Pancreatic Progenitor Cells—Recent Studies

Abstract

Past studies of pancreatic progenitor cell biology relied mostly on histological analyses. Recent studies, using genetic labeling and tracing of progenitors, direct single cell analyses, colony assays, and enrichment of the minor population of progenitor cells through the use of cell surface markers, have strongly suggested that pancreatic progenitor cells with various frequency and lineage potentials, including the multipotent progenitors that give rise to endocrine, exocrine, and duct cells, exist in the developing and adult pancreas. In this review, it is therefore proposed that pancreatic progenitor cells may be organized in a hierarchy, in which the most primitive pan-pancreatic multipotent progenitors are at the top and rare, and the monopotent progenitors are at the bottom and abundant. This model may explain why only drastic injuries lead to effective activation of the progenitor cell compartment of the higher hierarchy, whereas under steady state, pregnancy, and milder injuries, recruitment of preexisting mature cells or their immediate monopotent progenitors could be sufficient to restore metabolic homeostasis. It is also proposed that the morphologically defined ductal cells are likely to be functionally heterogeneous and that endocrine progenitor cell activity should be determined based on functional analyses rather than histological locations.

A FUNCTIONAL PANCREAS consists of two types of tissue: exocrine and endocrine. The exocrine tissue mainly consists of acinar cells, which secrete bicarbonate and digestive enzymes. These secretions are collected by the pancreatic ductal system, which begins with centroacinar cells that are directly in contact with acinar cells. The prolongation of the terminal ducts, or alveoli, are lined by centroacinar cells and gradually merge into a main duct that drains into the duodenum. The endocrine tissue is organized as islets and contains cells that secrete glucagon, insulin, somatostatin, pancreatic polypeptide, or ghrelin (1,2). These endocrine hormones are released directly into the blood stream in response to metabolic signals.

Recently there has been an intense interest in identifying pancreatic stem or progenitor cells, especially the endocrine progenitor cells, for the purpose of replacement therapy of type I diabetes, a disease in which the insulin-secreting β-cells are specifically destroyed by autoimmunity. The pancreatic progenitor cell field has slowly evolved over time but has made exciting advances in recent years. This minireview will present a focused perspective of how this field has advanced over the past 10–15 yr by examining the experimental data available so far and will provide some suggestions as to how this field may move forward in the near future.

Progenitor Cell Studies Based on Histological Analysis

The first notion of what may constitute an endocrine progenitor population in the pancreas came from histological studies of developing rodent embryos and regenerating adult pancreases under certain injury models (reviewed in Ref. 3). In the pancreatic progenitor cell field, the terms progenitor cells and stem cells are used interchangeably by some investigators, although the ability of any cell type in the pancreas to self-renew, which is a defining property of true stem cells, has not been vigorously tested.

The murine embryonic pancreas, which consists of dorsal and ventral buds, develops from the endoderm-derived duodenal region of the foregut (1,4). The pancreatic dorsal bud evaginates from the foregut at embryonic day (E) 9.0 of the mouse embryo, and the ventral bud evaginates at E9.5. Histologic and immunohistochemical analyses demonstrate that ductal structures and their branches form by E13.5, and by E17.5 prominent budding islets can be seen adjacent to, or in the vicinity of, the ducts (5). The single cells in the newly formed islets often express multiple combinations of endocrine hormones both in embryos (5,6,7,8,9,10) and adults (11). Based on these findings, it was proposed that embryonic endocrine progenitor cells originate from the ducts and express multiple hormones before becoming terminally differentiated.

Similar to the embryonic pancreas, the ductal cells in adults are thought to be the endocrine progenitor cells, based on histological studies after pancreas injury. In general, long-term regeneration of tissue after injury is indicative of stem cell activity. Several methods that induce pancreas damage have shown endocrine progenitor cell activity in the mature pancreas, including 90% pancreatectomy in rats (12,13), wrapping a part of the hamster or monkey pancreas with cellophane (14,15) or a murine autoimmune destruction model for β-cells in which γ-interferon, expressed under the control of the insulin promoter, led to inflammation-induced loss of islets and acinar cells (16). Pulse-chase experiments using bromodeoxyuridine immunohistochemistry (12,13,16) or tritiated thymidine autoradiography analyses (14) revealed that proliferation occurs first in the ducts and then in the islets (12,14), and extensive formation of ducts and budding of islets in the vicinity of the ducts is observed (12,13,14,15,16). Together, these findings led to the conclusion that ductal cells are the progenitors of islets.

Progenitor Cell Studies Based on in Vitro and in Vivo Lineage-Tracing Analysis

Extrapolating progenitor cell activity with morphological and histological analyses is at most inconclusive, as demonstrated by studies using lineage-tracing strategies and single-cell analyses (to be described in the following sections). Using lineage tracing, or more precisely progenitor tracing, the first theory that was challenged was the germ layer origin of the pancreatic islets. Because endocrine islets express many neuronal genes, it was thought that the endocrine cells were of neuroectodermal origin and migrated into the pancreatic anlage. This hypothesis was disproved through the use of chick/quail (17,18) and transgenic mouse (19) tissue recombination studies. In the mouse study, early (E11.5) embryonic pancreatic epithelia that were labeled with constitutively expressed β-galactosidase were dissected. These epithelia gave rise to both amylase- and insulin-expressing cells in an in vitro explant assay (19), providing convincing evidence that exocrine and endocrine cells share a common origin of pancreatic endoderm.

The second theory that was challenged is the origin of the adult endocrine cells within the developing endocrine system. The start of endocrine commitment, which is known as the first wave of endocrine differentiation, occurs when glucagon-positive cells develop at the dorsal bud (E9.5 of the mouse embryos). This is followed by the emergence of a population that simultaneously expresses insulin and glucagon at E11.5 (5,6,10). It was previously thought that this first wave of insulin- and glucagon-expressing cells was the progenitor of later endocrine lineages (5,6,10). However, using genetic lineage ablation (20) and tracing (21) approaches, Herrera et al. found that this first wave of insulin⁺glucagon⁺ cells does not give rise to the majority of the β-cells that later populate the islets in a second wave of differentiation, which occurs at approximately E13.5.

Although studies on the cellular origins of pancreatic progenitor cells have progressed at a relatively slow pace, much has been learned over the past decades concerning the genetic regulation of the developing pancreas by transcription factors. Many transcription factors are important for the morphogenesis of pancreatic buds and the formation of endocrine islet cells (reviewed in Ref. 22). In particular, the homeodomain-containing pancreatic and duodenal homeobox gene (Pdx)-1 and the basic helix-loop-helix-containing neurogenin3 (Ngn3) are expressed early during the development of the pancreas [Pdx-1, E8.5 (5,23); Ngn3, E9.5 (24)] and are required for the formation of the entire (23,25) and the endocrine (24) pancreas, respectively.

Gu et al. (26) were first to design an inducible marking method that uses double transgenic mice carrying a reporter gene and either a Pdx-1 promoter-driven cyclization recombinase (cre)-estrogen receptor (Pdx-1-creER) or Ngn3-creER transgene. This allows a fraction of progenitor cells of interest to be labeled by injecting a small molecule inducer (tamoxifen) to the mice at a given developmental time point. The fate of the marked progenitor cells is determined by the presence of a reporter gene in cells of adult tissues. Consistent with the knockout mouse studies (23,24,25), Pdx-1-expressing progenitor cells, when labeled before E12.5, gave rise to adult exocrine, endocrine, and ductal cells but not pancreatic endothelial and smooth muscle cells, whereas marked Ngn3-expressing progenitor cells (E8.5–13.5 and 3–8 wk old) gave rise to only islet cells in the adults (26). Surprisingly, Pdx-1-expressing cells labeled after E12.5 gave rise only to exocrine and endocrine, but not ductal, lineages in adult pancreas (26), suggesting that the embryonic progenitors of adult ductal cells are committed early (before E12.5) and that restriction of the lineage potential of Pdx-1-expressing cells occurs at later time points (after E12.5). Use of carboxypeptidase A1 (Cpa1)-cre-ER/reporter double-transgenic mice to label the subdomains of the early (E9.5–12.5) Pdx-1-expressing cells further demonstrated that the Cpa1⁺Amylase⁻Pdx-1⁺ cells are the population capable of giving rise to multiple pancreatic lineages (27). Together these results unequivocally demonstrate a qualitative difference in lineage potential between the earlier vs. the later Pdx-1⁺ or Cpa1⁺ cells, and between the Pdx-1-expressing and the Ngn3-expressing cells. However, it is not clear from these studies whether single multipotent progenitor cells and/or multiple monopotent progenitor cells are labeled because a cohort of cells, not single cells, is simultaneously marked in this system.

In the same study, Gu et al. (26) also used Ngn3-cre but not the inducible Ngn3-creER, double-transgenic mice to mark all of the intermediate endocrine progenitor cells, or their ancestors, that had already expressed Ngn3 and were destined to become hormone-expressing cells. All of the adult islets, but not ducts or acinar cells, expressed the reporter gene, as expected. Some of the embryonic (E14.5–E16.5) ductal structures had mosaic or no expression of the reporter gene, suggesting that only some, but not all, of the embryonic duct-like cells are endocrine progenitors. Thus, this study (26) is the first to demonstrate that embryonic ductal cells are functionally heterogeneous. Consistent with this idea, Ngn3-expressing cells are found in some, but not all, of the CD133-expressing E13.5 pancreatic ductal epithelia and can give rise to hormone⁺ cells in vitro (28).

The role of adult pancreatic ductal cells in islet regeneration has been controversial in recent years. Progenitor tracing methods, such as in vivo lineage-tracing with inducible rat insulin II-creER/reporter double-transgenic mice (29) and a DNA analog-based technique (30), were used to study whether β-cells or specialized progenitor cells contribute to the new and replicating β-cells under steady state (29,30), pregnancy (30), 50% (30) or 70% (29) pancreatectomy, or 70–80% β-cell-specific ablation (31). In these studies, it was shown that the majority of the new β-cells came from preexisting insulin-expressing cells. Based on these observations, Dor et al. (29), Nir et al. (31), and Teta et al. (30) concluded that there is little or no progenitor cell activity in the adult pancreas.

However, this idea has been challenged recently by a study that used the duct ligation injury model to show significant activation of the adult endocrine progenitor cell population (32). For mechanisms that are not clear, but may be attributed to the microenvironment or niche, this injury method is the only model shown to date that demonstrates activation of Ngn3⁺ endocrine progenitor cells in the adult pancreas. These Ngn3⁺ cells do not arise from de-differentiation of committed endocrine cells and are able to generate endocrine cells, similar to the embryonic Ngn3-expressing cells (32). In addition, the activated Ngn3⁺ cells are scattered among the ductal structures, the islet structures that are Ngn3⁺hormone⁻, and other pancreatic tissues. This may explain why in certain prior studies, the progenitor cell activity could be found in the islet- (33), ductal- (34), or the epithelial cell-enriched (35) populations. Furthermore, the Ngn3-expressing cells do not colocalize with the amylase⁺ cells (32), which may suggest that cell-cell interaction is important and that acinar cells may inhibit the activation of the endocrine progenitor cells or the differentiation of these endocrine progenitor cells from their progenitor cells. This hypothesis is supported by studies showing that, in rats, islet mass increased only when acinar cells were specifically eliminated by a copper-deficient diet (36,37), and the increased mass of those islets was maintained when acinar cells were regenerated by feeding the animals a normal diet (36,38). Taken together, these results suggest that, similar to embryonic ductal-like cells, adult ductal cells are also heterogeneous and that endocrine progenitor cells in the adult pancreas are likely associated with several histological locations.

Progenitor Cell Studies Based on Population vs. Single Cell Analyses

In the hematopoietic system, the most immature self-renewing stem cells and the committed multipotent progenitor cells are usually a minor population, whereas the intermediate oligopotent progenitor cells are a larger population, and the most mature monopotent progenitor cells are the major population (39). This progenitor cell hierarchy is believed to be responsible for an efficient response to a variety of blood cell damage and need signals.

One may hypothesize that a similar progenitor cell hierarchy (Fig. 1), with the very rare multipotent stem cells at the top and the abundant monopotent progenitor cells at the bottom, is a preferred system for the adult pancreas to efficiently respond to injury and metabolic need signals. Then, perhaps under steady state, specific physiological conditions [e.g. postnatal growth (40) and pregnancy (41,42)] and milder injuries (e.g. 50–70% pancreatectomy that leads to only 20–30% recovery of the total pancreatic mass or 70–80% β-cell ablation), recruitment to the cell cycle of mostly preexisting insulin-expressing cells or their immediate lineage-specific monopotent progenitors from the lower hierarchy may be sufficient to restore the glucose homeostatic balance. The observations that both duct ligation (43) and 90% pancreatectomy, a method that could result in 80% recovery of the total pancreatic mass (44), eliminate a majority of acinar and endocrine cells of the affected pancreatic lobes may suggest that other pancreatic cell types play integral roles in activating endocrine progenitors from the higher compartment of the hierarchy.

Figure 1.

A model for the pancreatic development hierarchy. The existence of self-renewing pancreatic stem cells (PSC) has not been vigorously tested. In this model, the early multipotent progenitor cells (in blue) are rare (45,46) and are at the top of the hierarchy. For example, murine neonatal multipotential progenitors can be enriched by sorting with c-Met⁺Flk-1⁻c-Kit⁻CD45⁻TER119⁻ cells, which comprise only 0.59% of the total pancreatic population (45).The oligopotent progenitors (in red) (45) are larger in number and are at the middle of the hierarchy. The late monopotent lineage-restricted progenitor cells (in green) are abundant and are at the bottom of the hierarchy. Under steady state (30), specific physiological conditions [e.g. postnatal growth (40) and pregnancy (30)] and milder injury conditions [e.g. 50–75% pancreatectomy (30,48) and 70–80% β-cell ablation (31)], replication of existing mature effector cells (in orange), and/or recruitment of lineage-specific monopotent progenitor cells to the cell cycle and subsequent production of new effector cells, may be sufficient to meet the metabolic need. Progenitor cells from the higher hierarchy compartment may be activated in cases of more severe injury involving multiple pancreatic cell lineages of the affected lobes [e.g. duct ligation (32) and 90% pancreatectomy (44)].

To distinguish the activities of the rare progenitor cells from the vast majority of other cell types, we must enrich for these infrequent cells and perform subsequent functional analyses. Two recent reports have used several cluster of definition cell surface markers to enrich for pancreatic progenitor cells and identify the murine neonate and adult multipotent progenitor cells as c-Met⁺Flk-1⁻c-Kit⁻CD45⁻TER119⁻ (45) and E13.5 fetal endocrine progenitor cells as CD133⁺CD49f^low (28).

To unequivocally discern the lineage potential of various progenitor cells, single-cell analysis and/or colony assays are required. Much of the ambiguity of the current progenitor cell studies, including those claiming transdifferentiation and dedifferentiation, stems from the use of a mixed population, and not single cells, from various parts of the pancreas or other sources in these experiments. Using single-cell sorting (45) or plating of single cells by limiting dilutions (46), two recent studies have unequivocally demonstrated the existence of adult murine pan-pancreatic multipotent progenitors that can give rise to endocrine, exocrine, ductal, and possibly other lineages in vitro and in vivo. In addition, it is noteworthy that some single progenitors are oligopotent and give rise to several, but not all, of the pancreatic lineages in vitro (45). We have contributed to this field by establishing a quantitative colony assay using semisolid media that allow dissociated single β-cell-like progenitors derived from murine embryonic stem cells to form insulin-expressing colonies in vitro (47). These insulin-expressing colony-forming progenitor cells could be enriched by the selection of cells that express an Ngn3 promoter-driven enhanced green fluorescent protein reporter gene (47).

Conclusion and Future Directions

Experimental data available to date have convincingly demonstrated that, first, the very early murine embryonic pancreatic Pdx-1⁺Cpa1⁺ cells (26,27), as a population, have the potential to give rise to all lineages, whereas the embryonic and adult Ngn3-expressing cells have the potential to give rise to only endocrine pancreas (26), supporting a view that sequential activation of transcription factors is accompanied by progressive restriction of the lineage potentials of pancreatic progenitor cells. Second, multipotent pan-pancreatic progenitor cells do exist as a very rare population in normal adult murine pancreas (45,46). In addition, oligopotent (45) and monopotent β-cell-like progenitor cells (47) are found in populations of murine adult and embryonic stem cell-derived cells, respectively, suggesting that a spectrum of progenitor cells with various lineage potentials exist in the pancreas. Third, the adult murine pancreas is activated under certain drastic (32), but not milder (48), injury models that lead to activation and differentiation of endocrine progenitor cells that expressed Ngn3. Lastly, the morphologically defined ductal epithelia are likely to be functionally heterogeneous (26,28,32), and endocrine progenitor cell activity should be determined based on functional analyses rather than histological locations.

Several important questions remained to be answered in future studies. Can various embryonic and adult pancreatic progenitor cells, especially the multipotent progenitors, be identified in vivo and in humans? What are their frequencies in steady-state and injury conditions? How can the rare progenitor cells be effectively enriched, isolated, and cultured at the single-cell level? Do the multipotent progenitor cells self-renew over a prolonged period of time in vitro and in vivo? What soluble factors, cell-to-cell, and cell-to-matrix signals are required to regulate progenitor cell activation and how? What are the cellular and molecular mechanisms that control the steps of lineage commitment of the primitive pancreatic progenitor cells? To answer these questions will take many years, but a precise knowledge of the pancreatic progenitor cell biology will translate to a better defined and efficient clinical treatment for type I diabetes patients in the future.