Stem Cell-Derived Pancreatic Beta Cells for the Study and Treatment of Diabetes

Abstract

Patients suffering from Type 1 Diabetes rely on the exogenous supply of insulin. Cell replacement therapy employing cadaveric islets cells has demonstrated a proof of principle for a practical cure, rendering patients insulin independent for prolonged periods of time. However, challenges remain before this innovative therapy can be widely accessed by diabetic patients. Availability of cadaveric donor islets is limited, necessitating the generation of an abundant source of insulin-producing pancreatic beta cells. Immunological rejection of the allogeneic transplant and recurring autoreactivity contribute to eventual graft failure in all transplant recipients. In the current chapter, we summarize past and current efforts to generate functional beta cells from pluripotent stem cells and highlight current knowledge on graft immune interactions. We further discuss remaining challenges of current cell replacement efforts and highlight potentially innovative approaches to aid current strategies.

1. Introduction

1.1 The beta cell and type 1 diabetes

Global regulation of the essential metabolite, glucose, relies on the proper action of small clusters of endocrine cells within the pancreas known as the islet of Langerhans. Mostly comprised of alpha, beta, delta, gamma, and epsilon cells, these endocrine populations release a tightly synchronized set of hormones to maintain systemic glycemic control [1, 2]. Most importantly beta cells respond to increased glucose levels by secreting the hormone insulin, which triggers glucose uptake by peripheral tissues such as the muscle and adipose tissue [3]. To counter regulate the beta cell’s suppressive impacts on blood glucose levels, alpha cells produce the hormone glucagon, which acts on the liver to increase glucose production and raise glucose concentrations [4]. While less studied, literature suggests that somatostatin-producing delta cells, pancreatic polypeptide-producing gamma cells, and ghrelin-producing epsilon cells, also have counterregulatory impacts on hormone secretion from the beta and alpha cells and help provide satiety signals to modulate gastric processes [5, 6, 7]. When this delicate balance of hormonal signals is dysregulated, particularly when the release of insulin from beta cells is impaired, systemic glucose homeostasis is perturbed. The loss of functional insulin-producing beta cells results in chronic hyperglycemia, a key hallmark of the disease state known as diabetes mellitus.

Type 1 diabetes (T1D) is distinguished from other types of diabetes mellitus by the involvement of autoreactive immune cells in the destruction of the pancreatic beta cells. While T1D only accounts for 5–15% of total clinical cases of diabetes, incidence rates are steadily increasing worldwide [8]. Despite this concerning trend, the field currently lacks the ability to identify future T1D patients early in disease progression before glucose homeostasis is affected. Clinical diagnosis usually presents with severely elevated peripheral glucose measurements combined with positive titers for autoantibodies against known beta cell antigens such as insulin, glutamic acid decarboxylase-65 (GAD-65), islet antigen-2 (I-A2), or zinc transporter 8 (ZnT8) [9]. At the time of diagnosis, it is estimated that around 60–90% of the total beta cell mass has already been lost and circulating insulin levels are severely reduced [10]. Therefore, once identified, patients are dependent on exogenous insulin injections to maintain some glycemic control. Insulin can be administered through various technologies including insulin pens and closed-loop insulin systems or bi-hormonal pumps. However, the mental and financial burden of these therapies can be extremely costly to patients [11, 12]. While these innovative delivery approaches have significantly improved the precision of exogenous insulin, they fail to reach the exquisite responses provided by native pancreatic beta cells to the most subtle changes in systemic glucose levels. Therefore over time, severe secondary complications including nephropathy, neuropathy, cardiovascular disease, and retinopathy are likely [13]. In addition to potentially devastating long term affects experienced by patients with T1D, the constant risk of life-threatening hypoglycemic episodes, induced by administering too much insulin, is a considerable burden on quality of life. While not perfect, in most patients with T1D, exogenous insulin therapies are beneficial, but alternative approaches that better recapitulate native beta cell function are needed to significantly improve the health and welfare for patients with T1D.

1.2 Beta cell replacement

A promising treatment strategy to more faithfully restore euglycemia for patients with T1D is beta cell replacement therapy. The first true success was documented in 2000, when a group of investigators from the University of Alberta published their findings from a small clinical trial following seven patients undergoing cadaveric islet transplantation [14]. This protocol, which would come to be known as the Edmonton Protocol, succeeded in maintaining insulin independence within this small cohort of individuals for nearly 5 months after islet infusion into the portal vein of the liver and systemic nonsteroid immune suppression. Clinically, this procedure is nowadays most usually performed, although whole pancreas transplants are also performed. Islet transplantation is especially beneficial for individuals that are poorly controlled by or are unresponsive to traditional insulin therapies [15]. Islet transplantation awards significant benefits to patients with T1D, most notably protection from hypoglycemic events, complete or some independence from insulin injections, and overall improved quality of life, however donor availability and immunological challenges persist that limit widespread implementation of this therapy [16].

After utilization of the Edmonton Protocol at several prominent trial centers, the field of islet transplantation made significant strides as the various mechanisms that can contribute to islet graft rejection became better understood. One of the key hurdles thought to contribute to the ultimate failure of transplanted islets is immune-mediated rejection. This process begins within minutes after transplantation and involves islet-derived stress responses that quickly recruit and activate innate immune subsets [17, 18]. This acute reaction contributes significantly to early graft loss immediately after transplant but can also serve to activate the adaptive immune system.

During clinical islet transplantation into patients with T1D, unlike other organ replacement surgeries such liver transplants, the islet graft is at risk from two separate adaptive immune reactions: (1) genetic mismatch responses and (2) autoreactive responses against islet antigens. As transplanted islets are from a genetically distinct donor, the immune system within the recipient will mount a response against the islet graft in an allogeneic (or HLA-mismatch) context. This immunological response has been associated with failure of the islet graft, as detection of alloreactive T cells and humoral responses precedes the loss of graft function [19]. The use of both induction and maintenance immunosuppression regimens, such as anti-thymocyte globulin (ATG), are commonly utilized to combat alloreactive immune responses effectively [20]. However, the use of these immunosuppressive drugs can leave a patient at risk for opportunistic infections, often requiring the temporary cessation of immunosuppression to clear these infections, thus leaving the islet graft unprotected at times [19].

Additionally, unlike other transplantation contexts, patients with T1D have autoreactive immune responses to beta cell antigens. After the initial destruction of endogenous beta cells within the pancreas during the time of disease onset, a subset of autoreactive T and B cells is converted into long-lived memory cells [21]. These memory populations can persist within the patient for decades and can become reactivated after islet transplantation, when a new source of beta cell antigen is introduced. The recurrence of autoimmunity toward the islet graft has been correlated with reduced graft survival, even in the absence of detectable allogeneic responses [22, 23]. Commonly utilized immunosuppressive drug regimens are not effective in combating autoimmunity and may even induce proliferation of autoreactive memory T cells during islet transplantation [23, 24]. Therefore, to achieve successful beta cell replacement, immunosuppressive treatment strategies must not only suppress allograft rejection, but also prevent the reactivation of T-cell mediated autoimmunity.

As the study and understanding of these immune reactions has been expanded, recent clinical islet transplant studies can now achieve rates of insulin independence in upwards of 44% of patients 3 years post-transplantation with >90% of patients seeing complete protection from severe hypoglycemic events beyond 5 years post-transplantation [25]. Some of the key factors that have contributed to improved islet transplantation success include the processing of the islet preparations [26], islet culture and manipulation prior to transplantation [27], the acute treatment of the patient at the time of transplantation [20, 28], and the maintenance immunosuppression protocols utilized long term [16]. However, nearly all islet transplants eventually fail to maintain glycemic control, and this treatment strategy still suffers from a lack of abundant, high-quality cadaveric islets to treat all individuals with T1D using this approach. Therefore, alternative sources of functional insulin-producing beta cells and novel immunosuppressive approaches are needed to allow a widespread, effective implementation of cell replacement therapy.

1.3 Stem cells for regenerative medicine approaches

One of the most promising alternatives to limited cadaveric human islets for transplantation is to generate large pools of glucose responsive, insulin-producing cells from human stem cell sources (Figure 1). Stem cells are separated into five general categories according to the extent of their ability to differentiate into other cell types: totipotent, pluripotent, multipotent, oligopotent, and unipotent [29]. Totipotent cells can generate an entire organism from a single cell, encompassing all cell types found in the body, as well as extraembryonic tissues [30]. Pluripotent cells can differentiate into all cells of the three germ layers, endoderm, mesoderm, and ectoderm. Multipotent cells include all progenitor type cells that can differentiate into a limited number of cell types that are derived from that progenitor population, while oligopotent cells can generate cells of only one specific differentiation fate. Finally, unipotent cells are proliferating cells that can only produce more of the same cell type.

Figure 1.

Generating functional beta cells from human stem cells. Through a series of sequential stages, human pluripotent stem cells (hPSC) are directed differentiated into functional stem cell-derived beta-like cells (sBC). sBC can undergo glucose-stimulated insulin release. GLUT1; glucose transporter 1, OxPhos; oxidative phosphorylation, K_ATP; ATP-dependent potassium channels, Ca²⁺; calcium. Figure generated using Biorender.

Human embryonic stem cells (hESC), a form of pluripotent stem cells (PSC) which are derived from the inner cell mass of in vitro fertilized embryos at the blastocyst stage, were first described in 1998 [31]. hESC can rapidly and indefinitely proliferate and differentiate into all cells of the human body, given appropriate signals are provided, thereby representing an attractive, quasi unlimited source of human cells that could be utilized for a multitude of regenerative medicine approaches [32]. To date, hundreds of hESC lines have been generated from donated, otherwise discarded blastocysts. Nevertheless, the use of human blastocysts to generate hESC has raised ethical concerns regarding their utilization in many countries.

The discovery of induced pluripotent stem cells (iPSC) generated from easily accessible patient tissues or cells such as skin, peripheral blood, and urine not only lessened the ethical concerns surrounding human PSC but also allowed for increased genetic diversity by allowing the generation of patient specific iPSC [33, 34, 35]. The first successful attempt to reprogram somatic fibroblasts into iPSC utilized retroviral transduction of pluripotency genes Oct4, Sox2, Klf4, and c-myc into murine embryonic fibroblasts or tail-tip fibroblasts [36]. Similar approaches were utilized to reprogram human fibroblasts isolated from skin into human iPSC. Importantly, established iPSC display comparable features including differentiation potential to hESCs [33]. Collectively, either ESCs or iPSCs, represent a potentially unlimited supply of any desired cell type for cell replacement therapy, and therefore could address some of the challenges facing beta cell replacement strategies for the treatment of T1D.

2. Differentiation of stem cell-derived beta cells

To differentiate PSC into a pure population of beta cells, the delicate and time-specific process of pancreatic organogenesis must be closely replicated in vitro. The first attempts to generate insulin positive cells from PSC resulted in loosely defined pancreatic cell populations with some endocrine cells that were usually poly-hormonal and lacked the ability to secrete insulin in response to glucose [37]. Indeed, such differentiated cultures contained a fairly high number of non-endocrine populations such as pancreatic progenitors and exocrine cell types [38]. However, transplantation of these heterogenous pancreatic cells into immunodeficient mice promoted their differentiation into single hormone, insulin positive beta like cells after a few months. In vivo generated stem cell derived beta cells (sBC) are functional and secrete insulin in response to elevated glucose levels, providing the first proof of principle demonstration that PSC can differentiate into functional beta cells [39]. Furthermore, after chemical ablation of endogenous mouse beta cells using streptozotocin, human grafts containing functional sBC could maintain euglycemia, suggesting an important therapeutic potential to restore glycemic control [40, 41]. These early experiments provided evidence that directed differentiation of hPSC can produce functional beta cells, encouraging the research community to devise strategies to build upon earlier protocols with the goal of generating functional sBC in vitro [41, 42, 43, 44]. Using these protocols, investigators have now better defined the key signals required to induce hPSCs commit to the appropriate differentiation trajectories in vitro that are essential to generate hormone positive endocrine cells.

2.1 From stem cell to endocrine progenitors

The first major fate decision needed to begin the process of directed differentiation is the induction of the endoderm lineage. WNT/β-catenin and the transforming growth factor beta (TGF-beta)/Nodal signaling pathways are the main drivers of endoderm differentiation [45]. Activin A, a member of a TGF-beta superfamily, is used in vitro to activate Nodal signaling. To initiate WNT signaling, CHIR-99021, an inhibitor of the suppressive kinase glycogen synthase kinase-3 beta (GSK3-beta), is nowadays commonly used. While initially researchers employed recombinant proteins to elicit certain signaling pathway modulation, small molecules provide more reproducible results and are far more cost effective. Overall, providing these signals in low serum media to hPSC induces the robust differentiation to endoderm, with upwards of 97% of cells upregulating key endoderm lineage markers [37].

Next, further differentiation to consecutively become primitive gut tube then posterior foregut are required. Fibroblast growth factor (FGF7) is frequently used to induce the differentiation of definitive endoderm to a primitive gut tube stage. In some instances, certain iPSC lines require additional low-grade BMP inhibition [46]. This differentiation step is followed by a combination of retinoic acid, Sant1 (a SHH signaling inhibitor), PMA (a PKC activator), and LDN (a BMP inhibitor) to further direct gut tube like cells toward both posterior foregut and then pancreatic progenitor stages [47]. Some beta cell differentiation protocols rely on the expansion of pancreatic progenitors before continuing with the induction of endocrine differentiation at this stage. For this purpose, pancreatic progenitors are cultured in the presence of epidermal growth factor (EGF) and FGF7 to establish a large pool of cells prior to endocrine induction [37, 39, 41, 42, 43, 47]. However, this amplification step is not necessarily required and extends the culture time required before reaching target beta cells considerably.

2.2 From pancreatic progenitor to mature beta cells

Current protocols induce endocrine cell differentiation from pancreatic progenitors by exposing cells to a cocktail of TGF-beta inhibition (ALK5i), thyroid hormone (T3), BMP inhibition (LDN), and NOTCH inhibition through gamma-secretase inhibition (XXi) [40, 41, 47, 48, 49]. This combination results in short lived expression of the master endocrine regulator gene NEUROG3 and activation of expression of downstream endocrine fate markers NKX2.1 and NEUROD1 [50]. Endocrine differentiation results in the generation of hormone expressing cells including sBCs of varying proportions based on the protocols utilized. While the resulting cell populations are heterogenous, the majority (greater than 90%) usually exhibit an early endocrine, non-proliferative phenotype, with sBCs representing commonly between 20 and 60% of all cells. Other hormone expressing cells are present at variable ratios and off target enterochromaffin cells, usually confined to the gut, are also detected at appreciable percentages (~10–15%), indicating that further protocol optimization might be required. Initially differentiated sBCs exhibit a rather immature phenotype [51] but further mature in vitro over the course of 1- several weeks to give rise to functional sBCs that exhibit several key features of primary human beta cells, including hormone and transcription factor expression, insulin granule formation and maturation, and elevated insulin secretion when exposed to elevated glucose concentrations [40, 41, 47]. Importantly, sBCs also display functionality shortly after transplantation, a distinct benefit when compared with the longer time period required for in vivo differentiation of transplanted pancreatic progenitors into sBC [52].

The WNT pathway as well as TGF-beta signaling have established roles in beta cell development and maturation, but these roles may be time dependent [48, 53, 54]. During beta cell development, inhibition of the TGF-beta receptor ALK5 is important for endocrine differentiation from pancreatic progenitors [48]. However during sBC maturation, inhibition or suppression of TGF-beta signaling results in sBCs with diminished insulin release [48]. Additionally, there is some evidence that canonical vs. non-canonical WNT signaling may influence the ability to induce optimal differentiation into functional sBCs. Different subpopulations of in vitro stem cell-derived definitive endoderm cells that rely on WNT signaling driven by either the canonical WNT3A (CD275⁺ cells) or the non-canonical WNT45A and WNT4 (CD177⁺ cells) have altered capacity to specify toward the pancreatic lineage and therefore differed capacity to produce functional sBCs [55]. In this report, the non-canonical WNT signaling-driven DE population of CD177⁺ cells had greater capacity to produce glucose responsive sBCs, suggesting that canonical WNT signaling may impair the commitment to the pancreatic cell fate [55]. Alternatively, other investigators have data indicating that the primary reason for the observed functional impairment in sBCs is a disruption in glycolytic processes [56]. This study suggested that GAPDH and PGK activity is altered in sBCs, resulting in decreased metabolite entry into the TCA cycle [56].

However, external signals also contribute to beta cell maturation. During mouse pancreatic organogenesis, newly derived endocrine cells including beta cells migrate and cluster together into the mini organ islet of Langerhans harboring the endocrine cells [57]. A similar phenomenon has also been observed during sBC maturation [58]. Insulin positive cells will group together within the heterogenous 3D cell cluster in a process referred to as “capping”, resulting in improved function and increased expression of maturation markers MAFA and UCN3 [58]. Other laboratories have also confirmed that sorted insulin positive cells from later stages of a differentiation protocol will reaggregate together, leading to sBC with a more mature phenotype [38, 59]. These studies further highlight the importance of glucose metabolism and sBC architecture in functional maturation.

The timing and location of developmental signals are highly important aspects for pancreas organogenesis. Interactions with different nonendocrine cell populations such as endothelial cells or mesenchymal cells and the integration of both vascular and neuronal systems is also required to develop a proper functional pancreas, including the endocrine islets of Langerhans. Additionally, other non-beta cell endocrine populations within the islet have displayed the ability to more accurately fine tune the responsiveness of beta cells. Through release of molecules such as glucagon, glucagon-like peptide-1 (GLP-1), and acetylcholine, the alpha cell can enhance the secretion of insulin above that of glucose alone [60]. Somatostatin-producing delta cells also contribute to the regulation of beta cell function, as somatostatin is an inhibitor of both glucagon and insulin release [61]. A current potential barrier to efficient and consistent generation of hPSC-derived beta cells is that fully mature supporting cell types are largely absent form current direct differentiation approaches. Ongoing and future studies aim at incorporating multiple cell types to recapitulate in vivo developmental cell-cell interactions more closely under cell culture conditions. However, if the field lacks the ability to make fully mature beta cells, the real impact of the presence or absence of other cell types might be hard to assess.

Another potential confounding factor is that the current field’s knowledge on pancreas organogenesis is largely derived from model systems, most notably frog, zebrafish, and mouse. Despite many key similarities between model systems and humans, a multitude of both structural and physiological differences between human and rodent islets have been described, leading to the recognition of vital differences in the complex dynamics of cell-cell interactions and the resultant insulin responsiveness of species specific beta cells [62]. This means that not all aspects of animal studies can or should be utilized as a reliable source of information that can be harnessed for the study of human physiology and associated model systems. To improve current protocols, further efforts should be focused on developing a comprehensive map of human beta cell development. This might be accomplished by combining an in-depth analysis of donated fetal pancreas tissues with an unbiased single cell or population analysis of both fetal and adult human samples. To accomplish this goal, PSC could represent a powerful tool to study human development in a controlled manner by providing access to otherwise inaccessible developmental timepoints. The hPSC differentiation model could provide key insights at later developmental stages since fetal pancreatic tissues past 20 weeks conception is not available. This should be done with caution, however, as many of the differentiation protocols used are aimed at driving the predominant phenotype of the beta cell, which could mask some of the delicate signaling balances that are surely present during organogenesis. More importantly, understanding the developmental signals and cues required to generate pancreatic cell types and potentially large pools of beta cells will accelerate our ability to treat human patients with type 1 diabetes.

3. Challenges to utilizing stem cell-derived beta cells for human disease

Once further optimized, generating large numbers of genetically identical and functional human beta cells in the laboratory will allow researchers to address the challenges facing cell replacement therapies in a reproducible manner. This possibility is improved by the rapidly evolving field of genome engineering that now allows for the establishment of novel human model systems that were previously largely restricted to animal models [63]. Since the first description of sBCs, researchers have proposed using these cells to address various aspects of beta cell replacement such as the ideal location to transplant, protecting grafts from ischemia during the peri-transplant period, shielding grafts from allogeneic and autoimmune recognition, and providing additional safety mechanisms to mitigate any off-target effects.

3.1 Generating immune-privileged beta cells

In addition to the optimization of directed differentiation of hPSCs into mature sBC, the field of T1D research has a vested interest in understanding the immune/beta cell interface [64]. This area of investigation includes trying to identify not only the early inflammatory events that lead to the activation of autoreactive immune cells, but also the consecutive interactions that ultimately cause specific beta cell destruction. Attaining such knowledge will not only help in protecting sBC grafts but also might provide insights as to how prevent autoimmunity directed against beta cells from developing in the first place. Precise and efficient genetic engineering usually harnesses targeted double strand DNA breaks to induce desired site-specific editing. Key advances in genome engineering have recently been described and technologies are centered on clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9, Zinc finger, and Trans activator like nuclease families. Such genome engineering now enables efficient modification of hPSCs including the knockout (KO) of genes, insertion or knock-in (KI) of transgene cassettes, inducing point mutations, and changing the chromatin landscape locally to name a few of the most widely used applications [65].

Harnessing these approaches, various investigators have sought to generate hypo-immunogenic stem cell lines that could suppress the ability for cells to be detected by the immune system (Figure 2). Several groups have reported that HLA class I molecules are upregulated on the surface of beta cells in T1D organ donors and upon exposure to T1D mimicking, proinflammatory culture conditions [66, 67, 68]. HLA class I molecules are an integral piece of the process to activate cytotoxic CD8⁺ T cells through their engagement with the CD8 T cell receptor. Therefore, genetically manipulating the expression of HLA class I molecules on beta cells could be an ideal target to abrogate both alloimmune and autoimmune cytotoxic responses toward beta cell grafts.

Figure 2.

Main mechanism of beta cell destruction and overview of approaches targeting immune suppression. An autoreactive T cell can recognize autoreactive peptides presented by the beta cell within HLA molecules. These HLA/peptide complexes interact with the autoreactive T cell receptor (TCR) and trigger the cytotoxic activity of the T cell. Through a deletion of all HLA-I molecules or overexpression of suppressive molecules such as programmed death-ligand 1 (PD-L1), this cytotoxic T cell action is suppressed. To also inhibit the innate immune-mediated destruction of the beta cell by natural killer (NK) cells, the specific expression of HLA-E and overexpression of inhibitory receptor CD47 have been utilized. Figure generated using Biorender.

To neutralize the negative impact of HLA class I expression on beta cells, various approaches have been assessed. hPSCs with a genetic knock out of the beta 2-microglobulin chain (B2M KO), shared among all HLA class I molecules has been utilized [67, 69]. B2M KO hPSC have been showed to exhibit reduced immune activation in an allogenic model [70, 71]. Generating B2M KO sBCs has demonstrated the ability to suppress autoreactive CD8⁺ T cell activation in a fully matched human model system [67]. However, other studies have suggested an increase of natural killer (NK) cell-mediated cell death due to a lack of HLA-E/G expression required for normal immune surveillance by NK cells [72]. To overcome this challenge, lentiviral overexpression of HLA-E within B2M KO cells was able to suppress allogeneic T cell proliferation and activation without inducing NK cell activation [73]. As an alternative to maintaining HLA-E/G, another approach to reduce NK cell activity is the overexpression of CD47, which is a ubiquitously expressed immunomodulatory suppressive gene [74, 75]. CD47 is extremely effective at inhibiting NK cells killing of MHC-deficient iPSCs in immunocompetent mice [75].

Another popular method to suppress immune-mediated destruction of sBCs is the activation of the program cell death protein 1 and ligand (PD1/PD-L1) axis. In the normal adaptive immune response, PD1 is expressed on T and B cells, whereas its ligand PD-L1 is expressed on a variety of cells including antigen presenting cells [76]. The interaction between PD1 and PD-L1 results in a downregulation of the adaptive immune response [77]. To harness this suppressive pathway, overexpression of the ligand PD-L1 on hPSCs is utilized. PD-L1 overexpressing hPSCs displayed the ability to inhibit the activation of autoreactive immune cells in vitro [67, 78]. However, targeting a potent immunosuppressive pathway can have its downsides that must be considered. There are some reports that inhibition of the PD1/PD-L1 axis can cause the development of autoimmunity, including T1D [79, 80, 81, 82, 83]. Therefore, the safety of these approaches must be carefully assessed before they can be harnessed for clinical use.

Other investigators have sought to develop a strategy which addresses both adaptive and innate immune responses through a combination of genetic modifications to knockout the HLA molecule expression followed by knock-ins (KI) to express the immunomodulatory factors PD-L1, CD47 and HLA-G [70]. This study demonstrated that these modifications led to significant reduction in allogenic immune responses with respect to T cell, NK cell, and macrophage-mediated killing in vitro assays. Many of the above studies primarily investigated these genetic alterations in non-differentiated cell types using allogenic assay systems. It remains to be determined if these mechanisms will effectively protect stem cell-derived beta cells from recognition after transplantation into an autoimmune environment. Additionally, one important consideration for the utilization of hypo-immune cell populations is the potential negative side effect of increasing the likelihood of cancerogenic transformation of such cells. To directly address these issues and provide an emergency kill switch in case transplanted cells become tumorigenic, the inclusion of an inducible ‘suicide gene’ that can be triggered at will is likely essential for the safety of these approaches [84].

3.2 Modeling human beta cell/immune interactions

Once ‘designer sBCs’ have been generated, there is a critical need for ways to assess the in vitro and in vivo immune responses to this potential cell replacement therapy prior to clinical trials. Currently, the in vitro assessment of immune recognition and activation relies heavily on co-cultures of primary or PSC-derived human beta cells with allogenic human PBMCs. With the advent of iPSC technology this can now be done in a personalized, matched manner. Somatic cells from a T1D donor patient or control individual can be reprogrammed into iPSC followed by differentiation into sBCs. These sBC can then be co-cultured with immune cells isolated from the same respective donor. Control or T1D donor cells would be used to either exclude or include the autoreactive component of the immune response [70, 85]. However, such innovative co-culture approaches still have their limitations. Perhaps the most important of them being the fact that even within patients diagnosed with T1D, the frequency of autoreactive T cells in circulation is extremely low, preventing a proper assessment of the autoimmune recognition of sBCs with these models [86]. One method to circumvent reduced frequencies of autoreactive T cells is to genetically engineer autoreactive TCRs identified within the pancreas of T1D tissue donors into avatar T cells. This approach allows the generation of large numbers of T cells with known TCR reactivity for downstream co-culture experiments, improving access and reproducibility [87]. Notably, this approach allows for an HLA-matched assessment of autoreactive T cell destruction. Yet, even using these engineering approaches, in vitro cultures will only supply a partial, isolated understanding of the underlying mechanisms at play. Pre-clinical in vivo systems are also required to provide a more wholesome understanding of immune/beta cell interactions. While considerable strides have been made toward developing novel humanized animal models, further improvements are required, especially to accurately assess autoimmunity in these models.

Mice containing aspects of the human immune system, also known as ‘humanized mice,’ offer an amenable pre-clinical model of the human immune response and have been used for a variety of transplantation immunology studies [88, 89, 90, 91]. There are number of humanized models available [92], but most advanced for transplant of hPSCs or their derivatives are those models which incorporate human hematopoietic stem cells (HSC) and thymic fragments into immune deficient mice to facilitate human T cell developmental in the animals. The bone-marrow-liver-thymus (BLT) model [93] and NeoThy model [94] are two popular models. Both contain de novo generated human HLA-restricted T cells, and a host of other adaptive and innate immune cell types useful for the assessment of transplantation success. Newer iterations include mouse models that have introduced mutations to various immunodeficient strains such as the NSG that remove the requirement for irradiation-based myeloablation before HSC transplant [95]. However, in the context of beta cell replacement therapies, these models only allow for the assessment of allogeneic graft rejection, therefore ignoring the key aspect of autoreactive destruction of a potential sBC graft. In addition, a common complication of humanized mice also present within these sophisticated models is the impaired trafficking of human T cells due to species differences of surface proteins.

To determine the true efficacy of a genetic alteration of sBCs, it is paramount to assess the autoimmune component. Potentially the closest animal system that currently exists to allow for the assessment of autoreactive immune cell recognition of a human sBC graft is a humanized model based on the non-obese diabetic (NOD) mouse. The NOD mouse is the most widely used pre-clinical T1D model, and unlike many other autoimmune disease models, the disease is spontaneous [95]. Importantly, disease risk is associated with numerous gene polymorphisms, many of them also found in T1D patients (MHC-II, PTPN22, PTPN2, CTLA4, IL10, CTSH, CD226, IL2, RGS1, TAGAP) [96]. To optimize the ‘humanization’ of this mouse model, all mouse MHC class I molecules were knocked out by genetic ablation of the conserved beta 2-microglobulin chain (NOD.β2m^−/− mice) [97]. This deletion caused a resistance to the autoimmunity observed within normal NOD mice, highlighting again the importance of class I HLA molecules. Then, the most abundant T1D-associated class I susceptibility variant in humans HLA-A*02:01 was inserted (HHD). When introduced into normally disease-resistant NOD.β2m^−/− mice, HHD transgene expression of HLA-A*02:01 restored the presence of pathogenic CD8⁺ T cells and T1D progression [98, 99]. These NOD-HLA-A2/HHD mice and their immunodeficient counterparts, NSG-HLA-A2/HHD mice, allow for a limited, but essential assessment of HLA-matched in vivo responses in an autoimmune environment. If using HLA-A*02:01 restricted hPSCs, this animal system allows for the assessment of autoimmune responses toward engineered sBCs [69]. However, presentation of mouse peptides during T cell development on human HLA-A2 molecules might still contribute a xenogeneic component to this model that should be considered when interpreting data. By continuously improving current humanized mouse models, the use of hPSC-derived beta cells along with humanized in vitro systems will allow for a true investigation of the human beta/immune cell interface in a matched autoimmune environment and open the door for a better understanding of possible clinical T1D interventions.

4. Summary

Since the initial development of directed differentiation protocols to generate human beta cells in vitro, the field has rapidly evolved. Early clinical trials involving transplantation of stem cell-derived pancreatic cells have shown some promising early results, reducing exogenous insulin requirements after transplantation [100]. Stem cell-derived beta cells can now be generated on a large scale, effectively removing the shortage of cadaveric islets previously preventing widespread utilization of cell therapy for T1D. First clinical trials using sBCs are currently in progress but peer reviewed reports as to the outcomes are still awaiting publication. While these are exciting times, seeing the implementation of stem cell technology for the treatment of diabetes in human patients, a few important challenges remain. Improvements to the directed differentiation process must be made to ensure the production of sufficient amounts of mature insulin-producing cells, as current protocols cannot achieve comparable glucose-responsive insulin release to cadaveric islets. Advances in the use of in vitro generated sBCs to model previously inaccessible aspects of human disease must be further explored, opening the potential for a better understanding of the human beta cell/immune interaction in the future. This will be critical knowledge to allow long term function of grafts. Harnessing the virtual limitless potential of genetic engineering to supply effective protection of sBC grafts from allogenic and recurring autoimmune rejection without systemic immunosuppression is an attractive prospect to allow for long-term survival and function of transplanted cells, and finally ensuring sufficient safety measures in case the engineered cells become aberrant are required for clinical trials. Addressing these remaining hurdles will allow for a well-tolerated beta cell replacement therapy for a large patient population affected by diabetes.