered regularly-interspaced short palindromic repeats (CRISPR) technology driven by Cas9 nuclease activity (aka ‘Crispr/Cas9’). These systems were first applied to PSCs back in 2009 when two different groups utilized ZFNs to edit genomic sequences at a variety of discrete loci within hESCs and iPSCs [118,119]. Since then, both TALEN-based gene editing and CRISPR/Cas technology have been applied to PSCs by various groups (reviewed in [107]). Multiplexing capability as well as an observed increased targeting efficiency of CRISPR/Cas over TALENs capability has made CRISPR technology perhaps the most popular choice for gene editing [120]. In addition, the combination of TALENS and CRISPR technologies has been used to create an inducible multiplex gene targeting system called ‘iCrispr’ for temporal control over gene editing at discrete stages of differentiation in iPSC-based disease modeling [121].

An alternative to nuclease-based gene editing is the use of adeno-associated viruses (AAVs); various serotypes, strains and recombinant AAVs have been developed to facilitate gene therapy as well as gene editing endeavors in various cell types, including PSCs [122]. Although AAV use in PSCs has not been as popular as that of TALENS or CRISPR/Cas, they appear to have an excellent safety record with more than 100 clinical trial testing AAV variants for therapeutic purposes ([123,124]).

Gene editing & PSCs: goals

Regardless of the editing system employed, the objectives of PSC-based gene editing endeavors fall into two major categories: improving disease models and drug screening systems through the creation of isogenic controls, and gene editing for cell-based therapies. For the former, isogenic controls created through gene editing technology can facilitate the study of mutations and complex diseases in a properly controlled and isolated manner. For example, ZFN technology was used to create isogenic controls for the study of long QT syndrome in iPSC-derived cardiomyocytes [125]. In another study, hetero- and homozygous mutations in the genes for apolipoprotein and presenilin 1 were generated in iPSCs with Crispr/Cas technology and their differentiation into cortical neurons has proven to recapitulate specific features of AD [108]. On a larger scale, the UK-based company, Horizon Discovery has created a genome-editing platform for the generation of several hundred pairs of isogenic cell lines using their HAP1 cell line [126]. A similar large scale platform could also be applied for PSC-based isogenic controls.

The second major undertaking for gene editing in PSCs involves cell-based therapies, particularly for monogenic diseases. Proof of principle studies includes a report where Crispr/Cas gene editing was used to correct the mutation of the β-globin gene in iPSCs from a β-thalassemia patient [109]. These corrected iPSCs displayed improved differentiation capacity into various types of hematopoietic progenitors and may be one day used as a source of autologous hematopoietic stem cells for transplantation and repopulation of the hematopoietic system. Similarly, Crispr/Cas9 was used to correct a mutation in the gene encoding the RP GTPase regulator in iPSCs derived from a patient with X-linked RP [127]. These corrected cells could in principle be differentiated into photoreceptors or their progenitors and used in cell replacement strategies for RP patients.

The concept of generating PSC banks to match the diversity of HLA phenotypes has been discussed for several years as a logical way to avoid immunogenicity of PSC-based therapies. Yet, gene editing is now being used to create ‘universal’ PSC lines and their subsequent differentiation into nonimmunogenic cell types. In proof of principle experiments, knocking out expression of β-2-microglobulin, which is the common light chain molecule to class I A, B, C molecules, can reduce immunogenicity in hESCs [110,111]. This approach has also been used to generate universal platelets from human iPSCs, which could be used as a potential strategy for the management of platelet refractoriness [112].

Various efforts are underway to reduce or eliminate immunogenicity of cells in a more comprehensive manner. Chad Cowan's group has stated they are utilizing Crispr/Cas technology to generate universal donor PSCs by eliminating expression of genes involved in immunogenicity in PSCs and enhancing expression of genes promoting immune tolerance such as PD-L1 or HLA G, which can help avoid natural killer cell-mediated lysis [113]. The company Universal Cells, whose work is based on intellectual property generated at the University of Washington, is using a recombinant adeno-assciated virus (rAAV)-based gene editing strategy to knock-out both HLA class I and II expression, while at the same time knocking in expression of the tolerance-inducing HLA E or G and a suicide gene to safeguard against uncontrolled proliferation or other potential untoward effects of cells after transplantation [128]. Creating universal PSCs would obviate the need and expense of generating HLA-matched PSC banks for regenerative medicine. If successful, it could help reduce or avoid the need for harsh immunosuppression and improve the engraftment or persistence of PSC-derived therapeutic cells, particularly for those indications where large numbers of cells are needed.

Common challenges to the use of PSCs

Regardless of the cell source (hESC or iPSC) or type of cell being developed for clinical use, disease modeling or drug discovery, it should be noted that a common challenge for all applications of PSCs is maintaining genomic stability through the culture and/or differentiation process. Various sources of stress (e.g, culture conditions, enzymatic passaging, among others) can lead to the acquisition of chromosomal abnormalities and the potential for tumorigenicity or distortions of cell-based models. Whole genome screening methods such as comparative genomic hybridization are now being used to supplement conventional cytogenetic detection methods (such as karyotype analysis) for identifying genetic abnormalities (reviewed in [129]) and will help safeguard PSC endeavors.

Future perspective

The dramatic progress made over the past decade will almost certainly translate into exciting new advancements in decades to come (Figure 4). First-in-man PSC-based clinical trials have thus far shown that PSC-derivatives are safe to use in humans, and provide the impetus for continued clinical trial testing. To date, trials have almost exclusively employed hESCs, yet that is likely to change in the future. Improvements in iPSC quality should enable these ethically sound alternatives to hESCs to catch up or even pass hESC usage in clinical trials. As differentiation procedures and 3D technologies improve, PSCs will become ever more integral to drug screening efforts and disease modeling, although it is unlikely they will ever fully replace the use of in vivo disease models. Another major advancement that will likely drive PSC research in years to come involves the marriage of gene editing technology with PSCs. The ability to precisely correct disease-causing mutations, create isogenic controls and potentially eliminate immunogenicity of PSC derivatives make gene editing in PSCs an incredibly important endeavor. The PSC field will likely produce additional exciting breakthroughs in the coming decade – advancements that could one day make incurable diseases curable.