New sources of pancreatic Beta-cells (Nature Biotechnology – Vol. 23 – July 2005)
Susan Bonner-Weir & Gordon C Weir
Two major initiatives are under way to correct the p-cell deficit of diabetes: one would generate Beta-cells ex vivothat are suitable for transplantation, and the second would stimulate regeneration of Beta-cells in the pancreas. Studies of ex vivo expansion suggest that Beta-cells have a potential for dedifferentiation, expansion, and redifferentiation. Work with mouse and human embryonic stem (ES) cells has not yet produced cells with the phenotype of true Beta-cells , but there has been recent progress in directing ES cells to endoderm. Putative islet stem/progenitor cells have been identified in mouse pancreas, and formation of new Beta-cells from duct, acinar and liver cells is an active area of investigation. Peptides, including glucagon-like peptide-1/exendin-4 and the combination of epidermal growth factor and gastrin, can stimulate regeneration of Beta-cells in vivo. Recent progress in the search for new sources of Beta-cells has opened promising new opportunities and spawned clinical trials.
An axiom of diabetes treatment is that the devastating complications of the disease can be prevented by normalization of blood glucose levels. Although modern insulin regimens are improving outcomes1, the successes achieved over the last few decades by transplantation of whole pancreas and isolated islets suggest that diabetes can be cured by replenishment of deficient Beta-cells . Whole-organ pancreas trans-plants have been performed at many centers since the 1970s, whereas transplantation of isolated islets in humans was undertaken seriously only in the 1990s. The latter was not considered successful until the demonstration that patients with type 1 diabetes could be predictably freed from insulin treatment using the so-called Edmonton protocol2. It is now apparent, however, that the clinical benefit of this protocol can be provided only to a small minority of patients, is not lasting and is accompanied by significant side effects3. Nonetheless, the promis-ing results afforded by transplantation of whole pancreases and iso-lated islets, coupled with the shortage of cadaver pancreases relative to the potential demand, have lent strong impetus to the search for new sources of insulin-producing cells.
This short review cannot attempt to cover the complex and often frustrating pursuit of the dream of curing diabetes; it is thus limited to summarizing progress of the last two years in research aimed at generating a clinically useful supply of new Beta-cells . Our focus will be on efforts to produce Beta-cells suitable for transplantation. Promising strategies include expansion of existing Beta-cells , differ-entiation of embryonic stem (ES) cells to Beta-cells , and conversion of either pancreatic or nonpancreatic adult stem/progenitor cells to Beta-cells (Fig. 1). We shall also consider an alternative, pharmacologi-cal approach that seeks to regenerate Beta-cells in the pancreas, either by replication of existing Beta-cells or by the generation of Beta-cells from other cell types4.
Expansion of Beta-cells ex vivo
Beta-cells have long been known to have a substantial capacity for replication, as best shown in rodents with various ex vivo and in vivo model sys-tems5. The replication rate of human Beta-cells , however, is much lower than that of rodent Beta-cells , although it can be modestly stimulated by transplantation of Beta-cells into insulin-resistant mice6. Considerable attention is now being focused on ways to enhance p-cell replication to generate cells for clinical application7-10. Recent data from three groups suggest that Beta-cells of cultured islets can dedifferentiate and expand, and then be directed to redifferentiate back toward a p-cell phenotype. The changes resemble the process of epithelial-mesenchymal transition (EMT)11. As cells expand from cultured human islet preparations, they develop a serpiginous appearance and express nestin and the mesenchy-mal marker vimentin, but not islet hormones7,8. They are not completely mesenchymal, however, because the endodermal proteins amylase, carbonic anhydrase II and albumin continue to be expressed8. In all three studies cited above, multiple passages were achieved, suggesting that the Beta-cells were capable of self-renewal. A variety of maneuvers were used to force redifferentiation, including serum-free media, nicotinamide, gluca-gon like-peptide 1 (GLP-1)/exendin-4, activin A, betacellulin, hepatocyte growth factor, Ly294002 (an inhibitor of PI3-kinase), and aggregation of cells, which led to expression of an assortment of islet markers at low levels. However, in all three studies the insulin mRNA levels were only about 0.01% that of human Beta-cells 12.
A caveat with all of these studies is the difficulty of distinguishing between the dedifferentiation of Beta-cells and the expansion of pancreatic stem/progenitor cells that are not of p-cell origin. Even the purest human islet preparations have many contaminating duct, acinar, stromal and endothelial cells, all of which proliferate actively. There is clearly a need for rigorous marking techniques, such as lineage tracing, to define the extent to which Beta-cells themselves can serve as precursors. Another rigorous approach would be to perform studies with clonal populations of human primary cells.
It is interesting to compare these studies with those of PANC 1 cells, a human pancreatic duct cell line, which have been shown to undergo a similar morphological transition with changes in cul-ture conditions. When aggregated in serum-free media with a high glucose concentration (17 mM), the cells express very low levels of islet hormone mRNA and protein13.
Embryonic stem cells
ES cells hold promise as a source of new Beta-cells , but realizing this potential has proven more difficult than expected. Beginning in 2000, papers began to appear that raised hopes prematurely. Progeny of mouse and human ES cells were reported to contain insulin and even to have regulated insu-lin secretion14,15. Several approaches have been used to obtain enriched populations: cell trapping with antibiotic resistance driven by the Nkx6.1 or insulin promoter to select cells14,16; selection by manipulating the cul-ture conditions17-19; and forced expression of key transcription factors such as pax4 and pdx-120-22. In one of the few reports on human ES cells, the combination of basic fibroblast growth factor (bFGF) followed by nicotinamide, low glucose concentration and suspension culture led to increased numbers of insulin-expressing cells23.
Although these studies provided evidence that cells containing insulin and various other p-cell markers could be generated from ES cells, a recent report demonstrated that insulin staining could be artifactual, reflect-ing insulin uptake by apoptotic cells from culture media containing high concentrations of insulin24-26. The resulting controversy, which has been thoughtfully addressed by Kania et al.27, has pushed the field to measure not only insulin but also C-peptide, which is excised from proinsulin, to prove that insulin has been synthesized. With this insulin artifact placed in perspective, it appears from published and unpublished work that various manipulations of ES cells can produce cells that do in fact produce insulin. Considering that human neural stem cells are capable of producing small amounts of insulin28, it is possible that the insulin found in ES cell-derived cells arises from aberrant neuronal differentiation. However, these cells have no more than a tiny fraction of the insulin content of a normal p-cell and exhibit incomplete expression of p-cell markers, thus falling far short of full differentiation to Beta-cells .
Great care must be taken in defining how closely such cells resemble normal Beta-cells , which are very well characterized with regard to their gene expression, metabolism, growth potential and secretory function29. Many believe that insulin-producing cells destined for therapeutic replace-ment must match the extraordinary performance of normal Beta-cells , with their ability to store large quantities of insulin and secrete it precisely to meet the complex demands of meals, exercise and fasting30,31. It must be recognized that there are developmental pathways found in tissues such as fetal liver, yolk sac, brain and other neuronal tissues32-34 that generate insulin-containing cells that will never become true Beta-cells . In rodents, Beta-cells normally express two insulin genes, insulin I and insulin II, but insulin-expressing cells outside the pancreas typically express only insulin II and not insu-lin I. Insulin II is not a suitable marker of p-cell phenotype in rodents, and in humans there is only one insulin gene.
The lack of success of these early attempts at differentiating ES cells has focused attention on the fundamentals of normal embryonic devel-opment to better understand the early stages of endoderm formation. One complication of this approach is the need to be sure that ES cells are directed to definitive endoderm rather than visceral endoderm, which has similar markers, but is a diferent pathway35. A potentially important recent contribution is a protocol in which a ‘pulse’ of serum-containing medium, followed by serum-free conditions with activin A, resulted in enrichment of endoderm markers from mouse ES cells36. Using a modi-fication of this protocol, Ku et al. were able to obtain 2.7% insulin-positive cells compared with less than 1% in controls37. Not only was insulin I mRNA found, but most insulin-positive cells were also C-peptide- and glucagon-positive, strongly suggesting a pancreatic pathway.
Pancreatic adult stem/progenitor cells
Islet neogenesis, the budding of new islets from pancreatic stem/progeni-tor cells located in or near ducts, has long been assumed to be an active process in the postnatal pancreas (Fig. 1). However, a recent study has challenged the view that neogenesis from ducts or any other progenitor cell takes place. Using genetic marking for lineage tracing with the insulin promoter, Dor et al. concluded that no new islets were formed in mice after birth or following 70% pancreatectomy, but that new Beta-cells could be generated by replication of existing Beta-cells 38. Although this study sup-ports the concept that p-cell replication is the dominant mechanism for p-cell expansion in adult mice, it remains controversial because it does not convincingly prove that new islets are not formed during neonatal life nor after regeneration-inducing maneuvers such as partial pancreatectomy or duct ligation.
Putative adult pancreatic stem/progenitor cells that could be clonally expanded and that express low levels of insulin and other pancreatic mark-ers have been found in mouse pancreas39,40. These cells are rare, and their localization within the pancreas remains undetermined; importantly, it has not been demonstrated that they can differentiate to fully functional islet cells41. In the paper by Suzuki et al.39, putative stem/progenitor cells were isolated with flow cytometry from neonatal pancreas using cell sur-face markers, in particular c-met, the hepatocyte growth factor receptor. The expanded colonies expressed islet and acinar characteristics as well as the duct marker cytokeratin 19 (CK-19) and nestin, an intermediate filament protein thought to be a marker of neural stem cells. Although the full extent of differentiation of these insulin-containing cells remains unknown, markers of hepatocytes and gastrointestinal cells were also expressed. In the study by Seaberg et al.40, rare single clonal cells obtained from adult mouse pancreas also formed colonies expressing markers of both neurons and pancreatic islet cells. These pancreas-derived cells appear to have limited capacity for self-renewal, lack the stem cell markers Oct4 and Nanog, and are of neither mesodermal nor neural crest origin. After further differentiation of single clones, protein markers for neural elements and islets were identified. These pancreatic proteins were expressed in 4-6% of the cells. Furthermore, by RT-PCR, the islet markers he endoderm markers GATA-4 and HNF3p and the pancreatic ductal cell marker cytokeratin were not.
We have favored the concept that pancreatic ductal epithelial cells have the potential to regress to a less differentiated progenitor cell capable of producing new islets and acini42. This hypothesis is supported by our experiments in rats with 90% partial pancreatectomy, a well-established model in which regeneration occurs by two pathways: replication of preexisting endocrine or exocrine cells, and proliferation of ductules and their subsequent differentiation into whole new lobes of pancreas that become indistinguishable from pre-existing ones. As the duct cells replicate, they transiently express the protein PDX-1, a transcription factor expressed widely in the embryonic pancreatic progenitors but by birth restricted to p and 8 cells. We hypothesize that a mature duct cell with rapid replication can transiently assume a less differentiated, and less restricted, phenotype g that can then redifferentiate into any of the pancreatic cell types. Such 13 plasticity results in the possibility of abundant multipotent progenitors © in adult pancreatic ducts that participate in natural renewal, a process that could be enhanced by pharmacological intervention. Duct cells are best thought of as progenitor cells rather than as true stem cells with a capacity for self-renewal. Whether these duct cells have any relationship to the cloned progenitor cells described above remains to be seen.
A variation of this hypothesis, the transdifferentiation of acinar cells to islets, has been championed by Bouwens and colleagues43,44. Infusion of gastrin into rats whose main pancreatic ducts were ligated produced changes consistent with transdifferentiation of acinar cells to duct cells and subsequent p-cell neogenesis45. The observed colocalizations of amy-lase and the duct marker cytokeratin 20 or amylase and insulin supports the hypothesis of a direct transition from acinar to other pancreatic cell types43. However, using alloxan diabetic mice treated in vivo with epider-mal growth factor (EGF) and gastrin, Bouwens and coworkers concluded that the observed normalization of blood glucose and increased islet mass resulted from increased neogenesis from ducts46. This conclusion was based on the findings of transitional cells expressing both cytokeratin and insulin and increased ductal proliferation without increased p-cell proliferation.
A more complete pattern of regression to a less differentiated state was seen in a murine model of acinar regeneration after treatment with cae-rulein, a potent cholecystokinin-A receptor agonist. The regrowth of the acinar compartment came from surviving pancreatic exocrine cells that repressed the terminal exocrine phenotype and induced genes normally associated with undifferentiated pancreatic progenitor cells, su-cadherin, p-catenin and Notch components47. New formation of islets was not seen in this model.
Additional support for the dedifferentiation concept comes from ex vivo studies with human pancreatic tissue. Islet-depleted tissue was expanded in culture and then, with serum-free media and differentiation factors, coaxed to form glucose-responsive, insulin-containing islet tissue48,49. Even after expansion of nonislet epithelial cells, the insulin content and mRNA increased, and cells with mature islet phenotypes were found. Using human islet preparations that include both duct and acinar cells, Suarez-Pinzon et al. reported a doubling of Beta-cells after 4 weeks of treatment with a combination of epidermal growth factor (EGF) and gastrin50. They also observed increases in the number of cells expressing the duct marker cytokeratin 19 and the transcription factor IPF-1/PDX-1, suggest-ing that the increase in p-cell number was due to activation of neogenesis from the pancreatic ducts. Although these studies support the idea that duct cells can serve as progenitor cells, conclusive evidence for this sup-position will require lineage tracing.
Nonpancreatic progenitor cells and transdifferentiation
Bone marrow. The possibility that bone marrow cells are capable of transdifferentiation has raised fascinating questions about developmen-tal plasticity. The potential of bone marrow cells remains open, especially considering the finding that adult bone marrow cells can be pluripotent, capable of progressing to ectoderm, mesoderm and endoderm fates51. A provocative report suggested that cells derived from bone marrow marked with GFP can become glucose-responsive, insulin-secreting cells in islets in vivo52. Several other laboratories have not been able to confirm these findings53,54. Even so, clonal cell lines from murine bone marrow cultured in high glucose (23 mM) for 4 months were recently reported to contain cells expressing a number of p-cell genes, although their insulin content was less than 1% that of a normal p-cell55.
Rather than serving as progenitors, bone marrow cells could facilitate new p-cell formation. In mice with streptozotocin-induced islet injury and resultant diabetes, homing of transplanted bone marrow cells to the pancreas was associated with normalization of glycemia and increased islet mass53,54. Many of these bone marrow-derived cells were found to have an endothelial phenotype, but there was no evidence of their turning into Beta-cells . How these infusions of bone marrow cells promoted such notable islet regeneration is still a mystery.
In other studies, temporary islet transplantation combined with infu-sions of either bone marrow56 or splenocytes and complete Freund’s adjuvant57 reversed diabetes in new-onset diabetic NOD mice. These and other similar studies have provided evidence for some islet regen-eration58,59, which we suspect was due to replication of residual Beta-cells . In the first study56 there was no evidence that bone marrow cells served as progenitors for new Beta-cells , but in the second57, it was suggested that splenocytes give rise to the Beta-cells . It is difficult, however, to be confident that the cells with splenocyte markers in islets were actually in the process of becoming Beta-cells .
Transdifferentiation. Another approach to the generation of Beta-cells is transdifferentiation of differentiated cells of endodermal origin, such as liver or intestinal cells60-66. In the first reports of transdifferentiation of liver cells, in vivo transduction of mice with an adenovirus expressing pdx-1 induced endogenous PDX-1 and expression of other p-cell markers as well as production of substantial amounts of insulin61,62. A similar approach using injections of helper-dependent adenoviruses showed that expression of both betacellulin (a mitogen) and neuroD (a p-cell tran-scription factor) in liver cells led to reversal of streptozotocin diabetes64. These results were interpreted as representing the stimulation of islet neogenesis in the liver rather than transdifferentiation of hepatocytes.
man telomerase (hTERT) and then pdx-1 produced cells with an impressively large amount of stored and secreted insulin65. These cells not only secreted insulin in a regulated manner but also reversed diabetes when transplanted into immunodeficient diabetic mice.
The question of which liver cell adopts the insulin-producing pheno-type remains unanswered. Murine hepatic oval cells were driven towards an islet phenotype by long-term exposure to high glucose67. In a fol-low-up study using a cell line (WB) with an oval cell phenotype stably transfected with an activated form of pdx-1, many genes of endocrine pancreatic development were induced, and an even more mature islet expression profile was obtained after exposure to high glucose levels68. However, even after 4 weeks of ex vivo culture at high glucose, the insulin content was less than 1% that of normal Beta-cells . Surprisingly, 4 months after transplantation the grafts were found to express the transcription factors ngn3 and pax4, which are only transiently expressed during devel-opment.
Regeneration of Beta-cells in the pancreas
There is great interest in the possibility that regeneration of Beta-cells in the pancreas could reverse both type 1 and type 2 diabetes. Little is known about the potential of p-cell replication or neogenesis in humans, but both possibilities will be explored in the near future. The capacity for in vivo expansion of p-cell mass in mice in response to insulin resistance is enormous69, and p-cell mass is known to be increased in human obesity70. Evidence for regeneration of Beta-cells after immunosuppression of new-onset diabetes in NOD mice, a model of type 1 diabetes, has been provided by several groups56-59. This regeneration, which has been observed only in mice with new-onset diabetes, may more likely result from expansion of existing Beta-cells than from neogenesis, given the known capacity for p-cell replication in mice.
Agents that are believed to stimulate neogenesis in rodents include the peptides INGAP71, GLP-1 and the GLP-1 receptor agonist exendin-4 (ref. 72), the combination of betacellulin and activin A73, and the combination g of EGF and gastrin46 (Fig. 2). INGAP, a peptide fragment of the pan-13 creatic REG protein, has been associated with regeneration in rodents71. © GLP-1/exendin-4 have incretin effects, enhancing insulin secretion; they also stimulate p-cell replication and neogenesis and have anti-apoptotic effects72. Betacellulin, a member of the EGF family, stimulates p-cell pro-liferation73. Both activin A, a member of the TGFp family73, and gastrin50 are thought to promote p-cell differentiation.
Successful trials of exendin-4 treatment in type 2 diabetes have now been reported74,75, and patients had a meaningful improvement in glyce-mic control associated with weight loss74. Recently the peptide has received FDA approval for clinical use. The relative contributions of weight loss, improved p-cell function and p-cell regeneration to the improved glucose control cannot be determined from the available results.
In addition to exendin-4, other ways to enhance GLP-1 action include the use of GLP-1 analogs and inhibition of its breakdown in blood with protease inhibitors. The latter approach, which is being examined by multiple companies, employs compounds that inhibit dipeptidyl peptidase IV and can be taken orally. Plans are underway to examine the effects of enhanced GLP-1 action on p-cell regeneration at various stages of type 1 diabetes, including the prediabetic phase before clinical onset, just after diagnosis, after long duration of diabetes, and with islet transplantation. Results of clinical trials using INGAP in both types 1 and 2 diabetes have recently been reported in abstract form, with the treatment showing modest increases in C-peptide levels but little clinical efficacy76,77. Trials are also being initiated with various combinations of gastrin, EGF and GLP-1 agonists, and no doubt other approaches to increasing p-cell mass will soon be developed.
The possibility that pharmacological agents might increase p-cell mass is tantalizing because a decrease in p-cell mass is the root cause of both types of diabetes. Encouraged by the hope that this approach could one day cure diabetes, the diabetes community eagerly awaits the results of these studies. Because of the intensity with which new agents are being tested in clinical trials, answers should emerge in the very near future.
Research to restore the p-cell deficiency of diabetes is being pursued aggressively on many fronts. There is optimism about disparate strategies for generating supplies of Beta-cells sufficient for transplantation, including expansion of primary Beta-cells , embryonic stem cells, pancreatic progenitor cells and transdifferentiation of non-pancreatic cells. Although all of these avenues have promise, each is difficult and none is clearly more promising than another. While embryonic stem cells seem to many, including the authors, to provide the best prospects, and transdifferentiation of liver cells to be a long-shot, there are no guarantees. One would be hard-pressed to predict which approach will cross the finish line first
Attempts to regenerate Beta-cells in vivo are also in the spotlight owing to impressive preclinical results that have led to clinical trials currently in progress. One must again balance optimism with concerns about whether it really will be possible to grow back enough of the missing Beta-cells . While some regeneration has been found in NOD mice with recently diagnosed diabetes, no success has been reported in mice with diabetes of longer duration, in which the Beta-cells are almost entirely wiped out. It seems a most daunting challenge to somehow stimulate what appears to be a very slow rate of neogenesis to provide just the right number or even a sufficient number of Beta-cells . There is also a concern that stimulation of pancreatic duct cells may promote pancreatic cancer. Yet even the most pessimistic naysayers must agree that possibilities for meaningful regeneration do exist and must be explored with the utmost vigor and imagination.
Thus, although expectations and hopes are very high, the difficulty of these approaches remains a sobering reality. Some are so doubtful about the prospects for surmounting these challenges in the near future that they are giving xenotransplantation using porcine islet sources a fresh look. A hard lesson about science is its unpredictability, which makes fore-casts about advances so inaccurate. Nonetheless, with the extraordinary potential of modern science and the many talented investigators working so intently on this critically important goal for diabetes, one must remain hopeful that clinical advances will come sooner rather than later.