In vitro pancreas formation from Xenopus ectoderm treated with activin and retinoic acid

Authors

  • Naomi Moriya,

    1. Core Research for Evolutional Science and Technology, Japan Science and Technology Corporation,
    Search for more papers by this author
  • Shinji Komazaki,

    1. Department of Anatomy, Saitama Medical School, 38 Morohongou, Moroyama-cho, Iruma-gun, Saitama 350-0495, Japan.
    Search for more papers by this author
  • Shuji Takahashi,

    1. Core Research for Evolutional Science and Technology, Japan Science and Technology Corporation,
    Search for more papers by this author
  • Chika Yokota,

    1. Core Research for Evolutional Science and Technology, Japan Science and Technology Corporation,
    Search for more papers by this author
  • Makoto Asashima

    1. Core Research for Evolutional Science and Technology, Japan Science and Technology Corporation,
    2. Department of Life Sciences (Biology), Graduate School of Arts and Sciences, The University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902 and
    Search for more papers by this author
    • *Author to whom all correspondence should be addressed.


Abstract

In the present study, isolated presumptive ectoderm from Xenopus blastula was treated with activin and retinoic acid to induce differentiation into pancreas. The presumptive ectoderm region of the blastula consists of undifferentiated cells and is fated to become epidermis and neural tissue in normal development. When the region is isolated and cultured in vitro, it develops into atypical epidermis. Isolated presumptive ectoderm was treated with activin and retinoic acid. The ectoderm frequently differentiated into pancreas-like structures accompanied by an intestinal epithelium-like structure. Sections of the explants viewed using light and electron microscopy showed some cells clustered and forming an acinus-like structure, including secretory granules. The pancreas-specific molecular markers insulin and XlHbox8 were also expressed in the treated explants. The pancreatic hormones, insulin and glucagon, were detected in the explants using immunohistochemistry. Therefore, sequential treatment with activin and retinoic acid can induce presumptive ectoderm to differentiate into a morphological and functional pancreas in vitro. When ectoderm was immediately treated with retinoic acid after treatment with activin, well-differentiated pronephric tubules were seen in a few of the differentiated pancreases. Treatment with retinoic acid 3–5 h after activin treatment induced frequent pancreatic differentiation. When the time lag was longer than 15 h, the explants developed into axial mesoderm and pharynx. The present study provides an effective system for analyzing pancreas differentiation in vertebrate development.

Introduction

Most vertebrates have a pancreas that consists of both endocrine and exocrine cells. This organ exhibits a histology and manner of development that is common to most vertebrates. Dorsal and ventral pancreatic buds arise from anterior endoderm during development, then fuse to become pancreas (reviewed by Slack 1995). Developing endodermal organs require signals from mesoderm located close to the endoderm ( Okada 1960; Yasugi 1993) and mesenchymal signals play an important role in pancreas organogenesis ( Golosow & Grobstein 1962). A notochord deletion study in the chick embryo has revealed that pancreas development requires the presence of notochord ( Kim et al. 1997 ). Notochord represses the expression of sonic hedgehog (Shh) in adjacent pancreatic endoderm, allowing the expression of differentiated pancreatic markers ( Hebrok et al. 1998 ; Kim & Melton 1998). However, notochord does not induce expression of these markers in endoderm posterior to the pancreatic anlage.

The expression of the homeobox gene pdx-1 (also known as ipf-1) is necessary for pancreas development in the mouse embryo. pdx-1–/– mice lack a pancreas and die postnatally ( Jonsson et al. 1994 ). These mutant mice, however, develop a pancreas bud and express glucagon ( Offield et al. 1996 ). XlHbox8, the Xenopus homolog of mouse Pdx-1, is expressed in the epithelial cells of pancreatic anlagen ( Wright et al. 1989 ) and transforming growth factor (TGF)-β signaling mediates the expression of XlHbox8 in endoderm ( Henry et al. 1996 ).

Retinoic acid regulates embryonic patterning along the anteroposterior axis ( Durston et al. 1989 ; Ruiz i Altaba & Jessell 1991a, b; Lopez & Carrasco 1992; Moriya et al. 1993 , 1998). Recently, we have reported that Xenopus dorsal lip cells treated with retinoic acid differentiate into pancreas. Although isolated dorsal lip cultured alone become axial mesoderm and pharynx, explants treated with retinoic acid develop a pancreas-like structure and express XlHbox8 and insulin (Moriya et al. 2000). Dorsal lip cells have self-differentiation activity, while presumptive ectoderm remains a mass of undifferentiated cells when cultured alone. Activin dose-dependently induces endomesoderm in presumptive ectoderm ( Asashima et al. 1990 ; Green & Smith 1990; Ariizumi et al. 1991 ; Asashima et al. 1999 ) and at high concentrations induces ectoderm to become dorsal lip-like cells ( Ninomiya et al. 1999 ).

In the present study, we treated presumptive ectoderm with activin and retinoic acid in an attempt to induce differentiation into pancreas. The results generated a simple model system suitable for analysis of the mechanisms of pancreas development in vitro.

Materials and Methods

Preparation of embryos and solutions

Xenopus laevis eggs were obtained as described previously ( Moriya et al. 1998 ). Human recombinant activin A (Ajinomoto Co. Inc., Kawasaki, Japan) was diluted with Steinberg’s solution containing 0.1% bovine serum albumin (BSA-SS) at a concentration of 100 ng/mL. All-trans-retinoic acid (CAT no. R2625; Sigma Chemical Co., St Louis, MO, USA) was dissolved in absolute ethanol at a concentration of 10–2M. For treatment, the solution was diluted with BSA-SS to concentrations varying from 10–9 to 10–4M.

Treatment and culture of explants

Presumptive ectoderm was isolated from stage 9 embryos ( Nieuwkoop & Faber 1956). For transient treatment, ectoderm was placed in an activin and retinoic acid mixture for 1 h. Ectoderm was twice washed with BSA-SS and then cultured in the same solution at 20°C. For continuous treatment, ectoderm was cultured in an activin and retinoic acid mixture at 20°C. For time-lag treatment, ectoderm was placed in an activin solution for 1 h and washed twice with BSA-SS. Ectoderm was kept in BSA-SS for various periods (time lag, 0–25 h) and then put in retinoic acid solution for 1 h. After washing twice ectoderm was cultured in BSA-SS at 20°C ( Fig. 1).

Figure 1.

Methods of ectoderm treatment and culture. Presumptive ectoderm (0.4 mm × 0.4 mm) was isolated from the Xenopus late blastula at stage 9. For short-term treatment, ectoderm was put in an activin/retinoic acid mixture for 1 h, washed twice and cultured in Steinberg’s solution including 0.1% bovine serum albumin (BSA-SS) for 10 days. For continuous treatment, ectoderm was cultured in an activin/retinoic acid (RA) mixture. For time-lag treatment, ectoderm was first put in activin solution for 1 h and washed twice with BSA-SS. Explants were kept in BSA-SS for each time lag, varying from 0 to 25 h, and then put in RA solution for 1 h. After being washed twice they were cultured in BSA-SS. Differentiation of explants was examined by microscopy, reverse transcription–polymerase chain reaction and immunohistochemistry.

Histology and reverse transcription–polymerase chain reaction analysis

The explants cultured for 10 days were examined histologically using light and electron microscopy as previously described (Moriya et al. 2000). Expression of molecular markers in the explants cultured for 3 days was also analyzed as previously described ( Moriya et al. 1998 ).

Immunohistochemistry

The explants cultured for 10 days were fixed with 0.1 M morpholinopropanesulfonic acid/2 m M EGTA/1 m M MgSO4/3.7% formaldehyde. After washing with phosphate-buffered saline (PBS; pH 7.4), blocking was carried out with PBS containing 2% skim milk and 1% bovine serum albumin. Explants were incubated overnight with the following primary antibodies: anti-insulin guinea-pig IgG (1:1000, CAT no. A0564; Dako Corporation, Carpinteria, CA, USA), anti-insulin guinea-pig IgG (1:1000, CAT no. 4010-01; Linco Research Inc., St Charles, MO, USA) and antiglucagon mouse IgG (1:2000, CAT no. G-2654; Sigma Chemical Co.). After washing with PBS, they were incubated with the following secondary antibodies coupled with alkaline phosphatase: antiguinea-pig IgG antibody (1:500; CAT no. 61-4622, Zymed Laboratories Inc., San Francisco, CA, USA) and antimouse IgG antibody (1:500, CAT no. AQ160A; Chemicon International Inc., Temecula, CA, USA). They were washed and stained blue with nitrobenzoic acid and bromochloroindolylphosphate in alkaline phosphate buffer (pH 9.5). They were fixed again with Bouin’s fluid, dehydrated through a graded series of ethanol and xylene, embedded in paraffin and cut into 10 μm sections. Pigments in the cells were bleached with methanol containing 10% H2O2 and observed.

Results

Transient treatment with activin and retinoic acid

Untreated ectoderm was cultured for 4 days and formed atypical epidermis ( Fig. 2A). The explants survived for 4–5 days. Ectoderm treated with activin (100 ng/mL) alone developed into axial mesoderm and anterior endoderm, such as notochord, muscle and pharynx ( Fig. 2B). Some neural tissues were also observed. When ectoderm was treated with a mixture of 100 ng/mL activin and retinoic acid at various concentrations (10–7–10–4M), retinoic acid had a dose-dependent effect on the pattern of formation of mesodermal tissue. As the concentration of retinoic acid was increased, the rate of notochord formation first decreased. Muscle formation was suppressed as the concentration further increased, while pronephric tubule formation was gradually induced ( Table 1). The frequency of pronephros formation reached a plateau around retinoic acid concentrations of 10–5–10–4M (67–61%). Some intestinal epithelium-like structures were induced at high concentrations of retinoic acid (23% at 10–4M). Some pancreas-like structures were also observed at high retinoic acid concentrations, but only at a low frequency (16% at 10–4M).

Figure 2.

Light micrographs of explants. Ectoderm was isolated from stage 9 blastulae and treated as follows. Ectoderm was cultured for 10 days except for untreated ectoderm, which was only cultured for 4 days and examined histologically. (A) Untreated presumptive ectoderm cultured in Steinberg’s solution including 0.1% bovine serum albumin (BSA-SS). (B) Ectoderm treated with activin alone (100 ng/mL) for 1 h. (C) Ectoderm treated with an activin/retinoic acid mixture (100 ng/mL and 10–4M, respectively) for 1 h. (D) Ectoderm first treated with activin (100 ng/mL) for 1 h, kept in BSA-SS for 5 h and then treated with retinoic acid (10–4M) for 1 h. (E) Ectoderm first treated with activin (100 ng/mL) for 1 h, kept in BSA-SS for 25 h and then treated with retinoic acid (10–4M) for 1 h. int, Intestinal epithelium-like structure; neu, neural tissue; not, notochord; pan, pancreas- like structure; pha, pharyngeal epithelium-like structure; pro, pronephric tubules. Bar, 100 μm.

Table 1.  Differentiation of presumptive ectoderm treated with activin and retinoic acid (short-term treatment)
 Retinoic acid concentration ( M)
 010–710–610–510–4
  1. Figures in parentheses indicate the number of specimens as a percentage of the total number.

No. specimens3129333031
Atypical epidermis0 (0)0 (0)0 (0)0 (0)0 (0)
Epidermis30 (97)25 (86)28 (85)26 (87)25 (81)
Neural tissue12 (39)12 (41)15 (45)10 (33)5 (16)
Notochord4 (13)2 (7)0 (0)0 (0)0 (0)
Muscle25 (81)26 (90)18 (55)6 (20)1 (3)
Pronephric tubules7 (23)6 (21)11 (33)20 (67)19 (61)
Pharyngeal epithelium-like structure12 (39)18 (62)23 (70)19 (63)23 (74)
Intestinal epithelium-like structure0 (0)0 (0)2 (6)5 (17)7 (23)
Pancreas-like structure0 (0)1 (3)0 (0)2 (7)5 (16)

Continuous treatment with activin and retinoic acid

Short-term treatment of ectoderm with activin and retinoic acid for 1 h immediately after isolation affected the differentiation of mesoderm, but not endoderm. This may be because determination of endodermal tissues occurs later than that of mesodermal tissues. To eliminate any possible effect of differential timing of determination of cell types, we cultured ectoderm continuously in the presence of 100 ng/mL activin and retinoic acid at various concentrations ranging from 10–9 to 10–4M. As the concentration of retinoic acid increased, the rate of notochord and muscle formation decreased and pronephric tubule formation increased ( Table 2). Intestinal epithelium-like structures developed at a low frequency (20%) at 10–5M retinoic acid. No pancreas-like structures appeared at any concentration of retinoic acid. All explants cultured in medium containing 10–4M retinoic acid died during the culture period.

Table 2.  Differentiation of presumptive ectoderm treated with activin and retinoic acid (continuous treatment)
  Retinoic acid concentration ( M)
 010–910–810–710–610–5
  1. Figures in parentheses indicate the number of specimens as a percentage of the total number.

No. specimens101110121110
Atypical epidermis0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
Epidermis9 (90)11 (100)10 (100)10 (83)10 (91)5 (50)
Neural tissue10 (100)8 (73)3 (30)7 (58)7 (64)0 (0)
Notochord3 (30)3 (27)1 (10)2 (17)0 (0)0 (0)
Muscle10 (100)11 (100)9 (90)10 (83)9 (82)2 (20)
Pronephric tubules0 (0)0 (0)0 (0)0 (0)7 (64)5 (50)
Pharyngeal epithelium-like structure5 (50)7 (64)8 (80)9 (75)9 (82)9 (90)
Intestinal epithelium-like structure0 (0)0 (0)0 (0)0 (0)0 (0)2 (20)
Pancreas-like structure0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)

Time-lag treatment with activin and retinoic acid

We treated ectoderm cultures first with activin and then with retinoic acid and examined the effects of various time lags between each treatment. Presumptive ectoderm was first exposed to activin (100 ng/mL) for 1 h and then to retinoic acid (10–4M) for 1 h at various times after the activin treatment. When ectoderm was exposed to activin and retinoic acid at the same time (activin/retinoic acid mixture) or sequentially with no time lag (ectoderm was treated with retinoic acid immediately after activin treatment), ectoderm frequently differentiated into pronephric tubules ( Fig. 2C; Table 3). With a time lag of 3–5 h, the frequency of pronephros formation decreased and pancreas-like structures appeared in over 80% of explants ( Fig. 2D). When the time lag was greater than 15 h, notochord and pharynx were mainly observed ( Fig. 2E).

Table 3.  Differentiation of presumptive ectoderm treated with activin and retinoic acid (time lag treatment)
 Time lag (h)
  * 0351525
  1. All samples were treated with activin for 1 h. All samples were treated with retinoic acid for 1 h after the time lag specified, except for *treatment with a mixture of activin and retinoic acid and treatment with activin alone. Figures in parentheses indicate the number of specimens as a percentage of the total number.

No. specimens59653135383365
Atypical epidermis0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
Epidermis45 (76)48 (74)13 (42)18 (51)28 (74)29 (88)59 (91)
Neural tissue5 (8)0 (0)0 (0)0 (0)1 (3)1 (3)32 (49)
Notochord0 (0)0 (0)0 (0)2 (6)5 (13)17 (52)34 (52)
Muscle3 (5)8 (12)12 (39)8 (23)23 (61)16 (48)37 (57)
Pronephric tubules38 (64)39 (60)10 (32)4 (11)2 (5)0 (0)5 (8)
Pharyngeal epithelium-like structure50 (85)59 (91)30 (97)27 (77)32 (84)29 (88)37 (57)
Intestinal epithelium-like structure11 (19)24 (37)19 (61)29 (83)9 (24)3 (9)3 (5)
Pancreas-like structure20 (34)27 (42)25 (81)29 (83)8 (21)1 (3)0 (0)

The optimal condition for induction of pancreas differentiation was a time lag of 3–5 h between activin and retinoic acid treatment, which led to development of pancreas-like structures in 80% or more of explants. The frequency of pancreas formation increased up to a time lag of 5 h, but with a time lag longer than 5 h the frequency decreased as the interval increased ( Table 3).

Pancreas-like structures in the explants displayed acinus-like structures under light microscopy. Some cell aggregates had a lumen. These cells had a number of features in common with pancreatic cells in vivo. The shapes of the cells constituting the acinus were pyramidal and their nuclei were located on the opposite side to the lumen. The cytoplasm of cells on the luminal side were stained well by eosin ( Fig. 2D).

Electron microscopy

Explants were treated with retinoic acid 5 h after activin treatment and then cultured for 10 days and examined by electron microscopy. Many cell clusters that looked like acini covered with the basal lamina were observed in the explants. The clusters had a lumen in the center and the nuclei were located opposite the lumen. There were many large and electron-dense granules (0.2–1.0 μm in diameter) in the cytoplasm of the luminal side ( Fig. 3A). These cells resembled the cells of exocrine pancreas in vivo ( Lozano et al. 1999 ). In addition to the acinus-like structures, two types of cells with differing secretory granules were found around the acinus-like structures. One contained electron-dense secretory granules (0.1–0.3 μm in diameter; Fig. 3B) similar to those in the glucagon cells in the islets of Langerhans and another contained secretory granules (0.2–1.0 μm in diameter; Fig. 3C) with an electron-dense core similar to those in insulin cells ( Leone et al. 1976 ; Lozano et al. 1999 ).

Figure 3.

Electron micrographs of explants treated with activin and retinoic acid. Isolated ectoderm was first treated with activin (100 ng/mL) for 1 h, kept in Steinberg’s solution including 0.1% bovine serum albumin (BSA-SS) for 5 h and then treated with retinoic acid (10–4M) for 1 h. Samples were cultured for 10 days and the sections were examined by electron microscopy. (A) An acinus-like structure in the explant. Large electron-dense granules (arrows) were observed. lu, Lumen of the acinus-like structure. (B) B cell-like cells containing small granules (arrows) with a round or irregular electron-dense core. (C) Pancreatic A cell-like cells containing small granules (arrows) with a round dense core. Bars, 5 μm (A); 1 μm (B,C).

Expression of pancreatic molecular markers

Total RNA was extracted from ectoderm explants treated with activin and then retinoic acid and cultured for 3 days. The expression of the pancreas-specific markers insulin and XlHbox8 was determined by reverse transcription–polymerase chain reaction (RT-PCR). Untreated ectoderm and retinoic acid-treated ectoderm did not express either marker ( Fig. 4, lanes 1 and 3). Activin induced some very low expression of these markers ( Fig. 4, lane 2) and increased expression was induced in the explants simultaneously treated with activin and retinoic acid ( Fig. 4, lane 4). Both markers were strongly expressed in explants treated with activin and then 5 h later with retinoic acid ( Fig. 4, lane 5), but expression was repressed in explants treated with activin and then retinoic acid 25 h later ( Fig. 4, lane 6).

Figure 4.

Expression of molecular markers in explants. Isolated ectoderm was treated as follows and cultured for 3 days. Total RNA was extracted and analyzed by reverse transcription– polymerase chain reaction. Lane 1, untreated ectoderm; lane 2, ectoderm treated with activin alone (100 ng/mL) for 1 h; lane 3, ectoderm treated with retinoic acid alone (10–4M) for 1 h; lane 4, ectoderm treated with activin (100 ng/mL) and retinoic acid simultaneously for 1 h; lane 5, ectoderm treated with activin (100 ng/mL) for 1 h, put in Steinberg’s solution with 0.1% bovine serum albumin (BSA-SS) for 5 h and then treated with retinoic acid (10–4M) for 1 h; lane 6, ectoderm treated with activin (100 ng/mL) for 1 h, put in BSA-SS for 25 h and then treated with retinoic acid (10–4M) for 1 h. The expression of EF-1α is shown as a loading control.

Immunohistochemistry

Immunoreactivity for the classic islet hormones insulin and glucagon is already found in the pancreas of larva at stage 41 ( Shuldiner et al. 1991 ; Maake et al. 1998 ). We confirmed insulin and glucagon immunoreactivity in the pancreas of larvae at stage 46 (data not shown).

Ectoderm explants treated with activin and then 5 h later with retinoic acid were cultured for 10 days and examined immunohistochemically. The sibling embryo reached stage 46. Specific immunoreactivity for insulin was observed in many explants ( Fig. 5B,C) and specific glucagon immunoreactivity was seen in even more cells ( Fig. 5D,E).

Figure 5.

Immunohistology of the explants treated with activin and retinoic acid. Isolated ectoderm was treated with activin (100 ng/mL) for 1 h, kept in Steinberg’s solution including 0.1% bovine serum albumin (BSA-SS) for 5 h and then treated with retinoic acid (10–4M) for 1 h. Explants were cultured for 10 days and examined immunohistologically. Ectoderm incubated with secondary antibody alone (antiguinea-pig IgG), as a negative control for B and C, was not stained. (B,C) Ectoderm incubated with anti-insulin antibody and secondary antibody (antiguinea-pig IgG). (D) Ectoderm incubated with secondary antibody alone (antimouse IgG) as a negative control for E and F also showed no staining. (E,F) Ectoderm incubated with anti-insulin antibody and secondary antibody (antimouse IgG). The secondary antibodies were coupled with alkaline phosphatase and visualized with nitrobenzoic acid/bromochloroindolylphosphate (arrows). Bar, 100 μm.

Discussion

Effects of retinoic acid on tissue differentiation

Retinoic acid has a posteriorization/lateralization activity on the central nervous system and mesodermal tissue ( Durston et al. 1989 ; Ruiz i Altaba & Jessell 1991a, b; Lopez & Carrasco 1992; Moriya et al. 1993 , 1998). Treatment of the isolated dorsal lip of Xenopus early gastrula with retinoic acid induces pancreas differentiation (Moriya et al. 2000). Activin at high concentrations induces isolated presumptive ectoderm of Xenopus blastula to differentiate into dorsal lip-like cells that also possess inductive activity similar to genuine dorsal lip ( Ninomiya et al. 1999 ). Thus, this suggests that treatment with activin and retinoic acid may induce presumptive ectoderm to differentiate into pancreas.

Ectoderm treated with activin alone became axial mesoderm and anterior endoderm, such as notochord, muscle and pharynx ( Table 1; Fig. 2B). These tissues are normally derived from the dorsal lip region in vivo, suggesting that activin at the concentration used here induces dorsal lip-like cells in ectoderm explants. Treatment with activin at this concentration and retinoic acid at various concentrations for 1 h, however, did not induce pancreas formation. Retinoic acid had a dose-dependent effect on the pattern of differentiation of mesodermal tissue ( Table 1). Notochord formation was first repressed and muscle formation was then repressed, while the formation of pronephros was induced as the concentration of retinoic acid increased. Retinoic acid in combination with activin had a small effect on endodermal tissue differentiation. In normal development, endoderm undergoes morphogenesis, and therefore probably determination, later than mesoderm ( Nieuwkoop & Faber 1956; Chalmers & Slack 1998). Short-term treatment with differentiating factors immediately after isolation of ectoderm may therefore only correspond to the timing of determination of mesodermal tissues, but not endodermal tissues. To avoid this effect, we cultured ectoderm continuously in the presence of activin and retinoic acid, but obtained similar results to short-term treatment. However, with continuous treatment, ectoderm explants were exposed to retinoic acid during the determination of endodermal tissues, which did not alter the pattern of endodermal differentiation ( Table 2). In contrast, when isolated dorsal lips were treated with retinoic acid for 1–3 h immediately after isolation they differentiated into pancreas (Moriya et al. 2000), but dorsal lips had already begun to differentiate into endomesoderm. Therefore, when undifferentiated cells are exposed to activin and retinoic acid at the same time, they may be induced to become lateral mesoderm, but not endomesoderm. In addition, retinoic acid may only be able to induce pancreas in cells that have already differentiated into endomesoderm.

Effects of the time lag between activin and retinoic acid treatments

To examine the effect of retinoic acid on endomesoderm, presumptive ectoderm was first exposed to activin (100 ng/mL) for 1 h and was then exposed to retinoic acid (10–4M) for 1 h at various times after the activin treatment. A time lag of 3–5 h between activin and retinoic acid treatment induced frequent pancreas formation in more than 80% of explants ( Table 3). These findings indicate that once ectoderm has been induced to differentiate by activin, there is a window of 3–5 h in which retinoic acid can induce a very high rate of differentiation of pancreas-like structures. However, beyond this window retinoic acid has little effect and ectoderm differentiation is driven by activin alone, leading to the development of axial mesoderm and pharynx. Therefore, retinoic acid may only be able to drive endomesoderm to form pancreatic tissues and once ectoderm has differentiated into endodermal tissue, which occurs at approximately 15 h after initiation of activin treatment, retinoic acid can no longer affect these cells.

It has been reported that activin alone induces Xenopus ectoderm to express XlHbox8, a marker of anterior endoderm including the pancreas anlage ( Gamer & Wright 1995). XlHbox8 is generally expressed in the posterior foregut region at stages 33–41 ( Wright et al. 1989 ). In the present study, morphological analysis showed that explants treated with activin alone differentiated into pharynx, an anterior endodermal tissue, and did not develop pancreas-like structures ( Tables 1–3). Thus, these findings suggest that activin alone may induce anterior endoderm, but not pancreas.

Acinus-like structures were observed using light microscopy in ectodermal explants treated with activin and then 5 h later with retinoic acid. Cells resembling exocrine or endocrine cells containing secretory granules of normal embryonic pancreas ( Leone et al. 1976 ) were seen under electron microscopy. Production of the islet hormones, insulin and glucagon, has already begun in larva at stage 41 ( Shuldiner et al. 1991 ; Maake et al. 1998 ). In the present study, ectoderm treated with activin and retinoic acid with a 5 h time lag produced both insulin and glucagon ( Fig. 5), suggesting that these explants had differentiated into endocrine pancreas, which was functionally similar to normal pancreas in vivo.

The pancreatic tissues that developed in the explants were covered by intestinal epithelium-like tissue in most cases. Endodermal epithelium forms a duct from mouth to anus and its morphology gradually varies along this duct in vivo. In the present study, epithelia observed in explants were classified into two types, pharyngeal epithelium-like structures and intestinal epithelium-like structures. Intestinal epithelium is very thick, while pharynx is thin ( Chalmers & Slack 1998). These two types of epithelium were distinguished according to the height/width ratio of epithelial cells. A layer with a ratio lower than 3 was classified as a pharyngeal epithelium-like structure and a layer with a ratio higher than 3 was scored as an intestinal epithelium-like structure. Ectoderm treated with activin alone formed a pharyngeal epithelium-like structure, while cells treated with activin and retinoic acid differentiated into pancreas and intestinal epithelium-like structures. Pharynx locates anteriorly, while pancreas and intestine locate posterior to the pharynx in vivo.

It has been reported that Xenopus blastulae treated with retinoic acid lose head structures ( Durston et al. 1989 ). Retinoic acid can repress anterior markers and induce posterior markers ( Ruiz i Altaba & Jessell 1991a, b; Lopez & Carrasco 1992; Kolm et al. 1997 ). Retinoic acid and its receptor are localized to the posterior of the embryo ( Ellinger-Ziegelbauer & Dreyer 1991; Chen et al. 1994 ). Thus, it is possible that ectoderm is first initiated into anterior endomesoderm by activin and then the cells are modified posteriorly by retinoic acid and differentiate into pancreas. Pancreatic structures actually appeared to be accompanied by intestinal epithelium under microscopy ( Fig. 2D).

The effect of retinoic acid on endoderm in Xenopus embryos has been reported previously ( Zeynali & Dixon 1998). Xenopus embryos at stages 22–33 were exposed to retinoic acid, which led to abnormal morphogenesis in the digestive tract, while pancreas developed normally. We treated Xenopus blastula with retinoic acid and looked for pancreas induction in vivo. The embryo lacked a head structure, but all endodermal organs developed normally. There were no organ-specific defects or enlargement with retinoic acid treatment. The location of endodermal organs was, however, abnormal; the organs were located close to each other along the anteroposterior axis (data not shown). These findings suggest that retinoic acid, which drives posteriorization, can only induce correct pancreas formation in ectoderm that has begun to differentiate into anterior endoderm, via, for example, treatment with activin.

Influence of mesoderm on pancreas development

Moriya et al. (2000) have induced pancreas formation in dorsal lip explants by retinoic acid treatment and have also demonstrated that the frequency of pancreas formation increases as the concentration of retinoic acid increases and that the pancreas is covered by intestinal epithelium. Pancreas was accompanied by notochord in these dorsal lip explants, whereas notochord formation was repressed under the conditions that frequently induced pancreas in ectodermal explants ( Table 3). Notochord induces pancreas in the prepancreatic endoderm chick embryo ( Kim et al. 1997 ). Thus, it is possible that the pancreas that developed in dorsal lip was induced by notochord present in the same explant. In the present study, however, pancreas was induced without notochord. Therefore, notochord formation is not necessary for pancreas differentiation.

In the chick embryo, notochord has been shown to repress Shh in prepancreatic endoderm and permit pancreas development; treatment with an Shh inhibitor causes a presumptive stomach region to differentiate into pancreas ( Kim et al. 1997 ; Hebrok et al. 1998 ; Kim & Melton 1998). Candidate notochord-derived inhibitors of Shh expression in prepancreatic endoderm include Shh, activin B and fibroblast growth factor (FGF)2 ( Kim & Melton 1998). It is possible that activin A in the present study may also have inhibited the expression of Shh. Activin alone, however, did not induce pancreas in ectoderm ( Tables 1–3); pancreas differentiation in ectoderm requires both activin and retinoic acid in this system.

Some studies of the effect of retinoic acid on development have suggested that retinoic acid affects Shh expression in the chick limb bud and mammal craniofacial morphogenesis ( Helms et al. 1994 ; Ogura et al. 1996 ; Helms et al. 1997 ). In Xenopus ectoderm, retinoic acid may also have an effect on Shh expression and pancreas differentiation.

In summary, pancreas may be frequently induced by activin and retinoic acid treatment of Xenopus presumptive ectoderm. This is a simple and effective in vitro system for analysis of the mechanisms of pancreas differentiation and endodermal patterning.

Acknowledgements

This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan and by Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Corporation.

Ancillary