Regeneration of the amphibian intestinal epithelium under the control of stem cell niche




The epithelium of the mammalian digestive tract originates from stem cells and undergoes rapid cell-renewal throughout adulthood. It has been proposed that the microenvironment around the stem cells, called ‘niche’, plays an important role in epithelial cell-renewal through cell-cell and cell-extracellular matrix interactions. The amphibian intestine, which establishes epithelial cell-renewal during metamorphosis, serves as a unique and good model for studying molecular mechanisms of the stem cell niche. By using the organ culture of the Xenopus laevis intestine, we have previously shown that larval-to-adult epithelial remodeling can be organ-autonomously induced by thyroid hormone (TH) and needs interactions with the surrounding connective tissue. Thus, in this animal model, the functional analysis of TH response genes is useful for clarifying the epithelial-connective tissue interactions essential for intestinal remodeling at the molecular level. Recent progress in culture and transgenic technology now enables us to investigate functions of such TH response genes in the X. laevis intestine and sheds light on molecular aspects of the stem cell niche that are common to the mammalian intestine.


All of the epithelial cells in the mammalian small intestine originate from multipotent stem cells, which are localized near the base of the crypt. Descendants of the stem cells actively proliferate and differentiate into major absorptive cells, goblet cells, enteroendocrine cells and Paneth cells along the crypt-villus axis (Bjerknes & Cheng 1981). Finally, all kinds of cells, except for Paneth cells, undergo apoptosis at the tip of the villi (Fig. 1). It is generally known that the microenvironment around the stem cells, known as ‘niche’, plays a key role in this epithelial cell-renewal (Potten et al. 1997; Mills & Gordon 2001). Morphologically, several types of cells have been proposed as ‘niche players’ which affect the stem cells and their descendants, for example, subepithelial myofibroblasts (Powell et al. 1999), endothelial cells of the microvasculature (Paris et al. 2001), intraepithelial lymphocytes (Komano et al. 1995) and enteric neurons that send their projections near the crypt epithelium (Bjerknes & Cheng 2001). Recently, signaling pathways such as the Wnt and Notch pathways have been shown to be involved in the epithelial cell-renewal of the mammalian intestine. The Wnt signaling pathway controls proliferation, differentiation and migration of epithelial cells through β-catenin and T-cell factor-4 (Soubeyran et al. 2001; Battle 2002; van de Wetering et al. 2002), while the Notch signaling pathway decides cell fate to be either absorptive cells or secretory cells, which include goblet cells, enteroendocrine cells and Paneth cells, through Hes-1 and Math-1 (van den Brink et al. 2001). However, it still remains largely unknown how these and other signaling pathways correlate with the niche players and how the stem cell niche develops (Fuchs et al. 2004).

Figure 1.

Comparison between the amphibian intestine from larval to adult form and the mammalian intestine. During amphibian metamorphosis, the simple larval intestine is remodeled into the more complex one, similar to the mammalian intestine. The brush border of the amphibian larval intestine is longer than that in the adult intestine and in the mammalian intestine. The amphibian adult epithelium undergoes cell-renewal along the trough-crest axis of intestinal folds, similar to that along the crypt-villus axis in the mammalian intestine. All of the mammalian intestinal epithelial cells originate from stem cells in the crypt, where several kinds of cells act as niche players.

In the present review, we focus on sonic hedgehog (Shh) and bone morphogenetic protein-4 (BMP-4) signaling and discuss their roles in the development of the intestinal epithelium originating from the stem cells, based on our studies using the amphibian intestine as a model.

Amphibian intestine as a model for the study of the stem cell niche

Organ remodeling during amphibian metamorphosis bears many similarities to postembryonic organ regeneration in mammals. During metamorphosis, to adapt to the terrestrial carnivorous life, the long larval intestine changes dramatically into a shorter but more complex one analogous to the mammalian intestine (Fig. 1). At the cellular level, almost all of the larval epithelium undergo apoptosis (Ishizuya-Oka & Ueda 1996), whereas a small number of adult progenitor cells, which are morphologically undifferentiated and express Musashi-1, a candidate marker for stem cells and their immediate descendants in the mammalian intestine (Kayahara et al. 2003; Potten et al. 2003; Okano et al. 2005), appear and newly form the adult epithelium by active cell proliferation and subsequent differentiation (Hourdry & Dauca 1977; McAvoy & Dixon 1977). The adult epithelium after metamorphosis consists of all cell types of mammalian intestinal epithelial cells except for Paneth cells (McAvoy & Dixon 1978) and undergoes cell renewal along the trough-crest axis of intestinal folds (Fig. 1; McAvoy & Dixon 1977; Shi & Ishizuya-Oka 1996) similar to that along the crypt-villus axis in the mammalian intestine (Cheng & Bjerknes 1985; Madara & Trier 1994). Adult epithelial cells proliferate in the trough of the intestinal folds and undergo apoptosis in their crest (McAvoy & Dixon 1977; Ishizuya-Oka & Ueda 1996). These results suggest that the adult progenitor cells include multipotent stem cells analogous to the mammalian ones (Ishizuya-Oka et al. 2003). Thus, the amphibian intestine during metamorphosis serves as a model for studying regeneration of the mammalian intestinal epithelium that originates from the stem cells (Ishizuya-Oka & Shi 2005).

In the Xenopus laevis intestine we previously reported that the connective tissue, which is immature before metamorphosis (Marshall & Dixon 1978), suddenly develops concomitantly with the start of larval-to-adult epithelial remodeling (Ishizuya-Oka & Shimozawa 1987). When the adult progenitor cells become detectable as small islets (adult epithelial primordia), which consist of undifferentiated stem cells and their descendants, between the larval epithelium and the connective tissue (Fig. 2A) (McAvoy & Dixon 1977), the connective tissue begins active cell proliferation. Characteristically, subepithelial fibroblasts surround the adult epithelial primordia with a high cell density and, through the modified basal lamina, often make direct cell contact with the adult epithelial primordia (Fig. 2B) but not with the larval epithelium undergoing apoptosis. In addition, almost all of the subepithelial fibroblasts possess well-developed rough endoplasmic reticulum (Fig. 2C), suggesting their active protein synthesis. These morphological results lead us to consider that cell-cell contact between the two tissues and endogenous factors produced by the subepithelial fibroblasts play some role in adult epithelial development. More importantly, our previous recombination experiments have demonstrated that the connective tissue isolated from the anterior part of the intestine, where the larval connective tissue is most localized, has an inductive action on adult epithelial development (Ishizuya-Oka & Shimozawa 1992; Shi & Ishizuya-Oka 1996). Thus, the connective tissue of the remodeling amphibian intestine is expected to serve as the microenvironment known as ‘niche’ (Potten et al. 1997; Mills & Gordon 2001).

Figure 2.

Adult epithelial primordia surrounded by subepithelial fibroblasts during Xenopus laevis metamorphosis. (A) Primordia of the adult intestinal epithelium (AE) consist of undifferentiated cells, whose cytoplasm is poor in cell organelles. They appear as islets between the degenerating larval epithelium (LE) and the connective tissue (CT). (B) Higher magnification of the site indicated by the box in (A). Subepithelial fibroblast (Fb) makes cell contact (arrowhead) with the adult epithelial cell through the basal lamina (bl), which becomes thick and amorphous. (C) Fibroblasts just beneath the adult epithelial primordia possess well-developed rough endoplasmic reticulum (rer). Bars, 1 µm.

It still remains unclear as to what cells in the tadpole intestine is the origin of adult stem cells. However, increasing circumstantial evidence supports the hypothesis that the stem cells originate from at least ‘partially’ differentiated larval cells (Marshall & Dixon 1978; Amano et al. 1998; Schreiber et al. 2005). In our previous observations with electron microscopy, all of the larval epithelial cells facing the lumen possess the long striated border and no undifferentiated cells are morphologically detectable in the epithelium before metamorphosis. Therefore, this amphibian model is expected to be useful for identifying the unknown signals that induce the intestinal epithelial cells to dedifferentiate into the stem cells as well as the signals that control existing stem cells.

Strategy to clarify molecular mechanisms of amphibian intestinal remodeling

It is well known that amphibian metamorphosis is triggered by a single hormone, thyroid hormone (TH) (Dodd & Dodd 1977; Yoshizato 1989; Kikuyama et al. 1993). In the X. laevis intestine, larval-to-adult epithelial remodeling can be organ autonomously reproduced by TH (Ishizuya-Oka & Shimozawa 1991, 1992). Therefore, the functional analysis of TH response genes that are endogenously expressed in the X. laevis intestine will finally lead to clarification of the molecular mechanisms of the intestinal epithelial remodeling (Shi 1999). Since the 1990s, a large number of TH response genes have been isolated from the X. laevis intestine using subtractive differential screening (Shi & Brown 1993; Amano & Yoshizato 1998; Shimizu et al. 2002). To assess their possible functions during intestinal remodeling, the expression profiles of TH response genes have been examined by in situ hybridization (ISH). Among the genes whose expression is tissue-specific and correlates well with adult epithelial development (Fig. 3), there are signaling molecules including Shh (Stolow & Shi 1995) and BMP-4 (Ishizuya-Oka et al. 2001a), and matrix metalloproteinases (MMPs) such as collagenases, gelatinases, stromelysins and membrane type-MMPs (Shi & Brown 1993; Patterton et al. 1995; Stolow et al. 1996; Hasebe et al. 2006), which are generally known to degrade various extracellular matrix (ECM) components (Woessner 1991; Matrisian 1992; Birkedal-Hansen et al. 1993). These suggest that the remodeling process is controlled by complicated cell-cell and cell-ECM interactions between the epithelium and the connective tissue. Although the expression analysis of TH response genes is still in progress, the availability of the organ culture and recently developed transgenic technology in the X. laevis (Fu et al. 2002; Buchholz et al. 2004b) have made it possible to analyze the function of the genes during intestinal remodeling (Fig. 4).

Figure 3.

Schematic drawing showing thyroid hormone (TH) response genes that are expressed either in the epithelium or the connective tissue during Xenopus laevis intestinal remodeling. Some of the genes, including Shh and ST3, are directly upregulated by TH, whereas the others are indirectly upregulated.

Figure 4.

Strategy for the functional analysis of thyroid hormone (TH) response genes. If genes encode the secretory protein, their effects can be examined by adding the encoded protein and its antagonist to the medium in the organ culture system (A), where tubular intestines isolated from tadpoles are slit open lengthwise and cultured at 26°C. If genes encode the transcription factor, a gene transfer system using electroporation (B) is available. In the intestine transfected with green fluorescent protein (GFP)-expressing plasmid and then cultured, its expression attains its maximum around day 3, when the GFP protein is detectable uniformly in the epithelium. Intestines isolated from transgenic frogs (C) are also useful for the organ culture system. CT, connective tissue; E, epithelium; M, muscle.

In the X. laevis organ culture system we previously established (Ishizuya-Oka & Shimozawa 1991), larval-to-adult intestinal remodeling can be reproduced in vitro by the inductive action of TH just like in vivo. In contrast, in the absence of TH, the tadpole epithelium remains a larval type. Thus, if TH response genes encode the secretory protein, we can easily examine their effects by adding the protein and its antagonist to the culture medium (Fig. 4A). In addition, for the genes that encode transcription factors, we have recently developed a gene transfer system by using electroporation (Fig. 4B), which has been shown to be a very useful method for gene transfer in chick embryos both in ovo (Funahashi et al. 1999) and in vitro (Fukuda et al. 2000). When the X. laevis intestine is transfected with the green fluorescent protein (GFP)-expressing plasmid by electroporation and then cultured, the expression of GFP is upregulated in the epithelium in vitro, attains its maximum around day 3, and thereafter is gradually downregulated (Ikuzawa et al. 2005). Because most of the TH response genes are transiently upregulated during metamorphosis, the period of time during which GFP is expressed is usually long enough to investigate the function of the genes. Finally, transgenic Xenopus frogs are now available for the functional analysis of TH response genes (Fig. 4C; Kroll & Amaya 1996; Huang et al. 1999; Buchholz et al. 2004a; Das & Brown 2004). A double promoter construct, which expresses some genes under the heat shock-inducible promoter and GFP gene under the γ-crystalline promoter, is a powerful tool for studies with the transgenic frogs (Fu et al. 2002; Buchholz et al. 2004b). The transgenic tadpoles with this construct are easily detected by GFP expression in their eyes and express TH response genes at any desired time by being heat-shocked. Furthermore, the use of the transgenic frogs for the organ culture system makes it possible to investigate the function of genes and their interactions more precisely.

Although such functional analysis of TH response genes has just begun in the X. laevis intestine, that of stromelysin-3 (ST3; MMP11) has already made steady progress in recent years (Ishizuya-Oka & Shi 2005). ST3 was first identified as a gene associated with human breast cancer carcinomas, and its mRNA was reported to be highly expressed in fibroblasts surrounding invasive breast cancers (Basset et al. 1990) and the epithelium undergoing apoptosis (Lefebvre et al. 1992). However, there is no direct evidence for the role of ST3 in epithelial apoptosis and cancer progression in mammalian tissues. Using the organ culture system of the X. laevis intestine (Ishizuya-Oka et al. 2000) and transgenic tadpoles expressing ST3 (Fu et al. 2005), we have experimentally shown that ST3 promotes apoptosis of the larval epithelium and the invasion of the adult epithelial primordia into the connective tissue, possibly by the cleavage of the receptor for laminin, a major ECM component of the basal lamina (Amano et al. 2005). For a full understanding of the molecular mechanisms of cell-ECM interactions during intestinal remodeling, it is worth clarifying the functions of several other MMPs identified as TH response genes and their interactions (see review by Fu et al. 2007).

Signaling molecules involved in epithelial-connective tissue interactions

BMP-4 affects not only the connective tissue but also the adult epithelium

In contrast to the recent progress in studies of MMPs, far less is known regarding the signaling molecules involved in cell-cell interactions. Among the TH response genes we have examined thus far by ISH, the expression profile of BMP-4 mRNA agrees best with the subepithelial fibroblasts possessing well-developed rough endoplasmic reticulum described above (Ishizuya-Oka et al. 2001a). Characteristically, when the level of BMP-4 mRNA is the highest, its mRNA becomes more abundant with a gradient toward the adult epithelial primordia actively proliferating, that is, signals detected by ISH become stronger the closer fibroblasts are to the adult epithelial primordia (Fig. 5A). This expression pattern suggests that BMP-4 may function as a morphogen important for adult epithelial development. Because BMP-4 is a secretory protein, we analyzed its functions using the culture system of the X. laevis intestine by the addition of BMP-4 protein and its antagonist, Chordin, which binds BMP-4 with a high affinity (Piccolo et al. 1996). Although there are more than seven known antagonists of BMP-4 in vertebrates (Balemans & Hul 2002), we used Chordin to inhibit BMP-4 activity, because its mRNA is transiently expressed in vivo during intestinal remodeling (Ishizuya-Oka et al. 2006). Our results indicated that BMP-4 suppresses cell proliferation of the connective tissue during the intestinal remodeling just like BMP-4 suppresses that of the mesenchyme in the embryonic gut of higher vertebrates (Smith et al. 2000; Batts et al. 2006). Therefore, it seems likely that BMP-4 function in the mesenchymal tissue may be conserved among the species and between the embryonic and postembryonic intestines. In contrast, the function of BMP-4 in the endoderm and/or epithelium is still less defined.

Figure 5.

BMP-4 mRNA expression in the Xenopus laevis intestine during metamorphic climax (A) and the stage 57 tadpole intestines cultured for 5 days in the presence of thyroid hormone (TH) (B–D). (A) When adult epithelial primordia (AE) grow in size, signals in the connective tissue (CT) become stronger with a gradient toward the primordia. (B) Intact intestine. BMP-4 expression is upregulated in the connective tissue just beneath the epithelium. (C, D) Epithelium-free intestines in the absence (C) and presence of Shh (500 ng/mL) (D). Shh induces BMP-4 expression only in the connective tissue. M, muscle. Scale bars, 20 µm.

In the organ culture of the intestine isolated from X. laevis tadpoles at stage 57 (before metamorphic climax; Nieuwkoop & Faber 1994), adult epithelial primordia become detectable by the inductive action of TH and actively proliferate after 5 days of cultivation (Fig. 6A; Ishizuya-Oka & Shimozawa 1991, 1992). Then, after 7 days, they differentiate into a single layer of epithelium expressing intestinal fatty-acid binding protein (IFABP), a marker of absorptive epithelial cells (Fig. 6D; Ishizuya-Oka et al. 1997). The addition of exogenous BMP-4 to the culture medium caused precocious differentiation of the adult epithelial primordia into a single layer of epithelium and the upregulation of IFABP expression (Fig. 6B,E), while Chordin inhibited epithelial differentiation (Fig. 6C,F; Ishizuya-Oka et al. 2006). This strongly suggests that BMP-4 promotes the differentiation of the intestinal absorptive epithelium.

Figure 6.

Effects of exogenous BMP-4 and Chordin proteins on the X. laevis intestine at stage 57 and cultured for 5 days (A–C) and 7 days (D–F) in the presence of thyroid hormone (TH). Sections were stained with methyl green-pyronin Y (A–C) and immunostained with anti-IFABP antibodies (D–F). (A, D) Control intestines cultured without BMP-4 and Chordin. Primordia (arrowheads) of the adult epithelium (AE) become detectable between the larval epithelium (LE) and the connective tissue (CT) (A) and then differentiate into the absorptive epithelium expressing IFABP (D). (B, E) In the presence of BMP-4 (100 ng/mL). The single layer of the adult epithelium is precociously differentiated on day 5 (B) and its immunoreactivity for IFABP is much higher (E) than that in the control intestine (D). (C, F) In the presence of Chordin (1 µg/mL). Adult progenitor cells positive for pyronin Y (arrows) are fewer (C) than those in the control intestine on day 5 (A) and IFABP expression remains undetectable until day 7 (F). Scale bars, 20 µm.

Recently, in the mammalian intestine, it has been reported that the BMP-4 signaling pathway represses crypt formation and polyp growth (Haramis et al. 2004; Batts et al. 2006) by suppressing the Wnt signaling pathway involved in epithelial cell proliferation (He et al. 2004). However, in the amphibian remodeling intestine, exogenous BMP-4 had no effect on cell proliferation of the adult epithelial primordia (Ishizuya-Oka et al. 2006). Because a number of TH response genes, which include genes that activate epithelial cell proliferation (Ikuzawa et al. 2005), are concomitantly expressed along with BMP-4 by the action of TH, interactions with such genes may make it difficult to detect the function of BMP-4 alone. In contrast, excessive Chordin, that is, lack of endogenous BMP-4, results in a decrease of adult progenitor cells in number and proliferation. This suggests that a certain amount of BMP-4 may be necessary for the maintenance and/or self-renewal of the intestinal stem cells, as proposed in other organs such as the ovary (Xie & Spradling 1998; Fujiwara et al. 2001) and embryonic stem cells (Ying et al. 2003; Qi et al. 2004). To address functions of BMP-4 in the epithelial cell proliferation precisely, further studies should distinguish the effects of BMP-4 on stem cells themselves from those on their descendants.

As a TH response gene that is related to BMP-4, we previously identified a Xenopus homologue of Drosophila Tolloid closely related to BMP-1 (Tolloid/BMP-1) (Shimizu et al. 2002), which regulates the activity of BMP-4 by degrading Chordin during the early development of X. laevis embryos (Piccolo et al. 1997; Wardle et al. 1999; Blitz et al. 2000; Balemans & Hul 2002). In the remodeling X. laevis intestine, both the mRNA expression of Tolloid/BMP-1 and that of Chordin are upregulated slightly earlier than that of BMP-4. Therefore, it seems likely that the activity of BMP-4 in vivo may be spatio-temporally regulated by Tolloid/BMP-1 and Chordin, both of which are TH-inducible.

Shh induces BMP-4 expression and acts as an epithelial signaling molecule

In the digestive tract of higher vertebrates, Shh expression is localized in the endoderm of the embryonic gut and upregulates the expression of BMP-4 mRNA in the mesenchyme (Roberts et al. 1995, 1998; Sukegawa et al. 1999; Roberts 2000). Studies with Shh mutant mice showed that Shh plays important roles in organogenesis of the digestive tract (Litingtung et al. 1998; Ramalho-Santos et al. 2000), although its precise mechanisms have not yet been clarified. In the X. laevis intestine, Shh has been identified as a TH response gene whose expression is directly induced in the epithelium by TH (Stolow & Shi 1995). To clarify if BMP-4 expression is induced by Shh in the remodeling intestine, we used the organ culture system (Ishizuya-Oka et al. 2006). In the tadpole intestine at stage 57, TH upregulated the connective tissue-specific expression of BMP-4 mRNA in vitro (Fig. 5B) but did not in the intestine deprived of the epithelium (Fig. 5C). When the Shh protein was added to the medium, BMP-4 mRNA became detectable in the connective tissue of the epithelium-free intestine (Fig. 5D). Therefore, in the X. laevis intestine, TH first induces the epithelium-specific expression of Shh, which in turn acts on the surrounding fibroblasts to express BMP-4 mRNA.

In the human adult intestine, Shh expression is recently reported to be restricted to the region corresponding to the epithelial stem cells (van den Brink et al. 2001; de Santa Barbara et al. 2003) and to be upregulated when the stem cells proliferate to compensate for the epithelial damage in the neoplastic or inflammatory intestines (Berman et al. 2003; Nielsen et al. 2004). Interestingly, in the amphibian remodeling intestine, the Shh protein is most intensely expressed in the adult epithelial primordia (Ishizuya-Oka et al. 2001b), which consists of the stem cells and their descendents and actively proliferate. This expression pattern of Shh predicts the involvement of Shh in the maintenance and/or proliferation of the stem cells in the postembryonic intestine. In addition, we previously indicated that the addition of excessive Shh protein to the culture medium inhibits differentiation of the intestinal epithelium and often causes anomalies of the epithelial structure (closure of the lumen) (Ishizuya-Oka et al. 2001b). This suggests that downregulation of Shh expression is necessary for the epithelial differentiation in the postembryonic intestine just like that seen in the X. laevis embryonic gut during organogenesis (Zhang et al. 2001). Although we have obtained only fragmentary information on the multifunctional protein Shh, recent studies strongly suggest key roles of Shh as an epithelial signaling molecule in the digestive tract from the embryonic period to adulthood. In the postembryonic intestine, it seems likely that Shh may develop the stem cell niche by acting on the surrounding tissues to express target genes such as BMP-4.


It has been amply shown that epithelial-connective tissue interactions play important roles in organogenesis, homeostasis and regeneration of the avian and mammalian digestive tract (Fukamachi & Takayama 1980; Kedinger et al. 1988; Yasugi 1993; Mizuno 1994). Their complicated molecular mechanisms are interesting from the viewpoint of developmental biology, but are just beginning to be studied in recent years. In particular, clarification of the stem cell niche in the digestive tract is a difficult but urgent problem that needs to be tackled from the viewpoint of the therapeutic application of stem cells. In this field of study, the amphibian intestine provides a unique and excellent model because: (1) stem cells of the adult epithelium have many characteristics in common with mammalian counterparts; (2) all of the genes essential for the intestinal remodeling can be identified as TH response genes; and (3) functions of TH response genes are now easily and directly analyzed using the organ culture and transgenic technology, which has been recently established. Although the functional analysis of TH response genes has just begun in the X. laevis intestine, there is a growing body of evidence that Shh and BMP-4 play important roles in the development of the adult epithelium originating from the stem cells, as an epithelial and a connective tissue signaling molecule, respectively. In the mammalian digestive tract, mutations in the members of Shh and BMP-4 signaling pathway are associated with different malformations and diseases such as pre-cancerous juvenile polyposis syndrome (Howe et al. 2001; de Santa Barbara et al. 2002; He et al. 2004). To establish a general view of the stem cell niche in the digestive tract, our next step should be directed at clarifying the functions of the members of the Shh and BMP-4 signaling pathway more precisely and their interactions with MMPs and other signaling pathways using this animal model.


I would like to express my gratitude to Dr Yun-Bo Shi, National Institute of Health, USA, for his valuable advice and collaboration. This research was supported in part by JSPS Grants-in-Aid for Scientific Research (C).