Expression patterns of Wnts, Frizzleds, sFRPs, and misexpression in transgenic mice suggesting a role for Wnts in pancreas and foregut pattern formation



It is well established that gut and pancreas development depend on epithelial-mesenchymal interactions. We show here that several Wnt, Frizzled, and secreted frizzled-related protein (sFRP) encoding mRNAs are present during mouse pancreatic morphogenesis. Wnt5a and 7b mRNA is broadly expressed in foregut mesenchyme starting around embryonic day 10 in mice. Other members expressed are Wnt2b, Wnt5b, and Wnt11. In addition, genes for the Wnt receptors, Frizzled2, 3, 4, 5, 6, 7, 8, and 9 are expressed. To understand potential Wnt functions in pancreas and foregut development in vivo, we analyzed transgenic F0 mouse fetuses expressing Wnt1 and 5a cDNAs under control of the PDX-1 gene promoter. In PDX-Wnt1 fetuses, the foregut region normally comprising the proximal duodenum instead resembles a posterior extension of the stomach, often associated with complete pancreatic and splenic agenesis. Furthermore, the boundary between expression domains of gastric and duodenal markers is shifted in a posterior direction. In PDX-Wnt5a fetuses, several structures derived from the proximal foregut are reduced in size, including the pancreas, spleen, and stomach, without any apparent shift in the stomach to duodenum transition. In these fetuses, overall pancreatic morphology is changed and the pancreatic epithelium is dense and compact, consistent with Wnt5A effects on cell movements and/or attachment. Taken together, these results suggest that Wnt genes participate in epithelial-mesenchymal signaling and may specify region identity in the anterior foregut. © 2002 Wiley-Liss, Inc.


The mammalian pancreas is a complex organ, which secretes hormones and enzymes that are responsible for the regulation of blood glucose levels and food digestion, respectively. This organ develops from the gut tube, near the stomach as two independent evaginations (Pictet and Rutter, 1972). In the lung, which also develops from the foregut, the mesenchyme is known to produce factors such as hedgehogs, fibroblast growth factors (FGFs), bone morphogenic proteins (BMPs), and Wnts, which act on the developing epithelia to influence cell division, cellular fate decisions, and differentiation (Warburton et al., 1999; Hogan, 1999). Recent data have suggested that many of these factors are also acting in the pancreas (Wells and Melton, 1999, Kim and Hebrok, 2001).

FGF-1, 7, and 10 have been shown to be expressed in the developing pancreas and to stimulate the growth of the exocrine tissue in mesenchyme-free explant cultures of pancreas (Miralles et al., 1999). Receptors for some members of the transforming growth factor (TGF) family of ligands are expressed during pancreatic development (Crisera et al., 1999), and addition of TGF-β1 to explant cultures effects the development of both endocrine and exocrine tissue (Sanvito et al., 1994). Inactivation of the activin receptors IIA and IIB disrupts the development of the posterior foregut organs and results in hypoplastic pancreatic islets (Kim et al., 2000). Sonic hedgehog (Shh) is another factor that is expressed in the gut endoderm during development (Echelard et al., 1993; Wells and Melton, 2000; Edlund, 1999), and has been shown recently to convert the pancreatic mesenchyme to gut-like mesoderm, suggesting that Shh affects the fate of the adjacent mesoderm at different regions of the gut tube (Apelqvist et al., 1997). In addition, inhibition of Shh outside the pancreatic region of the gut results in the ectopic expression of pancreas-specific genes (Kim and Melton, 1998). All this evidence taken together suggests that multiple signaling molecules from the mesenchyme regulate the formation and fates of the pancreas cell types.

Wnt gene products constitute a large family of secreted factors, which play important roles as signaling molecules in development (Nusse and Varmus, 1992; Parr and McMahon, 1994; Gavin et al., 1995). Gene knockouts in the mouse have suggested that the Wnts play diverse roles in central nervous system pattering, control of asymmetric cell division, tissue polarity, and segmentation (McMahon and Bradley, 1990; Stark et al., 1994; Parr and McMahon, 1995; Yamaguchi et al., 1999; Yamaguchi, 2001). Frizzled (Frz) proteins have been identified as the receptors for Wnts (Wang et al., 1996; Wodarz and Nusse, 1998), possibly working in complex with the low density lipoprotein-related receptors (LRP) (Pinson et al., 2000; Tamai et al., 2000). There are at least eight unique Frizzled genes in the mouse (Wang et al., 1996, 1999), but little is known about the specificity of Wnt–Frz interactions. Additionally, in the mouse, four secreted Frizzled-related proteins (sFRPs) have been cloned (Rattner et al., 1997), some of which have been shown to bind Wnts and antagonize Wnt-mediated signaling (Wang et al., 1997a,b; Rattner et al., 1997). Because multiple Wnts and Frizzleds appear to be expressed in similar patterns, it has been suggested that the sFRPs may locally regulate which Wnt may interact with a specific receptor to create a local morphogen gradient (Wodarz and Nusse, 1998). Another class of Wnt signal modifiers are the Dickkopf (DKK) genes (Zorn, 2001). DKK1 has been shown to bind the extracellular domain of LRP6 and inhibit Wnt-Frz–induced signaling, whereas DKK2 and 3 do not inhibit Wnt signaling (Krupnik et al., 1999; Wu et al., 2000; Mao et al., 2001).

Here, we describe the expression levels and spatial arrangement of several Wnts, Wnt receptors (Frizzled), and Wnt antagonists (sFRP) during pancreatic development. In addition, we show how the overexpression of Wnt1 and Wnt5a in the developing pancreatic epithelium in transgenic mice alters the cellular fate and organization of the pancreas, suggesting that these molecules are important in the regulation of the cell fate decisions in the pancreas. The PDX-Wnt1 and PDX-Wnt5a phenotypes are the first known pieces of evidence demonstrating that Wnt genes function in regulating mammalian foregut development.


Kinetics of Wnt Gene Expression Throughout Pancreas Development

Gene expression profiles were examined by using cDNA prepared from RNA from mouse pancreata from embryonic day (E) 12–17 and postnatal (P) days 0 and 7. Wnt1 was expressed at very low levels and was not detectable before E15 (data not shown). Wnt2b levels were highest at E14 and decreased until E16, after which expression increased, being the highest in the P7 pancreas (Fig. 1A). Wnt5a has a very similar expression profile to Wnt2b but with higher levels detected during early development (Fig. 1A). Wnt5b is expressed at low levels but showing the same pattern of expression Wnt2b. Wnt7b expression levels are highest at E12 and decline to a low but detectable level throughout development (Fig. 1A). Wnt11 is detected at low levels throughout development (Fig. 1A). Wnt15 was barely detectable at any time points tested, whereas Wnt16 was expressed at low levels during development and at P0 and P7 (data not shown). Of interest, the expression levels of Wnts decrease dramatically between E15 and E17 as the amount of mesenchyme in the pancreas decreases and the epithelium enlarges.

Figure 1.

Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of the Wnt, Frizzled (Frz), and secreted Frizzled-related protein (sFRP) gene families in embryonic mouse pancreas. Expression was analyzed from RNA prepared from staged embryonic mouse pancreata from embryonic day (E) 12 to E17 and postnatal days 0 and 7. The ratio between the analyzed genes and the internal standard glucose-6-phosphate dehydrogenase are plotted. RT-PCR was performed as described in the Experimental Procedures section.

Kinetics of Frizzled Gene Expression Throughout Pancreas Development

Frizzled2, 3, 4, 5, 6, 7, 8, and 9 were detected by reverse transcriptase-polymerase chain reaction (RT-PCR) and the kinetics studied on cDNA prepared from RNA from mouse pancreata from E12 to E17 and P0 and P7. Frz2 is the most highly expressed Frz gene with high level expression at E12–E13 declining thereafter (Fig. 1B). Of interest, the gene cloned as mouse Frz10 is the same gene as mouse Frz2 (Malik and Shivdasani, 2000). Frz3, 4, 6, and 8 show little dynamic expression and are expressed throughout development and at days P0 and P7 (Fig. 1B,C). Frz5 is expressed at low levels at E12 and E13 but increased at the end of gestation (E17). Frz9 expression was very low throughout embryonic development but increases at the end of gestation and after birth. Wnt coreceptors LRP 5 and 6 were expressed at high levels from E12 to E18 (data not shown).

Wnt Inhibitors sFRP Are Expressed During Pancreas Development

It has been shown previously that sFRP 1 is expressed in the human pancreas (Hu et al., 1998). To determine which signaling inhibitors were expressed in the mouse embryonic pancreas, PCR analysis was performed for sFRP1-4, Wnt-inhibiting factor (WIF), and the DKK genes as described above (Fig. 1D). sFRP-1 expression was highest at E12–E14 and declined thereafter (Fig. 1D). sFRP-2 was expressed at E12 but showed a major peak in expression at E14, after which it declined to detectable but low levels. sFRP-3 was expressed at low but constant levels throughout development. sFRP-4 expression was very low throughout development, showing a peak of expression at P7. WIF was expressed at very low but detectable levels throughout development (data not shown). DKK-2 and DKK-3 were expressed between E12 and E18, with highest expression at E15.5 (data not shown). DKK-1 was detected at low levels only at E12 and E15 (data not shown).

Localization of Wnts and Frizzleds in the Developing Pancreas

By using in situ hybridization, a more detailed picture of the expression in the pancreas could be observed. Whole E11 embryos were sectioned and in situ hybridizations performed on slides. We observed that the Wnt5a gene was strongly expressed in the surrounding mesenchyme and epithelium of the pancreas, whereas Wnt7b was exclusively in the mesenchyme (Fig. 2). A similar pattern of expression was observed at E10 for both genes (data not shown). Frizzled 2 and 7 were found to be expressed in both the surrounding mesenchyme and pancreatic epithelium, whereas Frz3 was only expressed in the surrounding mesenchyme (Fig. 2).

Figure 2.

RNA in situ hybridization analysis of Wnt, Frizzled (Frz), and secreted Frizzled-related protein (sFRP) gene expression in embryonic day (E) 11 mouse embryonic pancreas. In situ hybridizations were performed on 10-μm sections prepared from E11 embryos as described in the Experimental Procedures section. Wnt5a (A) is expressed strongly in the mesenchyme and epithelium in the pancreas, whereas Wnt7b (B) is only found in the surrounding mesenchyme. Frz2 (C) was expressed in both pancreatic mesenchyme and epithelium, whereas Frz3 (D) was mesenchymal. E: No probe control. PE, pancreatic epithelium; PM, pancreatic mesenchyme; D, duodenal epithelium. Original magnification: A–D, 200×; E, 100×.

By using whole dissected gut regions, which contain pancreas, spleen, stomach, and the most proximal portion of the duodenum, expression was analyzed at E15.5 and E17.5 in the mouse by using whole-mount in situ hybridizations. At E15, Wnt1 was expressed in the mesenchymal cells surrounding the developing endocrine clusters stained for insulin and glucagon (Fig. 3A–D). At E17.5, Wnt5a was expressed in the mesenchyme and areas of the insulin and glucagon expressing cells (Fig. 3G–H). At E15.5, Wnt5b was expressed mainly in the mesenchyme area (Fig. 3I). By E17.5, the signal was detected around the insulin and glucagon cells as well as colocalized with some insulin and glucagon (Fig. 3J–L). Wnt11 expression was mainly in the mesenchyme (Fig. 3M) but showed a weak colocalization with some of the insulin-immunoreactive cells but not the glucagon cells at E17.5 (Fig. 3N–P).

Figure 3.

RNA in situ hybridization analysis of Wnt gene expression in mouse embryonic pancreas. Depicted are sections through whole-mount (pancreas-gut) preparations at embryonic day (E) 15.5 or E17.5 that were hybridized with antisense RNA probes indicated at the top of the column. A: Insulin and glucagon double stainings were performed on the same sections. Wnt1 is expressed in the mesenchyme at E15.5. Also depicted is Wnt1 expression in the area surrounding (B) a developing endocrine cell cluster that is immunostained for insulin (C) and glucagon (D). Wnt5a is expressed mainly in the mesenchyme (E,F) but is also colocalized with some insulin cells (G) and possibly glucagon (H) as well at E17.5. At E15.5, Wnt5b is strongly expressed in the mesenchyme (I). At E17.5 (J), it is more strongly expressed in the insulin (K) than the glucagon-immunoreactive cells (L). Wnt11 is mainly expressed in the mesenchyme (M) but also weakly in the areas of endocrine cells (N–P). Arrows denote areas of colocalization. m, mesenchyme; e, epithelium.

Expression patterns for some of the Frz genes were examined at E17.5. Interestingly, Frz2, Frz3, and Frz7 were all found to colocalize with the endocrine cells in the pancreas at this time point (Fig. 4A–I). sFRP-1 was strongly expressed in the mesenchyme surrounding the pancreas epithelium at E11 (data not shown). In situ hybridizations for sFRP2 and sFRP3, showed that these inhibitors of Wnt binding are expressed with the endocrine cells at E17.5 (Fig. 4J–M).

Figure 4.

RNA in situ hybridization analysis of Frizzled (Frz) and secreted Frizzled-related protein (sFRP) gene expression in mouse embryonic pancreas. Depicted are sections through whole-mount (pancreas-gut) preparations at embryonic day (E) 17.5 that were hybridized with antisense RNA probes indicated at the left of the column. Insulin and glucagon double immunostainings were performed on the same sections. Frz2 (A–C), Frz3 (D–F), and Frz7 (G–I) all colocalize with the endocrine cell clusters at E17.5. sFRP2 (J,K) and sFRP3 (L,M) expression colocalizes with the endocrine cell clusters. Arrows denote areas of colocalization.

Overexpression of Wnt1 and Wnt5a to the Developing Pancreatic Epithelium Disrupts Normal Development

Based on similarities between vertebrate and fly gut development (Bienz, 1994), coupled with information that was known about the biology of Wnt growth factors (Nusse and Varmus, 1992) and results from our in situ hybridization and RT-PCR experiments, we proposed that Wnts would be involved in gut and pancreas development. Wnt gene products have been implicated as mediators of mesenchymal–epithelial inductive interactions in several systems (Nusse and Varmus, 1992) and so, are likely to be important mediators in the development of the pancreas. The PDX-1 gene promoter was chosen to direct the expression of the transgene to the developing foregut epithelium, which includes the pancreas, pyloric stomach, and the most proximal portion of the duodenum (Offield et al., 1996; Stoffers et al., 1999). Wnt1 was an obvious choice to test as it will stimulate the canonical β-catenin pathway, whereas Wnt5a, has been shown to possibly activate alternative Wnt pathways and is the major Wnt expressed in the pancreas. Several Wnt family members were tested (Wnt1, Wnt2, Wnt4, Wnt5a, Wnt6, and Wnt7a) by using transgenic misexpression, and we have shown that Wnt1 (n = 5) and Wnt5a (n = 4) modify pancreas development in interesting and important ways. The other Wnts did not produce reproducible phenotypes (data not shown).

Misexpression of Wnt1 under the control of the PDX-1 promoter causes a repatterning of the stomach-duodenum region in the foregut, with the diameter of the proximal duodenum being greatly enlarged (Fig. 5A–D). There is no obvious pyloric sphincter, and there is a posterior shift in the transition from gastric to duodenal epithelium as indicated by the transition from alkaline phosphatase (AP)-negative to AP-positive cells (Fig. 5E–H). In some cases, it appears that a second distended stomach-like structure develops immediately posterior to the true stomach (Fig. 5B,D). When this occurs, there is an atypical constriction where the pyloric sphincter should be and the epithelium of the abnormal posterior distended structure has features that are more characteristic of stomach than duodenum (Fig. 5L). In addition, immunostaining for the parietal cell marker GPC shows that there is a clear shift in the boundary of the stomach in a more posterior direction (Fig. 5I–N). In PDX-Wnt1 fetuses, pancreas and spleen development is arrested, and in some cases, they are entirely absent (Fig. 5A–F). When the pancreas and spleen are present, they are reduced in size and may appear diffuse and abnormally shaped. Sometimes, pancreatic tissue does form and small clusters of pancreatic exocrine cells were observed on the luminal side of the smooth muscle layer of the proximal duodenum (data not shown). When abundant pancreatic tissue was observed, it had normal amylase immunoreactivity (data not shown). Of interest, we observed that several of the insulin-immunoreactive cells lacked PDX-1 immunoreactivity (data not shown). β-Catenin immunoreactivity in the pancreas was greatly increased in the transgenic mice, indicating the likelihood that the transgene was expressed at high levels (data not shown).

Figure 5.

Effects of misexpression of Wnt1 on pancreatic and proximal gut morphology. A: Macroscopic appearance of the gut from embryonic day (E) 19 wild type (WT) and transgenic mouse embryo. White arrows denote the position of the pancreas. In the transgenic (TG) mice, there is nearly a complete lack of pancreatic tissue. Also note the greatly enlarged stomach and lack of normal stomach to duodenal transition. B,C: Different views of the same embryo depicted in B. White arrow in C denotes the very small spleen. D: Macroscopic appearance of the gut from a different E19 transgenic mouse embryo. White arrow denotes a piece of liver attached to the sample. E,F: Different views of a different TG embryo. Note the absence of the spleen and pancreas. G,H: Hematoxylin-stained WT (E) and TG (F) section from E9 embryo. White arrows denote the position of the pancreas, which is completely lacking in the TG animal (F,H). The lack of a pyloric sphincter and normal transition from stomach to duodenum is also evident. G,H: Insulin (blue) and glucagon (brown) staining of WT (G) and TG (H) sections. WT pancreas stains for both hormones, whereas TG mice lack immunoreactivity for both hormones. The dark blue staining in the lining of the proximal duodenum is due to endogenous alkaline phosphatase (AP) activity, which serves as a reliable indicator of the transition from gastric to duodenal epithelium. The image in H clearly shows the posterior shift in the transition from AP-negative to AP-positive epithelium. I: Low-magnification image of section from E19 WT embryo cut through the stomach and duodenum and stained with GPC (FA125) antiserum, a parietal cell marker. Boxes denote the specific areas for the following high magnification images. K: Specific staining for the GPC antigen in the stomach of a WT embryo. M: Shows the lack of GPC immunoreactivity in the duodenum of the WT animal. J: Low-magnification image of section from E19 TG embryo cut through the stomach and duodenum and stained with GPC (FA125) antiserum, a parietal cell marker. Boxes denote the specific areas for the following high magnification images. L: Specific staining for the GPC antigen in the stomach of a TG embryo. N: GPC immunoreactivity in the duodenum of the TG animal provides further evidence for a posterior shift in the stomach epithelium in the TG animal.

Misexpression of Wnt5a also reprograms gut and pancreas development but in a manner different from Wnt1. Misexpression of Wnt5a leads to a dramatic reduction in size of the pancreas, spleen, and stomach (Fig. 6A–E). Unlike PDX-Wnt1 fetuses, the stomach and spleen appear to have relatively normal overall architecture, shape, and histologic appearance. The pancreas is greatly reduced in size and does not have its normal gross tissue architecture or histologic appearance. Pancreatic tissue in these fetuses exists as multiple small, dense masses attached to the outer surface of the proximal duodenum and stomach (Fig. 6C). Sometimes they appear as a series of masses along one face of the stomach (Fig. 6C). Histologically, the masses of pancreatic tissue are made up of densely packed exocrine structures and compact atypical endocrine-like cells (Fig. 6E,F), only a portion of which express insulin or glucagon (Fig. 6H,I). Insulin-immunoreactive cells, which lack the transcription factor Isl-1, were observed in some of the endocrine clusters (Fig. 6M). Some small, almost normal islet structures could sometimes be observed which contain a significant number of insulin- or glucagon-positive cells (data not shown). Amylase immunoreactivity was observed, despite the poorly formed exocrine acini (Fig. 6K). Pancreatic ducts appear never to be formed. In PDX-Wnt5a fetuses, the duodenum appears normal and the shift from AP-negative gastric epithelium to AP-positive intestinal epithelium occurs as it should at the pyloric sphincter (Fig. 6G,H). This finding is an important distinction between the PDX-Wnt1 and PDX-Wnt5a phenotypes. In several fetuses, pancreatic tissue grew into the body of the spleen, dividing the spleen into two separate structures (Fig. 6K).

Figure 6.

Effects of misexpression of Wnt5a on pancreatic and proximal gut morphology. A–C: Macroscopic appearance of the gut from embryonic day (E) 19 wild-type (WT) and transgenic mouse embryo. White arrows denote the position of the pancreas. In the transgenic (TG) mice, there is nearly a complete lack of pancreatic tissue. Note the extremely small size of the stomach and spleen. B: High-magnification view of a portion of the pancreas in which the tissue has formed a tight ball-like structure. C: View from the other side, highlighting the pancreatic tissue, which appears as small dense masses attached to the foregut wall. Also illustrated is the smaller stomach, spleen, and pancreas. D: Hematoxylin and eosin-stained representative gut section from a single E19 PDX-1-Wnt5a. Arrow denotes an area of endocrine cells in the densely packed epithelium. E: Hematoxylin and eosin–stained representative gut section from a single E19 PDX-1-Wnt5a. Arrows denote location of the pancreatic epithelium. F: High-magnification view of the densely packed pancreatic epithelium depicted in E. G: Low-magnification of the gut region of an E19 wild-type embryo immunostained for insulin (original magnification, 40×). H: Low magnification (original magnification, 40×) of the gut region of a transgenic embryo immunostained for insulin (blue) and glucagon (brown). Arrows denote the pancreatic areas. I: Higher magnification (original magnification, 200×) of insulin and glucagon immunostaining. Many cells are present, but they are not clustered into islet structures. J,K: Amylase immunoreactivity in wild-type (J) and Wnt5a transgenic mice (K). The exocrine pancreas fails to form acini but does express amylase (K). L: Expression of ISL-1 in WT mice (original magnification, 400×). M: Lack of transcription factor Isl-1 (red nuclei) in most of the insulin-immunoreactive (green) β-cells (original magnification, 200×). sp, spleen; p, pancreas; st, stomach.


Multiple Members of the Wnt, Frz, and sFRP Family Are Expressed During Pancreas Development

The expression of five members of the Wnt family in the pancreas as well as eight members of the Frizzled receptor family in developing pancreas suggests a major role for Wnts in many aspects of pancreas development. Both Wnt and Frizzled gene expression is dynamic during development, with most genes expressed at their highest levels before E15 and show a decrease around E15–E16, when endocrine differentiation is occurring and the mesenchyme component is shrinking. Of the Wnt signaling components examined, Dishevelled1, 2, 3, and β- and δ-catenin were all found to be expressed as well as GSK3-α and -β (data not shown). Interestingly, the downstream signaling components also show this dramatic decrease in expression around E15–E16 and a later increase suggesting some importance of down-regulating Wnt signaling around the time of endocrine differentiation.

Wnt and Frizzled gene expression was examined by in situ hybridization. At E11, we observed that Wnt5a was broadly expressed across the epithelium and mesenchyme with a more intense signal in the mesenchyme. Wnt7b was clearly only expressed in the mesenchyme. It was reported recently that Wnt2b is expressed in the pancreatic mesenchyme at E13.5 (Lin et al., 2001) and localizes to the islets after birth (Heller and Jensen, unpublished data). Frz receptors 2, 3, and 7 were examined by in situ hybridization and localized with the developing endocrine cells at E17.5. This finding is consistent with our findings that several of the insulin-producing cell lines express Frz genes (Heller and Jensen, unpublished data). Expression of Wnt5a in the mesenchyme and possibly the exocrine pancreas is consistent with Wnt expression in the developing mesenchyme of the lung (Lako et al., 1998), both which develop as invaginations from the endoderm and both undergo branching morphogenesis (Slack, 1995; Hogan, 1999; Warburton et al., 1999).

Repatterning of the Stomach-Duodenum Region in Wnt1 Transgenic Mice

Wnts are expressed in the mesenchyme and signal to the epithelium in several organ systems (Nusse and Varmus, 1992; Lako et al., 1998). The use of the PDX-1 promoter provided a Wnt signal from the epithelium directly from the earliest time points of pancreas bud formation. In addition, the PDX-1 promoter is expressed in the developing epithelia of the stomach and proximal duodenum (Offield et al., 1996; Stoffers et al., 1999). The effect of misexpression of Wnt1 in the developing epithelia resulted in a posterior shift in the gastric epithelia into the duodenum. This finding is consistent with a role of Wnts in midgut patterning in drosophila (Bienz, 1994).

Of interest, misexpression of Shh in transgenic mice using the PDX-1 promoter caused the pancreatic mesoderm to develop into smooth muscle and interstitial cells characteristic of the intestine, rather than into pancreatic mesenchyme and spleen (Apelqvist et al., 1997). Overexpression of Wnt7b represses Shh gene expression in the oral ectoderm, thus suggesting that Wnt7b repression acts to maintain the boundary between oral and dental ectoderm cells (Sarkar et al., 2000). These studies together could suggest that Wnt regulation of Shh expression in the gut may alter the patterning and partially explain the Wnt1 transgenic phenotype we observe. Further evidence for a role of Wnts in patterning comes from studies showing that inhibition of BMP and Wnt signaling is essential for patterning of the anterior neural plate (Niehrs, 1999). A role for Wnts in the gut has long been suggested by the mom gene mutations identified in C. elegans, which are mutations in several of the components of the Wnt pathway that directly effect the formation of the single cell that gives rise to the entire gut endoderm (Han, 1997).

Misexpression of Wnt1 and 5a Alters Spleen Development

The observation that spleen development is arrested in PDX-Wnt1 transgenic fetuses suggests either that Wnt genes play an essential role in development and differentiation of the splanchnic mesoderm present around the early foregut or that Wnt-induced alterations of the pancreatic and gut epithelium lead to secondary alterations of the surrounding mesenchyme. In addition to the PDX-Wnt1 phenotype, the involvement of the spleen in the PDX-Wnt5a phenotype suggests a role for Wnt proteins or their absence in patterning the splanchnic mesoderm surrounding the foregut. For PDX-Wnt5a, the effect may be due to alterations of cell movements or adhesion (e.g., abnormal growth of pancreatic tissue into the spleen, multiple discrete pancreatic masses distributed along one face of the stomach, and more densely packed pancreatic epithelium and mesenchyme), whereas for PDX-Wnt1, the effect seems to be a more general reprogramming of segmental identity, not unlike a posterior homeotic shift, in the proximal foregut. This reasoning is consistent with known activities of the Wnt1 class, which promotes axis duplication, whereas Wnts of the 5a class alter morphogenic programs in Xenopus development assays (Torres et al., 1996), suggesting different intracellular pathways being activated by these two Wnts (Kühl et al., 2000).

Misexpression of WNT5a With the PDX-1 Promoter Reduces the Size of All Organs in the Foregut

The size of the pancreas, stomach, and duodenum were all greatly reduced in the Wnt5a transgenic mice. Spleen and stomach appeared histologically normal, but the pancreas was greatly compacted. Amylase was expressed but the exocrine acinar structures were abnormal in appearance. We speculate that pancreatic bud formation was never completed correctly as the pancreas tissue appears as dense masses along the stomach and duodenum. How overexpression of Wnt5a in the PDX-1 expression domain limits the growth of the organs is not known. It suggests that excessive Wnt signaling in the epithelia limits the expansion of both the mesenchyme and epithelium and, thus, not allowing complete growth of the organs.

Misexpression of Wnts Alters Pancreatic Development

The misexpression of Wnt1 in the PDX-1 expression domain had powerful effects on the pancreas. In many fetuses, no pancreas was formed at all, suggesting a complete block of PDX-1 cell expansion and differentiation. In the animals in which pancreas was formed, we saw a reduction in endocrine cell number, lack of real islet formation, and the absence of the endocrine transcription factors Isl-1 and PDX-1 (data not shown) in some of the insulin-immunoreactive cells. Amylase immunoreactivity was detected in the transgenic mice. In less dramatically affected fetuses, a compacted pancreas structure that appeared to lack ductal structures and had poorly formed exocrine acini was observed. Although some endocrine development occurred, normal islet architecture was disrupted. The compact nature of the pancreas could be due to reported effects of Wnts on cytoskeletal restructuring (Shibamoto et al., 1998), cell migration (Torres et al., 1996), or the lack of pancreatic ducts. Of interest, it was observed that small clusters of pancreatic exocrine cells were found on the luminal side of the smooth muscle layer of the proximal duodenum (data not shown). This observation suggests either that intestinal epithelial cells can be converted into pancreatic epithelium or that a small number of pancreatic progenitors form but were incorporated into the intestine, perhaps as a result of inhibition of pancreatic bud formation and expansion. Examination of embryos at earlier time points could address this finding.

Our data suggest that Wnt expression from the epithelium of the pancreas by using the PDX-1 promoter has negative effects on endocrine and exocrine cell formation. The differences between the PDX-Wnt1 and PDX-Wnt5a phenotypes may be due to differences in binding of Wnt1 and Wnt5A to subtypes of Frizzled proteins, now known to function as Wnt receptors (Bhanot et al., 1996; Kühl et al., 2000). This mechanism would appear likely as it is known that Wnt1 activates the β-catenin pathway (Nusse, 2001), whereas Wnt5a does not, working by means of a protein kinase C–dependent pathway (Kühl et al., 2000). The complex patterns of expression of multiple Wnts and Frizzled genes during pancreas development suggest important functions for Wnts in the pancreas. Finally, the PDX-Wnt1 and PDX-Wnt5a phenotypes are the first known pieces of evidence demonstrating that Wnt genes function in regulating mammalian foregut development.


Multiplex PCR

The multiplex RT-PCR methods described previously in detail (Jensen et al., 1996), allows the co-amplification of several cDNA products from total RNA preparations in a single tube. Briefly, RNA was prepared from three independent staged mouse embryo pancreata by using RNAzol (Cinna Biotecx), according to the manufacturer's instructions. cDNA was synthesized from 2 μg of RNA, and samples were diluted to the same final concentration. PCR was performed for 18–28 cycles, depending on the primer sets used. Internal standards for every PCR reaction were included, and all data were normalized to the internal standard. Data (n = 3) were quantitated from the Phosphoimage by using the ImageQuant program. Data are plotted as the mean plus the standard deviation.

PCR Primers

The PCR primers were primarily designed to mouse sequences, but some rat genes were included as well and have been denoted in Table 1.

Table 1. Polymerase Chain Reaction Primersa
GeneUpstream primerDownstream primerProduct size (bp)Species
  • a

    sFRP, secreted Frizzled-related protein; WIF, Wnt-inhibiting factor; G6PDH, glucose-6-phosphate dehydrogenase.

Wnt-15′ ctg ggt ttc tactac gtt gcgtt ctg gtc gga tca gtc g199Mouse
Wnt-2b5′ tgg agg gca ctc tca gac ttc cgcc ttg tcc aag aca cag tag t188Rat/mouse
Wnt-45′ cgc gct aaa gga gaa ttt gaccca cag cac agt agc tcg ca240Mouse
Wnt-5A5′ ctt ccg caa ggt ggg cga tgcttg cac agg cgt ccc tgc gtg204Rat/mouse
Wnt-5B5′ gtg gct gct gac ctc aag accttc ttc atg cag aag tca gga g170Rat/mouse
Wnt-7b5′ cga gcc agc cgc ctg cgc catgc gtg cct acg ctg ccc gtg147Rat/mouse
Wnt-115′ gtg gct gct gac ctc aag accttc ttc atg cag aag tca gga g160Rat/mouse
Wnt-155′ gta agt gcc atg gtg tgt cgtg tca tag cgt agc ttc ag103Mouse
Wnt-165′ aca gca tcc aga tct cag acact aca tgg gtg ttg tag cc238Mouse
Frizzled-25′ tcc tgc caa gcc tag tca ctc gtatcg gag cga ggg cta gag180Rat/mouse
Frizzled-35′ tag caa tgg agc cct tcc accctc cat atc ttc agg cca cgg200Mouse
Frizzled-45′ gct tgt gct atg ttg gga acc cacaca ggt tgc agg aac cgt230Mouse
Frizzled-55′ gtc tgt gct gtg ctt cat cagt gac aca cac agg tag ca123Mouse
Frizzled-65′ ttt gtc ttt gtg caa ctc tggct tct cgt ctg cct tat ta131Mouse
Frizzled-65′ tga agg aga gaa gca atg gat ctga aca ggc aga gat gtg gag160Mouse
Frizzled-75′ ggc cat cga ggc caa ctc gcacgc aat cga tcc aca cta gac150Mouse
Frizzled-85′ ttt gtg ctg gcg cca ctg gtttag agc acg gtg aag agg ccc170Mouse
Frizzled-85′ cta ctc gca gta ctt cca ccgcg gat cat gag ttt ttc ta289Mouse
Frizzled-95′ aaa tct tca tgt ct tgg tgatg ttc tag agg tgt gtg gg256Mouse
Frizzled-105′ tgt ccg gtt gct aca cca tgg gctgcc agg aac cag gtg agg220Rat/mouse
sFRP15′ aga ctc tgc ttt tct gca aggca ctt tag cct caa aag aa149Mouse
sFRP25′ aat gac ttc gca ctg aaa atgtt gat gtc gtt cat ctc ct138Mouse
sFRP35′ att tgg tgt tct gta ccc tgcgt ttc ctc ata aaa tgc ttc268Mouse
sFRP45′ tca tga aga tgt ata acc aca ggcc act cat aac aca tga tta g453Mouse
sFRP45′ acg tgg tgg cta ata tgt tccac aac ata ggc aca gag tg149Mouse
WIF5′ atg ggg agt gtt aga gag gtaga gta aca gca agg gtg aa196Mouse
G6PDH5′ gac ctg cag agc tcc aat caa ccac gac cct cag tac caa agg g214Rat/mouse

In Situ Hybridization

Wnt-encoding plasmids were provided by A.P. and J. MacMahon, Harvard University, Cambridge, MA. The mouse Frizzled and sFRP plasmids were provided by J. Nathans, John Hopkins University, Baltimore, MD. Antisense RNA probes were synthesized from plasmids by using the digoxigenin (DIG) RNA synthesis kit from Boehringer Mannheim, following the manufacturer's instructions. Whole-mount in situ hybridization was performed essentially as previously described (Jensen et al., 2000). Briefly, fetuses or whole gut regions (included pancreas, stomach, and proximal duodenum) were fixed in fresh 4% paraformaldehyde overnight at 4°C. Fetuses were then rinsed in PBS and stored at −20°C in 100% methanol. Fetuses were rehydrated in Tris buffered saline with Tween (TBST), treated with proteinase K, fixed in paraformaldehyde, rinsed, and prehybridized for 3 hr and hybridized at 68°C overnight with antisense RNA probes in hybridization buffer. Fetuses were rinsed in hybridization fluid, followed by TBST and incubated with sheep anti-DIG or anti-fluorescein antiserum (1:5,000) conjugated to alkaline phosphatase overnight at 4°C. Fetuses were rinsed in TBST followed by NTMT (0.1 mol/L NaCl, 0.02 mol/L Tris [pH 9.5], 0.01 mol/L MgCl2, and 1% Tween-20) and incubated with the alkaline phosphatase substrate for 1–72 hr. The reaction was stopped by rinsing in TBST. Fetuses were embedded in OCT compound and processed for frozen sections.

In situ hybridization on frozen sections was performed as previously described (Schaeren-Wiemers and Gerfin-Moser, 1993; Gradwohl et al., 1996). Briefly, in situ hybridization was performed on 10-μm frozen sections. Slides were removed from −80°C and allowed to come to room temperature for 30 min. Prehybridization buffer was added for 15–30 min while probes where denatured at 75°C in hybridization buffer. Slides were hybridized overnight at 65°C. The unhybridized probe was removed with subsequent rinses in 1× sodium chloride-sodium citrate (SSC), 50% formamide, and 0.1% tween 20 at 65°C for 3× 30 min. Slides were rinsed two times in maleic acid buffer (MABT: 100 mM maleic acid, 150 mM NaCl, 0.1% Tween 20, pH 7.5) for 20 min each. Slides were blocked with 2% Boehringer Mannheim blocking reagent, 20% sheep serum in MABT for 1 hr at room temperature. Alkaline phophatase-conjugated Fab fragments (Boehringer Mannheim) diluted 1/2,500 was added to the slides overnight at room temperature. Slides were rinsed 5× 20 min MABT, followed by two rinses for 10 min each in NTMT. The reaction mixture was added (5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium [BCIP-NBT; Boehringer Mannheim] in NTMT). Reactions were allowed to proceed for 2 hr to 3 days at room temperature. The reaction was stopped by several rinses in Tris–ethylenediaminetetraacetic acid, and slides were mounted and cover-slipped.


Immunohistochemistry was performed as previously described (Blume et al., 1993). Briefly, sections were dewaxed in xylene and rehydrated through a descending ethanol series. Antigen retrieval was accomplished through microwave treatment (two times for 5 min at 600 W in 0.01 mol/L citrate buffer, pH 6.0) followed by three washes in PBS. Nonspecific binding was blocked with 10% donkey nonimmune serum. For double immunofluorescence, sections were incubated with primary antibodies overnight. Secondary antibodies (fluorescein isothiocyanate–, Cy-2–, and Texas Red–conjugated) were obtained from Jackson ImmunoResearch (West Grove, PA). Double alkaline phosphatase and horseradish peroxidase stainings were performed by using different substrates (diaminobenzidine for glucagon and BCIP-NBT for insulin). GPC (FA125) antiserum, a parietal cell marker antiserum was purchased from the Binding Site (Birmingham, UK). Guinea pig anti-insulin and mouse monoclonal anti-glucagon were obtained from Nordisk Gentofte A/S (Gentofte, Denmark). Amylase (1:1,000) was purchased from DAKO (Glostrup, Denmark). Rabbit anti–Isl-1 (1:200) antiserum was a gift from Dr. Helena Edlund (Umeå, Sweden). Anti–PDX-1 253 (1:1,000) antiserum was a gift from Dr. Joel F. Habener (Boston, MA). Double immunofluorescence on in situ hybridizations was performed as previously described (Jensen et al., 2000). Images were captured by using a Hammamtsu digital camera and Image Pro Plus 4.5 software interfaced with a PC computer.

Transgene Construction

Transgenes were assembled that contain approximately 4.5 kb of the murine PDX-1 gene 5′ flanking region, including the putative transcription start site (isolated by hybridization screening of a SV129 murine genomic library [Stratagene, Inc., La Jolla, CA] by using a 0.6 kb 5′ fragment of the rat PDX-1 cDNA [Miller et al., 1994] as a probe), splicing and polyadenylation regulatory cassettes from the rabbit beta globin gene (Sasaki and Hogan, 1994), and Wnt-encoding cDNAs (kindly provided by J. McMahon, Harvard University). This PDX-1 promoter fragment has been shown previously to faithfully reproduce developmental and tissue-specific expression in mice (Stoffers et al., 1999).

To construct PDX-Wnt transgenes, cDNAs encoding murine Wnt1 and Wnt5A were inserted into the pIT2 polylinker. For PDX-Wnt1, a 2.0-kb Xba1/blunt-ended fragment containing the Wnt1 cDNA was inserted between the NheI and SwaI sites of pIT2. For PDX-Wnt5A, a 2.5-kb HindIII/SpeI Wnt5A fragment was inserted into the HindIII and SpeI sites of pIT2. For PDX-Wnt5A, NheI and SalI sites were introduced by site-directed mutagenesis just 5′ to the start codon and 3′ to the stop codon, respectively, leaving the native Kozak sequence intact. The NheI/SalI fragment was then ligated into the NheI and SalI sites of pIT2. It was necessary to trim the Wnt5A cDNA down to the open reading frame to observe the clear phenotype shown for PDX-Wnt5A. An initial round of transgenic fetuses generated by using the complete Wnt5A cDNA (open reading frame plus approximately 500 bases of 5′ UT and 1,600 bases of 3′ UT) yielded several bona fide transgenics but none with any obvious phenotypes (pancreatic or otherwise). Transgene fragments were separated from the plasmid backbone (pBluescript II-KS(+); Stratagene, Inc.) by restriction digest and gel purification, followed by gel-based fluorometric estimation of DNA concentration by using a Fluorimager SI (Molecular Dynamics, Inc., Sunnyvale, CA).

Transgenic Misexpression

The transgenic misexpression assay involves producing and analyzing F0 transgenic mouse fetuses for pancreas and gut abnormalities. Transgene fragments were microinjected at transgenic mouse facilities at the Beth Israel Hospital, Boston (BITF), and University of Massachusetts, Worcester (UMass). On average, 12% of the fetuses produced at each facility are PCR-positive transgenics. A total of 60–90% of the PCR-positive fetuses exhibited phenotypes. For all experiments, fetuses derived from microinjected one-celled fetuses were allowed to develop for 12 or 19 days in utero, then were delivered by cesarean section, killed, eviscerated, and examined for gross and foregut-specific abnormalities. Tail biopsies were taken for extraction of genomic DNA, and viscera were gently removed, examined, and fixed overnight at 4°C in fresh 4% paraformaldehyde. Genomic DNA was prepared by proteinase K digestion followed by phenol-chloroform extraction and ethanol precipitation (Hogan et al., 1994). Genomic DNA was isolated from embryonic tail samples, then subjected to PCR genotyping by using primers that anneal to the mouse PDX-1 gene promoter (5′ primer: 5′-AACTGTCAAAGCGATCTG-3′) and the rabbit beta globin splice cassette (3′ primer: 5′-ACATGGTTAGCAGAGGGGCC-3′), generating a unique 500-bp product in transgenic samples. Viscera from transgenic fetuses were processed for embedding in paraffin (Shandon Hypercenter XP; Shandon Lipshaw, Inc., Pittsburgh, PA). Five-micron sections were cut for hematoxylin and eosin staining and immunohistochemistry.


The authors thank H. Sasaki, Vanderbilt University, Nashville TN, for rabbit beta globin splice and polyadenylation cassettes; A.P. and J. MacMahon, Harvard University, Cambridge MA, for Wnt-encoding cDNAs; J. Nathans, John Hopkins University, Baltimore, MD, for the mouse Frizzled plasmids; A. Rattner for the mouse sFRP plasmids; J. Gosselin, University of Massachusetts-Worcester Transgenic Mouse Facility, and J. Lawitts and K. Herzberg, Beth Israel Hospital, Boston MA, for transgenic microinjection services. They also thank Erna Engholm Pedersen, Tove Funder-Nielsen, and Heidi Ingemann Jensen for technical assistance and Bob Gimlich, Mark Williamson, Genevieve Arnold, Lisa Anne Whittemore, the Members of the Embryonic Growth and Regulatory Proteins Group, Discovery Research, and the Genetics Institute for their helpful comments and suggestions. Thanks also to P. Bouchard, Preclinical Research, Genetics Institute, for histology and pathology consultations.