The 3R principle: advancing clinical application of human pluripotent stem cells

The first derivation of human embryonic stem cells brought with it a clear understanding that animal models of human disease might be replaced by an unlimited supply of human cells for research, drug discovery, and drug development. With the advent of clinical trials using human pluripotent stem cell-based therapies, it is both timely and relevant to reflect on factors that will facilitate future translation of this technology. Human pluripotent cells are increasingly being used to investigate the molecular mechanisms that underpin normal and pathological human development. Their differentiated progeny are also being used to identify novel pharmaceuticals, to screen for toxic effects of known chemicals, and to investigate cell or tissue transplantation strategies. The intrinsic assumption of these research efforts is that the information gained from these studies will be more accurate, and therefore of greater relevance, than traditional investigations based on animal models of human disease and injury. This review will therefore evaluate how animals and animal-derived products are used for human pluripotent stem cell research, and will indicate how efforts to further reduce or remove animals and animal products from this research will increase the clinical translation of human pluripotent stem cell technologies through drug discovery, toxicology screening, and cell replacement therapies.


Introduction
Human pluripotent stem (PS) cells are providing a plentiful supply of diff erentiated human cell types for developmental biology, drug discovery, and clinical applications. Th e very fact these are human cells brings an implicit opportunity to decrease our reliance upon traditional animal research models of human disease and injury. In the USA alone, over 1 million dogs, cats, primates, rabbits, and other large animals are used for research each year [1]. When rats and mice are included, the number is estimated to be in the tens of millions of animals per year. In Europe, almost 10 million vertebrate animals are used annually for research, and changes to the European Commission REACH legislation (Registration, Evaluation, Authorisation and Restriction of Chemical substances) could result in an additional 3.9 million test animals being required over the next 11 years [2].
Given the time and fi nancial costs associated with animal models, as well as the fact that they often do not accurately predict the toxicity or effi cacy of new treatments and products in humans [3][4][5], it is timely to analyse the various fi elds of human PS cell research while keeping in mind the 3R principle of ethical animal use in research: to reduce, replace, or refi ne animal usage [6]. When viewed from this perspective there are clear avenues to help realise more quickly the commercial potential of human PS cells in biotechnology and clinical research, while at the same time reducing our reliance upon imperfect animal research models.
A variety of human PS cell types are available for research, including embryonic stem (ES) cells, induced pluripotent stem (iPS) cells, embryonal carcinoma cells, and embryonal germ cells. Human ES and iPS cells are the most likely of these pluripotent cells to have direct clinical application; however, all four human pluripotent cell types are considered relevant to investigations of molecular development and novel drug screening. Use of these human PS cells can be divided into three broad yet overlapping areas: pluripotent cell generation and mainte nance; generation and purifi cation of diff erentiated cell types; and research and clinical application of human PS cells and their diff erentiated derivatives. Each of these research areas has used, or continues to use, animals and/ or animal-derived products. However, there is growing recognition that, where possible, fi nding non-animalbased alternatives will help to more quickly achieve the full potential of human PS cells in each research area [7][8][9][10]. As described below, the translation of PS cell tech nology from academic research to biotechnology

Abstract
The fi rst derivation of human embryonic stem cells brought with it a clear understanding that animal models of human disease might be replaced by an unlimited supply of human cells for research, drug discovery, and drug development. With the advent of clinical trials using human pluripotent stem cellbased therapies, it is both timely and relevant to refl ect on factors that will facilitate future translation of this technology. Human pluripotent cells are increasingly being used to investigate the molecular mechanisms that underpin normal and pathological human development. Their diff erentiated progeny are also being used to identify novel pharmaceuticals, to screen for toxic eff ects of known chemicals, and to investigate cell or tissue transplantation strategies. The intrinsic assumption of these research eff orts is that the information gained from these studies will be more accurate, and therefore of greater relevance, than traditional investigations based on animal models of human disease and injury. This review will therefore evaluate how animals and animal-derived products are used for human pluripotent stem cell research, and will indicate how eff orts to further reduce or remove animals and animal products from this research will increase the clinical translation of human pluripotent stem cell technologies through drug discovery, toxicology screening, and cell replacement therapies.
appli ca tions has already begun. Th is includes use of the embryonic stem cell test (EST) to predict embryotoxicity [9] and clinical trials using human ES cell-derived oligodendro cytes and retinal pigment epithelium (RPE) [11,12]. Th ese early successes suggest that further application of the 3R principle to human PS cell research will be economically, clinically, and ethically useful.

Removal of animal products delivers high-quality pluripotent cell cultures
Traditionally, maintenance of human PS cells required co-culture with mouse embryonic fi broblasts in medium containing bovine serum in order to keep the cells in a pluripotent state [13]. Diffi culties with this system -in particular, sourcing suitably supportive batches of fi broblasts and serum -made this approach highly specialised, time consuming, and expensive. Even with validated batches of these reagents, variable levels of spontaneous diff erentiation routinely occurred [7]. Th is meant that maintaining reproducibly undiff erentiated cultures on a weekly basis was challenging. Initial attempts to minimise or eliminate these batch-variable animal products were partially successful, and included elimination of direct feeder cell contact through the use of feeder conditioned medium [14], development of a moderately less variable serum replacement [8], and the establishment of human feeder cells [15].
Continued research into the growth factor requirements of human PS cells led to the publication of a variety of defi ned, feeder-independent media, some of which are fully xeno free [7]. A multi-laboratory international comparison of eight diff erent feeder-free human PS cell media recently demonstrated that two of these media are able to support mid-term to long-term culture of multiple human PS cell lines across multiple groups [16]. Th ese reagents have greatly reduced inter-laboratory variability and thus have made human PS cell research more effi cient and reliable.
Coincident with the development of xeno-free human PS cell culture media, eff orts have also been focused on developing xeno-free culture surfaces. Initial reliance upon mouse embryonic fi broblasts for cell attachment was replaced with animal-derived extracellular matrices [14] or human feeder cells [15]. More recently, xeno-free chemical or recombinant extracellular matrices have been developed that are suitable for use with xeno-free human PS cell culture media [17]. While the cost of these media and culture surfaces was initially a factor limiting their widespread use, this is becoming less of an issue as multiple competing products are being commercialised.
Application of the 3R principle has thus removed animal products from human PS cell maintenance media and culture surfaces while having multiple additional benefi ts, including increased ease of handling, improved culture quality (that is, markedly reduced levels of spontaneous diff erentiation), and generation of a more clinically appropriate culture environment.

Xeno-free human pluripotent stem cell generation
Xeno-free culture reagents for human PS cell maintenance are now being applied for the generation of clinically relevant, xeno-free human ES and iPS cell lines [7,18]. Th ese cell lines display defi ning characteristics of pluripotency, such as self-renewal, tri-lineage diff er en tiation capacity, and karyotypic stability. For over a decade, the teratoma assay has been the most widely used and accepted method to demonstrate multi-lineage diff erentia tion capacity of human PS cells [19,20]. To perform this assay, PS cells are transplanted into immunocompromised mice (under the kidney capsule, into the testis, or subcutaneously) to induce their diff erentiation to derivatives of endoderm, mesoderm, and ectoderm. However, the lack of sensitivity of this assay means it is essentially nonquantitative with respect to the input cell number [19,20]. Moreover, the assay is labour and time intensive (taking 2 to 3 months to complete), and analysis of the diff erentiated cells produced is almost always qualitative (that is, simply reporting whether or not certain cell types are found but not the relative ratios of these). Accordingly, there is a growing consensus that these limitations necessitate the development of improved replacement pluripotency assays [21]. Th is is particularly relevant given the advent of human iPS cells, and the associated pluripotency assessment that will be required to enable clinical application of such patientspecifi c cells.

Xeno-free human pluripotent stem cell characterisation
Replacing the teratoma assay will require the establishment of methods that comprehensively and quantitatively defi ne a broad range of phenotypic and functional parameters characteristic of pluripotent cells. Attempts to do this via genome-wide expression profi ling of undiff erentiated human ES and iPS cells has led to in silico methods, such as PluriNet [22], which can catalogue and compare the gene expression profi les of newly derived cell lines. Encouragingly, these methods can accurately distinguish pluripotent cells from non pluripotent cells, suggesting they may provide an alternate method to teratoma formation for pluripotent cell classifi cation. Not all human PS cells have equivalent diff erentiation potential, however, even though they have similar gene expression profi les [23]. Accurately assessing functional pluripotency (that is, the capacity to produce particular diff erentiated cell types) is thus currently not possible on the basis of gene expression or epigenetics alone. To address this, animal-free in vitro methods of teratoma formation are being explored. One recent study examined the spontaneous diff eren tiation of human ES cells in a perfused three-dimensional multi-compartment bioreactor [24]. Th is system reported generation of embryoid bodies with histological and transcriptional features similar to those that develop in the mouse teratoma assay. However, this approach still relies upon histological analysis to infer pluripotency from the identifi cation of multiple diff erentiated cell types. As histological assessment is time and labour inten sive and does not assess diff erentiated cell function, this may limit the utility of in vitro teratoma formation for human PS cell characterisation.
Interestingly, an alternate animal-free approach for assessing the diff erentiation capacity of new human PS cell lines is emerging. With this approach one or two desired diff erentiated cell types are rigorously and quantitatively analysed using a broad range of cell-type specifi c assays. For example, human PS cell-derived RPE is beginning to be assessed using a range of assays to establish how closely its characteristics resemble primary human RPE. Th is has led to the emergence of the 4P criteria for RPE: (i) pigmentation and (ii) polygonal morphology, assessed via light microscopy and immunofl uores cence (for example, for TRP-1 and claudin 19, respectively); (iii) cell polarity, assessed via immunofl uores cence (for example, Ezrin), transmission electron microscopy (to look for apical projections), or trans epithelial resistance measurements; and (iv) phago cytic capacity, assessed by the ability to uptake fl uorescently-labelled beads or photo receptor fragments [25,26]. Importantly, such comprehensive assess ment provides a detailed phenotypic and functional baseline for human PS cellderived RPE compared with the known behaviour of primary human RPE. In doing so, this approach goes a signifi cant way towards defi ning the suitability of human PS cell-derived RPE for both research and clinical applications. Future development of similar assay panels for other diff erentiated cell types will enable a reduction in the use of animals in research and, equally importantly, will enable robust phenotypic and functional cell characterisation that increases confi dence in the data generated [9].

Improved drug discovery and development using pluripotent stem cells
Validated panels of PS cell-based in vitro assays are also being applied to drug discovery and toxicology assessments [10,[27][28][29][30][31][32][33][34][35][36][37]. Th is application is due not only to the ethical and economic issues associated with animal experi mentation, but also because animal models do not always accurately predict the effi cacy and/or safety of substances in humans [3][4][5]. Various groups are investigating the potential of human PS cell-derived hepatocytes and cardiomyocytes for drug discovery [32,33], while human iPS cell-derived cardiomyocytes are already commercially available for research and drug screening (for example, Cellular Dynamics, GE Healthcare, Lonza). Improved methods for generating and purifying a wider variety of diff erentiated cell types from human PS cells may ultimately enable in vitro toxicity testing of all human cell and/or tissue types.
Additional strategies are also being developed to eliminate inappropriate drugs, chemicals and cosmetics earlier during commercial development according to their toxicity profi le in PS cell-based assays. Traditionally, a range of animal models have been used to test the effi cacy and toxicity of these substances, including mice, rats, rabbits, pigs, dogs, and non-human primates [38]. However, signifi cant international interest from scientists, ethicists, and regulatory bodies is driving the develop ment of replacement in vitro assays [37,39]. For example, the mouse EST has already been validated by the European Centre for the Validation of Alternative Methods as a viable alternative assay for embryotoxicity [9,10]. Th is 7 to 10-day assay makes use of undiff erentiated ES cells, ES cell-derived cardiomyocytes, and NIH 3T3 cells, together with statistical assessment of endpoints that include cell morphology, viability, beating frequency, and more recently gene expression and proteomic analyses.
Application of this assay to a range of known toxicants predicted embryotoxicity with an overall accuracy of 80%, with 100% accuracy for strong embryotoxicants [10,40]. While the overall predictive capacity is lower than the ideal of 100%, it has nevertheless been suffi cient (relatively high true positives, low false positives, and low false negatives) for its use as an animal replacement assay by a number of large international pharmaceutical companies (for example, Pfi zer) [10]. Moreover, its appli cation to compliance testing for the European Commission REACH legislation is predicted to reduce animal usage by ~30% (that is, 1.3 million animals), and thus greatly reduce the cost of this testing while still providing the necessary testing rigour [2].
Th e success of the EST has led to other studies aimed at incorporating a wider array of diff erentiated cell types to enable toxicity assessment for other human tissues. For instance, incorporation of ES cell-derived neural and bone cells is being investigated for the EST [9,41]. Th e drug development opportunities associated with the ability to reproducibly generate large, pure populations of normal (or diseased) diff erentiated human cells are clear. Moreover, as improvements in diff erentiation strategies enable the production of functional human tissue in vitro, the opportunities for drug and toxicology screening will increase. Given that all human organs can experience adverse reactions due to exposure to inappropriate drugs or chemicals, there is much scope for continued develop ment and commercial application of novel human PS cell-based toxicology systems to replace currently used animal models [42].

Xeno-free diff erentiation of human pluripotent stem cells
Establishing simple and effi cient methods for diff eren tia ting human PS cells into desired specialised cell types has become a major focus of human PS cell research. Initial eff orts in this area relied mainly upon adapting methods used for mouse PS cell diff erentiation [41]. While these studies showed that in vitro diff erentiation of human PS cells was possible, the yield and purity of the desired diff erentiated cells were routinely low due to the uncontrolled production of other, undesired cell types. One reason for this indiscriminate diff erentiation was (and still is) the use of poorly characterised animal products to initiate and/or propagate the diff erentiation process (for example, serum, feeder cells, cell substrates, and so forth) [41].
Th ese initial diff erentiation eff orts have evolved into attempts at developing multi-step diff erentiation protocols that aim to mimic in vitro the cell type-specifi c sequence of events that occur during embryogenesis. For example, this directed approach to PS cell diff erentiation has been used to obtain pancreatic cells [43], neurons [44], RPE [25,26], and ocular lens cells [45]. To a greater or lesser extent, these protocols have been partially successful in that they produce in a step-wise manner at least some of the cellular precursors required as well as the fi nal cell type of interest. However, these directeddiff erentiation protocols remain suboptimal due to the simultaneous production of varying numbers of undesired cell types.
Until these methods are optimised, purifi cation strategies are needed to obtain the desired cell type free from undesired contaminating cells. Cell purifi cation is typically achieved via manual excision, fl ow cytometry, and/or reporter plasmids (that is, cDNA constructs containing cell type-specifi c promoters that drive expression of antibiotic-resistance and/or fl uorescence genes). Th ese purifi cation strategies add to the cost, processing time and specialised expertise required to obtain the cell types of interest. Further optimisation of directed-diff eren tiation protocols, by continued replacement of poorlydefi ned animal products with consecutive application of defi ned recombinant growth factors or small molecules, will improve the yield and purity of desired diff erentiated cell types [41]. Th is process of protocol optimisation will probably continue to be iterative, as empirical testing of protocol modifi cations leads to further rounds of cell biology discovery and subsequent protocol optimisation.
While it will be important to further optimise methods for diff erentiating human PS cells into specifi c individual cell types, there is also much interest in using pluripotent cells to generate complex functional tissues in vitro. To date, in vitro tissues that are morphologically and functionally similar to the liver [46], skin [47], retina [48], heart [49], lung [50], and ocular lens [51,52] have been generated from various pluripotent or other cell sources. Th ese eff orts face the challenge of not only producing the correct cell types present in a particular tissue, but also having them form the correct three-dimensional structure required for normal tissue functionality. Critical to these eff orts is the development of appropriate extracellular matrix analogues that can be fabricated into eff ective tissue shapes [50]. Th ese matrices may also need to stimulate localised growth of diff erent cell types (for example, to allow formation of functional blood vessels where necessary within an in vitro tissue). Replacing animal-derived growth factors and extracellular matrices with defi ned recombinant reagents or small molecules will accelerate these eff orts at in vitro tissue production. In turn, this will open up a wide range of new possibilities for cell replacement and drug discovery.

Validation of pluripotent stem cell-derived technology
In the short term to mid-term, animal models of development will continue being investigated to provide clues for the molecular events involved in cell and tissue formation. Additionally, live animal cell-function assays will continue being used to defi ne the functional properties of human PS cell progeny, by assessing whether these cells can properly integrate and correct a disease or injury in vivo [41]. Th is assessment is evidenced by the increasing number of human PS cell derivatives that are being assessed in animal models, including neurons to correct spinal cord injury [53], blood cells for understanding haematopoietic diseases [54], and retinal cells for blindness [55]. In the foreseeable future, many fi elds will continue to require transplantation into animal models to assess diff erentiated cell function, particularly if the diff eren tiated cells are for clinical transplantation.
Given the cellular and molecular diff erences between humans and animals, expanded assay panels for in vitro cell phenotype and function will become increasingly important in reducing the use and cost of animal-based assays while increasing the quality of data obtained [9]. Th e EST for embryotoxicity, new derivative ESTs, and the 4P assays for PS cell-derived RPE are examples of how PS cell-based in vitro assays are beginning to replace in vivo animal models. Key technical factors driving this replacement are an ability to purify large quantities of the test cell type for observation and measurement, an under standing of how the test cell type normally functions in vivo, and defi ned assay endpoints that accurately detect cell phenotype or function. Together, these elements enable comparison of the in vitro cell performance with known in vivo behaviour of the same cell type, thus enabling a determination of whether a proposed PS cell-based assay can be validated as a replacement of a current animal-based assay. Technological improvements are also aiding development of these in vitro animal replacement assays, such as highcontent imaging, novel cell culture substrates, and the ability to produce complex in vitro tissues that mimic performance of whole tissues as opposed to isolated individual cell types [46][47][48][49][50]. Rigorous assessment of these emergent in vitro assays will be critical for ensuring that only assays with demonstrable utility and accuracy gain acceptance. Moreover, development of these animal replacement assays will be aided by consultation with end users (for example, pharmaceutical companies, policy advisory bodies, and so forth), who will ultimately determine whether such assays provide suffi cient fi nancial and/or accuracy advantages over existing animalbased assays to warrant widespread use [39].

Clinical application of human pluripotent stem cell-derived therapies
One of the most widely anticipated applications of human PS cell technology is the development of new cell replacement therapies. Early concerns about the clinical translation of human ES cell derivatives revolved around the potential transfer of pathogens and immunogens from animal products such as the mouse embryonic fi broblasts traditionally used for pluripotent cell maintenance [56]. For example, human ES cells maintained using mouse embryonic fi broblasts and/or animalderived serum replacements were shown to incorporate an immunogenic nonhuman sialic acid (Neu5Gc) capable of inducing an immune response in vitro [56]. Although other studies questioned the extent of Neu5Gc contamination in human ES cell cultures and its immunogenic potential [57][58][59], signifi cant eff orts continue to be made to enable animal product-free culture of human pluripotent cells and their derivatives. As mentioned above, this has led to the development of a variety of animal product replacements including human feeder cells [15], defi ned media [7], and chemical or recombinant extracellular matrices [17].
Th ree clinical trials have to date been initiated using purifi ed human ES cell derivatives: oligodendrocytes for spinal cord injury [11], and RPE for age-related macular degeneration and Stargardt's macular dystrophy [12]. Th e phase I oligodendrocyte trial recently ceased taking new enrolments due to Geron's controversial strategic fi nancial decision to move out of the stem cell fi eld in favour of developing in-house anti-cancer drugs [60]. However, new enrolments in the two RPE trials are set to continue on the basis of encouraging preliminary safety data from a single patient in each trial [12]. As these trials progress, continued human PS cell research aimed at further eliminating animal product will both increase the effi ciency of diff erentiation protocols and limit the potential adverse clinical consequences of animal product use. In turn, these studies will aid the clinical translation of a wider variety of human PS cell-derived cell types.

Conclusion
Th e potential for human PS cells to increase our understanding of human development and to provide diff erentiated cells for drug screening, toxicology studies, and cell replacement therapies is widely acknowledged. Signifi cant technical hurdles must still be overcome, but realisation of this potential is increasingly becoming reality. A small number of commercially produced diff erentiated human PS cell types are already in use for drug development or clinical applications, and it is reasonable to expect that this will increase as a greater variety of human PS cell-derived cells and tissues become available over the coming years. Owing to the experimental variation, cost, and clinical barriers that arise through the use of animals and animal products, continued development of human PS cell technology will signifi cantly benefi t from refi nement, reduction, and/or replacement of these animal-based methods. Th e goals of many human PS cell researchers are therefore highly consistent with the 3R principle for the ethical use of animals in research. Moreover, optimisation of human PS cell technology based on the 3R principle will increasingly provide clear commercial opportunities for the bio technology and clinical sectors, including: development of chemically defi ned reagents for human PS cell generation, maintenance, and diff erentiation; improved cell detection and purifi cation strategies; and purifi ed cell types for transplantation and in vitro drug discovery and toxicity assays.

Competing interests
MDO has previously received research funding from the Medical Advances Without Animals Trust.