|Home | About | Journals | Submit | Contact Us | Français|
Hedgehog signaling is critical in gastrointestinal patterning. Mice deficient in Hedgehog signaling exhibit abnormalities that mirror deformities seen in the human VACTERL (vertebral, anal, cardiac, tracheal, esophageal, renal, limb) association. However, the direction of Hedgehog signal flow is controversial and the cellular targets of Hedgehog signaling change with time during development. We profiled cellular Hedgehog response patterns from embryonic day 10.5 (E10.5) to adult in murine antrum, pyloric region, small intestine and colon.
Hedgehog signaling was profiled using Hedgehog pathway reporter mice and in situ hybridization. Cellular targets were identified by immunostaining. Ihh-overexpressing transgenic animals were generated and analyzed.
Hedgehog signaling is strictly paracrine from antrum to colon throughout embryonic and adult life. Novel findings include: mesothelial cells of the serosa transduce Hedgehog signals in fetal life; the hindgut epithelium expresses Ptch but not Gli1 at E10.5; the two layers of the muscularis externa respond differently to Hedgehog signals; organogenesis of the pyloric sphincter is associated with robust Hedgehog signaling; dramatically different Hedgehog responses characterize stomach and intestine at E16; after birth, the muscularis mucosa and villus smooth muscle (SM) consist primarily of Hedgehog responsive cells and Hh levels actively modulate villus core SM.
These studies reveal a previously unrecognized association of paracrine Hedgehog signaling with several gastrointestinal patterning events involving the serosa, pylorus and villus smooth muscle. The results may have implications for several human anomalies and could potentially expand the spectrum of the human VACTERL association.
Organogenesis of the gut relies on soluble signals that pass bi-directionally between endodermal and mesodermal layers (reviewed in 1). The Hedgehog (Hh) signaling pathway participates in this process at multiple sites along the developing gut2. Indeed, Hedgehog signaling is part of an ancient gut sculpting program, as components of this pathway in gut tissues of Drosophila3, Amphioxus4, leech5, sea urchin6, zebrafish7, Xenopus8, chicken9 and mouse10 11 coordinate morphogenic patterning events that are specific to each regional address along the anterior/posterior axis of the gut tube.
In vertebrates, Hh ligands include Sonic Hedgehog (Shh), Indian Hedgehog (Ihh) and Desert Hedgehog (Dhh). All three are expressed in the developing gut tube. Shh and Ihh are epithelially expressed and do not overlap with Dhh, which is expressed in Schwann cells, peripheral nerves and endotheial cells10. The three ligands bind to the receptors, Patched-1 (Ptch-1) and Patched-2 (Ptch-2). In the absence of ligand, the unoccupied Ptch receptor inhibits another membrane protein, Smoothened (Smo), deactivating the pathway. Hh ligand binding to Ptch relieves this repression, activating the pathway as measured by transcriptional modulation of target genes. The Gli transcription factors (Gli1, Gli2 and Gli3) represent the downstream effectors of Hh signaling in vertebrates (reviewed in 12) All three of these factors are expressed in the gastrointestinal tract13.
Significant GI pathology results from reduction of Hh ligand levels. Shh-/- or Ihh-/- mice exhibit malrotation of the GI tract, decreased muscularis propria and enteric neuron abnormalities14, 15. Other aspects of the phenotypes of these two ligand knockouts are distinct and include: esophageal atresia with tracheal esophageal fistula, gastric overgrowth and imperforate anus in Shh deficient animals; and Hirschprung's-like dilation of the colon as well as epithelial stem cell defects in Ihh null mice16. Reducing the combined (Shh+Ihh) Hh signal from the epithelium either by expression of a soluble form of the Hedgehog inhibitor protein, Hhip 17 or by injection of an anti-Hedgehog antibody18, results in a distinct phenotype that includes branched villi and vacuolated epithelium as well as disrupted mesenchymal patterning.
Loss of any of the Gli factors also has pathological consequences. Gli2 null animals exhibit malformations of the esophagus and hindgut while Gli3 deficient mice present with anal stenosis and overgrowth of the distal stomach, without apparent small intestinal phenotype 19. Gli1-/- mice show no apparent gut abnormalities 20, but a full complement of Gli1 activity is important in coping with inflammatory stress: a Gli1 variant in the human population (E1100Q) is implicated in inflammatory bowel disease and the Gli1+/- mouse is highly sensitive to chemically-induced colitis21.
Likewise, in humans, perturbed Hedgehog signaling is implicated in malformations of the GI tract. Pallister-Hall syndrome, which includes limb defects, hypothalamic hamartomas and imperforate anus, is due to a frameshift in the Gli3 protein22. In large part, however, the association of human deformities with the Hedgehog pathway has been based on the similarity of these malformations to those described in mouse Hh pathway mutants: the human VACTERL association (which includes vertebral, cardiac, tracheo-esophageal and/or anorectal malformations) mirrors foregut and/or hindgut phenotypes seen in Shh, Gli2 or Gli3 null mice14, 23.
Despite the importance of the Hedgehog signaling program to gut development and disease, conflicting reports exist as to the direction of Hh signals. Some studies support a strictly paracrine Hh signal (from epithelium to mesenchyme)10, 14, 17, while others suggest that the epithelium can also participate in an autocrine signaling program, especially during later development and adulthood24-26. Indeed, autocrine Hh signals have been proposed to mediate Paneth cell differentiation26, control colonic epithelial cell differentiation25 and promote epithelial regeneration in a setting of chronic inflammation24.
In this study, we mapped regional expression of Hh signaling molecules (Shh, Ihh, Ptch1, Gli1, Gli2, Gli3) throughout development in the antrum, small intestine and large intestine. The data reveal that the Hh signaling pathway is strictly paracrine at all times. The dynamic expression domains of Hh pathway activators suggest that in addition to previously assigned roles, Hh signaling might be involved in formation of the pylorus, patterning of antral epithelium, emergence of intestinal cell identity, and development of the serosal layer. Analysis of transgenic mice that overexpress Ihh revealed that Hh levels control SM populations in the cores of the villi.
Gli1+/LacZ, Gli2 +/LacZ, Ptch1+/LacZ mice have been described20, 27-29. Shh+/LacZ mice were used in a previous study30; their derivation will be described elsewhere (Gonzalez and Kottman, in preparation). Heterozygous mice were mated with C57Bl/6 mice and the morning of vaginal plug was counted as E0.5. Genotyping was performed as previously described13, 17, 31, 32. Protocols for X-gal staining, immunostaining, in situ hybridization, Q-RT-PCR and generation of Ihh transgenic mice are detailed in Supplementary Methods (www.gastrojournal.org).
Although Gli1, Gli2 and Ptch1 are all components of the Hh pathway, they reveal different aspects of the Hh signaling network. Gli1 is a direct target of Hh and its expression is dependent upon active Hh signaling27. Ptch1 is also a Hh target gene, but its transcription is not solely dependent upon Hh. Finally, Gli2 is an important mediator of activation27, but its expression is not transcriptionally regulated by Hh. Thus, Ptch1LacZ/+ and Gli2LacZ/+ mice indicate cells capable of responding to Hh signals while Gli1LacZ/+ mice reveal the cells that are actively responding. We used these reporters to map Hh pathway components in the developing and adult GI tract. In situ hybridization was used to confirm the reporter findings and examine additional Hh pathway molecules (Gli3, Ptch2).
Both Ihh and Shh are expressed robustly in the multilayered E10.5 endoderm throughout midgut and hindgut as documented previously10. Ptch1LacZ/+ is expressed in mesenchymal cells of the presumptive antrum (data not shown) and midgut (Figure 1A). However, on the dorsal side of the hindgut epithelium and post-anal portion of the tail gut, Ptch1LacZ/+ staining is clearly seen in the epithelium (Figure 1B,C). In contrast, Gli1LacZ/+ staining is strictly mesenchymal in all of these tissues (Figure D-F). Since Gli1 expression requires Hh ligands while Ptch1 may be expressed independently33, 34, we conclude that Hh signaling is paracrine in these tissues at this time. Gli3 is also highly expressed in the early intestinal mesenchyme and is progressively down-regulated during fetal development (Supplementary Information 1).
An opposing gradient of Shh and Ihh expression has been previously documented in E11 stomach10. We detected three different patterns of Hh expression in E14.5 stomach: in forestomach epithelium, Shh is expressed strongly while Ihh is undetectable; in distal stomach that will give rise to corpus and antrum, Ihh expression is strong, and overlaps with the weaker, but detectable expression of Shh (Figure 2A, B); finally, on the lesser curvature near the esophagus, a small region of epithelium expresses both Shh and Ihh at high levels (line, Figure 2A,B). Ptch1 expression is also high in this region (Figure 2C), as observed previously35. The lesser curvature of the stomach derives from the ventral gut tube36 and is clinically important since it is a common site for the development of gastric tumors37.
Ptch1, Gli1 and Gli2 are all exclusively mesenchymal at E14.5 in stomach and proximal duodenum (Figure 2C-H). The inner layer of the muscularis externa (ME) is positive for Gli1 (Figure 2D, F), while both layers are positive for Gli2 (Figure 2E,G). The signals for Ptch1 (Figure 2C, H), Gli1 (Figure 2D) and Gli2 (Figure 2E) are prominent surrounding the pyloric border. The forming pyloric sphincter is seen as a spur of Gli1, Gli2 and Ptch1 positive nuclei that extends inward from the inner circular layer of the ME. Interestingly, enteric neurons in the stomach and small intestine are unstained in all three reporter models (Supplementary Information 2).
As previously reported10, both Shh and Ihh are highly expressed in the intestinal epithelium (Figure 3A,B). Ptch1, Gli1 and Gli2, are expressed only in the mesenchyme (Figure 3C-G). Both in situ hybridization using a Ptch1 specific probe (Figure 3C) and X-gal stained sections from the Ptch1LacZ/+ reporter mice (Figure 3D) reveal that Ptch1 expression is high near the epithelium, falls off sharply in mesenchymal cells that lie 8-10 cell diameters away from the epithelium, and then rises again at the inner side of the ME (Figure 3C,D, asterisks). Gli1 is expressed throughout the loose mesenchyme, with particularly robust expression at the epithelial surface and at the inner layer of the ME (Figure 3E,F). Gli1 expression is also apparent in some cells of the serosal layer (arrowheads).
The pattern of Gli2 expression is similar to that of Gli1, but more cells are stained. In addition, both inner and outer muscle layers are strongly positive for Gli2 (Figure 3G), while Gli1 is expressed only in the inner muscular layer (Figure 3E, F). Most, but not all cells of the serosa are also positive for Gli2. Enteric neurons are strikingly negative (Figure 3G, arrow). Gli3 expression is much weaker at E14.5 than at E12.5, but remains detectable in the ME (Supplemental Information 1E,F). In E14.5 colon, staining patterns for Gli1, Gli2 and Ptch1 are entirely similar to those of the small intestine; no epithelial expression is detectable (data not shown).
As antral gland morphogenesis begins, both Shh and Ihh mRNA are detected throughout the epithelium of the antral villus-like structures (Figure 4A,B). Ptch1, Gli1 and Gli2 are restricted to the underlying mesenchyme (Figure 4C-F). Ptch1 is expressed between the invaginating epithelial folds, in the submucosa and in the innermost cells of the inner circular layer of the ME (Figure 4C). Gli1 is expressed similarly, but is additionally found within entire circular muscle as well as some cells of the serosa (Figure 4D). Gli2 is detected throughout the lamina propria and in both layers of the ME (Figure 4E). In cross sections of the pyloric border region, Gli1 expression and, by implication, active Hh signaling, is now much more prominent in the antrum and pyloric border region itself than in the adjacent duodenal tissue (Figure 4F). This dramatic difference in Hh signaling between antrum and duodenum was not present at E14.5.
Intestinal villus morphogenesis is initiated at approximately E15.5 and proceeds in an anterior to posterior wave. As villi grow, proliferating epithelial cells become concentrated at the intervillus base (Figure 5A). Shh expression resolves to encompass only the region occupied by the proliferative cells at the base of forming villi (Figure 5B); this differs from the pattern seen in antral epithelium (Figure 4A). In contrast, Ihh is expressed throughout the epithelium in newly formed villi; as villi lengthen, Ihh message is reduced slightly at the tips (Figure 5C).
Ptch2, Gli1, Ptch1 and Gli2 are expressed in the cores of growing villi (Figure 5D-I). Ptch2 (Figure 5D) and Gli3 (Supplementary Information 1H) are weakly expressed in mesenchyme. Gli1 is expressed in the circular SM of the ME (Figure 5E,H) while Gli2 is expressed in both externa layers (Figure 5I). Both Gli1 and Gli2 are expressed in some serosal cells. Enteric neurons of Auerbach's plexus are negative for Hh signal transduction components (Figure 5G-I, Supplementary Information 2). Colonic expression of all reporters at E16.5 is also mesenchymal and is similar to the pattern seen in small intestine (Supplementary Information 3).
At birth (P0), Ptch1, Gli1 and Gli2 are robustly expressed in mesenchymal cells, with prominent activity in antrum and pylorus. At postnatal day 10, a distinct muscularis mucosa, containing many PtchLacZ/+ and Gli1LacZ/+ stained cells also becomes clearly visible (Supplementary Information 4). By P0, Gli3 activity is undetectable in antral stomach (Supplementary Information 1I,J).
In newborn intestine, a layer of Gli1 positive cells lies just below the intervillus epithelium prior to crypt formation (Figure 6A, arrows). Ten days later, the intervillus epithelium has started to reshape to form crypt like structures. A nearly continuous layer of X-gal positive cells with elongated nuclei is seen immediately beneath the emerging crypts; this is the developing MM (Figure 6B, arrowhead). Some Hh responsive cells within the villus cores are connected with this layer; these cells express desmin and αSMA and represent the developing SM cores within the villi (Figure 6C-E). Budding crypts remain in close contact with a cuff of Hh responsive cells (Figure 6F-H). Enteric neurons are devoid of Gli1 and Gli2, but for the first time, a subset of these cells is positive for Ptch1 (Supplementary Information 2). Gli3 is undetectable after P0 (Supplementary Information 1K,L).
As in the fetus, analysis of adult mice revealed no evidence for autocrine Hh signaling in these portions of the gut tube (Supplementary Information 5-8). Though antral stomach was previously thought to be devoid of Hh signaling38, we found consistent evidence of Shh expression and Hh signal transduction in this tissue (Supplementary Information 5 and 6). In adult intestine (Figure 7) and colon (Supplementary Figure 7), Hh responsive cells included SM precursors, differentiated SM cells of the villus cores and MM, myofibroblasts and pericytes. Gli3 was not detectable in adult tissues, though high background in colon could have obscured weak expression (data not shown).
The studies above indicated that SM precursors and differentiated SMCs in the MM and villus cores transduce Hh signals, suggesting that hedgehog might be important in the development or maintenance of these populations. Indeed, our earlier studies of a mouse model of Hh inhibition indicated that reduced Hh signaling results in loss of differentiated SMC from villus cores17. To examine whether increased levels of Hh ligand would promote SM differentiation, we used the mouse villin promoter31 to generate transgenic mice that overexpress Ihh in small intestinal epithelium. Figure 7J shows that 12.4KVil-Ihh mice exhibit expanded smooth muscle in villus cores. Together, these findings indicate that Hh signals control the size of the smooth muscle population in the villus cores.
This analysis of Hh signal response in the developing and mature GI tract provides a cellular basis for Hh function in this tissue and suggests new avenues for exploration. Novel findings include: a) shortly after gut tube formation, epithelial cells of the hindgut and tailgut express Ptch1, but not Gli1; b) serosal cells respond to Hh signals during fetal life; c) the developing MM contains Hh responsive cells at P10 and continues to receive Hh signals during adult life; d) Hh levels control the amount of smooth muscle in villus cores; e) enteric neurons are not responsive to Hh signals in fetal life; they express Ptch1, but not Gli1 in neonatal and adult life; f) Hh responding cells are concentrated at the pyloric border during formation of that sphincter and are prominent in the antrum, and much less so in the intestine after E16.5; g) epithelial cells of the antrum, small intestine and colon do not express Gli1 and therefore do not respond to Hh signals at any point during embryonic, fetal or adult life. This last finding is in conflict with previous reports that propose autocrine epithelial hedgehog signaling25, 26 (discussed in Supplementary Information 6-8). Here, we further enlarge upon the potential implications of these data in the context of gastrointestinal development and disease.
It is interesting that transient expression of Ptch1, but not Gli1 is detectable in E10.5 hindgut and tailgut epithelium as well as in postnatal enteric neurons. Ptch1 expression in the absence of Gli1 has also been seen in neural tissue33, 34. Although Ptch1 is normally a Hh target gene, it can also act as a dependence receptor, to promote apoptosis in the absence of a Hh signal. It has been proposed that such apoptotic activity may help to shape the neural tube during development34. Perhaps transient expression of Ptch1 in these hindgut cells that clearly are not transducing Hh signals (since they are Gli1 negative) plays a role in preventing the apoptosis of these cells.
Hh signal transduction has not previously been demonstrated in the MM, but our data indicate that the cells of the forming MM are Hh responsive. Morphological studies indicate that the MM arises from the inner circular layer of ME39. Indeed, we detected apparent connections between the MM and cells of the inner circular layer of ME (Figure 6) and the SM cells of the villus core. It will be important to lineage trace this developing SM network to confirm these apparent origins. Our transgenic studies further reveal that villus core SM is highly sensitive to the level of Hh ligand. Though MM did not appear amplified, this might be due to the fact that the strength of the villin promoter is greater in villus tips than in crypts31. Alternatively, MM cells might be under separate control, even though these cells are also responsive to Hh signals as measured by their expression of Gli1.
Interestingly, from an early time point, the signaling properties of the inner circular and outer longitudinal muscles of the ME are different: only the former is Gli1 positive, indicative of active Hh signaling. Such different properties could potentially play a role in the differential response of these two muscles in disease. For example, familial type IV visceral myopathy presents with hypertophy of the inner circular and atrophy of the outer longitudinal muscle of the small bowel and is a rare cause of chronic intestinal pseudoobstruction40, 41. Given the ability of Hh signaling to modulate villus core smooth muscle as shown above and in our earlier studies17, it will be important to examine whether increased Hedgehog signaling plays any role in the etiology of this rare but usually fatal pathological condition.
The finding that the serosal mesothelium responds to Hh signals throughout development suggests that a source of Hh ligand may lie in the peritoneal cavity and/or that these cells receive Hh signals as they migrate onto the surface of the gut tube at E11.542. Recent studies indicate that mesothelial cells undergo epithelial to mesenchymal transition and migrate into the gut tube, differentiating into endothelial cells, vascular SM cells and pericytes42. Whether the vasculogenic activity of gut serosal mesothelium requires Hh signaling has not been directly tested, but it is noteworthy that Smo null embryos, which lack Hh response, have major vascular defects43 and administration of Shh blocking antibodies or a chemical inhibitor of Hh signaling produces vascular malformations and impaired vascular remodeling44, 45. Confirmation of the connection between Hedgehog signaling and the serosal mesothelium of the gut that is suggested here could potentially have therapeutic implications in the context of a rare developmental anomaly called “apple peel bowel”, a syndrome of intestinal wasting that is associated with loss of the serosa and its associated blood vessels46, 47.
Our data also suggest a possible indirect role for Hh in the establishment of intestinal vs. stomach epithelial identity. Exactly at E16.5, the precise time when a clear-cut epithelial boundary is being generated between stomach and intestine32, a dramatic difference in Hh signaling is generated in the mesenchyme: Hh signal transduction is robust in the antrum and much less prominent in small intestine. This difference is not visible at E14.5. This finding fits well with previous functional studies in Xenopus, where Hh signaling is also downregulated during differentiation of the intestine48. Indeed, constitutive activation of Hh signaling in the midgut results in arrested cytodifferentiation and poor growth of that tissue49. Thus, there may be an evolutionarily conserved requirement for downregulation of Hh signaling to permit intestinal cell differentiation.
We show here that in contrast to earlier conclusions50, Hh signal transduction is quite active in the antrum. Though its function in this adult tissue is not clear, a role for Hh in patterning the perinatal antral epithelium was previously suggested by the antral overgrowth phenotype seen in Gli3 and Shh null mice14, 19. Since signal transduction in the antrum is exclusively paracrine, this effect must represent an altered relay from epithelium to mesenchyme and back to epithelium. In this context, it is of interest to consider a common developmental abnormality of the pyloric stomach known as hypertrophic pyloric stenosis (HPS). In a study of over 100 infants with HPS, Hernanz-Schulman et al. found that robust overgrowth of the antral mucosa was responsible for obstruction of the pyloric opening51. Interestingly, a survey of the literature shows that HPS is often associated with other anomalies51, 52, including: abdominal malrotation or volvulus53, 54, cardiac anomalies55, imperforate anus, tracheal esophageal fistula56-58 and hydronephrosis of the kidney52, 59. Each of these malformations is seen in the VACTERL association, a constellation of abnormalities involving vertebral, anal, tracheal, esophageal, renal and limb development that are thought to be linked to defects in Hh signaling23. Since both Shh null mice and humans with HPS exhibit antral epithelial overgrowth, HPS might represent yet another associated abnormality within the VACTERL spectrum.
Grant support: NIH P01 DK62041, NIH R01 DK065850, T32-HD007505, T32-HL07622
The authors acknowledge support from NIH R01 DK065850 (DG), NIH P01 DK62041 (DG, JM, AD); the Organogenesis Training Program, T32-HD007505 (WZ and AG and ÅK); and the Hematology Training Program, T32-HL07622 (KW). Excellent technical support was provided by the Organogenesis Morphology Core and the Microscopy and Image Analysis Laboratory.
Financial Disclosures: Nothing to disclose
Publisher's Disclaimer: 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.