|Home | About | Journals | Submit | Contact Us | Français|
Intestinal stem cells (ISCs) are regulated by the mesenchymal environment via physical interaction and diffusible factors. We examined the role of Indian hedgehog (Ihh) in mesenchymal organization and the mechanisms by which perturbations in epithelial–mesenchymal interactions affect ISC fate.
We generated mice with intestinal epithelial-specific disruption of Ihh. Gross and microscopic anatomical changes were determined using histologic, immunohistochemical, and in situ hybridization analyses. Molecular mechanisms were elucidated by expression profiling and in vitro analyses.
Deletion of intestinal epithelial Ihh disrupted the intestinal mesenchymal architecture, demonstrated by loss of the muscularis mucosae, deterioration of the extracellular matrix, and reductions in numbers of crypt myofibroblasts. Concurrently, the epithelial compartment had increased Wnt signaling, disturbed crypt polarity and architecture, defective enterocyte differentiation, and increased and ectopic proliferation that was accompanied by increased numbers of ISC. Mechanistic studies revealed that Hh inhibition deregulates bone morphogenetic protein signaling, increases matrix metalloproteinase levels, and disrupts extracellular matrix proteins, fostering a proliferative environment for ISCs and progenitor cells.
Ihh regulates ISC self-renewal and differentiation. Intestinal epithelial Ihh signals to the mesenchymal compartment to regulate formation and proliferation of mesenchymal cells, which in turn affect epithelial proliferation and differentiation. These findings provide a basis for analyses of the role of the muscularis mucosae in ISC regulation.
In the intestine, epithelial cells undergo repeated progenitor cell proliferation, terminal differentiation and cell death, a process that requires intestinal epithelial stem cells (ISCs) to engage in a continuous dialogue with neighboring epithelial and mesenchymal cells.1, 2 Recent cell lineage tracing experiments have identified Lgr5 and Bmi1 as ISC markers, although they appear to represent two distinct ISC populations.3, 4 The regenerative capacity of ISCs is directed by structural and biochemical cues received from the ISC microenvironment.5 This microenvironment is a complex structure that modulates intestinal homeostasis by maintaining a fine balance between ISC self-renewal and downstream differentiation. While multiple cell types, including endothelial cells, lymphocytes, and muscle cells may contribute to ISC regulation, the cell type generally considered the most important to ISC regulation are intestinal subepithelial myofibroblasts (ISEMFs) due to their close proximity to ISCs.6 These mesenchymal cells secrete various factors that favor or restrict ISC self-renewal including cytokines, matrix proteins, and growth factors such as BMP antagonists Noggin and Gremlin.7, 8 Yet how ISEMFs are regulated within the ISC microenvironment, their precise role in fostering ISC self-renewal and proliferation, and whether they are the only major contributors to the mesenchymal ISC microenvironment remain unclear.
The Hedgehog (Hh) signaling pathway plays a critical role during gut development.9 Expression of the Hh ligands, Sonic Hedgehog (Shh) and Indian Hedgehog (Ihh) has been detected exclusively in the intestinal epithelium, while expression of Hh target genes, Patched (Ptch1) and Gli1 has been observed in the mesenchyme, including the villus core, muscularis mucosae and pericryptal myofibroblasts.10, 11 Both Shh- and Ihh- null mice display marked gastrointestinal abnormalities including attenuated smooth muscle layers and intestinal malrotation.10 In mice, overexpression of Hedgehog interacting protein (Hhip), a negative regulator of Hh signaling, in the gut leads to the mislocalization of ISEMFs and expansion of immature smooth muscle cells (SMCs).12 Furthermore, these mice showed increased cell proliferation and aberrant crypt-like structures, as well as enhanced Wnt activity. In converse experiments, enhanced Hh signaling due to the conditional deletion of Ptch1 resulted in accrual of colonic myofibroblasts and colonic crypt hypoplasia.13
Despite these recent advances, the exact role of Ihh in ISC regulation and gut development still remains unclear. Here we find that loss of Ihh signaling causes profound morphological changes to the intestinal mesenchymal compartment with the most consistent and severe changes occurring near crypt bottoms where ISCs are located. Our data suggest that the muscularis mucosae might be a novel component of the mesenchymal microenvironment, which acts with myofibroblasts to limit the size of the crypt and ISC pool. Altogether, the results support that Ihh functions as a critical regulator of ISC self-renewal through epithelial-mesenchymal interactions.
The Ihhflox/flox mice were provided by Dr. Beate Lanske of Harvard University.14 Villin-Cre mice15 and Smoflox/flox mice16 were obtained from the Jackson Laboratory. Ihhflox/flox and Villin-Cre mice were mated and the offspring were backcrossed to generate Villin-Cre;Ihhflox/flox mice. Genotyping was performed by polymerase chain reaction on genomic DNA from tail clips as previously described.14, 15 The Villin-Cre:Ihhflox/flox pups suffer early lethality and were sacrificed when they displayed lethargy and inability to feed. All mice were housed, fed, and treated in accordance with protocols approved by the committee for animal research at the University of California, San Francisco.
Animals were euthanized and their intestine was removed and flushed with PBS. Fixed tissue was embedded in paraffin. Immunostaining was performed using standard protocols after heat-mediated antigen retrieval. Olfm4 in situ hybridization was performed as previously described.17
Student's t-test was used to evaluate statistical significance. Values of P < 0.05 were considered significant.
A partial description of Ihh's role in intestinal development has been provided by two studies of Ihh-/- mice: one study described a loss of proliferative epithelial cells in Ihh-/- mice,10 while the second study noted an expansion of proliferative epithelial cells in Ihh-/- mice.18 However, Ihh-/- mice die at birth, precluding any analysis of Ihh's role during postnatal intestinal development. To examine the full role of Ihh in intestinal development including the critical postnatal period when crypt structures are established, we generated a conditional Ihh-deficient mouse line by breeding Ihhflox/flox mice to Villin-Cre mice. This cross generated Villin-Cre;Ihhflox/flox mice that had intestinal Ihh mRNA expression levels that were 4% of control mice, and an absence of Ihh at the protein level (Supplemental Figure 1A and 1B). The mRNA expression level of Ptch1 and Gli1, two direct transcriptional targets of Hh signals were 19% and 9% of the control mice, respectively (Supplemental Figure 1A), suggesting that Ihh is the key Hh molecule mediating Hh signaling in intestinal tissues, and Shh or Desert hedgehog cannot replace its function.
Villin-Cre;Ihhflox/flox mice were born alive and appeared normal. However, at P3 it was apparent that the mutant mice were not thriving as they were noticeably smaller than their control littermates (detailed gross phenotypes are described in Supplemental Figure 1). The majority of Villin-Cre;Ihhflox/flox mice died between P7 and P10, although some managed to survive until P30. The early lethality of Villin-Cre;Ihhflox/flox mice is likely caused by malnourishment (Supplemental Figure 1C and 1D).
It has previously been reported that Ihh signals in a paracrine direction, moving from its origin in intestinal epithelial cells towards Hh signaling effectors in mesenchymal cells.11 We investigated the myofibroblast and smooth muscle changes by α-smooth muscle actin (α-SMA) immunostaining. The most striking and consistent change observed in the Villin-Cre;Ihhflox/flox mice was the loss of a horizontal layer of α-SMA-positive cells at the crypt base in the small intestine and colon that corresponds to the muscularis mucosae (Figure 1A and 1B; Supplemental Figure 2). The complete loss of muscularis mucosae cells was observed soon after birth and persisted throughout development (Supplemental Figure 2). Since the cellular components of the muscularis mucosae in mice have not been well-characterized, we performed double-labeling experiments for α-SMA and desmin, as well as α-SMA and vimentin. Myofibroblasts can be distinguished from fibroblasts by their expression of α-SMA, and separated from SMCs by their expression of vimentin and lack of desmin expression. The staining results revealed that in control mice, the cells comprising the thin layer below small intestinal crypts expressed α-SMA and vimentin, but not desmin (Figure 1A; Supplemental Figure 2). The results indicate the muscularis mucosae layer absent from the small intestine of Villin-Cre;Ihhflox/flox mice is predominantly composed of myofibroblasts. In control colon, α-SMA and desmin were co-expressed along the muscularis mucosae underneath colonic crypts (Figure 1B). Noticeably, this cell layer does not express the fibroblast marker vimentin (Figure 1B), demonstrating the missing muscularis mucosae layer in the colon of Villin-Cre;Ihhflox/flox mice is composed of SMCs (Figure 1B; Supplemental Figure 2).
Since ISEMFs, especially pericryptal myofibroblasts have traditionally been considered the key mesenchymal cell type that regulates ISCs, we examined the composition of these cells in mutant mice. In the colon, we consistently observed diminished numbers of pericryptal myofibroblasts at the crypt base (Figure 1B; Supplemental Figure 2). In the small intestine, loss of pericryptal myofibroblasts was more variable. Additionally, the mislocalization of pericryptal myofibroblasts as described in Villin-Hhip mice12 was not observed in Villin-Cre;Ihhflox/flox mice (Figure 1A; Supplemental Figure 2). Examination of villus core mesenchymal cells revealed a consistent reduction in desmin-positive smooth muscle cells and central core myofibroblasts (Supplemental Figure 2). However, some villi exhibited an expansion of myofibroblasts in regions with superimposed inflammation, suggesting myofibroblast numbers may also be affected by secondary changes. Notably, in P5 control mice, we detected active proliferation of mesenchymal cells surrounding newly forming crypts. In mutant mice, however, mesenchymal cell proliferative activity was profoundly attenuated at this stage (Figure 1C). These data suggest that Ihh loss might impair ISEMF proliferation. To test this hypothesis, we examined in vitro ISEMF cell growth in the presence of GANT61, a small-molecule antagonist of GLI-mediated transcription.19 We found GANT61 significantly inhibits ISEMF cell viability and proliferation (Figure 1D and 1E). Thus, these in vivo and in vitro studies support that Hh signaling is required for ISEMF growth.
Altogether our data indicate that deletion of Ihh results in the disappearance of the muscularis mucosae and fewer ISEMFs surrounding the crypt base, disrupting the key mesenchymal cells that surround ISCs.
As we detected marked differences in the mesenchymal compartment surrounding ISCs between control and mutant mice, we sought to determine whether these differences were accompanied by changes in the intestinal epithelium. We found that elimination of Ihh resulted in crypts that appeared wider, loosely organized, and crowded with nuclei (Figure 2A). Villus branching accompanied by aberrant crypt-like structures was also detected in the small intestine of Villin-Cre;Ihhflox/flox mice (Figure 2A). The ectopic crypt-like structures contained cells that were positive for the proliferation marker Ki67 (Figure 2A). Furthermore, proliferating cells in the mutants greatly outnumbered those in control littermates with their distribution extending beyond the normal confines of the crypt (Figure 2A). Crypt fission is a process in which new crypts are produced and is believed to occur in response to stem cell expansion. While crypt fission normally occurs in neonatal mice, we observed a significantly higher incidence of crypt fission in the jejunum of P9 Villin-Cre;Ihhflox/flox mice compared to control, revealing major proliferation abnormalities (Figure 2B). In P30 Villin-Cre;Ihhflox/flox jejunum, we observed marked elongation of crypts with florid proliferation (Figure 2C). Additionally, in some areas, we saw a loss of epithelial maturation, characterized by the absence of villus architecture and proliferative cells reaching the luminal surface (Figure 2C). A similar epithelial phenotype was noted in the colon of Villin-Cre;Ihhflox/flox mice in which crypts were dilated with frequent branching and had disturbed orientation and a high degree of proliferation (Supplemental Figure 3). Furthermore, in one P30 Villin-Cre;Ihhflox/flox colon, a lesion mimicking a small adenoma with mild dysplasia was detected in the midst of disorganized and disoriented crypts (Figure 2C). These observations suggest that loss of Ihh expression induces florid proliferative events. To determine whether this may eventually culminate in neoplastic transformation would require a study of aged Ihh mutants; however, the early death of Villin-Cre;Ihhflox/flox mice makes such a study infeasible in our setting.
It is known that a small amount of mosaic Cre expression exists in the colon of Villin-Cre transgenic mice.15 In cases in which Villin-Cre;Ihhflox/flox mice survived beyond P15, we noticed normal crypt structures adjacent to dilated crypts in the colon (Figure 2D). Interestingly, beneath the normal crypt structures α-SMA staining detected an intact muscularis mucosae (Figure 2D), whereas below neighboring dilated crypts, no α-SMA staining was detected, revealing an absence of the muscularis mucosae (Figure 2D). Furthermore, Ki67-positive cells were restricted to crypt bottoms in regions where the muscularis mucosae was present (Figure 2D); however, in the absence of the muscularis mucosae, differentiated cells occupied crypt bottoms and proliferative cells were found at crypt bottoms and tops (Figure 2D). These results suggest the muscularis mucosae might influence crypt epithelial fate and polarity, and contribute to ISC regulation.
Given that we observed several manifestations in the Villin-Cre;Ihhflox/flox mice that were comparable to those seen in mice with increased Wnt signaling, including enhanced epithelial cell proliferation, branched villi, and enlarged crypts, we sought to analyze whether mutant mice displayed increased Wnt activity.20, 21 Typically, the expression of Wnt/β-catenin target genes (e.g., Cd44, Sox9, EphB2) is restricted to the crypt proliferative compartment (Figure 3A; Supplemental Figure 4). However, in mutant mice Cd44, Sox9, EphB2 expression was highly expressed throughout the crypt and along the villus in the small intestine as well as along the entire crypt length in the colon (Figure 3A; Supplemental Figure 4). Furthermore, staining for β-catenin in mutant mice showed increased cytoplasmic staining in crypts and villi, providing additional evidence that the mutants have increased Wnt activity (Supplemental Figure 4).
We next addressed whether the loss of Ihh affected the ISC population by performing in situ hybridization for Olfm4, a marker for small intestinal stem cells.17 Villin-Cre;Ihhflox/flox mice showed an increase in expression of Olfm4 as well as an increase in the number of Olfm4+ cells per crypt compared to control mice (Figure 3B). A previous study of Ihh-/- mice suggested that a complete loss of Ihh diminishes the number of ISCs, whereas a second study suggested the opposite.10, 18 However, both studies lacked a definitive ISC marker and based their findings solely on a cell proliferation marker. Thus, by utilizing a specific ISC marker, we provide compelling evidence that deactivation of Ihh leads to ISC expansion.
Next, we investigated whether the expansion of epithelial proliferation occurred to the detriment of intestinal epithelial differentiation in Villin-Cre;Ihhflox/flox mice. We examined the differentiation pattern of the four different intestinal cell types (enterocytes, goblet, enteroendocrine, and Paneth cells) using specific markers for each cell lineage, as well as electron microscopy (Supplemental Figures 5-7). Our analyses demonstrated that there is an expansion of the secretory cell lineages at the expense of the enterocyte differentiation program in Villin-Cre;Ihhflox/flox mice. Furthermore, we detected an accumulation of vacuolated cells at crypt-villus junctions in the small intestine, indicating the Ihh mutants have some maturation arrest at the enterocyte precursor stage (Supplemental Figure 6).
A recent report22 showed that chronic Hh inhibition in adult mice results in villus atrophy and profound inflammatory responses in the intestine. We noted patchy and variable inflammatory responses in the small intestine and colon of Villin-Cre;Ihhflox/flox mice. Early inflammatory changes were observed in regions lined by vacuolated cells at crypt-villus junctions, where breach of epithelial integrity was accompanied by neutrophil infiltration. In acutely inflamed regions, we noticed villus sloughing, resulting in villus loss, mucosal ulceration, dense neutrophil infiltration, and fibrosis (Supplemental Figure 8). Similar surface ulceration with neutrophil infiltration was also noted in the colon (Supplemental Figure 8).
While some studies suggest Hh signals in an autocrine direction, acting directly on intestinal epithelial cells,18, 23 more recent reports indicate that Hh signals predominantly in a paracrine fashion in the intestine.11, 24 To exclude the possibility that the phenotypes we observed in Villin-Cre;Ihhflox/flox mice were due to autocrine Hh signaling, we deleted the required hedgehog receptor Smoothened (Smo) in intestinal epithelial cells by crossing Villin-Cre mice with Smoflox/flox mice to generate Villin-Cre;Smoflox/flox mice. We found that Villin-Cre;Smoflox/flox mice were born at normal frequencies, and were healthy with no gross abnormalities up to 15 months of age. Microscopic examination revealed normal intestinal and colonic architecture (Supplemental Figure 9A). All four epithelial cell lineages were well developed, and normal Wnt signaling was observed (Supplemental Figure 9B). The results provide strong genetic evidence that Hh signaling functions strictly in a paracrine manner during gut morphogenesis.
To investigate the molecular mechanisms underlying the disruption of the mesenchymal compartment and how it leads to abnormal ISC proliferation, we performed expression array analysis of colon samples from control and Villin-Cre;Ihhflox/flox mice. Statistical analysis identified 508 transcripts, including 298 named genes up-regulated and 532 transcripts, including 429 named genes down-regulated in Villin-Cre;Ihhflox/flox mice (Supplemental Table 1). As expected, all Hh signaling targets, such as Gli1, Ptch, and Hhip, were significantly down-regulated in colon samples from mutant mice (Figure 4A).
Among the genes that were up-regulated in mutants were Wnt targets, including c-Myc, Sox-9, and MMP7, as well as the ISC marker Lgr5 (Figure 4B). Furthermore, comparison of genes up-regulated in Villin-Cre;Ihhflox/flox mice with Lgr5 stem cell genes identified several genes that overlap, including Lgr5, Acot1, Adora1, Sox9, Soat1, and Slc12a2 (Figure 4B).25 Genes involved in gut hormones (CCK and glucagon), reflecting changes in enteroendocrine cells as well as goblet cell marker genes (Spdef, Spink4, Muc2, Gcnt3, and Foxa3) were up-regulated in Villin-Cre;Ihhflox/flox mice (Figure 4C), which is consistent with the expansion of secretory cell lineages observed in the mutants. Intriguingly, several MMPs (MMP3, MMP7, MMP8, and MMP10) which are known to degrade ECM proteins and connective tissues were up-regulated in mutant colon samples (Figure 4D).
Interestingly, the most prominent genes down-regulated in the mutants encode proteins that help support and maintain the intestinal epithelium (Figure 4D and 4E). For example, genes that may provide structural support to ISCs such as genes involved in smooth muscle development (Myh10, Myh11, myocardin, Mef2c, desmin, etc.) were down-regulated in Ihh mutant colons (Figure 4E). Furthermore, genes encoding ECM proteins (multiple isoforms of collagen and laminin; as well as fibronectin, osteoglycin, versican, nidogen 1, Ecm2, etc.) which provide support, organization, and mechanical signals to the epithelium were extensively down-regulated in Villin-Cre;Ihhflox/flox mice (Figure 4D). Additionally, several integrins (Itga1, Itga8, Itga9, Itgav, Itgb6) which attach epithelial cells to the ECM and mediate epithelial cell-matrix interactions were down-regulated in mutant mice (Figure 4D). Overall, the expression analysis demonstrates that the gut mesenchymal compartment is compromised when Ihh is deleted in the intestine, suggesting the possibility that disruption of the mesenchymal cells surrounding ISCs may be the key mechanism that leads to abnormal ISC expansion in Villin-Cre;Ihhflox/flox mice.
We next analyzed the signaling pathways that have been implicated in gut development including Notch, BMP, and Ras/MAPK pathway genes. We found that the BMP pathway was one of the major targets for Ihh signaling during gut morphogenesis (Figure 4F). For example, BMPs including BMP2, BMP4, and BMP5 were all down-regulated. BMP antagonists showed a more complicated pattern of expression: some were up-regulated, such as Gremlin1 and Chordin-like-2, while others were down-regulated, such as Gremlin2 and Twsg1. Nevertheless, analysis of transcriptional targets of BMP signaling, including ID1, ID2, and ID4, revealed that all these genes were down-regulated in the mutant mice. Furthermore, immunostaining of phosphorylated Smad1/5 (p-Smad1/5) showed decreased expression of p-Smad1/5 in the epithelial cells of mutant mice (Figure 5), confirming an overall decrease in BMP signaling in Villin-Cre;Ihhflox/flox mice.
Given that the gene expression profiling data suggested that the smooth muscle character and ECM components undergo marked changes in Villin-Cre;Ihhflox/flox mice, we decided to investigate how the loss of Hh signaling directly affects expression of muscle and ECM-related genes in ISEMFs. Quantitative RT-PCR performed on ISEMFs treated with the Gli transcription inhibitor GANT61 found a significant down-regulation of Mef2c and Myocd, two transcription factors known to regulate smooth muscle development (Figure 6A). These results are consistent with the microarray analysis, suggesting cultured ISEMFs closely resemble the in vivo situation. Furthermore, these findings imply that Hh signals may promote the differentiation of fibroblasts towards a myofibroblast phenotype in the intestine. We next examined the ability of ISEMFs to produce matrix proteins in the absence of Hh signals. Expression of laminin4 was significantly down-regulated in GANT61-treated ISEMFs (Figure 6A), while other ECM genes such as fibronectin and collagen showed no significant change (data not shown). Strikingly, MMPs, including MMP3 and MMP10 were significantly up-regulated in GANT61-treated ISEMFs (Figure 6A). These data suggest that loss of Hh signaling amplifies the expression of MMPs in intestinal stromal cells, potentially leading to the degradation of ECM components.
These studies indicate that ECM components may be significantly compromised in Villin-Cre;Ihhflox/flox mice. In normal colon, collagen IV expression was detected throughout the lamina propria. In mutant mice, collagen IV staining was significantly reduced in the colon (Figure 6B). Another basement membrane protein, laminin, was expressed in the lamina propria with intense staining at the epithelial-mesenchymal interface corresponding to the basement membrane in control colon. In contrast, laminin staining appeared diffusely weak and completely absent from the basement membrane in the mutant mice (Figure 6B). Thus, inactivation of Ihh in the gut leads to the down-regulation of ECM genes and matrix protein degradation, resulting in a weakened ECM that is vulnerable to crypt expansion.
The epithelial phenotypes we observed in the Villin-Cre;Ihhflox/flox mice are overall consistent with other mouse models that disrupt Hh signaling during gut morphogenesis, such as mice overexpressing the Hh inhibitor Hhip, or mice with a conditional deletion of Ptch1.12, 13 However, none of the previous studies addressed the critical question of whether paracrine Hh signaling affects ISC self-renewal or whether the epithelial phenotypes are due to the disruption of trans-amplifying/progenitor cells near the crypt base. Our study showed that in Villin-Cre;Ihhflox/flox mice there is a clear expansion of the ISC compartment, providing solid evidence that Ihh regulates ISCs during gut development. Furthermore, our findings indicate that Ihh is the key Hh ligand mediating the observed epithelial phenotypes in the small intestine and colon after birth. Additionally, our studies with conditional epithelial Smo knockout mice demonstrated that Ihh regulates ISC fates strictly in a paracrine fashion. This statement is corroborated by the fact that there is a profound and consistent disruption of the mesenchymal compartment, especially at the crypt base surrounding ISCs in Villin-Cre;Ihhflox/flox mice. Altogether these data suggest that ablation of Ihh leads to the significant deterioration of the microenvironment surrounding ISCs, which in turn leads to the expansion of ISCs and altered epithelial cell differentiation programs. In addition, previous studies have suggested that the mesenchymal cells surrounding ISCs are the major source of Wnt and function primarily to maintain ISC proliferation. Here, we propose a different role of the ISC microenvironment, which is to restrain crypt size and prevent abnormal stimulation. Thus, a delicate balance between the proliferative and restrictive activity by mesenchymal cells surrouding ISCs likely exists to refine the shape, size, and function of the gut epithelium to form proper crypt-villus structures.
Pericryptal myofibroblasts are generally considered the key mesenchymal cells that regulate ISCs. On the other hand, cells within the muscularis mucosae layer, despite its vicinity to the ISCs, are not known to be involved in regulating ISCs. In our current study, we find the development of muscularis mucosae cells is strictly dependent on epithelial Ihh signaling, as deletion of Ihh leads to total ablation of the muscularis mucosae in both the colon and small intestine. Our studies suggest that loss of the muscularis mucosae layer may contribute to ISC expansion and deregulation of intestinal epithelial cell differentiation. These findings provide a basis for further investigation of the role of the muscularis mucosae in ISC regulation. For example, in vitro co-culture of muscularis mucosae cells with Lgr5+ ISCs will provide additional evidence of how the muscularis mucosae modulates ISC self-renewal and expansion. In addition, using genomic approaches, one could identify secretory factors produced by the muscularis mucosae. The functions of these secretory factors in regulating ISCs should then be further explored using either knockout mouse models or through in vitro analysis of Lgr5+ ISCs.
What are the molecular mechanisms behind the expansion of the ISC compartment upon Ihh loss during gut morphogenesis? Our genomic analyses indicate that Ihh likely regulates ISC self-renewal and cell fate determination via multiple mechanisms. The reduction of BMP signaling upon Ihh deletion could constitute one possible mechanism that leads to the described phenotype, as BMP signaling normally acts to inhibit ISC self-renewal and repress crypt formation in the gut.7, 26 Nevertheless, the reduced BMP signaling does not account for all the phenotypes observed in Villin-Cre;Ihhflox/flox mice. For example, in villin-noggin mice in which BMP signaling is completely abrogated, no morphological alternations are detected till 4-weeks of age.26 Other factors that likely contribute to the severe phenotypes seen in mutant mice are the complete loss of muscularis mucosae cells and the disruption of the ECM that surrounds ISCs. One can imagine muscularis mucosae cells likely provide solid structural support for ISCs at the crypt base. It is also likely that muscularis mucosae cells secrete additional factors that maintain proper ISC number and crypt structure such as the Wnt antagonist Sfrp2, a down-regulated gene in Villin-Cre;Ihhflox/flox mice identified by our microarray analysis. Additionally, our microarray analysis showed a profound loss of ECM gene expression in Villin-Cre;Ihhflox/flox mice at the RNA level. The ECM components were further impaired by the up-regulation of MMPs, the major enzymes that degrade ECM proteins. Altogether, these mesenchymal compartment changes provide a pro-growth microenvironment for ISCs, promoting ISC expansion and subsequent expansion of the transit-amplifying compartment.
These findings imply a very interesting possibility that Ihh deletion, resulting in the loss of structural ECM integrity, the loss of the muscularis mucosae, and expansion of ISCs may predispose to neoplastic transformation. This is in contrast to many other organs where abnormal activation of the Hh pathway promotes tumor development.27, 28 Indeed, the role of the Hh pathway in colorectal cancer remains controversial with numerous conflicting reports.18, 24, 29-31 Clearly, the precise role of Hh signaling in colorectal tumorigenesis needs to be further clarified.
In a recent study, Zacharias et al reported that chronic Hh inhibition in adult mice resulted in villus atrophy and profound inflammatory responses in the intestine, resembling human celiac disease.22 These data suggest that Hh signaling acts as an important anti-inflammatory factor in the gut. In our study, we also noted the presence of acute inflammation in the intestine of Villin-Cre;Ihhflox/flox mice, but mostly in a patchy manner and centered around regions lined by vacuolated cells at the crypt-villus junction in the small intestine and crypt-surface in the colon. We speculate that in the absence of generalized inflammation, the acute inflammatory changes in Villin-Cre;Ihhflox/flox mice most likely results from a weakened mucosal barrier caused by defective enterocyte differentiation. It is possible that the inflammatory changes may also be aggravated by proinflammatory responses released by stromal cells in the absence of Ihh signaling. Our microarray analysis also demonstrated an increased acute inflammatory response in Villin-Cre;Ihhflox/flox mice. For example, surface markers expressed by neutrophils and/or macrophages such as CD11β/ITGAM, CD14 and CD177, and proinflammatory cytokines or chemokines, such as IL1β, CCL6, CCL9, CCL25 and CXCL5 were all up-regulated in Villin-Cre;Ihhflox/flox mice (Supplemental Table 1). Many of these genes overlap with the genes identified by Zacharias et al. However, we detect no resemblance of the inflammatory process in Villin-Cre;Ihhflox/flox mice to human chronic inflammatory bowel diseases. Specifically, lymphocytic infiltration which is characteristically seen in celiac disease or Crohn's disease, and significant cryptitis or crypt abscess formation which is typically seen in ulcerative colitis are all absent in Villin-Cre;Ihhflox/flox mice. Altogether, the analysis suggests that Hh signaling may play distinct roles in regulating inflammatory responses during gut development versus homeostasis.
We thank Dr. Beate Lanske of Harvard University for providing us with the Ihhflox/flox mice; and Sandra Huling of the UCSF Liver Center for histology and electron microscopy support.
Grant Support: This work is supported in part by NIH grants R21DK069309 and R01CA136606 to X.C; R01CA127229 and R01DK55783 to D.W.P; a grant from the Research Grants Council of the Hong Kong Special Administrative Region (Project No. HKU7524/06M) to S.Y.L. and X.C; a grant from Strategic Research Theme on Cancer from The University of Hong Kong to S.Y.L. C.K. is supported by a stem cell fellowship provided by the California Institute for Regenerative Medicine; and D.E.S. is supported by Deutsche Forschungsgemeinschaft (DFG, Sta1065/1-1). The funding agencies have no role in data collection, analysis, and interpretation.
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.
Disclosure: No conflict of interest exists for any author.
Transcript Profiling: The raw data of microarray experiment are available at Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/projects/geo) under the accession number of GSE18393.