In vitro organogenesis from undifferentiated cells in Xenopus


  • Makoto Asashima,

    Corresponding author
    1. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan
    2. Organ Development Research Laboratory, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 4, Tsukuba, Ibaraki, Japan
    • Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan
    Search for more papers by this author
  • Yuzuru Ito,

    1. Organ Development Research Laboratory, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 4, Tsukuba, Ibaraki, Japan
    Search for more papers by this author
  • Techuan Chan,

    1. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan
    Search for more papers by this author
  • Tatsuo Michiue,

    1. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan
    Search for more papers by this author
  • Mio Nakanishi,

    1. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan
    Search for more papers by this author
  • Kan Suzuki,

    1. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan
    Search for more papers by this author
  • Keisuke Hitachi,

    1. Institute for Comprehensive Medical Sciences, Fujita Health University, Toyoake, Aichi, Japan
    Search for more papers by this author
  • Koji Okabayashi,

    1. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan
    Search for more papers by this author
  • Akiko Kondow,

    1. Institute for Comprehensive Medical Sciences, Fujita Health University, Toyoake, Aichi, Japan
    Search for more papers by this author
  • Takashi Ariizumi

    1. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, Meguro-ku, Tokyo, Japan
    Search for more papers by this author


Amphibians have been used for over a century as experimental animals. In the field of developmental biology in particular, much knowledge has been accumulated from studies on amphibians, mainly because they are easy to observe and handle. Xenopus laevis is one of the most intensely investigated amphibians in developmental biology at the molecular level. Thus, Xenopus is highly suitable for studies on the mechanisms of organ differentiation from not only a single fertilized egg, as in normal development, but also from undifferentiated cells, as in the case of in vitro organogenesis. Based on the established in vitro organogenesis methods, we have identified many genes that are indispensable for normal development in various organs. These experimental systems are useful for investigations of embryonic development and for advancing regenerative medicine. Developmental Dynamics 238:1309–1320, 2009. © 2009 Wiley-Liss, Inc.


Many researchers are conducting studies in the fields of regenerative science and regenerative medicine. In these fields, the establishment of experimental systems or protocols for the in vitro differentiation of various tissues and organs is important not only for investigations of the mechanisms of embryonic development and organ differentiation, but also for the screening and identification of genes, proteins, and chemical substances that might be useful for drug discovery and therapeutic applications.

We attempted to define the optimal conditions for the in vitro induction of various tissues and organs from the undifferentiated cells (animal cap cells) of amphibians, especially the African clawed frog Xenopus laevis, using treatment with activin and other factors. In amphibians, the ectodermal cell masse of the mid-blastula embryo (animal cap) is pluripotent. Animal caps can be excised easily with sharpened forceps or needles under a stereomicroscope. The caps can be cultured as “explants” in simple saline solution under serum-free conditions for more than a week, and their differentiation can be controlled easily in vitro (animal cap assay).

Several research groups have identified activin and homologues thereof as factors that induce all mesodermal tissue types in the animal cap assay (Asashima et al.,1989,1990a,b; Smith et al.,1990; Sokol et al.,1990; van den Eijnden-Van Raaij et al.,1990). Activin, which is one of the transforming growth factor β (TGF-β) family proteins, stimulates the activin/Nodal signal pathway, one of the key signaling pathways in embryonic development. Activin A induces the mesoderm in a dose-dependent manner (Nakano et al.,1990; Green and Smith,1990; Ariizumi et al.,1991; Green et al.,1992; Ariizumi and Asashima,1994). For example, lower concentrations of activin A (0.5–1.0 ng/ml) act on animal caps to induce ventral mesodermal cells, including blood cells, medium concentrations of activin A (5–10 ng/ml) induce the more dorsal mesodermal cells, including muscle cells, and higher concentrations of activin A (50–100 ng/ml) induce the most dorsal mesodermal structures, such as the notochord (Okabayashi and Asashima,2003). We have also found that Xenopus animal cap cells treated with high concentrations of activin differentiate into anterior endodermal tissue, and can function as the cells of the organizer region when the activin-treated animal cap is transplanted into early gastrula (Ninomiya et al.,1999). Using activin as a key factor, we investigated the establishment of an experimental system that would allow the induction of various mesodermal and endodermal cells and tissues of the heart, kidney, pancreas, and so on. We succeeded in inducing various organs and tissues from undifferentiated cells of Xenopus.

The organs and tissues induced from animal caps are very useful for elucidating the mechanisms of organ development, because the in vitro–induced organs and naturally developing organs are very similar and almost equivalent in terms of gene expression and histologic features. Many of the genes identified by differential screening using in vitro–induced organs of Xenopus have been found to function in the same organs during mammalian embryonic development, including in mice and humans. Recently, we studied in vitro organogenesis using mammalian stem cells and compared the optimal conditions for induction. We found that the animal cap assay remains a powerful tool for not only developmental biology studies, but also for studies in regenerative medicine. In this review, we report on recent advances in understanding the mechanisms of organ development based on in vitro organogenesis in X. laevis.


Research on cardiac development in vertebrates has adopted various approaches, including the vitro induction of myocardial cells from pluripotent cells, such as mammalian embryonic stem (ES) cells (Boheler et al.,2002; Sachinidis et al.,2003; Caspi and Gepstein,2004; Yuasa et al.,2005; Honda et al.,2006), myeloid stem cells (Dengler and Katus,2002), and umbilical cord blood stem cells (Kogler et al.,2004). Numerous experiments have been carried out to transplant myocardial cells obtained in this manner into hearts that have undergone myocardial infarction, with the aim of improving cardiac function (Gepstein,2002; Dowell et al.,2003; Gerecht-Nir and Itskovitz-Eldor,2004). However, there are no published reports describing induced tissues that have progressed beyond the transplanted cell level. The creation of higher-order structures unique to the heart, such as atria, ventricles, and valves, is an important goal. Considering regeneration of the heart, we asked whether the organ as whole could be regenerated rather than simply creating aggregations of cells, which represents a shift from tissue engineering to organ engineering.

Xenopus embryos are valuable research tools because they are available in large numbers and are sufficiently large and robust to be cultured after simple microsurgery. In addition, Xenopus eggs are fertilized externally and all the stages of embryo development are easily accessible, making it relatively easy to study the functions of individual gene products during heart development by microinjection into embryonic cells (Jacobson and Sater,1988; Lohr and Yost, 2000; Mohun et al., 2003). Blastomere-specific microinjection allows the disruption of genes specifically in differentiating heart tissues. Furthermore, because heart function is nonessential until tadpoles begin feeding, even treatments that severely disrupt cardiogenesis can be studied in this model (Ito et al.,2008; Monzen et al.,2008; Zhu et al.,2008).

In Xenopus, cardiogenesis is initiated by the combined action of Wnt antagonists from the Spemann organizer and several endodermal factors (Nascone and Mercola,1995; Shi et al.,2000; Schneider and Mercola,2001). Cardiac progenitor cells are formed bilaterally on mesodermal tissues adjacent to the Spemann organizer during the early gastrula stages. The precardiac mesoderm (PCM) extends laterally as neurulation proceeds, and eventually fuses at the ventral midline in the tail bud stage. The subsequent differentiation of cardiac tissue and formation of the beating heart occur in this ventral location (Fig. 1). To elucidate the mechanism of heart differentiation, the cloning and functional analysis of heart-specific genes are effective strategies. However, it is difficult to isolate manually the endogenous PCM owing to its small size (Fig. 1).

Figure 1.

Cardiac formation during Xenopus early embryogenesis. A: Precardiac mesoderm (PCM, blue) is formed on the prospective dorsal side of the early gastrula (stage 10). B–G: The region of PCM formation is defined by the expression of Nkx2.5 (arrowhead; Tonissen et al.,1994). B: By the end of the gastrula (stage 13), the PCM is fused and lies at the anterior end of the embryo. C: At the neurula (stage 18), the PCM is located on the ventral side behind the cement gland. D: At the early tail bud (stage 23), the PCM is separated. At the mid–tail bud (stage 28), the PCM is re-fused. F: At the late tail bud (stage 34), a tubular and beating heart is formed. G: At this point, the tadpole heart separates into the atrium and ventricle. A: Vegetal view, with dorsal upward. B–G: Ventral view, with anterior toward the left.

Pluripotent cells, as constituents of the animal cap, are present in the X. laevis blastula. This structure comprises approximately 1,000 cells, and is known to be capable of inducing the differentiation of myocardial cells by activating activin (Logan and Mohun,1993; Ariizumi et al.,1996) and causing the overexpression of proteins, such as GATA4 (Latinkic et al.,2003). Moreover, explants of the noncardiogenic ventral marginal zone (VMZ) formed beating structures through the overexpression of Xwnt11, crescent or dkk1 (Schneider and Mercola,2001; Pandur et al.,2002). However, in all these investigations, the induction rate for myocardial cells was not high, and the induced tissues did not structurally resemble the heart.

We established an experimental system that generates beating hearts with almost 100% certainty, using a new procedure for the temporal dissociation of cells from animal caps treated with activin (Ariizumi et al.,2003; Fig. 2A). When these cells are cultivated, rather than simply form a mass of myocardial cells, they form a tubular structure. Whole-mount in situ hybridization of the XcTnI (Xenopus cardiac troponin I; Drysdale et al.,1994) gene, which is ordinarily expressed specifically in the myocardium, revealed that the beating tissues were definitely heart-like (Fig. 2B). Next, we performed an ectopic transplantation experiment to determine whether explants: (1) form higher-order heart structures in the recipient; and (2) play some role in blood circulation (Fig. 2A). If ectopic hearts contribute to hemodynamics and contain cardiac structures, such as atria and ventricles, this would represent a breakthrough in understanding cardiogenesis and organ engineering. When myocardial cells induced by this method were transplanted ectopically into the abdomen of a recipient embryo they contributed to systemic circulation in the same manner as the host cardiac cells (Fig. 2C). Moreover, the second hearts contained higher-order structures, such as atria and ventricles, and were morphologically, histologically (Fig. 2D), and electrophysiologically identical to the original hearts. However, unlike ES cells, these cells did not become tumorigenic. Although immediate clinical applications are not readily apparent, the finding that a heart induced in vitro forms higher-order structures in the body and functions as a circulatory organ represents a fundamental advance in heart regeneration research.

Figure 2.

Cardiac induction from animal caps and in vivo transplantation. A: Animal caps excised from blastulae were placed in physiologic saline that lacked calcium, to loosen the cell adhesions. The culture medium was replaced with 100 ng/ml activin dissolved in physiologic saline that contained calcium, and the cells were dissociated by pipetting. Individual cells were then reaggregated, and the reaggregates began to beat approximately 2 days later. Reaggregates that had been cultured for 1 day were broken into appropriate sizes and transplanted through an incision that had already been made anterior to the cloaca in the neurula. B:XcTnI, a marker gene for myocardial differentiation, is expressed specifically in the myocardium of a normal 3-day-old embryo (left). The dissociated animal cap cells form spherical reaggregates, regardless of whether they have been treated with activin. XcTnI signals are detected exclusively in the activin-treated reaggregates (right). C: The internal anatomy of a 1-year-old frog with a well-developed ectopic heart. The ectopic heart adjacent to the host's intestine is incorporated into the host's vascular system. The blood from the host's mesenteric artery (black arrows) flows into the host's anterior abdominal vein (white arrows) by means of the ectopic heart (h). D: Histologic section of an ectopic heart. The heart can be divided into multiple chambers: a ventricle-like chamber (v) that comprises a thick and deeply penetrating layer of myocardium; and an atrium-like chamber (a) that is surrounded by a thin layer of myocardium.

The above-mentioned in vitro heart differentiation systems will be useful for analyzing cardiogenesis at the molecular level in vertebrates. These systems enable the acceleration of cardiogenesis through the use of an undifferentiated animal cap or VMZ as the starting material, in combination with a simple procedure to dissociate the animal cap. These systems should enable the isolation of novel genes that are involved in the earliest stages of cardiogenesis, the analysis of which has been difficult to achieve in previous experimental systems that used the presumptive cardiac region as the source material. For example, we originally set out to investigate genes that are involved in cardiogenesis. Subsequently, we cloned the Xenopus hyaluronan and proteoglycan-binding link protein 3 (XHAPLN3) gene, which is expressed in the cardiac region during early embryogenesis (Ito et al.,2008). HAPLN family proteins commonly contain an aggrecan-family binding domain and two hyaluronan-binding domains. The localization of XHAPLN3 transcripts overlapped with those of both Xhas2 (Köprunner et al.,2000, Nardini et al.,2004) and Xversican in the cardiac anlagen of early tail bud embryos (Fig. 3A–D). Furthermore, knockdown of XHAPLN3 or Xhas2 with morpholino antisense oligonucleotides resulted in structural deficiencies in the hearts of developing tadpoles (Fig. 3E–G). Further analysis indicated that XHAPLN3 helps to maintain the hyaluronan matrix that surrounds the cardiac anlage, thereby contributing to cardiogenesis. Thus, screening with these in vitro–induced hearts has contributed new knowledge as to when and how the components of the hyaluronan matrix function in cardiogenesis. These systems also permit more detailed analyses of the roles of Wnt signaling (Marvin et al.,2001; Schneider and Mercola,2001; Tzahor and Lassar,2001) and transcription factors, such as Nkx2.5 (Durocher et al.,1997; Reecy et al.,1999) and GATA4 (Grepin et al.,1995,1997; Jiang et al.,1999) in cardiogenesis. In the recent study, we examined a comprehensive set of cardiogenesis-related genes using an in vitro Xenopus-based system and microarrays. This screening process uncovered several cardiogenesis-related genes that show promise in terms of advancing our current knowledge of heart formation (Fig. 4).

Figure 3.

Developmental expression profiles and knockdown analysis of XHAPLN3 and Xhas2. A–D: Whole-mount in situ hybridization of Nkx2.5 (A), XHAPLN3 (B), Xversican (C), and Xhas2 (D). Lateral views are shown, with the anterior toward the left and dorsal side upward. The arrowhead indicates the precardiac mesoderm (PCM). E–G: The effects of depleting XHAPLN3 and Xhas2 are validated by the pattern of XcTnI expression. Normal tadpole heart structure (E, arrow). F,G: However, both XHAPLN3-morphant (F) and Xhas2-morphant (G) show no XcTnI expression. Ventral views are shown, with anterior upward.

Figure 4.

Already-known genes that were identified using the microarray. Gene names are indicated in red and arranged by embryonic stage. Nkx2.5 and GATA4/5/6 are among the earliest heart-related transcription factors (Tonissen et al.,1994, Afouda et al.,2005). Moreover, Nkx2.5 expression is regulated by both XHMGA2 and Smads (Monzen et al.,2008). The RNA-binding protein hermes also regulates Nkx2.5 expression (Gerber et al.,2002). Tbx5, Tbx20, myocardin, and islet1 are transcription factors that interact with Nkx2.5 and GATAs and regulate gene expression in the precardiac region (Stennard et al.,2003, Small et al.,2005, Brade et al.,2007). Both the Notch signal (Notch1, Serrate1, and Delta1) and hyaluronan matrix (Xhas2, Xversican, and XHAPLN3) genes are essential for the correct specification of myocardial cell fates (Rones et al.,2000, Ito et al.,2008). Finally, terminal differentiation marker genes (cardiac troponin I/C/T, MHCa/b, MLC2, cardiac actin, and ANF), which encode principally structural proteins, are expressed (Gurdon et al.,1989, Drysdale et al.,1994, Small and Krieg,2003). Moreover, cardiogenesis is maintained by several secreted factors (Eroshkin et al.,2006, Zhu et al.,2008).

Moreover, as the ectopically transplanted cardiac primordium undergoes morphogenetic processes (e.g., looping at the transplantation site) that ultimately determine the atria and ventricles, this system allows the elucidation of inductive interactions with proximal tissues (Nascone and Mercola,1995; Branford et al.,2000; Lough and Sugi,2000). In the transplantation experiment, the transplanted cardiac primordium not only differentiated into a beating heart, but this heart communicated with the vascular system of the host and was incorporated into the systemic circulation. Thus, these systems may also be useful for analyzing the mechanisms of communication between the heart and vasculature.


Hematopoietic cells and vascular endothelial cells share a common precursor, the hemangioblast (Xiong,2008). In the first stage of blood vessel development, vascular endothelial cells are differentiated from the hemangioblast and form a vascular structure, with the main blood vessels being formed. This process is called “vasculogenesis.” In the second stage, which is called “angiogenesis,” new vascular structures sprout from pre-existing blood vessels, elongate, and become interconnected, so that a network of blood vessels is generated. The blood vessel is a fundamental structure that is crucial for the normal formation and growth of all organs. We attempted to establish the optimal conditions for inducing vascular endothelial cells in animal cap explants, with the dual aims of investigating the mechanisms of blood vessel development and discovering ways to manipulate blood vessel formation in vitro.

It has been previously reported that treatment with lower concentrations of activin promotes the differentiation of ventral mesodermal cells, including blood cells, coelomic epithelial cells, and mesenchymal cells in the explants of animal caps (Green and Smith,1990; Ariizumi et al.,1991). We assume that these blood cells are differentiated from hemangioblasts in the activin-treated explants. To define the optimal conditions for the differentiation of vascular endothelial cells only, we dissociated the animal cap cells in Ca2+- and Mg2+-free Steinberg's solution, so as to ensure the uniform treatment of the animal cap cells with activin. We also used an additional inducer, angiopoietin-2 (Ang-2), which is known to be important during the later stages of blood vessel differentiation and remodeling. Co-treatment of the dissociated animal cap cells with 0.4 ng/ml activin and 100 ng/ml Ang-2 resulted in the expression of vascular endothelial markers, such as X-msr and Xtie2 (Devic et al.,1996; Iraha et al.,2002), whereas there was no expression of hematopoietic markers, such as GATA-1, in the reaggregated explants (Nagamine et al.,2005). Microarray analysis was used to compare the gene expression profiles of the activin-treated and activin-Ang-2–treated animal cap cells, with the result that vascular-specific genes, including XRASGRP2 (Nagamine et al.,2008), were identified. We also isolated a novel vascular-related gene under other experimental conditions, and named the gene Ami (“mesh” in Japanese). Ami mRNA is expressed in the endothelium of the forming blood vessel (Fig. 5; Inui and Asashima,2006). These two genes are candidate marker genes for angiogenesis and/or maturation, although their functions are unknown. Improvements to this in vitro induction method are expected to identify novel vascular endothelial cell markers.

Figure 5.

Temporal expression patterns of already-known genes that are related to blood vessel differentiation in Xenopus. Dashed arrows indicate the temporal expression of each gene in prospective and forming regions of the blood vessel, as follows: DLP, dorsal lateral plate; AA, aortic arches; PCV, posterior cardinal vein; ISV, intersomitic vein. The differentiation processes of blood vessel formation are indicated in the middle. The timing of definitive differentiation of the hemangioblast is not well understood, but it is known that the hemangioblasts located in the DLP differentiate into adult blood cells and vascular endothelial cells. Fli1 is known to be a marker of hemangioblasts (Liu et al.,2008). The start of vasculogenesis and angiogenesis is clearly observed as the expression of Xmsr in the AA or PCV and the expression of Xtie2 in the sprouting ISV, as determined by whole-mount in situ hybridization. We identify two novel potential marker genes, Ami and XRASGRP2, for the late differentiation of blood vessels, angiogenesis and/or maturation.


To control drainage and to ensure the excretion of waste throughout their lifespan, vertebrates develop three different forms of the kidney, that is, the pronephros, mesonephros, and metanephros. All three forms consist of a basic unit, the nephron, although they differ with respect to the number and organization of the nephrons within the kidneys. The metanephros, which is the most complex and terminal kidney form, contains one million nephrons. Most nephrons are integrated around the cortex of the metanephros. The waste collected by nephrons is excreted by means of a system of collecting ducts (Saxén,1987). The pronephros is a simple excretory organ of Xenopus larvae. Unlike the metanephros, the pronephros contains open nephrons that consist of tubules, a duct, and a glomus. The anlagen of the pronephros derived from the intermediate mesoderm are determined by 12 hr postfertilization, and pronephros differentiation is completed and the organ is functioning within 3 days (Nieuwkoop and Faber,1994). Kidney tissue induction assays have been developed for studies of kidney development. For studies of early pronephros development in Xenopus, a simple and efficient in vitro induction assay system is available (Moriya et al.,1993). As described previously, animal caps isolated from the late Xenopus blastulas represent useful and reliable materials for the study of cell induction. When animal caps were treated with activin and retinoic acid (RA), pronephros differentiation was observed in the explants (Fig. 6A). The in vitro–induced pronephros was histologically equivalent to the pronephros seen in normal embryos, and the expression patterns of the pronephros-related genes in the explants paralleled those of normal pronephros (Uochi and Asashima,1996; Brennan et al.,1999; Osafune et al.,2002). A transplantation experiment performed with this induced pronephros demonstrated that it was functionally active. The anlagen of the pronephric tubules were removed, and explants that had been treated with activin and RA were transplanted into the site from which the pronephric primordia had been removed (Fig. 6A). These transplanted explants suppressed the onset of edema, which is a consequence of pronephrectomy. The results of the transplantation experiments demonstrate that the induced pronephros in animal caps functions in the same way as a normal pronephros to maintain water balance in vivo (Chan et al.,1999). This induction method established for the animal cap assay has been applied to other cell induction experiments. Recently, it has been reported that ES cells treated with activin and RA differentiate into nephric tissues (Kim and Dressler,2005; Vigneau et al.,2007).

Figure 6.

In vitro induction of pronephros or pancreas from animal caps treated with activin and retinoic acid (RA). Animal caps were excised and treated with activin and/or RA in Steinberg's solution (SS) supplemented with 0.1% bovine serum albumin (BSA). Explants were washed with SS and cultured in SS with 0.1% BSA. All the treatments and culturing were performed at 20°C. A: When animal caps were excised and treated with 10 ng/ml activin and 104 M RA simultaneously for 3 hr, washed and cultured for 4 days. The explants mainly differentiated into pronephric tissues. After this explant was transplanted to a pronephrectomized Xenopus embryo of stage 20–25, the transplanted embryo developed normally. In contrast, the pronephrectomized Xenopus embryo showed edema at the tadpole stages. B: When animal caps were treated with 100 ng/ml activin for 1 hr, washed and cultured for 3–5 hr in SS with 0.1% BSA, and then treated with 104 M RA for 1 hr, the explants differentiated into pancreatic tissues at a high frequency.

The results obtained using the in vitro induction assay provide important information on the mechanism of pronephros development. Using a modified time-lag induction method for the animal cap assay, RA was shown to induce pronephros within a 12-hr window after activin treatment (Chan et al.,2000). Activin is known to induce the differentiation of mesodermal tissues in animal caps. Furthermore, pronephros development occurs from the late gastrula to the early neurula stage, which corresponds to the period from 6 hr to 12 hr after animal cap isolation. This result suggests that RA-mediated modification of mesodermal tissues is necessary for early pronephros development. Recently, it has been shown that the RA-induced signal is necessary for both pronephros and kidney development (Cartry et al.,2006).

To reveal the molecular mechanism underlying pronephros development, various studies have been conducted (Fig. 7). The data from these studies suggest that the expression of molecular signals is accompanied by morphologic changes in the pronephros. The Xenopus pronephros is derived from the intermediate mesoderm at the late gastrula stage. Lim-1 and Pax-8 are the first molecular signals to be expressed in the pronephros anlagen in the late gastrula stage. Lim-1, which is a LIM domain-containing homeobox transcription factor, is required for patterning of the anterior neural tissues (Taira et al.,1994). Dominant-negative mutant analysis and time-lag induction in the animal cap assay revealed that the expression of Lim-1 was necessary for tubule development (Chan et al.,2000). Other studies have implicated the combination of Lim-1 and Pax8 in the initiation of pronephros development. The overexpression of Lim-1, Pax-8, and Pax-2 leads to ectopic differentiation of tubules and enlargement of the ducts in the lateral mesoderm of Xenopus larvae (Carroll and Vize,1999). However, factors other than Lim-1 and Pax-8 are required for the initiation of pronephros development. Injection of Pax-8 and Lim-1 does not induce pronephros formation in isolated mesodermal tissues, such as activin-treated animal caps (Chan, unpublished data). Recently, it has been shown that Osr1 and Osr2 are necessary for early pronephros development, and that Osr1 and Osr2 can induce ectopic tubule differentiation (Tena et al.,2007). Because both Osr1 and Osr2 are expressed in the lateral mesoderm of the late gastrula, these proteins are considered to be candidates for the induction of pronephros development with Lim-1 and Pax-8.

Figure 7.

Schematic representation of pronephros development in X. laevis. The expression of molecular signals is accompanied by the differentiation of the tubule, duct, and glomus. The expression of Lim-1 and Pax-8 in the pronephric mesoderm determines the analgen of pronephros at stage 12.5. At stage 20, the anlagen of the pronephros thickens and becomes morphologically discernible, and the expression of Pax-2, Wnt-4, WT1, and other molecules begins at pronephros. By stage 25, the genes involved in the patterning and morphogenesis of the pronephros are activated. The pronephros is separated from the lateral plate, and further differentiation commences at the late tail-bud stage. The genes shown in blue, green, and red letters represent genes that are involved in the development of the tubule, duct, and glomus.

From the late neurula stage, the pronephros anlagen start to differentiate into the tubule, duct, and glomus. Changes in the shape of the pronephros are accompanied by the expression of Pax-2, Wnt-4, WT1, Xfz8, and other molecular signals. These molecules are considered to regulate the differentiation of the tubules, duct, and glomus (Wallingford et al.,1998; Saulnier et al.,2002; Satow et al.,2004). It has been suggested that factors, for example, XC3H-3b, present in the mesenchymal tissue that surrounds the pronephros are necessary for pronephros development (Kaneko et al.,2003). The expression of XC3H-3b is restricted to the mesodermal tissues that surround the pronephros. The morpholino antisense oligonucleotide-mediated knockdown of XC3H-3b resulted in reduced pronephros development and the expression of pronephric marker genes, such as Lim-1, Pax-2, Pax-8, Wnt-4, and WT1. Furthermore, XC3H-3b was found to regulate the signaling activities of FGF and bone morphogenetic protein (BMP; Chan, unpublished data). The TTP/TIS CCCH zinc-finger protein is reported to play a role in regulating the activity of JNK/MAPK signaling. Therefore, XC3H-3b may regulate tubulogenesis by means of JNK/MAPK signaling during Xenopus pronephros development. FGF and BMP are also reported to regulate pronephros development (Urban et al.,2006; Bracken et al.,2008). The FGF signal is necessary for the condensation and transition of the nephric mesenchyme during early tubulogenesis, and the BMP signal is suggested to regulate the differentiation of the tubule and duct. Concomitant with the expression of these genes, patterning of the tubules and extension of the ducts begin. From the late tail-bud stage, the anlagen of the tubule undergo further morphogenesis, and the entire pronephros is luminized and starts to function from stage 35/36.


A combination treatment with activin and RA has been shown to induce pancreatic tissues in animal caps (Moriya et al.,2000). In this method, following treatment with 100 ng/ml activin, the Xenopus animal caps were cultured for 5 hr, and then treated with 0.1 mM RA (Fig. 6B). As a result, pancreas differentiation from the animal caps occurred at a high frequency. The induced pancreas was found to have the normal pancreatic structure, including both exocrine cells and endocrine cells, as observed by transmission electron microscopy. Immunohistochemical staining with anti-insulin and anti-glucagon antibodies showed that the induced tissues secreted pancreatic hormones in a manner similar to a normal embryonic pancreas (Moriya et al.,2000). Using this in vitro pancreas-induction system, screening for novel pancreatic genes was carried out and four novel genes were found to be expressed specifically in the developing pancreas (Sogame et al.,2003).

RA and TGF-β signaling play key roles in the regulation of pancreatic organogenesis (Kim and Hebrok,2001; Kumar and Melton,2003). Studies on Xenopus embryos suggest that the level of TGF-β signaling is directly responsible for the segregation of the endodermal and the mesodermal germ layers. High-level signaling is required for specification of the endoderm, while a lower level of signaling promotes mesoderm formation (Jones et al.,1995; Henry et al.,1996; Clements et al.,1999; Piccolo et al.,1999; Sun et al.,1999; Yasuo and Lemaire,1999; Gritsman et al.,2000).

Studies of Xenopus animal caps have also suggested an important role for TGF-β signaling in the anterior–posterior (AP) patterning of the endoderm. In overexpression experiments with Xenopus, the maternal TGF-β signals (Vg1, Xnr1-2) induced ectopic expression of the anterior endoderm marker XlHbox8, while the posterior endoderm marker IFABP was not affected (Henry et al.,1996; Zorn et al.,1999). Inhibiting TGF-β pathways with a dominant-negative form of ActRIIB blocked the expression of endogenous anterior endoderm markers (Zorn et al.,1999). These results indicate that anterior endoderm specification requires endoderm-specific factors that are mediated by TGF-β signaling.

Another important mediator of pancreas differentiation is sonic hedgehog (Shh); its down-regulation by signals from the notochord is essential for initiating differentiation of the dorsal pancreas (Kim et al.,1997). In tissue culture experiments using endoderm isolated from chicken embryos, activin inhibited Shh expression and induced the transcription of pancreatic marker genes, thereby mimicking the effects of notochord signaling (Kim et al.,1997; Hebrok et al.,1998).

RA signaling is also involved in patterning along the AP axis of the endoderm during the late gastrula stage in zebrafish, Xenopus, and quail. While inhibition of RA signaling in the late stages has little effect on the expression of anterior mesodermal markers, it abrogates pancreatic marker expression. In addition, zebrafish and Xenopus embryos treated with exogenous RA showed enlargement of the pancreas, as well as enlargement of the liver in zebrafish, in the anterior direction (Stafford and Prince,2002; Chen et al.,2004; Stafford et al.,2004). A recent study reported that RA is sufficient to induce pancreas-specific gene expression in the dorsal but not the ventral endoderm (Pan et al.,2007). Mouse embryos deficient for the RA-synthesizing enzyme retinaldehyde dehydrogenase 2 (RALDH2), if rescued from early lethality by maternal RA, lack active RA signaling in the foregut region. In the resulting mutants, development of the lungs (Desai et al.,2004; Wang et al.,2006), stomach, duodenum, liver (Wang et al.,2006), and dorsal pancreas (Molotkov et al.,2005) is severely affected.

Recently, using an approach similar to that used for the induction of pancreas from the animal caps of Xenopus embryos, a few groups have succeeded in using RA and activin to differentiate from murine ES cells a pancreas-like tissue that contains endocrine cells, exocrine cells, and duct-like structures (Shi et al.,2005; Jiang et al.,2007; Nakanishi et al.,2007). These advances show that the experimental system using animal caps of Xenopus is a useful tool to screen for novel substances that promote the differentiation of vertebrate stem cells into endodermal tissues and organs.


As described above, activin A induces both mesodermal and endodermal tissues, such as those of the heart, kidney, and pancreas, from undifferentiated cells. In addition, activin treatment induces neural tissues. However, the rate of induction of neural tissues from a single cap treated with a single concentration of activin is not high. Therefore, to induce neural tissues, several caps treated with different concentrations of activin should be combined. Previously, we induced the production of a head-like structure from animal caps using a “sandwich” method, whereby activin-treated caps were wrapped in two nontreated caps. Importantly, the induced structure was determined by the length of exposure to activin. When an animal cap that had been incubated with activin A for 12 hr was sandwiched, a trunk–tail structure was induced, whereas incubation with activin A for 24 hr caused the sandwiched explants to differentiate into a head structure (Ariizumi and Asashima,1995). These results are very interesting in that they mirror the outcomes of a classic experiment, in which transplantation of a dorsal rip at the early gastrula stage into the ventral region of another gastrula induced a head-like structure, while transplantation of a dorsal tip at the late gastrula stage induced a trunk–tail (Spemann,1931).

To date, many of the genes involved in anterior neural structure formation during early embryogenesis have been identified (Fig. 8), and the molecular mechanisms underlying the induction of the head region have been clarified. For the determination of AP neural patterning, multiple regulatory events, which are mediated by several signal pathways, are important. For example, the inhibition of canonical Wnt signaling by means of several secreted factors is required for formation of the head structure, including the brain and eye structures (Glinka et al.,1997). Indeed, many of the genes that are specifically expressed in the head region are regulated by canonical Wnt signaling. In our microarray analysis, we identified many novel genes that showed head-specific transcription and the expression of which was altered by modulation of Wnt signaling. One of these genes, xCyp26c, is expressed in restricted sets of rhombomeres (Tanibe et al.,2008). The xCyp26c gene encodes an RA-metabolizing protein, which suggests crosstalk between Wnt signaling and RA signaling. Further information on AP neural patterning might be derived from extended analyses of these genes.

Figure 8.

List of genes implicated in brain development. Representative genes involved in brain formation are indicated for each developmental stage. During the gastrula stage, neural tissue is induced and the anterior–posterior (AP) axis is determined, resulting in fundamental neural tissue patterning. At the neurula stage, the anterior neural plate is subdivided into the telencephalon (red), diencephalon (yellow), mesencephalon (green), and rhombencephalon by the restricted expression of several brain-specific genes. At the tadpole stage, the brain region assumes a three-dimensional structure.


As mentioned in the previous sections, activin A-treated animal cap cells differentiate into various mesodermal tissues in a concentration-dependent manner (Ariizumi et al.,1991). An advantage of this in vitro induction system is that it generates high numbers of cells that have the same fate. This makes animal cap cells a valuable tool for screening genes that are required for organogenesis. In this section, we introduce a screening method for novel somitogenesis-related genes in a cDNA pool prepared from activin A-treated animal caps. We also provide information on (1) the molecular functions of the isolated gene products, and (2) the termination mechanism for vertebrate somite segmentation.

To construct the cDNA library, X. laevis animal caps were treated with activin A (100 ng/ml) for 1 hr, and after 5 hr in culture, the caps were sampled. Under this condition, animal caps differentiate into explants that consist of multiple types of tissues, including the notochord and muscle (Ariizumi et al.,1991). Because somites are known to differentiate into muscle cells, we selected this condition to screen for somitogenesis-related genes. Clones were selected based on mRNA localization in the presomitic mesoderms of the blastula and neurula of X. laevis. Through this screening process, several somitogenesis-related genes, such as paraxial protocadherin and mesogenin 1 (msgn1), as well as a novel gene bowline (Kim et al.,2000; Yoon et al.,2000; Kondow et al.,2006), were isolated. Two bowline-related genes, ledgerline and xRipply3, were also isolated from X. laevis (Chan et al.,2006; Hitachi et al.,2008). Counterparts have been identified in the zebrafish and mouse (Ripply1, 2, 3), human (DSCR6), and amphioxus (AmphiSom) (Shibuya et al.,2000; Kawamura et al.,2005; Li et al.,2006; Chan et al.,2007; Morimoto et al.,2007).

For formation of the somite boundaries, bowline, ledgerline, and their counterpart in zebrafish, ripply1, are required (Kawamura et al.,2005; Chan et al.,2006; Kondow et al.,2006). Knockout of the murine Ripply2 gene results in segmental defects and rostralization of the somites (Chan et al.,2007; Morimoto et al.,2007). Therefore, the bowline/ripply family genes are commonly required for somite boundary formation in vertebrates.

The Bowline/ripply family proteins mediate interactions between Tbx6/tbx24, which are transcriptional activators that are commonly required for somite segmentation program (Nikaido et al.,2002; Yasuhiko et al.,2006), and the transcriptional corepressors XGrg-4/Groucho (Kondow et al.,2007; Kawamura et al.,2008). At the end of somitogenesis, it is likely that this interaction results in the suppression of Tbx6/tbx24-dependent transcription of genes that are essential for the somite segmentation program. The mechanism for termination of the somite segmentation program is thus described by the identification and characterization of the bowline genes isolated using the in vitro induction system. Further analyses of the molecular mechanisms of vertebrate development will be facilitated by the ability of this system.


Using Xenopus animal caps, various protocols for the induction of tissues and organs have been established. The differentiation of mesodermal/endodermal tissues from animal caps through dose-dependent induction by activin is considered to mirror the mesoderm induction and organizer formation that occur during normal development through activation of the activin/Nodal signal pathway. RA is known to have posteriorization activity in the differentiation of mesodermal and neural tissues (Durston et al.,1989; Ruiz i Altaba and Jessell,1991). In our pancreatic tissue induction system, for example, the animal caps differentiate into anterior endodermal tissues after treatment with high concentrations of activin, whereas the direction of differentiation may be modified subsequently by the posteriorization activity of RA. Although the molecular mechanisms for these induction systems are not fully understood, these systems are useful for investigating the mechanisms of tissue/organ differentiation and for screening functional genes or substances.

We and other researchers have described the in vitro induction of mammalian tissues. In studies using mammalian stem cells, we reconfirmed that the advantages of the animal cap assay are that it is simple and uses serum-free cultures. In mammalian stem cell cultures, serum is normally used as a supplement and source of nutrition, although its use in assays designed to determine the level of cell differentiation confers poor reproducibility, as serum contains many soluble factors, such as growth factors, vitamins, and extracellular matrices (ECMs). Recently, we established a serum-free medium for murine ES cells (Furue et al.,2005), to avoid the above-mentioned problems. However, compared with Xenopus embryos, early embryonic development in the mouse is slow and very difficult to observe and manipulate through embryonic surgery. In summary, the X. laevis model is valuable for studies of embryonic development and organogenesis and contributes significantly to the fields of developmental biology and regenerative medicine.


The authors acknowledge the financial assistance provided by Grants- in-Aid for Scientific Research from the Ministry of Education, Science, Sports, Culture and Technology of Japan, and by ICORP (International Cooperative Research Project) of the Japan Science and Technology Agency.