Volume 247, Issue 7 p. 924-933
Research Article
Free Access

Role of the Wilms' tumor suppressor gene Wt1 in pancreatic development

Laura Ariza

Laura Ariza

Department of Animal Biology, Faculty of Science, University of Málaga, Málaga (Spain) and Andalusian Center for Nanomedicine and Biotechnology (BIONAND), Malaga, Spain

Search for more papers by this author
Ana Cañete

Ana Cañete

Department of Animal Biology, Faculty of Science, University of Málaga, Málaga (Spain) and Andalusian Center for Nanomedicine and Biotechnology (BIONAND), Malaga, Spain

Search for more papers by this author
Anabel Rojas

Anabel Rojas

Andalusian Center of Molecular Biology and Regenerative Medicine (CABIMER) and Centro de Investigación Biomédica en Red de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM), Sevilla, Spain

Search for more papers by this author
Ramón Muñoz-Chápuli

Ramón Muñoz-Chápuli

Department of Animal Biology, Faculty of Science, University of Málaga, Málaga (Spain) and Andalusian Center for Nanomedicine and Biotechnology (BIONAND), Malaga, Spain

Search for more papers by this author
Rita Carmona

Corresponding Author

Rita Carmona

Department of Animal Biology, Faculty of Science, University of Málaga, Málaga (Spain) and Andalusian Center for Nanomedicine and Biotechnology (BIONAND), Malaga, Spain

Correspondence to: Rita Carmona, Department of Animal Biology, Faculty of Science, University of Málaga, 29071 Málaga, Spain. E-mail: [email protected]Search for more papers by this author
First published: 30 April 2018
Citations: 11

Abstract

The Wilms tumor suppressor gene (Wt1) encodes a transcription factor involved in the development of a number of organs, but the role played by Wt1 in pancreatic development is unknown. The pancreas contains a population of pancreatic stellate cells (PSC) very important for pancreatic physiology. We described elsewhere that hepatic stellate cells originate from the WT1-expressing liver mesothelium. Thus, we checked if the origin of PSCs was similar. WT1 expression is restricted to the pancreatic mesothelium. Between embryonic day (E) 10.5 and E15.5, this mesothelium gives rise to mesenchymal cells that contribute to a major part of the PSC and other cell types including endothelial cells. Most WT1 systemic mutants show abnormal localization of the dorsal pancreas within the mesentery and intestinal malrotation by E14.0. Embryos with conditional deletion of WT1 between E9.5 and E12.5 showed normal dorsal pancreatic bud and intestine, but the number of acini in the ventral bud was reduced approximately 30% by E16.5. Proliferation of acinar cells was reduced in WT1 systemic mutants, but pancreatic differentiation was not impaired. Thus, mesothelial-derived cells constitute an important subpopulation of pancreatic mesodermal cells. WT1 expression is not essential for pancreas development, although it influences intestinal rotation and correct localization of the dorsal pancreas within the mesogastrium. Developmental Dynamics 247:924–933, 2018. © 2018 Wiley Periodicals, Inc.

ABBREVIATIONS

  • DAPI
  • 4′,6-diamidine-2-phenylidole-dihydrochloride
  • E
  • embryonic day
  • EMT
  • epithelial–mesenchymal transition
  • GFAP
  • glial fibrillary acidic protein
  • GFP
  • green fluorescent protein
  • PBS
  • phosphate buffered saline
  • PSC
  • pancreatic stellate cells
  • TRITC
  • tetramethylrhodamine isothiocyanate
  • YFP
  • yellow fluorescent protein
  • Introduction

    The Wilms tumor suppressor gene (Wt1) encodes a C2H2-type zinc-finger transcription factor that can appears in mammals under different isoforms, participating in transcriptional regulation, RNA metabolism and protein–protein interactions (Hohenstein and Hastie, 2006; Morrison et al., 2008; Toska and Roberts, 2014). Wt1 has been involved in the development of several organs, including kidneys and gonads (Kreidberg et al., 1993), spleen (Herzer et al., 1999), adrenals (Moore et al., 1999), liver (IJpenberg et al., 2007; Asahina et al., 2011), heart (Norden et al., 2010; Martinez-Estrada et al., 2010; Cano et al., 2016), lungs (Cano et al., 2013), and diaphragm (Carmona et al., 2016). However, there are no reports in the literature about the role played by Wt1 in pancreatic development, despite the parallelisms between liver and pancreas organogenesis.

    In mouse, pancreas develops from dorsal and ventral primordia derived from the endoderm (reviewed in Gittes, 2009). The dorsal primordium emerges through evagination of the endodermal epithelium starting by embryonic day (E) 9.5. The ventral pancreatic bud appears later, by stage E10.5. Splanchnic mesodermal cells induce growth and branching of these buds. The endodermal cells give rise to the main pancreatic cell types, the exocrine acinar cells, ductal cells and endocrine cells from the islets of Langerhans. Mesodermal cells contribute to the vascular and connective tissue of the pancreas, and also to a pancreas-specific cell type, the pancreatic stellate cells (PSCs) (Kim and Hebrok, 2001; Gittes, 2009).

    In normal adult pancreas, the pancreatic stellate cells (PSCs) are quiescent, star-shaped cells with a periacinar distribution. They contain vitamin A-rich lipid droplets in the cytoplasm and exhibit positive immunostaining for desmin and glial fibrillary acidic protein (GFAP) (Omary et al., 2007). In normal pancreas, PSCs are involved in keeping the epithelial integrity of the pancreatic acini by means of maintenance of the basement membrane (Riopel et al., 2013; Means, 2013). Furthermore, when activated by profibrogenic stimuli such as inflammatory cytokines or oxidant stress, PSCs proliferate and transform into myofibroblast-like cells, becoming the major source of extracellular matrix. PSC have also recently become a focus of interest because they play a key role in pancreatic cancer, a malignancy with a high prevalence and poor prognosis (Chronopoulos et al., 2016; Karnevi et al., 2016; Pothula et al., 2016). Despite this capital importance in normal and pathological conditions, the embryonic origin of PSCs remains unknown.

    PSCs are similar to the hepatic stellate cells located in the perisinusoidal space of the liver. These cells share many features with PSCs and they are also involved in the fibrotic process. It has been described that the liver mesothelium contributes to the hepatic stellate cells and vascular endothelium during development (IJpenberg et al., 2007; Asahina et al., 2011). The local emergence of mesenchymal cells through a localized epithelial–mesenchymal transition (EMT) of the coelomic epithelium or mesothelium lining developing organs has been extensively studied in some cases, such as the heart, where the epicardium gives rise to epicardial-derived cells that contribute to the cardiac vascular and connective tissues (Ruiz-Villalba and Pérez-Pomares, 2012). In other organs, such as the lungs, liver and gut, the developmental fate of the mesothelial-derived mesenchyme and their importance for visceral morphogenesis has been also demonstrated (for a review, see Ariza et al., 2016). Thus, we aimed to check if pancreatic mesothelium also supplies mesenchymal cells to the developing pancreas and if these mesenchymal cells account for the origin of PSCs.

    Our results confirmed that WT1 protein is only expressed in the mesothelium of the developing pancreas, allowing for tracing of the mesothelial-derived cells with the Cre/LoxP system. During the early stages of pancreas morphogenesis, this mesothelium shows the typical features of EMT. Mesothelial-derived cells, identified by yellow fluorescent protein (YFP) expression in our model, differentiate into a major part of the PSCs and contribute to other cell types, including a part of the vascular endothelium. Conditional deletion of WT1 by the time in which the EMT process is generating the PSC, causes a delay in the growth of the ventral pancreatic bud, but the development of the dorsal bud and the PSC are normal. Systemic WT1 mutants die during the early stages of pancreatic development, but they show defects in the mesogastrium, anomalous localization of the dorsal pancreatic bud in the mesentery, intestinal malrotation and reduced proliferation of acinar cells. Thus, mesothelial-derived cells originated by an EMT constitute a significant subpopulation of mesodermal cells during pancreas development and modulate its growth, although WT1 seems to be basically dispensable for pancreatic differentiation.

    Results

    Pancreatic Mesothelium Expresses WT1 and Gives Rise to Mesenchymal Cells During Development

    WT1 expression in the pancreas is basically restricted to the mesothelium in all the stages studied. A faint WT1 immunoreactivity can be seen in a few submesothelial cells from early (E12.5–E13.5) embryos suggesting down-regulation of WT1 expression during the EMT of the mesothelium, as described in other organs (Cano et al., 2013). However, no WT1 expression was detected in the mesoderm surrounding the pancreatic epithelium (Fig. 1A–C). Thus, the driver Wt1Cre is a reliable marker for the mesothelial-derived cells during pancreatic development.

    Details are in the caption following the image

    Origin of pancreatic stromal cells from the embryonic mesothelium. A–C: Wt1Cre;R26REYFP embryo, E12.5 (A), E13.5 (B) and E15.5 (C). Wt1 protein is shown in red and the YFP reporter of the Wt1 lineage is shown in green. Actual Wt1 expression is restricted to the pancreatic mesothelium in all cases (arrows), although cells of the Wt1 lineage are already present in the pancreatic stroma by E12.5, where a faint Wt1 immunoreactivity can be seen in a few submesothelial cells (arrowheads in A). D: Wt1Cre;R26REYFP embryo, E12.5. Laminin immunoreactivity is shown in red. The basal lamina is disorganized in areas where YFP + cells are more abundant (insert) suggesting EMT. E: Wt1Cre;R26REYFP embryo, E12.5. Cytokeratin immunoreactivity (in red) shows a characteristic punctiform pattern in mesenchymal cells, suggesting collapse of the cytokeratin cytoskeleton, another feature of EMT (insert). F: Wt1Cre;R26REYFP embryo, E12.5. E-cadherin immunoreactivity (red) is lacking in areas of the mesothelium by this stage (upper insert) while other mesothelial cells show E-cadherin immunoreactivity (lower insert). Loss of E-cadherin expression in an epithelium is another feature of EMT. G: Wt1Cre;R26REYFP embryo, E15.5. Laminin (red) is detected by this stage in a continuous basal lamina of the mesothelium (arrows). H: Immunolocalization of E-cadherin (red) and RALDH2 (green) in an E15.5 embryo. Epithelialization of the mesothelium at this stage is demonstrated by expression of E-cadherin between mesothelial cells (arrows). I: Wt1CreERT2;R26REYFP embryo, E14.5. Expression of the reporter YFP (green) was induced at E9.5. Only a few cells appear labelled in the pancreatic stroma, suggesting a later origin for the mesothelial-derived cells. However, positive cells are very abundant in the Sertoli cells of the testis (T). J: Wt1CreERT2;R26REYFP embryo, E15.5, with reporter induction between the stages E9.5 and E11.5. Mesothelial-derived cells, belonging to the Wt1-expressing lineage, are far more abundant than in the previous case. I, intestine; LI, liver; P, pancreas; ST, stomach. Scale bars = 50 μm in B–G; 25 μm in A,H; 100 μm in I.

    Morphological features of EMT were observed in the pancreatic mesothelium at the earliest stages studied (E12.5 and E13.5), including discontinuous laminin immunoreactivity, reduced E-cadherin expression, loose lateral adhesion between mesothelial cells and basal cytoplasmic processes (Fig. 1D–F). These signs coincided with a punctate pattern of cytokeratin immunostaining in the submesothelial cells, suggesting collapse of the epithelial-type cytoskeleton (Fig. 1E). Continuous laminin immunoreactivity appears in the basal lamina of the pancreatic mesothelium by E15.5, and mesothelial cells express E-cadherin in later stages (Fig. 1G,H), suggesting a down-regulation of the EMT process. RALDH2, the main enzyme for retinoic acid synthesis in the mesoderm, was immunolocalized in the pancreatic mesothelium at all the stages studied (Fig. 1H).

    Time lapse video recording of pancreatic explants obtained from an E12.5 Wt1Cre;R26REYPF embryo and cultured for 48 hr, showed areas of EMT where YFP+ cells were migrating from the mesothelium toward the growing acini (Fig. 2A,C and Supplemental video 2). In the initial stage of the explant (E12.5), many YFP + cells are already located in the pancreatic stroma, because mesothelial EMT is active between E9.5 and E11.5, as described below. Other areas of the explant showed a thick, quiescent mesothelium lacking of signs of EMT (Fig. 2B and Supplemental Video S1, which is available online). Interestingly, YFP+ cells were observed migrating and intercalating between the branching acini (Fig. 2D and Supplemental Video S3).

    Details are in the caption following the image

    Frames obtained from time lapse videos of a E12.5 Wt1Cre;R26REYPF pancreatic explant after 48 h in culture. A: Clear field image of the explant, marking the areas where the frames shown in panels B, C, and D were obtained. B: Area of quiescent mesothelium, which appears thick and lacks of signs of EMT. C: Area of active mesothelium, showing signs of EMT. The cells marked with colored dots are migrating toward the acini located at the right of the figure. D: Group of developing acini. Cytoplasmic processes of YFP + cells can be seen intercalating between the branching endodermal tissue and defining the nascent acini (arrows). Scale bar = 100 μm in A.

    Cell counts on confocal images taken from two E12.5 Wt1Cre;R26REYPF embryos (four images each) show that 58.6 ± 2.1% of the mesenchymal cells of the pancreas are YFP+, and thus they belong to the WT1-expressing cell lineage. Induction of the Cre-recombinase expression by E9.5 in the Wt1CreERT2;R26REYFP model provoked the staining of very few cells in the pancreatic stroma of E14.5 embryos (Fig. 1I), contrasting with the massive staining of Sertoli cells in the testis. However, induction between E9.5 and E11.5 greatly increased the number of stromal YFP+ cells by E15.5 (Fig. 1J). This suggests that most cells derived from the WT1-expressing cell lineage (i.e., putative mesothelial-derived cells) originate after E10.5, when the pancreatic buds start growing.

    A Major Part of Pancreatic Stellate Cells Derives From the WT1-Expressing Cell Lineage

    YFP colocalized with desmin, a marker of PSC, in many periacinar mesenchymal cells from the Wt1Cre;R26REYPF embryos from E14.5 on (Fig. 3A,B). Some of these cells project thin cytoplasmic projections between the acinar cells, a characteristic feature of PSC (Fig. 3A). WT1-expressing lineage cells also contribute to the subpopulation of desmin-positive, islet PSC (Fig. 3B). Colocalization of YFP with GFAP, another marker of pancreatic stellate cells, confirmed the mesothelial origin of islet PSC in older embryos, approximately E18.5 (Fig. 3C). No differences were found between dorsal and ventral pancreatic buds in the contribution of WT1-expressing lineage cells (Fig. 4)

    Details are in the caption following the image

    Developmental fate of the Wt1-expressing cell lineage in Wt1Cre;R26REYFP embryos. A,B: Desmin, a marker of PSC, colocalizes with YFP (arrows) in many cells of the developing pancreas by the stages E14.5 (A) and E18.5 (B). The thin prolongations of PSC can be seen between acinar cells (arrowheads in A). In the E18.5 embryo YFP+/Desmin + cells can be seen in a developing islet (IS) (arrowhead in B). These probably are islet stellate cells. Note colocalization around a vessel (V). The musculature of the intestine (I) is also desmin immunoreactive. C: GFAP also colocalizes with YFP in some presumptive islet pancreatic stellate cells by the stage E18.5 (arrows). GFAP + nerve fibers (NF) can be seen around a vessel. D: Neuron glia antigen-2 (NG2), a marker of pericytes and other perivascular cells colocalizes with YFP in the wall of the developing pancreatic vessels (V) and capillaries (arrows) (stage E18.5). E,F: Colocalization of Pecam-1/CD31 with YFP in the pancreatic endothelium (arrows) by E16.5 (E) and E18.5 (F). Note the presence of YFP + cells in the media of the developing arteries (arrowheads). Scale bars = 50 μm.

    Details are in the caption following the image

    Wt1Cre;R26REYPF E16.5 embryo. No differences were found in the contribution of YFP + cells in the dorsal and ventral pancreas. In both cases we can see a similar contribution to desmin-expressing and endomucin-expressing cells (arrows). Scale bars = 50 μm.

    The contribution from mesothelial-derived cells to the vascularization of the pancreas was also studied. Several perivascular cells expressing NG2 (a pericyte marker) are YFP+ (Fig. 3D), and YFP also colocalized with the endothelial marker CD31/Pecam-1 (Fig. 3E,F).

    To confirm and to quantify the contribution of YFP+ cells to the pancreatic endothelium, we performed analytic flow cytometry of disaggregated pancreas obtained from four E14.5 embryos. The results (Fig. 5) show that approximately a third of all the pancreatic cells by this stage belongs to the WT1-expressing cell lineage (34.7% ± 1.9; mean ± SEM; n = 4). 3.1% ± 0.2 of the total cells analyzed are endothelial and from them, 9.2% ± 1.1 are YFP+.

    Details are in the caption following the image

    Analytical cytometry of developing pancreas at the stage E14.5. Representative experiment of the data obtained from four embryos. Approximately 10% of all the endothelial cells are YFP + (i.e., derived from a Wt1-expressing cell lineage). approximately 35% of the pancreatic cells, by this stage, are derived from the same lineage.

    WT1 Deletion in Mesothelium Causes a Delay in Ventral Pancreas Growth, Defective Mesogastrium, Abnormal Dorsal Pancreas Location, and Intestinal Malrotation

    To check if WT1 expression is required for pancreatic development we studied both, embryos with systemic deletion of WT1 (Wt1GFP/GFP; green fluorescent protein) and Wt1CreERT2;Wt1flox embryos induced with tamoxifen at different times. Most systemic WT1-null embryos die before the stage E13.5 and their pancreas are too small to detect anomalies in their development. The eight Wt1GFP/GFP embryos obtained by the stages E13.5 and E14.5 showed pancreas of similar size when compared with the control heterozygote littermates (Wt1GFP/+) (Fig. 6A–H). However, these embryos with loss of WT1 function showed defective development of the mesogastrium, leading to an abnormal location of the dorsal pancreas, that was embedded in the mesenteric wall. In one case, pancreatic tissue was observed inside the liver, due to this anomalous position (Fig. 6G). Surprisingly, in six of the eight mutant embryos analyzed the duodenum was located close to the larger curvature of the stomach, at the left side, suggesting an intestinal malrotation (Fig. 6E–H). In some of these cases, the hindgut was abnormally placed at the right side (Fig. 6E,F).

    Details are in the caption following the image

    Pancreatic phenotype after systemic (Wt1GFP/GFP) and conditional (Wt1 Wt1CreERT2;Wt1flox) deletion of WT1. A–H: E13.5–E14.5 embryos with systemic deletion of Wt1 (B,D–H) and control littermate (A,C). The dorsal mesogastrium (MG) is thickened in the mutant shown in B and remains largely attached to the mesentery (ME). The dorsal pancreatic bud (DP) is covered by the mesothelium (M) in the control (C), but this occurs only partially in this mutant, that shows the pancreas partially located in the mesentery (ME) (D). Note the abnormal development of the mesonephros (MN in A, arrows in B), gonads (G) and adrenals (AD) in the mutant embryo, as well as the hypoplastic liver (L). In the mutant embryos shown in panels E–H, the phenotype is more severe, the ventral mesogastrium is absent and the dorsal pancreas is embedded in the mesentery. In one case, the dorsal pancreas grows into the liver (panel G). In these mutant embryos the intestine is abnormally rotated, and the duodenum (D) is located at the left side, close to the stomach (ST). In two mutant embryos the hindgut (HG) is also malpositioned at the left side (panels G,H) VP: ventral pancreatic bud. I,J: WT1 is efficiently deleted in the mesothelium of the ventral bud of the pancreas of E15.5 Wt1CreERT2;Wt1flox embryos after tamoxifen treatment between E9.5 and E11.5. K,L: Deletion of WT1 by tamoxifen treatment between E9.5 and E12.5 causes a delay in the development of the ventral pancreatic bud, whose acini appear more disperse in this E16.5 embryo. No differences can be found in the dorsal pancreatic bud, mesogastrium or duodenum. Note the lack of kidneys (K) and the normal mesogastrium in the mutant embryo. Scale bars = 500 μm in A,B,E–H,K,L; 100 μm in C,D; 50 μm in I,J.

    Tamoxifen treatment of Wt1CreERT2;Wt1flox embryos between E9.5 and E11.5 (three doses) and fixed at stages E15.5 and E16.5 suppressed WT1 expression in the pancreas (Fig. 6I,J), but this deletion did not cause differences in pancreatic development (not shown). However, WT1 deletion induced between E9.5 and E12.5 (four doses) and analyzed at E16.5 showed a significant delay in the development of the ventral pancreatic bud, characterized by a smaller number and smaller density of acini as compared with the controls (Figs. 6K,L, 7A,B). However, anomalies were not found in dorsal pancreatic buds, intestine or mesogastrium in these mutant embryos. (Fig. 6K,L).

    Details are in the caption following the image

    A,B: Results of image analysis on histological sections from E16.5 Wt1CreERT2;Wt1flox and control embryos, treated with tamoxifen between E9.5 and E12.5. A: Average number of acini in the dorsal (D) and ventral (V) pancreatic buds from four conditional mutant and four control embryos. Between 4 and 6 sections were analyzed from each embryo (39 sections in total). B: Acinar density calculated by number of acini/pancreatic surface. Both, number and density of acini in the mutant ventral bud were significantly lower as compared with the controls (*P < 0.05; ** = p<0.01). No differences were found between the dorsal pancreatic buds. C: Box plots of the frequency of PH3 + cells by acinar surface in dorsal and ventral pancreas of E13.5–E14.5 WT1 systemic mutants. The horizontal bar represents the median, the boxes contain the values between the Q1 and Q3 quartiles and the vertical bars represent the full range of the values. A significant reduction (P < 0.05, Mann-Whitney U-test) was found when the data from dorsal and ventral pancreas were gathered and compared with the controls (N = 6 mutants, 8 controls). Proliferation index was also lower in dorsal and ventral pancreas separately, but the differences were not significant at the P < 0.05 level (P = 0.06 and 0.08, respectively).

    We then studied the proliferation of the pancreatic acinar cells in systemic Wt1GFP/GFP mutant embryos by immunolocalization of phosphohistone-H3 (Fig. 7C). The mitotic rate (expressed as number of PH3 + cells by acinar surface) was significantly lower than the obtained from control Wt1GFP/+ littermates when we considered dorsal and ventral pancreas together (P = 0.012; Mann-Whitney U-test; Fig. 7C). Dorsal and ventral mutant pancreas considered separately showed a trend to reduced proliferation, very close to statistical significance (P = 0.06 and P = 0.08, respectively).

    All E16.5 control and mutant embryos with conditional deletion of WT1 showed similar immunoreactive patterns of amylase, insulin, glucagon, Ptf1a, mucin, desmin, E-cadherin, PDX1, neurogenin, and RALDH2 (some examples shown in Fig. 8). This suggests that the developmental delay did not affect the pancreatic differentiation. Particularly, the number of desmin+, PSCs was similar in both, conditional WT1 mutant and control embryos.

    Details are in the caption following the image

    A–F: Histological sections from E16.5 Wt1CreERT2;Wt1flox and control embryos, treated with tamoxifen between E9.5 and E12.5. Despite the conditional deletion of WT1, the differentiation of the pancreas appears normal as suggested by the expression of mucin, amylase, insulin, desmin or RALDH2. Scale bars = 100 μm in A–D; 50 μm in E,F.

    Adult Mesothelium Does Not Contribute to Renewal of the Pancreatic Stellate Cell Population

    To check if the adult mesothelium is still contributing to the stellate cell population in adult mice we induced Wt1 reporter expression in a Wt1CreERT2;R26REYPF adult mouse (4 weeks old) and then we checked the presence of YFP + stellate cells in the pancreas after one month. The pancreatic mesothelium was YFP+, but no positive cells were found inside the pancreas, thus suggesting the lack of a postnatal contribution of the mesothelium to the pancreatic stellate cells (Fig. 9).

    Details are in the caption following the image

    Induction of Wt1 reporter expression in a Wt1CreERT2;R26REYPF adult mouse (four weeks old). After 1 month, the pancreatic mesothelium was YFP+, but no positive cells were found inside the pancreas, suggesting lack of postnatal contribution of mesothelial-derived cells to this organ. Scale bar = 50 μm

    Discussion

    We show in this study that the pancreas, as described for other viscera (for a review, see Ariza et al., 2016), receives a substantial contribution of mesothelial-derived cells originated by a process of EMT. Our findings confirm, using a genetic model of lineage tracing, the recently published results obtained by direct labeling of the pancreatic mesothelium (Angelo and Tremblay, 2018). We have also shown that WT1 is expressed in the embryonic mesothelium of the pancreas and it is rapidly down-regulated as mesenchymal cells arise from it and migrate into the developing organ, intercalating between the branching endodermal tissue. Thus, Wt1Cre is a reliable driver for cell tracing of the mesothelial-derived cells during pancreatic development. The main part of this process of generation of pancreatic stroma occurs between E10.5 and E14.5, as suggested by the scarce labelling of mesenchymal cells when reporter activation is performed by E9.5 and by the epithelialization of the pancreatic mesothelium and the disappearance of EMT signs by E15.5. Mesothelial-derived cells contribute to the vascularization of the developing pancreas giving rise to a minor part of the endothelium (approximately 10% by midgestation) and perivascular cells.

    Furthermore, as it has been also described for other developing organs, mesothelial-derived cells contribute to an organ-specific cell type, the pancreatic stellate cells, including the subtype known as islet stellate cells (Zha et al., 2014). This contribution is significant, accounting for more than a half of the PSC by midgestation. In adults, we have seen that approximately 15–30% of the PSC still belong to a WT1-expressing lineage (not shown), but a postnatal contribution of the mesothelium to the PSC population seems not significant (Fig. 9). However, we cannot conclude that the adult PSC of the WT1-expressing lineage have originated from the embryonic mesothelium, because postnatal expression of WT1 occurs during PSC activation (Regel et al., 2015, and our unpublished observations). Anyhow, the embryonic origin of the PSC shows commonalities with the origin of the similar hepatic stellate cells, many of them also emerging from the embryonic mesothelium (IJpenberg et al., 2007; Asahina et al., 2011).

    Despite this significant contribution of WT1-expressing cells to the pancreatic stroma, WT1 seems to be basically dispensable for pancreas development, differently to many other viscera, as detailed in the introduction. After conditional WT1 deletion between E9.5 and E12.5, pancreatic bud develops normally at least until E16.5, although the growth of the ventral pancreatic bud growth is delayed, showing a 30% reduction in the number and density of acini. This delay can be explained by the decrease in proliferation of endodermal cells that we have observed in systemic WT1 mutants between E13.5 and E14.5. The reduced proliferation suggests some kind of WT1-dependent signaling mechanism acting in the mesothelial-derived cells. In fact, the liver size is also hypoplasic in WT1-null mice (IJpenberg et al., 2007).

    Pancreatic size was normal in E13.5 and E14.5 WT1 systemic mutants (the oldest stage reached by these embryos). However, the mesogastrium showed defective development and, as a consequence, the dorsal pancreatic bud was mislocated into the dorsal mesentery. Unexpectedly, in six of eight WT1 systemic mutants, we observed an intestinal malrotation, with the duodenum placed at the left side, along the greater curvature of the stomach. This phenotype does not appear in the conditional WT1 deletion performed between E9.5 and E12.5. Thus, early WT1 expression (before E9.5) seems to be required for the development of the mesogastrium and the correct rotation of the intestine. A similar phenotype of intestinal malrotation has been described in mice with loss of function of both, the homeodomain transcription factors BARX1 and PITX2 (Shiratori et al., 2006; Jayewickreme and Shivdasani, 2015). Importantly, BARX1 loss of function leads to decreased expression of WT1 (Kim et al., 2007). Thus, WT1 is involved in L-R patterning of the intestine, probably through its role in the development of the mesogastrium and mesenteries. In fact, WT1 function is critical for the organization of other coelomic components, such as the pleuropericardial membranes (Norden et al., 2010) and the diaphragm (Carmona et al., 2016).

    We could not study further pancreatic development in older systemic WT1 mutant embryos due to the mortality caused by defects in other viscera (basically the heart). Thus, we cannot know if the difference between dorsal and ventral buds can expand in later stages, or if the formation of the endocrine pancreas might be compromised by the lack of WT1.

    In summary, the pancreas becomes a new addition to the list of organs whose morphogenesis is regulated by WT1, although in this case the impact of the WT1 loss of function is very low as compared with other viscera as the heart, kidney, adrenals, gonads or spleen, i.e., the mesodermal organs. The situation is similar to that described in other endodermal organ, the liver, where the lack of WT1 only results in a reduced size and an abnormal lobing (IJpenberg et al., 2007). The pancreatic stroma, as described for many other viscera, shows also a mixed origin, from the splanchnopleural mesoderm and from WT1-expressing mesothelial-derived cells. Finally, we have shown that an organ-specific cell type, the PSC, can also originate from embryonic mesothelium, as occurs in the testicle (Sertoli cells), ovary (granulosa cells), or intestine (Cajal interstitial cells) (for a review, see Ariza et al., 2016).

    Experimental Procedures

    The animals used in our research program were handled in compliance with the institutional and European Union guidelines for animal care and welfare. The procedures used in this study were approved by the Committee on the Ethics of Animal Experiments of the University of Malaga (procedure code 2015-0028).

    The Tg(Wt1-cre)#Jbeb (Wt1Cre) mouse line has been used in previous studies to trace or delete specific genes in WT1-expressing cells (Del Monte et al., 2011; Wessels et al., 2012; Carmona et al., 2013; Casanova et al., 2013; Cano et al., 2013, 2016). For lineage tracing studies, homozygote Wt1Cre+/+ were crossed with Rosa26EYFP (B6.129X1-Gt(ROSA)26Sortm1(EYFP)Cos/J) mice to generate permanent reporter expression in Wt1-expressing cells.

    The Wt1tm2(Cre/ERT2)Wtp/J allows for inducible Cre-recombinase expression in WT1-expressing cells after tamoxifen treatment (Chau et al., 2011). We have used this line crossed with Rosa26EYFP for inducible reporter expression in the WT1 lineage and also to delete expression of Wt1 in this lineage when the driver was crossed with a Wt1LoxP mice (Martinez-Estrada et al., 2010). For inducible reporter expression mice were intraperitoneally injected at different stages with tamoxifen (Sigma, T5648) dissolved in corn oil (10 mg/ml) at a dose of 0.1 mg/g body weight together with 0.05 mg/g body weight of progesterone to reduce abortion risk. In one experiment, a single dose was injected at the stage E9.5, and the embryos were fixed at E14.5. In a second experiment, three doses of tamoxifen were injected at E9.5, E10.5, and E11.5, and the embryos were fixed at E15.5. For conditional deletion of WT1, tamoxifen was first injected at stages E9.5, E10.5, E11.5 (three doses) and the embryos fixed at E15.5 and E16.5. In a second experiment, pregnant females were injected at E9.5, E10.5, E11.5, and E12.5 (four doses) and the embryos were fixed at E16.5. We always used as controls littermates either Cre- or lacking of floxed Wt1 alleles. No differences were observed between both types of controls.

    The Wt1GFP/+ knockin line in which the exon 1 of one Wt1 allele has been replaced by the GFP sequence (Hosen et al., 2007), was also used as an independent reporter for active Wt1 transcription and also as a model of WT1 loss of function in homozygosity (Wt1GFP/GFP).

    Embryos were staged from the time point of vaginal plug observation, which was designated as the stage E0.5. Whole embryos and the viscera of neonates were excised, washed in phosphate buffered saline (PBS), and fixed in 4% fresh paraformaldehyde solution in PBS for 2–8 hr. Then, the embryos were paraffin-embedded or washed in PBS, cryoprotected in sucrose solutions, embedded in OCT and frozen in liquid N2-cooled isopentane. Ten-micrometer cryosections were stored at -20 ºC until use.

    Immunofluorescence was performed using routine protocols. Deparaffinized sections or cryosections were rehydrated in Tris-PBS (TPBS) and blocked for nonspecific binding with SBT (16% sheep serum, 1% bovine albumin, 0.1% Triton X-100 in TPBS). When biotinylated secondary antibodies were used, endogenous biotin was blocked with the Avidin-Biotin blocking kit from Vector. Single immunofluorescence was performed incubating the sections with the primary antibody overnight at 4 ºC, washing in TPBS and incubating with the corresponding fluorochrome-conjugated secondary antibody. Double immunofluorescence was performed by mixing both primary antibodies (rabbit polyclonal and mouse or rat monoclonal), and incubating overnight at 4 ºC. We then used a Cy5-conjugated and a biotin-conjugated secondary antibody, followed by 45-min incubation with tetramethylrhodamine isothiocyanate (TRITC)-conjugated streptavidin. Nuclei were counterstained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; Sigma D-4592). Details of the antibodies used in this study are provided in Table 1.

    Table 1. Antibodies Used in This Study
    Antibody Supplier Clone or ref. Dilution
    Monoclonal rat anti-mouse CD31 (PECAM) Pharmingen Ref. 550274 1/20
    Monoclonal mouse anti alpha smooth muscle actin Sigma Clone 1A4 Ref. A2547 1/100
    Rabbit polyclonal anti-pan cytokeratin Dako Ref. Z0622 1/100
    Rabbit polyclonal anti-laminin Sigma Ref. L9393 1/200
    Rabbit polyclonal anti-alpha NG2 Abcam Ref. ab 5320 1/50
    Rabbit polyclonal anti-ALDH1A2 Abcam Ref. ab75674 1/200
    Rabbit polyclonal anti-phosphohistone H3 Millipore Ref. 06-570 1/100
    Chicken polyclonal anti-GFP Abcam Ref. ab 13970 1/200
    Mouse monoclonal anti-E-cadherin BD Ref 610181 1/200
    Mouse monoclonal anti-Wt1 Millipore MAB4234 1/100
    Chicken polyclonal anti-GFAP Millipore Ref. ab5541 1/200
    Mouse monoclonal anti-desmin Sigma Ref D1033 1/75
    Hamster anti-mucin Thermo Sci. HM-1630 1/300
    Mouse anti-amylase Santa Cruz Sc-46657 1/200
    Mouse anti-insulin Sigma I2018 1/500
    Rat anti-endomucin Sta. Cruz Sc-65495 1/500

    For flow cytometry analysis, pancreas from Wt1Cre/Rosa26REYFP embryos were excised, dissociated for 15 min at 37 ºC in 0.1% collagenase solution in PBS and homogenized by repeated pipetting. Cell suspension was washed in PBS plus 2% fetal bovine serum and 10 mM HEPES. Then, cells were incubated on ice at the dark with Cy5-conjugated rat Anti-mouse CD31 (Pecam-1). After washing, the cells were analyzed in a MoFlo cell sorter.

    For time-lapse video recording, E12.5 Wt1Cre/ROSA26REYFP mouse embryos were dissected in sterile PBS. Caudal stomach and proximal duodenum were isolated and cultured as previously described (Puri and Hebrok, 2007). Basically, organ explants were transferred to coverglass bottomed dishes (Ibidi, Martinsried, Germany) coated with 20 μl Matrigel™ and placed in a 37 °C incubator for 20 min to solidify the Matrigel™. The rudiments were then covered with medium (DME H-16/F-12 1:1 supplemented with 10% fetal bovine serum, antibiotics, and insulin-transferrin-selenium). After 48 h of culture, images were captured in a Leica SP5 confocal microscope every 5 min for 6 hr. During the capture, the culture chamber was maintained at 37 °C in a 5% CO2 humidified atmosphere.

    Analysis of the differences between dorsal and ventral pancreatic buds was performed on images taken from paraffin sections of four E16.5 Wt1CreERT2;Wt1flox mutants and four control littermates (all injected with four doses of tamoxifen between E9.5 and E12.5), using between four and six images for each embryo. In total, 39 sections were analyzed. The contour of the dorsal and ventral pancreatic buds were drawn, and the acini were manually recorded for each section. Number of acini and surface of the pancreatic contours were calculated using ImageJ software. The mean values obtained for each embryo were compared using Student's t-test, because the KS test found the data consistent with a normal distribution.

    For the proliferation study, the acinar surface was measured as described above and the number of PH3+ acinar cells counted (N = 6 systemic mutants and 8 controls, 15 and 20 sections analyzed, respectively). Because one of the data sets was not consistent with a normal distribution, Mann Whitney U-test was used to compare the values obtained.

    Acknowledgments

    We thank Dr. John Burch (National Institutes of Health) for the gift of the Wt1-Cre mice, and David Navas (SCAI, University of Málaga) for technical help with confocal microscopy and flow cytometry.

      Author contributions

      R. Carmona, L. Ariza, and R. Muñoz-Chápuli designed research, analyzed data, and wrote the paper. R. Carmona, L. Ariza, A. Cañete, and A. Rojas performed research.