Endocrine Pancreas Development and Regeneration: Noncanonical Ideas From Neural Stem Cell Biology
Loss of insulin-producing pancreatic islet β-cells is a hallmark of type 1 diabetes. Several experimental paradigms demonstrate that these cells can, in principle, be regenerated from multiple endogenous sources using signaling pathways that are also used during pancreas development. A thorough understanding of these pathways will provide improved opportunities for therapeutic intervention. It is now appreciated that signaling pathways should not be seen as “on” or “off” but that the degree of activity may result in wildly different cellular outcomes. In addition to the degree of operation of a signaling pathway, noncanonical branches also play important roles. Thus, a pathway, once considered as “off” or “low” may actually be highly operational but may be using noncanonical branches. Such branches are only now revealing themselves as new tools to assay them are being generated. A formidable source of noncanonical signal transduction concepts is neural stem cells because these cells appear to have acquired unusual signaling interpretations to allow them to maintain their unique dual properties (self-renewal and multipotency). We discuss how such findings from the neural field can provide a blueprint for the identification of new molecular mechanisms regulating pancreatic biology, with a focus on Notch, Hes/Hey, and hedgehog pathways.
To the neuroscientist, the pancreas can seem like a highly plastic organ whose cells are able to undergo vast changes in the context of homeostatic control and regeneration (1–4). Such plasticity in the adult brain is exhibited only by endogenous neural stem cell (NSC) populations. How NSCs manage to maintain this plasticity is an intensely studied question that has led to the identification of unusual (noncanonical) signal transduction pathways that help them manage the cellular properties that define them as stem cells, namely the ability to self-renew and the potential to differentiate into mature cell types. Part of this noncanonical machinery is a novel branch of the Notch signaling pathway that has pronounced consequences in vitro and in vivo (5). It is logical to wonder whether other tissues also use these signaling pathways.
Plasticity in the pancreas is demonstrated by multiple examples of trans- and dedifferentiation among various cell types reported; these have been experimentally induced, whether by pharmacological or genetic manipulation or by the implementation of damage and regeneration models (1,6–12). An impressive amount of work has been done identifying the molecular mechanisms that control this plasticity. The patterns that emerge indicate that a variety of signaling molecules and transcription factors act during development to both suppress and induce fates as well as to maintain proper function of the adult cell types, including the Notch and hedgehog pathways and a number of transcription factors, including pancreatic and duodenal homeobox 1 (Pdx1), neurogenin 3 (Ngn3), and the hairy and enhancer of split-1 (Hes/Hey) family (1,13–42).
A binary view of signal transduction (where a particular pathway is either on or off) seems to be incomplete as emerging evidence points toward several examples where low activity of a particular pathway (e.g., Notch and hedgehog) leads to vastly different developmental and cellular outcomes. But a simply quantitative distinction between low and high activity of a signaling pathway may also not be sufficient. A cell that is exhibiting low activity of a canonical pathway may actually also exhibit high activity in noncanonical pathways that are only now becoming revealed. For example, whereas canonical Notch (caNotch) signaling is often assessed by the amount of Hes1 expression (a caNotch pathway target gene), it is possible that the cell exhibits high expression of, for example, Hes3, an indirect, noncanonical target of Notch (noncaNotch) with distinct outcomes in NSCs (43). Low activity of a signaling pathway, as assessed by measurements of its canonical functions, does not preclude that other branches of the same pathway are not highly active. Noncanonical branches may be unidentified simply because we have not yet designed assays for them.
NSCs are a treasure trove of noncanonical functions, perhaps because they are required to fit multiple properties into their signal transduction machinery (e.g., their self-renewal and cell fate–decision abilities), forcing them to come up with unique molecular solutions. Some of these have already proven important in pancreatic function. We will focus on lessons learned about noncaNotch and hedgehog signaling in NSCs as a paradigm for their potential roles in β-cell development and function. First, we discuss specific aspects of pancreatic development on which light can be shed by our emerging understanding of noncanonical signaling in NSCs. We do not aim to provide a comprehensive review of the developmental processes of the pancreas because this has been meticulously done elsewhere (1–4).
Overview of Pancreas Development and Involvement of Key Pathways
In this section we present a brief overview of specific aspects in pancreas development (Fig. 1). We focus only on particular pathways that are of direct relevance to the concepts discussed here. At approximately mouse embryonic day 8.5 (E8.5), a prepancreatic region is specified in the gut endoderm (44). For pancreatic specification to occur and pancreatic markers to be induced, sonic hedgehog (Shh) expression must be suppressed from the presumptive pancreatic endoderm (31,33,34). Subsequently, during early pancreatic development (∼E9.5; primary transition phase), the pancreatic bud forms, which contains precursor cells (multipotent pancreatic precursors [MPCs]) that are able to generate differentiated cells from all three major lineages of the adult pancreas (endocrine, duct, and acinar). These pancreatic progenitors express Pdx1, Sox9, and Hes1, a direct transcriptional target of caNotch signaling with multiple developmental roles (25,29,45,46). Notch signaling is critical in the maintenance of pancreatic progenitors (24,35,47,48). Specifically, loss of function of the Notch ligand Delta-like 1 (Dll1) results in premature endocrine differentiation and depletion of the pancreatic progenitors (24,35).
Conversely, overexpression of the Notch intracellular domain prevents differentiation and traps cells in a progenitor state (48). This appears to be due to caNotch signaling because similar phenotypes involving pancreatic growth arrest are observed with genetic mouse models deficient in recombining binding protein suppressor of hairless (RBPJ-κ), a transcriptional mediator of caNotch signaling (14) or Hes1 (24,35,46,49). Apart from Hes1, Notch signaling also regulates Sox9, a transcription factor that, like Hes1, has multiple developmental roles in many tissues (2,46,50–54). The molecular mechanisms that control Hes1 are very important for transcriptional control and the responsiveness of pancreatic progenitors to mitogenic support from the surrounding mesenchyme. In one example, Hes1 opposes expression of p57, which normally induces cell cycle exit (55). Notch activity is necessary for the cells to respond to fibroblast growth factor (FGF) 10, a mitogen expressed by the mesenchyme at ∼E9.5–11.5 (56). In fact, inhibition of Notch cleavage by a γ-secretase inhibitor renders cells nonresponsive to the proliferative effects of FGF10 (57–59).
Later in development, these progenitors become more restricted in their differentiation potential and separate into two cell types: unipotent acinar progenitor “tip cells” and bipotent “trunk epithelial” cells that can generate endocrine and duct cells. High caNotch signaling, as assessed by the involvement of RBPJ-κ and Hes1, promotes the trunk fate (47,48,60–64). High caNotch signaling in trunk cells activates Hes1 and Sox9 (a repressor and an activator of Ngn3, respectively). On the one hand, Hes1 activity apparently dominates this fight, leading to low Ngn3 levels and, subsequently, specification to the ductal fate. Low caNotch activity, on the other hand, leads to induction of only Sox9, leading to high Ngn3 levels and the acquisition of the endocrine fate (3,65). The interpretation of high versus low Notch activity is an important and recurring theme throughout this perspective.
In this article, we discuss data acquired from different developmental stages of the endocrine pancreas and also from the exocrine pancreas; these observations point toward novel regulatory signaling networks that may be of significant interest to pancreas biology. However, additional future work will be necessary to fully elucidate at which developmental stages these networks may be operational and with what precise consequences.
NSCs Point Us Toward Noncanonical Signaling Pathways That Also Control Pancreatic Islet Cells
In this section, we present work from the NSC field demonstrating alternative interpretations of common signal transduction pathways, and in subsequent sections, we address and speculate how these pathways may contribute to the explanation of open questions in pancreatic biology.
Special Cellular Requirements Force Special Signal Transduction Interpretations
NSCs appear to have acquired an inventive mechanism to fit both their self-renewal and multipotency properties into the existing signal transduction machinery. For example, they have allocated some of these different properties to different phosphorylation sites of the same molecule, such as those on signal transducer and activator of transcription 3 (STAT3). STAT3 has two phosphorylation sites, one on the tyrosine residue (STAT3-Tyr) at amino acid position 705 (numbering is for the mouse protein) and one on the serine residue at position 727 (STAT3-Ser) (66). On the one hand, tyrosine phosphorylation leads to the induction of differentiation, ending their stem cell state (67–69). Serine phosphorylation, on the other hand, is important for their growth and, therefore, self-renewal (5). In stark contrast, many other cell types that are only capable of self-renewal and therefore do not have issues of such choices (e.g., astrocytes), use both phosphorylations for growth and survival (with, typically, the tyrosine site being much more important than the serine site). In NSCs, a noncanonical branch of the Notch signaling pathway is a major activator of STAT3-Ser phosphorylation (5).
Notch Signaling in Neural Development
Notch signaling is a major regulator in brain development and the regulation of NSCs. The Notch signaling pathway is involved in a variety of biological processes, including cell fate specification of many different tissues, stem cell and progenitor growth, self-renewal, and differentiation; these functions are of direct consequence to development, regeneration, and cancer (14,50,60). The Notch family of genes consists of plasma membrane–spanning receptors that, upon activation by ligands (also membrane spanning) from adjacent cells, undergo a series of proteolytic cleavages that release the intracellular domain of the receptor into the cytoplasm. This subsequently translocates to the nucleus where it interacts with other proteins such as the transcriptional regulator RBPJ-κ (14). A well-studied target gene in many tissues is the transcription factor Hes1, a member of the Hes/Hey gene family of basic helix-loop-helix (bHLH) transcriptional repressors (50). The involvement of RBPJ-κ and the induction of Hes1 are often used as indicators of Notch activation. RBPJ-κ and Hes1 involvement typically denote the operation of caNotch signaling (70–72) (Fig. 2), although RBPJ-κ–independent functions of Notch have also been reported (but also Notch-independent RBPJ-κ functions).
The noncaNotch signaling branch that leads to STAT3-Ser phosphorylation also involves a number of other intracellular signaling components (e.g., phosphoinositide 3-kinase [PI3K], Akt, mammalian target of rapamycin [mTOR], etc.) as well as the transcription factor Hes3, a member of the Hes/Hey gene family like Hes1 and Hes5 (5,54,73). Therefore, Hes3 is also downstream of Notch signaling but is not a direct transcriptional target like Hes1 and Hes5; instead, Notch receptor activation leads to Hes3 induction via Akt, mTOR, and STAT3-Ser phosphorylation. Notch ligands that lead to the induction of Hes3 expression greatly increase NSC numbers in vitro and in vivo (5,74).
These two Notch signaling branches (ca/Hes1 and nonca/Hes3) have opposing functions in NSCs. Hes1 induces differentiation to the glial fate by directly activating prodifferentiation genes and indirectly by promoting the interaction of Janus kinase (JAK) with STAT3, which further promotes gliogenic differentiation (66–68,75). This involves the phosphorylation of STAT3 by JAK on the tyrosine residue. Hes3, in contrast, is induced by several factors that lead to STAT3-Ser phosphorylation in the absence of STAT3-Tyr phosphorylation (5).
Therefore, the Hes/Hey family of transcription factors has multiple functions in the nucleus and cytoplasm, including transcriptional repression, passive repression (through protein-protein interactions), and the mediation of signal transduction. Elucidating these highly complex functions for all members of this gene family is likely to contribute to our detailed understanding of a number of processes where these genes are involved. Because Hes/Hey proteins interact with each other, typically leading to the repression of their transcriptional activity, studying the expression patterns, subcellular localization, and function of all family members together in different biological systems may be useful.
Hes3 itself is a functional mediator of the noncanonical action of Notch signaling because NSCs from adult Hes3-null mice fail to respond to Notch activation that normally promotes survival (76). They also fail to respond to insulin, another stimulator of Akt, mTOR, STAT3-Ser, and Hes3 (Fig. 3A). Hes3 also promotes Shh expression, which is a mitogen for NSCs (5). Therefore, Hes3 is an indirect target of Notch and represents the operation of a noncanonical branch of Notch that involves a number of intracellular signaling components, including STAT3 when it is phosphorylated on the serine residue. In other words, this noncanonical branch of Notch is an input to the STAT3-Ser/Hes3 signaling pathway.
A Generalizable, Noncanonical Molecular Pathway?
Deleting Hes3 in mice that already lack the Hes/Hey gene family members Hes1 and Hes5 leads to precocious differentiation of the NSC population during brain development (77). Aspects of this pathway are not limited to NSCs but can be found in a number of plastic cell types, including putative cancer stem cells from glioblastoma multiforme patients and chromaffin progenitor cells from the bovine adrenal medulla (78,79). STAT3-Ser has been shown to drive carcinogenesis in models of prostate cancer (80). Hes3 expression in vivo has been implicated as an indicator of the efficacy of a γ-secretase inhibitor (that blocks Notch signaling) in models of breast cancer (81). Recent studies have implicated Hes3 (still largely in a correlative manner) in different reprogramming paradigms (mouse embryonic fibroblasts to induced pluripotent cells and direct adult to induced NSCs) by demonstrating Hes3 expression regulation during reprogramming and by correlating Hes3 transduction of adult cells with successful reprogramming to induced NSCs (82,83).
These observations challenge the question of how widespread the operation of this pathway may be in different cell types in the body, including in those of the pancreas. After all, STAT3-Ser phosphorylation is downstream of several pathways known to promote β-cell proliferation in various developmental stages of the pancreas, including FGF signaling (84,85), which has been shown to intercept Notch signaling components and to regulate cell proliferation and self-renewal in both NSC and pancreatic systems (58,86), several cell surface receptors, Akt, and mTOR. It is possible that Hes3 may be a mediator of some of the functions of these molecules. This would also raise the possibility that, being a bHLH factor, Hes3 may interfere with the actions of other bHLH factors of importance to pancreatic development and function such as Ngn3 (5,73,87,88). In another parallel between brain and pancreas, Ngn3 has important functions in the development of the brain (including hippocampal and hypothalamic neurons and glial progenitors), islet cell specification, and β-cell regeneration (18,89–91).
Early data from mice lacking Hes3 corroborate some of these ideas because these mice exhibit phenotypes of direct relevance to pancreatic function; specifically, whereas otherwise apparently normal, these mice exhibit a more pronounced loss of pancreatic β-cells, develop diabetes faster (43), and show impaired regeneration (as assessed by β-cell marker expression) after streptozotocin (STZ)-induced pancreatic damage (Fig. 3B–D). Further, after damage they fail to induce Ngn3 expression in the exocrine pancreas compartment 5 months after partial STZ damage compared with wild-type controls (92). These data invite further studies on the possible developmental roles of Hes3 and its regulators and mediators.
Why Did We Miss It?
The lack of an obvious phenotype in the Hes3-null mice (88) and the absence of Hes3 probes in some DNA microarrays may have hindered research focus on this gene because these have often favored other Hes/Hey gene family members that are commonly used to assay caNotch signaling. An additional difficulty in studying the function of this gene arises from its regulation at the molecular level. Hes3 exists in two isoforms generated by distinct promoters (87). The coding region in the Hes3-null mouse has been replaced by the lacZ gene (knock-in), and this was placed under the control of the “a” promoter, which produces the full-length, DNA-binding isoform. Therefore, despite the widespread use of this mouse line, it is limited because expression information is provided on only one of the two isoforms. Being passive repressors (i.e., repressors of transcription factor activity by protein-protein interactions), the “b” isoform may be hugely important, even though it does not have DNA binding capability. A role of the b isoform in regeneration may be supported by work in the developing brain demonstrating that the b isoform is preferentially expressed in neural progenitors of the developing brain (87).
Do “Alternative” Hes/Hey Genes Operate in Pancreatic Cells?
Early reports demonstrated the operation of the STAT3-Ser/Hes3 signaling axis in adult pancreatic cells in vitro and in vivo (43). It may be worth speculating how this pathway may intercept other established signaling pathways involved in pancreas development and function, including the mitogenic FGF pathway, Notch (canonical via Hes1, noncanonical via Hes3), and the hedgehog pathway. These testable hypotheses may provide a useful blueprint for the integration of multiple signaling pathways in pancreas biology.
Early Clues for the Involvement of Alternative Hes/Hey Genes in Pancreas Biology
Early experiments with genetically engineered mice suggested that complex interactions among members of the Hes/Hey gene family profoundly affect pancreatic development. In Hes1-null mice, ectopic pancreas formation is observed in areas of the developing primitive stomach, duodenum, and bile duct (93). Hes1 inactivation induced the miss-expression of pancreas-specific transcription factor 1α (Ptf1α), a bHLH that is important in the formation of the exocrine pancreas and the spatial organization of the endocrine pancreas in these regions (94). Using Hes1-null mice that were crossed with a Cre/loxP reporter mouse, allowing the tracing of Ptf1α cells, it was demonstrated that ectopic Ptf1α-expressing cells had acquired multipotent pancreatic progenitor properties in the Hes1-null mouse that differentiated into cells of the pancreatic lineage, including insulin-producing cells with characteristic nuclear Pdx1 expression. A model where Hes1 suppresses Ptf1α was thus suggested as a means of restricting pancreas formation in the appropriate location. However, Hes1 expression is much broader than the pattern of ectopic pancreas formation in the Hes1-null mice, suggesting that additional mechanisms could contribute to Ptf1α control or the control of other genes helping to limit ectopic pancreas generation. Hes3 and Hes5, genes with established roles in neural development and specification (5,74,82,83,95,96), are also expressed in the developing stomach and small intestine, respectively (24,97), so they could be candidates as cocontrollers of pancreas organogenesis. Another interesting member of the Hes/Hey gene family is Hes6, a proneural gene implicated in glioma progression (98) and prostate cancer aggression, possibly by opposing the hedgehog signaling inhibitor (via Gli1, at least) suppressor of fused (99), with roles in pancreatic islet cells where it opposes Hes1 expression, thus promoting the mature β-cell phenotype (100).
NSC Biology Suggests the Involvement of Noncanonical Signaling Pathways
Specifically for Hes3, however, the data from NSC biology may argue against a compensatory role for Hes1 deletion. This may suggest that other Hes/Hey genes or bHLH genes compensate for Hes1 loss. Assessing the roles of these signaling pathways in pancreatic development and adult pancreatic cells will be highly valuable. Putting together observations from the brain and the pancreas, one may suggest a molecular interaction decreasing its stability and inhibiting promoter activity (21,101), leading to largely mutually exclusive Hes1 and Ngn3 expression patterns.
This is based on a number of observations: 1) isolated adult pancreatic islets from Hes3-null mice have increased Hes1 mRNA levels (43); 2) whereas Hes1 opposes Ngn3 expression (21), Hes3 seems to be required for the increased Ngn3 expression in the exocrine pancreas after STZ-induced damage (92); and 3) the pancreatic phenotypes in the Hes1- and Hes3-null mice are very different (Hes1-null: hindered development ; Hes3-null: no obvious phenotype in undamaged animals ).
Hes3, therefore, may allow for regeneration-induced mechanisms to promote the induction of Ngn3 expression (Fig. 3D). How Hes6 may intercept this network is not yet well understood; however, as we discuss in a later section, Hes6 may activate components of the Shh pathway while suppressing Hes1 expression.
caNotch Levels Differentially Regulate the Balance Between Sox9 and Hes1
In trunk cells, caNotch signaling is an important regulator of Sox9 and Hes1 (102). The levels of activation of Sox9 and Hes1 determine whether Ngn3 will be subsequently induced (leading to endocrine commitment) or suppressed (leading to ductal commitment). These levels appear to be regulated by highly complex interactions among Sox9, Hes1, and Ngn3 (Sox9 induces Hes1 and Ngn3 expression; Hes1 opposes Ngn3 expression) as well as the levels of caNotch signaling (high caNotch leads to the induction of both Sox9 and Hes1; low caNotch signaling leads to the preferential induction of Sox9) (Fig. 4).
More specifically, to demonstrate the dose-dependent effects of Notch activity, whole embryonic (E12.5) pancreas explants were cultured under different concentrations of a γ-secretase inhibitor to oppose Notch receptor activation, and it was demonstrated that Sox9 transcription requires lower levels of Notch activity than Hes1 transcription (51). These data are supported by reports that the expression of Ngn3 and Hes1 is mutually exclusive in the developing pancreas (27,35), as is the expression of Sox9 and Ngn3 at the endocrine progenitor stage (46,103–105). Furthermore, Sox9 occupancy regions on the Ngn3 gene promoter have been identified (104,106). In addition, Sox9 may also regulate Ngn3 via Pdx1 (107,108). (In a seeming paradox, Sox9 also induces Hes1 expression , perhaps as a means of limiting its action on Ngn3.)
In support of a role of caNotch signaling opposing Ngn3 expression, mice deficient in the Notch receptor family ligand Dll1 and RBPJ-κ exhibit reduced Hes1 expression (35). In concert with these observations, Hes1 inhibits the expression of Ngn3 (15), Hes1-null mice exhibit increased Ngn3 activity (24), and Hes1 inhibition induces the redifferentiation of expanded human pancreatic β-cell–derived cells (109). Identifying new regulators of Hes1 and Sox9 may improve our understanding of pancreas development.
Possible Involvement of noncaNotch Branches
In contrast to the caNotch signaling pathway, the noncaNotch branch involves the indirect (likely via Akt, mTOR, and STAT3-Ser phosphorylation, as shown in NSCs ) induction of Hes3, which then suppresses Hes1 transcription (the mechanism for this is not yet known), potentially releasing Ngn3 from Hes1-induced inhibition. The contribution of such a putative mechanism in development is not yet demonstrated because no detailed studies have been performed, although in the adult Hes3-null mouse, no gross morphological phenotypes are obvious (43,88). Perhaps, this mechanism is particularly active during regeneration because Hes3-null mice exhibit increased damage and reduced regeneration of β-cells after STZ damage (43). A technical difficulty in separating caNotch from noncaNotch effects is that, at least in NSC cultures, both pathway branches are sensitive to γ-secretase inhibition (5).
Regardless, the possibility that noncaNotch is also involved in pancreas development and function prompts the question of whether low Notch is really simply low Notch or actually a different equilibrium between caNotch (Hes1-mediated) and noncaNotch (e.g., Hes3-mediated) activity. But is there evidence, even circumstantial, in line with this possibility?
Integrating FGF10 Signaling
The MPC mitogen FGF10 (56) could provide an entry point to Hes3; in fact, it may provide a Notch receptor–independent means of activating both Hes1 (a caNotch target) and Hes3 (a noncaNotch target). We speculate that, in a manner similar to high versus low Notch signaling (46), high versus low FGF10 signaling may also preferentially stimulate the Hes1 (59) versus the Hes3 Notch signaling branch.
During early pancreatic development, FGF10, derived from mesenchymal cells, leads to Sox9 induction via its receptor, Fgfr2b. Sox9 promotes the expression of Fgfr2b, helping to maintain the receptivity of MPCs to FGF10 (53). Pdx1 also promotes Fgfr2b expression, and this may be mediated by Sox9, because Sox9 and Fgfr2b expression are lost simultaneously in Pdx1-deficient pancreata. Therefore, Sox9 is regulated by both the Notch and FGF10 pathways. How these apparently disparate pathways may regulate the same gene is not well understood.
The STAT3-Ser/Hes3 signaling axis may possibly contribute because it provides a molecular mechanism that connects the Notch signaling pathway with classic mitogenic intracellular signals (e.g., Akt, mTOR), leading to the induction of Hes3 or still unidentified mediators. Whether this molecular mechanism is operational in MPCs is not yet reported (Fig. 5).
There are clues suggesting complex molecular mechanisms by which FGF10 may be connected to caNotch and noncaNotch signaling in MPCs: genetic perturbations of FGF10 or its receptor Fgfr2b show that FGF10 signaling leads to hyperplasia, the maintenance of the immature cell state (high Ptf1α, low Pdx1, low Nkx6.1 state), expression of Hes1, low Dll1, and low Ngn3 (58,59). At first sight, these data may argue against the involvement of Hes3 downstream of FGF10 signaling because Hes3 activation may be expected to result in low Hes1 and high Pdx1, at least based on data from mouse insulinoma cell lines and the adult pancreata from Hes3-null mice (43). However, FGF10 signaling can also activate the JAK-STAT pathway, which is a strong inhibitor of Hes3 as well as an inducer of Hes1 in other immature cell types (110). Similarly, in NSC cultures, the cytokine ciliary neurotrophic factor can be converted from a prodifferentiation signal suppressing Hes3 transcription to a mitogen promoting Hes3 transcription simply by cotreatment with a JAK inhibitor (78). These observations raise the intriguing possibility that FGF10 may, perhaps under different levels of activation or cellular and developmental contexts, intercept caNotch or noncaNotch signaling with complex downstream results.
Neural/Pancreas Progenitor Cell Equivalence?
Identifying the cell types in the developing pancreas that may use aspects of the STAT3-Ser/Hes3 signaling axis will require substantial experimentation. Issues to resolve include identifying which cell type expresses Hes3 (or a functional equivalent) and in which cell types STAT3 is phosphorylated on the serine residue and not on the tyrosine residue. As noted earlier, an integrated elucidation of the function of all members of the Hes/Hey gene family may be required.
Few detailed studies on Hes3 expression during development exist (24,73,88,93). Detailed analyses of adult tissues have demonstrated Hes3 expression in putative NSCs; such observations during development may be difficult to make by low-magnification inspection (e.g., whole-embryo immunostaining or in situ hybridization) because Hes3+ cells may be few and Hes3 expression can be low. Therefore, future studies may shed first light on Hes3 expression at different developmental stages of the pancreas.
Is it possible then that this system is operational in bipotent trunk cells? One may speculate that in the decision between the endocrine and ductal commitment, Hes3 could help bias a progenitor to take the endocrine route by opposing Hes1 expression. There is no evidence, to our knowledge, that Hes3 regulates Sox9 itself.
It may be interesting to study whether tip cells express Hes3, because they exhibit low caNotch signaling and because adult Hes3-null mice exhibit phenotypes in acinar cells (the inability to induce Ngn3 after partial STZ damage ). Tip cells express Sox9 (111), but there is currently no evidence that this may be mediated by Hes3; Sox9 induction may be a consequence of low caNotch signaling.
Obvious developmental deficits in Hes3-null mice are not known, although detailed studies are still lacking. This does not mean that Hes3 does not have developmental roles, because other factors (possibly Hes6, for example) may functionally compensate for the loss of Hes3 in these genetic mouse models. However, the current data point toward a clearer role for Hes3 in differentiated, insulin-producing cells: In MIN6 cells, Hes3 overexpression induces Pdx1 expression, and Hes3 RNA interference opposes growth, insulin expression, and insulin sensitivity. In line with the operation of the STAT3-Ser/Hes3 signaling axis (which in NSCs is opposed by JAK activity), these cells can be efficiently cultured in conditions that suppress JAK-STAT signaling, thereby supporting STAT3-Ser phosphorylation and Hes3 expression (43). These results are mirrored by observations reported in adult mice where antibody- and PCR-based techniques demonstrate Hes3 expression in islets; immunofluorescence experiments also suggest the expression of Hes3 in adult human islets. Further, Hes3–knock-in reporter mice confirm Hes3 promoter activity in adult mouse islets and demonstrate promoter activation after STZ-induced damage. The functional significance of pancreatic Hes3 is suggested by observations in the adult Hes3-null mice showing increased sensitivity to STZ damage and impaired regeneration, although no stark phenotypes are obvious in the uninjured adult Hes3-null mice.
Therefore, the STAT3-Ser/Hes3 signaling axis may not be a molecular mechanism that determines a particular progenitor cell type but a signaling module that is used by a variety of plastic cell types in the pancreas, both during development and in the adult. Modeling the operation of this signaling pathway in culture may provide new opportunities in basic science and drug discovery (discussed later).
Who Controls Shh?
Shh plays important roles in pancreatic development and function, but the mechanisms that regulate its expression are not fully understood. Similar to what we discussed for Notch signaling and possibly also for FGF signaling, Shh signaling shows dosage-dependent effects, intercepts both caNotch and noncaNotch signaling pathway branches (e.g., Hes5 and Hes3) (5,112), and NSC biology may contribute to our understanding of Shh regulation in the pancreas.
Shh Signaling in Neural and Pancreatic Development
In neural development, different levels of Shh result in different cellular outcomes, with low Shh levels more prone to promoting self-renewal/proliferation and high Shh more prone to morphogenetic/cell specification decisions (112). The dosage-dependent Shh effects may be linked to the dosage-dependent Notch effects. In fact, there is direct evidence that Shh intercepts both caNotch (e.g., Hes5) and noncaNotch signaling (e.g., Hes3) (5,112).
Early pancreatic development and pancreatic progenitor maintenance requires suppression of Shh and activation of caNotch signaling (involving RBPJ-κ and Hes1). At ∼E9.5, the Shh receptor Ptch1 is still not expressed in the pancreatic epithelium (31). However, it appears ∼1 day later, albeit at low amounts; after birth, Ptch1 expression increases and is widespread in the islets and ducts (113). Therefore, although hedgehog signaling must be suppressed for pancreatic specification, it is possible that it is required for optimal development starting soon after specification, even if at relatively low levels. Indeed, genetic deletion of the Shh mediator Smo in the pancreas epithelium induces delayed expansion of the early pancreatic epithelium and delayed β-cell mass development (113). Several other studies demonstrate a role of hedgehog signaling in the growth of the pancreas (31,33,34,114–118). These seemingly conflicting studies may be consolidated by taking into account the different mechanisms of hedgehog signaling modulation, such as ligand overexpression or Smo deletion. It also suggests that hedgehog signaling opposes the establishment of pancreas organ boundaries in the foregut while subsequently promoting the expansion of early pancreatic epithelium starting at ∼E12.5 (113). Adding to the context-dependent complexity of hedgehog signaling, the actions of this pathway may oppose proliferation of pancreatic epithelial cells at around midgestation. We will address the possibility that noncanonical signaling pathways may mediate some of the pleiotropic effects of Shh.
Effects of Shh on Pancreatic Function
In adult β-cells, conclusions on the roles of Shh are complex, partly because of the different biological systems (cell lines in vitro, in vitro vs. in vivo) used and the different experimental approaches used to modulate Shh (pharmacological or genetic manipulation of different components of the Shh pathway). However, despite differences, these studies show important roles of Shh on adult β-cells. β-Cell lines, the Shh signaling small-molecule inhibitor cyclopamine, and ectopic expression of Shh were used to show that hedgehog signaling promotes insulin secretion and content and insulin promoter activity (119,120). In a similar experimental system, Shh was also demonstrated to provide protective properties when cells were stressed by the addition of proinflammatory cytokines (121).
In vivo, a somewhat different story arises on the role of Shh on adult β-cells. A study that used a different approach to activate hedgehog signaling (using an active version of Gli2, a mediator of hedgehog signaling in β-cells devoid of primary cilia that normally negatively regulate hedgehog signaling ) showed that increased Shh signaling correlated with increased expression of the precursor markers Hes1 and Sox9, both direct targets that are normally excluded from β-cells (123–126). It should be noted that Gli1 and Gli2 overexpression leads to Hes1 induction in several cellular systems, providing a Notch-independent mechanism of inducing Hes1 (126). Therefore, hedgehog signaling in this in vivo system may be perturbed such that it preferentially leads to Hes1 induction at the expense of other target genes, suppressing Pdx1 expression. Overall, therefore, there are strong indications that hedgehog signaling may both promote and suppress Pdx1 expression. We propose a model, fusing data from both the NSC and pancreas fields to consolidate these observations.
As mentioned above, Shh mediates the nuclear localization of Pdx1 in cultured insulinoma cells (120) (Fig. 6A). Hes3 overexpression also induces Pdx1 expression and nuclear localization in insulinoma cells, and chromatin immunoprecipitation on-chip data suggest the direct regulation of the Pdx1 promoter as a putative mechanism (Fig. 6B) (43,127). In NSCs, Hes3 overexpression promotes Shh expression (Fig. 6C) (5). A hedgehog signaling branch may oppose Pdx1 expression through increased Hes1 expression in cells that overexpress Gli2 and lack primary cilia (122) (Fig. 6D). Put together, these observations suggest the possible (and testable) involvement of noncaNotch signaling components in the regulation of Shh and Pdx1 (Fig. 7).
These varied observations raise a number of questions:
Could Hes3 be both an inducer of Shh signaling activity as well as a compensator for the lack of Shh during early pancreatic development and perhaps also during regeneration?
If Gli2 does not induce Pdx1 expression, is there another Gli member that does (128)?
Are there Gli factors that preferentially lead to Hes1 and others to Pdx1 expression?
How important is the inhibitory effect of primary cilia in Pdx1 regulation?
Given the positive role of Sox9 in the formation of primary cilia (51) and the negative role of primary cilia in the mediation of hedgehog signaling (122), how significant might this Sox9/hedgehog signaling cross talk be in terms of development and function?
How much of the inhibitory effect of the hedgehog branch on Pdx1 is mediated by Hes1?
Can We Model Different Signaling States in Culture?
Answering these questions will require a more thorough understanding of the many states that pancreatic cells can adopt during development and disease. It would be a great advantage to researchers if these different cellular/signaling states could be modeled in vitro, contributing to our search for alternative growth pathways that may eventually lead to new therapies.
Again, there are valuable lessons from NSC biology that can be applied to β-cell research. MIN6 cells are a commonly used transformed mouse insulinoma cell line (129). Despite the caveats with transformed cell lines, they have proven a valuable tool to study evoked insulin release. Still, these cells exhibit low nuclear Pdx1 expression, relative to β-cells in vivo, suggesting that they do not model the in vivo state of β-cells perfectly. They also exhibit Hes1 expression, which is normally absent from β-cells. They are typically cultured in serum-containing medium, a common strategy to provide ample yet undefined factors required for cell maintenance and proliferation. NSC biology reveals that serum, a potent JAK/STAT pathway activator, suppresses Hes3 expression and, even more, nuclear Hes3 localization (5,43). This might explain why common insulinoma cell line culture conditions maintain the cells in a low Hes3, low Pdx1, high Hes1 state that does not model the in vivo state of adult β-cells.
We recently showed that indeed MIN6 cells can be cultured in defined (serum-free conditions) even in the presence of a JAK inhibitor, which suppresses STAT3-Tyr phosphorylation (43). Hes3 is expressed under these conditions, and a polyclonal antibody against Hes3 demonstrates nuclear localization. Nuclear Pdx1 expression incidence also increases, providing potential access to its gene targets and representing more accurately β-cells in vivo. Pdx1 expression is regulated by Hes3, because overexpression of Hes3 in these cells invariably leads to high expression of nuclear Pdx1 (43). The Hes3-Pdx1 relation may be direct because Hes3 has been demonstrated to bind to Pdx1 gene promoter regions in MIN6 cells (127). Hes3 RNA interference opposes cell growth and insulin release (43). Switching the same cells between culture conditions induces reversible changes showcasing the bidirectional ability of the cells to promote their growth via distinct signaling pathways. Such approaches may assist in experimental studies by locking β-cells in culture in particular signaling states representing specific aspects of their development and function.
It is possible that Shh and Hes3 signaling have similar outcomes to cells at different developmental stages or signaling states. Despite the similar effects of hedgehog and Hes3 signaling manipulation, the effects of hedgehog signaling were reported using cell culture conditions that place cells in a signaling state that is nonresponsive to Hes3 manipulation. Specifically, the effects of hedgehog signaling modulation were seen in commonly used serum-containing culture systems where Hes3 is excluded from the nucleus and Hes3 RNA interference does not significantly affect cell growth. In contrast, Hes3 manipulation has powerful roles in serum-free, defined culture conditions, where Hes3 is allowed in the nucleus. These results give a glimpse to strategies that may allow locking cells in different states to allow their study. Whether Shh manipulation affects cells in defined conditions is not yet known. It will be important to define cell states using distinct culture conditions and reevaluate the effects of the perturbation of canonical and noncanonical signaling pathway components.
The protection and regeneration of insulin-producing pancreatic islet β-cells of the endocrine pancreas is a major focus of diabetes research (3,4). Compared with other adult tissues, the endocrine pancreas exhibits considerable plasticity in that many stimuli, including age, nutrition, pregnancy, insulin sensitivity, excessive caloric intake, and various paradigms of damage, can affect β-cell proliferation and alter β-cell mass, at least in experimental rodent models. Understanding the signal transduction pathways behind these phenomena will provide potentially new therapeutic avenues. Clues are provided by reports of experimental interventions that manipulate β-cell mass using treatments with known downstream signaling pathways. These include hyperglycemia, incretins, Nodal (a TGF-β family member), vascular endothelial growth factor, Wnt pathway activators, and γ-aminobutyric acid, for example (4). Additional clues are provided by reports pointing out the importance of transcription factors in the specification of the β-cell fate, including Pdx1, MafA, and Ngn3 (130).
It is important to elucidate in detail how these signaling pathways regulate their downstream transcription factors to provide a deeper understanding of pancreatic development and function and to give more precise direction to drug discovery programs. Recent evidence suggests that the view that a signaling pathway is on/off or high/low may be too simplistic because it does not account for the operation of alternative (noncanonical) branches of these pathways that may be highly active while canonical branches may be suppressed. In an example of cross-disciplinary approach, the NSC field may provide a template to evaluate the involvement of such noncanonical signaling pathway branches in pancreatic development and function, providing new ideas that may help understand and manipulate better the plasticity of this organ.
Funding. This work was partly supported by the Helmholtz Alliance Imaging and Curing Environmental Metabolic Diseases (ICEMED), through the Helmholtz Association Initiative and Networking Fund grant 051_40001, Deutsche Forschungsgemeinschaft grant SFB 655 “Cells Into Tissues” Project A24, and Deutsche Forschungsgemeinschaft Clinical Research Unit grant KFO 252.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. J.M. and P.N. collected, assembled, analyzed, and interpreted the data and revised the manuscript. S.W.P. collected, assembled, analyzed, and interpreted the data; revised the manuscript; and provided a critical revision for important intellectual content. G.C., R.D.M., and S.R.B. analyzed and interpreted the data, revised the manuscript, and provided a critical revision for important intellectual content. P.M.J. analyzed and interpreted the data, wrote the manuscript, revised the manuscript, and provided a critical revision for important intellectual content. A.A.-T. conceived the topic, drafted the article, wrote the manuscript, and interpreted and analyzed data. All authors approved the final manuscript. A.A.-T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
- Received August 6, 2015.
- Accepted November 2, 2015.
- © 2016 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.