Centroacinar cells: at the center of pancreas regeneration

Abstract

The process of regeneration serves to heal injury by replacing missing cells. Understanding regeneration can help us replace cell populations lost during disease, such as the insulin-producing β cells lost in diabetic patients. Centroacinar cells (CACs) are a specialized ductal pancreatic cell type that act as progenitors to replace β cells in the zebrafish. However, whether CACs contribute to β-cell regeneration in adult mammals remains controversial. Here we review the current understanding of the role of CACs as endocrine progenitors during regeneration in zebrafish and mammals.

Introduction

Regeneration is the process of replacing missing cells to regain homeostasis. Three major mechanisms can contribute to this process. In cases where an extreme injury causes a drastic loss of tissue, such as the resection of a limb, cells nearby the damage dedifferentiate and proliferate to contribute to new tissue growth (Tornini and Poss, 2014). Alternatively, some cell types are known to transdifferentiate in response to acute injury by changing fate without an intermediate step (Bermingham-McDonogh and Reh, 2011). Resident progenitor/stem cell populations can also be called out of reserve to respond to a range of drastic and acute injuries (Tanaka and Reddien, 2011). This process has been well defined in planaria, which are able to regenerate any portion of the body because of a large population of multipotent adult stem cells called neoblasts (Reddien and Sanchez Alvarado, 2004). In other cases, adult progenitor/stem cell populations in tissues such as the retina and kidney (Gemberling et al., 2013; Sander and Davidson, 2014) respond to injury signals by self-renewing and differentiating to replace missing cells. Excitingly, research into each of these mechanisms can uncover new ways to treat diseases where entire tissues or specific cells need to be replaced.

Type 1 diabetes is caused by an autoimmune destruction of insulin-producing β cells in the pancreas. Thus, an effective cure for type 1 diabetes will address eliminating autoimmunity and replacing lost β cells. Attempts to replace β cells by transplanting donor cells or human stem cells differentiated in vitro have already found some success, but they are limited by the small number of functional β cells that can be made available for transplant (Pagliuca et al., 2014; Rezania et al., 2014; Russ et al., 2015; Shapiro et al., 2006; Vardanyan et al., 2010). Alternatively, an unaffected endocrine progenitor pool could be induced to differentiate and replace β cells endogenously. This strategy is non-invasive and could be achieved using drug treatment, making it an accessible and attractive solution. However, the regenerative response of the mammalian pancreas can vary depending on the injury model (Criscimanna et al., 2011; Kopp et al., 2011; Lee et al., 2006; Li et al., 2010; Nagasao et al., 2005; Xu et al., 2008), making uncovering mechanisms of regeneration in those animals difficult. On the other hand, as we will discuss here, the zebrafish pancreas displays a remarkable capacity for regeneration (Delaspre et al., 2015; Ghaye et al., 2015; Moss et al., 2009). A detailed understanding of how the fish pancreas regenerates is thus the critical first step towards achieving endogenous β-cell replacement therapy in humans.

Functionally, the pancreas can be separated into two major compartments. The exocrine pancreas includes digestive enzyme-secreting acinar cells arranged in functional units called acini, and ductal cells that direct the passage of these enzymes into the gut. The endocrine pancreas secretes hormones into the circulation and plays a major role in regulating glucose metabolism. The cells of this compartment are arranged in islets containing mostly insulin-producing β cells, glucagon-producing α cells, and δ cells making somatostatin.

In the zebrafish, a population of pancreatic Notch-responsive cells have been well recognized for their progenitor capacity during development (Dalgin and Prince, 2015; Delaspre et al., 2015; Delous et al., 2012; Huang et al., 2014; Manfroid et al., 2012; Ninov et al., 2012; Parsons et al., 2009; Rovira et al., 2011; Wang et al., 2015; Wang et al., 2011). Recently, these cells have been revealed to be specialized ductal cells called centroacinar cells (CACs) (Delaspre et al., 2015), which also serve as progenitors to regenerate β cells in the adult zebrafish (Delaspre et al., 2015; Ghaye et al., 2015). CACs are a cell type common to all vertebrates and have several defining characteristics: they are a ductal cell type positioned at the center of acini, with unique cell morphology, active Notch signaling, and expression of the endocrine differentiation regulator Sox9 (Kopp et al., 2011; Manfroid et al., 2012; Seymour, 2014; Seymour et al., 2008; Shih et al., 2012). Here, we review what is currently known about this intriguing cell type, highlighting the commonalities and differences between species and the implication for potential regeneration in humans.

Centroacinar cells are specialized ductal cells

Of all the cells of the exocrine pancreas, CACs are perhaps the most enigmatic. As their name suggests, CACs are positioned within the center of acini at the duct terminus (Figure 1A and 2A–C). In mammals, CACs are found at the proximal tips of the pancreatic ductal tree, which is composed largely of cuboidal and columnar epithelium. On the other hand, in the adult zebrafish, CACs appear to form the majority of the intrapancreatic ductal network (Figure 1D). Large ducts composed of distinct cuboidal epithelium are relatively rare and are found primarily near the head of the pancreas, close to the hepatopancreatic duct (Chen et al., 2007). The precise juxtaposition of CACs and the manner in which they form the ductal lumen are still unclear. In both zebrafish and mammals, transmission electron microscopy studies of CAC ultrastructure have detailed a unique ruffled nuclear morphology and the close association of CACs with both acinar and other ductal cells via tight junctions (Parsons et al., 2009; Pour, 1994) (Figure 3E). Interestingly, contrasting other duct cells that contribute to cuboidal and columnar epithelium, in all vertebrates examined CACs have long cytoplasmic extensions (Figure 1C and 2D) that extend along the duct to contact neighboring CACs and into the parenchyma to touch other neighboring cells, including islets (Delaspre et al., 2015; Leeson and Leeson, 1986; Pour, 1994). Notably, zebrafish CACs seem to have much longer extensions than mammalian CACs. What specific role these extensions play in CAC biology remains to be determined. However, since some extensions contact the endocrine compartment, it is intriguing to postulate that they play a role in the CAC response to endocrine regeneration.

Figure 1.

Centroacinar cells in the adult zebrafish. A) In adult fish, CACs (purple) posses extensions that extend throughout the parenchyma to contact other CACs, islets, and acinar cells (brown). Illustration depicts a small portion of the larger, multi-lobed organ. B) Adult CACs (arrows) express the Notch-responsive Tg(tp1:nuclear-mCherry) transgene (red) and retain the expression of the transcription factor Nkx6.1 (green) B’ and B” are magnifications of depicted area in B. C) CACs (Nkx6.1, red, arrows) also express Nkx2.2a, as indicated by the TgBAC(nkx2.2a:membrane-GFP) transgene (green) and Insulin (white). CAC extensions are indicated with arrowheads. D) 2F11 (green) marks all ductal epithelium, including CACs and other ductal cells, in adult zebrafish pancreas. Dashed lines show a 2F11-positive duct and arrows indicate 2F11-positive CACs also marked by Tg(tp1:nuclear-mCherry) (red) transgenic line. Scale bars = 20 μm. Images in B and C have been reproduced from (Delaspre et al., 2015) with permission of the American Diabetes Association.

Figure 2.

Centroacinar cells in the adult mammalian pancreas. A) Illustration of a portion of adult mouse pancreas containing acinar cells (brown), centroacinar cells (dark purple), terminal ductal cells (light purple), ductal cells (blue) and islets. B–C) Hematoxylin/Eosin staining of human pancreas (B) and mouse pancreas (C), dashed lines demarcate acinar units with one or two centroacinar cells (arrows). D) Immunofluorescence images of mouse adult pancreas showing Hnf1β-positive (red) and DBA-positive (green) centroacinar cells (arrows) with extensions (arrowheads in D′), DBA and Hnf1β mark all ductal cell types. D′ is a magnification of depicted area in D. Nuclear staining (DAPI, gray). Scale bar = 40 μm (B and C). Scale bar =100 μm (D). DBA = Dolichos Biflorus Agglutinin

Figure 3.

Centroacinar cells in the larval zebrafish. A) At 5 dpf, the zebrafish pancreas contains one large principal islet. At this time, secondary islets arise from CAC progenitor cells (purple), which have extensions that contact both acinar cells (brown) and islets. B) Live image of CACs (green) visualized using the Notch-responsive Tg(tp1:membrane-GFP) line, which make up the intrapancreatic duct at 5 dpf. CAC extensions are indicated with arrowheads. sb = swim bladder, m = muscle. C) 5 dpf CACs express the transcription factor Nkx6.1 (green) and the Notch-responsive Tg(tp1:nuclear-mCherry) transgene (red). D) The TgBAC(nkx2.2a:membrane-GFP) transgene (green) indicates that CACs (Nkx6.1⁺, red, arrows) also express Nkx2.2a at 5 dpf. CAC extensions are indicated with arrowheads. Dashed lines in B–D delineate the pancreas. Scale bars B–D = 50 μm. E) Transmission electron micrograph of a larval 10 dpf pancreas cross section. CACs (purple) line the intrapancreatic ductal lumen (asterisks) and closely associate with other CACs and acinar cells (brown) via tight junctions (arrows). Scale bars E = 10 μm, E′ = 1 μm, E″ = 500 nm. Images in C and D have been reproduced from (Delaspre et al., 2015) with permission of the American Diabetes Association. Images in E have been reproduced from (Parsons et al., 2009).

Physiologically, CACs play an important role in exocrine function by regulating the contents of the distal ductal lumen. CACs in mammals express the bicarbonate/proton pump carbonic anhydrase II (CAII) (Inada et al., 2008). Bicarbonate secretion serves to neutralize acid entering the duodenum from the stomach and avoid the aggregation of digestive enzymes in the pancreatic lumen. CACs in fish and mammals also express several aquaporins and calcium ion channels such as Cftr and Clcn1 (Burghardt et al., 2003; Delaspre et al., 2015; Marino et al., 1991; Trezise and Buchwald, 1991). Thus, CACs play an important role in maintaining the ionic content of the ductal lumen and keep it open for fluid flow. Interestingly, CACs and terminal ductal cells (TDCs) (Figure 2A) have long been known to express Cftr (Marino et al., 1991; Trezise and Buchwald, 1991). CFTR mutations cause the common recessive disorder cystic fibrosis. 85% of cystic fibrosis patients develop diabetes, possibly arising from blockage of the pancreatic duct, an accumulation of digestive enzymes and generalized destruction of pancreatic tissues (O’Shea and O’Connell, 2014).

Adult CACs also maintain the expression of genes and signaling pathways important during development, as we will see below. Thus, they have been included among cell types proposed as candidate adult endocrine progenitors in mammals and zebrafish (Parsons et al., 2009; Rovira et al., 2010; Stanger et al., 2005; Wang et al., 2011). During pancreas development in fish and mammals, Notch-responsive progenitors give rise to CACs, ductal, and endocrine cells. Aside from some endothelial cells, CACs are the only Notch-responsive pancreatic cell type in adult zebrafish (Figure 1B) (Delaspre et al., 2015; Parsons et al., 2009). In adult mammals, both CACs and TDCs are Notch-responsive, suggesting that CACs/TDCs are the mammalian equivalent to zebrafish CACs (Kopinke et al., 2011; Stanger et al., 2005). Notch signaling is crucial to maintain CAC plasticity, progenitor status, and stabilization of cell fate in both mammals and fish (Kopinke et al., 2012; Nakano et al., 2015; Parsons et al., 2009). Additionally, in mammals and fish CACs express Sox9, a well-known regulator of stem and progenitor cells (Delous et al., 2012; Kopp et al., 2011; Lioubinski et al., 2003; Manfroid et al., 2012; Seymour, 2014). Whether these characteristics confer a progenitor capacity to CACs that could be activated under certain injury conditions in mammals remains unclear.

Centroacinar cells during pancreas development

In the larval zebrafish at 5 days post fertilization (dpf) the pancreas is a spoon-shaped organ. A large rudimentary islet called the principal islet is found in the head of the pancreas, and the intrapancreatic duct composed entirely of CACs extends into the tail (Figure 3A–B). The same arrangement of CACs in the adult pancreas likely also serves to form the ductal lumen in larvae, however this remains to be determined. Nonetheless, by this stage CACs already display their distinct cellular extensions (Figure 3B and D) (Delaspre et al., 2015; Leeson and Leeson, 1986; Parsons et al., 2009; Pour, 1994). The majority of the endocrine lineage arises from these ductal CACs during an event called the secondary transition. This wave of differentiation becomes most easily observable at 5 dpf, as secondary islets appear along the duct in the tail of the pancreas (Parsons et al., 2009). During the secondary transition, Notch signaling maintains progenitor status in CACs. Blocking Notch activity leads to precocious differentiation of CACs into NeuroD-expressing endocrine precursors (Dalgin and Prince, 2015; Parsons et al., 2009). Importantly, larval zebrafish CACs express several major transcriptional regulators of endocrine differentiation such as Sox9b, Nkx6.1, and Nkx2.2a (Delaspre et al., 2015; Ghaye et al., 2015; Manfroid et al., 2012; Pauls et al., 2007) (Figure 3C–D). Inducible genetic lineage tracing of CACs at 5 dpf using the Notch-responsive tp1 promoter has demonstrated that larval CACs give rise to ductal and endocrine cells in the larval and adult pancreas (Wang et al., 2011). Thus, larval CACs are a source of β-cell neogenesis in larvae and give rise to adult endocrine tissue.

At first glance, early pancreas morphogenesis in the mouse looks very different when compared to the zebrafish (Shih et al., 2013). Difficulty matching equivalent stages due to the astonishing speed of early pancreatogenesis in the zebrafish may account for some disparity. However, the cellular and molecular mechanisms regulating endocrine differentiation are well conserved; a population of ductal, Notch-responsive, Nkx6.1-expressing progenitors also acts as the embryonic source of the majority of endocrine cells in the mammalian pancreas (Kopinke et al., 2011; Kopinke and Murtaugh, 2010; Sander et al., 2000; Schaffer et al., 2010). In the mouse, the secondary transition begins around E12.5. At this stage, the pancreas contains a branched ductal epithelium that is compartmentalized into tip and trunk domains (Figure 4A) (Afelik et al., 2012; Villasenor et al., 2010; Zhou et al., 2007). The Ptf1a-expressing tip domain will give rise to acinar cells, while trunk cells retain both duct and endocrine differentiation potential (Hald et al., 2008; Schaffer et al., 2013; Taylor et al., 2013). As in the zebrafish, Notch signaling retains progenitors by restricting endocrine differentiation in the trunk domain. High levels of Notch induce expression of the transcription factor Hes1, which in turn promotes expression of Nkx6.1 and Sox9 while inhibiting expression of the endocrine determinant Ngn3 (Figure 4B). (Afelik et al., 2012; Apelqvist et al., 1999; Hald et al., 2003; Kopinke et al., 2012; Shih et al., 2012). The expression of these transcription factors – Hes1, Sox9, and Nkx6.1 – eventually becomes restricted to cells immediately adjacent to the tip domain. Although the developmental origin of mammalian CACs is unknown, on account of their tip-trunk junction position and their unique transcriptional signature, some have hypothesized that these cells eventually give rise to adult CACs/TDCs (Cleveland et al., 2012; Seymour, 2014). Nonetheless, genetic programs active in these progenitors suggest that they are the embryonic mammalian equivalent to zebrafish CACs. However, the lack of unique centroacinar specific markers in mammals has hindered a direct test of this hypothesis by lineage tracing.

Figure 4.

Tip/trunk model of mouse pancreas development. A) During the secondary transition in mouse beginning at ~E12.5, the branching pancreatic epithelium contains acinar progenitors in the tip domain (brown) and bipotent endocrine and ductal progenitors in the trunk domain (purple). B) Immunofluorescence images of mouse E13.5 embryonic pancreas showing tip-trunk regionalization delineated by trunk markers Nkx6.1 (green) and Sox9 (blue), and tip marker Carboxypeptidase A1 (CPA1, red). High levels of Nkx6.1 and Sox9 are detected in the trunk region (demarcated by dashed lines) while the tip region expresses CPA1 and lower levels of Nkx6.1 and Sox9. Nuclear staining (DAPI, gray). Scale bar =100 μm.

During the secondary transition, bipotent progenitors found in the duct of both the fish and mouse pancreas run a highly conserved genetic program. Therefore, further dissecting the mechanisms regulating the larval zebrafish progenitor population during the secondary transition can also help us understand β-cell neogenesis in mammals. These efforts can inform methods to differentiate β cells in vitro for transplantation into type 1 diabetes patients. Alternatively, understanding the mechanisms regulating adult zebrafish CACs may be useful in developing ways to induce endogenous β-cell regeneration in humans.

Centroacinar cells as adult progenitors

CACs have recently been shown to be a source of β-cell neogenesis in the adult zebrafish pancreas, mirroring their progenitor function during development. Two independent groups utilized the NTR/MTZ system to specifically ablate >99% of β cells in the adult (Delaspre et al., 2015; Ghaye et al., 2015). Following ablation, single β cells and β-cell clusters not associated with islets appear along ducts and near CACs, suggesting that β-cell neogenesis is one mechanism by which the pancreas regenerates (Delaspre et al., 2015). CACs proliferate (Delaspre et al., 2015; Ghaye et al., 2015) and respond dynamically during regeneration by moving closer and even into recovering islets (Delaspre et al., 2015). Using either the Notch-responsive tp1 or the nkx6.1 promoter driving GFP expression, each group observed GFP perdurance in cells beginning to express Insulin, suggesting that CACs themselves contribute to β-cell regeneration (Delaspre et al., 2015; Ghaye et al., 2015). To determine definitively whether CACs were acting as progenitors, Delaspre, et al. used inducible genetic lineage tracing via the Notch-responsive tp1 promoter to label CACs before β-cell ablation. They reported that adult CAC progeny contributed to β-cell regeneration, with 43% of regenerated β cells labeled. In the same study, RNAseq of adult zebrafish CAC markers revealed that larval and adult CACs are indistinguishable and that CACs retain expression of developmental endocrine program genes Nkx6.1 and Nkx2.2a (Figure 1B–C) (Delaspre et al., 2015). Thus, advantages of the larval zebrafish model can be used in conjunction with adult studies to further explore the regulation of CACs and β-cell neogenesis.

One important note from these studies is that CACs are likely not the only cellular source of β-cell regeneration in the zebrafish. β cells proliferate as part of the regenerative response (Delaspre et al., 2015; Ghaye et al., 2015; Moss et al., 2009). Moreover, other sources of β-cell regeneration in the fish are known; following ablation in larva, new β cells in the principal islet arise via transdifferentiation from α cells (Ye et al., 2015). The possibility remains that other pancreatic cell types may also contribute to β-cell regeneration in the adult.

In the mammalian pancreas, β-cell neogenesis contributes to regeneration by varying degrees. Continued apoptosis of β cells in older type 1 diabetes patients suggests that some new β cells continually arise during the course of the disease (Meier et al., 2005). In the mouse, the extent of pancreas regeneration seems to be highly dependent upon the extent and type of injury. Following β-cell-specific ablation, new β cells seem to arise mostly from β-cell proliferation and/or transdifferentiation from α cells, δ cells, and, to a lesser extent, ε cells (Arnes et al., 2012; Chera et al., 2014; Courtney et al., 2013; Nir et al., 2007; Thorel et al., 2010). However, partial ductal ligation leads to the appearance of Ngn3-expressing endocrine precursors in regenerating adult mouse pancreata (Xu et al., 2008). Interestingly, using genetic lineage tracing with several ductal markers, these Ngn3-expressing cells have in some studies been observed to give rise to new β cells but do not do so in others (Al-Hasani et al., 2013; Kopp et al., 2011; Van de Casteele et al., 2014; Xiao et al., 2013a; Xiao et al., 2013b). Subtotal pancreatectomy, removal of 90% of the pancreas, leads in rats to extensive regeneration of both exocrine and endocrine compartments, suggesting β-cell neogenesis is a part of the regenerative process (Bonner-Weir et al., 1993). Adult rat pancreatic duct cells recapitulate aspects of embryonic pancreas differentiation in response to this type of injury and thus may contribute to regeneration of the pancreas (Bonner-Weir et al., 1993; Li et al., 2010).

Whether CACs or other ductal progenitors play a role in mammalian pancreas regeneration is currently under contentious debate. Much evidence exists to support the role of CACs or other ductal cells in β-cell neogenesis in adult mammals. As in the fish, rodent CACs proliferate in response to pancreas injury (Gasslander et al., 1992; Hayashi et al., 1999; Hayashi et al., 2003). Several other studies have observed endocrine cells near to or embedded in the ductal tree during normal growth or regeneration (Bonner-Weir et al., 1993; El-Gohary et al., 2012; Gu and Sarvetnick, 1993; Li et al., 2010; Phillips et al., 2007; Rooman and Bouwens, 2004; Rosenberg, 1995; Weaver et al., 1985; Xu et al., 1999; Xu et al., 2008; Yoneda et al., 2013). Indeed, Ngn3-positive cells lining the ducts contribute to β cell expansion induced by overexpressing Pax4 in α cells (Al-Hasani et al., 2013). Moreover, ductal cells can be converted into α cells, δ cells and β cells following the inactivation of SCF-type E3 ubiquitin ligase substrate component Fbw7. Deletion of Fbw7 stabilizes Ngn3 and seems to reawaken an endocrine developmental differentiation program in adult pancreatic ductal cells (Sancho et al., 2014). El-Gohary et al., have also observed that β cells can arise from specialized ductal structures called inter-islet ducts following murine partial pancreatectomy (El-Gohary et al., 2016). Intriguingly, these ductal cells appear to have thin cellular extensions reminiscent of CAC extensions. Recently, lineage-tracing studies using Sox9-CreER transgenic mice have shown that adult pancreatic ductal cells can differentiate into β cells under medium hyperglycemic conditions and long-term administration of low doses of gastrin and epidermal growth factor (Zhang et al., 2016).

On the other hand, genetic lineage tracing experiments using fluorescent “timers” (Miyatsuka et al., 2014) or driven by a plethora of ductal and CAC promoters – including Hnf1β (Solar et al., 2009), Hes1 (Kopinke et al., 2011), Sox9 (Kopp et al., 2011), Mucin1 (Kopinke and Murtaugh, 2010), and CAII (Inada et al., 2008) – have also examined whether adult ductal cells or CACs act as endocrine progenitors. With the exception of Inada, et al., most of these studies have concluded that endocrine and exocrine cells arise from ductal cells throughout development but not in the adult pancreas. However, several caveats accompany this conclusion. Many of the genetic lineage tracing models labeled only a small proportion of cells (as low as 20% of the target population) (Kushner et al., 2010). Additionally, to label potential progenitors some of these studies used regulatory sequences controlling expression of factors important for the maintenance of an undifferentiated state (i.e. Sox9 or Hes1) (Kopinke et al., 2012; Kopinke et al., 2011; Kopp et al., 2011). These factors would be expressed at their highest levels – and their regulatory elements be more likely to induce labeling – in those cells least likely to differentiate, possibly leading to false negative results. Furthermore, the potential to differentiate into β cells could be limited to only a subset of ductal cells not uniformly labeled by known ductal markers. Indeed some studies have suggested that CACs are a heterogeneous population (Matsuda et al., 2013; Pour, 1994; Rovira et al., 2010). Lastly, the regenerative mechanisms utilized in all mammalian species may not be the same, and CACs or other ductal progenitors may contribute to β-cell neogenesis more readily in some than in others.

Setting aside the controversy, if no neogenesis occurs from ductal cells in mice, what are the mechanisms of mammalian β-cell regeneration? In mice, as in zebrafish, β cells proliferate in response to injury (Dor et al., 2004; Gasslander et al., 1992; Hayashi et al., 1999). A variety of other cell types can also act to replenish the β-cell pool by transdifferentiation such as acinar cells, α cells, and δ cells; these studies have been well reviewed elsewhere (Puri et al., 2015; Ziv et al., 2013).

Remaining questions

The field of pancreas regeneration is relatively unexplored, and several important questions remain. The signaling molecules regulating the CAC response to injury are unknown, and analysis of CACs during both regeneration and homeostasis can begin to address this issue. Furthermore, several groups have hypothesized the existence of a pancreas stem cell population (Belo et al., 2013; Furuyama et al., 2011; Parsons et al., 2009; Stanger et al., 2005; Wang et al., 2011). CACs do express Sox9, a known regulator of hair follicle and intestinal stem cells (Bastide et al., 2007; Blache et al., 2004; Kadaja et al., 2014; Nowak et al., 2008). Additionally, zebrafish CACs display behavior characteristic of an adult stem cell population, both proliferating to self-renew and giving rise to differentiated progeny. However, the question remains whether they are a long-term self-renewing stem cell population. This can be experimentally addressed by transplanting isolated CACs, as has been used to demonstrate the long-term self-renewing properties of mammalian hematopoietic stem cells (Keller and Snodgrass, 1990) and adult zebrafish nephron stem cells (Diep et al., 2011). Alternatively, several serial rounds of β-cell ablation and regeneration, similar to an approach used by Al-Hasani, et al., could also potentially demonstrate the self-renewal properties of CACs in the zebrafish (Al-Hasani et al., 2013).

The more limited regenerative capacity of the mammalian pancreas is in striking contrast to the zebrafish, which can recover from near total β-cell ablation in a matter of days. Is there some facet of zebrafish biology that underlies such ability? Adult zebrafish CACs retain expression of Nkx6.1 and Nkx2.2a, which are not expressed in mouse CACs (Delaspre et al., 2015; Ghaye et al., 2015). Considering Nkx6.1 and Nkx2.2 are expressed in mammalian islet cells and are master regulators of the endocrine program (Binot et al., 2010; Papizan et al., 2011; Sander et al., 2000; Schaffer et al., 2013; Sussel et al., 1998), the idea that their retained expression in adult zebrafish CACs may drive enhanced regenerative capacity is compelling. Conditional zebrafish knockouts of nkx6.1 or nkx2.2a may address this question. Additionally, a comprehensive examination of transcriptional and epigenetic changes that occur in the mammalian pancreatic ductal population between embryogenesis and adulthood, and in comparison to zebrafish CACs, might reveal the mechanisms that restrict CAC/TDC differentiation potential in the adult mammal.

The conserved differentiation programs and enhanced regenerative capacity of the fish makes it an invaluable model for understanding CACs and β-cell neogenesis. This understanding could one day inform therapeutic strategies to replace β cells in diabetic patients by promoting differentiation of an unaffected ductal progenitor population. As larval and adult zebrafish CACs are so far indistinguishable, recent advances in high-throughput drug screening in whole zebrafish larvae will be useful in identifying drugs that can be used to specifically induce a regenerative response (Wang et al., 2015). A diabetes cure that could be achieved using an amenable and cost-effective solution like drug treatment is essential for a global solution to diabetes. Therefore, future investigations addressing CACs and pancreas regeneration will be greatly impactful.

• Centroacinar cells (CACs), specialized pancreatic ductal cells, are defined.

• CACs are endocrine progenitors during zebrafish development and regeneration.

• Whether CACs or other ductal cells are progenitors in adult mammals is discussed.

• Understanding CACs during pancreas regeneration can inform methods to treat disease.