|Home | About | Journals | Submit | Contact Us | Français|
The transplantation of insulin producing β–cells derived from human embryonic stem cells and induced pluripotent stem cells (collectively termed pluripotent stem cells; PSCs) holds great promise for therapy of diabetes mellitus. The purpose of this review is to summarize recent advances in this area, emphasizing the importance of studies of endocrine pancreas development in efforts to direct PSC differentiation into endocrine cells, as well as to outline the major challenges remaining before transplantation of PSC-derived β-cells can become a reality.
Although several protocols to generate glucose-responsive pancreatic β–cells in vitro have been described, the most successful approaches are those that most closely mimic embryonic development of the endocrine pancreas. Until recently, cells generated by these methods have exhibited immature pancreatic endocrine phenotypes. However, protocols that generate more functional β–like cells have now been described. In addition, small molecules are being used to improve protocols to direct differentiation of PSCs into endoderm and pancreatic lineages.
Advances over the past decade suggest that generating function β-cells from human PSCs is achievable. However there are aspects of β-cell development that are not well understood that are hampering generation of PSC-derived β-cells. In particular the signaling pathways that instruct endocrine progenitor cells to differentiate into mature and functional β-cells are poorly understood. Other significant obstacles remain including the need for safe and cost-effective differentiation methods for large-scale generation of transplantation quality β–cells, methods to prevent immune rejection of grafted tissues, and amelioration of the risks of tumorigenesis.
Diabetes mellitus is a major worldwide health crisis, affecting more than 200 million people. In the US, nearly 25 million people have been diagnosed with diabetes, and it is predicted that this figure will increase to nearly 50 million by the year 2050 . Diabetes is characterized by aberrant blood glucose regulation resulting in prolonged hyperglycemia associated with significant long-term health complications. Type 1 diabetes results from autoimmune destruction of the insulin-producing β–cells of the islets of Langerhans; whereas type 2 diabetes is characterized by peripheral insulin resistance and the inability to produce enough insulin to overcome this resistance. Less common forms of diabetes such as gestational diabetes, maturity onset diabetes of the young, neonatal diabetes mellitus and loss of islets in pancreatitis are also associated with impaired insulin production.
At the core of current type 1 diabetes treatment is injection of exogenous insulin, which provides some level of control over blood glucose levels and has significantly reduced diabetes morbidity . Even in the best cases, glucose monitoring and insulin injections cannot fully compensate for loss of β–cells. While technological advances in glucose monitors and insulin pumps will likely improve efforts to deliver exogenous insulin , a more physiological solution is the replacement of β–cells by whole pancreas or islet transplantation . Transplantation-based approaches such as the Edmonton protocol have proven effective , however the widespread use of pancreas or islet transplantation is impossible due to a chronic shortage of donors. Consequently, human pluripotent stem cells  are a promising, renewable source of cells from which β–cells can be derived for replacement therapy (FIGURE 1) [7, 8].
In this review we will summarize the recent progress in generation of β-cells from human PSCs and outline some of the major practical challenges remaining before the widespread clinical use of PSC-derived β–cells can become a reality.
Pluripotent stem cells exhibit stable self-renewal in culture and have the potential to differentiate into all somatic cell types, including pancreatic β-cells. The challenge in directed differentiation of PSCs into β-cells is to specifically generate only one of the ~210 somatic cell types. In the past decade, numerous approaches to generate insulin-producing β–cells from embryonic stem cells have been reported. Initial attempts using mouse  and human embryonic stem cells (hESCs) [10, 11] included spontaneous in vitro differentiation following selection of cells positive for nestin  or coupled with ectopic expression of transcription factors known to be important for β–cell development in vivo [13-15]. These early studies often resulted in some insulin positive cells, but there was little evidence that these were bona-fide, functional endoderm-derived pancreatic β-cells. The limited success using these protocols prompted a more physiologically-based approach that utilized signaling pathways that are required during embryonic β-cell development in vivo [16-19]. The more successful of these efforts have directed hESC differentiation in a stepwise fashion that recapitulates all the major stages of β–cell development [20-23], and have resulted in the production of definitive endoderm-derived, mature, glucose responsive β–cells (FIGURE 1). The developmental basis of this approach is described below and reviewed in .
β-cell development can grossly be broken down into four steps: endoderm formation, pancreas specification, endocrine specification and beta-cell maturation [24-26] (FIGURE 1). Differentiation into definitive endoderm (DE) is the first obligatory step in generating pancreatic endocrine cells from PSCs. DE is generated in vivo by the process of gastrulation, where naïve cells are instructed to form the three primary germ layers: the ectoderm, mesoderm and endoderm. The molecular control of endoderm formation is highly conserved across vertebrate species and involves the Nodal signaling pathway. Nodal is a TGFβ ligand whose activity initiates a series of downstream signaling events that culminates in the activation of an evolutionarily conserved transcriptional network that regulates DE development (reviewed in ). Protocols to differentiate hESCs into DE utilize the nodal related protein activin, which is commercially available, is highly bioactive, and mimics nodal activity [28-31].
Definitive endoderm gives rise to a diverse array of cells and tissues that contribute to vital organs including the pancreas, liver, lungs, stomach and the epithelial lining of the alimentary tract. Remarkable progress has been made in defining the embryonic processes that direct a subset of endodermal cells into the pancreatic progenitors that arise from the posterior embryonic foregut. Signaling pathways that direct regionalization of the foregut in mouse involves FGF, BMP and retinoic acid (RA) signaling [32-35]. FGF2 and activin mediate suppression of sonic hedgehog signaling in the posterior foregut, which is required for initiation of pancreas gene expression . Efficient differentiation of hESC-derived DE into the pancreatic lineage has been accomplished by temporally manipulating the FGF, BMP, RA and hedgehog-signaling pathways, thereby directing DE cells first into a foregut (HNF1β, HNF4) and then into a pancreatic fate (Pdx1, Nkx6.1, Hnf6, Sox9)(FIGURE 1) [21, 22].
The next stages of pancreas development involve the proliferation of pancreatic progenitors and their segregation into either exocrine or endocrine cell types, required for digestion and glucose homeostasis, respectively. Expansion of pancreatic progenitor cells involves FGF10 signaling  and the decision to become an endocrine or exocrine cell requires the Notch signaling pathway, and endocrine progenitor specification is marked by the expression of the transcription factor NGN3 [38, 39]. The expansion and organization of endocrine cells into islets in vivo involves EGF signaling . In successfully generating and expanding pancreatic endocrine cells from hESCs, FGF ligands (FGF7 and 10) [21, 22] were used to expand pancreatic progenitors, inhibition of notch signaling was used to generate pancreatic endocrine progenitor cells , and EGF was used to expand endocrine progenitors .
One of the final stages of pancreas development involves maturation of endocrine progenitors to mature hormone-producing cells. Engrafting hESC-derived pancreatic endocrine cells into mice promotes their maturation into β–like cells , demonstrating that these cells have the intrinsic ability to form insulin-expressing cells with some functionality. However, in vitro attempts to differentiate pancreatic endocrine cells into glucose-responsive, insulin-secreting cells have had mixed success, in part due to our lack of understanding the signaling pathways that direct β–cell maturation in vivo. Most of the focus has been placed on signals that are known to affect postnatal β–cell function such as the incretin signaling pathway mediated by GLP1. Activation of this pathway with Extendin 4, which mimics the activity of GLP1, results in an increase in the formation of insulin-positive cells in vitro . However these cells are largely fetal/neonatal in character and do not appear to be bona fide β–cells. It has recently been shown that islet innervation  correlates with the maturation of fetal β–cells. Moreover, β–cells express the receptor for nerve growth factor (NGF)  and NGF promotes the maturation of fetal β–cells . This serves to highlight why a better understanding of fetal β–cell maturation in vivo will continue to guide efforts to direct maturation of PSCs into β–cells in vitro.
Although the concepts and framework for the directed differentiation of β–cells from pluripotent human stem cells are well developed, significant hurdles remain before this approach can be widely used for diabetes therapy. In the following section we will discuss some of these remaining challenges and outline potential solutions to some of these concerns.
Current protocols rely on the use of up to 10 recombinant proteins (cytokines and inhibitors) for directed differentiation of PSCs [21, 23]. Given the manufacturing cost of generating high quality recombinant proteins with consistent biological activity, this approach is prohibitive to generate sufficient numbers of cells for human transplantation . As such, a focus has been shifted to the possible use of small molecules for large-scale, reproducible, directed PSC differentiation under good manufacturing practice (GMP) conditions.
Recent progress has been made with the discovery of small molecules that can modulate various stages of β-cell differentiation from PSCs. Borowiak et al. (2009) screened a chemical library and identified compounds that would mimic activin A and induce an endoderm-specific Sox17 promoter in mouse ESCs . The two compounds (IDE1 and IDE2) induced mouse and human ESC differentiation into definitive endoderm cells. IDE1 and IDE2 were shown to induce SMAD2 phosphorylation, indicating that these compounds promote DE formation via the nodal/activin pathway. Furthermore, DE generated using IDE1 and IDE2 initiated pancreatic differentiation in response to RA, FGF10 and cyclopamine (an inhibitor or hedgehog signaling). Another screen of chemical libraries identified Indolactam V, which promoted pancreatic progenitor formation from DE .
These findings demonstrate that the availability of chemically diverse small molecule libraries and robust biological assays can lead to the discovery of compounds that can either replace recombinant proteins or enhance their activity  in the directed differentiation of PSCs to β–cells. The use of chemical compounds instead of recombinant proteins will facilitate development of GMP protocols for generating therapeutic quality β–cells.
Of the safety concerns associated with transplantation of pluripotent stem cell-derived tissues (reviewed by ), one main issue is tumorigenicity. Teratomas, which are benign growths containing tissue types from all three germ layers, have been reported in numerous PSC-derived cell transplantation studies. Studies using hESC-derived insulin-producing cells in some cases reported teratomas  where as others observed no tumors . These teratomas are largely thought to derive from undifferentiated hESCs that persist in the differentiated cultures, and cell-sorting based approaches to remove hESCs have helped to reduce the incidence of teratoma formation . Additional approaches to eliminate undifferentiated cells include the use of molecules or antibody-toxin conjugates that selectively kill undifferentiated PSCs. In all cases, PSC-derived and sorted cells will need to be tested for residual tumor generating cells. The current approach, teratoma formation assay in immunocompromised mice, may not be sensitive enough to detect low level PSC contamination and may not recapitulate all the unique biological contexts of different transplantation sites (reviewed in ). Given these limitations, the development of additional in vitro and in vivo techniques that detect residual undifferentiated cells will be critical. While there is little evidence for malignant cell types arising from PSCs, this is something that needs to be fully investigated using animal models.
The ideal β–cell transplantation site would be one that supports the long-term function and survival of grafted cells and is easily accessible for maximal patient safety (for review see ). Current clinical practice is to transplant islets into the liver via the portal vein, with the rationale that the majority of insulin released from the pancreas is utilized in the liver and the easy accessibility of this site by a minimally invasive procedure. However, half of the beta-cells die shortly after transplantation , and this is thought to be due to low oxygen tension, an active immune response, and high levels of toxins and drugs in the liver. In addition, the instant blood mediated inflammatory reaction (IBMIR) encapsulates transplanted islets in a fibrin clot and enhances the immune reaction against the graft . Therefore, several alternative sites for islet transplantation have been tested including the kidney capsule, omentum, and subcutaneous, which may be best for patient safety but not ideal from a functional perspective due to systemic release of insulin . An improved understanding of the optimal anatomical sites for islet transplant is important for enhancing the survival and function of islet or PSC-derived β-cells following transplantation.
Despite reports that hESCs and their differentiated progeny may be non-immunogenic [54, 55], recent studies have documented the development of immune responses against transplanted hESC-derived tissues in immune competent mice [56, 57]. Therefore, patients receiving allogeneic hESC-derived β–cells will require life-long immunosuppressive therapy to prevent graft rejection. Additionally, almost all commonly used immunosuppressive drugs are deleterious to β–cell function, replication and survival . Furthermore, although the evolution of induced pluripotent stem cell (iPSC) technology is a promising strategy for generating autologous, patient-specific β–cells (as discussed below), even autologous transplantation for type I diabetes will require the suppression of the preexisting autoimmunity. One promising treatment approach for combating autoimmunity is the use of targeted immunotherapies such as CD3-specific antibodies for induction of long-term tolerance to auto-antigens .
In the case of transplant of allogeneic PSC-derived β–cells, several strategies to minimize or eliminate the requirement for immunosuppression are under investigation, including microencapsulation techniques to protect grafted tissues from the immune system (reviewed by ), and transplantation of cells into immune privileged sites such as the anterior chamber of the eye or the testis. However, recent studies have demonstrated rejection of islets transplanted in the testis of immunocompetent mice [61, 62], indicating that this approach may still require immunosuppression. Furthermore, it is unclear whether these locations will meet the requirements for β–cell transplantation sites discussed above.
Another strategy would be to generate large banks of hESCs for the purpose of matching their HLA-phenotype to recipient transplant patients. This would require generating hundreds to potentially thousands of discrete HLA-typed hESC lines to ensure adequate matching for most individuals [63, 64]. This approach would require enormous regulatory oversight and stringent quality control analyses to test for pluripotency and that each line is competent to generate functional β–cells before it could be widely used for cell replacement therapies .
The discovery that human somatic cells can be reprogrammed into an embryonic stem cell-like state that appears to be phenotypically and functionally equivalent to hESCs represents a seminal moment for regenerative medicine [66, 67]. Induced pluripotent stem cells (iPSCs) may be an ideal source for cell replacement therapies because they can be derived from patients, including those with diabetes [68, 69], and eventually used to generate autologous therapeutic β-cells for transplantation (FIGURE 1). In fact two recent studies describe the generation of insulin producing cells from human iPSCs, supporting the feasibility of this approach to generate patient-specific β-cells for autologous transplantation [23, 69].
Original protocols for human iPSC generation employed the retroviral or lentiviral-mediated expression of four transcription factors involved in stem cell pluripotency [66, 67]. This approach is not suitable for generation of therapeutic quality cells because of risks from insertional mutagenesis and the use of the oncogene cMyc, which caused tumorigenesis in chimeric mice derived from these cells [66, 67]. Subsequent studies have reported iPSC generation using non-integrating methods of gene delivery such as plasmid transfection , episomal plasmid transfection , the piggyBac transposon system , adenoviral transduction [73, 74], Cre-excisable viral vectors [75, 76], and most recently by direct introduction of membrane soluble reprogramming factor proteins [77, 78]. With proper, rigorous quality controls  and the rapid advances in iPSC technology it is likely that iPSCs will factor heavily into future efforts for cell replacement therapy for diabetes.
There has been significant progress in our understanding of normal pancreas development and this information has been central to recent successes at directing the differentiation of PSCs to β-cells in vitro . Further research, particularly into mechanisms of β-cell maturation in vivo, will likely prove essential for generation of glucose-responsive, mature β-cells suitable for cell replacement therapy for diabetes. Continued advances in establishing GMP-compliant protocols, enhanced transplantation techniques, and methods for regulation of the immune response will also factor into the success of these endeavors. Moreover, rapid progress in induced pluripotent stem cell technology is likely to prove transformational in future efforts in generating patient-specific β-cells.
We apologize to our colleagues whose work we could not include due to space constraints. We are grateful to the Wells and Zorn labs for discussions, and to Jason Spence and Suh-Chin Lin for comments on the manuscript. JMW and CNM are supported by grants from the NIH (NIGMS and NIDDK) and Juvenile Diabetes Research Foundation (JDRF).
The authors have no financial conflicts.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.