In vertebrates, the process of gastrulation takes place very early during the development of the embryo. This process reorganizes the embryo's cells into 3 layers: ectoderm, endoderm, and mesoderm. The ectoderm forms the skin and the central nervous system; the mesoderm gives rise to the cells from which blood, bone, and muscle are derived; and the endoderm forms the respiratory and digestive tracts. The embryonic endoderm, taking the shape of the primitive gut tube, serves as a template for the gastrointestinal tract from which the embryonic pancreas eventually buds. It has been shown that the branching morphogenesis of the pancreatic bud gives rise to the ducts and the acinar components of the gland. Endocrine progenitors, proliferating from the budding ducts, then form aggregates of differentiated cells known as the islets of Langerhans (Figure 1). While the pancreatic acini, composed of cells dedicated to the secretion into the intestine of enzymes that will participate in the digestive process of the ingested food, constitute the exocrine component of the gland, the islets are made up of 4 cell populations, organized in a stereotypical topological order, which constitute instead the endocrine component of the gland. In the islet, the α cells produce glucagon; the β cells, insulin; the γ cells, pancreatic polypeptide; and the δ cells, somatostatin (Figure 1).
Type 1 diabetes is the clinical consequence of the destruction of the insulin-producing β cells of the pancreas, mediated by autoreactive T cells specifically directed against β cell determinants. The loss of the majority of the β cell population, evident at the onset of the disease, requires daily subcutaneous injections of quantities of insulin that should be proportional to the quantity of glucose present in the blood at each moment in time. The physical replacement of the β cell mass constitutes the islet transplantation. Although it was recently demonstrated that islet cell transplants can be performed with greater chances of success than just a few years ago, constrains under which this is clinically possible are still too numerous to allow the broad application of this procedure to permanently cure the disease. The immunosuppressive drug regimen necessary to protect islets from a recurrent autoimmune response and allorejection may, with time, irreversibly damage kidney function, while the process of islet isolation itself, even if drastically improved during the last few years, damages transplantable islets and, consequently, two to three donors are necessary in order to obtain the minimal cell mass sufficient for transplantation into a single recipient.

Figure 1. Cross-section of the pancreas.
The pancreas houses 2 distinctly different tissues. Its bulk comprises exocrine tissue, which is made up of acinar cells that secrete pancreatic enzymes delivered to the intestine to facilitate the digestion of food. Scattered throughout the exocrine tissue are many thousands of clusters of endocrine cells known as islets of Langerhans. Within the islet, α cells produce glucagon; β cells, insulin; γ cells, somatostatin; and δ cells, pancreatic polypeptide – all of which are delivered to the blood.

In considering the utility of stem cells for the regeneration of the β pancreatic cell, we currently face some major questions: in the adult endocrine pancreas, do multipotent progenitors still exist? If so, where exactly are they located? What are the best markers for recognizing and isolating them? What are the stimuli able to activate their differentiation pathway(s)? How large is the time window in which they can still respond to the needs of an aging tissue? In the absence of pancreatic progenitors, are there adult pluripotent stem cells – cells not committed to a specific lineage that may differentiate into all types of cells and tissues, with the exception of extraembryonic tissues – located elsewhere in the body that may have the capability to regenerate β cell mass? If these cells are all lineage-committed precursor cells, can we instead use ES cell lines derived in vitro to substitute the lost β cells of the pancreas?
Increases in β cell mass may occur through increased β cell replication, increased β cell size, decreased β cell death, and differentiation of possibly existing β cell progenitors. It has been shown that occasional endocrine cells can be found embedded in normal pancreatic ducts. However, these cells are rare. The number of these duct-associated endocrine cells physiologically increases as the consequence of severe insulin resistance in obese individuals or during pregnancy. Similar histological changes are observed under conditions of tissue injury and repair after partial pancreatectomy, duct ligation, cellophane wrapping of the gland, or IFN-γ overexpression. Even then within the ducts, only a small number of cells become insulin positive. This suggests that even if some hypothetical precursors exist, the process of formation of endocrine cells out of the islet (neogenesis) would not be a frequently observed property of the duct epithelium. On the other hand, the fact that α and β cells develop from a possibly common, non-hormone-expressing, yet Pdx1-positive precursor (Pdx1 being a transcription factor required for pancreatic development) suggests that all cell types found within the islet may originate from a bona fide, common endocrine progenitor.
These endocrine progenitors may be located close to the duct but may not actually be components of the ductal epithelium. The progenitor cells could be mesenchymal in origin, or they could be cells differentiated from an unknown cell type. If the number of these progenitors is extremely small, lineage analysis becomes very difficult because of the lack of known appropriate markers. Moreover, if these cells are as rare as they appear to be, it becomes difficult to quantify their contribution to normal endocrine cell turnover. It was shown that single murine adult pancreatic precursor cells can generate progeny with characteristics of pancreatic cells, including β cells. These rare (1 in 3,000-9,000 cells) pancreas-derived multipotent precursors (PMPs) do not seem to be bona fide pluripotent ES cells, since they lack, for example, the Oct4 and Nanog markers that direct the propagation of undifferentiated ES cells; nor are these cells of clear ectodermal, mesodermal, or endodermal origin, since they failed to express other markers considered specific for precursors of each of the embryonic cell types. Because, surprisingly, these PMPs also lacked some β cell markers (e.g., HNF3β) as well as ductal epithelium markers (e.g., cytokeratin), but were able to generate differentiation products with neural characteristics along with α, β, δ, and acinar pancreatic cells, the existence of a new and unique ectodermal/endodermal precursor cell present during embryonic development that could persist in adult tissues was proposed.
All of studies, even with their somewhat divergent outcomes, seem to support the conclusion that endocrine precursor cells of some kind exist in the pancreas. They are present not only in the duct, but also within the islets themselves, since both subpopulations were independently used as the source of the isolated single cell precursors. On the one hand, this conclusion supports the working hypothesis of those who propose that pancreatic ductal cells can transdifferentiate into β cells and that this is a physiologic process generally more efficiently activated by increased metabolic demand and tissue injuries.
Precursors of a perhaps unconventional type can be located both in close proximity to and inside the endocrine tissue and that they can be activated by increased metabolic demand or by still-unknown secreted factors, normally able to accelerate the process that guarantees homeostasis of islets of Langerhans under normal conditions. The physiologic equilibrium between lost and newly generated cells can be altered by the action of β cell-specific, autoreactive T cells in instances in which autoimmunity develops. Once T cell killing activity overcomes the regenerative compensatory activity of the organ, the number of functional β cells progressively decreases until they become too few to maintain the glucohomeostasis of the entire body. The time of transition over this metabolic threshold becomes immediately evident with the presentation of the characteristic signs of the clinical onset of type 1 diabetes. During the course of disease, even if the regenerative properties of the pancreas remain functional, the continued presence of diabetogenic, autoreactive T cells consistently nullifies the reparative effort. The fact that these autoreactive T cells remain present in the body of the diabetic patient for a long time is proven by experiments in which healthy islet cells transplanted into syngeneic, long-term diabetic mice or humans were quickly killed by these same autoreactive T cells.
The autoimmune response is successfully averted in the NOD mouse either by directly eliminating the majority of the autoreactive T cells with anti-T cell antibodies or by substituting all or part of the immunocompetent cell repertoire with bone marrow cells obtained from diabetic-resistant donors. The treatment of overtly diabetic NOD mice with anti-lymphocyte serum (ALS) abrogated autoimmunity but achieved only partial clinical remission. Transient treatment of overtly diabetic NOD mice with ALS and exendin-4, a potent insulinotropic hormone that promotes replication and differentiation of β cells in vitro and in vivo, achieved instead complete remission of 88% of the treated animals within 75 days, accompanied by progressive normalization of glucose tolerance, improved islet histology, increased insulin content in the pancreas, and almost normal insulin release in response to a glucose challenge. These results show that exendin-4 synergistically augments the remission-inducing effect of ALS, possibly by promoting differentiation of β cell precursors.
Also, the successful induction of a mixed allogeneic chimerism obtained after transplanting bone marrow from a diabetes-resistant donor into a diabetic animal following a sublethal dose of irradiation is sufficient to block and eventually also revert the systematic invasion and inflammation of the islets by the autoreactive T cells that result in insulitis (Figure 2).

Figure 2. Regeneration of the β cell in diabetic NOD mice.
(A) In NOD mice, the infiltration of autoreactive T cells into the islets of Langerhans (resulting in insulitis) begins at around 4 weeks of age. At 20 to 23 weeks, approximately 85% of female mice are diabetic, i.e., their glycemia is greater than 300 mg/dl. (B) When it is successfully transplanted with bone marrow from a non-diabetes-prone donor and hematopoietic chimerism is established, the NOD mouse no longer show signs of autoimmune activity. However, while there is no more evidence of insulitis in the endogenous pancreas, there is also no sign of insulin production (no red staining). (C) Insulinpositive cells in the islets can be seen to be dividing (yellow arrows); i.e., they are insulin (blue) and BrdU (red) positive. (D) Three to 4 months after bone marrow transplantation, new insulin-positive cells (shown in red) are present throughout the endogenous pancreas. Thus, when the islets transplanted under the kidney capsule in order to maintain euglycemia while regeneration takes place are removed by nephrectomy, the mice remain nondiabetic.

Within the endocrine pancreas, once the insult of autoimmunity is abrogated, the physiologic process of regeneration can continue efficiently, eventually replenishing the population of insulin-producing cells to a number sufficient to maintain euglycemia, thus curing the diabetic recipient (Figure 2D). A subject of ongoing debate is whether either or both the transplanted bone marrow and the cotransplanted β cells are necessary for promoting an efficient regenerative process, independent of their ability to block autoimmunity or preserve euglycemia, respectively. For example, they may secrete factors such as glucagon-like peptides, which are useful in order to sustain an efficient regenerative process. Strong evidence suggests that the hematopoietic precursors present in the bone marrow cell population do not directly participate in the reparative process of the insulin-producing cell population (Figure 3) (32, 33).

Figure 3.
Using a GFP-transgenic mouse as donor, it is possible to observe how the majority of the transplanted bone marrow cells do not directly participate in the regeneration of the endogenous pancreas. As shown here, there are no double-positive (orange) cells in the newly formed islets. The donor cells (green) appear to be located close to possibly existing juxta-ductal precursor cells, which may be activated by bone marrow cell-secreted factors. Insulin-positive cells are red.

A different source of donor cells, for example, the spleen, might be able to block autoimmunity and also provide mesenchymal β cell precursors. However, the hypothesized presence in the mouse spleen of embryonic mesenchymal cells that lack surface expression of CD45 and are able to differentiate into endothelial and endodermal cells remains to be confirmed.
Even when regeneration of the β cells from precursor cells is definitively proven, issues still to be resolved will include the time frame and physiological conditions necessary for regeneration to occur and reach completion, as well as the circumstances that facilitate or limit the regeneration process and the ability of clinicians to promote or avoid these, to more efficiently achieve the desired therapeutic results. In addition, assuming the existence of β cell precursors, we still do not know whether these cells are immortal or subject to senescence, a situation that would leave a narrow window of time for intervention. This matter may be especially relevant in diabetic individuals in whom the reparative process has been repressed by autoimmune surveillance for a long period. If successful immunoregulatory intervention cannot be initiated immediately after the clinical onset of the disease, a full recovery of the endocrine function of the gland via the physiologic regeneration route may become impossible. However, if the regenerative process is proven to be irreversibly compromised at a certain point, it may still be possible to transplant into diabetic patients functional precursor cells from nondiabetic donors or to artificially convert the patient's own cells from other tissues or lineages into insulin-producing β cells.
Using a cocktail of specific antibodies able to recognize a combination of distinct cell surface markers (c-kit^highThy^lowLin^-Sca-1⁺) on a single cell, the investigators physically isolated HSCs from bone marrow and transplanted them into the fumarylacetoacetate hydrolase-deficient mouse, an animal model of fatal hereditary tyrosinemia type I. Liver function in 4 out of 9 mutant mice was restored to near normal (transaminase and tyrosine levels were slightly increased), when, 3 weeks after transplantation, the standard treatment with 2-(2-nitro-4-trifluoromethylbenzyol)-1,3-cyclohexane-dione was discontinued, and these mice survived for an additional 6 months without signs of progressive liver failure or renal tubular damage. When the experiment was interrupted at 7 months after transplantation, 30-50% of the liver mass showed cells expressing donor-derived markers. Similarly, by transferring precursors into the brain of a recipient mouse, purified HSCs differentiate into Purkinje neurons. Like HSCs, once transduced with a retrovirus to express bone morphogenetic protein 4 (BMP4), muscle-derived cell (MDC) precursors were able to dramatically improve the healing of a spontaneously irreparable bone fracture.
These data suggest that the signals sent via host-secreted factors or by cell-to-cell contacts are powerful enough to guide the transplanted precursors to differentiate into the same type of cells surrounding them, even across different lineages.
Instead of promoting cell-to-cell fusion in an effort to restore certain physiologic functions lost with the death of specific cells, it has been considered perhaps more efficient to transfect cells belonging to a certain lineage with genes able to convert them into cells carrying the characteristics of those lost, even if they belong to a completely different lineage. In particular, in the field of diabetes research, gut K cells of the mouse were induced to produce human insulin by transfecting the human insulin gene linked to the 5'-regulatory region of the gene encoding glucose-dependent insulinotropic polypeptide (GIP). Also, research demonstrating that hepatocytes transfected with the Pdx1 gene under the control of the rat insulin 1 promoter were able to produce insulin attracted significant attention and has inspired new hope. In these studies, sufficient levels of insulin were secreted to satisfy the needs of a diabetic mouse, which, when treated, became and remained steadily euglycemic. These studies, however, have not yet been successfully repeated by other groups.

Our young diabetic patients must check their blood glucose levels and be injected with insulin at least 4 times a day. Concurrently, they live with the constant threat of incidents secondary to unpredictable acute hypoglycemic episodes and the ever-present worry of chronic complications. Although human ES cell research carries with it enormous scientific potential in the treatment and possible cure of many diseases, in the near future, the advances in the realm of immunoregulation may precede those in the stem cell arena. To me, finding safe ways to block autoimmunity seems to be the first goal we should achieve in order to give our patients a reliable solution to their heavy, lifelong burden, since it is a prerequisite for both the efficient use of an ES cell-based therapy and the reestablishment of euglycemia capitalizing on the pancreatic regenerative pathway.