Wnt5a is essential for intestinal elongation in mice

doi:10.1016/j.ydbio.2008.11.020

Dev Biol. Author manuscript; available in PMC 2010 February 15.

Published in final edited form as:

Dev Biol. 2009 February 15; 326(2): 285–294.

Published online 2008 December 7. doi: 10.1016/j.ydbio.2008.11.020

PMCID: PMC2654720

NIHMSID: NIHMS95931

Wnt5a is essential for intestinal elongation in mice

Sara Cervantes,¹ Terry P. Yamaguchi,² and Matthias Hebrok¹^*

¹ Diabetes Center, Department of Medicine, University of California, San Francisco, CA 94143, USA

² Cancer and Developmental Biology Laboratory, Center for Cancer Research, National Cancer Institute-Frederick, NIH Frederick, MD 21702, USA

^* To whom correspondence should be addressed: Matthias Hebrok, Phone: (415) 514-0820, Fax: (415) 564-5813, Email: mhebrok/at/diabetes.ucsf.edu

The publisher's final edited version of this article is available at Dev Biol

See other articles in PMC that cite the published article.

Publisher's Disclaimer

Summary

Morphogenesis of the mammalian small intestine entails extensive elongation and folding of the primitive gut into a tightly coiled digestive tube. Surprisingly, little is known about the cellular and molecular mechanisms that mediate the morphological aspects of small intestine formation. Here, we demonstrate that Wnt5a, a member of the Wnt family of secreted proteins, is essential for the development and elongation of the small intestine from the midgut region. We found that the small intestine in mice lacking Wnt5a was dramatically shortened and duplicated, forming a bifurcated lumen instead of a single tube. In addition, cell proliferation was reduced and re-intercalation of post-mitotic cells into the elongating gut tube epithelium was disrupted. Thus, our study demonstrates that Wnt5a functions as a critical regulator of midgut formation and morphogenesis in mammals.

Keywords: Wnt, Wnt5a, endoderm, midgut development, small intestine elongation

Introduction

The small intestine derives from the definitive endoderm, one of the three germ layers formed during gastrulation. At this stage, the mouse embryo is organized in a U-shape rather than a flat sheet of cells. The mouse definitive endoderm is located at the outside of the embryo and is contiguous with the visceral endoderm that gives rise to the yolk sac. At the early somite stage (E8.5), the U-shaped mouse embryo turns, thereby reversing its topography and as a consequence the endoderm relocates into the embryonic body cavity. As a result of these morphogenetic movements the definitive endoderm forms the three-dimensional primitive gut tube by mouse embryonic day 9 (E9.0). Shortly thereafter, the primitive gut tube gives rise to the digestive organs, establishing a functional gastrointestinal tract that is divided into distinct regions: esophagus, stomach, small intestine and large intestine. The small intestine originates from the midgut region by E10.0 as a hairpin loop that grows towards the ventral side of the embryo. Over the next 5 days, the gut extends while rotating to form the convoluted intestinal tract. Due to space constraints, the intestine grows outside the embryonic peritoneal cavity during this period, thereby forming the physiological umbilical hernia. The elongated and tightly wound small intestine withdraws into the peritoneal cavity by E15.5, possibly due to contraction of the musculature of the duodenum and proximal jejunum (Kaufman and Bard, 1999). Thus, generation of the mature intestinal tract entails dramatic morphogenetic changes. However, despite important advances in the understanding of embryonic development, the cellular and molecular mechanisms that underlie elongation of the small intestine remain obscure.

One pathway that is involved in intestinal organogenesis is Wnt signaling. The multiple Wnt ligands described in vertebrate animals signal through two distinct mechanisms: the Wnt/β-catenin and the non-canonical Wnt pathways. Of the two pathways, the Wnt/β-catenin pathway is well understood, whereas the different arms of the non-canonical Wnt pathway are still not fully elucidated. Originally, different Wnt ligands were assigned to one or the other pathway according to their signaling properties in specific assays. However, more recent evidence suggests that signaling through one or the other pathway is dictated by the composition of the Wnt receptors on receiving cells.

One of the Wnt ligands that traditionally had been assigned to the non-canonical pathway is Wnt5a. In frogs and zebrafish Wnt5a regulates convergence and extension movements during gastrulation in a process likely mediated by Ror2, RhoA GTPase, Jun kinase (JNK), and Ca²⁺ release. Moreover, Wnt5a has been shown to inhibit the canonical Wnt/β-catenin pathway in different systems, although the mechanisms that mediate this activity remain controversial. In contrast, other results indicate that in the presence of the appropriate receptors, Wnt5a signals through the canonical Wnt/β-catenin pathway. Similar to Wnt5a mutants in frog and zebrafish, mice deficient in Wnt5a show a profound defect in posterior elongation and morphogenesis of outgrowing structures with no alterations in cell fate. In any case, the exact mechanisms of Wnt5a signaling remain unknown and downstream effectors in mice have not been identified.

Here, we have analyzed the requirement for Wnt5a function during formation of the intestinal tract. During mouse development, Wnt5a has been shown to be expressed in the gut mesenchyme. Given its role in other tissues, we hypothesized that Wnt5a might govern intestinal elongation in mouse embryos. Here, we show that Wnt5a mutants display a dramatic shortening of the small intestine accompanied by an aberrant bifurcation of the midgut. This phenotype results from a combination of defective closure of the primitive gut tube at E10.0 and abrogated midgut elongation starting at E10.5. Notably, Wnt5a is not required for the differentiation of the diverse intestinal cell types or for the activation of the canonical Wnt/β-catenin pathway. In contrast, Wnt5a is essential to maintain the architecture of the growing epithelium by regulating re-intercalation of post-mitotic cells into the epithelium after cell division, and by controlling cell proliferation during midgut elongation. Thus, Wnt5a mutant mice reveal critical information about the cellular basis and dynamics of small intestine development.

Materials and methods

Mice

Mice used in this study were maintained in a barrier facility according to protocols approved by the Committee on Animal Research at the University of California, San Francisco. Mice were kept on a 12-hour light/dark cycle. Wnt5a heterozygous mice were generated and described previously and maintained on a C57BL/6J background. Wnt5a^+/⁻ were intercrossed to generate Wnt5a⁻^/⁻ at different embryonic stages. Vangl2^Lp/+ mice were obtained from Jackson laboratories and kept on a mixed background to generate Vangl2^Lp/Lp. TCF/LEF reporter mice (TOPGAL mice) have been described elsewhere. The morning of the appearance of a vaginal plug was considered as embryonic day 0.5. Mutant embryos were identified by their external appearance at all embryonic stages analyzed.

In situ hybridization and immunohistochemistry

Embryos derived were dissected in Phosphate Buffered Saline (PBS) and fixed in 4% paraformaldehyde (PFA). Wnt5a RNA detection was performed as published (). For immunohistochemistry, embryos derived from Wnt5a^+/⁻ and Vangl2^Lp/+ crosses were dissected in PBS and fixed overnight in 4% PFA. For analysis of tissue sections, embryos were dehydrated, embedded in paraffin wax, and sectioned at 5 μm. For general histology, sections were stained with hematoxylin and eosin. Immunofluorescence and immunohistochemistry analyses were performed as described previously. For Vangl2 immunofluorescence, wild-type embryos were processed as described previously (Montcouquiol et al., 2006). Staining using phalloidin and DBA was performed on 10μm cryosections without antigen retrieval. For whole mount immunofluorescence, whole embryos (E9.5) or the dissected abdominal region (E10.5–11.5) were blocked for at least overnight in blocking solution: 4% Bovine Serum Albumin (BSA) in PBS 0.3–0.5% Triton X-100 at 4°C. All subsequent steps were also carried out at 4°C. Tissues were incubated with the indicated primary antibodies for 36–48 hours, rinsed in washing solution (2% BSA in PBS 0.3–0.5% Triton X-100) at least three times, and incubated overnight with the corresponding secondary antibodies. Alternatively, E-cadherin whole mount immunofluorescence was performed as described previously (Ahnfelt-Rønne et al., 2007). Finally, embryos were rinsed in washing solution before mounting in Vectashield (Vector Labs) for confocal imaging. The following primary antibodies were used: mouse anti-Cdx2 diluted 1:100 (Biogenex, San Ramon, CA), rat anti-E-cadherin clone ECCD2 diluted 1:100 for staining on sections and 1:1000 for whole mount staining (Calbiochem, La Jolla, CA), mouse anti-E-cadherin diluted 1:200 (BD biosciences, San Jose, CA), rabbit anti-Sox9 diluted 1:200 (Chemicon, Temecula, CA), rabbit anti-phosphorylated Histone H3 diluted 1:200 (Upstate, Charlottesville, VA), rabbit anti-Ki67 diluted 1:200 (Novocastra, Vision BioSystems, Norwell, MA), mouse anti-α-smooth muscle actin diluted 1:200 (Sigma, Saint Louis, MI), rat anti-BrdU diluted 1:200 (Serotec, Raleigh, NC), mouse anti-Synaptophysin diluted 1:200 (Biogenex, San Ramon, CA), rabbit anti-lfabp diluted 1:50 (Novus Biologicals, Inc., Litleton, CO), mouse anti-villin diluted 1:100 (Immunotech, Marseille, France) and rabbit anti-Vangl2 diluted 1:500 was a kind gift from M. Kelley (Montcouquiol et al., 2006). Biotinylated Dolichos Biflorus Agglutinin (DBA) was obtained from Vector Laboratories (Burlingame, CA) and used at a 1:200 dilution, and phalloidin conjugated to Alexa Fluor 488 was obtained from Molecular Probes (Carlsbad, CA). Avidin-FITC and avidin-Cy3 (Jackson Immunoresearch, West Grove, PA) were used at a 1:200 dilution. The secondary antibodies used were goat anti-rabbit Alexa Fluor 633, goat anti-rat Alexa Fluor 488, goat anti-mouse Alexa Fluor 568, goat anti-rabbit Alexa Fluor 488 (Molecular Probes, Carlsbad, CA), goat anti-mouse biotinylated, and goat anti-rabbit biotinylated (Vector Laboratories, Burlingame, CA). Bright-field images were acquired using a Zeiss Axio Imager D1 microscope; fluorescent images were captured using a Zeiss Axiophoto2 plus and a Leica SL or SP2 confocal microscope.

X-gal staining

E10.5 and E11.5 embryos derived from crosses of Wnt5a^+/⁻ with Wnt5a^+/⁻; TOPGAL mice were dissected in PBS, and the peritoneal cavity was opened to facilitate penetration of the reagents. Tissues were fixed for 1 hour in 2% PFA and 0.25% glutaraldehyde in PBS at 4°C and stained according to previous protocols. After staining, tissues were washed in PBS and fixed overnight in Zinc-buffered formalin (Anatech, Hayward, CA). The foregut-midgut areas, including the lungs as a positive control, were further dissected for visualization and imaging on a Leica MZ FL3 equipped with a Leica IM500 system.

Quantitative real-time PCR

E11.5 midgut tissues were individually harvested in ‘RNA later’ (Ambion, Austin, TX) and stored at −20°C. After genotyping, 2–3 Wnt5a^+/⁻ and 1–3 Wnt5a⁻^/⁻ tissues from the same litter were pooled together according to their genotype for RNA extraction. A total of 5 litters were analyzed. RNA was prepared with the ‘QIAeasy RNA extraction kit’ (QIAgen, Valencia, CA) and contaminating traces of genomic DNA were digested using Q1 DNase (Promega, Madison, WI). After further clean up with the ‘QIAeasy RNA kit’, 1 μg of total RNA was used for cDNA synthesis using the ‘Superscript First-Strand Synthesis System’ (Invitrogen, Carlsbad, CA). PCR reactions were performed as previously described. The following primers were used: Gus: ACGGGATTGTGGTCATCGA and TCGTTGCCAAAACTCTGAGGTA; CyclophilinA: TCACAGAATTATTCCAGGATTCATG and TGCCGCCAGTGCCATT; Axin2: AAAACGGATTCAGGTCCTTCAA and GCAAAGACATAGCCGGAACCTA; Lef1: TCCTCTCAGGAGCCCTACCA and GGCCTCCGTCTGGATGCT; Tcf1: GCTGCCATCAACCAGATCCT and AGTTCATAGTACTTGGCCTGCTCTTC.

BrdU pulse chase

In vivo BrdU pulse chase labeling of El1.5 embryos was performed as published () with some minor modifications. Pregnant females were injected intraperitoneally with a single dose of BrdU (20μg/gram body weight) and labeled for 30 minutes. Labeled cells were chased by injection of 500μg Thymidine/gram body weight every 2 hours for 8 hours. After dissection, embryos were fixed overnight in Zinc-buffered formalin and processed for immunofluorescence as described above. For BrdU staining, antigen retrieval was performed in 1N HCl for 8 min at 65°C.

Quantification of cell proliferation

We performed immunofluorescence stainings using anti-Cdx2 and anti-phosphorylated Histone H3 antibodies to mark gut cells and cells in mitosis, respectively. Four pairs of control and Wnt5a mutant E11.5 littermates were stained and all Cdx2 positive and phosphorylated Histone H3 positive cells were counted along the midgut region. The percentage of phosphorylated Histone H3 positive cells was calculated for each embryo and the values obtained for each Wnt5a mutant embryo were normalized to the values obtained for the corresponding control littermate, which was considered to be 100%. Thus, each pair of littermates was treated as an independent experiment. A Student’s t-test was applied to determine whether the differences were statistically significant.

Results

Wnt5a is expressed in the midgut region at the onset of gut elongation

Previous work has shown that Wnt5a is expressed in the gut mesenchyme during mouse intestinal development and that this expression is maintained in the small intestine throughout its development. To determine whether Wnt5a is required for intestinal elongation, we first investigated whether Wnt5a is expressed specifically during midgut elongation. Wnt5a expression was first detected in the mesenchyme in the midgut region at E9.5 (Fig. 1A,C) and, as the midgut elongated to form the hairpin loop that later extends into the umbilical physiological hernia (E10.0), Wnt5a expression increased (Fig. 1B,D,E,G). Wnt5a expression was also easily detected in the posterior-most hindgut endoderm and mesenchyme of the tailbud (data not shown). At E12.5, Wnt5a was expressed in the mesenchyme of the midgut and hindgut but excluded from the cecum (Fig. 1F,H). Thus, Wnt5a expression in the mesenchyme surrounding the endoderm correlates with morphogenesis of the small intestine and with gut regions undergoing elongation.

Fig. 1

Wnt5a is expressed during midgut elongation

Wnt5a mutants show multiple defects along the gastrointestinal tract

To address whether Wnt5a is required during development of the gastrointestinal tract, we obtained tissues from Wnt5a mutants at E18.5 (Wnt5a ⁻^/⁻ mice die at birth, preventing an analysis of postnatal gut maturation and function). We found that the overall length of the gastrointestinal tract was severely reduced in mutant embryos, although the stomach, small intestine, cecum and large intestine were present. Side–by-side comparison of the control and mutant gastrointestinal tracts from the stomach to the anus revealed that the small intestine was most severely affected (Fig. 2A). In addition, quantification of the intestinal tract length confirmed that whereas the large intestine reached 37% of its normal length, the small intestine was reduced by almost 80% (Fig. S1). Similar differences were detected earlier in development at E14.5 (Fig. S1). Higher magnification bright-field images showed a secondary gut tube that formed parallel to the main gastrointestinal tract along the anterior-posterior axis (Fig. 2B). Moreover, Wnt5a mutants had a shorter cecum and imperforated anus (Fig. 2E–F and 2I), and displayed a mild distension of the small intestine, which appeared 30% wider than that of control embryos (Fig. S1). Despite these profound morphological defects, the general cellular architecture of the intestine was not affected. Hematoxylin and eosin staining of sections showed a well developed tissue architecture, including the presence of the characteristic villi, in both the main gut tube and the duplication of the small intestine (Fig. 2C–D). Similarly, the cellular composition of the large intestine appeared intact, with readily recognizable crypts of Lieberkühn (Fig. 2G–H).

Fig. 2

Multiple defects along the gastrointestinal tract in Wnt5a mutants

To examine possible defects in the differentiation state of the cell types within the small intestine and the duplicated region, we performed immunohistochemical analysis. Staining against Ki67, a marker of proliferating cells, demonstrated that the intervillus regions remained highly proliferative (Fig. 3A–B). In addition, Sox9, a marker of stem/progenitor cells which is necessary for Paneth cell differentiation (Mori-Akiyama et al., 2007; Bastide et al., 2007), was appropriately expressed at the base of the villi (Fig. S3). These observations suggest that Wnt5a is neither required to maintain the progenitor cell pool nor to regulate cell differentiation in the gastrointestinal tract after villi formation. These findings are in contrast to those obtained from mice in which the canonical Wnt/β-catenin signaling pathway is perturbed. We further found that smooth muscle actin (SMA) was expressed in the submucosa of the small intestine in control and Wnt5a mutant embryos, as well as within the duplication. However, the smooth muscle layers appeared reduced in Wnt5a mutant embryos as compared to control littermates (Fig. 3C–D). Moreover, the number of subepithelial SMA-positive fibroblasts was lowered in Wnt5a mutant embryos (Fig. 3C–D). We quantified the number of proliferating cells (positive for Ki67) in the smooth muscle and SMA-positive fibroblasts at E18.5 but failed to detect any changes (Fig. S2). Stainings against lfabp and villin demonstrated that enterocytes differentiated in the absence of Wnt5a (Fig. 3C–D, S3). In addition, expression of Cdx2, a homeobox transcription factor and intestinal marker, was unaltered in Wnt5a mutant embryos (Fig. 3E–F). Dolichos Biflorus Agglutinin (DBA), which binds to mucins secreted by goblet cells, revealed the presence of this cell type in both the control littermates and Wnt5a mutant tissues (Fig. 3E–F). To investigate the presence of neural derivatives in the gut, including enteroendocrine cells and the submucosal nerve plexus, we used the pan-neural marker synaptophysin. As shown in Fig. 3H, the main and duplicated branch of the Wnt5a mutant small intestine were innervated and contained abundant enteroendocrine cells. In short, the small intestine of Wnt5a mutant embryos harbors all the different intestinal histological layers and the three intestinal cell lineages that form during embryogenesis (enterocytes, goblet cells and enteroendocrine cells). Therefore, Wnt5a does not appear to control cell specification or differentiation. Rather, the shortened cecum, imperforated anus and truncated/bifurcated small intestine of mutants all point to a requirement for Wnt5a function in the regulation of gut morphogenesis.

Fig. 3

Loss of Wnt5a does not alter cell differentiation in the small intestine

Wnt5a function is required for gut closure

During formation of the primitive gut, the gut endoderm in the midgut area is contiguous with the yolk sac endoderm, a tissue derived from the visceral endoderm. By E9.5, the connection between the midgut endoderm and the yolk sac, known as the vitelline or omphalomesenteric duct, progressively narrows into a rod-like structure (Kaufman & Bard, 1999). In order to determine the temporal requirements for Wnt5a function during morphogenesis of the small intestine, we first investigated midgut closure and vitelline duct formation. We compared control and Wnt5a mutant embryos at E9.5 and E10.0, when midgut elongation is not yet initiated, by performing whole mount indirect immunofluorescence stainings against E-cadherin and Cdx2 to mark the endoderm epithelium, and DBA lectin to detect the gut mucins present within the vitelline duct and gut lumen. Our results show that in 20–22 somite embryos (~E9.5) the control midgut consisted of a well formed tube connected to an almost closed vitelline duct (Fig. 4A, C). In contrast, the midgut of Wnt5a mutant embryos was not as closed and the lumen of the vitelline duct appeared much wider at the junction with the midgut (Fig. 4B, D). By E10.0 (24–26 somite-embryos) the lumen of the vitelline duct in control embryos was dramatically reduced and almost completely closed proximal to the junction with the midgut (Fig. 4E). In Wnt5a mutant embryos, the midgut stayed open and the lumen of the vitelline duct remained expanded when compared to control embryos (Fig. 4F), indicating that the tubular structure of the primitive gut is not fully formed at the onset of midgut elongation. Despite the perturbed morphology of the vitelline duct, its epithelial architecture, which consisted of a monolayer of squamous cells, was preserved in Wnt5a mutant embryos (Fig. 4G–H). These observations demonstrate that the closure of the midgut and the vitelline duct are perturbed in Wnt5a mutant embryos. As a consequence, the morphology of the primitive gut tube is aberrant at the onset of midgut elongation.

Fig. 4

Defects in vitelline duct closure in Wnt5a mutant embryos

Midgut duplication and elongation in Wnt5a mutants

Morphogenesis of the small intestine starts after midgut closure at E10.5. During this process, the midgut extensively elongates and loops to eventually form the umbilical physiological hernia. Given the above described defects in vitelline duct closure and the intestinal shortening and branching observed at E18.5 in Wnt5a mutant embryos, we tested for additional defects in small intestine morphogenesis. To this end, whole mount indirect immunofluorescence was performed against Cdx2 to mark the endoderm epithelium and DBA lectin was used to detect gut mucins present within the vitelline duct and gut lumen. The first sign of a parallel branching extending from the midgut endoderm was observed between E10 and E10.5 in the region of the forming midgut (Fig. 5B,D). In control embryos, the midgut forms a loop structure that grows ventrally between the prospective duodenum and the hindgut at E10.5 (Fig. 5A,C). At E11.5, this loop adopts a hairpin shape with two morphologically distinct portions: the anterior portion, which later gives rise to the small intestine (Fig. 5G, red), and the posterior portion, which subsequently will form the large intestine (Fig. 5G, red dots). The junction between the small and large intestine is marked by the cecum (Fig. 5G, green). The duplicated branch in Wnt5a mutant embryos arose from the area of the midgut caudal to the stomach that is normally destined to form the loop structure (Fig. 5B,D). In E11.5 wild-type embryos, the midgut hairpin loop has adopted the characteristics of the small and large intestine, separated by the cecum (Fig. 5E). The elongation of the midgut region, the budding of the cecum and the formation of the hairpin-shaped loop were not observed in Wnt5a mutants. Instead, the midgut region was dysmorphic and harbored a duplication, which continued to grow in an anterior-posterior direction as a secondary branch away from the main gastrointestinal tract (Fig. 5F,H). All Wnt5a mutant embryos analyzed from E10.5 onwards displayed this duplication, and its localization within the gastrointestinal tract was strikingly conserved.

Fig. 5

Midgut elongation is abnormal in Wnt5a mutants

To further define the precise location of the abnormal bifurcation, we analyzed the position of the vitelline duct during midgut elongation. We performed whole mount fluorescence stainings using DBA to mark the vitelline duct and found that in E10.5 embryos the vitelline duct is located at the tip of the bifurcation. At this stage, the vitelline duct appears as a rod-like structure in both control and Wnt5a mutant embryos (Fig. S4). In E11.5 wild-type embryos, the vitelline duct was located in the anterior branch of the hairpin loop (Fig. 5I and 5G, blue). We also noted that at these stages, the gut tube appears wider than in control littermates. We quantified these defects on sections through the midgut region in control and Wnt5a mutant embryos and found that the diameter of the midgut was increased by 42.5% in the absence of Wnt5a (Fig. S9). Interestingly, in Wnt5a mutant embryos the vitelline duct was always connected to the tip of the bifurcated branch (Fig. 5J and 5H, blue). Thus, instead of elongating the forming gut tube, Wnt5a mutant embryos develop a duplication that originates at the junction of the midgut and the yolk sac.

Wnt5a does not interfere with canonical Wnt/β-catenin signaling

Previous studies have demonstrated that Wnt5a is capable of both activating and inhibiting the canonical branch of the Wnt pathway,. To investigate whether changes in the canonical Wnt/β-catenin pathway could regulate intestinal elongation downstream of Wnt5a, we crossed Wnt5a mutant mice and TOPGAL mice. TOPGAL mice carry the β-galactosidase gene under the control of the canonical Wnt-responsive Tcf promoter sequence and serve as reporters for β-catenin dependent Wnt signaling. At E10.5 and E11.5, strong activity was detected in the trachea, lung buds, esophagus, fore-stomach and dorsal pancreatic bud. However, β-galactosidase activity was undetectable in the duodenum and midgut regions in both control and Wnt5a mutant embryos (Fig. 6A–B), strongly suggesting that canonical Wnt/β-catenin is predominantly inactive during the initiation of midgut elongation. Furthermore, we could not detect any changes in the overall β-galactosidase activity levels or distribution in the foregut and midgut areas of the Wnt5a mutant embryos at those embryonic stages (Fig. 6A–B). These observations were confirmed on sections from E10.5 and E11.5 midguts stained for β-galactosidase activity (data not shown). Next, quantitative PCR analysis was performed to determine the expression levels of Wnt/β-catenin target genes in the midgut. Axin2, Lef1 and Tcf1, considered to represent ubiquitous Wnt/β-catenin target genes, were chosen to assess changes in canonical Wnt/β-catenin pathway. Lef1 showed a minor increase (2-fold) in the Wnt5a mutant that was not statistically significant, whereas axin2 and Tcf1 remained unchanged (Fig. 6C). The absence of β-galactosidase activity during midgut elongation and the absence of robust changes in canonical Wnt signaling target genes indicate that Wnt5a does not regulate the Wnt/β-catenin pathway during this process.

Fig. 6

Wnt/β-catenin pathway during gut elongation

Wnt5a genetically interacts with Vangl2, one of the core components of the planar cell polarity (PCP) pathway during mouse cochlear lengthening and neurulation. Vangl2, the ortholog of the Drosophila strabismus gene, also regulates PCP and convergence and extension movements in frog, zebrafish, and mouse. To assess whether the gut defects observed in Wnt5a mutant embryos are due to changes in the PCP pathway, we first analyzed Vangl2 expression during midgut elongation and found that Vangl2 is present at E11.5 in the midgut epithelium (data not shown). We then examined looptail mice (Vangl2^Lp/Lp) that carry a loss-of-function point mutation in the Vangl2 gene. We found that, in contrast to Wnt5a mutants, elongation of midgut structures proceeded normally in Vangl2^Lp/Lp mutants (Fig. S5). Thus, Vangl2 is not required for midgut elongation.

Wnt5a regulates epithelial architecture in the forming midgut

Small intestine morphogenesis from the midgut area in Wnt5a mutant embryos results in the formation of a bifurcated and shortened gut tube. To gain insights into the mechanisms by which Wnt5a controls midgut elongation, we analyzed the cellular aspects of this process. Immunofluorescence staining against E-cadherin, a marker of cell membranes, revealed that the primitive gut tube was formed by a single layer of cuboidal cells in both control and Wnt5a mutant embryos at E9.0 (Fig. 7A–B). As the midgut tube elongated at E10.5 (Fig. S6) and E11.5 (Fig. 7C, S6, S7), the cells in the epithelium of control embryos adopted a columnar shape. In Wnt5a mutant embryos, the epithelial lining of the gut was thickened (Fig. 7D, S6, S7) and the overall diameter of the tube was visibly wider (Fig. S6). A closer inspection revealed that the mutant epithelium was composed of 2 to 4 cell layers (Fig. 7D, S6, S7) compared to the 1 to 2 cell layers found in control tissues (Fig. 7C, S6, S7). Furthermore, the gut epithelial cells in mutant embryos adopted aberrant morphologies that differed from the columnar shape found in control littermates (Fig. 7C–D, S7, S8). Such alterations in cell and tissue architecture were not detected in more posterior gut regions, suggesting that they are specific for the elongating midgut (data not shown). Despite these changes, E-cadherin localization was maintained in the basal-lateral sides of cells in both the control littermates and Wnt5a mutants, suggesting that epithelial cell polarity remained intact (Fig. 7C–D, S6, S7). In support of this hypothesis, atypical protein kinase C (aPKC), an apical marker, and laminin-1, a basement membrane protein and basal marker, were also found to be appropriately localized in Wnt5a control and mutant gut epithelium (Fig. S8). Furthermore, intestinal mucins and actin filaments were appropriately accumulated on the apical side of the epithelial cells (Fig. S8). As development progressed, the mutant intestinal epithelium stratified at E14.5 (data not shown) and formed villi with normal looking appearance at E15.5 (Fig. 7E–F). Thus, loss of Wnt5a function temporally disturbs the formation of an organized gut epithelium during the initiation of midgut elongation. At later stages, Wnt5a function is dispensable and the multi-layered epithelium acquires a normal morphology.

Fig. 7

Architecture of the gut epithelium is disrupted in the absence of Wnt5a

Wnt5a regulates cell proliferation and reintegration of post-mitotic cells into the epithelial layer during midgut elongation

The cellular basis of small intestine elongation in mammals is unknown, but the dramatic increase in length over a short period of time suggests that cell proliferation plays an important role. Using phosphorylated Histone H3 as cell proliferation marker in Wnt5a control and mutant midguts we found that cell proliferation was reduced by 36% and 16.2% in the epithelial and mesenchymal compartments respectively at E11.5 (Fig. 7G, S9) in the absence of Wnt5a. Furthermore, mitotic cells adopted a rounded shape (data not shown) and were located on the luminal side of the gut epithelium in both the control and Wnt5a mutants (Fig. 7H–I), indicating that they transiently delaminate to divide. In order to maintain the single-layered structure observed in the control embryos, the newly post-mitotic cells would have to re-integrate into the expanding layer. To test this hypothesis, we performed pulse-chase experiments over an eight-hour period during midgut elongation (E11.5). The 30 min pulse consisted of an initial administration of BrdU to label those cells currently in S phase. Subsequently, pregnant females were injected with an excess of Thymidine every 2 hours (chase) (Fig. 8A). We detected post-mitotic cells by BrdU staining and found that in control tissues cells reintegrated in the growing epithelial monolayer upon cell division (Fig. 8A,B), strongly suggesting that indeed, cell proliferation followed by re-intercalation drives midgut elongation. In contrast, post-mitotic cells remained predominantly in the most apical layers in the Wnt5a mutant epithelium (Fig. 8A,C). In some cases, BrdU positive cells formed a ring in the inner side of the gut tube (Fig. 8C), suggesting that cells did not re-intercalate after completing cell division. This pattern was consistent throughout the midgut region, including the duplicated area. We quantified re-intercalated cells by calculating the proportion of BrdU positive cells located in the outermost layer of the epithelium and found a reduction from 64.6% in control embryos to 37.6% in Wnt5a mutant embryos over several sections along the midgut (Fig. 8D). Thus, these results demonstrate that re-intercalation of the post-mitotic cells in the gut epithelium is disrupted in Wnt5a mutant embryos. As a consequence, the originally mono-layered epithelium becomes stratified rather than elongated along the anterior-posterior axis. Interestingly, the differences in the ratio of re-intercalated cells attenuated as the midgut epithelium became stratified in both control and Wnt5a mutant embryos at E13.5 (Fig. S10). Of note, changes in the spatial orientation of cell division were not found in Wnt5a mutant embryos (data not shown).

Fig. 8

Re-intercalation of post-mitotic cells is perturbed in Wnt5a mutant gut epithelium

Discussion

Results presented in this study demonstrate that Wnt5a is required for different processes necessary for proper gut morphogenesis. We show that Wnt5a function is essential for primitive gut closure before E10.0 and elongation of the midgut region starting at E10.5. Loss of the gene in mutant embryos results in the development of a prominent bifurcation within this region, and a dramatically shortened small intestine.

Mechanisms of bifurcation and intestinal shortening

The analysis of the Wnt5a mutant midgut phenotype unravels a previously unappreciated requirement for this factor during proper midgut closure and formation of the primitive gut tube. During normal development, the gut elongates from a closed, well-formed tube and assumes its final shape through a complicated process of bending and coiling. In Wnt5a mutants, the midgut region does not fully close and the vitelline duct is dilated at the onset of midgut elongation. As a consequence, as the intestinal tract expands, the still open midgut branches at the junction with the vitelline duct, thereby disrupting the normal intestinal morphology. The bifurcation observed in Wnt5a mutant embryos does not regress over time and is still clearly visible at E18.5, indicating that early defects during midgut closure are not rectified at later stages.

Several mechanisms could explain the gut elongation defect observed in Wnt5a mutants. Our findings point to changes in cell proliferation as the underlying problem accounting for the rapid elongation of the midgut in wild-type embryos. Similarly, defects in the outgrowth or elongation of the limb and primary body axis in Wnt5a mutant embryos were associated with changes in proliferating cells (Yamaguchi et al., 1999). Not only did we detect a measurable reduction in cell proliferation in Wnt5a mutant embryos, but our analysis also revealed defects in re-integration of post-mitotic cells into the growing gut epithelium. Dividing cells in both wild-type and Wnt5a mutant embryos partially delaminate towards the luminal side of the epithelium and adopt a rounded shape. In order to maintain the gut monolayer and contribute to its elongation, the newly generated cells re-intercalated into the epithelium, thereby extending the existing single cell layer of the gut tube. In Wnt5a mutant embryos, post-mitotic cells do not re-intercalate properly and accumulate on the luminal side of the epithelium. As a consequence, the midgut epithelium of the Wnt5a mutants appears disorganized and contains 3 to 4 layers of cells. The accumulation of cells in the midgut of Wnt5a mutant embryos is reminiscent of defects in convergence and extension, a cellular process used in several organs to promote narrowing along the medio-lateral axis while elongating along the anterior-posterior axis. Thus, the defects in re-intercalation found in Wnt5a embryos could be interpreted as a disruption in the convergence and extension-like mechanism responsible for midgut elongation. Surprisingly, despite these dramatic epithelial defects observed at E11.5, the intestinal epithelium in Wnt5a mutants recovers by the end of embryogenesis and cell differentiation appears normal, even though the intestine is still dramatically shortened. Thus, the reduction in cell proliferation, the transient formation of a stratified epithelium, and the growth of a secondary branch of the small intestine, account for the profound shortening of the small intestine in Wnt5a mutant mice.

Wnt5a does not regulate gut morphogenesis via the canonical Wnt/β-catenin pathway

The exact mechanism of action by which Wnt5a regulates aspects of mouse development is currently unknown. Data obtained from diverse in vivo and in vitro systems suggest that Wnt5a can signal through different pathways, including the canonical arm (Wnt/β-catenin) and non-canonical arms (Wnt/PCP, Wnt/JNK and Wnt/Ca²⁺) of Wnt signaling. Recent studies have provided evidence that Wnt5a can either activate or inhibit the canonical Wnt pathway in different experimental systems. The absence of reporter activity in control TOPGAL and Wnt5a⁻^/⁻;TOPGAL reporter mice demonstrates that Wnt/β-catenin signaling is not activated during midgut elongation. In addition, the largely unchanged expression levels of canonical Wnt target genes indicate that Wnt5a does not modulate the Wnt/β-catenin pathway during this process. Our results are consistent with previous studies in TOPGAL and Axin^LacZ mice showing undetectable β-galactosidase activity before E16.25. These observations indicate that Wnt/β-catenin pathway becomes activated only during later developmental processes, such as villi formation and differentiation of the different intestinal cell types. Results from other experiments in which Wnt/β-catenin activity was manipulated are also in agreement with these findings. Specifically, inhibition of the Wnt/β-catenin pathway results in the destruction of intestinal villi at birth, whereas its activation leads to massive hyperproliferation of intestinal crypts, defects not observed in Wnt5a mutants. Thus, we exclude a role for Wnt5a in the regulation of Wnt/β-catenin signaling during midgut elongation.

Mouse mutants in Wnt/PCP genes show phenotypes similar to that of the Wnt5a mutants, consisting of a shortened and wider body axis. Wnt5a and Vangl2, one of the core proteins of the Wnt/PCP pathway, genetically interact to regulate neurulation, cochlear elongation and stereociliary bundle orientation. In contrast to these other organs, the requirement of the Wnt/PCP pathway in the mouse endoderm has not been explored. Given that Vangl2 is the most accepted regulator of Wnt/PCP, we hypothesized that Vangl2^Lp/Lp mice might reveal whether Wnt/PCP regulates gastrointestinal elongation. Our study demonstrates that the midgut grew normally in Vangl2^Lp/Lp mutants at E11.5 and did not consist of a multi-layered epithelium (Fig. S5), suggesting that Wnt5a regulates this process independently of Vangl2. Alternatively, the absence of a midgut phenotype in Vangl2 mutants could be due to the redundant function of other Wnt/PCP components that might mask the role of PCP in midgut elongation and future studies would have to address this notion.

Formation of the mature intestinal tract involves not only differentiation of the diverse cell types, but also extensive elongation of a tubular structure. These different aspects of gastrointestinal morphogenesis are conserved in mammals, including humans, where the intestinal tract elongates, rotates and transiently herniates outside the body cavity. Although rare, duplications in several regions of the intestinal tract are found in humans. Newborns can present with congenital short bowel syndrome, a condition that can become life-threatening when diagnosis is delayed (Schalamon et al., 1999; Hasosah et al., 2008). Moreover, some cases of bowel diverticulum are diagnosed as Meckel’s diverticulum even though they might actually have other etiologies, such as tubular duplication of the intestine. In this context, our findings place Wnt5a as a critical molecule governing early steps of gut formation and open up new avenues to explore the origin of these conditions.

Supplementary Material

Supplemental Figure 1

Click here to view.^{(53K, pdf)}

Supplemental Figure 2

Click here to view.^{(133K, pdf)}

Supplemental Figure 3

Click here to view.^{(146K, pdf)}

Supplemental Figure 4

Click here to view.^{(43K, pdf)}

Supplemental Figure 5

Click here to view.^{(176K, pdf)}

Supplemental Figure 6

Click here to view.^{(207K, pdf)}

Supplemental Figure 7

Click here to view.^{(156K, pdf)}

Supplemental Figure 8

Click here to view.^{(227K, pdf)}

Supplemental Figure 9

Click here to view.^{(174K, pdf)}

Supplemental Figure 10

Click here to view.^{(61K, pdf)}

Acknowledgments

We thank M. Kelley for the gift of the anti-Vangl2 antibody. We thank Dr. Patrick Heiser for initial observations and members of the Hebrok lab for helpful discussion. We also thank Drs. Sapna Puri, Grace Wei, David Cano and Limor Landsman for critical reading of the manuscript. We are indebted to Heather Heiser for assistance with animal genotyping, Cecilia Austin for help with tissue processing, and Jimmy Chen for assistance with graphics. Work in MH’s lab is supported by an ADA grant and grants from the NIH (DK60533, CA112537). Research in TPY’s lab was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research. SC is supported by postdoctoral fellowships from the Ministerio de Educación, Ciencia y Deportes from the Spanish government and the National Pancreas Foundation (NPF).

Footnotes

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

Ahnfelt-Rønne J, Jørgensen MC, Hald H, Madsen OD, Serup P, Hecksher-Sørensen J. An improved method for three-dimensional reconstruction of protein expression patterns in intact mouse and chicken embryos and organs. J Histochem Cytochem. 2007;55:925–30. [PubMed]
Bastide P, Darido C, Pannequin J, Kist R, Robine S, Marty-Double C, Bibeau F, Scherer G, Joubert D, Hollande F, Blache P, Jay P. Sox9 regulates cell proliferation and is required for Paneth cell differentiation in the intestinal epithelium. J Cell Biol. 2007;178:635–48. [PMC free article] [PubMed]
Brandvold KA, Ewert DL, Kent SC, Neiman P, Ruddell A. Blocked B cell differentiation and emigration support the early growth of Myc-induced lymphomas. Oncogene. 2001;20:3226–34. [PubMed]
Ciruna B, Jenny A, Lee D, Mlodzik M, Schier AF. Planar cell polarity signalling couples cell division and morphogenesis during neurulation. Nature. 2006;439:220–4. [PMC free article] [PubMed]
DasGupta R, Fuchs E. Multiple roles for activated LEF/TCF transcription complexes during hair follicle development and differentiation. Development. 1999;126:4557–68. [PubMed]
Dessimoz J, Bonnard C, Huelsken J, Grapin-Botton A. Pancreas-specific deletion of beta-catenin reveals Wnt-dependent and Wnt-independent functions during development. Curr Biol. 2005;15:1677–83. [PubMed]
Du SJ, Purcell SM, Christian JL, McGrew LL, Moon RT. Identification of distinct classes and functional domains of Wnts through expression of wild-type and chimeric proteins in Xenopus embryos. Mol Cell Biol. 1995;15:2625–34. [PMC free article] [PubMed]
Gordon MD, Nusse R. Wnt signaling: multiple pathways, multiple receptors, and multiple transcription factors. J Biol Chem. 2006;281:22429–33. [PubMed]
Goto T, Keller R. The planar cell polarity gene strabismus regulates convergence and extension and neural fold closure in Xenopus. Dev Biol. 2002;247:165–81. [PubMed]
Hasosah M, Lemberg DA, Skarsgard E, Schreiber R. Congenital short bowel syndrome: a case report and review of the literature. Can J Gastroenterol. 2008;22:71–4. [PMC free article] [PubMed]
He X, Saint-Jeannet JP, Wang Y, Nathans J, Dawid I, Varmus H. A member of the Frizzled protein family mediating axis induction by Wnt-5A. Science. 1997;275:1652–4. [PubMed]
Ishitani T, Kishida S, Hyodo-Miura J, Ueno N, Yasuda J, Waterman M, Shibuya H, Moon RT, Ninomiya-Tsuji J, Matsumoto K. The TAK1-NLK mitogen-activated protein kinase cascade functions in the Wnt-5a/Ca(2+) pathway to antagonize Wnt/beta-catenin signaling. Mol Cell Biol. 2003;23:131–9. [PMC free article] [PubMed]
Jessen JR, Topczewski J, Bingham S, Sepich DS, Marlow F, Chandrasekhar A, Solnica-Krezel L. Zebrafish trilobite identifies new roles for Strabismus in gastrulation and neuronal movements. Nat Cell Biol. 2002;4:610–5. [PMC free article] [PubMed]
Kaufman MH, Bard JBL. The anatomical basis of mouse development. Academic Press; San Diego, California, USA: 1999.
Keller R. Shaping the vertebrate body plan by polarized embryonic cell movements. Science. 2002;298:1950–4. [PubMed]
Kim SK, Hebrok M, Melton DA. Pancreas development in the chick embryo. Cold Spring Harb Symp Quant Biol. 1997;62:377–83. [PubMed]
Kim BM, Mao J, Taketo MM, Shivdasani RA. Phases of canonical wnt signaling during the development of mouse intestinal epithelium. Gastroenterology. 2007;133:529–38. [PubMed]
Kohn AD, Moon RT. Wnt and calcium signaling: beta-catenin-independent pathways. Cell Calcium. 2005;38:439–46. [PubMed]
Korinek V, Barker N, Moerer P, van Donselaar E, Huls G, Peters PJ, Clevers H. Depletion of epithelial stem-cell compartments in the small intestine of mice lacking Tcf-4. Nat Genet. 1998;19:379–83. [PubMed]
Kuhl M, Sheldahl LC, Malbon CC, Moon RT. Ca(2+)/calmodulin-dependent protein kinase II is stimulated by Wnt and Frizzled homologs and promotes ventral cell fates in Xenopus. J Biol Chem. 2000;275:12701–11. [PubMed]
Lickert H, Kispert A, Kutsch S, Kemler R. Expression patterns of Wnt genes in mouse gut development. Mech Dev. 2001;105:181–4. [PubMed]
Mikels AJ, Nusse R. Purified Wnt5a protein activates or inhibits beta-catenin-TCF signaling depending on receptor context. PLoS Biol. 2006;4:e115. [PMC free article] [PubMed]
Montcouquiol M, Sans N, Huss D, Kach J, Dickman JD, Forge A, Rachel RA, Copeland NG, Jenkins NA, Bogani D, Murdoch J, Warchol ME, Wenthold RJ, Kelley MW. Assymetric localization of Vangl2 and Fz3 indicate novel mechanisms for planar cell polarity in mammals. J Neurosci. 2006;26:5265–75. [PubMed]
Moon RT, Campbell RM, Christian JL, McGrew LL, Shih J, Fraser S. Xwnt-5A: a maternal Wnt that affects morphogenetic movements after overexpression in embryos of Xenopus laevis. Development. 1993a;119:97–111. [PubMed]
Moon RT, Christian JL, Campbell RM, McGrew LL, DeMarais AA, Torres M, Lai CJ, Olson DJ, Kelly GM. Dissecting Wnt signalling pathways and Wnt-sensitive developmental processes through transient misexpression analyses in embryos of Xenopus laevis. Dev Suppl. 1993b:85–94. [PubMed]
Mori-Akiyama Y, van den Born M, van Es JH, Hamilton SR, Adams HP, Zhang J, Clevers H, de Crombrugghe B. SOX9 is required for the differentiation of paneth cells in the intestinal epithelium. Gastroenterology. 2007;133:539–46. [PubMed]
Pasca di Magliano M, Sekine S, Ermilov A, Ferris J, Dlugosz AA, Hebrok M. Hedgehog/Ras interactions regulate early stages of pancreatic cancer. Genes Dev. 2006;20:3161–73. [PubMed]
Qian D, Jones C, Rzadzinska A, Mark S, Zhang X, Steel KP, Dai X, Chen P. Wnt5a functions in planar cell polarity regulation in mice. Dev Biol. 2007;306:121–33. [PMC free article] [PubMed]
Sansom OJ, Reed KR, Hayes AJ, Ireland H, Brinkmann H, Newton IP, Batlle E, Simon-Assmann P, Clevers H, Nathke IS, et al. Loss of Apc in vivo immediately perturbs Wnt signaling, differentiation, and migration. Genes Dev. 2004;18:1385–90. [PubMed]
Schalamon J, Schober PH, Gallippi P, Matthyssens L, Höllwarth ME. Congenital short-bowel; a case study and review of the literature. Eur J Pediatr Surg. 1999;9:248–50. [PubMed]
Schambony A, Wedlich D. Wnt-5A/Ror2 Regulate Expression of XPAPC through an Alternative Noncanonical Signaling Pathway. Dev Cell. 2007;12:779–92. [PubMed]
Sekine S, Gutierrez PJ, Lan BY, Feng S, Hebrok M. Liver-specific loss of beta-catenin results in delayed hepatocyte proliferation after partial hepatectomy. Hepatology. 2007;45:361–8. [PubMed]
Slusarski DC, Yang-Snyder J, Busa WB, Moon RT. Modulation of embryonic intracellular Ca2+ signaling by Wnt-5A. Dev Biol. 1997;182:114–20. [PubMed]
Smith LJ, Stein KF. Axial elongation in the mouse and its retardation in homozygous looptail mice. J Embryol Exp Morphol. 1962;10:73–87. [PubMed]
Solnica-Krezel L. Conserved patterns of cell movements during vertebrate gastrulation. Curr Biol. 2005;15:R213–28. [PubMed]
Stern LE, Warner BW. Gastrointestinal duplications. Semin Pediatr Surg. 2000;9:135–40. [PubMed]
Topol L, Jiang X, Choi H, Garrett-Beal L, Carolan PJ, Yang Y. Wnt-5a inhibits the canonical Wnt pathway by promoting GSK-3-independent beta-catenin degradation. J Cell Biol. 2003;162:899–908. [PMC free article] [PubMed]
Torres MA, Yang-Snyder JA, Purcell SM, DeMarais AA, McGrew LL, Moon RT. Activities of the Wnt-1 class of secreted signaling factors are antagonized by the Wnt-5A class and by a dominant negative cadherin in early Xenopus development. J Cell Biol. 1996;133:1123–37. [PMC free article] [PubMed]
Van Twisk R. A congenital diverticulum of the small intestine; Meckel’s diverticulum or duplication? Neth J Surg. 1984;36:107–11. [PubMed]
Veeman MT, Axelrod JD, Moon RT. A second canon. Functions and mechanisms of beta-catenin-independent Wnt signaling. Dev Cell. 2003;5:367–77. [PubMed]
Wells JM, Melton DA. Vertebrate endoderm development. Annu Rev Cell Dev Biol. 1999;15:393–410. [PubMed]
Wilkinson DG, Nieto MA. Detection of messenger RNA by in situ hybridization to tissue sections and whole mounts. Methods Enzymol. 1993;225:361–73. [PubMed]
Wong GT, Gavin BJ, McMahon AP. Differential transformation of mammary epithelial cells by Wnt genes. Mol Cell Biol. 1994;14:6278–86. [PMC free article] [PubMed]
Yamaguchi TP, Bradley A, McMahon AP, Jones S. A Wnt5a pathway underlies outgrowth of multiple structures in the vertebrate embryo. Development. 1999;126:1211–23. [PubMed]
Yamanaka H, Moriguchi T, Masuyama N, Kusakabe M, Hanafusa H, Takada R, Takada S, Nishida E. JNK functions in the non-canonical Wnt pathway to regulate convergent extension movements in vertebrates. EMBO Rep. 2002;3:69–75. [PubMed]
Ybot-Gonzalez P, Savery D, Gerrelli D, Signore M, Mitchell CE, Faux CH, Greene ND, Copp AJ. Convergent extension, planar-cell-polarity signalling and initiation of mouse neural tube closure. Development. 2007;134:789–99. [PMC free article] [PubMed]
Zhu S, Liu L, Korzh V, Gong Z, Low BC. RhoA acts downstream of Wnt5 and Wnt11 to regulate convergence and extension movements by involving effectors Rho kinase and Diaphanous: use of zebrafish as an in vivo model for GTPase signaling. Cell Signal. 2006;18:359–72. [PubMed]