Cell Replacement Strategies Aimed at Reconstitution of the β-Cell Compartment in Type 1 Diabetes

Regenerative medicine promises to contribute to the advancement of β-cell replacement strategies through the development and implementation of microencapsulation technology, the production of insulin-producing cells either by regeneration from endogenous cells or reprogramming from nonendocrine adult cell sources, and, more recently, the exploitation of bioengineered microenvironments (1). Considering the shortage of available pancreata, the search for alternative cell sources has recently been categorized within one of the following three “Rs” of pancreatic β-cell replenishment: replacement (from stem cells or xenoislets), regeneration (from endogenous progenitor cells), and reprogramming (from nonendocrine adult cell sources) (2).

The purpose of this article is to comprehensively and critically review the main regenerative medicine−based strategies that are currently being developed to treat type 1 diabetes. The first part of the article will illustrate and discuss islet encapsulation technology, which has been extensively investigated over the past 40 years (3) and represents one of the most promising regenerative medicine−based approaches to achieve immunoisolation of transplantable organs. The second part will focus on cell bioengineering, where different types of progenitor and differentiated cell sources are manipulated to ultimately yield insulin-producing cells. The last part of the article will summarize the most recent advancements in the study of the pancreatic extracellular matrix (ECM) as a platform for endocrine pancreas bioengineering.

Encapsulation technology has been used to protect islets from the host immune system. This strategy holds the promise to allow implantation of islets without immunosuppression not only across genetically different individuals from the same species, but also across species barriers.

The first approach to immunoisolate cells consists of the microencapsulation of one to three islets per semipermeable immunoprotective capsule. The spherical configuration of these microcapsules results in a higher surface-to-volume ratio and a higher diffusion rate (4). Furthermore, microcapsules can be injected in large numbers, are durable, and are difficult to disrupt mechanically (5,6). In this area, alginate is and has certainly been the most commonly used polymer to encapsulate islet cells.

There have been sporadic reports of patients receiving alginate microencapsulated human islets during the last two decades. Soon-Shiong (7) reported on one patient in 1994 who stopped insulin treatment for 9 months concomitant with production of C-peptide (from 0.1 to 0.6 ng/mL) and a decline in HbA_1c level from 9.3 to 7.6% following receipt of microencapsulated islets. In 2009, Tuch et al. (8) reported four patients with single or multiple microencapsulated human islets infusions into the peritoneum without significant correction of glycemia or insulin needs, but urinary C-peptide was detectable up to 2.5 years later in one patient. A humoral response was induced and cytotoxic antibodies were found in recipient sera 4−8 weeks following infusion associated with necrosis of the islets at 16 months.

More recently in 2011, Calafiore and colleagues (9) similarly reported four patients who received single or multiple alginate microencapsulated islets infusions with a discrete correction of HbA_1c level (−1%) and decrease in exogenous insulin needs. A C-peptide response was evidenced during basal assessment and after stimulation and no alloantibodies were detected during a follow-up of 3 years.

These clinical reports have demonstrated that intraperitoneally infused microencapsulated human islets can be considered safe for up to 3 years (8,9). Although insulin independence was not achieved, glycemic control was slightly improved, with reduction of insulin daily requirements in some cases. The main question remains as to why such a technique has not been broadly used and expanded in well-controlled clinical studies in order to determine whether microencapsulation could be an efficient way to graft human islets without the use of immunosuppression. The lack of human pancreas cells certainly plays a major role and it is not obvious or ethical to use rare human islets in uncertain clinical pilot studies, especially when 1) preclinical microencapsulation studies in large animal models are lacking and 2) solid results have been obtained in some clinical studies with human islet transplantation following the Edmonton protocol. In fact, for diabetic patients who are already under immunosuppression (following kidney transplant), the advantage of encapsulation is not clear. At present, microencapsulation should be considered only for patients who will not be subjected to immunosuppression and are not candidates to receive either an islet or pancreas transplant. A second barrier to the large use of encapsulation could be the long-term stability of microencapsulation, which has not been broadly investigated in preclinical studies. But it has been recently reported that intact alginate microcapsules were present in the peritoneum of one patient for 9.5 years (10,11). A third barrier in the past was inadequate purification of alginate. Currently, however, alginate can be produced with almost no endotoxin contamination, which could make a difference in terms of foreign body reaction, biocompatibility, and stability.

For these reasons, clinicians believe that encapsulation could be considered when the use of an alternative source of insulin-producing cells becomes available. Interestingly, in recent years there have been more patients receiving microencapsulated pig islets rather than allogeneic human islets.

In 2007, Elliott et al. (10) reported the case of a 41-year-old labile diabetic patient who in 1996 received a single implantation of microencapsulated pig islet cells into the peritoneal cavity to help regulate blood glucose levels and control diabetes. This patient's insulin requirements were reduced 10–30% for over 14 months, and 9.5 years later, a laparoscopic examination revealed intact capsules and some living and functioning pig islet cells in the peritoneal cavity. These findings demonstrated the long-term safety, viability, and functionality of encapsulated porcine islets in one patient, without the use of immunosuppression.

Although larger clinical studies with microencapsulated pig islets grafts are still being conducted (New Zealand, Argentina) or have been completed (Russia) by the same team and Living Cell Technologies (http://www.lctglobal.com), only partial data have been disclosed (12). However, these data are promising as they show that severe hypoglycemia and hypoglycemia unawareness were significantly reduced in patients receiving intraperitoneal pig islets implantation.

Until recently, the use of microencapsulated pig islets in humans was not really supported by strong preclinical data, except one historical study in 1996, which demonstrated following transplantation of microencapsulated adult pig islets into spontaneously diabetic cynomolgus monkeys that blood glucose became normalized and the monkeys became insulin-independent for periods ranging from 120 to 804 days (13). Although these results were very encouraging for the clinical application, they may have been dependent on the primate model. In fact, the article does not report a precise status of the “spontaneous” diabetic monkeys used and particularly not the primate C-peptide levels from these animals prior and during the follow-up.

Two more recent studies describing transplantation of microencapsulated neonatal pig islets in an alginate matrix confirmed their biocompatibility in primates as well as their capacity to partially regulate diabetes very transiently (14,15). One study confirmed the biocompatibility for up to 8 weeks after transplantation of encapsulated pig islets in nondiabetic animals; the second demonstrated that these microcapsules could regulate the diabetic state of diabetic recipients. Although the latter showed that daily exogenous insulin requirements were reduced by a mean of 43% compared with control animals transplanted with empty capsules, neither showed significant changes in weekly blood glucose levels (14). The absence of solid consistent data on glucose metabolism renders these experimental studies difficult to consider as evidence for going to clinical trials. More recently, Gianello and colleagues (16) presented data showing that in nondiabetic primates some of the microencapsulated pig islets survived for up to 6 months and were able to respond in vitro to glucose challenge 135 and 180 days after implantation under the kidney capsula. In addition, the microcapsules were intact under the kidney capsule.

Although microencapsulation seems able to protect islet cells from the immune attack, the previous data are insufficient to provide evidence for clinical applications. More interesting data could come from macroencapsulation. In the first preclinical reports of macroencapsulation, a large number of islets were immunoisolated between flat-sheet double membranes (10,11). Although several types of biomaterial have been used to produce macrocapsules, including nitrocellulose, alginate, acrylonitrile, and agarose, these devices activated nonspecific foreign body reactions, resulting in fibrotic overgrowth with subsequent necrosis of the encapsulated tissue (17). Interestingly, a subcutaneously transplanted macrodevice, 4 cm in length, shaped like a teabag, and made of bilayered polytetrafluoroethylene membranes, was found to be biocompatible (11,18).

In humans, Valdés-González et al. (19) reported among 12 nonimmunosuppressed adolescents that half had reduced insulin requirement up to 4 years after transplantation of porcine islets encapsulated in hollow fibers with porcine Sertoli cells, which likely have immunomodulating properties. Two patients became insulin-independent for several months, and porcine insulin was detected following glucose stimulation in three patients. However, this study was done in young patients without preclinical results and was received with skepticism by the scientific community (20).

Recent data using macroencapsulation of pig islets in nonhuman primates are encouraging. In fact, a “monolayer” configuration of macroencapsulated pig islets (monolayer cellular device) implanted subcutaneously has been found to significantly improve diabetes control (HbA_1c <7%) in primates for 6 months without any immunosuppression. In this macroencapsulation system, islets were seeded as a monolayer on an acellular collagen matrix, enhancing their interactions with a biologic membrane and increasing islet concentration per unit surface area. In addition, diabetes was controlled for up to 1 year in two diabetic nonhuman primates after retransplantation with new monolayer cellular devices (21). Interestingly, in the same model, diabetic control was obtained for more than 32 weeks after the cotransplantation of adult pig islets and either bone marrow or adipose mesenchymal stem cells (MSCs) (22). The cotransplantation of MSCs increased the vascularization of the monolayer device in terms of increased number of vessels and production of vascular endothelial growth factor (Fig. 1). By using this subcutaneous monolayer cellular device, a phase 1 clinical study is currently ongoing at the Université Catholique de Louvain to assess the safety of this device for allotransplantation of encapsulated islets into humans.

There are two main categories of polymers that can be used either as water-soluble polymers, such as alginate, or water-insoluble polymers, such as poly(hydroxyethyl methacrylate-methyl methacrylate) (17), but the latter requires organic solvent usually interfering with cellular function (23). Despite their solubility in aqueous solutions, alginate-based capsules have been shown to remain stable for several years in small and large animals and in humans (11,21,24–27). In most studies to date, alginate beads were coated with a second layer to reduce the porosity of the capsule membrane (17,28); in some studies, alternating layers of poly-l-lysine and polyornithine were used to surround the alginate core but mechanical instability was described, thereby limiting their application (17,29,30). Several groups have recently reported that encapsulation in simple alginate microbeads can protect pig pancreatic cells against xenorejections in diabetic mice (5,6,31). Although several chemical formulations of alginate (e.g., high-mannuronic/guluronic and high/low viscosity, with or without additional peptide sequences) have been proposed for islet immunoisolation, high-mannuronic alginate was the most suitable to obtain selective impermeability for molecules over 150 kDa and optimal biocompatibility associated with surrounding angiogenesis and sufficient oxygen tension (up to 40 mmHg) (32). This type of alginate was biocompatible not only in rodents but also into a pig-to-primate model of xenotransplantation for up to 8 months (16,33,34).

Islets perform well in a number of ectopic locations and as long as they get appropriately vascularized, islets will be able to function correctly. A site in which encapsulated islets are in close contact with the bloodstream is crucial for clinical applications. Unfortunately, it is difficult to find such a site for encapsulated islets due to the large graft volume needed. Sites reported to allow successful nonencapsulated islet transplantation, such as the liver and spleen, do not meet these requirements because these sites are unable to tolerate the large volumes of capsules (of diameter >600 μm) required for transplantation. Therefore, most transplantations of encapsulated pig islets into preclinical models were intraperitoneal (13,14). Although the technique is easy by laparoscopy, the peritoneal site could be not optimal. In fact, studies in mice found that macrophages and lymphocytes are involved in the rapid degradation of encapsulated pig islets after their transplantation into the peritoneum (31,35–38). The peritoneum is a preferential site for inflammation and immunologic reactions. In fact, peritoneal mesothelial cells facilitate the action of powerful innate immune mechanisms by producing (39) significant levels of cytokines, such as tumor necrosis factor-α, interleukin (IL)-1β, IL-10, and intracellular adhesion molecule-I (40). Studies in mice showed that immunosuppression had beneficial effects, improving the biocompatibility and prolonging the survival of encapsulated pig islets transplanted into the peritoneum (35,36,41), but the combined use of encapsulation and immunosuppression probably reduces to nihil the interest of encapsulation.

Encapsulated pig islets transplanted under the kidney capsule and under the skin demonstrated better biocompatibility than capsules transplanted into the peritoneum in both rodents and primates (33). Subcutaneous and kidney capsule implantation resulted in very weak cellular immune reactions (low macrophages recruitment) against encapsulated pig islets, along with improved porcine islet viability, optimal insulin secretion after glucose challenge, and an acceptable oxygen tension (20–40 mmHg) compatible with the function and survival of encapsulated islets (32).

The process of differentiating pluripotent cells toward pancreatic endocrine cells has not been easy or without controversy, but there is a consensus that we have finally reached the “plateau of productivity” often cited when describing the evolution of most technical innovations (42,43). The protocols developed over the second half of the last decade by scientists at ViaCyte (formerly Novocell) (44,45) have paved the way for what is widely expected to be the first human embryonic stem cell−based clinical trials within the next few years (Fig. 2). These methods have circumvented a bottleneck still in place today (our inability to generate fully functional β-cells in vitro [44]) by calling for the transplantation of these cells at the β-cells progenitor stage, long before they are capable of producing insulin (46). By doing so, the recipient’s body provides a microenvironment that is permissive for their efficient functional maturation, even if the process is neither short nor safe (47). Indeed, maturation may typically take anywhere from 2 to 3 months, during which carry-over undifferentiated cells may give rise to teratomas. While further preclinical refinements of the protocol have reportedly taken care of the latter problem (which could also be addressed by selecting only nontumorigenic cells for transplantation [46]), the former is still of concern for those who would rather have a fully characterized cell product prior to transplantation. Regarding the issue of rejection, the first clinical trials are likely to adopt the embodiment proposed by ViaCyte (www.viacyte.com), in which allogeneic progenitor cell populations will be seeded within a subcutaneously implanted immunoisolation device (see Macroencapsulation), which would serve the dual purpose of preventing allorejection and providing a physical containment for any potential tumors that may arise. As this approach does not require immunosuppression of the patient, any conceivable breakage of the device (e.g., by tumorigenic cells) would result in the prompt rejection of the graft.

As for the star newcomers in the field, namely induced pluripotent stem (iPS) cells (48,49), they have also proved their worth at generating pancreatic endocrine cells in vitro (50). However, as fast as iPS cells rose to relevance, concerns about their stability have somewhat cooled down their prospects as potential replacements for human embryonic stem cells. In particular, it has been reported some of their derivatives senesce prematurely (51), that they maintain some residual epigenetic memory of their parental tissues (52–55), and that the formation of potentially oncogenic mutations might be inherent to the process of reprogramming (56–59).

The hypothesis that the ductal system may harbor such progenitor cells has been considered for decades now, and is still being studied very actively. More recently, a unique population of cells within the extrahepatic biliary tree has been characterized for its capacity to turn into both hepatocytes and pancreatic endocrine cells (60,61). Their existence, rather than a mere remnant of the embryonic stage at which liver and pancreas diverted, has been explained in the context of a theoretical process of organogenesis throughout life (62,63). Whether or not these cells can be expanded in sufficient amounts for clinical use is the subject of active investigation.

In parallel to the above efforts, many groups are pursuing the most ubiquitous of all adult stem cells, the MSCs (or stromal cells). MSCs are easily expandable in plastic and can be isolated from most tissues, including the pancreas (64). In a recent review of the subject, Domínguez-Bendala et al. (65) echoed the growing perception that these mesodermal cells are intrinsically incapable of becoming bona fide endodermal β-cells. Their directed differentiation and subsequent transplantation has met only with partial success in preclinical models (66–75). However, when directly infused in an undifferentiated state onto diabetic subjects, a variable degree of endogenous regeneration has been observed (76–79). It is now broadly believed that MSCs are at their therapeutic best when used in this context, probably due to their well-studied immunomodulatory, anti-inflammatory, proangiogenic, and trophic properties (80–87).

Hematopoietic stem cells are another variety of adult stem cells that have been studied for their potential use to treat type 1 diabetes. To an even higher degree than MSCs (because they do it systemically), the main usefulness of hematopoietic stem cells relies on their ability to modulate immunity. Thus, they have been successfully tested for their ability to reset the immunological clock of diabetes (88,89). Cord blood−derived multipotent stem cells (CB-SCs) are also the basis of the novel “stem cell educator” concept, in which lymphocytes of type 1 diabetic patients are circulated through a device preseeded with CB-SCs from healthy donors and reinfused into the patient after a quick process of “re-education” that has been reported to reverse type 1 diabetes (90,91). These findings, however, still require independent confirmation.

In spite of the concerns previously pointed out, the reprogramming techniques originally developed for iPS cell generation are here to stay, and will probably be used more and more for other applications. From a technical perspective, the use of retroviruses to deliver the reprogramming agents appears now downright primitive in view of the more translational friendly alternatives of synthetic mRNAs, DNA minicircles, episomal vectors, or even transducible proteins (92,93). Early pioneers of the field of pancreatic reprogramming made use of the master gene Pdx1, which, when ectopically expressed in the liver of diabetic mice, led to reversal of hyperglycemia (94). Immunohistochemical analysis of such livers revealed insulin-positive cells in close proximity to blood vessels. It was subsequently found that Pdx1 expression persisted in the liver long after the expected clearance time of the adenoviruses used to carry the gene, suggesting that the initial ectopic expression of the gene may have primed endogenous gene networks that sustain long-term reprogramming (95). Whether these observations were reflective of the reprogramming of terminal hepatocytes or the differentiation of resident progenitor/stem cells was not clear at the time, and in vitro experiments did not prove to be much more informative (96). Additional reports established that combining Pdx1 with other transcription factors, such as BETA2/NeuroD, Ngn3, or MafA, had a synergistic effect (97–103). Pancreatic ductal cells, which have been historically thought to contain putative β-cell progenitors, have also been subjected to similar transfection regimens, leading to significant upregulation of insulin expression. An intriguing observation in the field has been that often times reprogramming is seemingly dependent on the activation of immune/innate responses in the host cells. This was evidenced by the fact that the adeno-associated virus (AAV)-mediated delivery of such factors failed to induce liver-to-pancreas transdifferentiation, whereas plasmids plus an irrelevant adenovirus (which are much more immunogenic than AAVs) did (103).

Just around the time when enthusiasm for the directed transdifferentiation to pancreas had started to wane, a new breakthrough rejuvenated the field again. This time, however, the starting material was the exocrine compartment of the pancreas, rather than the liver. Exocrine cells are known to be highly malleable, as shown by their reported conversion to liver (104), ductal (105,106), and β-cell−like cells (107–111). Zhou et al. (112) came up with a combination of three key pancreatic transcription factors, Pdx1, Ngn3, and MafA (also used with some success in earlier liver-to-pancreas reprogramming attempts), that, when shuttled aboard adenoviral vector and injected into the pancreatic parenchyma, gave rise to new β-cells in vivo. Upon treatment, diabetic mice exhibited a marked amelioration of their glycemic levels.

More recently, the above results were confirmed in vitro using first the AR42J acinar cell line (113,114), and subsequently primary human pancreatic exocrine cells cultured in conditions that prevented epithelial-to-mesenchymal transition (114). While not as ready as hematopoietic stem cells for clinical prime time, reprogramming of human exocrine tissues offers a most promising alternative to the use of stem cells, especially considering that the exocrine compartment of the pancreas makes up to 95% of the pancreas and is now routinely discarded after every clinical islet preparation.

The pancreatic ECM is a three-dimensional, structural framework of proteins in a state of “dynamic reciprocity” with the cells of the endocrine pancreas (115). Once thought to be a simple scaffold, the ECM has recently being shown to regulate several aspects of islet biology, including development, morphology and differentiation, intracellular signaling, gene expression, adhesion and migration, proliferation, secretion, and survival. The ECM mediates these functions via three-dimensional structure (116,117), signaling molecules, and the secretion and storage of growth factors and cytokines (118). The ECM is therefore essential for effective glucose-stimulated insulin secretion and normoglycemia, the primary function of the endocrine pancreas. Thus, the ECM has become a valuable platform for regenerative medicine investigations, and intact ECM from animal or human whole organs is currently being evaluated for organ bioengineering purposes.

The functional significance of the ECM has been highlighted by the underperformance of isolated islet transplantation (117,119–122). The seminal Edmonton protocol (123) attempted to correct hyperglycemia through the transplantation of isolated islets from multiple cadaveric donors. Although the majority of patients reached insulin independence following transplantation, few managed to maintain euglycemia at 2-year follow-up (119). Subsequent studies suggest that the observed transience of therapeutic effect may be due to the loss of native ECM during the islet isolation process. Disruption of the native ECM appears to compromise survival, engraftment, and revascularization of transplanted islets. This implies that preservation of the native ECM may be a necessary condition for successful islet transplantation protocols.

Several features of the ECM appear to be essential for the function of the endocrine pancreas. The three-dimensional structure of native ECM determines the topographical arrangement of pancreatic endocrine cells, which has been shown to influence islet secretory activity (124) and survival (125). Furthermore, the constituent elements of the ECM, including collagens, glycoproteins, and glycosaminoglycans, have been shown to be independently capable of preventing B-cell apoptosis induced by loss of cellular adhesion, termed anoikis (116,124,126–130). ECM components have also been shown to enhance insulin secretion, even in the absence of glucose (125). Finally, the ECM is capable of binding, storing, and regulating the activity of growth factors including transforming growth factor-β1, which plays a role in development (131), function, and regeneration (132) of pancreatic islets. The dysregulation of these essential ECM−growth factor interactions have been shown to underlie a variety of pancreatic pathologies, including pancreatitis, fibrosis, and adenocarcinoma.

The permissive effects of ECM structure on islet survival and function have led researchers to investigate the use of biomaterial carriers, which mimic the structure of native ECM. These carriers serve the functions of delivery platform, transitory structural support, and mechanical immune barrier, thus enabling islet transplantation into heterotopic sites (117). The properties of these carriers can also be precisely manipulated, allowing for the study of individual ECM components. Several islet carriers have been investigated, including poly-lactic-coglycolic acid (PLGA), poly vinyl alcohol (PVA), poly(ethylene glycol) (PEG), poly(N-isopropylacrylamide) (NIPA), and biopolymer films. These carriers vary in their structure, strength, stability, biocompatibility, growth factor binding, and amenability to manipulation. However, insights from the fields of bioengineering and regenerative medicine suggest that organ-specific ECM better maintains cell function and phenotype. The use of native ECM has been made possible through the refinement of decellularization protocols capable of removing DNA, cellular material, and cell surface antigens while leaving the attachment sites, structural integrity, and vascular channels intact. Decellularization protocols involve the repeated irrigation of cadaveric tissues with detergents or acids through the innate vasculature, although organs with higher fat content, like the pancreas, often require the addition of lipid solvents, such as alcohol (133).

Pancreas bioengineering lags behind other organs in the field, as only three studies to date report successful recellularization of native pancreatic ECM. De Carlo et al. (134) report the successful subcutaneous implantation of PVA/PEG tubular devices containing slices of rat pancreas and liver recellularized with differentiated murine islets. The pancreatic matrix significantly extended the duration of insulin function, suggesting that pancreas-specific matrix favors physiological response to glucose over the long term. Upon transplantation, the islet devices effectively reduced hyperglycemia, although normalization was not achieved. Conrad et al. (135) report abbreviated findings describing the successful recellularization of murine pancreatic matrix with human islet cells and supportive MSCs. The islets showed preserved glucose-stimulated insulin response, cell viability, subcellular anatomy, and attachments. However, complete findings have yet to be published. In a recent report from the Wake Forest School of Medicine, whole-organ acellular ECM scaffolds were generated through detergent-based decellularization of porcine pancreata (7). The method described therein achieved effective removal of the cellular material and DNA content of the native organ, while preserving ECM proteins and the native vascular tree. The scaffolds were seeded thereafter with human stem cells and porcine pancreatic islets, demonstrating that the decellularized pancreas can support cellular adhesion and maintenance of cell functions (Fig. 3). Similar data have been recently produced in a rodent model (136). However, the successful decellularization and subsequent recellularization of the human pancreatic ECM has not been reported in the bioengineering literature.

We have reviewed key aspects of regenerative medicine technologies as they relate to the endocrine pancreas and the reconstitution of the β-cell compartment. Advances in islet encapsulation, stem cell therapy, and organ bioengineering are intersecting with the promise to resolve the dire shortage of transplantable organs. Despite steady progress, pancreatic bioengineering lags behind other organs in the field due to the complex architecture and physiology of the native pancreas, and a poor understanding of the interactions between β-cells, vascular supply, growth factors, and the pancreatic ECM. While current research endeavors are promising, the transition to safe and effective clinical implementation faces significant obstacles. Collaborative efforts are required to drive the field toward the production of a bioengineered pancreas capable of restoring endocrine function in patients with end-stage disease.

Duality of Interest.No potential conflicts of interest relevant to this article were reported.

Author Contributions. G.O. conceived and designed the manuscript, wrote the section on ECM technology, edited the manuscript, and accepted its final version. P.G. wrote the second section, edited the manuscript, and accepted its final version. M.S. cowrote the third section, edited the manuscript, and accepted its final version. R.J.S. and S.S. edited the manuscript and accepted its final version. C.R. and J.D.-B. wrote the first section, edited the manuscript, and accepted its final version.