Generation of Beta Cells from Acinar Cells
It is becoming increasingly clear that differentiated adult somatic cells retain the capacity to be reprogrammed into other cell types. In the case of the pancreas, a switch from an acinar to a β-cell phenotype in vitro can be induced by soluble agents, such as growth factors and cytokines. We found that the combination of epidermal growth factor and leukemia inhibitory factor stimulated the transdifferentiation of rat acinar cells into β cells in culture. The transdifferentiation, or cellular reprogramming, appears to recapitulate embryonic events, such as expression of the transcription factor NGN3, which is characteristic of pancreas proendocrine progenitor cells. The NOTCH-signaling pathway, whose activity is normally restricted to embryonic pancreas development, is also reactivated during transdifferentiation. Inhibition of this pathway in the same experimental model leads to further stimulation of β-cell neogenesis from adult acinar cells. Engraftment of the acinar-derived β cells results in correction of glycemia in alloxan-diabetic mice. The phenotype of the transdifferentiated cells is stable in vivo, resulting in normal and safe function following transplantation. This approach opens ways for β-cell replacement therapy by transplantation or regeneration.
Recent work by Takahashi et al. on reprogramming of adult fibroblasts to pluripotent embryonic stem cells (Takahashi et al., 2007) has sparked an increased interest in the capacity of adult somatic cells to be reprogrammed into a different phenotype. Transdifferentiation, the reprogramming of “terminally” differentiated cells into another differentiated phenotype, has been studied for many years in different types of tissues (Brockes and Kumar, 2002; Thowfeequ et al., 2007; Tsonis et al., 2004). This process is also referred to as metaplasia, cell plasticity, or lineage switching. It is important to distinguish cell fate switching as a consequence of nuclear transfer to a new cytoplasm or as a result of gene transfer from cell fate changes induced by cell environment. In this chapter we focus on the latter type of transdifferentiation.
Transdifferentiation usually involves a dedifferentiation intermediate stage. Cells in this stage may enter the cell cycle and either proliferate before activating the new differentiation program or proceed directly into differentiation without replication.
A decade ago we proposed transdifferentiation as an alternative to stem cells for β-cell neogenesis (Bouwens, 1998). Although most studies suggest that in the adult pancreas new β cells are generated from duct cells (Bonner-Weir and Weir, 2005), the idea that islet cells could also originate from pancreatic acinar cells is quite old. The French histopathologist E. Laguesse, who was the first to introduce the term “islets of Langerhans” for the pancreatic endocrine microorgans, suggested this possibility (Laguesse, 1896). Later histopathological studies confirmed the existence of transitional or intermediate forms between acinar and islet cells under particular experimental or pathological conditions and in many animal species.
So-called “mixed cells” or “acinar-islet cells,” which contain both exocrine and endocrine granules, have been reported by many electron microscopists (Leduc and Jones, 1968; Setalo et al., 1972). Another piece of evidence supporting the occurrence of acinar-to-β-cell transdifferentiation was provided by immunohistochemical studies describing cells that coexpress the acinar enzyme amylase and insulin following experimental pancreatic injury in rats (Bertelli and Bendayan, 1997; Lardon et al., 2004a). One of these procedures, pancreatic duct ligation, consists of a surgical intervention in which the ducts draining the tail part of the pancreas are obstructed by a silk thread, provoking a pancreatitis-like phenomenon in the splenic part of the organ. However, a recent genetic lineage-tracing study using the elastase promoter, which allows permanent marking of cells of acinar origin, failed to demonstrate an acinar origin of β cells in mice that were subjected to three different procedures of pancreatic injury, including duct ligation (Desai et al., 2007). This study employed the elastase-CreERT transgene, which drives tamoxifen-inducible acinar-cell-specific DNA recombination in a reporter transgene.
In contrast, in another recent genetic lineage-tracing study, in-vivo virus-mediated transfer of three genes encoding transcription factors that play key roles in normal β-cell development, NGN3, PDX1, and MAFA, was found to robustly reprogram acinar cells into β-like cells (Zhou et al., 2008). The resulting cells expressed genes essential for β-cell function and were able to ameliorate hyperglycemia in experimental mouse models of diabetes. Although it is uncertain whether these cells can sustain a precise glycemic control as well as normal β cells, these findings hold promise for future treatment of diabetes mellitus. The newly-formed β cells remain scattered within the pancreas and do not form new islet structures, a feature that may affect their survival and functionality. However, this may be less problematic in humans, since scattered β cells are found within the normal human pancreas (Bouwens and Pipeleers, 1998). Another challenge to overcome in this approach is the elimination of adverse effects associated with intraparenchymal injection and virus infection (German, 2008).
Genetic lineage tracing is likely to remain a powerful tool in ascertaining the origin of new β cells from acinar cells in future in-vivo experimental models. There is also an obvious need for defining the extracellular signals involved in normal differentiation of acinar cells, which may lead to new approaches for their reprogramming to β cells.
7.2 In-Vitro Dedifferentiation of Acinar Cells
Thus, from a developmental point of view acinar cells may be more attractive candidates for endocrine progenitors than duct cells, although the latter have been studied much more extensively in this respect. Moreover, adult acinar tissue displays a remarkable plasticity in vitro. When acinar cells are isolated from adult pancreas, they undergo a spontaneous dedifferentiation into ductlike cells. This has been shown in different species (Arias and Bendayan, 1993; Vila et al., 1994; Rooman et al., 2000) and has also been confirmed by genetic lineage tracing in Elastase-CreERT mice (Means et al., 2005). We showed that during this dedifferentiation rat acinar cells lose their exocrine characteristics, such as expression of digestive proenzymes, within the first days in culture. They also downregulate the acinar transcription factors p48/PTF1A and MIST1. On the other hand, they activate expression of genes characteristic of embryonic-fetal pancreatic progenitor cells, such as the gene encoding the transcription factor PDX1 and genes expressed in both ductal cells and embryonic-fetal pancreatic epithelium, such as the one encoding cytokeratin-20 (Rooman et al., 2000). Another embryonic/ductal activated gene encodes for the transcription factor SOX9, which is also considered a pancreatic progenitor marker. The cells also activate expression of receptors for growth factors such as gastrin (Rooman et al., 2001), vascular endothelial growth factor (VEGF) (Rooman et al., 1997), and leukemia inhibitory factor (LIF) (De Breuck et al., 2006), which may play roles in pancreas development. These factors were found to affect proliferation or differentiation of the dedifferentiated cells in culture. These cells also express receptors for netrin, a secreted factor that plays a role in guiding migration of pancreatic progenitor cells during development (De Breuck et al., 2003). Another interesting receptor that is reexpressed by dedifferentiated acinar cells is NOTCH, along with its downstream target HES1 (Miyamoto et al., 2003; Rooman et al., 2006). Interfering with the NOTCH-signaling pathway in mouse acinar cell cultures had an inhibitory effect on the dedifferentiation process (Miyamoto et al., 2003). During early pancreas development NOTCH signaling maintains the progenitor population via HES1, which blocks expression of the cell cycle inhibitor p57 (Apelqvist et al., 1999; Murtaugh et al., 2003; Georgia et al., 2006) and prevents differentiation into specific lineages. Acinar cell dedifferentiation can be modulated by histone deacetylase inhibitors, such as sodium butyrate and trichostatin-A, and by the ADP-ribosylase inhibitor nicotinamide (Rooman et al., 2000), indicating that chromatin remodeling is involved in the observed changes in gene expression.
The reactivation of embryonic transcription factors, receptors, and signaling pathways supports the hypothesis that acinar cells dedifferentiate into an “uncommitted” progenitor state. This state may be induced by the absence of differentiation maintenance signals, which are normally present in the adult pancreatic environment and are lost during cell isolation. Alternatively, tissue dissociation or tissue injury may activate processes that induce these changes in gene expression. It was recently reported that the metalloproteinase MMP7 is involved in the activation of the NOTCH pathway during acinar dedifferentiation (Sawey et al., 2007). Minami et al. recently reported on the role of cell-to-cell contacts during the in-vitro dedifferentiation of acinar cells (Minami et al., 2008). It is known that alterations in tissue architecture and cellular communications can fundamentally change the differentiation state of cells, and that the response to reprogramming stimuli can differ significantly when cells are taken out of their normal environment. Minami et al. showed that the loss of E-cadherin-mediated cell-to-cell contact was an important step during acinar cell dedifferentiation into a progenitor-like state and that its recovery played an important role during in-vitro transdifferentiation of acinar cells into endocrine cells. PI3-kinase plays an essential role in transducing the E-cadherin signal.
Acinar cells can also transdifferentiate into ductal cells in vivo (Lardon and Bouwens, 2005). An acinar-to-adipocyte transdifferentiation in vivo has also been reported (Bonal et al., 2008). What may be the biological role of this plasticity? One possibility is that in cases of pancreatic injury, such as in ductal obstruction or some forms of pancreatitis, acinar cells dedifferentiate or die to prevent damage caused by activation of exocrine enzymes. In acute pancreatitis, the transient dedifferentiation of acinar cells is well-documented, with temporary acquirement of fetal-ductal characteristics (Jensen et al., 2005). The reactivation of the NOTCH pathway in this model is essential for restoration of the acinar differentiation (Siveke et al., 2008). Another interesting possibility is that this acinar cell plasticity is an evolutionary relic. It was recently shown that when Xenopus tadpoles undergo metamorphosis, their exocrine pancreas remodels by massive dedifferentiation of exocrine acinar cells into a progenitor cell phenotype. The acinar cells lose their zymogen granules and activate Pdx1, Notch1, and Hes1 gene expression, after which they redifferentiate into acinar and duct cells. In this way, the ductal system, which is absent in the tadpole, is reformed in the adult amphibian pancreas (Mukhi et al., 2008). These changes in gene expression and differentiation are similar to those observed in rodent acinar cells during dedifferentiation (see above).
7.3 In-Vitro Transdifferentiation of Acinar Cells into Beta Cells
Our group was the first to report that transdifferentiation of primary rat acinar cells into endocrine β-like cells can be induced ex vivo with soluble factors, epidermal growth factor (EGF), and LIF (Baeyens et al., 2005). A second group, led by S. Seino, reported on the induced transdifferentiation of murine acinar cells into β-like cells in the presence of EGF and another differentiation-inducing agent, nicotinamide (Minami et al., 2005). Whereas the latter group used genetic lineage tracing to demonstrate the acinar origin of the newly generated murine β cells, we developed a lectin-tracer-based method to confirm the acinar origin of such cells in other species, such as rats (Baeyens L et al., unpublished results). Lectin tracing was performed by intraparenchymal injection of a fluorescent wheat germ agglutinin (WGA) before pancreas dissociation and cell isolation. In this way, WGA labels only the acinar cells and none of the other cell types, such as duct cells, centroacinar cells, islet cells, blood vessels, or mesenchymal cells. The label is maintained in the cytoplasm and is stable for more than 10 days. Following in-vitro transdifferentiation, newly formed β cells contained this acinar tracer, which demonstrated their acinar origin. It has also been shown that β cells can be generated from acinar tissue collected from mouse models of type 1 diabetes, demonstrating that metabolic disorders do not interfere with the capacity of acinar cells to transdifferentiate into β cells (Okuno et al., 2007).
Transdifferentiation was also reported for the acinar-derived rat tumor cell line AR42J. In the presence of hepatocyte growth factor, glucagon-like peptide-1, or the combination of activin A and betacellulin, these cells could be converted into insulin-producing cells (Mashima et al., 1996; Mashima et al., 1996a; Zhou et al., 1999). Interestingly, the embryonic proendocrine transcription factors NGN3 and PAX4 were induced in these cells (Zhang et al., 2001; Kanno et al., 2006). This cell line was also shown to transdifferentiate into hepatocyte-like cells in the presence of dexamethasone (Tosh et al., 2002) or following expression of the liver-specific transcription factor C/EBPβ α (Shen et al., 2000).
7.4 Mechanism of Acinar-to-Beta-Cell Transdifferentiation
We investigated the molecular mechanism involved in transdifferentiation of acinar cells into β cells induced by EGF and LIF. LIF exhibits a wide range of biological activities, including induction of proliferation and differentiation of different cell types (Kurzrock et al., 1991). It is also known to maintain pluripotency of murine embryonic stem cells. During neurogenesis, LIF regulates the differentiation of neural precursors into neurons or glial cells (Viti et al., 2003). In the latter study, it was also shown that EGF increases the competence of LIF as an inducer of astrocyte differentiation. Furthermore, LIF is considered a key signal for injury-induced nerve regeneration in the adult (Niwa et al., 1998; Chambers and Smith, 2004). We demonstrated an increased expression of LIF and its receptor in injured pancreas tissue (De Breuck et al., 2006). In light of the many similarities between neurogenesis and pancreatic islet formation, it seems likely that signals regulating differentiation in the neural system may exhibit a similar effect in the pancreas.
In our in-vitro transdifferentiation model, a strong inhibition of β-cell neogenesis can be observed after blocking EGF or LIF signal transduction, namely by inhibiting the EGF receptor or the JAK2 or STAT3 mediators of cytokine receptor signaling. The strongest effect was achieved by preventing STAT3 activation in this model. The latter observation is in accordance with the effects described during neuronal development, in which EGF signaling was shown to amplify the responsiveness of LIF-mediated signaling at the level of STAT3 activation (Baeyens et al., 2006).
In EGF/LIF-treated rat acinar cells (Baeyens et al., 2006), as well as in EGF/nicotinamide-treated mouse acinar cells (Minami et al., 2005), reexpression of NGN3 was noted during transdifferentiation (Fig. 7.2). This transcription factor is not expressed postnatally, but is known to be crucial for the development of endocrine cells in the embryonic pancreas during secondary transition (Gradwohl et al., 2000; Schwitzgebel et al., 2000; Gu et al., 2002). We found a transient upregulation of Ngn3 mRNA and protein expression immediately preceding, and partially overlapping, the expression of insulin (Baeyens et al., 2006). The upregulation of NGN3 and the ensuing expression of insulin and other β-cell markers could be significantly inhibited by specific chemical inhibitors of JAK2 and STAT3 signaling (Baeyens et al., 2006). RNA interference with specific siRNA to silence Ngn3 expression led to a strong inhibition of β-cell neogenesis (Baeyens et al., 2006). These findings demonstrate that NGN3 expression is just as necessary for endocrine differentiation in adult pancreatic cell reprogramming as it is for islet development. It is noteworthy that following EGF/LIF-induced differentiation most NGN3-expressing cells gave rise to insulin-positive cells and that only very few cells expressing glucagon or other islet hormones appeared in the culture. This might be explained by the upregulation of specific transcription factors downstream of NGN3.
The signaling cascade leading to final endocrine cell differentiation is tightly controlled by the balance among the different cell-specifying transcription factors. If the relative abundance of one set of transcription factors outweighs the others, the equilibrium within the endocrine progenitor pool shifts toward one particular cell type. An example is given by the opposite actions of PAX4 and ARX (Fig. 7.1). Although originating from the same NGN3-positive progenitor, an excess of PAX4 will push the cells toward the β/δ cell lineage. On the other hand, if ARX outweighs PAX4, the progenitor cell will progress into the α/PP cell lineage. Beta-cell neogenesis induced by EGF and LIF is characterized not only by a reexpression of NGN3, but also by the specific upregulation of its downstream target PAX4 (unpublished observations). If we take into account that no changes were observed in ARX expression, this may help explain why this treatment generates predominantly β cells.
Immediately preceding NGN3 expression in EGF/LIF-treated acinar cells a transient increase in Hnf6 transcript and protein levels was noted. HNF6 is another transcription factor associated with pancreas progenitor cells during embryonic development and is known to transactivate the Ngn3 promoter prior to initiation of endocrine differentiation (Jacquemin, et al., 2000). The appearance of HNF6 is accompanied by a striking downregulation of the NOTCH signaling factor HES1 and a rapid increase in HES6. Although not demonstrated in the pancreas, HES6 is a known inhibitor of HES1 during neuronal development (Bae et al., 2000). HES1 expression in embryonic pancreas cells is induced by neighboring cells expressing NOTCH ligands, such as DELTA or JAGGED. In cells expressing the NOTCH receptor HES1 represses Ngn3 and thus prevents endocrine differentiation. Instead, these cells remain in the progenitor pool and may differentiate to exocrine cells at a later point in time when NOTCH and HES1 are downregulated. This process, which is also known as lateral inhibition, is responsible for restricting the number of pancreatic epithelial cells that differentiate into endocrine cells (Apelqvist et al., 1999; Gu et al., 2002; Murtaugh et al., 2003). Therefore, the NOTCH–HES1 pathway can be considered an endocrine–exocrine gatekeeper. This gatekeeper function also operates during adult acinar cell transdifferentiation. Indeed, in acinar cell cultures treated with EGF and LIF, only 10% of the cells can be reprogrammed into β cells and exhibit NOTCH–HES1 signaling activity (Baeyens L et al., unpublished results). Hyperactivation of NOTCH signaling by an excess of ligands enhances antiendocrine signaling and renders the dedifferentiated acinar cells insensitive to the growth factor treatment. In contrast, specific inhibition of active NOTCH signaling by RNA interference releases the cells from this inhibition and amplifies the potential of the cells to respond to proendocrine stimuli. This effect in manifested in a pronounced increase in the number of cells adopting a β-cell phenotype. The same effect is obtained by treatment with a soluble form of the extracellular domain of NOTCH1 in the culture medium. By competitively inhibiting the interaction of NOTCH ligands with their receptor on acinar cells, this treatment increases the fraction of insulin-positive cells from 10% to about 33%. Thus, at least one-third of acinar cells can be reprogrammed into β cells by culturing them in the presence of EGF, LIF, and soluble NOTCH1. In combination with RNA interference an even higher efficiency can be obtained. Thus, transdifferentiation from acinar cells represents a very robust method for β-cell neogenesis.
Taken together, after a first step of partial dedifferentiation in vitro, adult acinar cells become responsive to agonists of the JAK2–STAT3 signaling pathway, such as EGF and LIF. This pathway may activate HNF6 and, subsequently, NGN3 expression, resulting in a cascade of transcription factors that determine the β-cell phenotype (Fig. 7.2). The endocrine reprogramming is controlled by the NOTCH1–HES1 gatekeeper system.
7.5 Phenotype and Function of Beta Cells Generated from Acinar Cells
In-vitro analyses revealed that the β-like cells obtained from transdifferentiated acinar cells expressed insulin, C-peptide, and PDX1 proteins at levels similar to those observed in normal islet β cells (Baeyens L et al., unpublished results). The insulin content of the β-like cells was similar to that of islets, but glucose-induced insulin secretion was 50% lower. When examined by immunocytochemistry for other β-cell markers, such as the glucose transporter GLUT2, the secretory granule components chromogranin-A and islet amyloid polypeptide, and the transcription factor MAFA, the newly-generated β cells appeared immature compared with islet β cells (Baeyens L et al., unpublished results). Engraftment of these cells under the kidney capsule of diabetic animals led to maturation of the grafted insulin-positive cells within 1 week. They acquired normal levels of the β-cell markers noted above, suggesting further maturation in vivo. Transplantation of only 105 insulin-positive cells obtained from acinar cell cultures was sufficient to normalize glycemia in severely hyperglycemic recipients (Baeyens L et al., unpublished results). This number of cells is approximately the same as the number required for normalizing glycemia with islet β cells. This finding indicates that as yet unidentified factors in vivo induce further maturation of acinar-derived β cells to the normal functional state and raises hopes for using these cells in cell replacement therapy for diabetes.
7.6 Translation to the Clinic
Treatment of type 1 diabetes by islet transplantation is hampered by, among other problems, the shortage of organ donors. This scarcity is aggravated by the fact that more than one organ is needed to obtain sufficient islets for one recipient. The failure of many grafts to maintain normoglycemia for more than a year post-transplantation may be explained by a suboptimal number of transplanted cells. Therefore, developing alternative sources of transplantable β-like cells is important for efficient application of this therapy. At present acinar cells, the most abundant cell type in the pancreas, are discarded during human islet isolation, but they represent such a source.
Thus far, transdifferentiation of acinar cells to β cells has been studied only in rats and mice. Preliminary attempts to reproduce our work on rodent cells in human acinar cells have not yet been successful (unpublished observations). It is possible, but unlikely, that human acinar cells do not possess a transdifferentiation capacity. More likely, however, is the possibility that species-related differences exist at the level of signaling factors or pathways involved in the regulation of transdifferentiation/reprogramming events. Species-related differences are known to exist even between evolutionarily close species, such as mouse and rat. Indeed, the protocol that we employed successfully with rat acinar cells does not seem to work with mouse acinar cells. The protocol that was reported by Minami and colleagues (Minami et al., 2005) to work with mouse cells does not work with rat cells (unpublished observations). Therefore, before this approach can be developed into a cell replacement therapy, protocols and agents that can effect human acinar cell reprogramming must be identified. Previous studies have already shown that human acinar cells possess some differentiation plasticity, as judged by their ability to undergo a phenotypical switch to duct-like cells (Hall and Lemoine, 1992). However, others have reported that in cultures of human exocrine pancreas tissue acinar cells selectively die by apoptosis, leaving cultures enriched in duct cells (Street et al., 2004; Klein et al., 2008). Thus, finding appropriate culture conditions in which viable human acinar cells can be maintained is a first step in the translation of this research.
7.7 Future Work
The abundance of acinar tissue that can be obtained from donor pancreata makes efforts toward neogenesis of β cells from acinar cells highly worthwhile. In-vitro studies with rat acinar cells unequivocally demonstrated the occurrence of acinar- to-β-cell transdifferentiation. It is possible that such transdifferentiation occurs in other experimental models but, if so, its contribution has been overlooked. Many reports, including by our own group, have described β-cell neogenesis from duct cells. In these studies the conclusion regarding the ductal origin of β cells was based primarily on the appearance of insulin-positive cells within the ductal epithelium surrounding a lumen. A commonly used marker for duct cells is a specific pattern of cytokeratin expression, but this pattern is also acquired by dedifferentiated acinar cells (Rooman et al., 2000). A similar problem faced investigators attempting to identify the cell type from which pancreatic adenocarcinoma originates. Recent genetic lineage-tracing studies have proposed that the cancer cells may originate from acinar cells that undergo dedifferentiation following a chronic injury, rather than from duct cells (Guerra et al., 2007; De La O et al., 2008; Habbe et al., 2008). Along the same line of thought, the role of centroacinar cells in different models of β-cell neogenesis has to be addressed. It is thought that these cells are not labeled in lineage-tracing studies employing promoters of genes encoding acinar enzymes, but this possibility has not been rigorously evaluated. In a recent study by Guerra et al. (Guerra et al., 2007), Cre recombinase driven by the elastase-promoter labeled centroacinar cells. Thus, an adequate lineage-tracing method for labeling both duct cells and centroacinar cells has to be developed and tested side by side with the acinar cell lineage tracing in the available experimental models of β-cell neogenesis. Such studies may provide further support for a physiological role of acinar-to-β-cell differentiation.
We still lack understanding of the molecular mechanisms that underlie the dedifferentiation phase of the transdifferentiation process. Dedifferentiation is assumed to be a prerequisite for the reprogramming of acinar cells into β cells. In our in-vitro rat acinar cell transdifferentiation model, induction of dedifferentiation upon cell isolation and culture is obvious, as judged by downregulation of several acinar cell markers. This aspect has not been addressed in detail in the in-vivo mouse acinar cell transdifferentiation model using viral transduction of transcription factors (Zhou et al., 2008). The molecular signals and mechanisms involved in dedifferentiation should therefore be further studied. Elucidating them may allow manipulation of this process in vivo without gene transfer. Dedifferentiation can also predispose the cells to neoplasia, thus providing further impetus for fully understanding its role and mechanisms.
- Bonal C, Thorel F, Ait-Lounis A, et al. (2008) Pancreatic inactivation of c-Myc decreases acinar mass and transdifferentiates acinar cells into adipocytes in mice. Gastroenterology [Epub ahead of print].Google Scholar
- De La OJP, Emerson LL, Goodman JL, et al. (2008) Notch and Kras reprogram pancreatic acinar cells to ductal intraepithelial neoplasia. Proc Natl Acad Sci USA [Epub ahead of print].Google Scholar
- Habbe N, Shi G, Meguid RA, et al. (2008) Spontaneous induction of murine pancreatic intraepithelial neoplasia (mPanIN) by acinar cell targeting of oncogenic Kras in adult mice. Proc Natl Acad Sci USA [Epub ahead of print].Google Scholar
- Klein T, Heremans Y, Heimberg H, et al. (2008) Investigation and characterization of the duct cell-enriching process during serum-free suspension and monolayer culture using the human exocrine pancreas fraction. Pancreas [Epub ahead of print].Google Scholar
- Laguesse E. (1896) Recherches sur l’histogenie du pancreas chez le mouton. J Anat Physiol. 32:209–255.Google Scholar