INTRODUCTION

Cells of the endoderm are fated to form the lining of the gut,that is the pharynx, oesophagus, stomach and intestines. Theepithelium of the liver, gall bladder, pancreas and therespiratory system also forms from the endoderm. In Xenopusthe endoderm originates from the vegetal hemisphere of theembryo (Dale and Slack, 1987), and there has recently beenprogress showing that both secreted molecules, for examplenoggin, Vg-1 and cerberus (Sasai et al., 1996; Joseph andMelton, 1998; Bouwmeester et al., 1996) and transcriptionfactors, such as Mix 1, Mixer, Xsox17α and VegT (Lemaire etal., 1998; Henry and Melton, 1998; Hudson et al., 1997; Zhanget al., 1998 and references therein) play a role in endodermformation.

As well as the endoderm-derived epithelium, the organs ofthe gut also have an outer layer of mesoderm-derivedmesenchyme that forms connective tissue and smooth muscle.Work in other vertebrates has shown that signals between themesenchyme layer and the underlying epithelium are importantin the regional specification of the endoderm (reviewed inHaffen et al., 1987; Yasugi and Mizuno, 1990; Rawdon andAndrew, 1993). For example, recombination experiments in thechick have shown that intestinal mesenchyme can cause therespecification of stomach epithelium into intestinal epithelium(Yasugi and Mizuno, 1978; Ishizuya-Oka and Mizuno, 1984;Andrew and Rawdon, 1990). More recent work has started tosuggest a molecular basis to these interactions, as the anterior

boundaries of Hox gene expression in the intestinal smoothmuscle were found to match anatomical boundaries in theintestinal epithelium (Roberts et al., 1995; Yokouchi et al.,1995). This raises the possibility that the Hox genes set upboundaries in the smooth muscle that are somehow transferredonto the underlying epithelium by local inductive signals.Consistent with this, ectopic expression of Hoxd-13 in theanterior intestinal smooth muscle causes a partialtransformation of the underlying epithelium into a moreposterior fate (Roberts et al., 1998). This corresponds well withthe situation in Drosophila where homeotic genes expressed inthe visceral mesoderm play a role in patterning the larvalmidgut (reviewed in Bienz, 1994).

Classical recombination experiments in amphibian embryoshave also shown that the early mesoderm is able to respecifyexplanted endoderm to a different regional character (Okada,1957, 1960). A limitation to these experiments is that they werecarried out before the introduction of lineage tracers, makingit impossible to eliminate the possibility of contaminating cellsin the recombinations. They were also done without anaccurate fate map for the endodermal and mesodermalcomponents of the gut. In order to investigate the specificationof the endoderm an accurate fate map is required so that theresults of any explant or transplantation experiment can becompared with the presumptive fate for that piece of tissue.There are a number of existing amphibian endoderm fate maps(particularly that of Tahara and Nakamura, 1961), but thesesuffer from the limitations of spreading and fading associated

381Development 127, 381-392 (2000)Printed in Great Britain © The Company of Biologists Limited 2000DEV1459

We have produced a comprehensive fate map showingwhere the organs of the gut and respiratory system arederived from in the early Xenopus laevis endoderm. We alsoshow the origin of the associated smooth muscle layer on aseparate fate map. Comparison of the two maps shows thatfor most organs of the gut the prospective epithelium andsmooth muscle do not overlie each other in the earlyembryo but come together at a later stage. These fate mapsshould be useful for future studies into endodermspecification.

It was not previously known how the elongation of theendoderm occurs, how the single-layered dorsal and many-layered ventral endoderm gives rise to the single layeredepithelium, and whether or not the archenteron cavity

actually gives rise to the gut lumen. Using a variety oflabelling procedures we show firstly, that radialintercalation occurs in the gut transforming a short thicktube into a long thin tube; secondly, that the archenteronlining does not become the definitive gut lumen. Instead thearchenteron cavity almost closes at tailbud stages beforeproviding a nucleus for the definitive gut cavity, whichopens up during elongation. Based on this work we presenta model explaining the morphogenesis of the gut.

Key words: Xenopus laevis, Endoderm, Mesoderm, Smooth muscle,Fate map, Radial intercalation, Morphogenesis, Gut, Pancreas, Liver,Intestine, Epithelium

SUMMARY

The Xenopus tadpole gut: fate maps and morphogenetic movements

Andrew D. Chalmers and Jonathan M. W. Slack*

Developmental Biology Programme, Department of Biology and Biochemistry, University of Bath, Bath BA2 7AY, UK*Author for correspondence (e-mail: bssjmws@bath.ac.uk)

Accepted 1 November; published on WWW 20 December 1999

382

with vital dyes and, furthermore, none of these older studieswere carried out using Xenopus embryos. In order tounderstand the development of the endoderm it is importantalso to understand the origin of the smooth muscle layer andthe role it plays in specifying the endoderm. At present it isnot known where the smooth muscle layer originates from inthe Xenopus mesoderm.

As well as regional specification, the development of anorgan system also involves morphogenesis: that is, how theshape of the finished organ is formed from the cells of the earlyembryo. In other systems, cell rearrangements have beenshown to be important in morphogenesis (reviewed in Keller,1987). Little is known about the morphogenesis of theendoderm and there are several important questions that havenot been answered. At neurula stages the Xenopus endodermlines the archenteron cavity. The dorsal endoderm consists ofa single cell layer while the ventral endoderm has several layersof large yolky cells. By stage 45 (5-day-old tadpole) theendoderm has undergone enormous elongation and formed thesingle-cell-layered epithelium of the intestine. It is not knownhow the elongation of the endoderm is accomplished. It is alsonot clear how the thin dorsal layer and thick ventral layers ofendoderm produce the single layer of cells that forms the gutepithelium. It is sometimes suggested that the large yolkyventral cells disintegrate and are digested during development(Nieuwkoop and Faber, 1967; Mathews and Schoenwolf,1998), implying that that the outer endodermal cells lyingcloser to the mesoderm form the gut epithelium, although thishas not been proved. Finally, it is not known whether thearchenteron cavity of the neurula really gives rise to the gutcavity of the later tadpole. It has been proposed (Goette, 1875),and is generally assumed, that the gut cavity is a continuationof the archenteron. A problem with this model is that thearchenteron narrows considerably and appears to close throughmuch of the gut at early tadpole stages. This means therecannot be a simple transition from archenteron to gut cavity. Ithas also been proposed, based on vital dye experiments, thatthe gut cavity is not a continuation of the archenteron but opensup de novo during development (Tahara and Nakamura, 1961).So, although it is widely assumed, it is not known whether thearchenteron does in fact give rise to the gut cavity.

We have used grafts of fluorescent labelled tissue to producea new and comprehensive fate map showing which parts ofthe early endoderm give rise to which organs in the gutand respiratory system. This demonstrates that duringmorphogenesis of the endoderm the axes (anterior/posterior,

dorsal/ventral and left/right) of the early embryo aremaintained in the later gut. We have also produced a fate mapfor the smooth muscle layer showing where in the earlymesoderm the smooth muscle layer of the gut originates from.Comparison of the two fate maps shows that the origins of thefuture epithelia and smooth muscle layer are only partiallyoverlapping in the early embryo. Therefore, continuousinteractions between the two germ layers could not start untilthe two layers move into accord later on in development. Thesetwo fate maps represent a resource for future explant andtransplantation studies into the specification of the endoderm.

In order to investigate the morphogenetic movementsassociated with gut formation we used the lipophilic dye DiIspecifically to label the large yolky cells on the floor of thearchenteron and the cells in the middle of the ventral endoderm(between the floor of the archenteron and very ventral endodermthat lies next to the mesoderm). Both groups of cells were foundto be incorporated into the gut epithelium and so cannot be

A. D. Chalmers and J. M. W. Slack

Fig. 1. The stage 46 Xenopus tadpole gut.(A) Schematic diagram (ventral view/anteriorat the top). (B) Whole-mount drawing(ventral and lateral views). The gut has beenshaded dark grey and the liver and pancreaslight grey. oe, oesophagus; lv, liver; gb, gallbladder; pa, pancreas; sia, proximal smallintestine; sib, external coil of small intestine;sic, internal coil of small intestine; lic,internal coil of large intestine; lid, distal largeintestine; st, stomach; pr, proctodaeum; ht,heart. Figure adapted from Chalmers andSlack (1998), with permission. Bar, 1 mm.

Fig. 2. The 14 regions used for fate mapping. (A) Left view of astage 14 whole mount. (B) Dorsal view of a stage 14 whole mount.(C) Parasagittal section of a stage 14 embryo. (D) Transverse sectionof a stage 14 embryo. The grafting regions are shown on the embryosusing the following labels: 1, extreme anterior; 2, anterior dorsal; 3,anterior right; 4, anterior left; 5, anterior ventral; 6, middle dorsal; 7,middle right; 8, middle left; 9, middle ventral; 10, posterior dorsal;11, posterior right; 12, posterior left, 13; posterior ventral; 14extreme posterior.

383Fate maps and morphogenesis in Xenopus gut

digested during development. We then used double labelling(DiI + fluorescein dextran amine) to show that these cells areincorporated into the epithelium by a process of radialintercalation. We believe that the occurrence of radialintercalation drives the elongation of the endoderm. Using asecond approach, the cells lining the entire surface of thearchenteron were labelled with biotin, allowing the embryoniccavity to be followed during development. The archenteron wasshown to narrow during development before reopening to splitthe ventral endoderm and give rise to the definitive gut cavity.Based on this work we present a new model explaining themorphogenesis of the gut.

MATERIALS AND METHODS

Lineage tracingFDA labellingXenopus laevis embryos were obtained using standard procedures(Godsave et al., 1988), cultured in normal amphibian medium (NAM)

(Beck and Slack, 1999) and staged according to Nieuwkoop and Faber(1967).

Fate mapping was carried out using Fluorescein dextran amine(FDA, Molecular Probes) (Gimlich and Braun, 1985). During the firstcleavage donor embryos were placed in NAM plus 5% Ficoll andinjected once into each of the two blastomeres with 0.25 ng (4.6 nl of50 mg/ml) FDA. At stage 13/14 an orthotopic graft, a rectangular pieceof tissue approximately 400 µm × 600 µm containing all three germlayers, was cut from a labelled donor embryo at one of the 14 standardregions (see Results) and used to replace the equivalent region froman unlabelled host embryo. Embryos that developed normally aftergrafting were fixed at stage 46 in 10% formalin in 70% PBS for 24 or48 hours at room temperature. They were then embedded, sectionedand mounted in DPX (BDH). The FDA-labelled organs were identifiedbased on our previous work (Chalmers and Slack, 1998). Sections fromtypical individuals were drawn with the aid of a drawing tube (Leica)and the position of the FDA label was then added to the drawings. Asmall number of the mesoderm cases were scored in whole-mount guts,as described for the DiI labelling, rather than in sections.

DiI labellingTo specifically label small populations of endoderm cells a fixable

Fig. 3. Examples of FDA labelling. For each example a labelled drawing of a section (A,C,E,G,I) and a photograph of the labelled section(B,D,F,H,J) are shown. A whole-mount drawing is also included at the top of the figure to show the position of the sections in the tadpole.(A,B) Labelled pharynx. (C,D) Labelled pancreas and small intestine. (E,F) Labelled small intestine. (G,H) Labelled proctodaeum.(I,J) Labelled smooth muscle in the large intestine. (K) Smooth muscle layer in the tadpole gut. In B,D,F,H and J, arrowheads highlight labelledepithelia and arrows highlight smooth muscle. ph, pharynx; tn, tongue; sia, proximal small intestine; pa, pancreas; st, stomach; lid, distal largeintestine; pr, proctodaeum. Bar, 200 µm (B); 120 µm (F,H); 95 µm (D,J); 30 µm (K).

384

derivative of the lipophilic dye DiI was used (Cell Tracker CM-DiI,Molecular Probes, referred to simply as ‘DiI’). This was dissolved at3 mg/ml in ethanol + 100 mg/ml phosphatidylcholine, heated to 50°C,diluted 1/10 in 0.2 M sucrose at 50°C, centrifuged to remove anyprecipitate, then a 4.6 nl pulse was fired at the desired position usinga Drummond Microinjector. Stage 14 embryos were labelled with DiIat one of four positions (shown in Fig. 6A). (1) Floor of thearchenteron: a piece was cut from the middle dorsal position to exposethe floor of the archenteron, which was then labelled with DiI beforereplacing the dorsal piece. (2) The ventral endoderm: a piece ofectoderm and mesoderm was cut from the mid-ventral positionexposing the endoderm, which was labelled with DiI and the piecereplaced. (3) The mid-ventral endoderm: a thick piece of tissue wascut from the mid-ventral position. This exposed the middle endoderm,which was then labelled with DiI and the piece of tissue replaced. (4)The dorsal endoderm: a piece was cut from the middle dorsal positionand turned over to expose the archenteron roof, which was labelledwith DiI and then replaced. DiI labelled-embryos at stages 14 and39/40 were fixed, embedded, sectioned and scored as previouslydescribed for the FDA labelling. Stage 45 DiI-labelled embryos werescored as isolated gut preparations. They were dissected as previouslydescribed (Chalmers and Slack, 1998) and confocal images were thencaptured using a Zeiss 510 laser scanning microscope.

FDA + DiI labellingTo label the floor and dorsal endoderm simultaneously a piece oftissue was cut from the dorsal roof (equivalent to middle and posteriordorsal locations) of a host embryo and the archenteron floor labelledwith DiI as above. A graft from an FDA-labelled donor embryo wasthen used to replace the dorsal roof. To label the floor and ventralendoderm simultaneously a shallow graft (containing ectoderm,mesoderm and approximately 1-2 layers of endoderm cells) was cutfrom a donor embryo and used to replace the equivalent piece froman unlabelled host. The embryo was left to heal and then the floor waslabelled with DiI as described above. To label the middle and ventralpositions simultaneously a piece was cut from the mid-ventralposition as described before and the middle endoderm labelled withDiI. The ventral piece was then replaced with a graft of FDA-labelledtissue. The three sets of embryos were then cultured, dissected andscored as described for the DiI labelling.

Labelling of the archenteron liningThe entire external surface of late blastula (stage 9/10) embryos waslabelled with biotin using sulfo-NHS-LC-biotin (Pierce) as previouslydescribed (Minsuk and Keller, 1997). The only modification was thatthe sulfo-NHS-LC-biotin was made up in NAM/10 and the labellingtime was reduced to 5 minutes. The biotin-labelled embryos werefixed at the required stage and the biotin detected using alkalinephosphatase-conjugated streptavidin (Vector Labs) as previouslydescribed (Minsuk and Keller, 1997).

RESULTS

The Xenopus tadpole gut and respiratory systemIn our previous work we published a detailed description of thenormal development of the Xenopus tadpole gut (Chalmers andSlack, 1998). Similar to the mammalian gut, it consists of thepharynx, oesophagus, stomach, intestines, pancreas and liver(Fig. 1). A striking feature is that the intestine forms a doublecoiled structure. The double coil is formed by an exterior coilconsisting of anticlockwise loops of intestine followed by aninterior coil consisting of clockwise loops of intestine. Weproposed a nomenclature to describe the different parts of thedouble coiled intestine, from the proximal to the distal end.

‘sia’ is the most proximal part of the small intestine that islocated before the double coil; ‘sib’, is the small intestine thatforms the external coil and ‘sic’ is the small intestine in theinternal coil. ‘lic’ is the proximal part of the large intestine inthe internal coil and ‘lid’ is the distal part of the large intestinethat runs to the proctodaeum (pr). In this study we rely on ourprevious work to identify the organs labelled by the fatemapping and use the nomenclature when describing the coiledintestine.

The tadpole respiratory system consists of the gills, whichare located in the lateral pharynx, and the trachea, which splitsfrom the posterior pharynx and then bifurcates to give rise tothe two lung buds (Nieuwkoop and Faber, 1967; Chalmers andSlack, 1998).

The 14 embryonic regions used for the fate mappingThe early neurula stage Xenopus embryo was divided up into14 regions that covered the entire embryo (Fig. 2). Region 1 atthe very anterior is called ‘extreme anterior’. Posterior to thisthe embryo was split into 3 anterior/posterior levels, termed‘anterior’, ‘middle’ and ‘posterior’. Each of these levels wassplit into a dorsal, right, left and ventral region. Finally the mostposterior region, the ‘extreme posterior’ region, number 14, liesopposite the extreme anterior region. Each of these regions waslabelled using orthotopic grafts of fluorochrome-labelled tissueand the embryos were left to develop to stage 46. They werethen fixed and sectioned and the labelled epithelia scored.Examples of the experiments are shown in Fig. 3, where adrawing of each section is shown along with a photograph ofthe FDA label (FDA is highlighted by arrowheads). A whole-mount drawing shows where the sections lie in the tadpole. Thesmooth muscle layer in the Xenopus tadpole gut consists of asingle very thin layer of cells (Fig. 3K, arrow; Kordylewski,1983; Chalmers and Slack, 1998). As the grafts also containedthe mesoderm layer it was possible to score the label in thesmooth muscle layer that surrounds the gut (Fig. 3J, comparewith the unlabelled smooth muscle in Fig. 3F).

Presenting the endoderm fate map The results for the endoderm fate mapping are presented inthree ways. Table 1 shows the organs that were labelled fromgrafts in each region. If a particular region labelled a particularorgan in at least 50% of cases then that organ was consideredto arise from that region and was included in the fate map(shaded grey in Table 1). A limitation of this method is that itgives the same score if a graft labels a small or large proportionof cells of an organ. To overcome this we also show the labellingpattern of typical examples for eight of the 14 regions (Fig. 4).The lateral regions label a similar proximal/distal part of the gutto the ventral regions and so were not included here. Drawingsof a number of sections from each of the eight typical examplesare shown, with the position of the FDA label marked in green.Above the typical examples are a set of standard sections anda drawing of a whole-mount embryo. The organs labelled ineach drawing can be identified by reference to the set ofstandard sections and the position of the section in the tadpolebody can then be established by reference to the drawing of thewhole mount. These diagrams give an indication of the amountof labelling in each organ. Finally we present the data in twotypes of summary diagram (Fig. 5A-D). The first shows adrawing of a neurula-stage embryo labelled with the organs that

A. D. Chalmers and J. M. W. Slack

385Fate maps and morphogenesis in Xenopus gut

the extreme anterior, dorsal, ventral and extreme posteriorregions are fated to form (Fig. 5A). For the sake of clarity thelateral regions, which gave similar results to the ventral regions,have not been included in the diagram (they are included in Fig.5D, see later). The second type of diagram shows how the earlyendoderm projects on to the later gut (Fig. 5B,C). Theprojection of the ventral endoderm (with extreme anterior andextreme posterior regions) and the dorsal endoderm (withextreme anterior and posterior regions) is shown (Fig. 5B,C). Itis important to realise that the shaded regions of the gut aremeant to show that labelled cells were found in these regionsrather than every cell in these regions was labelled.

Overview of the endoderm fate mapEach region of the neurula was found to be reproducibly fatedto form part of the gut or respiratory system of the tadpole. Thisshows that all regions of the endoderm do contribute to thetadpole gut or respiratory epithelia. Not surprisingly, theanterior endoderm was found to be fated to form proximal gutepithelia while the posterior endoderm formed distal epithelia.To the resolution of this study there is a smooth projectionwithout discontinuities from early to late stages relative to allanatomical axes: anterior/posterior, dorsal/ventral and left/right.However, there is some quantitative deformation in that thedorsal endoderm was generally found to be fated to form moreanterior structures than the ventral endoderm (discussed in moredetail below). So while the dorsal endoderm remains oppositethe ventral endoderm it ends up opposite ventral cells thatoriginated from a more anterior position in the endoderm. Thelater gut has a left/right asymmetry and there is currently a lotof interest in how this is established (e.g. Campione et al., 1999;reviewed by Yost, 1998). We were interested in whether therewould be any left/right asymmetry in the fate of the endodermbut found no convincing difference between the fates of the leftand right grafts. Presumably therefore the asymmetry must arise

without large differences in cell fate or migration between theright and left sides of the embryo.

Specific regions of the endoderm fate mapAnterior regionsThe extreme anterior region (1) labelled the epithelium of thepharynx and the tongue (orange in Fig. 5). All the anterior grafts(2-5, red in Fig. 5) labelled the epithelium of the pharynx,oesophagus, stomach and proximal small intestine (sia).However, the anterior ventral (5) and lateral grafts (3+4) labelledless pharynx and more distal parts of sia than the anterior dorsalgrafts (2) (compare Fig. 5B with C). So, the dorsal endoderm isfated to form more anterior structures than the lateral and ventralendoderm. The anterior ventral (5) grafts also labelled thetrachea, lungs, liver, gall bladder, pancreas and bile duct,showing that a large number of organs are fated to form fromthis small anterior ventral portion of the endoderm. The anteriorleft and right (3+4) grafts also labelled the pancreas but not theliver. This shows that the ventral pancreatic rudiment but not theliver has a lateral as well as a ventral component. However, finergrain fate mapping studies will be required to establish to whatextent the ventral pancreatic rudiment extends laterally and sohow different in origin it is from the rudiment of the liver.

Middle regionsThe middle dorsal region (6, blue in Fig. 5) labelled the sia andthe pancreas. This means that the dorsal pancreas forms froma more posterior region than the ventral pancreas. The middleright (7), left (8) and ventral (9) regions labelled the distalportion of sia and a large part of sib (blue in Fig. 5). Therefore,like the anterior regions, the middle lateral and ventral graftslabelled more distal parts of the gut than the dorsal grafts.

Posterior regionsThe posterior right (11), left (12) and ventral regions (13)

Position n A ph P ph tn gills tr lungs liver gb pa bd oe st sia sib sic lic lid pr(1) Extreme anterior 9 89 89 78 33 44 44 11 22 22 22 22 22 11

(2) Anterior dorsal 8 25 100 13 38 75 63 63(3) Anterior right 6 50 17 33 67 50 83 100 83(4) Anterior left 6 33 17 17 17 17 50 33 83 83(5) Anterior ventral 9 67 67 44 55 55 100 67 89 89 78 78 100 22

(6) Middle dorsal 6 100 17 34 100 17(7) Middle right 6 17 17 100 83 67(8) Middle left 7 14 86 100 43(9) Middle ventral 8 75 100

(10) Posterior dorsal 6 50 50 83 83 33 17(11) Posterior right 6 83 100 50 33(12) Posterior left 10 10 90 100 40 30(13) Posterior ventral 6 50 83 67 67

(14) Extreme posterior 6 50 100 100

The percentage of cases where each organ was labelled is shown for each of the 14 regions. Organs that were labelled in at least 50% of the cases were included in the fate map and shaded grey in this table. A ph, anterior pharynx; P ph, posterior pharynx; tn, tongue; tr, trachea; gb, gall bladder; pa, pancreas; bd, bile duct; oe, oesophagus; st, stomach; sia, proximal small intestine; sib, external coil of the small intestine; sic, internal coil of the small intestine; lic, internal coil of the large intestine; lid, distal large intestine; pr, proctodaeum.

Table 1. The endoderm fate map: organs labelled from grafts in each region

386

(green in Fig. 5) labelled the distal part of sib, sic and a smallpart of the large intestine. The posterior dorsal region (13)labelled cells in a large span of the intestine from sia throughto sic. Therefore, the fact that the anterior and middle dorsalendoderm is shifted to the anterior compared with the lateral

and ventral endoderm is compensated for by the spreading outof the posterior dorsal endoderm over a large part of theintestine (compare Fig. 5B,C). The extreme posterior graft (14,black in Fig. 5) labelled the most distal parts of the gut: thelarge intestine and the proctodaeum.

A. D. Chalmers and J. M. W. Slack

Fig. 4. Representative examples of the fatemapping. Drawings of sections fromrepresentative examples of eight of the 14regions are shown. The position of the FDAlabel has been added to the drawings (green).A whole-mount drawing and drawings ofstandard sections are also shown to aidinterpretation of the labelled sections (adaptedfrom Chalmers and Slack, 1998). ph, pharynx;tn, tongue; ov, otic vesicle; no, notochord; ht,heart; lv, liver; tr, trachea; sia, proximal smallintestine; sib, external coil of small intestine;sic, internal coil of small intestine; lic, internalcoil of large intestine; lid, distal largeintestine; pa, pancreas; bd, bile duct; lu, lungs;nd, nephritic ducts; pr, proctodaeum.

387Fate maps and morphogenesis in Xenopus gut

The fate map for the smooth muscle layer The fate map for the smooth muscle layer is shown in Table 2and Fig. 5E. The summary diagram shows which organs of thegut the lateral and ventral mesoderm is fated to cover in smoothmuscle (Fig. 5E). The lateral rudiments are shown in themiddle of the dorsal/ventral axis to depict their origin from theside of the embryo. The mesoderm fate map can be comparedwith the endoderm fate map (Fig. 5D). For comparison withthe smooth muscle fate map this diagram shows only theorgans of the digestive system that are covered in smoothmuscle and, unlike Fig. 5A, includes the lateral rudiments,shown in the middle of the embryo.

The middle right (7) and left (8) grafts labelled the smoothmuscle layer surrounding the oesophagus, stomach and sia. Themiddle ventral grafts (5) labelled the posterior part of sia and sib.The posterior right (11) and left grafts (12) labelled the distal partof sib, sic and the proximal part of the large intestine. Theposterior ventral graft (13) labelled the majority of the large

intestine. The other regions were not found reproducibly tocontribute to the smooth muscle of the gut but were fated to formother mesodermal tissues such as the notochord (this is consistentwith a previous fate map of the early mesoderm; Keller, 1976).These results show that there were several differences betweenthe origins of the epithelia and the smooth muscle. All regions inthe endoderm were found to form some part of the gut epitheliumwhile the smooth muscle originated from just a subset of regionsin the early mesoderm. Another difference between themesoderm and endoderm is that the lateral and ventral endodermlabelled the same proximal/distal segment of the gut. In contrast,the lateral mesoderm is fated to form more proximal smoothmuscle than the ventral mesoderm. These differences mean thatthe cells of the future epithelium and smooth muscle layer in themiddle of the gut overlay each other in the early embryo. Incontrast, the cells of the epithelia and smooth muscle layer at theproximal end, for example oesophagus and stomach, and distalend, for example large intestine, do not.

Fig. 5. The endoderm and smoothmuscle layer fate maps. (A) Theendoderm fate map. The organ rudimentsof the extreme anterior (1) (yellow),anterior dorsal (2) and ventral (5) (red),middle dorsal (6) and ventral (9) (blue),posterior dorsal (10) and ventral (13)(green) and extreme posterior (14)(black) endoderm are labelled at theposition they form on a drawing of astage 14 embryo. For the sake of claritythe lateral rudiments have not beenincluded. (B) Projection of the earlyventral endoderm onto the tadpole gut.The extreme anterior (1) (yellow),anterior ventral (5) (red), middle ventral(9) (blue), posterior ventral (13) (green)and extreme posterior regions (14)(black) are shaded in the embryo and inthe regions they will give rise to in thetadpole gut. The drawings are viewedfrom the ventral side with anterior at thetop. (C) Projection of the early dorsalendoderm. As in B but for the dorsalrather than ventral regions. (D) Locationof the dorsal, lateral and ventral digestivetract rudiments for comparison with thesmooth muscle layer fate map. Thelateral rudiments are shown in themiddle of the dorsal/ventral axis torepresent their position on the side of theembryo. (E) Smooth muscle layer fatemap. The position in the mesoderm ofthe lateral and ventral rudiments for thesmooth muscle layer of the gut is shownon a drawing of a stage 14 embryo. Noneof the dorsal regions were fated to formsmooth muscle. (F) Presumptive geneexpression domains in the dorsal andventral Xenopus endoderm. Regions ofthe endoderm that will give rise totissues that express Xlhbox8 or IFABP innormal development are highlighted withcoloured diagonal lines. ph, pharynx; tn, tongue; tr, trachea; lu, lungs; lv, liver; gb, gall bladder; pa, pancreas; bd, bile duct; oe, oesophagus; st,stomach; si, small intestine; sia, proximal small intestine; sib, external coil of small intestine; sic, internal coil of small intestine; li, largeintestine; lic, internal coil of large intestine; lid, distal large intestine; pr, proctodaeum.

388

The yolky cells on the floor of the archenteron areincorporated into the gut epithelium by radialintercalationAt neurula stages the Xenopus endoderm consists of a singlelayer of dorsal cells and many layers of ventral cells (Fig. 6A).At stage 39/40 the archenteron lumen in the trunk region hasbecome a very narrow cavity and often appears completelyoccluded (Fig. 6B). By stage 45 the endoderm has undergonemassive elongation and formed the single layered epitheliumthat surrounds the gut cavity (Fig. 6C). Measurement of thepharynx and dissected gut from embryos showed that betweenstage 14 and stage 45 the endoderm increased in lengthapproximately 5 times. From stage 41 to stage 45, when thegut is transformed from a straight tube to a coiled tube, the gutincreases in length approximately 3.5 times. It is not clear howthis transformation from the short, many layered, embryonicendoderm to the long, single layered, gut epithelia takes place,so we went on to investigate endoderm morphogenesis in moredetail.

A crucial question of gut formation is whether cells fromeach of the four positions shown in Fig. 6A are incorporatedinto the gut epithelium. If the dorsal, floor, middle and ventralcells (Fig. 6A) are all incorporated, then massive cellrearrangement must be occurring. In order to investigate this, asmall clump of cells in the middle of the archenteron floor werelabelled with DiI (Fig. 6D, arrow; the archenteron is shrunkbecause of the replacement of the roof). At stage 39/40, whenthe archenteron has narrowed, the labelled floor cells were stillfound in a quite dorsal position (Fig. 6E) but had become spreadout over a short stretch of the anterior/posterior axis. If a cleararchenteron cavity was present the label was normally foundabutting the ventral side of the lumen, although occasionally alabelled cell was seen slightly more ventrally, separated from

the lumen. At stage 45, once the intestinal epithelium hasformed, the labelled cells were found incorporated into theintestinal epithelium (Fig. 6F). The middle endodermal cellsthat lie between the floor of the archenteron and the very ventralendoderm were then labelled with DiI (Fig. 6G). The middlecells had remained in a middle position by stage 39/40 (Fig.6H) and were incorporated into the gut at stage 45 (Fig. 6I).This shows that these yolky cells of the archenteron floor andmiddle endoderm do not disintegrate during development butbecome an integral part of the gut.

The extreme ventral and dorsal endoderm were then labelledto show how these regions move relative to the floor and middlecells. The ventralmost cells of the endoderm that lie next to themesoderm were labelled (Fig. 6J). At stage 39/40 they werefound to have maintained their ventral position next to themesoderm (Fig. 6K) and at stage 45 were found incorporatedinto the intestinal epithelium (Fig. 6L). The proximal/distalposition of the label in the intestine was consistent with the FDAfate mapping. The dorsal endoderm cells were then labelled(Fig. 6M). These cells were still dorsal at stage 39/40 (Fig. 6N)and were also incorporated into the epithelium at stage 45 (Fig.6O). As expected from the FDA fate mapping the dorsalendoderm labelled more proximal small intestine than theventral endoderm, and often labelled the pancreas as well. Ineach case (dorsal, floor, middle and ventral) a small coherentpatch of labelled cells spread out to form a proximal/distal stripof labelled cells interspersed with unlabelled ones. Since cellsfrom each position became incorporated into the single layeredepithelium we inferred that radial intercalation of endodermalcells must be occurring.

To prove that radial intercalation was occurring and toestablish in which directions it was occurring, we carried outdouble labelling. The floor cells were labelled with DiI and theventral endoderm labelled with a shallow graft (containingectoderm, mesoderm and 1-2 layers of endoderm; see Materialsand Methods) of FDA-labelled tissue. Surprisingly, this showedthat the floor cells and the ventral cells ended up on oppositesides of the gut tube at stage 45 (Fig. 7B). Conversely, we foundthat the cells of the archenteron floor end up on the same sideof the gut tube as those from the dorsal roof. This was shownby labelling the floor cells with DiI and the dorsal cells with agraft of FDA tissue (equivalent to middle and posterior dorsalregions). The dorsal and floor cells were later foundintermingled on the same side of the gut tube (Fig. 7C) showingthat the floor cells undergo radial intercalation with the dorsalcells as the epithelium is forming. The middle cells were thenlabelled with DiI and the ventral cells labelled with FDA. Atstage 45 these cells were found intermingled on the same sideof the gut tube (Fig. 7D). The DiI-labelled cells were also foundjust lateral to the FDA-labelled cells (Fig. 7D, a couple of DiI-labelled cells can be seen to the left of the gut in a more lateralposition). The double labelling showed that radial intercalationwas occurring between the floor and dorsal cells and the middleand ventral cells. Radial intercalation increases the surface areaof a tissue (Keller, 1980, 1987; Wolpert, 1998) and so could bedriving the elongation of the gut tube (see Discussion).

The archenteron narrows during developmentbefore reopening to split the ventral endoderm andproduce the definitive gut cavityThe double labelling demonstrates that the cells on the roof and

A. D. Chalmers and J. M. W. Slack

Position n oe st sia sib sic lic lid(1) Extreme anterior 9

(2) Anterior dorsal 8(3) Anterior right 7(3) 29 14 29(4) Anterior left 7(2) 43(5) Anterior ventral 6

(6) Middle dorsal 6 16 16 16(7) Middle right 4 50 75 100 25(8) Middle left 5(1) 60 60 60 20(9) Middle ventral 6 50 100

(10) Posterior dorsal 6(11) Posterior right 6(4) 67 67 83 50(12) Posterior left 7(3) 43 100 43 14(13) Posterior ventral 7 29 57 100

(14) Extreme posterior 6

The percentage of cases where the smooth muscle for each organ was labelled is shown for each of the 14 regions. Organs that were labelled in at least 50% of the cases were included in the fate map and shaded grey in this table. The number of cases scored as whole mounts rather than in sections is shown in brackets. oe, oesophagus; st, stomach; sia, proximal small intestine; sib, external coil of the small intestine; sic, internal coil of the small intestine; lic, internal coil of the large intestine; lid, distal large intestine.

Table 2. Gut smooth muscle fate map: organs labelled from grafts in each region

389Fate maps and morphogenesis in Xenopus gut

floor of the archenteron end up on one side of thegut cavity while the cells of the middle and ventralendoderm end up on the other side. This result isconsistent with the archenteron closing and thedefinitive gut cavity opening up de novo in a positionthat is ventral and completely separate from theremaining archenteron cavity. However, the resultscould also be explained by the archenteronnarrowing and then widening to split the ventralendoderm and form the definitive gut cavity. Todistinguish between these two possibilities it isnecessary to label the cells surrounding thearchenteron and to follow the progress of the cavityduring development.

Sulfo-NHS-LC-biotin has previously been used tolabel the superficial cells of pre-gastrulation Xenopusembryos (Muller and Hausen, 1995; Minsuk andKeller, 1997). After gastrulation these superficialcells will give rise to the cells that line thearchenteron cavity (Keller, 1975; Smith andMalacinski, 1983; Minsuk and Keller, 1997) and sothis method can be used to label the archenteron cells.The biotin treatment gave good strong labelling ofthe archenteron cells (Fig. 8B, arrow) with nolabelling in the rest of the endoderm or in untreatedcontrol embryos (Fig. 8A). At later stages endogenous stainingwas seen in the epidermis of untreated control embryos but notin the gut (Fig. 8C). By stage 38 the archenteron had closed toa narrow slit through much of the gut (Fig. 8D). At this stage anumber of labelled cells can be seen to have separated from theresidual archenteron cavity and now lie ventral to it (Fig. 8D,E,arrow). The number of these cells varies slightly betweenindividuals and also along the anterior/posterior axis with fewerisolated cells seen in the posterior gut. This shows that closureof the archenteron occurs by cells becoming separated from thearchenteron. At stage 40 the archenteron has narrowed furtherso that it often appears to have completely closed. Despite this,a small ring of the biotin-labelled cells was always clearlyvisible (Fig. 8F, arrow). The cells that separated from theclosing cavity, clearly visible at stage 38, appear to have lost

their label by stage 40. This is probably because as these cellsleave the epithelium they lose their polarisation and there is aconsequent increase in the turnover of cell surface proteins thatwould remove the biotin label.

In the more posterior regions of the stage 40 intestine thearchenteron cavity does not close up so much (Fig. 8G) and itappears that in certain regions it has started to widen in aventral direction. This gives rise to a cavity lined with a regionof labelled cells (arrow) and one of unlabelled cells(arrowhead). At stage 42 the archenteron in the middle of thegut is still very small while in the posterior of the gut the cavityhas continued to widen (Fig. 8H). Once the gut cavity hasformed the biotin labelling can be seen in one small segmentof the circumference (Fig. 8I).

This biotin labelling study shows that the archenteron

Fig. 6. DiI labelling of the Xenopus endoderm.(A) Section of control stage 14 embryo. The fourlabelling positions are marked with arrows. (B) Sectionof control stage 40 embryo. (C) Whole-mount stage 45control gut. (D) Section of floor DiI labelling at stage 14(the archenteron is shrunk because of the replacement ofthe roof). Insert shows high magnification view.(E) Section of floor DiI labelling visualised at stage 39.(F) Floor DiI labelling visualised in a stage 45 wholemount. (G) Section of middle labelling at stage 14.(H) Section of middle DiI labelling visualised at stage 39.(I) Middle DiI labelling visualised in a stage 45 wholemount. (J) Section of ventral DiI labelling at stage 14.(K) Section of ventral DiI labelling visualised at stage 39.(L) Ventral DiI labelling visualised in a stage 45 wholemount. (M) Section of dorsal DiI labelling at stage 14.Insert shows high magnification view. (N) Section ofdorsal DiI labelling visualised at stage 39. (O) Dorsal DiIlabelling visualised in a stage 45 whole mount. Arrowshighlight DiI label. no, notochord; ar-r, archenteron roof;ar-f, archenteron floor.

narrows to a very small cavity and some of its lining cellsbecome dispersed into the ventral endoderm. Later thedefinitive gut cavity is formed by enlargement of a splitoriginating from the archenteron remnant. This split divides theendoderm down the middle and the radial intercalation,demonstrated by the DiI labelling, transforms the short thickundifferentiated gut tube into the long thin epithelium.

DISCUSSION

Origin of the gut and respiratory epitheliaThe endoderm fate map shows that the axes of the endodermare maintained in the later gut and that all regions of theendoderm give rise to part of the gut or respiratory epithelium.This means that the mechanisms of gut morphogenesis producea smooth projection from the neurula to the tadpole stage.Although the axes of the endoderm were maintained, theanterior and middle dorsal endoderm were found to form moreproximal gut structures than the ventral endoderm. The anteriorshift in the anterior and middle dorsal endoderm wascompensated for by the posterior dorsal endoderm, whichspread out over a large part of the gut. Convergent extensionof the dorsal axial tissues continues to occur during neurulastages (reviewed in Keller, 1987) and could possibly accountfor this distortion of the dorsal endoderm. A similar anteriorshift has also been shown to occur in the Cerberus expressing

dorsal leading edge of the gastrulating endomesoderm(Bouwmeester et al., 1996). These cells move anteriorly andthen ventrally to end up in the ventral foregut region.

In general, the presumptive organ rudiments are found fairlyevenly spread across the embryonic endoderm but an exceptionto this is the anterior ventral region, which gives rise to a largenumber of organ rudiments. This region is also of interestbecause it contains the Cerberus expressing cells (see above). Itwill be important to understand how so many organs are inducedto form from the anterior ventral region. For example, it wouldbe interesting to know if this region contains precursor cells withmore than one fate or consists of closely spaced small regionswith a single fate, two possibilities that this study does notdistinguish. The fate map also shows the difference in originbetween the rudiments of the dorsal and ventral pancreas. Thereis currently a lot of progress being made in the study of pancreasdevelopment and it will be interesting to see how the differentorigins of the dorsal and ventral pancreas are reflected indifferent mechanisms of induction (e.g. Ahlgren et al., 1997;Hebrok et al., 1998; Kim and Melton, 1998; Harrison et al.,1999; Li et al., 1999).

The anatomy of the early chick embryo is very different to theanatomy of early Xenopus embryos, but despite these differencesthere are a number of similarities between the endoderm fatemap presented here and the fate map for the chick endoderm(Matsushita, 1996). The anterior/posterior organisation of the

390 A. D. Chalmers and J. M. W. Slack

Fig. 7. DiI/FDA double labelling of theXenopus endoderm. (A) Section of stage 14control embryo. The four labelling positionsare marked with arrows. (B) Floor DiI andventral FDA double labelling visualised in astage 45 whole mount. (C) Floor DiI anddorsal FDA double labelling visualised in astage 45 whole mount. (D) Middle DiI andventral FDA double labelling visualised in astage 45 whole mount.

Fig. 8. The archenteroncavity during development.The superficial cell layer thatgives rise to the cells liningthe archenteron was labelledat stage 9/10. The labelledcells lining the archenteroncould then be followedduring development.(A) Unlabelled stage 14embryo. (B) Biotin-labelledcells at stage 14 (arrow).(C) Unlabelled stage 38embryo showingendogenous signal in theepidermis but not the gut.(D,E) Biotin-labelled cells(arrow) at stage 38 showingpartial closure ofarchenteron. (F) Biotin-labelled cells (arrow) atstage 40 showing smallpersistent archenteron. Thelabel is lost from the internalised cells. (G) Biotin-labelled cells in the posterior gut of a stage 40 embryo showing new cavity opening (arrow,biotin; arrowhead, unlabelled). (H) Biotin-labelled cells in the posterior gut at stage 42 showing cavity is only partially labelled (arrow, biotin;arrowhead, unlabelled). (I) Biotin-labelled cells (arrow) in stage 44 gut showing partial labelling of new cavity. Bar, 100 µm; 50 µm (E,G-I).

391Fate maps and morphogenesis in Xenopus gut

chick endoderm is maintained in the later gut, and the dorsalendoderm also gives rise to more proximal epithelia than thelateral endoderm. A fate map showing where the endodermordinates from in the early Zebrafish embryo has also recentlybeen published (Warga and Nusslein-Volhard, 1999), but it is notpossible to make detailed comparisons as it is from acomparatively much earlier stage than our fate map.

The endoderm fate map and gene expressiondomainsThe intestinal fatty acid binding protein (IFABP) is expressed inthe tadpole small intestine and Xlhbox8 is expressed in the tadpolepancreas (Wright et al., 1988; Shi and Hayes, 1994; Chalmers andSlack, 1998). The fate map shows which parts of the endodermwill form these regions of the embryo, making it possible topredict which regions of the endoderm will express these genes inlater development (Fig. 5F). As more genes are shown to beexpressed in the tadpole gut they can be added to the fate map.

Origin of the smooth muscle layer and signallingbetween the mesodermal and endodermal tissues The prospective regions for the epithelial tissues are spreadthroughout the endoderm. In contrast the prospective region forthe smooth muscle layer is restricted to only part of themesoderm, meaning that for most organs of the gut theprospective epithelium and smooth muscle do not overlie eachother in the early embryo. The chick fate maps (Matsushita,1995, 1996) show that the presumptive rudiments of theendoderm are located in a more anterior position than thecorresponding ones in the mesoderm. So in both the chick andXenopus the two sets of rudiments do not overlie each other inthe early embryo. This suggests that reciprocal signallingbetween the two germ layers cannot begin until the two tissuelayers come into accord later in development. This does notpreclude early signals from the mesoderm to the endoderm, butit does mean that any early signals must be different from thewell known late signals responsible for the regional-specificinductions from the mesenchyme (Haffen et al., 1987; Yasugiand Mizuno, 1990; Rawdon and Andrew, 1993).

A model for morphogenesis of the gutBased on the combined evidence from the endoderm fate map,the DiI labelling and the biotin surface labelling experiments, wepropose a model for the morphogenesis of the gut (Fig. 9). Fromneurula stages to stage 38 the archenteron gradually closes. Thisis achieved by cells leaving the cavity lining to be incorporatedinto the deeper endoderm. These movements leave the originaldorsal (red), floor (blue), middle (black) and ventral (green) cellgroups in similar relative positions. From stage 38 to stage 45 theremains of the archenteron expands to split the ventral endoderm.This gives rise to the definitive gut cavity, which contains the cellsof the original archenteron and the more ventral endoderm.

Cell rearrangements have been shown to be a powerful forcein morphogenesis (reviewed in Keller, 1987; Wolpert, 1998). Agood example is gastrulation in Xenopus embryos, where bothradial and mediolateral intercalations drive elongation (Keller,1980, 1984; Keller et al., 1985). The DiI labelling shows that thefloor endoderm intercalates with the dorsal endoderm and themiddle endoderm intercalates with the ventral/lateral endoderm.This radial intercalation would cause the opening of the gutcavity from the remains of the archenteron and would alsoincrease the surface area of the gut accounting for at least partof the gut elongation.

Cell rearrangements can be an active, force-generatingprocess, or a passive process that responds to external forces(Keller, 1987). The DiI labelling shows that the radialintercalation occurs between stage 40 and stage 45, when the gutis transformed from a straight tube to a complex coiled structureand elongates approximately 3.5 times. Once coiling has startedthe elongation of the endoderm can no longer be linked to theelongation of the body axis and so must occur by a separateactive process. We propose that this process is the radialintercalation of the endoderm and that it drives both theelongation of the gut and the opening of the gut cavity.

The authors would like to thank the Wellcome Trust for providing aPrize Studentship for A. D. C. and for providing the confocal facilityin the department (grant no. 049452). We would also like to thankCaroline Beck, Bea Christen, Marko Horb, Benjamin Rawdon andAndrew Ward for comments on the manuscript.

Fig. 9. Morphogenesis of the gut epithelium. From stage 14 to stage38 the embryonic archenteron narrows to a small cavity. During thistime the dorsal (red), floor (blue), middle (black) and ventral (green)cell populations retain their relative positions. After stage 38 the gutcavity starts to open from the remains of the archenteron to split theventral endoderm. This produces a cavity, part of which originatedfrom the archenteron (solid line) and part is the new cavity (dashedline). By stage 45 the gut cavity is fully open. The splitting of theventral endoderm places the cells of the archenteron (red + blue) onone side of the cavity and the middle (black) and ventral endodermcells (green) on the other side. Radial intercalation may drive boththe opening of the gut cavity and the elongation of the gut.

392

REFERENCES

Ahlgren, U., Plaff, S. L., Jessell, T. M., Edlund, T. and Edlund, H. (1997).Independent requirement for ISL1 in formation of pancreatic mesenchymeand islet cells. Nature 385, 257-260.

Andrew, A. and Rawdon, B. B. (1990). Intestinal mesenchyme provokesdifferentiation of intestinal endocrine cells in gizzard endoderm.Differentiation 43, 165-174.

Beck, C. W. and Slack, J. M. W. (1999). A developmental pathway controllingoutgrowth of the Xenopus tail bud. Development 126, 1611-1620.

Bienz, M. (1994). Homeotic genes and positional signalling in the Drosophilaviscera. Trends Genet. 10, 22-26.

Bouwmeester, T., Kim, S., Sasai, Y., Lu, B. and De Robertis, E. M. (1996).Cerberus is a head-inducing secreted factor expressed in the anteriorendoderm of Spemann’s organizer. Nature 382, 595-601.

Campione, M., Steinbeisser, H., Schweickert, A., Deissler, K., Bebber, F. V.,Lowe, L. A., Nowotschin, S., Viebahn, C., Haffter, P., Kuehen, M. R. andBlum, M. (1999). The homeobox gene Pitx2: mediator of asymmetric left-rightsignalling in vertebrate heart and gut looping. Development 126, 1225-1234.

Chalmers, A. D. and Slack, J. M. W. (1998). Development of the gut inXenopus laevis. Dev. Dynam. 212, 509-521.

Dale, L. and Slack, J. M. W. (1987). Fate map for the 32-cell stage of Xenopuslaevis. Development 99, 527-51.

Gimlich, R. L. and Braun, J. (1985). Improved fluorescent compounds fortracing cell lineage. Dev. Biol. 109, 509-514.

Godsave, S. F., Isaacs, H. V. and Slack, J. M. W. (1988). Mesoderm-inducingfactors: a small class of molecules. Development 102, 555-566.

Goette, A. (1875). Die Entwicklungsgeschichte der Unke. Leipzig: LeopoldVoss.

Haffen, K., Kedinger, M. and Simon-Assmann, P. (1987). Mesenchyme-Dependent Differentiation of Epithelial Progenitor Cells in the Gut. J. Ped.Gastroenterol. Nutr. 6, 14-23.

Harrison, K. A., Thaler, J., Pfaff, S. L., Gu, H. and Kehrl, J. H. (1999).Pancreas dorsal lobe agenesis and abnormal islets of Langerhans in Hlxb9-deficient mice. Nature Genetics 23, 71-75.

Hebrok, M., Kim, S. K. and Melton, D. A. (1998). Notochord repression ofendodermal sonic hedgehog permits pancreas development. Genes Dev. 12,1705-1713.

Henry, G. L. and Melton, D. A. (1998). Mixer, a Homeobox Gene Requiredfor Endoderm Development. Science 281, 91-96.

Hudson, C., Clements, D., Friday, R. V., Stott, D. and Woodland, H. R.(1997). Xsox17a and -B Mediate Endoderm Formation in Xenopus. Cell 91,397-405.

Ishizuya-Oka, A. and Mizuno, T. (1984). Intestinal cytodifferentiation in vitroof chick stomach endoderm induced by the duodenal mesenchyme. J.Embryol. Exp. Morph. 82, 163-176.

Joseph, E. M. and Melton, D. A. (1998). Mutant Vg1 ligands disrupt endodermand mesoderm formation in Xenopus embryos. Development 125, 2677-2685.

Keller, R. E. (1975). Vital dye mapping of the gastrula and neurula of Xenopuslaevis. I. prospective areas and morphogenetic movements in the superficiallayer. Dev. Biol. 42, 222-241.

Keller, R. E. (1976). Vital dye mapping of the gastrula and neurula of Xenopuslaevis. II. Prospective areas and morphogenic movements of the deep layer.Dev. Biol. 51, 118-137.

Keller, R. E. (1980). The cellular basis of epiboly: an SEM study of deep-cellrearrangement during gastrulation in Xenopus laevis. J. Embryol. Exp. Morph.60, 201-234.

Keller, R. E. (1984). The cellular basis of gastrulation in Xenopus laevis: Active,postinvolution convergence and extension by mediolateral interdigitation. Am.Zool. 24, 589-603.

Keller, R. E. (1987). Cell Rearrangement in Morphogenesis. Zool. Sci. 4, 763-779.

Keller, R. E., Danilchik, M. and Gimlich, R. (1985). The function andmechanism of convergent extension during gastrulation of Xenopus laevis. J.Embryol. Exp. Morph. 89, 185-209.

Kordylewski, L. (1983). Light and Electron Microscope Observations of theDevelopment of Intestinal Musculature in Xenopus. Z. Mikrosk.-anat. Forsh97, 719-734.

Kim, S. K. and Melton, D. A. (1998). Pancreas development is promoted bycyclopamine, a hedgehog signaling inhibitior. Proc. Natl. Acad. Sci. USA 95,13036-13041.

Lemaire, P., Darras, S., Caillol, D. and Kodjabachian, L. (1998). A role forthe vegetally expressed Xenopus gene Mix. 1 in endoderm formation and inthe restriction of mesoderm to the marginal zone. Development 125, 2371-2380.

Li, H., Arber, S., Jessell, T. M. and Edlund, H. (1999). Selective agenesis ofthe dorsal pancreas in mice lacking homeobox gene Hlxb9. Nature Genetics23, 67-70.

Mathews, W. W. and Schoenwolf, G. C. (1998). Atlas of DescriptiveEmbryology. New Jersey: Prentice-Hall.

Matsushita, S. (1995). Fate mapping study of the splanchnopleural mesodermof the 1.5-day-old chick embryo. Roux’s Arch. Dev. Biol. 204, 392-399.

Matsushita, S. (1996). Fate mapping study of the endoderm of the 1.5 day-oldchick embryo. Roux’s Arch. Dev. Biol. 205, 225-231.

Minsuk, S. B. and Keller, R. E. (1997). Surface mesoderm in Xenopus: arevision of the stage 10 fate map. Dev. Genes Evol. 207, 389-401.

Muller, H. J. and Hausen, P. (1995). Epithelial cell polarity in early Xenopusdevelopment. Dev. Dynam. 202, 405-420.

Nieuwkoop, P. D. and Faber, J. (1967). Normal Table of Xenopus laevis(Daudin). Amsterdam: North-Holland.

Okada, T. S. (1957). The Pluripotency of the Pharyngeal Primordium inUrodelan Neurulae. J. Embryol. Exp. Morph. 5, 438-448.

Okada, T. S. (1960). Epithelio-Mesenchymal Relationships in the RegionalDifferentiation of the Digestive Tract in the Amphibian Embryo. Roux’s Arch.Ent. Mech. 152, 1-21.

Rawdon, B. B. and Andrew, A. (1993). Origin and differentiation of gutendocrine cells. Histol. Histopath. 8, 567-580.

Roberts, D. J., Johnson, R. L., Burke, A. C., Nelson, C. E., Morgan, B. A.and Tabin, C. (1995). Sonic hedgehog is an endodermal signal inducingBmp-4 and Hox genes during induction and regionalization of the chickhindgut. Development 121, 3163-3174.

Roberts, D. J., Smith, D. M., Goff, D. J. and Tabin, C. J. (1998). Epithelial-mesenchymal signaling during the regionalization of the chick gut.Development 125, 2791-2801.

Sasai, Y., Lu, B., Piccolo, S. and DeRobertis, E. (1996). Endoderm inductionby the organiser-secreted factors chordin and noggin in Xenopus animal caps.EMBO J. 15, 4547-4555.

Shi, Y.-B. and Hayes, W. P. (1994). Thyroid Hormone-Dependent Regulationof the Intestinal Fatty Acid-Binding Protein Gene during AmphibianMetamorphosis. Dev. Biol. 161, 48-58.

Smith, J. C. and Malacinski, G. M. (1983). The origin of the mesoderm in ananuran, Xenopus laevis, and a urodele, Ambystoma mexicanum. Dev. Biol. 98,250-4.

Tahara, Y. and Nakamura, O. (1961). Topography of the PresumptiveRudiments in the Endoderm of the Anuran Neurula. J. Embryol. Exp. Morph.9, 138-158.

Warga, M. R. and Nusslein-Volhard, C. (1999). Origin and development ofthe Zebrafish endoderm. Development 126, 827-838.

Wolpert, L. (1998). Principles of Development. Oxford: Oxford UniversityPress.

Wright, C. V., Schnegelsberg, P. and De Robertis, E. M. (1988). XlHbox 8:a novel homeo protein restricted to a narrow band of endoderm. Development104, 787-794.

Yasugi, S. and Mizuno, T. (1978). Differentiation of the digestive tractepithelium under the influence of the heterologous mesenchyme of thedigestive tract in the bird embryos. Dev. Growth Diff. 20, 261-267.

Yasugi, S. and Mizuno, T. (1990). Mesenchymal-Epithelial Interactions in theOrganogenesis of Digestive Tract. Zool. Sci. 7, 159-170.

Yokouchi, Y., Sakiyama, J. and Kuroiwa, A. (1995). Coordinated Expressionof Abd-B Subfamily Genes of the HoxA Cluster in the Developing DigestiveTract of Chick Embryo. Dev. Biol. 169, 76-89.

Yost, H. J. (1998). Left-right development in Xenopus and Zebrafish. Sem. CellDev. Biol. 9, 61-66.

Zhang, J., Houston, D. W., King, M. L., Payne, C., Wylie, C. and Heasman,J. (1998). The role of maternal VegT in establishing the primary germ layersin Xenopus embryos. Cell 94, 515-524.

A. D. Chalmers and J. M. W. Slack