Pancreatic Mesenchyme Regulates Islet Cellular Composition in a Patched/Hedgehog-Dependent Manner

Pancreas development requires restrained Hedgehog (Hh) signaling activation. While deregulated Hh signaling in the pancreatic mesenchyme has been long suggested to be detrimental for proper organogenesis, this association was not directly shown. Here, we analyzed the contribution of mesenchymal Hh signaling to pancreas development. To increase Hh signaling in the pancreatic mesenchyme of mouse embryos, we deleted Patched1 (Ptch1) in these cells. Our findings indicate that deregulated Hh signaling in mesenchymal cells was sufficient to impair pancreas development, affecting both endocrine and exocrine cells. Notably, transgenic embryos displayed disrupted islet cellular composition and morphology, with a reduced β-cell portion. Our results indicate that the cell-specific growth rates of α- and β-cell populations, found during normal development, require regulated mesenchymal Hh signaling. In addition, we detected hyperplasia of mesenchymal cells upon elevated Hh signaling, accompanied by them acquiring smooth-muscle like phenotype. By specifically manipulating mesenchymal cells, our findings provide direct evidence for the non-autonomous roles of the Hh pathway in pancreatic epithelium development. To conclude, we directly show that regulated mesenchymal Hh signaling is required for pancreas organogenesis and establishment of its proper cellular composition.


Results
Increased Hh signaling in the embryonic pancreatic mesenchyme upon Ptch1 deletion. To manipulate Hh signaling in the pancreatic mesenchyme, we set to inhibit the expression of Ptch1, a negative regulator of Hh signaling transduction, which its deletion was shown to increase expression of target genes 17 . To this end, we generated Nkx3.2-Cre;Ptch1 flox/flox embryos, in which the two copies of Ptch1 are deleted in the Nkx3.2-Cre lineage in a Cre/lox dependent manner 28,29 . We have previously shown that the Nkx3.2-Cre line allows for manipulation of mesenchymal cells, but for no other cell types (including epithelial, endothelial and neuronal) in the developing pancreas 11 . To assess the level of Hh signaling activation, we analyzed the expression levels of Gli1, a target gene of this pathway, in the embryonic pancreas of Nkx3.2-Cre;Ptch1 flox/flox and non-transgenic control littermates (Ptch1 flox ; Cre-negative) at e14.5. Our analysis revealed increased expression of Gli1 in pancreatic tissue of transgenic embryos (Fig. 1A), indicating elevated levels of Hh signaling.
To assess the localization of Hh signaling activation, we monitored Ptch1 expression in the embryonic pancreas. As Ptch1 is a Hh signaling target gene, Ptch1 LacZ mouse line, in which a copy of this gene was knocked in by a LacZ cassette, serves as a reporter for Hh signaling activation 30 . In Ptch1 LacZ e14.5 embryos, LacZ activity is abundant in the mesenchyme surrounding the developing duodenum epithelium (Fig. 1B). However, despite the removal of one copy of Ptch1, and in agreement with previous studies 22,24 , this activity was below detection in the pancreatic mesenchyme (Fig. 1B). Next, we generated Nkx3.2-Cre;Ptch1 LacZ/flox compound embryos, in which one copy of Ptch1 was knocked in by the LacZ transgene, and the other (Ptch1 flox ) depleted in the Nkx3.2-Cre lineage, as described above. Analysis of LacZ activity in Nkx3.2-Cre;Ptch1 LacZ/flox e14.5 embryos pointed to increased Hh signaling in their pancreatic mesenchyme, but not in the epithelium (Fig. 1B). Thus, deleting Ptch1 using the Nkx3.2-Cre mouse line allowed increased mesenchymal Hh signaling in the developing pancreas.
Deletion of Ptch1 in the pancreatic mesenchyme results in reduced pancreatic mass. To analyze the resultant effect of increased mesenchymal Hh signaling on pancreatic development, we analyzed Nkx3.2-Cre;Ptch1 flox/flox embryos and non-transgenic control littermates (Ptch1 flox ; Cre-negative). Nkx3.2-Cre;Ptch1 flox/flox mice died upon birth, likely due to non-pancreatic expression of the Nkx3.2-Cre line in the embryonic gut and stomach mesenchyme and in skeletal somites 29,31 . However, at e18.5 transgenic embryos exhibited comparable appearance and body weight to their non-transgenic control littermates ( Fig. 2A and B). As previously reported upon elevated Hh signaling in the Nkx3.2-Cre lineage 32 , the gastrointestinal tract of transgenic mice was drastically deformed, with short and dilated intestine and misshapen stomach (Fig. 2C). Of note, we were unable to detect splenic tissue, a derivative of the embryonic pancreatic mesenchyme 33 , in transgenic mice. Nonetheless, pancreatic tissue could be detected in Nkx3.2-Cre;Ptch1 flox/flox embryos (Fig. 2C). Pancreatic tissue of transgenic embryos was significantly smaller than that of their non-transgenic littermates (Fig. 2D), implicating that increased Hh signaling in the pancreas mesenchyme affects proper organogenesis.
Elevated Hh signaling leads to expansion of the pancreatic mesenchyme. To further analyze the effect of deregulated mesenchymal Hh signaling on pancreas development, we dissected pancreatic tissue from e18.5 Nkx3.2-Cre;Ptch1 flox/flox and non-transgenic littermate control embryos. While non-transgenic pancreatic tissue contained mostly epithelial cells, histological analysis revealed abundant non-epithelial cells in pancreatic tissue of Nkx3.2-Cre;Ptch1 flox/flox embryos (Fig. 3A).
Scientific RepoRts | 6:38008 | DOI: 10.1038/srep38008 Hh signaling was shown to support proliferation of mesenchymal cells lining the developing gastrointestinal tract 32 . In order to test if elevated Hh signaling leads to changes in the mesenchymal cell layer, we labeled these cells using the pan-mesenchymal marker desmin. In control embryos, desmin-expressing cells formed a thin, cell-wide layer surrounding acinar lobes (Fig. 3B). In contrast, transgenic pancreatic tissues displayed multi-cellular desmin-positive mesenchymal layer that extended away from the epithelium (Fig. 3B). Of note, expanded mesenchymal layer was detected also in between acinar cells. Thus, our results point to hyperplasia of the pancreatic mesenchyme upon increased Hh signaling in these cells.
Pancreatic mesenchymal cells were shown to acquire smooth muscle fate, typical to mesenchymal cells surrounding the gut, in response to increased Hh signaling 19 . To study potential changes in these cells, we employed the R26R-YFP transgenic mouse line, which allows for YFP expression in a Cre-dependent manner 34 . As Nkx3.2-Cre express mesenchymal cells from early stages of pancreas development 11,31 , the inclusion of a Cre-dependent YFP reporter allowed us to trace these cells, regardless of potential phenotypical changes. To analyze for potential acquisition of smooth muscle fate upon elevated mesenchymal Hh signaling, we analyzed the expression of αSMA in YFP-expressing cells. As expected, in control Nkx3.2-Cre;R26R-EYFP;Ptch1 flox/+ pancreatic tissue, αSMA expression was observed in vascular smooth muscle cells (vSMCs), embedded in the pancreatic tissue ( Fig. 3C and C') 35 . However, in Nkx3.2-Cre;R26R-EYFP;Ptch1 flox/flox embryos, this marker was expressed by YFP-expressing cells that extended away from the pancreatic epithelium in transgenic embryos ( Fig. 3C and C'). Furthermore, the fusiform shape of αSMA-expressing cells (Fig. 3C') further suggests pancreatic mesenchymal cells acquire smooth muscle-like phenotype upon deregulated Hh signaling, as previously suggested 19 .
To conclude, in agreement with previous reports 19, 32 , we observed hyperplasia of the pancreatic mesenchyme upon elevated Hh signaling, accompanied by acquisition of smooth muscle-like phenotype.
Reduced epithelial area upon deletion of Ptch1 in the pancreatic mesenchyme. To analyze the resultant effect of increased mesenchymal Hh signaling on pancreatic epithelial development, we analyzed for the presence of the most abundant pancreatic epithelial cell types, β-, αand acinar cells, in Nkx3.2-Cre; Ptch1 flox/flox and non-transgenic littermate controls at e18.5. Immunofluorescence analysis revealed the presence of all three cell types in transgenic embryos (Fig. 4A), with endocrine cells embedded in exocrine tissue. However, morphometric analysis revealed that the combined area of the three epithelial cell population was significantly smaller in transgenic embryos (Fig. 4B). Note that the reduction of epithelial area was more profound than the reduction in pancreatic weight, likely representing the contribution of mesenchyme hyperplasia to the latter (compared Figs 2D and 4B). In addition, the ratio between endocrine (combined insulin-and glucagon-positive area) and exocrine area (amylase-positive area) was smaller in transgenic embryos (Fig. 4C), implicating these two cellular compartments were differentially affected by increased mesenchymal Hh signaling.
Morphology of exocrine tissue was disrupted in transgenic embryos, with more compacted cellular distribution (Fig. 4A,D and E). Our analysis indicated that the typical acellular areas normally found between adjacent lobes is lost in transgenic embryos, and is filled by mesenchymal cells (Figs 3B and 4E). As expected from the reduced epithelial area, morphometric analysis revealed smaller amylase-positive area in Nkx3.2-Cre;Ptch1 flox/flox e18.5 embryos as compared to littermate control (Fig. 4F). To conclude, our findings indicate that deregulated mesenchymal Hh signaling impairs growth and morphology of the exocrine pancreas.
Abnormal islet morphology in Nkx3.2-Cre;Ptch1 flox/flox pancreas. Our analysis indicates that endocrine mass is affected more so than that of exocrine from elevated Hh signaling (Fig. 4C). To analyze for potential changes in endocrine cells upon mesenchymal Ptch1 deletion, pancreatic tissues of Nkx3.2-Cre;Ptch1 flox/flox and non-transgenic littermates e18.5 embryos were immune-stained for insulin, glucagon, and somatostatin. Our analysis revealed that while all three endocrine cell populations are present in transgenic embryos, islet morphology was abnormal (Fig. 5A). Of note, Pdx1 expression by insulin-positive, but not by glucagon-positive cells in transgenic pancreatic tissue pointed to appropriate cell fate acquisition 36 (Fig. 5B).
In mice, β-cells populate the islet core, while αand δ-cells are found in the islet periphery, forming a mantle 2 . As shown in Fig. 5A, transgenic islets displayed a disrupted organization, with some having a α-cell core and βand δ-cell mantle. To quantify the observed morphological changes, we divided islets into three groups based on their core: distinct β-cell core, distinct α-cell core, or with a core formed by both cell types ('mixed'; Fig. 5C). A vast majority of islets in non-transgenic pancreatic tissue had the typical β-cell core, with few having a core containing both βand α-cells (Fig. 5C). In contrast, less than one-third of islets in e18.5 Nkx3.2-Cre;Ptch1 flox/flox pancreatic tissue had a defined β-cell core, with about one-half of their islets exhibited a core containing both βand α-cells. Interestingly, while we observed no islets with a α-cell core in non-transgenic pancreatic tissues, around one-fifth of transgenic islets had a distinct α-cell core (Fig. 5C). Thus, our findings indicate that deregulated mesenchymal Hh signaling leads to disrupted islet morphology.

Abnormal endocrine composition and mass upon deletion of mesenchymal Ptch1. To analyze
if the observed abnormal islet morphology is associated with changes in islet cellular composition, we directly compare the portion of endocrine cell types. To this end, we measured insulin + , glucagon + and somatostatin + area in pancreatic tissues of Nkx3.2-Cre;Ptch1 flox/flox transgenic and control e18.5 embryos. As shown in Fig. 5D, non-transgenic tissues exhibit the expected endocrine cell ratio 2 , with ~70% being insulin + , ~24% being glucagon + and the remaining ~6% being somatostatin + . However, transgenic pancreatic tissues had a reduced insulin + portion to ~50% of endocrine cells, whereas the portion of glucagon + increased to ~41% and that of (D) Bar diagram (mean ± SD) summarizing normalized pancreas weight of transgenic (orange bar) to nontransgenic (cyan bar; set to '1') littermates. n = 5. p value: **P < 0.01, as compared to non-transgenic control, determined using Student's t-test.
To directly analyze for potential changes in αand β-cell mass upon deregulation of mesenchymal Hh signaling, we analyzed their area. Morphometric analysis revealed a significantly reduced area in both these cell populations in e18.5 Nkx3.2-Cre;Ptch1 flox/flox embryos, as compared to littermate control ( Fig. 5E and E'). Furthermore, while the reduction of glucagon + area was proportional to the reduced epithelial area in transgenic embryos, the reduction in insulin + area was more profound than the reduction in total epithelial area (Fig. 5F). In agreement with other studies [23][24][25] , our analysis indicates that β-cell mass is affected more so than that of α-cells from elevated Hh signaling.
Regulated mesenchymal Hh signaling is required for endocrine cell-specific proliferation rates. Our analysis implicates that manipulating Hh signaling in the pancreatic mesenchyme differentially affected αand βcell development. To directly test this possibility, we calculated the ratio between insulin + and glucagon + area at e14.5, e17.5, and e18.5 in Nkx3.2-Cre; Ptch1 flox/flox embryos and littermate controls. As shown in Fig. 5G, in control embryos, insulin + area was slightly larger (by ~1.2 fold) than glucagon + area at e14.5, likely representing increased frequency of precursor differentiation toward a β-cell fate 13 , combined with larger β-cell size (Table 1) 37 . However, insulin/glucagon ratio significantly increased with age in non-transgenic embryos; while insulin + area was 2.4-fold bigger than glucagon + area at e17.5, this difference grew to 2.9-fold at e18.5 (Fig. 5G). This analysis indicates that β-cell population grows at a higher rate than α-cell population during normal pancreas development.
In e14.5 Nkx3.2-Cre;Ptch1 flox/flox embryos, insulin/glucagon ratio was comparable to that of control (ratio of 1.2) (Fig. 5G). Of note, transgenic embryos displayed comparable insulin + and glucagon + area to their littermate controls at this age (Fig. 5H). Interestingly, this ratio remained constant in transgenic embryos (ratio of ~1.2) at the three analyzed ages (Fig. 5G), implicating similar growth rates of the two cell populations upon deregulated mesenchymal Hh signaling.
Our analysis indicates that β-cell population grows at a higher rate than the α-cell population during normal development, but not upon increased mesenchymal Hh signaling (Fig. 5G). As this is observed also between e17.5 and e18.5, after differentiation of the two populations from common endocrine precursors had ceased 1 , we analyzed for potential differences in cell proliferation at e17.5. To this end, we analyzed the portion of αand βcells expressing the proliferation marker phosphorylated Histone H3 (pHH3) in Nkx3.2-Cre;Ptch1 flox/flox and non-transgenic e17.5 embryos. As shown in Fig. 5I, the portion of pHH3 + β-cells in non-transgenic embryos was significantly higher (~4-fold) than that of α-cells. In contrast, the portions of pHH3 + β-cells and α-cells in transgenic embryos were comparable (Fig. 5I). Of note, while the rate of β-cell proliferation at e17.5 decreased in transgenic embryos as compared to control, the rate of α-cell proliferation increased.
This set of experiments indicate that during normal pancreas development, β-cells grow at a higher rate than α-cells to establish proper islet cellular composition. Furthermore, our results suggest that this cell specific growth rates require regulated Hh signaling in mesenchymal cells.
Deregulated Hh signaling leads to elevated mesenchymal mass. Mesenchymal cells surround islet of Langerhans, and their depletion abrogated endocrine cell growth 11 . To analyze whether endocrine-associated   Fig. 6A, while mesenchymal cells formed a thin, cell-wide layer surrounding the islet of Langerhans in control embryos 11 , this layer was considerably thickened in Nkx3.2-Cre;R26R-EYFP;Ptch1 flox/flox embryos (Fig. 6A), similarly to the observed thickening of this layer in the exocrine tissue (Fig. 3).
Endothelial cells were shown to regulate β-cell development 38,39 . Although endothelial cells are not targeted by the Nkx3.2-Cre mouse line 11,40 , they might be affected by changes in the pancreatic mesenchyme. We therefore stained pancreatic tissues of Nkx3.2-Cre;Ptch1 flox/flox and non-transgenic e18.5 embryos for the endothelial marker Platelet Endothelial Cell Adhesion Molecule 1 (PECAM1). Our analysis indicated the presence of endothelial cells in and around islets of both transgenic and control embryos, with similar distribution (Fig. 6B). To conclude, our findings point to increased mesenchymal mass, without apparent change in endothelial area, around islet of Langerhans upon deregulated Hh signaling.

Discussion
Regulation of Hh signaling is essential for proper pancreas development. Here, we analyzed the contribution of mesenchymal Hh signaling to this process. Our results indicate that the ability of the pancreatic mesenchyme to support organogenesis depends on proper regulation of Hh signaling in these cells. To increase Hh signaling in the pancreatic mesenchyme, we generated Nkx3.2-Cre;Ptch1 flox/flox embryos, in which two copies of Ptch1 were deleted in this tissue. Deregulated Hh signaling in mesenchymal cells was sufficient to disrupt epithelial growth, affecting both the endocrine and the exocrine pancreas. However, mesenchymal growth was increased, leading to hyperplasia of this cell layer. We further observed disrupted endocrine cellular composition, with a reduced β-cell portion and abnormal islet morphology. Thus, our findings indicate that the cell-specific growth rates of epithelial cell populations depend on the pancreatic mesenchyme, and requires regulated Hh signaling activity in this cell layer. To conclude, we showed that mesenchymal Hh signaling is required for pancreatic growth and establishment of its cellular composition.
Islets of Langerhans display a characteristic cellular composition, determined during development 2,13 . Our results indicate that in the mouse embryo, pancreatic endocrine cells exhibit specific growth rates, with the β-cell population growing at a higher rate than the α-cell population. In part, this could be an outcome of a higher tendency of endocrine precursors to differentiate to β-cells than to alternative cell fates 13 . In addition, our results suggest that cell-specific proliferation rate might contribute to the stereotypical islet composition, when β-cells proliferate at a higher rate than α-cells do. Deregulated Hh signaling in pancreatic mesenchymal cells, achieved by deletion of Ptch1 in these cells, led to similar βand αcell growth rates toward end of gestation, likely contributing to the observed abnormal islet composition. While we observed abnormal cell proliferation rates in transgenic embryos, this could not fully explain the dramatic reduction in βand α-cell mass. It is therefore possible that endocrine cells proliferate at a higher rate at earlier developmental stages. Alternatively, although normal βand αmass was observed at e14.5, their differentiation rate was affected by deregulated mesenchymal Hh signaling. Of note, β-cell development was shown by others to be more affected than α-cells from deregulated pancreatic Hh 25 , further suggesting their specific growth rate is dependent on restrained Hh signaling. While β-cell function was shown to require cells in the islet microenvironment 38,40 , the postnatal lethality of Nkx3.2-Cre;Ptch1 flox/flox mice prevents us from being able to directly study the role of mesenchymal Hh signaling in this process. Nevertheless, the abnormal endocrine composition observed in transgenic mice would have likely affected the levels of secreted hormones.
Mouse islets have a distinct β-cell core, whereas the core of human islets were reported to be populated by both αand β-cells 2,41 . Recent studies suggest that the morphology of human islets depends on their size, whereas small islets resemble the morphology of mouse islets, with a distinct β-cell core 42,43 . Furthermore, the arrangement of endocrine cells within the islets allows for proper homotypic and heterotypic interactions, essential for proper functioning 2,44 . We observed that increased mesenchymal Hh signaling leads to abnormal islet organization, with most islets lacking the typical β-cell core. It could therefore be possible that the αto βcell ratio dictates islet morphology, and that the abnormal endocrine cell ratio in transgenic embryos leads to the abnormal islet morphology. This possibility is supported by the abnormal islet morphology observed upon increased β-cell death during development 45 . Alternatively, similarly to neurons 46 , mesenchymal cells may directly dictate pancreatic endocrine cell arrangement and islet morphology.
Pancreas morphogenesis and growth depend on proper interactions of the developing epithelium with cells in its surrounding. Mesenchymal-epithelial interaction was shown to be promote epithelial branching 11,47 . Our findings indicate that deregulated mesenchymal Hh signaling leads to hyperplasia of mesenchymal cells and abrogated epithelial expansion and branching. The pancreatic endothelium was shown to restrain exocrine growth, when hyper-vascularization repress pancreas expansion and branching [48][49][50] . It is therefore possible that, similarly to endothelial cells, mesenchymal hyperplasia does not allow proper mesenchymal-epithelial interactions, leading to the observed morphological phenotype. Alternatively, yet to be identified Hh-dependent mesenchymal cues may regulate pancreatic growth and branching. positive areas at e14.5, represented in arbitrary units. n = 3. (I) Bar diagram (mean ± SD) shows the percentage of insulin (Ins + ), glucagon (Gluc + ) -positive cells co-expressing phosphorylated Histone H3 (pHH3). n = 3. P values: *p < 0.05, **p < 0.01, ***p < 0.005, NS = non-significant. Comparison between transgenic and nontransgenic embryos are marked with black font, while comparison between samples of the same genotype are marked with orange (transgenic embryos) and cyan (non-transgenic embryos) fonts, all determined using student's t-test.
Scientific RepoRts | 6:38008 | DOI: 10.1038/srep38008 The role of Hh signaling in pancreas development was established by a series of studies that manipulated this pathway in both epithelial and mesenchymal compartments 19,21,[23][24][25] . These studies showed that restrained pancreatic Hh signaling is crucial for epithelial expansion and proper β-cell mass and function. Epithelial and β-cell specific manipulations of this pathway recapitulated some of these phenotypes, including reduced endocrine mass and impaired β-cell function, indicating a cell intrinsic role of Hh signaling 22,26,27 . By manipulating this pathway in the pancreatic mesenchyme, but not in its epithelium, we directly showed a cell extrinsic roles of Hh signaling in epithelial expansion. Of note, we could not observe ectopic pancreas in Nkx3.2-Cre;Ptch1 f/f embryos, further indicating that this phenomena is an outcome of deregulated Hh signaling in the pancreatic epithelium 51 . Importantly, neither epithelial nor mesenchymal -specific elevation of Hh signaling fully recapitulate the pancreatic agenesis phenotype observed upon pancreas-wide manipulation of this pathway 19,[23][24][25] . The different manipulations of this pathway (removal of regulatory elements, ectopic ligand, and transcription factors' expression) may lead to different levels of pathway activation. Alternatively, deregulating Hh signaling in both the epithelium and the mesenchyme might have a synergetic, negative effect on epithelial growth. The severity of pancreatic phenotype observed upon systemic manipulation of Ptch1 expression 24 , as compared to the phenotype described here upon mesenchymal manipulation of this gene, supports the requirement of regulated Hh signaling in both pancreatic epithelium and mesenchyme.
Hh signaling was shown to be required for proliferation of mesenchymal cells of the gastrointestinal tract 32 . While along the gut tube mesenchymal cells form the smooth muscle layer that controls its local movement, the adult pancreas lacks this layer and contains relatively few mesenchymal cells (including pancreatic stellate cells, vSMCs, and pericytes) 35,52 . Therefore, the expression of Hh ligands along the gut tube, and their exclusion from  the developing pancreas, may reflect a differential need for mesenchymal expansion 19,20,32 . This notion was first suggested by Apelqvist and colleagues in 1997, in a seminal study reporting acquisition of a gut-like phenotype by pancreatic mesenchymal cells upon ectopic Shh expression 19 , and was further supported by others [23][24][25] . Furthermore, Hh signaling was shown to promote stroma expansion during the progression of pancreatic ductal adenocarcinoma (PDAC) 53 . Here, we were able to directly show that elevated Hh signaling leads to expansion of the mesenchymal layer in a cell-autonomous manner. Hence, regulated Hh signaling may be required for establishing a proper epithelial-mesenchymal ratio in the digestive system, allowing for proper size and functioning of these organs.

Morphometric quantifications.
For all measurements presented in this study the following regimen was applied: the entire pancreatic tissue, including both dorsal and ventral buds, was embedded in paraffin wax and cut into 5 μm thick sections. For e18.5 and e17.5 embryos, every fifth section (20% of total tissue) was immuno-stained with indicated antibodies (as described above), where each transgenic tissue was processed and stained in parallel with its littermate control. For e14.5 embryos, half of the sections was immuno-stained with indicated antibodies (as described above), where each transgenic tissue was processed and stained in parallel with its littermate control. Sections were automatically imaged using Keyence BZ-9000 microscope (Biorevo). For all quantifications, with the exception of the measurement of islet morphology and pHH3 expression, all acquired images were analyzed using imageJ software (NIH). For analysis of islet morphology, 50-70 islets from each embryo were manually scored, blind to genotype. For analysis of percentage of pHH3 expressing cells, at least 150 cells from each analyzed cell type in each embryo were manually scored, blind to genotype.