453Development 124, 453-460 (1997)Printed in Great Britain © The Company of Biologists Limited 1997DEV9489

Induction of the primary dorsalizing center in Xenopus by the Wnt/GSK/β-

catenin signaling pathway, but not by Vg1, Activin or Noggin

François Fagotto, Kathleen Guger and Barry M. Gumbiner*

Cellular Biochemistry and Biophysics Program, Memorial Sloan-Kettering Cancer Center, New York, 1275 York Avenue, Box 564,New York, NY 10021, USA

*Author for correspondence

The molecular nature of the primary dorsalizing inducingevent in Xenopus is controversial and several secretedfactors have been proposed as potential candidates: Wnts,Vg1, Activin and Noggin. Recent studies, however, haveprovided new insight into the activity of the dorsalizingregion, called the Nieuwkoop Center. (1) The activity of thisdorsalizing center involves an entire signal transductionpathway that requires maternal β-catenin (Heasman, J.,Crawford, A., Goldstone, K., Garner-Hamrick, P.,Gumbiner, B., McCrea, P., Kintner, C., Noro, C. Y. andWylie, C. (1994) Cell 79, 791-803). (2) A transcription factorwith potent dorsalizing activity, Siamois, is expressedwithin the Nieuwkoop Center (Lemaire, P., Garrett, N. andGurdon, J. B. (1995) Cell 81, 85-94). We have used thesetwo properties of the Nieuwkoop Center to evaluate thedorsalizing activity of the four secreted factors Wnt8, Vg1,Activin and Noggin. The requirement for β-catenin wastested by coexpressing a cadherin, which sequesters β-catenin at the cell membrane and specifically blocks itsintracellular signaling activity (Fagotto, F., Funayama, N.,Glück, U. and Gumbiner, B. M. (1996) J. Cell Biol. 132,1105-1114). Induction of Siamois expression was detectedby RT-PCR. Of the four growth factors, only Wnt was

sensitive to inhibition of β-catenin activity and only Wntcould induce Siamois expression. Therefore, Wnt is able toinduce a bonafide Nieuwkoop Center, while Vg1, Activinand Noggin probably induce dorsal structures by adifferent mechanism.

To order the steps in the Nieuwkoop Center signalingcascade, we have tested the relationship between β-cateninand GSK, a serine-threonine kinase that has been impli-cated in axis formation in a step downstream of Wnt. Wefound that GSK acts upstream of β-catenin, similar to theorder of these components in the Wingless pathway inDrosophila. We have also examined the relationshipbetween the Wnt/β-catenin pathway and Siamois. We showthat β-catenin induces expression of Siamois and that thefree signaling pool of β-catenin is required for normalexpression of endogenous Siamois. We conclude that the sequence of steps in the signaling pathway isWnt | GSK | β-catenin→Siamois.

Key words: axis formation, dorsoventral patterning, NieuwkoopCenter, Spemann Organizer, growth factor, signal transduction,Siamois, Xenopus, Wnt, β-catenin, Activin, Noggin

SUMMARY

INTRODUCTION

Determination of the dorsoventral axis in Xenopus eggs occursat fertilization by an asymmetric redistribution of an unknowndeterminant within the vegetal pole (Gerhart et al., 1989). Theresulting dorsalizing region, called the Nieuwkoop Center, theninduces the overlying mesodermal cells to become theSpemann Organizer, which orchestrates gastrulation and thepatterning of axial structures. Various secreted factors havebeen shown to have dorsalizing activity, as detected by theirabilities to induce a secondary axis when expressed ventrallyand to rescue an axis in UV-treated, ventralized embryos.Although such phenotypes are characteristic of bothNieuwkoop Center and Spemann Organizer activities, thesetwo signaling centers can be distinguished according to twocriteria: (1) the activity of the Nieuwkoop Center is maximalin the lower vegetal region, while the activity of the Spemann

Organizer is restricted to the equatorial region, and (2) the cellsof the Spemann Organizer themselves become part of theinduced dorsal mesoderm, independently of their origin, whilecells expressing Nieuwkoop Center-like activities are notrecruited into axial structures, but maintain their original fates(endoderm in the case of the vegetal pole) (Smith and Harland,1991). According to these criteria, several members of the Wntfamily, as well as Vg1, Activin and Noggin (Christian et al.,1991; Smith and Harland, 1991; Sokol et al., 1991; Ku andMelton, 1993; Thomsen and Melton, 1993; Thomsen et al.,1990; Steinbeisser et al., 1993; Smith and Harland, 1992),appear to mimic the activity of the primary dorsalizing center,i.e. the Nieuwkoop Center. Despite their common abilities toact as dorsalizers, these growth factors have quite differentproperties. For instance, Vg1 and Activin, both members of theTGFβ family, are also mesodermal inducers (Thomsen et al.,1990; Green et al., 1992; Dale et al., 1993; Thomsen and

454 F. Fagotto, K. Guger and B. M. Gumbiner

Melton, 1993; Kessler and Melton, 1995), while Noggin is apotent neuralizer (Lamb et al., 1993). Some Wnts (Wnt1,Wnt8, Wnt8b) have strong dorsalizing activity, but none caninduce mesoderm (Moon, 1993). On the other hand, Wnts areprobably also involved later during development in variousinducing events, for example patterning of the ventralmesoderm during gastrulation (Wnt8, Christian and Moon,1993) and patterning of the central nervous system (Wnt1,McMahon and Bradley, 1990; Thomas and Cappechi, 1990).

Several intracellular signaling molecules, β-catenin,Glycogen Synthase Kinase-3 (GSK-3) and Dishevelled, havealso been recently implicated in the activity of the NieuwkoopCenter (Heasman et al., 1994; Funayama et al., 1995; Gugerand Gumbiner, 1995; He et al., 1995; Pierce and Kimelman,1995; Sokol et al., 1995). Interestingly, the Drosophila homo-logues of these three molecules (Armadillo, Zeste-white-3 andDisheveled, respectively) are components of a well-character-ized signaling pathway activated by Wingless, a DrosophilaWnt homologue (Peifer, 1995). One of these intracellularproteins, β-catenin, has been shown to be required both fornormal axis formation in Xenopus and for the dorsalizingactivity of Wnt (Heasman et al., 1994). β-catenin is a cyto-plasmic protein that associates with cadherin cell-adhesionmolecules (Kemler, 1993), but also has an independent intra-cellular signaling activity (Fagotto et al., 1996).

In Drosophila, the signaling activity of the β-cateninhomologue, Armadillo, is negatively regulated by aserine/threonine kinase, Zeste-white-3/Shaggy (Peifer, 1995)and Wingless activates the signaling pathway by antagonizingZeste-white-3. Similarly, in Xenopus embryos, the homologueof Zeste-white-3, GSK-3, has ventralizing activity, and adominant negative mutant form, GSK-3K→R, induces axisduplication (He et al., 1995; Pierce and Kimelman, 1995).Overexpression of GSK can inhibit axis induction by Wnt,indicating that GSK acts downstream of Wnt, similar to Zeste-white-3 acting downstream of Wingless.

Another intracellular molecule with strong dorsalizingactivity is Siamois, a novel Hox-type transcription factorexpressed in the region of the Nieuwkoop Center (Lemaire etal., 1995). Its potent activity, its transient expression at thebeginning of the midblastula transition, and its restriction tothe dorsal-vegetal region of the late blastula, strongly suggestthat Siamois is an important component of the NieuwkoopCenter.

The Nieuwkoop Center appears not to be simply the site ofsecretion of a dorsalizing inducer, but to involve an entire intra-cellular signal transduction pathway. The established require-ment for β-catenin and the strong evidence of the involvementof GSK suggest that this pathway is similar to the DrosophilaWingless→Armadillo pathway. However, the order of thesecomponents along the pathway has not been established inXenopus and the endogenous activator of this pathway has notbeen identified. The possibility remains that growth factorsother than Wnts are able to activate the pathway. Finally, therelationship between the novel transcription factor Siamois andthese signaling pathways has not yet been investigated. Wehave therefore evaluated the dorsalizing activity of fourcandidate growth factors, Wnt8, Vg1, Activin and Noggin,according to the two newly established properties of theNieuwkoop Center: (1) requirement for β-catenin activity,which can be experimentally blocked by co-expressing

cadherins (Fagotto et al., 1996), and (2) the ability to activateSiamois expression.

MATERIALS AND METHODS

Wnt8 and Noggin expression plasmids have been kindly provided byDr R. Harland, UCBerkeley, and Activin and BMP2-Vg1 plasmids byDr D. Melton, Harvard. Siamois (Lemaire et al., 1995) was cloned byRT-PCR using RNA from stage 10G Xenopus embryos and insertedinto the pCS2+MT expression vector (Rupp et al., 1994; Turner andWeintraub, 1994). mRNAs were synthesized in vitro using the SP6RNA polymerase (Promega Corp., Madison, WI) and were dissolvedat the appropriate concentration in DPEC-treated water.

For cadherin coinjection experiments, mRNAs were injected intothe vegetal region of a ventral blastomere at the 4- to 8-cell stage(Fagotto et al., 1996). For this set of experiments, the amount ofmRNA for each of the various factors was titrated down to the smallestamount required for maximal activity, measured as maximalfrequency and completeness of secondary axis formation. The amountof mRNA injected (in pg) was: Wnt8: 10-20; Noggin: 200; Activin:7.5-15; BMP2-Vg1: 100-200; GSK-3-K→R: 1000-3000; Siamois:50-100. Higher amounts of Wnt8 cause dorsalization, and thusreduction and disappearance of a body axis. Higher amounts ofActivin resulted in uninterpretable phenotypes, probably caused bygeneralized mesodermal induction over most of the embryo. 4-5 ngof C-cadherin mRNA were used for coinjections. Embryos wereallowed to develop at room temperature in 0.1× MMR (Kay and Peng,1991) and axis duplication was scored at the neurula stage. Thenumber of embryos scored in Fig. 1 (pooled from several experiments)were (+,−) C-cadherin respectively: Wnt8: (104,95); Noggin: (95,84);Activin: (68,65); BMP2-Vg1: (76,85); GSK-3-K→R: (46,44);Siamois: (119,125). BMP2-Vg1 is a chimeric molecule that isprocessed in the embryo into the active, secreted fragment of Vg1(Thomsen and Melton, 1993). GSK-3-K→R is a mutant of GSK-3βwithout kinase activity (Pierce and Kimelman, 1995).

For histological studies, tailbud-early tadpole stages (stage 25-29)were fixed in Dent’s fixative and embedded in gelatin as described(Fagotto and Gumbiner, 1994). Serial 10 µm frozen transversesections of the entire embryos were collected. Sections were immuno-labeled (Fagotto and Gumbiner, 1994) using the rabbit anti-N-CAMpolyclonal antibody R016 (affinity purified IgG, 2 µg/ml, generousgift of Dr U. Rutishauser) and the mouse monoclonal antibody Tor70,a notochord marker (ascites fluid, 1/5000 dilution, Bolce et al., 1992).Both antibodies were detected with FITC-labeled secondary anti-bodies (each at 50 µg/ml, Molecular Probes, Inc., Eugene, OR, USA).Nuclei were then stained with DAPI (20 µg/ml, Molecular Probes,Inc., Eugene, OR, USA) and the yolk was counterstained with 0.1%Eriochrome Black (Aldrich Chemical C0., Milwaukee, WI, USA).Triple exposure microphotographs were taken under an Axioplanmicroscope (Karl Zeiss, Inc., Thornwood, NY, USA) using a 20×objective and filter blocks for DAPI, fluoresceine and rhodamine (forbackground counterstain).

For RT-PCR experiments, mRNA was extracted from wholeembryos (Fig. 7) or dissected dorsal and ventral halves at earlygastrula stage (stage 10-10G). Animal caps were dissected at stage 8and incubated for further 2 hours (stage 10) before RNA extraction.mRNA was extracted either using the proteinase K treatment protocol(Wilson and Melton, 1994) or the Ultraspec RNA isolation system(Biotecx laboratories, Inc., Houston, TX, USA). Various specificmRNAs were detected by RT-PCR as described (Wilson and Melton,1994). Primers were: for Siamois: 5′: CGC GGA TCC ATG GCC TATGAG GCT GAA ATG GAG and 3′: GCT CTA GAG AAG TCA GTTTGG GTA GGG CT (Fig. 6) or 5′: TTG GGA GAC AGA CAT GAand 3′: GCT CTA GAG AAG TCA GTT TGG GTA GGG CT (Fig.7); for Goosecoid: 5: ACA ACT GGA AGC ACT GGA and 3: TCT

455Primary dorsalizing induction in Xenopus

SiamoisSiamois + C-cadherin

0

20

40

60

80

100

vestigial axis normal

complete partialsec. axis

vestigial axis

normal

+ C-cadherin

partialxis

ActivinActivin + C-cadherin

vestigial axis

normal0

20

40

60

80

100

complete partialsec. axis

(0) (0)

NogginNoggin + C-cadherin

0

20

40

60

80

100

Dvestigial axis normal

complete partialsec. axis

(0)

3K->R3K->R + C-cadherin

vestigial axis normal

partialxis

(0)

-Vg1-Vg1 + C-cadherin

vestigial axis normal partial

xis

B

F

% t

ota

l em

bry

os

% t

ota

l em

bry

os

% t

ota

l em

bry

os

f C-cadherin co-expression on axis duplication by Wnt8, Noggin,ctivin, GSK-3-K→R and Siamois. Embryos were scored in fourplete secondary axis (with cement gland), partial secondary axis (i.e.

axis lacking the cement gland), vestigial axis (very small posteriorigmented spot or line) and normal (only one axis). Axis duplication byGSK-3-K→R (E) was strongly inhibited by C-cadherin expression.on by Noggin (B), BMP2-Vg1 (C) and Activin (D) was unaffected byexpression. Axis induction by Siamois (F) was only marginallyost anterior structures were often lacking, but the secondary axis were consistently more complete than axis induced by Noggin or BMP2-uency of Siamois-induced axis duplication (complete and partial axiss unchanged in the presence of C-cadherin (70-75%).

TAT TCC AGA GGA ACC; for Chordin: 5 ATG CAG TGT CCCCCC ATC and 3: GCA GTG CAT AAC TCC GAA; for Noggin: 5:ATG GAT CAT TCC CAG TGC and 3: TCT GTG CTT TTT GCTCTG; for Xnr3: 5: ATG GCA TTT CTG AAC CTG and 3TCT ACTGTC ACA CTG TGA; for EF1: 5: CAG ATT GGT GCT GGA TATGC AND 3: AC TGC CTT GAT GAC TCC TAG.

RESULTS

Dependence of axis induction by Wnt, Noggin, Vg1and Activin on β-catenin signalingSince β-catenin is required for normal axis formation(Heasman et al., 1994), we tested which of the putative dor-salizing inducers Wnt8, Noggin, Vg1 or Activin require activeβ-catenin signaling. We have previously shown that signalingby β-catenin can be specifically inactivated by coexpressingcadherins (Fagotto et al., 1996). Indeed, C-cadherin inhibits β-catenin-induced axis duplication by sequestering β-catenin atthe plasma membrane and thus depleting the cytosolic pool ofβ-catenin that is active in intracellular signaltransduction. We therefore used cadherincoexpression as a simple method to inhibit thisstep in the endogenous dorsalizing signalingpathway. mRNAs encoding Wnt8, Noggin,BMP2-Vg1 and Activin were injected into theventral side of early cleaving embryos, eitheralone or coinjected with C-cadherin mRNA(see Materials and Methods for ratios). BMP2-Vg1 is a chimeric molecule that is processedinto an active, secreted fragment of Vg1 whenexpressed in Xenopus embryos (Thomsen andMelton, 1993). Axis duplication was scored atthe neurula stage (Fig. 1). Also, the histologyof secondary axial structures was analyzedfrom serial frozen sections of late tailbud-earlytadpole stages (Fig. 3).

C-cadherin coexpression strongly inhibitedWnt8 axis-inducing activity. While expressionof Wnt8 alone induced a complete secondaryaxis at high frequency (Figs 1A, 2A, 3A), mostembryos expressing both Wnt8 and C-cadherinmRNA developed normally (Figs 1A, 2B, 3B).At most, a short protrusion or small pigmentedarea was observed (vestigial axis), indicative ofa weak remnant of dorsalizing activity. Thus,axis induction by Wnt is sensitive to depletionof cytoplasmic β-catenin, similar to the resultsobtained by Heasman et al. (1994) usingantisense technology.

In contrast, C-cadherin coexpression wasfound to have no detectable effect on axisduplication induced by Noggin, BMP2-Vg1 orActivin. C-cadherin coexpression did notaffect the frequency of axis duplicationinduced by these three growth factors (Fig.1B-D). Moreover, the secondary axes inducedin the presence of C-cadherin were similar tothe axes induced by the growth factors alone,by both criteria of their external morphology(Fig. 2C-F) and of their internal organization

Wnt8Wnt8

0

20

40

60

80

100

complete sec. a

(0)

GSK-GSK-

0

20

40

60

80

100

complete sec. a

(0)

BMP2BMP2

0

20

40

60

80

100

complete sec. a

(0) (0)

A

C

E

% t

ota

l em

bry

os

% t

ota

l em

bry

os

% t

ota

l em

bry

os

Fig. 1. Effect oBMP2-Vg1, Acategories: comany secondaryprotrusion or pWnt8 (A) and Axis duplicatiC-cadherin coaffected: the mstill very long,Vg1. The freqcombined) wa

(Fig. 3C-F). Similar to previous reports, Noggin, BMP2-Vg1and Activin only induced partial secondary axes, but notcomplete axes (Figs 1B-D, 2C-F). These partial axes alwayslacked anterior structures and, in most cases, a differentiatednotochord was not detected (Fig. 3C,E,F, see legend of Fig. 3).However, the partial duplicated axes were otherwise well struc-tured, with a neural tube, dorsal mesoderm (muscle) and anintestinal lumen (Fig. 3C-F). Note that, in the cases in whichincomplete axis induction was generated by Wnt expression,the secondary axes also often lacked notochord (not shown).

Another indication that the dorsalizing activities of Noggin,Vg1 and Activin were not sensitive to cadherin inhibition wasthe appearance of a second blastoporal lip. This lip, which isparticularly prominent in BMP2-Vg1 and Activin-expressingembryos, was still present in embryos co-expressing C-cadherin (not shown). In contrast, no secondary lip wasobserved in embryos co-expressing Wnt and C-cadherin (notshown). We conclude that Wnt8 is the only one out of the fourgrowth factors tested that requires β-catenin signaling for axis-inducing activity.

456 F. Fagotto, K. Guger and B. M. Gumbiner

We have further confirmed these observations by evaluatingthe effect of C-cadherin coexpression on the capacity of thevarious dorsalizing factors to induce the expression ofGoosecoid, a marker of the Spemann Organizer (Cho et al.,1991). mRNA encoding the various factors were injected in theventral side, either alone or with C-cadherin. Embryos weredissected at early gastrula stage (10G) into dorsal and ventralhalves and the expression of Goosecoid was analyzed by RT-PCR. Dorsal halves embryos were used as controls for thenormal expression of Goosecoid in the endogenous SpemannOrganizer and uninjected ventral halves as negative controls.All four growth factors were able to induce Goosecoidexpression when injected alone. Coexpression of C-cadherinstrongly inhibited the induction of Goosecoid by Wnt8, but notthe induction by Noggin, BMP2-Vg1 or Activin (Fig. 4).Therefore, free b-catenin is required for the induction of the

Fig. 2. Examples of axis duplication by various dorsalizing growthfactors and dependence on β-catenin signaling activity. (A)Complete axis duplication, with two cement glands, by Wnt8.(B) Complete inhibition of Wnt-mediated axis duplication bycoexpression of C-cadherin. (C,E) Partial secondary axes inducedby Noggin and BMP2-Vg1, respectively. (D,F) No inhibition of axisduplication obtained when C-cadherin mRNA is coinjected withNoggin or BMP2-Vg1 mRNA. (G) Complete axis duplication bySiamois. (H) Axis duplication by Siamois is largely unaffected byC-cadherin coinjections. Although the formation of the mostanterior structures is inhibited, secondary axes are still very long,which is in contrast to the strong inhibition of Wnt-induced axisduplication (compare to B).

Spemann Organizer by Wnt, but not by the other growthfactors.

β-catenin is required for the dorsalizing activity ofGSK-3K→R but not the dorsalizing activity ofSiamoisGSK has been implicated in the Wnt signaling pathway leadingto axis induction. GSK-3K→R, a dominant negative mutantform of Xenopus GSK-3, induces axis duplication whenexpressed in the ventral side of early cleaving embryos (He etal., 1995; Pierce and Kimelman, 1995). To order the steps inthis pathway, we asked whether axis duplication by GSK-3K→R requires free β-catenin. Axis duplication was indeedstrongly inhibited by C-cadherin co-expression (Fig. 1E). Thisfinding indicates that GSK acts upstream of β-catenin, similarto the relationship between Zeste-white-3 and Armadillo inDrosophila.

We also tested whether β-catenin was required for axisduplication by Siamois, another factor with NieuwkoopCenter-like dorsalizing activity. In contrast to Wnt8, β-cateninor GSK-3K→R, the axis-inducing activity of Siamois wasbarely affected by C-cadherin overexpression (Figs 1F, 2G,H,3G,H). The only effect of C-cadherin coexpression was apartial reduction in anterior structures, while secondary axeswere otherwise reproducibly long and complete, including thepresence of well-differentiated notochords (Fig. 3H). In fact,these secondary axes were still more complete than themajority of the secondary axes generated by expressing Nogginor BMP2-Vg1. The effect of C-cadherin expression on Siamoisactivity was therefore very marginal compared to the strikinginhibition of Wnt, GSK-3K→R or β-catenin-inducedsecondary axes. The small effect observed might be due to aβ-catenin-independent effect of C-cadherin overexpression, forexample, a slight perturbation of gastrulation. Alternatively, β-catenin may also be involved in a later signaling step duringformation of head structures. Nevertheless, the axis-inducingactivity of Siamois does not seem to depend on free β-catenin,suggesting that it acts downstream in the pathway, or in parallelto it.

Similar results were obtained by analyzing the expression ofGoosecoid in the ventral side of injected embryos. Inductionof Goosecoid expression by GSK-3K→R was strongly reducedin the presence of C-cadherin, while Siamois induced similarlevels of Goosecoid transcripts in the presence or the absenceof C-cadherin (Fig. 4).

Induction of Siamois expression by a subset ofdorsalizing moleculesThe C-cadherin coexpression experiments have allowed us todetermine which of the known dorsalizing factors require β-catenin signaling in the Nieuwkoop Center. A second specificcharacteristic of the Nieuwkoop Center is the expression of thetranscription factor Siamois. We therefore asked which of thedorsalizing factors can induce Siamois expression. mRNAsencoding β-catenin and the four growth factors Wnt8, Noggin,BMP2-Vg1 and Activin were injected ventrally, and theembryos were dissected at early gastrula stage (the normalpeak of Siamois expression) into dorsal and ventral halves.Siamois expression was analyzed by RT-PCR (Fig. 5A). Dorsalhalves were used as internal controls for Siamois expression inthe endogenous Nieuwkoop Center and ventral halves from

457Primary dorsalizing induction in Xenopus

uninjected embryos were used as negative controls. Both β-catenin and Wnt8 induced strong expression of Siamois in theventral halves. However, none of the other secreted factors(Noggin, BMP2-Vg1, Activin) induced Siamois, even thoughall of them activated expression of Goosecoid, a marker for theSpemann Organizer, which is the overlying induced dorsalmesoderm (Cho et al., 1991). Similar injections were made inthe animal hemisphere, and animal caps were removed by dis-section and analyzed for Siamois expression. Although theanimal hemisphere is normally not involved in the dorsalizingprocess, it is a very useful tissue for analysis of mesoderm-inducing and dorsalizing activities. Again, both Wnt8 and β-catenin activated Siamois expression, but neither Noggin norBMP2-Vg1 induced its expression (Fig. 5B). Therefore,although all factors analyzed were able to induce Goosecoid inthe Spemann Organizer, only Wnt and β-catenin could induceexpression of Siamois, a marker for the Nieuwkoop Center.

β-catenin requirement for Siamois expression in theendogenous dorsalizing centerThe experiments described above suggest that Siamois is aspecific target of the Wnt-β-catenin pathway. We thereforeasked whether the normal expression of Siamois in the endoge-nous Nieuwkoop Center is dependent upon β-cateninsignaling. Normal axis formation can be significantly inhibited

by increasing the levels of C-cadherin, either globally inoocytes (Heasman et al., 1994), or specifically in the vegetaldorsal side of early embryos (Fagotto et al., 1996), similar tothe effect resulting from the depletion of maternal β-catenin(Heasman et al., 1994). Therefore Siamois expression wasanalyzed in embryos where β-catenin signaling in the endogen-ous Nieuwkoop Center was blocked by overexpression of C-cadherin. When 5 ng C-cadherin mRNA were injected into thetwo dorsal blastomeres of 4- to 8-cell-stage embryos, Siamoisexpression was greatly decreased compared to control unin-jected embryos (Fig. 6). Siamois expression was fully rescuedin embryos in which β-catenin mRNA (1.5 ng) was coinjectedwith C-cadherin mRNA, confirming that the inhibition ofSiamois expression by C-cadherin was specifically due to theinhibition of β-catenin signaling. Thus, Siamois expressiondepends on β-catenin signaling in the Nieuwkoop Center.

Siamois induces the expression of markers of theSpemann OrganizerThe selective activation of Siamois in the Nieuwkoop Centerby the Wnt/β-catenin pathway and the strong dorsalizingactivity of Siamois suggest that this pathway may account formost of the Nieuwkoop Center activity. To determine whetherthe activity of Siamois is sufficient to induce a SpemannOrganizer in the overlying mesoderm, we analyzed theexpression of several markers of the Organizer. In Fig. 4, wehave shown that ectopic expression of Siamois in the ventralside can induce the expression of Goosecoid. We further tested

Fig. 3. Internal organization of axial structures in embryos injectedwith various dorsalizing molecules in the presence and absence of C-cadherin coexpression. Injected embryos fixed at tailbud-earlytadpole stages were cut throughout in serial transverse frozensections. Both the neural tube (nt) and notochord (no) wereimmunolabeled, respectively, with an anti-N-CAM antibody andTor70, an antibody specific for the extracellular matrix of thenotochord. Both were detected with FITC-labeled secondaryantibodies (green). Nuclei were stained with DAPI (blue) and theyolk counterstained with Eriochrome Black (red). (A) Embryosinjected with Wnt8 developed with high frequency two completeaxes, with two neural tubes (nt), two notochords (no) flanked bysomitic mesoderm (muscle, ms). Two separate intestinal lumens (in)were often observed. Similar structures were observed when partialsecondary axes resulted from Wnt8 injections, except that thenotochord was generally missing (not shown). (B) Wnt8 and C-cadherin coinjection resulted in embryos with a normal morphologyhaving a single axis. (C-F) The general organization of the secondaryaxes (stars) induced by Noggin and BMP2-Vg1 wasindistinguishable in the presence or the absence of C-cadherin. Thesepartial axes displayed well differentiated axial structures such asneural tube, muscle and dorsal fin, and a secondary digestive tractwas often observed in the anterior part, eventually merging with thetract of the primary axis posteriorly. Differentiated notochordsdetected by Tor70 staining were found only in a few cases, generallyin the most posterior region of the embryo. One case of secondarynotochord is shown in D (Noggin + C-cadherin). In this case, thenotochord extended more anteriorly than in the other cases observed.In C, ot indicates a tangential section through the periphery of an oticvesicle stained with N-CAM. (G,H) Cross-sections through embryosexpressing exogenous Siamois or Siamois + C-cadherin showedidentical histological features, with well organized twin axes, withtwo neural tubes, two notochords extending quite anteriorly, and twointestinal lumen (which in the embryo shown in H have merged moreanteriorly).

458 F. Fagotto, K. Guger and B. M. Gumbiner

Fig. 4. Effect of C-cadherin co-expression on the induction ofGoosecoid, a marker of the Spemann Organizer, by Wnt8, Noggin,BMP2-Vg1, Activin, GSK-3-K→R and Siamois. mRNAs coding forthe various dorsalizing factors were injected vegetally into oneventral blastomere of 4- to 8-cell-stage embryos, alone (−) orcoinjected with 5 ng C-cadherin mRNA (+C). Ventral halves weredissected at stage 101/2 and analyzed for Goosecoid expression byRT-PCR. Dorsal (D) and ventral (V) halves from uninjected embryoswere used respectively as positive and negative controls for thenormal expression of Goosecoid. All dorsalizing factors were foundto ectopically induce Goosecoid when expressed alone. Coexpressionof C-cadherin strongly inhibited Goosecoid expression induced byWnt8 and GSK-K→R, but not by Noggin, BMP2-Vg1, Activin orSiamois. EF1 is an ubiquitous mRNA used as loading control.

mois expression in response to various dorsalizing factors. (A) Ectopic of factors in the vegetal-ventral region. mRNAs were injected

into the two ventral blastomeres of 4- to 8-cell-stage embryos. Totalf injected mRNAs are indicated. Dorsal (D) and ventral (V) halvescted at stage 10 and analyzed for Siamois and Goosecoid expression

R. β-catenin and Wnt8 induce high Siamois expression (compared topression in the dorsal side). No expression was observed in ventralm embryos injected with Noggin, Activin or BMP2-Vg1. All theseuced Goosecoid expression in the same embryos. (B) Ectopic in the animal hemisphere. mRNAs were injected into the animal pole

r blastomeres of 4- to 8-cell-stage embryos. Total amounts of injectedre indicated. Animal caps were dissected from injected embryos atd were analyzed at stage 10 for Siamois expression. β-catenin andpic expression induced Siamois expression in the animal hemisphere,n and BMP2-Vg1 did not.

B

whether other markers, Chordin, Xnr3 and Noggin, wereexpressed under similar conditions (Fig. 7). We found thatSiamois was able to induce ectopic expression of all themarkers tested, suggesting that Siamois is sufficient, at leastby the presently available criteria, to induce an entire SpemannOrganizer.

DISCUSSION

Several kinds of growth factors are able to induce asecondary body axis in Xenopus embryos. However,they may do so by different mechanisms. We haveused two independent objective criteria to evaluatewhich factors and/or signaling pathways can accountfor the primary dorsal induction events of theNieuwkoop Center, the earliest dorsalizing region inthe Xenopus embryo arising from cortical rotation ofthe egg after fertilization. Of the four growth factorstested, only a pathway triggered by Wnts satisfiesthese criteria. The first criterion, established byHeasman et al. (1994) is the requirement for β-cateninin the normal process of dorsalization and axis devel-opment. We confirm their findings that axis inductionby Wnts requires β-catenin, and go on to show thatnone of the other growth factors tested, Noggin, Vg1or Activin, require β-catenin for their axis-inducingactivities. A second criterion is the induction of theexpression of Siamois, a Hox protein with potent axis-inducing activity that is normally expressed in theendogenous Nieuwkoop Center (Lemaire et al.,1995). Only Wnts, but not Noggin, Vg1 or Activin,can induce the expression of Siamois when expressedectopically in the early embryo. Therefore, a Wnt/ β-

Fig. 5. Siaexpressionvegetally amounts owere disseby RT-PCnormal exhalves frofactors indexpressionof the foumRNAs astage 8 anWnt8 ectobut Noggi

A

catenin pathway leading to the induction of Siamois expressionis the only known signal transduction pathway that can accountfor the characteristics of the Nieuwkoop Center.

Although Vg1, Activin and Noggin have dorsalizing activi-ties, our results suggest that these growth factors do not act byinducing the primary dorsalizing Nieuwkoop Center. Indeed, therole of these three factors as endogenous dorsalizers had beenpreviously questioned, mainly because they could rarely inducecomplete secondary axes (see e.g. Dale et al., 1993; Lemaire etal., 1995). However, due to lack of more precise criteria, theissue has remained controversial. Complete versus partial axisinduction could be interpreted as resulting from differences inthe strength of the induction of injected constructs, rather thanfrom fundamental qualitative differences between the effects ofvarious factors. In the case of BMP2-Vg1, for example, it hasbeen argued that this exogenously expressed protein is poorlysecreted. In fact, another Vg1 chimeric molecule, which is moreefficiently processed, has been reported to cause a morecomplete axis duplication (Kessler and Melton, 1995). Also, ithas been reported that BMP2-Vg1 (Thomsen and Melton, 1993)and Noggin (Smith and Harland, 1992) can rescue complete axisin UV-treated embryos, which has been interpreted to mean thatthey possess full axis-inducing activity. We have not observedany compelling qualitative difference in the nature of thesecondary axes induced by the various dorsalizing factors (Fig.3). Indeed, the degree of internal organization, according to thepresence of differentiated axial structures such as neural tube,notochord or muscle, matches quite faithfully the degree ofcomplete induction as assessed by external morphology.Therefore, independent criteria are necessary to characterize theproperties of the various candidate dorsalizing inducers.

Our findings show that the dorsalizing activities of Vg1,

459Primary dorsalizing induction in Xenopus

Fig. 6. Inhibition of the normal expression of endogenous Siamois byoverexpression of C-cadherin in the Nieuwkoop Center. 5 ng C-cadherin mRNA (C) or 5 ng C-cadherin and 1.5 ng β-cateninmRNAs (C+β) were injected vegetally into both dorsal blastomeresof 4- to 8-cell-stage embryos. Siamois expression was analyzed atearly gastrula stages by RT-PCR. Uninjected embryos were used ascontrols (ctr). C-cadherin overexpression caused a dramatic, oftencomplete inhibition of Siamois expression, which could be rescued tohigh levels when excess β-catenin was coexpressed with C-cadherin.

Vg1ActivinNoggin

NieuwkoopCenter

SpemannOrganizer

Wnt

GSK-3

β-catenin

Siamois

Goosecoid

X ?

Fig. 8. Model for the dorsalizing inducting pathway in Xenopus. AWnt, or Wnt-related factor, initiates a signal transduction pathway inthe Nieuwkoop Center, involving sequentially the inhibition of GSK-3and the consequent stimulation of β-catenin signaling, leading toSiamois expression. The Nieuwkoop Center then secretes an unknownfactor or factors ‘X’, which induces the overlying mesodermal cells tobecome the Spemann Organizer, expressing Goosecoid and otherdorsal mesoderm markers. Vg1, Activin and Noggin can induce aSpemann Organizer, but they do not activate the β-catenin pathwayand Siamois expression. It is possible, however, that one or more ofthem could act at the step ‘X’, mediating the induction of the SpemannOrganizer by the Nieuwkoop Center (dotted arrow). Alternatively, Vg1and Activin may act in parallel to the Nieuwkoop Center activity, forinstance by participating in the mesodermal induction, in which casethey would normally act synergistically with the dorsalizing signal toinduce the Spemann Organizer in the dorsal marginal zone.

Activin and Noggin do not require β-catenin and do not induceSiamois expression. Therefore, the dorsalizing activity of Vg1,Activin and Noggin probably derives from direct induction ofcomponents of the Spemann Organizer, at sites either down-stream of Siamois or parallel with the Wnt/β-catenin/Siamoispathway (Fig. 8). One possibility is that they act at the secondinducing step, mimicking endogenous factors secreted by theNieuwkoop Center to induce the overlying Spemann Organizertissue (Figs 4, 5). Another possibility is that these or relatedfactors normally act synergistically with the signal emanatingfrom the Nieuwkoop Center (Watabe et al., 1995), but maypartially bypass the requirement for the Nieuwkoop Center

Fig. 7. Ectopic activation of markers of the Spemann Organizer bySiamois. Siamois mRNA was injected vegetally in one ventralblastomere at the 4- to 8-cell stage. Embryos were dissected at stage10-10G (early gastrula) into dorsal and ventral halves and markers ofthe Spemann Organizer were analyzed by RT-PCR. Gsc, Goosecoid;Xnr3, Xenopus nodal-related-3. Ventral halves from uninjectedembryos (V) were used as negative controls, and dorsal halves (D) aspositive controls. Ventral ectopic expression of Siamois (Sia) inducedexpression of all the markers tested, often to levels similar to thelevels found in the dorsal side.

signal steps when expressed at high levels. This would be con-sistent with a model in which the Spemann Organizer normallywould be determined by the co-ordination of a mesodermalinducing signal (by Vg1, Activin or a related molecule) and adorsalizing signal secreted by the Nieuwkoop Center(Kimelman et al., 1992).

Our results suggest that the dorsalizing center in Xenopus isestablished by a Wnt→GSK→β-catenin signaling cascadeleading to the induction of the transcription factor Siamois. Wehave confirmed that Wnt signaling requires β-catenin and wehave shown that GSK acts upstream of β-catenin in thispathway. This is similar to Zeste-white-3 being upstream ofArmadillo in Drosophila, and confirms the existence of aconserved Wnt/Wingless-activated pathway. Furthermore, wehave established that Siamois is a downstream target of theWnt/β-catenin pathway by the following criteria: (1) Siamoiscan be ectopically activated by Wnt and β-catenin; (2) normalexpression of Siamois in the endogenous Nieuwkoop Centerrequires free β-catenin, because it is inhibited by C-cadherinexpression and (3) the dorsalizing activity of Siamois is largelyunaffected by inhibition of β-catenin signaling by C-cadherin,which dramatically blocks signaling by Wnt and GSK-K→R.We conclude that Wnt, GSK and β-catenin act in sequence toinduce Siamois expression and axis formation.

Although a Wnt-activatable β-catenin signaling pathwayappears to be required for the activity of the Nieuwkoop Center,the endogenous secreted factor that initiates this signalingcascade has yet to be identified. The known maternallyexpressed Wnts are uncertain candidates, because they eitherhave only weak or partial dorsalizing activity (Wnt5a, Moon et

460 F. Fagotto, K. Guger and B. M. Gumbiner

al., 1993, and Wnt11, Ku and Melton, 1993), or do not appearto be properly localized (Wnt8b, Cui et al., 1995). Furthermore,it has been observed that Siamois expression can occur, at leastpartially, even when cells are kept dispersed (Lemaire et al.,1995). However, since Wnts are known to remain associatedwith the cell surface of the secreting cell (Bradley and Brown,1990; Papkoff and Schrywer, 1990), it is quite possible thatautocrine induction could still take place in isolated cells. It isalso conceivable that dorsalization normally may be initiatedcell autonomously by activation of an intracellular componentof the signaling pathway upstream of β-catenin, such as Dishev-elled or GSK. Nevertheless, it appears that if a secreted growthfactor is actually involved in the formation of the NieuwkoopCenter, it is likely to be some member of the Wnt family.

We are grateful to Dr R. Harland and Dr D. Melton for gift ofexpression plasmids, and Dr U. Rutishauser for gift of anti-N-CAMantibody. We thank E. Wong for cloning and constructing the Siamoisexpression plasmid. This work was supported by National Institute ofHealth grant GM37432 awarded to B. M. Gumbiner and by the CancerCenter Support grant NCI-P30-CA-08784. B. M. Gumbiner is arecipient of a Career Scientist Award from the Irma T Hirschl Trust.

REFERENCES

Bolce, M. E., Hemmati-Brivanlou, A., Kushner, P. D. and Harland, R. M.(1992). Ventral ectoderm of Xenopus forms neural tissue, includinghindbrain, in response to activin. Development 115, 681-688.

Bradley, R. S. and Brown, A. M. (1990). The proto-oncogene int-1 encodes asecreted protein associated with the extracellular matrix. EMBO J. 9, 1569-1575.

Cho, K. W. Y., Blumberg, B., Steinbeisser, H. and De Robertis, E. M.(1991). Molecular nature of Spemann’s organizer: The role of the Xenopushomeobox gene goosecoid. Cell 67, 1111-1120.

Christian, J. L., McMahon, J. A.,McMahon, A. P. and Moon, R. T. (1991).Xwnt-8, a Xenopus Wnt-1/int-1-related gene responsive to mesoderm-inducing growth factors, may play a role in ventral mesodermal patterningduring embryogenesis. Development 111, 1045-1055.

Christian, J. L. and Moon, R. T. (1993). Interactions between the Wnt-8 andSpemann organizer signalling pathways generate dorsoventral pattern in theembryonic mesoderm of Xenopus. Genes Dev. 7, 13-28.

Cui, Y., Brown, J. D., Moon, R. T.and Christian, J. L. (1995). Xwnt-8b: amaternally expressed Xenopus Wnt gene with a potential role in establishingthe dorsoventral axis. Development 121, 2177-2186.

Dale, L., Matthews, G. and Colman, A. (1993). Secretion and mesoderm-inducing activity of the TGF-β related domain of Xenopus Vg1. EMBO J. 12,4471-4480.

Fagotto, F. and Gumbiner, B. M. (1994). β-catenin localization duringXenopus embryogenesis: accumulation at tissue and somite boundaries.Development 120, 3667-3679.

Fagotto, F., Funayama, N., Glück, U. and Gumbiner, B. M. (1996). Bindingto cadherins antagonizes the signaling activity of β-catenin during axisformation in Xenopus. J. Cell Biol. 132, 1105-1114.

Funayama, N., Fagotto, F., McCrea, P. and Gumbiner, B. M. (1995).Embryonic axis induction by the armadillo repeat domain of β-catenin:evidence for intracellular signaling. J. Cell Biol. 128, 959-968.

Gerhart, J., Danilchik, M., Doniach, T., Roberts, S., Rowing, B. andSteward, R. (1989). Cortical rotation of the Xenopus egg: Consequences forthe anteroposterior pattern of embryonic dorsal development. Development107 Supplement, 37-51.

Green, J. B. A., New, H. V. and Smith, J. C. (1992). Responses of embryonicXenopus cells to activin and FGF are separated by multiple dose thresholdsand correspond to distinct axes of the mesoderm. Cell 71, 731-739.

Guger, K. A. and Gumbiner, B. M. (1995). β-catenin has Wnt-like activity andmimics the Nieuwkoop signaling center in Xenopus dorsal-ventralpatterning. Dev. Biol. 172, 115-125.

He, X., Saint-Jeannet, J.-P., Woodgett, J. R., Varmus, H. E. and Dawid, I. B.(1995). Glycogen synthase kinase-3 and dorsoventral patterning in Xenopusembryos. Nature 374, 617-622.

Heasman, J., Crawford, A., Goldstone, K., Garner-Hamrick, P., Gumbiner,B., McCrea, P., Kintner, C., Noro, C. Y. and Wylie, C. (1994).Overexpression of cadherins and underexpression of β-catenin inhibit dorsalmesoderm induction in early Xenopus embryos. Cell 79, 791-803.

Kay, B. K. and Peng, H. B. (1991). Xenopus laevis: Practical Uses in Cell andMolecular Biology. In Methods in Cell Biology. Vol. 36. New York:Academic Press, Inc.

Kemler, R. (1993). From cadherins to catenins: cytoplasmic proteininteractions and regulation of cell adhesion. Trends in Genetics 9, 317-321.

Kessler, D. S. and Melton, D. A. (1995). Induction of mesoderm by soluble,mature Vg1 protein. Development 121, 2155-2164.

Kimelman, D., Christian, J. L. and Moon, R. T. (1992). Synergisticprinciples of development: overlapping patterning system in Xenopusmesoderm induction. Development 116, 1-9.

Ku, M. and Melton, D. A. (1993). Xwnt-11: a maternally expressed Xenopuswnt gene. Development 119, 1161-1173.

Lamb, T. M., Knecht, A. K., Smith, W. C., Stachel, S. E., Economides, A. N.,Stahl, N., Yancopolous, G. D. and Harland, R. M. (1993). Neuralinduction by the secreted polypeptide noggin. Science 262, 713-718.

Lemaire, P., Garrett, N. and Gurdon, J. B. (1995). Expression cloning ofSiamois, a Xenopus homeobox gene expressed in dorsal-vegetal cells ofblastulae and able to induce a complete secondary axis. Cell 81, 85-94.

McMahon, A. P. and Bradley, A. (1990). The Wnt-1 (int-1) proto-oncogene isrequired for development of a large region of the mouse brain. Cell 62, 1073-1085.

Moon, R. T. (1993). In pursuit of the functions of the Wnt family ofdevelopmental regulators: insights from Xenopus laevis. BioEssays 15, 91-97.

Moon, R. T., Campbell, R. M., Christian, J. L., McGrew, L. L., Shih, J. andFraser, S. (1993). Xwnt-5A: a maternal Wnt that affects morphogeneticmovements after overexpression in embryos of Xenopus laevis. Development119, 97-111.

Papkoff, J. and Schrywer, B. (1990). Secreted int-1 protein is associated withthe cell surface. Molec. Cell. Biol. 10, 2723-2730.

Peifer, M. (1995). Cell adhesion and signal transduction: the Armadilloconnection. Trends Cell Biol. 5, 224-229.

Pierce, S. B. and Kimelman, D. (1995). Regulation of Spemann organizerformation by the intracellular kinase Xgsk-3. Development 121, 755-765.

Rupp, R. A., Snider, L. and Weintraub, H. (1994). Xenopus embryos regulatethe nuclear localization of XMyoD. Genes Dev. 8, 1311-1323.

Smith, W. C. and Harland, R. M. (1991). Injected Xwnt-8 RNA acts early inXenopus embryos to promote formation of a vegetal dorsalizing center. Cell67, 753-765.

Smith, W. C. and Harland, R. M. (1992). Expression cloning of noggin, a newdorsalizing factor localized to the Spemann organizer in Xenopus embryos.Cell 70, 829-840.

Sokol, S., Christian, J. L., Moon, R. T. and Melton, D. A. (1991). Injected wntRNA induces a complete body axis in Xenopus embryos. Cell 67, 741-752.

Sokol, S. Y., Klingensmith, J., Perrimon, N. and Itoh, K. (1995). Dorsalizingand neuralizing properties of Xdsh, a maternally expressed Xenopushomolog of dishevelled. Development 121, 1637-1647.

Steinbeisser, H., De Roberts, E. M., Ku, M., Kessler, D. S. and Melton, D.(1993). Xenopus axis formation: induction of goosecoid by injected Xwnt-8and activin mRNAs. Development 118, 499-507.

Thomas, K. R. and Cappechi, M. R. (1990). Targeted disruption of the murineint-1 proto-oncogene resulting in severe abnormalities in midbrain andcerebellar development. Nature 346, 847-850.

Thomsen, G., Woolf, T., Whitman, M., Sokol, S., Vaughan, J., Vale, W. andMelton, D. A. (1990). Activins are expressed early in Xenopusembryogenesis and can induce axial mesoderm and anterior structures. Cell63, 485-493.

Thomsen, G. H. and Melton, D. A. (1993). Processed Vg1 protein is an axialmesoderm inducer in Xenopus. Cell 74, 433-441.

Turner, D. L. and Weintraub, H. (1994). Expression of achaete-scutehomolog 3 in Xenopus embryos converts ectodermal cells to a neuronal fate.Genes Dev. 8, 1234-1447.

Watabe, T., Kim, S., Candia, A., Rothbächer, U., Hashimoto, C., Inoue, K.and Cho, K. W. Y. (1995). Molecular mechanisms of Spemann’s organizerformation: conserved growth factor synergy between Xenopus and mouse.Genes Dev. 9, 3038-3050.

Wilson, P. A. and Melton, D. A. (1994). Mesodermal patterning by an inducergradient depends on secondary cell-cell communication. Curr. Biol. 4, 676-686.

(Accepted 28 October 1996)