Development of the endoderm and gut in medaka, Oryzias latipes

Authors


*Author to whom all correspondence should be addressed.
Email: htakeda@biol.s.u-tokyo.ac.jp

Abstract

We performed an extensive analysis of endodermal development and gut tube morphogenesis in the medaka embryo by histology and in situ hybridization. The markers used in these analyses included sox17, sox32, foxA2, gata-4, -5, -6 and shh. sox17, sox32, foxA2, and gata-5 and -6 are expressed in the early endoderm to the onset of gut tube formation. Sections of medaka embryos hybridized with foxA2, a pan-endodermal marker during gut morphogenesis, demonstrated that gut tube formation is initiated in the anterior portion and that the anterior and mid/posterior gut undergo distinct morphogenetic processes. Tube formation in the anterior endoderm that is fated to the pharynx and esophagus is much delayed and appears to be independent of gut morphogenesis. The overall aspects of medaka gut development are similar to those of zebrafish, except that zebrafish tube formation initiates at both the anterior and posterior portions. Our results therefore describe both molecular and morphological aspects of medaka digestive system development that will be necessary for the characterization of medaka mutants.

Introduction

The digestive system provides a complex but rich organ system for the study of visceral pattern formations during development. In addition to being the site of nutrient digestion and absorption, the gut endoderm plays an essential role in inducing other organs such as the heart (Lough & Sugi 2000), and in providing the anlagen and the signals to form the many endoderm derived organs such as thyroid, liver and pancreas. The development of the digestive system has been extensively studied in amniotes (for a review, see Grapin-Botton & Melton 2000; Roberts 2000) in which gut formation initiates as a sequence of two invaginations of the endodermal sheet, one at the anterior end to form the foregut, followed by a posterior invagination to form the hindgut. Subsequently, the splanchnic mesoderm, which is closely associated with the endoderm, undergoes muscle differentiation around the endoderm. As development proceeds, axial growth of the foregut and hindgut from intervening endoderm, coupled with morphogenesis of the midgut, completes the formation of the continuous gut tube. However, different features of digestive tract development have been recently reported in zebrafish. In zebrafish, the most anterior portion (pharynx and esophagus) develops separately from the more posterior gut tube, whereas in amniotes the anlagen of the pharynx, esophagus and intestine primordia arise from the foregut. In addition, gut tube formation involves the rearrangement of newly polarized cells rather than the folding of an endodermal sheet in amniotes (Wallaec & Pack 2003).

These results suggest that teleost gut development has unique aspects in terms of its morphogenesis, although the molecular gene networks leading to endodermal and gut development seem to be conserved among vertebrates. For a better understating of digestive development in fish as a whole, and to provide basic knowledge for effective medaka mutant analysis, we have undertaken in our current study a thorough characterization of endodermal and gut development in the medaka embryo using key molecular markers, in combination with histological observations.

Our understanding of vertebrate endodermal formation at the molecular level has recently been much improved by studies mainly in Xenopus, zebrafish and mouse. These studies have uncovered a conserved set of transcription factors that are crucial to this process (Shivdasani 2002; Stainer 2002; Tam et al. 2003). sox17 is a member of the transcription factors that contain a high mobility group (HMG)-box, and was first implicated in endodermal formation in Xenopus (Hudson et al. 1997). This finding was supported by knockout experiments showing that the sox17-null mouse is defective in endodermal formation (Kanai-Azuma et al. 2002). Similarly, in zebrafish, sox17 is expressed specifically in the endoderm at the onset of gastrulation, although its function has yet to be determined (Alexander & Stainier 1999). Like other vertebrates, zebrafish endodermal cells also express foxA2/axial/HNF3β, a winged helix/forkhead transcription factor (Strahle et al. 1993; Schier et al. 1997). Zebrafish mutant analysis, however, has identified that a novel transcription factor, casanova (cas/sox32), which is closely related to sox17, operates upstream of these factors because the mutation or knockdown of the cas/sox32 gene completely abolishes all known endodermal markers, leading to the loss of endodermal cells (Dickmeis et al. 2001; Kikuchi et al. 2001; Sakaguchi et al. 2001).

gata factors have also been implicated in endodermal formation in organisms ranging from Caenorhabditis elegans to mammals. The gata family of genes encodes a transcription factor, which contains zinc finger DNA-binding domains and binds to the consensus DNA sequence (A/T)GATA(A/G). gata transcription factors regulate different developmental processes in a wide variety of cell types in a lineage-restricted manner (Patient & McGhee 2002). Based on their expression patterns, they have been divided into two subfamilies, gata-1, -2, -3 and gata-4, -5, -6, and the latter subgroup is expressed in the endoderm lineage (Molkentin 2000; Patient & McGhee 2002). Among this endoderm subgroup, a role for gata-5 in endodermal formation has been studied in detail in both Xenopus and zebrafish. In Xenopus, gata-5 is expressed in the endoderm during the gastrula stage and ectopic gata-5 expression induces ectopic formation of the endoderm (Weber et al. 2000). Consistently, in zebrafish, gata-5 found to be a causative gene for the faust (fau) mutant in which endodermal development is partially impaired (Reiter et al. 1999, 2001). In the later stages of zebrafish development, gata-4, -5 and -6 are also expressed in various endoderm-derived organs (Molkentin 2000).

The medaka Oryzias latipes is an emerging model vertebrate system that is complimentary to zebrafish (Ishikawa 2000; Wittbrodt et al. 2002). Recently, the large-scale isolation of expressed sequence tags (EST) from medaka embryos has provided a large number of genes that have been implicated in developmental processes (Kimura et al. 2004). This prompted us to analyze the development of medaka endoderm and endoderm-derived organs with the currently available molecular markers. Furthermore, mutagenesis screens have been successfully conducted by several groups (Ishikawa 2000; Loosli et al. 2000; Furutani-Seiki et al. 2004) and our laboratory has also isolated unique mutants showing defects in endoderm-derived organs as well as disrupted organ chiralities (H. T., pers. comm., 2003). Hence, there is now an urgent need for detailed anatomical and expression-based studies of endoderm-derived organs during the early stages of medaka development. In the present study, we characterize the expression patterns of sox17, sox32, gata-4, -5 and -6, foxA2, and sonic hedgehog (shh) in the medaka embryo using whole-mount in situ hybridization, and assess their significance in the progression of the morphological changes from endodermal sheet to organized gut tube.

Materials and methods

Animals

The cab strain of medaka was a generous gift from Dr Wittbrodt (EMBL, Heidelberg, Germany) and Dr Furutani-Seiki (ERATO, Japan). Fish stocks were maintained under an extended photoperiod of 14:10 h light : dark at 26°C. Under these conditions, medaka spawn daily within 1 h of the onset of light for a number of consecutive days.

Whole-mount in situ hybridization and histology

sox17 (GenBank accession nos BJ005072 and BJ019114), gata-4 (BJ005205 and BJ019205) and gata-6 (BJ003651 and BJ017891) cDNA were isolated by our laboratory EST project (Kimura et al. 2004). shh (AB007129) and foxA2 (AB001572) were kindly provided by Mr Okamoto and Dr Araki. gata-5 and sox32 were isolated by reverse transcription–polymerase chain reaction (RT–PCR; described below). We performed whole-mount in situ hybridization as previously described (Takasima et al., unpubl. data, 2005). Images were acquired using a Leica Wild M420 microscope and a HC-2500 digital camera (Fujifilm, Minamiashigara, Japan). The stained embryos were embedded in resin (Technovit 8100; Heraeus, Werheim, Germany) and sections were cut with a RM2245 microtome (Leica, Wetzlar, Germany) and analyzed using a BX51 microscope (Olympus, Tokyo, Japan) and Axiocam HRc digital camera with Axiovision software (both from Zeiss, Oberkochen, Germany).

Isolation of sox32 and gata-5 cDNA from medaka

Total RNA was isolated from medaka embryos at the late gastrula stage using ISOGEN (Nippongene, Tokyo, Japan), according to the manufacturer's instructions. First-strand cDNA was prepared from 5 µg total RNA using an oligo d(T) primer and SuperscriptII reverse transcriptase (Invitrogen, Carlsbad, CA, USA). sox32 and gata-5 cDNA fragments were obtained by polymerase chain reaction (PCR) with the degenerate oligonucleotide primers sox32-S1 (5′-AAYGCSTTYATSR TSTGG-3′), sox32-A2 (5′-VAGRTACTGBTCRAAYTC-3′), gata-5-S1 (5′-CATCCAYCCCSTCKGCCAGCT-3′) and gata-5-A1 (5′-CAGSCGTTTCTGYGGTTTGAT-3′). Both 5′- and 3′-rapid amplification of cDNA ends (RACE) for sox32 were performed using the SMART RACE cDNA Amplification Kit (Clontech, Mountain View, CA, USA) with the gene specific primers sox32-5′-GSP1 (5′-CTCAGGATAGGCGGGGTTGAGTGCATGG-3′), sox32-5′-GSP2 (5′-GGGAAATGTGTGGGGTCTGTATAGGCGC-3′) and sox32-3′-GSP1 (5′-TGGCGCAGCTCAACCCAGACCTAGAG-3′). PCR fragments were then cloned into the pCRII-TOPO vector (Invitrogen) and sequenced.

Linkage analysis

For placing endoderm-expressing genes on the medaka linkage map, we used a reference DNA panel from 39 backcross progeny cell lines (Naruse et al. 2000). The oligonucleotide primers and restriction enzymes used for mapping via restriction enzyme length polymorphism analysis are listed in Table 1.

Table 1.  Genetic position of medaka endodermal genes
GeneForward primerReverse primerRestriction enzymeLinkage group
  • Simple sequence length polymorphism.

sox17TAAGGATGAGCGCAAGAGGCAGGGTGTCGATGCAGGATGGTGTTHaeIII20
sox32GCAAACATGTTAGAAAGCGAAGGCCTGTATTTGTAGTTGGHaeIII20
foxA2GGAAATATGAGCGCGTCGAATGGAGAGCTTGACTTTGCTGSSLP11
gata4ATGATTACGCAGGAGTGAAACCAACTTTAGCTGCAACACATGMnlI24
gata5GCGCACGGATTCTCCTACACCTGATCACCCCGCTCTCGAAMspI 7
gata6CATCCTCCGCTGGTTATATGCACGGCTGGTGTAAGTCCCACTGABstUI17
shh20

Results

We divided our observations into three developmental periods: before, during and after gut tube formation. The first period corresponds to the stages from the gastrula to the early somite stages (stage 13–21; Iwamatsu 2004), whereas the second period is the mid-somite stage (stage 22–26) and the third period corresponds to stage 33. Additionally, in the present study, we focused on the endoderm lineage because, unlike the situation in amniotes, endodermal cells in medaka first undergo tube formation by themselves, with the lateral plate mesoderm joining at a later stage when it comes into contact with the endoderm to form the complete gut tube upon the initiation of looping of the tube (Horne-Badovinac, Rebagliati & Stainier 2003). In this paper, we define the portion posterior to esophagus as the gut tube (described below in detail).

1. Expression of endodermal markers from the gastrula to early somite stages of medaka development (stage 13–21).

Expression of sox-related genes and foxA2

Embryonic sox17 expression initiates at stage 13 in the dorsal margin (Fig. 1A,G). This expression then progressively extends toward the ventral region around the margin, and by stage 15, sox17 is expressed throughout the entire margin and the involuting hypoblast in a salt-and-pepper pattern (Fig. 1M,S). sox32 expression precedes sox17 expression during gastrulation in the blastoderm, and its embryonic expression can be detected in the entire margin at stage 13 in a salt-and-pepper pattern, with intense staining on the dorsal side (Fig. 1B,H). Moreover, sox32, but not sox17, is also expressed in the yolk syncytial layer (YSL) at stage 13 (Fig. 1H, arrowhead) and weak expression is maintained in the YSL at stage 15 (Fig. 1T). Plastic sections of medaka embryos revealed that the scattered expression patterns of sox32 and sox17 at stage 15 are restricted to cells in the hypoblast layer (Fig. 1S,T, inset), indicating that sox17 and sox32 are expressed in mesendodermal cells. However, the forerunner cells, a specialized subset of dorsal cells that were reported to express sox17 and sox32 in zebrafish (Alexander & Stainier 1999), could not be identified in the medaka embryo either morphologically or by the detection of sox32 and sox17 expression (Fig. 1G,H,S,T). At stage 17, the expression domains of sox17 and sox32 are divided into two parts, polster cells and the posterior endoderm (data not shown), and sox17 and sox32 are downregulated in the anterior side. Furthermore, this expression pattern persists at stage 20 (Fig. 2A,B). Plastic sections at stage 20 show that sox17 and sox32 are expressed in the endodermal sheet lying on the YSL (Fig. 2Aa,Ab,Ba,Bb; a, around the level of the anterior tip of the notochord; b, at the level of Kupffer's vesicle).

Figure 1.

sox17, sox32, foxA2, gata-4, -5, and -6 expression at gastrula stages in medaka embryo. (A,G,M,S) sox17. (B,H,N,T) sox32. (C,I,O,U) foxA2. (D,J,P,V) gata4. (E,K,Q,W) gata5. (F,L,R,X) gata6. (A–F,M–R) Animal view. The dorsal region is towards the bottom. (G–L,S–X) Dorsal views. (A–L) Stage 13. (M–X) Stage 15. Dotted lines represent the level of the sections, which are shown in the inset. The dorsal region is towards the left in the inset. sox17 expression is detectable in the dorsal margin and sox32 expression can be detected in the entire margin in a salt-and-pepper pattern, with intense expression evident on the dorsal side at stage 13 (A,B,G,H). sox32 is also expressed in the yolk syncytial layer (YSL) (H, arrowhead). Plastic sections revealed that sox32 and sox17 expression is restricted to cells in the hypoblast layer (stage15; S,T, inset). foxA2 is weakly expressed in scattered cells of the germ ring, in addition to strong expression in the dorsal cells fated to the axial mesoderm (stage 13; C,I), and then in a salt-and-pepper pattern in the endoderm (stage15; O,U). gata-4 expression in detected in the yolk syncytial layer (YSL) (D,J [arrowhead], P,V). gata-5 expression is detected in the dorsal margin, whereas gata-6 expression is detected in both the dorsal marginal (L, arrowhead) and the YSL at stage 13 (E,F,K,L) and throughout the entire margin at stage 15 (Q,R,W,X). Dotted lines in the insets indicate the border between hypoblast and YSL.

Figure 2.

sox17, sox32, foxA2, and gata-4, -5 and -6 expression at stage 20 of the medaka embryo. (A,D,G,J,M,P) Dorsal view. The levels of the sections are indicated by lines (a,b and c). (A,B) sox17 and sox32 are expressed in the endodermal sheet lying on the yolk syncytial layer (YSL) in the posterior region, (a) around the level of the anterior tip of the notochord, (b) at the level of Kupffer's vesicle. (C) foxA2 expression is detected in the endodermal sheet at all of the anterior–posterior levels examined, (a) around the level of the presumptive pharyngeal endoderm, (b) at the level rostral to the anterior tip of the notochord, (c) at the level of Kupffer's vesicle. (D) gata-4 is expressed in the endoderm and/or YSL, (a) at the level of the anterior tip of the notochord, (b) at the level of Kupffer's vesicle. (E,F) gata-5 and gata-6 are expressed in the lateral plate mesoderm, and weakly expressed in the anterior (a, at the level rostral to the anterior tip of the notochord), but more strongly in the posterior endoderm above the YSL (Eb, at the level about 150 µm anterior to Kupffer's vesicle; Fb, at the level about 200 µm anterior to Kupffer's vesicle). Kv, Kupffer's vesicle; nt, notochord; lpm, lateral plate mesoderm.

foxA2 is also known as a molecular marker of endodermal cells (Strahle et al. 1993). In addition to its strong expression in the dorsal cells fated to form the axial mesoderm, foxA2 is also weakly expressed in scattered cells of the germ ring at stage 13 (Fig. 1C,I), and then in a salt-and-pepper pattern in the endoderm at stage 15 (Fig. 1O,U). Between stages 15 and 17 of the developing medaka embryo, the expression domain of foxA2 dorsally converges and the most positively stained cells are located under the embryonic region on the dorsal side. In contrast to sox17 and sox32, however, foxA2 expression is maintained throughout the entire endoderm (stage 20, Fig. 2C level a). Plastic sections further revealed that foxA2 expression is detected in the endodermal sheet at any of the anterior–posterior levels examined (Fig. 2Ca, around the level of the presumptive pharyngeal endoderm; Cb, at the level rostral to the anterior tip of the notochord; Cc, at the level of Kupffer's vesicle). However, it is difficult to distinguish that foxA2 is only expressed in the endoderm or both of the endoderm and the head mesenchyme because an appropriate marker for head mesenchyme is not available in medaka (Fig. 2Ca).

Expression of the gata family members

gata-4, -5 and -6 are all expressed in gastrula stage embryos but in a different manner. Whereas gata-4 expression in the YSL persists during gastrulation, no endodermal expression is detectable (Fig. 1D,J,P,V). gata-5 expression initiates at stage 13 in the dorsal margin, whereas gata-6 expression is detected in both the dorsal margin and the YSL at this stage (Fig. 1E,F,K,L). At stage 15, gata-5 and -6 are expressed in the entire margin (Fig. 1Q,R,W,X), which persists during gastrulation. However, whilst gata-6 is expressed both in mesendodermal cells and in the YSL, gata-5 is expressed only in the mesendoderm (Fig. 1E,F,K,L,Q,R,W,X). At around stage 17, the expression of both gata-5 and -6 becomes confined under the embryonic body in a similar manner to sox17 and sox32, and the regions of expression are partially segregated (data not shown). Moreover, this pattern persists through stage 20 (Fig. 2E,F). As shown by analysis of plastic embryo sections at stage 20, gata-5 and gata-6 are expressed in the lateral plate mesoderm (Fig. 2Ea, Fa, at the level rostral to the anterior tip of the notochord) and strongly expressed in the posterior endoderm above the YSL (Fig. 2Eb, at the level about 150 µm anterior to Kupffer's vesicle; Fb, at the level about 200 µm anterior to Kupffer's vesicle). In contrast, gata-4 is not expressed in the lateral plate mesoderm, but is found in the endoderm and/or YSL (Fig. 2Da, at the level of the anterior tip of the notochord; Db, at the level of Kupffer's vesicle; at this stage, however, it is sometimes difficult to distinguish the YSL from the endodermal sheet because both tissues are very thin and contact each other closely).

2. Expression of endodermal markers from stages 21–26 of the developing medaka embryo

Whereas sox17, sox32, foxA2, and gata-4, -5 and -6 are expressed in early endodermal cells, they are expressed at later stages in endoderm-derived organs, exhibiting distinct patterns of expression. In the following section, we divide the endodermal sheet into the two parts along the anterior–posterior axis at the level of the first somite. The anterior part gives rise to the pharynx and esophagus while the remainder gives rise to the intestine. In this paper, we define the latter portion as the gut tube and this is further subdivided into the rostral, intermediate and caudal portions, therefore the rostral gut means the portion posterior to the esophagus. In a similar manner to zebrafish, the medaka does not develop a stomach and thus the intestine is directly connected to the esophagus.

The expression profile of foxA2 and the formation of the gut tube

Since foxA2 continues to be expressed throughout the entire endoderm from the precursor stage to the onset of gut tube morphogenesis, we attempted to analyze and characterize gut tube formation using the staining pattern of foxA2 expression. Medaka embryos from stages 21–26 were stained for foxA2 and sectioned at the esophagus (a) and at the rostral (b), intermediate (c) and caudal (d) levels of the presumptive gut endoderm. The expression of foxA2 in the endoderm, which is shown in Figure 2Cb and Cc, remains unchanged until stage 21, during which time no obvious structure is formed in the endodermal sheet at any anterior or posterior levels (Fig. 3Ab, at the level of the anterior tip of the notochord; Ac, 270 µm posterior to b; Ad, 60 µm anterior to the Kupffer's vesicle). Plastic sections revealed that foxA2-expressing endodermal cells retain a monolayer on the YSL and that their cell shape is elongated along the medio-lateral axis. The endoderm fated to form the esophagus is also comprised of a monolayer of cells (Fig. 3Aa). At stage 22, the rostral portion of the endoderm starts moving towards the midline to form a cell aggregate. foxA2-positive cells in the cell aggregate then form a dorso-ventrally flattened bilayer, and cells on the dorsal side become thickened, losing their medio-laterally elongated shapes, whilst those in the ventral remain flattened (Fig. 3Bb, at the level of the anterior tip of the notochord). In contrast, endodermal cells in the caudal portion still maintain a monolayer and expand laterally (Fig. 3Bd, 40 µm anterior to the Kupffer's vesicle). In the intermediate portion, as endodermal cells migrate to the midline, they form a bilayer structure (Fig. 3Bc, 150 µm posterior to b), but all cells retain a medio-laterally elongated shape (Fig. 3Bc, dotted line).

Figure 3.

Expression of foxA2 during gut tube formation in medaka. (A–F) Dorsal view of medaka embryos at stage 21 (A), 22 (B), 23 (C), 24 (D), 25 (E) and 26 (F) with corresponding histological sections (a) esophagus and (b) the rostral (c) intermediate and (d) caudal levels of the presumptive gut endoderm. (Aa–d) Stage 21; no obvious morphological structures are formed in the endodermal sheet at any anterior–posterior levels. (Ba–d) Stage 22; the rostral portion of the endoderm starts moving towards the midline to form a cell aggregate (b). The endodermal cells in the intermediate portion form a bilayer structure (c, dotted line, 150 µm posterior to b). The endodermal cells in the caudal portion still maintain a monolayer and expand laterally (d, 40 µm anterior to the Kupffer's vesicle). The endoderm in the esophagus region still retains a monolayer expanding laterally (a). (Cb–d) Stage 23; the rostral portion forms a rod-like structure (b, anterior tip of the notochord level). The intermediate portion becomes thick dorso-ventrally (c, 140 µm posterior to b). The caudal portion still maintains a monolayer structure (d, 80 µm anterior to the center of Kupffer's vesicle). (Db–d) stage 24; the rostral portion shows a radial organization with nuclei on the basal side (b, at the level of the liver bud). The liver bud is first observed at this stage (Db, arrowhead). The caudal portion forms a bilayered structure with a medio-laterally elongated shape (d, 340 µm posterior to c). (c) is 180 µm posterior to (b). (Eb–d) stage 25; the rostral and the intermediate portions adopt a radial configuration and the gut lumen is first visible in the rostral portion (b, at the level of the liver bud level; c, 340 µm posterior to b). Cells in the caudal portion start migration toward the midline at this stage (d, 200 µm posterior to level c). (Fb–d) stage 26; all parts of the gut finally acquire a lumen, and gut tube formation completes (b–d; b, liver bud level; c, 200 µm posterior to b; d, 360 µm posterior to c). The liver bud increases are indicated by arrowheads (Db,Eb,Fb).

The endoderm in the esophagus region still retains a monolayer expanding laterally (Fig. 3Ba) and at stage 23, the rostral portion forms a rod-like structure, but neither the radial structure nor the lumen is formed at this stage (Fig. 3Cb, anterior tip of the notochord level). In contrast, the caudal portion still maintains a monolayer structure, whereas cells in the intermediate portion, which retains a bilayer, become thickened dorso-ventrally (Fig. 3Cb–d; c, 140 µm posterior to b; d, 80 µm anterior to the center of Kupffer's vesicle). At stage 24, the rostral portion shows a radial organization with nuclei on the basal side, but the gut lumen is not yet visible (Fig. 3Db, at the level of the liver bud). The liver bud, positioned slightly left of the midline and projecting from the ventral side of the rostral gut endoderm, is first observed at this stage (Fig. 3Db, arrowhead). The caudal portion eventually forms a bilayered structure but the cells remain medio-laterally elongated in shape (Fig. 3Dc, 180 µm posterior to b; d, 340 µm posterior to c). At stage 25, the intermediate portion, in addition to the rostral portion, adopts a radial configuration and the gut lumen is visible for the first time in the rostral portion (Fig. 3Eb, at the level of the liver bud level; c, 340 µm posterior to b). Cells in the caudal portion begin migrating towards the midline at this stage (Fig. 3Ed, 200 µm posterior to level c). At stage 26, the most caudal part of the gut finally acquires a lumen, and gut tube formation has completed by this stage (Fig. 3Fb–d; b, liver bud level; c, 200 µm posterior to b; d, 360 µm posterior to c). The liver bud subsequently increases in size and becomes distinct from the nascent gut tube (Fig. 3Fb arrowhead). A detailed analysis of liver development in medaka will be described elsewhere.

Expression of gata family genes in the medaka embryo

The expression of the gata family of genes, detectable in the endodermal layer, is downregulated in most of the endoderm prior to the onset of gut tube formation (stage 21, data not shown). During gut tube formation, however, their expression becomes detectable again in the developing gut tube and its accessory organs. The expression of gata-4 and -6 are activated in the most rostral portion of the developing gut at stage 22 (Fig. 4A,F) and proceeds in an anterior to posterior direction (Fig. 4B,C,G,H). Finally, gata-6 expression level reaches the caudal end by stage 25 (Fig. 4I) and persists after stage 26 (Fig. 4J). In contrast, gata-4 expression level does not reach the caudal end and starts to disappear from the intermediate portion at stage 25 (Fig. 4D) and confined to the anterior portion at stage 26 (Fig. 4E). In addition to the gut, gata-4 and -6 are also expressed in the liver rudiment (Fig. 4E,J, arrow), whereas only gata-6 is expressed in the swim bladder (Fig. 4J, arrowhead). In contrast, gata-5 expression, though detected in the entire gut at stage 26, is much weaker than both gata-4 and -6 (data not shown).

Figure 4.

The expression domains of gata-4 and -6 are coincident with gut tube formation in the medaka embryo. (A–E) gata-4. (F–J) gata-6. Brackets indicate expression domains. gata-4 and -6 are activated in the most rostral portion of the forming gut at stage 22 (A,F). The expression of both gata-4 and -6 seems to expand posteriorly (stage 23–24; B,C,G,H). gata-4 and -6 expression are detected in the entire gut at stage 25 (D,I). gata-4 expression demarcates the anterior portions (4E) whereas gata-6 expression is detected throughout the entire gut at stage 26 (J). Arrows indicate liver rudiment and arrowhead indicates swim bladder rudiment.

3. Expression of endodermal genes in accessory organs at stage 33 of the medaka embryo

The early endodermal markers of the developing medaka embryo, sox32 and sox17, are switched off in the forming gut but reactivated in the restricted regions at stage 33. sox32 is expressed in the gallbladder (Fig. 5A, arrow) and sox17 is expressed in both the gallbladder (Fig. 5B, arrow) and anterior part of the swim bladder (Fig. 5A arrowhead). We then examined shh expression which is activated in the differentiating endodermal epithelium in other vertebrates (Apelqvist et al. 1997; Narita et al. 1998; Roberts et al. 1998). shh is not expressed in endodermal cells during the gastrulation and early somite stages and is first activated at stage 26 in both the rostral and caudal portions (the cloaca; data not shown). Both expression domains extend towards the intermediate portion and finally cover the entire gut until stage 31. shh expression is maintained until stage 33 (Fig. 5C, dotted line), and then decreases in the intermediate portion at stage 37 (data not shown). Unlike in zebrafish, shh is never expressed in the liver or the swim bladder throughout embryogenesis in medaka (data not shown). In contrast to the shh expression profile, foxA2 is expressed strongly in the liver and swim bladder at stage 26 of the medaka embryo (Fig. 3F), and these expression levels are maintained in both organs at stage 33 (Fig. 5D). Additionally, given the differential expression of gata-4, -5 and -6 at the early somite stages (Fig. 2D–F), their patterns of expression vary quite significantly at stage 33. Whereas gata-5 and gata-6 can be detected in the liver, swim bladder and entire gut, gata-4 is expressed only in the liver and rostral gut (Fig. 5E,F,G).

Figure 5.

Molecular markers reveal the endodermal organ primordia at stage 33 in the developing medaka embryo. Schematic illustration of the gut and its accessory organs are shown in (H). sox32 is expressed in the gallbladder (A, arrow) and sox17 is expressed in both the gallbladder (B, arrow) and anterior part of the swim bladder (B, arrowhead). The shh expression domain covers the entire gut (C, dotted line) and foxA2 is expressed strongly in the liver and swim bladder (F). gata-4 is expressed in the liver and rostral gut (E, dotted line), whereas gata-5 and gata-6 are detected in the liver, swim bladder and the entire gut (F,G).

4. Genetic mapping of endoderm-related genes and the evolution of sox17 during medaka development

We performed genetic mapping of the endoderm-related genes examined in this study. shh had already been mapped onto linkage group (LG) 20 (Naruse et al. 2000), and in our present study we have mapped sox17, sox32, gata-4, -5, -6, and foxA2 onto LG 20, 20, 24, 7, 17 and 11, respectively (Table 1). It is interesting to note that sox17 and sox32, which are closely related genes containing a HMG box, are aligned in tandem about 30 kb apart in the medaka genome (data was obtained from the UT genome browser, Medaka Genome Sequencing Project 2004). Comparison of this genomic region with those of fugu, zebrafish and human demonstrate that the region around sox17 displays conserved synteny, with the exception of sox32 which is absent in human, suggesting that sox17 was locally duplicated during the evolution in the fish lineage (Fig. 6). This is further supported by a comparison of the gene order around the sox17 and sox32 loci in medaka to that of fugu (Aparicio et al. 2002, version 3.0), zebrafish (Pre! Ensembl 2005) and human (National Center for Biotechnology Information 2005). The results show conserved synteny among medaka, fugu, zebrafish and human, but demonstrate that a part of this region is inverted in zebrafish (Fig. 6, dotted lines), probably due to an intrachromosomal rearrangements.

Figure 6.

Gene content of the regions of shared synteny around the sox17 locus of human 8q11–13 (National Center for Biotechnology Information, Homo sapiens genome view, Build 35 version 1, 2004), a medaka LG20 scaffold 516 (Medaka Genome Sequencing Project 2004, revision 200406), a Fugu scaffold 4 (Aparicio et al. 2002, version 3.0) and zebrafish chromosome 15 (Pre! Ensembl 2005 release 31). Genes are denoted by colored bars along a scale. The orientations of the transcripts are indicated by arrows. The gene order is inverted only in zebrafish (dotted lines). lypla1, lysophospholipase I; mrpl15, mitochondrial ribosomal protein L15; rp1, retinitis pigmentosa 1.

Discussion

Our current study was designed to further elucidate the principal events during the development of the endoderm and gut in developing medaka embryo for future studies. For this purpose, we have characterized the expression patterns of molecular markers that are known to be expressed in the endoderm and gut in other vertebrates. We also performed histological analyses to analyze the morphogenesis of the endodermal sheet above the YSL into the organized gut tube. Our findings are summarized in Figure 7.

Figure 7.

Schematic representation of gene expression patterns in the developing gut in medaka. Time chart showing gut morphogenesis and the expression of endodermal expressing genes. Gut tube formation is initiated in the rostral portion at stage 22, gradually expands posteriorly and finally reaches the caudal end at stage 26. In the rostral portion, the cells stack dorso-ventrally and form a rod-like structure, then acquire a radial structure. In the intermediate and caudal positions, the cells change their shape into a dorso-ventrally elongated form, and then form an interior canal. The expression of gata-4 and -6 expands posteriorly, concomitantly with an antero-posterior progression of tube morphogenesis. Once gut tube formation completes, gata-4 expression starts to disappear from the intermediate and caudal portions, whereas gata-6 expression persists. shh is first activated at stage 26 in both the rostral and caudal portions.

Expression of sox32, sox17 and foxA2 in the medaka endoderm and endoderm-derived organs

sox32 and sox17 are members of sox-type transcription factors and have been implicated in endoderm specification in zebrafish. In medaka, both genes exhibit a typical pattern of endodermal expression, the salt-and-pepper pattern, which is also found in zebrafish. Also similar to zebrafish, sox32 expression precedes sox17 and is expressed in the absence of sox17 in the YSL of medaka. These findings suggest the conserved roles of these genes in medaka endodermal development. sox32 was originally identified as the causative gene for zebrafish casanova mutant in which any endodermal cells fail to develop, and in which sox17 was placed downstream of cas/sox32 (Dickmeis et al. 2001; Kikuchi et al. 2001Sakaguchi et al. 2001). However, no orthologues of sox32 have so far been identified in mouse or human, suggesting that sox32 is specific to the fish lineage. Interestingly also, these two genes map to the same position on LG20 (Table 1), as previously reported in zebrafish (chromosome 15). Our comparative analysis with the corresponding genomic regions strongly suggests that sox17 experienced an evolutionary tandem-duplication after the fish lineage diverged from a common vertebrate ancestor, and that one of the duplicated genes evolved as sox32 and now functions upstream of sox17.

During gastrulation, endodermal cells express sox32, sox17 and foxA2 but at the early somite stages, sox32 and sox17 are downregulated in the anterior endoderm. Consequently, the anterior endoderm, which is fated to form the pharynx and esophagus, mainly expresses foxA2, while the posterior endoderm is positive for fox2, sox32, sox17, in addition to gata-5 and -6 (Fig. 2). Thus, these genes were suitable for the analysis of the regional specification of the endoderm along the anterior–posterior axis in the early medaka embryo. The overlapping expression of sox32 and sox17 was also observed in the gallbladder at later stages (Fig. 5A,B), suggesting that a similar gene circuit consisting of sox32 and sox17 functions in this process.

Expression of gata genes in the endoderm and endoderm-derived organs in medaka

In zebrafish, during embryogenesis, the endoderm expresses gata-5 and gata-4 is silent, whereas the YSL expresses both gata-5 and -6 (Reiter et al. 2001). In medaka, by contrast, gata-5 and -6 are activated in endodermal cells while the YSL expresses gata-4 and -6. Thus, it seems likely that medaka gata-6 and gata-4 are functional counterparts to zebrafish gata-5 and gata-6, respectively, during early endodermal development. Genomic analysis of each syntenic region will be required for a further understanding of the functional and structural changes of these gata family genes during fish evolution. In Xenopus, although gata-4, -5 and -6 are all involved in the formation of the endoderm, gata-6 appears to be a key regulator in this process and gata-4 and -5 play relatively minor roles, in part acting through gata-6 (Afouda et al. 2005). Further functional analysis will reveal whether each gata factor has a distinct or redundant function in medaka.

gata genes are also expressed in the forming gut and its accessory organs. Our temporal expression analysis demonstrates that the expression of gata-4 and -6 expands posteriorly. While gata-6 expression reaches to the caudal end of the gut at stage 25, gata-4 expression extends only to the intermediate region and not detected in the posterior region. Furthermore, gata-4 expression disappears in the intermediate portion and confined to the small anterior portion at stage 26. This differential expression between gata-4 and -6 may result in the regionalization of the gut tube along the anterior–posterior axis.

Medaka gut formation proceeds in a different manner between the rostral and intermediate/caudal gut region

Histological analysis using foxA2 as an endoderm marker enabled us to follow the overall gut formation events in the medaka embryo. Gut tube formation is initiated in the rostral portion at stage 22, gradually expanding posteriorly and finally reaching the caudal end, the cloaca, at stage 26. The first sign of gut morphogenesis is the migration of endodermal cells to the midline. The monolayer structure of the endoderm, which overlays the YSL, migrates medially and then changes into a bilayer configuration with medio-laterally elongated cell shapes. During the next step, the rostral and the remaining portions behave differently. In the intermediate and caudal positions, cells change their shape into a dorso-ventrally elongated form and then form an interior canal. On the other hand, in the rostral portion, the cells stack dorso-ventrally, form a rod-like structure and then acquire a radial structure. At this point, canalization has not yet occurred. During this process, accessory organs, such as the liver and swim bladder, can be identified histologically (Fig. 3Db). These different morphogenetic processes may reflect the fact that the rostral gut provides tissue anlagen such as the liver bud, while the remaining parts of the gut do not. It has been reported in zebrafish that the liver progenitors arise from endoderm that is anterior to the gut tube, independently of tube morphogenesis (Wallaec & Pack 2003). However, we histologically observed that, like amniotes, the liver rudiment in medaka is formed by budding from the gut tube. To address whether medaka liver progenitors can be identified before the onset of gut tube morphogenesis, further experiments using earlier markers such as hhex and pdx-1 will be required.

Conserved and unique aspects of medaka gut morphogenesis

We next describe the temporal sequence of gut tube formation in the medaka embryo. Medaka gut formation has both conserved and unique aspects when compared with zebrafish and mammalian gut morphogenesis. It has been reported that zebrafish gut morphogenesis initiates at both the rostral and caudal portions. In contrast, gut formation proceeds in one direction, from rostral to caudal, in the medaka embryo (Figs 3 and 7). Tube morphogenesis is accompanied by the expression of gata-4 and -6 (Fig. 4), which initiate their expression in the rostral gut and expand posteriorly. In contrast, the expression of shh in the gut epithelium begins at both ends when gut tube formation has completed (Figs 5C and 6 and data not shown), suggesting that the differentiation wave of the epithelium proceeds in a manner different to the morphogenetic wave.

Our histological analysis suggests that tube morphogenesis in the anterior endoderm of the medaka, fated to form the pharynx and esophagus, occurs independently of the rostral gut. In the esophagus region, tube formation is much delayed when compared with the gut region (Fig. 3Ba,b). This is consistent with the report of Wallaec and Pack (2003) that showed that the pharynx and esophagus develop independently of rostral gut morphogenesis in zebrafish. Furthermore, in medaka and zebrafish, the gut forms from an unorganized cell mass and/or bilayered endodermal sheet. These characteristic features of gut development may be shared by most teleosts, but are in sharp contrast with the amniotes. In amniotes, the gut is formed by the folding of the endodermal sheet, and the anlagen of the pharynx, esophagus and intestine arise from the foregut. Despite these differences, recent genetic analyses have revealed that a number of conserved molecular programs regulate vertebrate gut development.

Acknowledgements

We thank Dr Wittbrodt and Dr Furutani-Seiki for providing us with materials described in Materials and methods. We thank S. Minami, Y. Park, M. Sugimoto, K. Nakaguchi, A. Igarashi, K., Mochizuki Y., Ozawa, K. Oki and T. Obata for technical assistance. The work presented here was supported in part by Grants-in-Aid for Scientific Research Priority Area ‘Genome Science’ and ‘Organized Research Combination System’ from the Ministry of Education, Culture, Sports, Science and Technology of Japan. D. K. is a research fellow supported by the 21st century COE program of the University of Tokyo from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Ancillary