Nodal-related signals induce axial mesoderm and dorsalize mesoderm during
gastrulation
C. Michael Jones2,3, Michael R. Kuehn4, Brigid L. M. Hogan2, James C. Smith3 and Christopher V. E. Wright1
1Department of Cell Biology, Vanderbilt University Medical School, Nashville, TN 37232-2175, USA2Howard Hughes Medical Institute, Vanderbilt University Medical School, and Department of Cell Biology, Vanderbilt UniversityMedical School, Nashville, TN 37232-2175, USA3Laboratory of Developmental Biology, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA, UK4Experimental Immunology Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA
Mouse embryos homozygous for a null mutation in nodalarrest development at early gastrulation and contain littleor no embryonic mesoderm. Here, two Xenopus nodal-related genes (Xnr-1 and Xnr-2) are identified and shownto be expressed transiently during embryogenesis, firstwithin the vegetal region of late blastulae and later in themarginal zone during gastrulation, with enrichment in thedorsal lip. Xnrs and mouse nodal function as dose-dependent dorsoanterior and ventral mesoderm inducersin whole embryos and explanted animal caps. Using aplasmid vector to produce Xnr proteins during gastrula-tion, we show that, in contrast to activin and other TGFβ-
like molecules, Xnr-1 and Xnr-2 can dorsalize ventralmarginal zone explants and induce muscle differentiation.Xnr signalling also rescues a complete embryonic axis inUV-ventralized embryos. The patterns of Xnr expression,the activities of the proteins and the phenotype of mousenodal mutants, all argue strongly that a signaling pathwayinvolving nodal, or nodal-related peptides, is an essentialconserved element in mesoderm differentiation associatedwith vertebrate gastrulation and axial patterning.
Formation and patterning of the mesoderm is fundamental tothe establishment of the vertebrate body plan. In Xenopus,prospective mesoderm forms in the equatorial region (marginalzone) of the blastula as a result of signals originating in thevegetal hemisphere (reviewed in Smith, 1989a,b; Slack, 1994).The inducing signals from the presumptive dorsal and ventralvegetal regions appear to be quite distinct. This is based on theobservation that dorsal marginal quadrants explanted shortlyafter induction differentiate notochord and muscle, while theremaining three-fourths of the marginal zone forms onlyventral tissues (e.g. blood and loose mesenchyme; Boteren-brood and Nieuwkoop, 1973; Dale and Slack, 1987a).However, in the embryo, large amounts of muscle arise fromlateral marginal zone tissue that was originally specified asventral mesoderm (Dale and Slack, 1987b; Moody, 1987a,b).This is because the dorsal mesoderm (Spemann’s Organizer)produces a third signal(s) converting adjacent mesoderm fromventral into more dorsal mesoderm (see Slack, 1994 and ref-erences therein). This process, called dorsalization (Dale andSlack, 1987a), is well illustrated by the observation that ventralmarginal zone explants can be dorsalized towards a muscle fatein contact with dorsal marginal zone explants.
Many years of study have led to the identification of factorswith activities expected of each of the three signals discussed
above. Activin and BVg1 exhibit dorsal mesoderm inducingactivity, while bFGF and BMP4 act as ventral mesoderminducers (for review, see Slack, 1994). However, this is com-plicated by data from dominant negative FGF receptor exper-iments suggesting that dorsal induction is superimposed upona preexistent FGF signaling pathway (Cornell and Kimelman,1994; Amaya et al., 1993). The recently isolated secretedproteins noggin and chordin are candidates for the Organizer-derived dorsalizing activity (Smith et al., 1993; Sasai et al.,1994). Many lines of evidence lead to the proposal that a fourthsignal, mediated by BMP4 (and possibly Xwnt8), activelymaintains the ventral mesodermal fate. The ventral-dominantBMP4/Xwnt8 signal may also titrate the strength of the dor-salization signal in the prospective mesoderm so that the fullrange of dorsal/lateral/ventral mesodermal tissues is produced(Jones et al., 1992a; Dale et al., 1992; Christian and Moon,1993; Sive, 1993; Maeno et al., 1994; Suzuki et al., 1994; Graffet al., 1994; Schmidt et al., 1995; Fainsod et al., 1994). Whilethe above molecules are implicated in early patterning, but ourunderstanding of their precise role is still quite vague andincomplete, partly because of a lack of definitive geneticanalysis.
The mechanisms controlling mesodermal patterning inmammalian embryos are just beginning to be dissected, withcurrent ideas being based largely on gene expression patternsand fate mapping studies. However, with an increasing
3652 C. M. Jones and others
frequency, insight has come from analyzing mutant phenotypesresulting from gene targeting and insertional mutagenesis inembryonic stem cells (see Faust and Magnuson, 1993; Bed-dington and Smith, 1993 and references therein). A geneessential for embryonic patterning, called nodal, was isolatedbased upon its disruption by a proviral insertion in the 413.dmutant mouse strain. Embryos homozygous for the disruptedallele contain little or no mesoderm, fail to form a recogniz-able primitive streak and are arrested in development duringearly gastrulation (Conlon et al., 1991, 1994; Iannaccone et al.,1992). Undisrupted nodal signalling is thus essential for theformation of mesodermal cell lineages and subsequent pat-terning events (Zhou et al., 1993; Conlon et al., 1994). nodalRNA is detected by RT-PCR in very early streak stageembryos and, at gastrulation, transcripts are localized to cellsat the anterior tip of the primitive streak surrounding the node(Zhou et al., 1993; Conlon et al., 1994). The node is an orga-nizing center similar to the dorsal lip of frog embryos (Bed-dington, 1994), and cells arising from it populate most of theaxial mesoderm (Lawson and Pedersen, 1992; for review seeHogan et al., 1995).
The nodal gene encodes a TGFβ-type secreted molecule.The biologically active ligand derived from the C terminuscontains seven conserved cysteine residues found in the DVR(decapentaplegic/Vg-related) group of the TGFβ superfamily.The nodal sequence is slightly more similar to the BMPsubgroup (generally ventral mesoderm inducers) than toactivin (a dorsal mesoderm inducer), but is divergent enoughto occupy a separate branch of the family tree (Kingsley,1994). Thus, whether nodal acts as a ventral or dorsalmesoderm inducer, a node maintenance factor, or perhaps amigration-inducing factor, cannot be predicted either from theprotein sequence, or the phenotype of nodal null mutants.
Here, we demonstrate that mouse nodal is a dose-dependentinducer of notochord and axial mesoderm in Xenopus embryos,thus determining at least one likely biological activity of thispatterning molecule. Two novel Xenopus nodal-related genes,designated Xnr-1 and Xnr-2, respectively, have mesoderm-inducing and axial patterning activities essentially indistin-guishable from mouse nodal. Xnr-1 and Xnr-2 are transientlyexpressed during embryogenesis. Their transcripts are firstdetected uniformly over the vegetal hemisphere of the blastula,but at gastrula become localized to the marginal zone, withenrichment at the dorsal lip. Both Xnrs are dose-dependentinducers of dorsal and ventral mesoderm in animal capectoderm. Moreover, both Xnr proteins dorsalize ventralmarginal zone (VMZ) tissues during gastrulation, a functionshared by noggin and chordin, but not activin. The asymmet-ric expression of Xnr-1 and Xnr-2 in the dorsal marginal zoneduring gastrulation puts the gene products in the correct placeand time to be endogenous dorsalizing signals. Consistent withthese patterning activities, Xnr signalling also rescues acomplete embryonic axis in UV-ventralized embryos.
MATERIALS AND METHODS
Construction of Xβ-nodalGenomic DNA encoding the nodal N terminus was digested with NotIjust downstream of the signal sequence (Zhou et al., 1992), andannealed to a cDNA fragment encoding the rest of nodal cut with NotI
at the 5′ end. PCR was used to generate a 1 kb fragment encoding thecomplete protein. For insertion into pSP64-Xβm, the 5′ PCR primer(CCAAACAGCCCATCATGAGTGC) introduced a BspHI site at theinitiation codon, and the 3′ primer (reverse complement of GCCTCT-GACAGAGCCCGGGGGAGTGC) introduced SmaI downstream ofthe stop codon. PCR fragment was digested with BspHI-SmaI, andexchanged for the β-globin insert of pSP64-Xβm (Wright et al.,1989), prepared by BstEII digestion, blunt-ending and NcoI digestion.This construct, pXβ-nodal, contains zero nucleotides of nodal 5′ UTR;the translation initiation context is TTGGCCATGAGT. Fidelities ofPCR reactions and subcloning were checked by sequencing the entireinsert. pXβ-nodal was EcoRI linearized and capped RNA synthesizedwith SP6 polymerase (Jones et al., 1992a).
Isolation of nodal-related cDNAsPCR using degenerate oligonucleotides corresponding to mouse nodalsequence gave no nodal-related sequences from Xenopus genomicDNA. A 1 kb EcoRI-SphI fragment of cDNA ENH1 (Zhou et al.,1992), containing nodal protein-coding sequences, was also used toscreen gastrula cDNA and genomic libraries at low to moderate strin-gency. Approximately 15 million plaques were screened; all yieldedeither no positive clones or high nonspecific hybridization. Subse-quently, a radiolabeled PCR fragment (nucleotides 1069 to 1411) ofthe nodal mature region was used. Because nodal transcripts localizeto the node in mouse embryos (Zhou et al., 1992), a dorsal lip-specificλZAPII cDNA library (Blumberg et al., 1991) was screened atreduced stringency (Sambrook et al., 1989). Washes were 0.5× SSC/0.1% SDS/37°C. 250,000 plaques of the unamplified (UDL) library and1.5 million plaques of amplified (ADL) library were screened. Eightclones were isolated, falling into 2 groups. UDL1 encoded Xnr-1,while UDL2,3,5,6 and ADL1,6,8 encoded Xnr-2. ADL libraryrescreens with a 5′ UDL1 probe produced 16 additional Xnr-1 cDNAs.Complete sequence on both strands of the longest Xnr-1 and Xnr-2cDNAs was obtained with a Sequenase II kit (USB), specific primersand overlapping sequences generated from HaeIII-HinfI and RsaI-HinfI subclones.
Whole-mount in situ hybridization The protocol of Harland (1991) was used with minor modifications(R. Harland, personal communication) and the BM-purple substrate.Unhydrolyzed RNA probes corresponded to the entire pBluescriptcDNA inserts of Xnr-1 and Xnr-2.
Xnr injection constructscDNAs of Xnr-1 and Xnr-2 with the shortest 5′ UTRs (approx. 40 nt)were isolated with SmaI-XhoI (Xnr-1; cDNA clone ADL-15) or EcoRI(Xnr-2; clone UDL-3), blunt-ended and inserted into blunt-endedBglII-digested pSP64T (Krieg and Melton, 1984). Both resulting con-structs, pSP64T/Xnr-1 and pSP64T/Xnr-2, were linearized with SmaI,and capped mRNA generated using SP6 polymerase. pCSKA plasmidconstructs were generated by isolating coding sequences for mousenodal, Xnr-1, Xnr-2, or activin from the pXβ or pSP64T plasmidsdescribed above. Mouse nodal was isolated from pXβ-nodal withHindIII-SmaI. Xnr-1 and Xnr-2 were removed from their pSP64T con-structs by HindIII-SmaI, and activin from pSP64T-activin (gift fromD. Melton) using HindIII-EcoRI. Inserts were blunt-ended andinserted into EcoRV-cut pCSKA vector (gift of R. Harland). InjectedDNA was CsCl-purified supercoiled DNA.
RNAse protection probes and RNA isolationXnr-1: a 5′ 450 bp EcoRI fragment was subcloned into pBluescript.The construct was linearized with HindIII, and antisense RNA probegenerated with T7 polymerase. Xnr-2: two probes (HE4 and HH5) were used interchangeably. HE4is a 350 bp HinfI-EcoRI fragment from the 3′ end of the cDNA,subcloned into SmaI cut pBluescript, linearized with BamHI, tran-scribed with T3 polymerase. HH5 is a 200 bp HinfI fragment
3653Nodal-related signaling in mesoderm patterning
Fig. 1. Expression of dorsoanterior mesoderm markers in nodal-injected embryos. (A) goosecoid transcript levels are elevated duringearly gastrula stages and remain higher than controls at least intosibling neurula stages. ODC signal assesses RNA loading in thesamples. (B) Muscle-specific (ms) actin expression is upregulated innodal-injected embryos. Higher levels of ms-actin mRNA aredetected precociously in embryos expressing nodal. Cytoskeletalactin serves as a control for RNA integrity in the lanes. C, control; N,mouse nodal RNA-injected.
subcloned into SmaI cut pBluescript, linearized with BamHI, tran-scribed with T3 polymerase.
Probes for goosecoid, Xbra, Xhox-3, globin and actin were as pre-viously described (Jones et al., 1992a). RNAse digestions were withRNAses A and T1 for Xnr-1, and RNAse T1 alone for goosecoid,Xbra, actin and both Xnr-2 probes. Test RNA was isolated fromembryonic tissues by SDS/proteinase K digestion and selective pre-cipitation with lithium chloride (Jones et al., 1992a).
Embryo manipulations Xenopus embryos were obtained by in vitro fertilization, reared innormal amphibian medium (NAM; Slack, 1984) and staged accordingto Nieuwkoop and Faber (1967). For animal cap assays, RNAs wereinjected into 1-cell embryos in 5% Ficoll/75% NAM. Animal capswere isolated from stage 8 blastulae, cultured in 75% NAM untilappropriate stages, when tissues were either frozen (RNA isolation)or fixed (histological analysis). In VMZ dorsalization experiments,plasmids were injected at the 4-cell stage into both ventral blas-tomeres (50-70 pg total), close to the cleavage plane between them.Ventral was distinguished by the darker pigmentation (Nieuwkoopand Faber, 1967); correct assignment was greater than 97%. Embryoswere incubated to stage 10-10.25, when the blastopore lip clearlymarks the dorsal side. VMZs were then isolated, consisting of equa-torial tissues spanning a 60° arc centered on the midline. Explantswere cultured in 75% NAM until appropriate stages for molecular orhistological analyses.
Histological analysisTissues were fixed overnight at 4°C in 3.7% formaldehyde, 50%ethanol, 2% acetic acid, 40% NAM. After further overnight fixationin 3.7% formaldehyde/phosphate-buffered saline, tissues were dehy-drated, wax embedded, sectioned (7 µm) and stained withFeulgen/light green/orange G (Green et al., 1990).
RESULTS
Nodal induces axial mesoderm in Xenopus embryosMouse nodal was ectopically expressed in Xenopus embryos byinjection of in vitro synthesized RNA (Materials and Methods).Whole embryos expressing nodal develop normally until earlygastrula, but then exhibit grossly altered morphogeneticmovements, and later fail to develop a neural tube. Injectedembryos adopt a comma-shaped appearance, with no recogniz-able anteroposterior or dorsoventral axis, and a large ‘proboscis’similar to LiCl-hyperdorsalized embryos. Histological analysisreveals formation of extensive notochord immediately under avery thin epidermis, adjacent to very large somitic arrays, andno visible neural tissue (data not shown). The formation ofexcess notochord and muscle is apparently at the expense ofventral mesodermal tissues, suggesting that the whole embryo ishyperdorsalized. Fig. 1 demonstrates that dorsal mesodermmarkers, goosecoid and muscle Actin, are upregulated in nodal-injected embryos. We conclude that nodal-RNA injectionstrongly dorsoanteriorizes Xenopus embryos.
The mesoderm inducing activity of mouse nodal wasassayed in animal caps explanted from blastulae injected at the1-cell stage with nodal mRNA (Fig. 2). Such explants elongateextensively compared to controls (Fig. 2B), a behaviormimicking the convergent extension of axial mesodermal cellsin normal development (Keller and Tibbetts, 1989). Histolog-ical examination of nodal-injected explants confirms inductionof extensive axial mesoderm, characterized by massiveamounts of notochord and muscle (Fig. 2D).
Animal caps from injected embryos were then assayed forexpression of mesodermal markers: Xbra, a general mesodermmarker at early gastrula stages, and goosecoid and muscleactin, markers of dorsoanterior and paraxial mesoderm, respec-tively (Fig. 2E). nodal-injected animal caps express Xbra,showing that mesoderm is induced. Furthermore, goosecoidand muscle-specific actin are induced at high doses (1 ng per1-cell embryo), while lower doses (10-100 pg/embryo) induceactin, but not goosecoid. We conclude that nodal induces avariety of dorsal mesodermal cell types in a dose-dependentmanner. Induced tissues range from dorsoanterior mesoderm,marked by the presence of notochord and goosecoidexpression, to more lateral types of mesoderm, marked bymuscle differentiation and actin expression.
Two Xenopus nodal-related genesThe results of mouse nodal RNA injections suggested thatXenopus embryos contain an endogenous signalling pathwayrecognizing nodal signals. We therefore undertook a search forfrog nodal-related genes. cDNAs encoding two separate nodal-related proteins were isolated from a dorsal blastopore lip-specific cDNA library (see Materials and Methods). ExtensiveRT-PCR, cDNA and genomic library searches strongly suggestthat these genes are the Xenopus homologs of nodal (see Dis-cussion).
The two Xenopus nodal-related genes have been called Xnr-1 and Xnr-2 (for Xenopus nodal-related-1 and -2). Within thetwo groups of cDNAs, closely related A and B copies werefound for both Xnr-1 and Xnr-2, representing the pseudo-tetraploid nature of the Xenopus genome. Fig. 3 shows the
3654 C. M. Jones and others
Fig. 2. Characterization ofanimal caps induced byinjection of mouse nodalRNA. Morphological (A,B),and histological (C,D)analysis of control andnodal-loaded animal caps.(A,C) Control caps formatypical epidermis, but capsexpressing mouse nodal(B,D) elongate anddifferentiate dorsalmesodermal tissues, oftenconsisting almost entirely ofnotochord and small patchesof muscle. Abbreviations:epi, atypical epidermis; ms,muscle; no, notochord. (E) Analysis of geneexpression induced inanimal explants by differentnodal RNA doses. Highdoses (1 ng RNA/embryo)induce goosecoid (adorsoanterior mesoderm
marker), muscle-specific actin and Xbra. Decreased doses (100, 10, or 1 pg/embryo) do not induce goosecoid, but intermediate concentrationsinduce muscle-specific actin and Xbra. All samples are from the same injection experiment. goosecoid and Xbra were assayed at sibling stage10. 5, and muscle actin at stage 20. ODC and cytoskeletal actin serve as controls for RNA integrity and loading in samples.
Fig. 3. Alignment of amino acid sequences of mouse nodal with Xnr-1, Xnr-2 anda chicken nodal-related peptide. (A) Deduced amino acid sequences of Xnr-1 andXnr-2. A hydrophobic region (underlined) at the N terminus of each Xnrresembles a secretory signal sequence, with cleavage predicted according to thealgorithm of von Heijne (1986). Four potential N-linked glycosylation sites(consensus N-X-T/S) are present in each protein (centered on residues 72, 137,174, and 345 for Xnr-1, and residues 72, 160, 174, and 344 for Xnr-2). Three arepositionally conserved between Xnr-1/Xnr-2. Putative basic proteolyticprocessing sites (RRxRR, underlined) begin at residues 277 (Xnr-1) and 278(Xnr-2). Asterisks indicate identities, double dots represent conservative changes.(B) Alignment of C-terminal mature regions of Xnr-1, Xnr-2, mouse nodal and anewly isolated chick nodal-related sequence. Alignments begin at the putativebasic processing site of each molecule. The region of cysteine spacing unique tothe Xnr factors (C-X-X-C) is underlined. Vertical lines represent identities in allfour proteins. A consensus sequence is presented below the alignment. Dashesrepresent spaces introduced to optimize alignments.
3655Nodal-related signaling in mesoderm patterning
Fig. 4. Temporal and spatial expression of Xnr-1 and Xnr-2 duringXenopus development. (A) RNAse protection analysis of Xnr-1 andXnr-2 expression during Xenopus development. Transcripts are detectedduring late blastula (stage 9) and gastrula (stage 10 and 10. 5) stages. Avery low level of Xnr-1 RNA is detected during neurula stages (stage17), but no expression is detected for Xnr-2 after gastrulation (stage13). Stage 8 represents an RNA pool before zygotic transcriptionbegins, and transcripts for Xnr-1 and Xnr-2 are not detected. ODC is aloading control, and the tRNA lane demonstrates specificity of signal toembryo RNA. (B-H) Whole-mount in situ hybridization analysis of Xnr-1 and Xnr-2 expression. All embryos are cleared albino embryos,viewed from the vegetal surface with dorsal oriented upward. The dorsal lip is indicated by the black arrowhead. (B) Stage 9 embryos showpunctate perinuclear Xnr-1 signal over the entire vegetal region. Xnr-2 shows the same pattern (data not shown). (C) Xnr-1 signal at stage 10.25is restricted to the dorsal marginal zone (dark arc at bottom left is a background artefact). (D) Xnr-2 signal in stage 10 pregastrula is primarilylocated in the dorsal marginal zone, but also in adjacent dorsovegetal cells. (E) Xnr-2 signal in the stage 10.5 gastrula is highly concentratedjust above the dorsal lip, with a gradual decrease laterally and ventrally. (F) Whole-mount stained stage 10.25 embryo, split open along thedorsal/ventral plane and viewed internally to show Xnr-2 expression at the dorsal lip. Superficial and slightly deeper staining is observed. Someout-of-focus vegetal cells below the lip express Xnr-2 (white arrowhead). (G) noggin mRNA hybridization in a stage 10.5 embryo showsdeeper mesodermal expression extending anteriorly along the dorsal midline. (H) Xnr-2 sense strand control, stage 10.25 embryo. (I) RNAseprotection analysis of Xnr-1 and Xnr-2 distribution in dissected gastrulae. Lanes 1-3: at stage 10.25, Xnr-1 and Xnr-2 transcripts are detected inthe marginal zone, at greatly reduced levels in vegetal tissue, but are undetectable in animal tissue. Lanes 4 and 5: in stage 10 embryos, Xnr-1RNA and, to a lesser extent, Xnr-2 RNA, is enriched in dorsal halves of embryos compared to ventral halves. EF1-α assesses RNA integrityand loading.
protein sequences of Xnr-1 and Xnr-2, an alignment of theirpredicted mature ligand regions with that of mouse nodal anda novel chick nodal-related sequence. Both Xnrs have a stretchof hydrophobic amino acids at the N terminus characteristic ofa signal leader sequence found in secreted proteins. Xnr-1 andXnr-2 cDNAs encode proteins of 406 and 405 amino acids,respectively, with predicted unprocessed relative molecularmasses of approx. 46×103. Both Xnrs contain four potential N-
linked glycosylation sites, and a characteristic cleavagesequence (RRxRR) which, based on studies with relatedproteins, is a site of specific proteolysis releasing the biologi-cally active carboxyl-terminal ligand. Within the predictedligand sequence, Xnr-1 and Xnr-2 are 87% similar to eachother, and 78% and 73% similar to mouse nodal.
A key feature of proteins in the BMP/activin subgroup ofthe TGFβ superfamily is the conservation of seven cysteine
3656 C. M. Jones and others
Fig. 5. RNAse protection analysis of Xnr-1 and Xnr-2 expression inanimal caps treated with growth factors and LiCl- and UV-treatedembryos. (A) Xnr-1 and Xnr-2 expression is induced in animal capsby activin, but not FGF, indicating activation by dorsal mesoderm-inducing signals. Control protections show Xnr-1 and Xnr-2 RNA instage 10.5 sibling embryos. EF1-α assesses RNA loading in thesamples. (B) Expression of Xnr-1 and Xnr-2 at stage 10 in UV-ventralized or LiCl-dorsalized embryos. UV-ventralization greatlyreduces Xnr-1 and Xnr-2 transcript levels. In contrast, LiCldorsalization results in a 2-4 fold increase (evaluateddensitometrically) in Xnr-1 and Xnr-2 RNA compared to untreatedsiblings. EF1-α is used as a loading control. DAI, dorso-anteriorindex score of sibling embryos (Kao and Elinson, 1988).
residues in the mature region (Kingsley, 1994) that appearcritical for dimerization, secretion, receptor-binding and bio-logical activity (Mason, 1994). The Xnr-1 and Xnr-2 ligandsalso contain seven cysteines in similar positions, but thespacing in one region is altered compared to nodal and otherTGFβ-like factors. In nodal, two adjacent cysteine residues arefound approximately 35 amino acids from the C terminus,while in the Xenopus proteins, two cysteine residues arepresent, but separated by two amino acids (Fig. 3B). We notethat the chick nodal-related sequence reported here is also char-acterized by this ‘split-cysteine’ arrangement.
Expression of Xnr-1 and Xnr-2 during XenopusembryogenesisThe temporal expression of Xnr-1 and Xnr-2 was analyzed byRNAse protection. Both Xnrs are detected first at late blastula(stage 9), and expression peaks at early gastrula (Fig. 4A).After a decline to undetectability, a low signal for Xnr-1 isdetected in late neurulae (stage 17; Fig. 4A), but Xnr-2 tran-scripts are not detected by this method after gastrula stages(e.g. stage 13). Undetectable RNA levels for both genes atstage 8 suggest the absence of long-lived maternally storedtranscripts. Northern blot analysis shows that both genesproduce single 1.6 kb transcripts (data not shown), consistentwith the size of the longest Xnr-1 and Xnr-2 cDNAs.
Whole-mount in situ hybridization was used to analyze thespatial expression pattern of Xnr-1 and Xnr-2 RNA duringdevelopment (Fig. 4B-H). For unknown reasons, detection ofXnr-1 and Xnr-2 RNAs by this method requires greatlyextended color development (24-36 hours). Even then, weaksignals are generated, with the weaker, Xnr-1, being at thelimits of detectability. Xnr-1 and Xnr-2 signals are firstdetected as punctate perinuclear staining in vegetal cells overthe bottom third of the late blastula (stage 9; see Fig. 4B forXnr-1), but this disappears as development proceeds. In thestage 10.5 gastrula, Xnr-1 signal becomes localized to a 60°arc in the dorsal quadrant of the marginal zone (Fig. 4C). Moredetail is discernable regarding Xnr-2 expression. Just beforegastrulation (stage 10), signal is observed above and slightlybelow the position of the future dorsal lip, in an approx. 120-180° arc spanning the dorsal midline (Fig. 4D). Not all cellsdisplay Xnr-2 signal, but positive cells are more abundant atthe dorsal midpoint. In the stage 10.5 gastrula (Fig. 4E), Xnr-2 expression almost encircles the embryo, being most intenseand uniform at the pucker of the dorsal lip, in a domain encom-passing the Organizer region. Moving around the lateral andventral marginal regions, labeled cells become progressivelyless abundant; only very few positive cells are detected at theventralmost extreme. Xnr-2 is also expressed at lower levels ina thin band of pre-endodermal cells below the dorsal lip (Fig.4F). The arc of Xnr-2 expression at stage 10.5 is wider thanthe expression domain of ‘Organizer-specific’ markers such asnoggin (compare Fig. 4D,E to Fig. 4G). Because the in situhybridization signals for Xnr-1 and Xnr-2 RNA were veryweak, RNAse protection analysis of RNA from dissected wild-type embryos was performed to check the validity of thesedata. Fig. 4I shows that Xnr-1 and Xnr-2 RNAs are primarilyfound in marginal tissue of stage 10.25 gastrulae, are low orundetectable in animal and vegetal tissues, and that both areenriched in dorsal halves of the gastrula compared to ventralhalves.
Xnrs are induced in response to dorsal mesoderminducing signalsBecause Xnr-1 and Xnr-2 expression is detected in the gastrulamarginal zone, we tested for their induction by mesoderm-inducing factors. Animal caps were explanted from embryos atstage 8, cultured in the presence of activin or bFGF, and XnrRNA assayed at sibling stage 10, corresponding to the peak ofXnr-1 and Xnr-2 expression in normal embryogenesis. Exami-nation of sibling caps at tailbud stages confirmed that eachgrowth factor gave the appropriate mesoderm induction. Fig. 5Ashows that both Xnrs are induced by activin, a dorsoanteriormesoderm inducer, but not by bFGF, a ventrolateral mesoderminducer. Consistent with this, transcript levels of both Xnrsincrease in embryos hyperdorsalized by LiCl treatment duringcleavage stages, and are greatly downregulated in radially ven-tralized UV-treated embryos (Fig. 5B). We conclude thatexpression of Xnr-1 and Xnr-2 is enhanced by treatments leadingto differentiation of dorsoanterior mesoderm.
Xnr-1 and Xnr-2 induce axial mesodermInjection of mRNA encoding each Xnr into Xenopus embryosproduced massive hyperdorsalization, similar to the effects ofmouse nodal, as characterized by external morphology andgreatly increased expression of goosecoid and muscle-specificactin (data not shown). To test the mesoderm induction prop-erties of each Xnr, RNA was injected into the 1-cell embryoand animal caps explanted (Fig. 6). Xnr-2 RNA inducesextensive morphogenetic movement, similar to that caused bymouse nodal mRNA (compare Figs 6C and 2A). Animal caps
3657Nodal-related signaling in mesoderm patterning
receiving similar amounts of Xnr-1 RNA, while undergoingsignificant morphological changes compared to controls (Fig.6B), extend much less than Xnr-2 caps.
Histological analysis confirms the differential activities ofXnr-1 and Xnr-2 RNA. Xnr-2-loaded caps differentiate dorsalmesodermal tissues, including notochord and muscle (Fig. 6F).Xnr-1-loaded caps form notochord less often (data not shown),but consistently differentiate large blocks of striated musclesurrounded by loose mesenchyme (Fig. 6E). In this experi-ment, we note that it cannot be determined if equal amounts ofactive Xnr-1 and Xnr-2 ligand are produced per unit of RNA.Purified Xnr-1 and Xnr-2 mature protein will help distinguishif the apparent differences in mesoderm induction strengthbetween Xnr-1 and Xnr-2 RNAs truly reflect intrinsicallydifferent signaling activities.
Xnr signaling induces dorsal and ventral mesodermalmarkers in animal caps in a dose-dependent manner. Fig. 6Gshows that high doses of Xnr-2 RNA (100 pg/embryo) induceexpression of goosecoid, cardiac actin and the pan-mesoder-mal marker Xbra. Lower doses (1-10 pg/embryo) induce thedorsolateral mesodermal marker, muscle-specific actin, but nolonger induce goosecoid-expressing dorsoanterior mesoderm.Xnr-1 RNA induces a similar gene expression profile in animalcaps (data not shown). Fig. 6H shows that, at intermediate Xnr-2 doses (10 pg/embryo), the ventral markers Xhox-3 and globinare induced in place of the goosecoid induction at higher doses.
Xnr signals dorsalize ventral marginal zone duringgastrula stagesXnr-1 and Xnr-2 expression peaks during late blastula/earlygastrula stages and is maximal in the dorsal marginal zone. Wenext tested whether Xnr expression at gastrula stages couldmodify the type of mesoderm induced earlier in the ventralmarginal zone (Dale and Slack, 1987a; see Introduction). Tooverexpress Xnr-1/Xnr-2 during gastrula stages, we usedplasmids containing Xnr-1 or Xnr-2 sequences driven by acytoskeletal actin promoter (pCSKA:Xnr-1 and pCSKA:Xnr-2). pCSKA constructs begin to express high levels of RNAduring early gastrula stages (Christian and Moon, 1992; Smithet al., 1993). pCSKA:Xnr-1 and pCSKA:Xnr-2 plasmids wereinjected into a ventral marginal location at the 4-cell stage. Asimilar construct designed to express activin, pCSKA:activin,was injected as a control, because activin has been shown pre-viously not to cause dorsalization when applied to VMZexplants (Smith et al., 1993; Lettice and Slack, 1993). Fig. 7Ashows that explanted VMZs from control uninjected embryosdo not express muscle actin. However, both pCSKA:Xnr-1 andpCSKA:Xnr-2 induce muscle actin expression in VMZs (Fig.7A), while pCSKA:activin does not. In the same VMZexplants, Xnr-1 or Xnr-2 expression greatly decreases αT4globin transcript levels (a ventral mesodermal marker)compared to control VMZs (data not shown). We conclude thatVMZs exposed to high levels of Xnr-1 or Xnr-2 protein duringgastrula stages are dorsalized from ventral to dorsal fates.
Histological examination of explanted VMZs confirms themolecular analysis. Normal DMZ explants differentiate bothnotochord and muscle (Fig. 7C). Control VMZs differentiateloose mesenchyme and mesothelium (Fig. 7B), but VMZsexpressing Xnr-1 and Xnr-2 form large blocks of striatedmuscle (Fig. 7D,E). Notochord has not been observed in VMZsdorsalized by pCSKA:Xnr-1 or Xnr-2 (n = 20). pCSKA:activin
does not induce muscle differentiation in injected VMZs (datanot shown). We conclude that Xnrs expressed during gastrula-tion can function similarly to noggin, but differently fromactivin, in dorsalizing VMZs towards dorsal (muscle) fates.
Xnr signals rescue dorsal axes in UV-ventralizedembryosWe next performed a functional test for the ability of Xnr-1 andXnr-2 to regulate early embryonic patterning by measuring thedegree of axial rescue achieved after injecting mRNA intoradially UV-ventralized embryos (e.g. Smith and Harland,1992). Xnr-1 or Xnr-2 mRNA was injected into one blastomereof 4-cell stage UV-ventralized embryos. With 50 pg of Xnr-1RNA/blastomere, a complete axis including eyes and forebraindevelops in approximately 70% of surviving embryos (Fig. 8).In a representative experiment, 13 of 21 injected embryos(61%) were rescued to a DAI (Kao and Elinson, 1988) of 3-5,with 8 of these 13 having a DAI of 4-5, representing completeaxial rescue (a DAI ranking of 3 was recorded if definitivemelanized eye tissue was seen, while a DAI of 5 is a normaltadpole). UV-treated embryos injected with a similar amount ofXnr-2 mRNA do not form a normal body plan, but develop largeamounts of notochord (data not shown) – a hyperdorsalizedphenotype similar to that caused by injecting Xnr-2 RNA intonon-UV-treated embryos. When less Xnr-2 RNA (1-5 pg perblastomere) is injected into UV-treated embryos, recognizablepartial anteroposterior and dorsoventral axes form in a smallpercentage of embryos but, most consistently, these groups ofembryos still develop a hyperdorsalized phenotype (data notshown). This differential result could be considered consistentwith the apparently different mesoderm induction activities ofXnr-1 and Xnr-2 RNAs in animal cap assays described above.Nevertheless, we conclude that Xnr signalling is sufficient torescue complete embryonic development (Xnr-1), or notochordinduction (Xnr-2), in radially ventralized embryos.
DISCUSSION
While the phenotype of embryos homozygous for the 413.dinsertional mutation established that nodal is essential formesoderm formation and embryonic patterning in mammals, itprovided no direct clue as to the role that it played in theseprocesses. We have shown here that nodal and Xnr-1/Xnr-2 arepotent dose-dependent mesoderm inducers in Xenopusembryonic cells and, furthermore, that they can act during gas-trulation as dorsalizers of ventral mesoderm. Nodal-relatedsignaling can also rescue a complete embryonic axis in UV-ventralized embryos. While these experiments were beingcompleted, we reported that injection of nodal mRNA intozebrafish embryos resulted in embryonic axis duplicationspreceded by ectopic activation of Organizer-specific markerssuch as goosecoid and lim-1 (Toyama et al., 1995). These func-tional studies thus suggest that nodal and nodal-relatedsignaling pathways play an extensive role in mesodermalinduction and patterning, and provide critical information inrelation to models for the biological activity of nodal in ver-tebrate gastrulation.
Xnr-1 and Xnr-2 are new members of the TGFβsuperfamilyThe mature C-terminal ligands of TGFβ-like secreted signaling
3658 C. M. Jones and others
Fig. 6. Effects of Xnr-1 and Xnr-2 on animal capexplants. Morphological (A-C) and correspondinghistological analysis (D-F) of caps explanted fromXnr-1 or Xnr-2 RNA-injected embryos. (A,D) Controlexplants remain rounded, differentiating into atypicalepidermis. (B,E) Most Xnr-1-injected caps extendslightly compared to controls, and primarilydifferentiate blocks of striated muscle (ms).Notochord differentiation is observed in Xnr-1injected caps, but at lower frequency than Xnr-2.(C,F) Xnr-2 expressing explants elongate extensively,and consistently form dorsal mesodermal tissueincluding notochord (no) and striated muscle. (G) Induction of dorsal mesodermal markers in animalcaps by Xnr-2 mRNA. Explants from embryosinjected with high Xnr-2 RNA doses (100 pg)differentiate dorsoanterior mesoderm as marked by
goosecoid and muscle actin expression. Lower concentrations do not induce goosecoid, but still induceactin and the pan-mesodermal marker, Xbra. As little as 1 pg of Xnr-2 RNA injected into 1-cell embryosinduces actin and Xbra expression in animal caps (barely visible in this exposure). goosecoid and Xbrawere assayed at sibling stage 10.5, and actin at sibling stage 20. Sibling embryo RNAs provide positivecontrols, and EF1-α and cytoskeletal actin assess RNA integrity and loading. (H) While high Xnr-2RNA doses induce dorsal mesoderm markers (compare to G), intermediate doses of Xnr-2 RNA inducethe ventrolateral mesodermal markers Xhox-3 and globin. The highest levels are seen at 10 pg/embryo.Low levels of Xhox-3 are indicated by the arrowheads.
molecules are released by proteolytic cleavage from larger pre-cursors. Most members contain seven conserved cysteineswithin the mature region and are presumed to function asdimers (Kingsley, 1994). Crystallographic studies on TGFβ2suggest that the dimeric ligand forms a ‘cystine knot’ (Daopinet al., 1992; Schlunegger and Grutter, 1992; McDonald and
Hendrickson, 1993), in which the conserved cysteine residuesfunction to stabilize the structure. Mouse nodal has thecanonical cysteine spacing, but the Xenopus and chick nodal-related sequences described here, together with Xnr-3 (seebelow), are unique in that a pair of cysteines that are usuallyadjacent exhibit a split-cysteine spacing (CysXXCys; Fig. 3B).
3659Nodal-related signaling in mesoderm patterning
n expression in ventral marginal zone explants dorsalized by Xnr-1ically. (A) Ventral marginal zones (VMZs) isolated from stage 10d at the 4-cell stage with pCSKA:Xnr-1 or pCSKA:Xnr-2 expressctin), indicating dorsalization compared to control VMZ explants,
ctin. In four separate experiments (two separate examples are shown),ced detectable levels of ms-actin. Dorsal marginal zone (DMZ)
explants express high levels of ms-actin. (B-E) Histological analysisof explanted marginal zones. (B) Control VMZs differentiateventral-type tissues, includingmesothelium (mt) and loosemesenchyme. ym, yolk mass. (C) Control DMZs form notochord(no), striated muscle (ms), andneural tissue (nt). (D,E) VMZexplants preinjected withpCSKA:Xnr-1 or pCSKA:Xnr-2,respectively, are dorsalized anddifferentiate large blocks of striatedmuscle.
The displaced cysteine corresponds to the residue normallyforming an intermolecular disulfide bond in the dimer (Daopinet al., 1992; Schlunegger and Grutter, 1992), and mutation ofthis cysteine in activin prevents dimerization, abrogating bio-logical activity (Mason, 1994). Two other TGFβ-likemolecules, Vgr-2/GDF-3 and GDF-9, encode peptides inwhich this cysteine is not displaced, but replaced by valine(Jones et al., 1992b; McPheron and Lee, 1993). It is possiblethat the displaced cysteine in Xnr peptides is still appropriatelypositioned to make the intermonomer disulfide bond, but wehypothesize that the associated structural changes could sub-stantially affect the specificity of Xnr ligand/receptor interac-tions.
Xnr-1 and Xnr-2 as Xenopus cognates of mousenodalSeveral pieces of information imply that Xnr-1/Xnr-2 representthe Xenopus genes most closely related to mouse nodal. First,Southern blot analysis of Xenopusgenomic DNA with mature regionnodal probes detects no positivesignals at medium to high stringencyand only Xnr-1 or Xnr-2 signals atreduced stringency (C. V. E. W.,unpublished observations). Second,PCR and low stringency screeningwith an Xnr-2 mature region probe onpre- and postgastrulation mouseembryo libraries results only in reiso-lating nodal cDNAs, suggesting thatmouse Xnr-like sequences do not exist(C. M. J., unpublished observations).Third, chick nodal-related sequences(Fig. 3) also encode peptides with theXnr-like ‘split-cysteine’ pattern. Itappears that nodal-related peptidesexhibit more divergence than usualduring vertebrate evolution, becausemolecules like activin and BMP4 arealmost identical even in evolutionar-ily distant vertebrate species. Thus,nodal-related peptides in lower verte-brates may be more likely to resemblefrog Xnr than mammalian nodal.Fourth, functional data presented inthis manuscript demonstrate thatmouse nodal and Xnrs are essentiallyinterchangeable, each inducingsimilar patterns of gene expressionand mesoderm differentiation.
Another recently isolated Xenopusgene, Xnr-3, also encodes a nodal-related molecule with a split-cysteinemotif (Smith et al., 1995). Xnr-1, Xnr-2 and Xnr-3 therefore represent frogprototypes of a new group of theTGFβ superfamily. However, Xnr-3differs substantially from Xnr-1/Xnr-2 in other parts of the mature ligandregion, including an extended Cterminus. While Xnr-3 exhibits partial
Fig. 7. Muscle-specific actiand Xnr-2 expressed zygotgastrulae previously injectemuscle-specific actin (ms-awhich do not express ms-apCSKA:activin never indu
dorsalizing activity in UV rescue experiments, unlike Xnr-1/Xnr-2, it cannot induce axial mesoderm in animal caps(Smith et al., 1995). Xnr-3 is expressed in a pattern overlap-ping Xnr-1/Xnr-2 such that, together with the different activi-ties of the three proteins, the possibility is raised of their func-tional interplay during mesodermal patterning processes.
Xnr-1 and Xnr-2 and axial mesoderm patterning inXenopusMesoderm induction begins around the 32-64 cell stage ofXenopus development (Jones and Woodland, 1987). This isbefore zygotic transcription begins (Newport and Kirschner,1982) and induction therefore depends upon factors stored asRNA or protein in the oocyte. Xnr-1 and Xnr-2 transcripts arefirst detected after zygotic transcription begins, in the vegetalhemisphere of late blastulae, suggesting that neither Xnrfunctions in the initial mesoderm induction process. Nomaternal transcripts for either Xnr-1 or Xnr-2 are detected (data
3660 C. M. Jones and others
Fig. 8. Complete axial rescue of UV-ventralized embryos bylocalized Xnr-1 RNA injection. (A) Normal tadpole, (B) UV-treatedembryo (DAI=0; Kao and Elinson, 1988) at the same age ofdevelopment, (C) Embryo resulting from injection of Xnr-1 RNAinto one cell of a 4-cell stage UV-irradiated embryo. Except for asmall injection artefact in the belly, Xnr-1 rescued embryos areindistinguishable from sibling normal embryos. In a representativeexperiment, 13 of 21 injected embryos (61%) were rescued to a DAIof 3-5, and 8 of these 13 had a DAI of 4 or 5, representing completerescue. A DAI score of 3 was recorded if definitive melanized eyetissue was seen, while a DAI of 5 represents a normal tadpole.
not shown), but it is not known if Xnrs are available to theearly embryo as maternally supplied protein, as is found foractivin (Fukui et al., 1994). Studies with Xnr antibodies willbe needed to address this possibility.
Alternatively, the finding that Xnr RNA is first expressed inthe vegetal region of the blastula raises the possibility that Xnrsignaling at this stage may act as a relay factor, maintaining orintensifying the initial mesoderm induction signals to allowcontinued formation and differentiation of the mesoderm. Asimilar idea has been proposed for FGF and activin (Slack,
1994). In contrast, the vegetal Xnr expression could beinvolved in the specification of the endodermal fate. In eithercase, it is important to consider that differential processing ofXnr precursors, as has been proposed for Vg1 (Dale et al.,1993; Thomsen and Melton, 1993), could cause a dorsal-ventral skewing in the production of active Xnr ligand.
Xnrs dorsalize ventral marginal zone mesodermduring gastrulationDiversification of ventrolateral mesoderm into graded domainsthat turn into striated muscle, nephric tubules and lateral platemesoderm occurs during gastrulation through dorsalization – aprocess dependent upon signals from the Spemann Organizer(Dale and Slack, 1987a; Lettice and Slack, 1993). Potentmesoderm inducers such as activin cannot dorsalize ventralmesoderm during gastrulation (Smith et al., 1993; Lettice andSlack, 1993; this study). However, two unrelated non-mesoderm inducers, noggin and chordin, induce muscle differ-entiation in VMZ explants which otherwise would not formmuscle (Smith et al., 1993; Sasai et al., 1994). Resultspresented here demonstrate that Xnr-1 and Xnr-2, producedduring gastrulation, can also divert VMZ explants from aventral program to a muscle fate. Thus, Xnr-1 and Xnr-2 arenew candidates for mediators of the dorsalization process invivo, but differ from previously characterized signaling factors(noggin, chordin or activin) in being both mesoderm-inducersand dorsalizers of ventral mesoderm during gastrulation.
The expression patterns of Xnr-1 and Xnr-2 at the gastrulastage are compatible with their role as mediators of dorsaliza-tion. Xnr-1 and Xnr-2 expression is highest in the marginalzone and, as illustrated particularly well for Xnr-2, appears toform a gradient of expression with a dorsal maximum and aventral minimum (Fig. 4D,E). We speculate that the specifica-tion of different fates within the spectrum of dorsal-ventralmesoderm depends upon the local level of Xnr signaling. Itwill be important to understand the link between Xnrexpression and earlier patterning events in the embryo; forexample, if the level of Xnr-1 and Xnr-2 expression in themarginal zone is a direct response to the relative levels of dor-salizers (e.g. noggin, chordin) and ventralizers (e.g. BMP4). Inthis respect, it is relevant to note that Xnr-1 and Xnr-2 areinduced in animal caps treated with activin, but not bFGF, indi-cating a stronger activation by dorsoanterior, rather than ven-troposterior, mesoderm inducers.
Nodal-related signalling and the vertebrate bodyplanMouse nodal and Xenopus nodal-related molecules are dose-dependent mesoderm inducers, a finding consistent with thefailure to form mesoderm in mouse embryos homozygous nullfor nodal. In addition, the detection of mouse nodal transcriptsin the early primitive streak and then around the node is verysimilar to the pattern described here for Xenopus Xnr-1 andXnr-2. It is unclear why Xenopus has three nodal-related genes,while mouse carries out similarly complex developmentalprocesses with only a single nodal gene. Mouse nodal mayrepresent an evolutionary convergence in the same moleculeof the different activities of separate ancestral nodal-like genes.Alternatively, the number and type of nodal-like genes indifferent vertebrate classes may depend upon the differentstrategies developed for germ layer specification and gastrula-
3661Nodal-related signaling in mesoderm patterning
tion (including the degree to which separate ways of regulat-ing nodal activities must be achieved). Whether all threeXenopus nodal-related peptides (Xnr-1, Xnr-2 and Xnr-3) arerequired for normal Xenopus development, and if interplaybetween them is an important aspect of their function, is underanalysis.
Is nodal or activin more important in vertebrate embryonicdevelopment? The Xnr activities mimic well the effects ofactivin as a dose-dependent mesoderm inducer in animal caps(e.g. Green et al., 1992). In fish embryos, the effect of injectingdominant negative activin ligands implies that activin isrequired for mesoderm induction in vivo (Wittbrodt and Rosa,1994). Consistent with this, Hemmati-Brivanlou and Melton(1992) showed that a truncated dominant negative activinreceptor blocked mesodermal induction in Xenopus embryos.However, evidence against this point of view also exists. Thedominant negative activin receptor interferes with the signalingpathways of several other TGFβ-like ligands, and experimentswith follistatin, an activin antagonist, lead to the conclusionthat activin is not required for mesoderm induction (Schulte-Merker et al., 1994). Moreover, early embryogenesis isperfectly normal in mouse embryos carrying null mutations foreither, or both, activin subunits (Matzuk et al., 1995; Vassaliet al., 1994; Schrewe et al., 1994), although rescue by maternalactivin protein might obscure an early mesoderm induction role(Matzuk et al., 1995). In contrast, the severe mesodermaldefect in mouse embryos homozygous null for nodal clearlyplaces this gene in a separate category. Complementary studieswith mouse, zebrafish, and now Xenopus and chick, usingembryological, molecular and genetic approaches, stronglyargue that nodal and nodal-related signaling molecules play acentral role in axial mesodermal patterning.
The cDNA sequences for Xnr-1 and Xnr-2 are deposited inGenBank under accession numbers U29447 and U29448. We thankLaura Gamer for critical comments on the manuscript, Michael Rayfor excellent technical support, and Bill Smith and Richard Harlandfor communicating results prior to publication. We are indebted toAmanda Frisch, Karuna Sampath and Raj Ladher for in situ hybrid-ization data. In the initial stages of this work, C. M. J. was anAssociate of the Howard Hughes Medical Institute (HHMI), and isnow a Fellow of the Human Frontiers Science Program. B. L. M. H.is an Investigator of the HHMI, and J. C. S. is an International Scholarof the HHMI. This work was supported by HD-28062 to C. V. E. W.
REFERENCES
Amaya, E., Stein, P. A., Musci, T. J. and Kirschner, M. W. (1993) FGFsignaling in the early specification of mesoderm in Xenopus. Development118, 477-487
Beddington, R. S. P. (1994). Induction of a second neural axis by the mousenode. Development 120, 613-620
Beddington, R. S. P. and Smith, J. C. (1993). Control of vertebrategastrulation: inducing signals and responding genes. Current Opinion inGen. Dev. 3, 655-661.
Blumberg, B., Wright, C. V. E., De Roberts, E. M. and Cho, K. W. Y.(1991). Organizer-specific homebox genes in Xenopus laevis embryos.Science 253, 194-196.
Boterenbrood, E. C. and Nieuwkoop, P. D. (1973). The formation ofmesoderm in urodelean amphibians. V. Its regional induction by endoderm.Wilhelm Roux Arch. EntwMech. Org. 173, 319-332.
Christian, J. L. and Moon, R. T. (1993). Interactions between Xwnt-8 andSpemann organizer signalling pathways generate dorsoventral pattern in theembryonic mesoderm of Xenopus. Genes Dev. 7, 13-28.
Conlon, F. L., Barth, K. S. and Robertson, E. J. (1991). A novel retrovirally
induced embryonic lethal mutation in the mouse: assessment of thedevelopmental fate of embryonic stem cells homozygous for the 413.dproviral integration. Development 111, 969-981.
Conlon, F. L., Lyons, K. M., Takaesu, N., Barth, K. S., Kispert, A.,Hermann, B. and Robertson, E. J. (1994). A primary requirement for nodalin the formation and maintenance of the primitive streak in the mouse.Development 120, 1919-1928.
Cornell, R. A. and Kimelman, D. (1994) Activin mediated mesoderminduction requires FGF. Development 120, 453-462
Dale, L., Howes, G., Price, B. M. J. and Smith, J. C. (1992). Bonemorphogenetic protein 4: a ventralizing factor in early Xenopusdevelopment. Development 115, 573-585.
Dale, L., Matthews, G. and Colman, A. (1993). Secretion and mesoderminducing activity of the TGF-β related domain of Vg1. EMBO J. 12, 4471-4480
Dale, L. and Slack, J. M. W. (1987a). Regional specification within themesoderm of early embryos of Xenopus laevis. Development 100, 279-295.
Dale, L. and Slack, J. M. W. (1987b). Fate map for the 32-cell stage ofXenopus laevis. Development 99, 527-551.
Daopin, S., Piez, K. A., Ogawa, Y. and Davies, D. R. (1992). Crystal structureof transforming growth factor-beta 2: An unusual fold for the superfamily.Science 257, 369-373.
Fainsod, A., Steinbesser, H. and De Robertis, E. M. (1995) On the functionof BMP-4 in patterning the marginal zone of the Xenopus embryo. EMBO J.13, 5015-5025
Faust, C. and Magnuson, T. (1993). Genetic control of gastrulation in themouse. Current Opinion Gen. Dev. 3, 491-498.
Fukui, A., Nakamura, T., Uchiyama, H., Sugino, K., Sugino, H. andAsashima, M. (1994). Identification of activins A, AB, and B and follistatinproteins in Xenopus embryos. Dev. Biol. 163, 279-281
Graff, J. M., Thies, R. S., Song, J. J., Celeste, A. J. and Melton, D. A. (1994).Studies with a Xenopus BMP receptor suggest that ventral mesoderm-inducing signals override dorsal signals in vivo. Cell 79, 169-179.
Green, J. B. A., Howes, G., Symes, K., Cooke, J. and Smith, J. C. (1990).The biological effects of XTC-MIF: quantitative comparison with XenopusbFGF. Development 108, 173-183.
Green, J. B. A., New, H. V. and Smith, J. C. (1992) Response of embryonicXenopus cells to activin and FGF are separated by multiple dose thresholdsand correspond to distinct axes of the mesoderm. Cell 71, 731-739
Harland, R. M. (1991) In situ hybridization: an improved whole-mountmethod for Xenopus embryos. Meth. Cell Biol. 36, 685-695
Hemmati-Brivanlou, A. and Melton, D. A. (1992). A truncated activinreceptor inhibits mesoderm induction and formation of axial structures inXenopus embryos. Nature 359, 609-614
Hogan, B. L. M., Beddington, R., Costantini, F. and Lacey, L. (1995)Manipulating the Mouse Embryo. Second Edition. Cold Spring Harbor Press.
Iannaccone, P. M., Zhou, X., Khokha, M., Boucher, D. and Kuehn, M. R.(1992). Insertional mutation of a gene involved in growth regulation of theearly mouse embryo. Dev. Dynamics 194, 198-208.
Jones, C. M., Lyons, K. M., Lapan, P. M., Wright, C. V. E. and Hogan, B. L.M. (1992a). DVR-4 (Bone Morphogenetic Protein 4) as a posteriorventralizing factor in Xenopus mesoderm induction. Development 115, 639-647.
Jones, C. M., Simon, C. D., Guenet, J. L. and Hogan, B. L. M. (1992b).Isolation of Vgr-2, a novel member of the transforming growth factor-beta-related gene family. Mol. Endocrinol. 6, 1961-1968.
Jones, E. A. and Woodland, H. (1987). The development of animal cap cellsin Xenopus laevis: a measure of the start of animal cap competence to formmesoderm. Development 101, 557-563.
Kao, K. R. and Elinson, R. P. (1988) The entire mesodermal mantle behavesas Spemann’s Organizer in dorsoanterior enhanced Xenopus laevis embryos.Dev. Biol. 127, 64-77
Keller, R. and Tibbetts, P. (1989). Mediolateral cell intercalation in the dorsalaxial mesoderm of Xenopus laevis. Dev. Biol. 131, 539-549.
Kingsley, D. M. (1994). The TGF-β superfamily: new members, newreceptors, and new genetic tests of function in different organisms. GenesDev. 8, 133-146.
Krieg, P. A. and Melton, D. A. (1984). Functional messenger RNAs areproduced by SP6 in vitro transcription of cloned cDNAs. Nucleic Acids Res.12, 7057-7070.
Lawson, K. A. and Pedersen, R. A. (1992). Clonal analysis of cell fate duringgastrulation and early neurulation in the mouse. In PostimplantationDevelopment in the Mouse. Ciba Foundation Symposium 165, 3-26.
3662 C. M. Jones and others
Lettice, L. A. and Slack, J. M. W. (1993). Properties of the dorsalizing signalin gastrulae of Xenopus laevis. Development 117, 263-271.
Maeno, M., Ong, R. C., Suzuki, A., Ueno, N. and Kung, H. F. (1994). Atruncated bone morphogenetic protein 4 receptor alters the fate of ventralmesoderm to dorsal mesoderm: roles of animal pole tissue in thedevelopment of ventral mesoderm. Proc. Natl. Acad. Sci. USA 91, 10260-10264
Mason, A. J. (1994). Functional analysis of the cysteine residues of activin A.Mol. Endocrinol. 8, 325-332.
Matzuk, M., Kumar, T. R., Vassali, A., Bickenbach, J. R., Roop, D. R.,Jaenisch, R. and Bradley, A. (1995). Functional analysis of activins duringmammalian development. Nature 374, 354-356
McDonald, N. Q. and Hendrickson, W. A. (1993). A structural superfamily ofgrowth factors containing a cystine knot motif. Cell 73, 421-424.
McPheron, A. C. and Lee, S. J. (1993). GDF-3 and GDF-9: Two newmembers of the transforming growth factor-beta superfamily containing anovel pattern of cysteines. J. Biol. Chem. 268, 3444-3449.
Moody, S. A. (1987a). Fates of the blastomeres of the 16 cell Xenopus embryo.Dev. Biol. 119, 560-578.
Moody, S. A. (1987b). Fates of the blastomeres of the 32 cell Xenopus embryo.Dev. Biol. 122, 300-319.
Newport, J. and Kirschner, M. (1982). A major developmental transition inearly Xenopus embryos: II. Control of the onset of transcription. Cell 30, 687-696.
Nieuwkoop, P. D. and Faber, J. (1967). Normal Table of Xenopus laevis.(Daudin). Amsterdam: North Holland.
Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). Molecular Cloning - ALaboratory Manual. ed 2. Cold Spring Harbor Laboratory, Cold SpringHarbor.
Sasai, Y., Lu, B., Steinbesser, H., Geissert, D., Gont, L. K. and De Robertis,E. M. (1994) Xenopus chordin: a novel dorsalizing factor activated byorganizer-specific homeobox genes. Cell 79, 779-790
Schlunegger, M. P. and Grutter, M. G. (1992). An unusual feature revealedby the crystal structure at 2.2 A resolution of human transforming growthfactor-β2. Nature 358, 430-434.
Schmidt, J. E., Suzuki, A., Ueno, N. and Kimelman, D. (1995) LocalizedBMP-4 mediates dorsal/ventral patterning in the early Xenopus embryo. Dev.Biol. 169, 37-50
Schrewe, H., Gendronmaguire, M., Harbison, M. L. and Gridley, T. (1994)Mice homozygous for a null mutation of activin beta(B) are viable and fertile.Mech. Dev. 47, 43-51
Schulte-Merker, S., Smith, J. C. and Dale, L. (1994) Effects of truncatedactivin and FGF receptors and of follistatin on the inducing activities ofBVg1 and activin: does activin play a role in mesoderm induction? EMBO J.13, 3533-3541
Sive, H. L. (1993). The frog prince-ss: a molecular formula for dorso-ventralpatterning in Xenopus. Genes Dev 7, 1-12.
Slack, J. M. W. (1984). Regional biosynthetic markers in the early amphibianembryo. J. Embryol. Exp. Morph. 80, 289-319.
Slack, J. M. W. (1994). Inducing factors in Xenopus early embryos. CurrentBiology 4, 116-126.
Smith, J. C. (1989a). Mesoderm induction and mesoderm-inducing factors inearly amphibian development. Development 105, 665-677.
Smith, J. C. (1989b). Induction and early amphibian development. CurrentOpinion in Cell Biology 1, 1061-1070.
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
Smith, W. C., A. K. Knecht, M. Wu, and R. M. Harland. (1993). Secretednoggin protein mimics the Spemann organizer in dorsalizing Xenopusmesoderm. Nature 361, 547-549.
Smith, W. C., McKendry, R. M., Ribisi, S. and Harland, R. M. (1995). Anodal-related gene defines a physical and functional domain within theSpemann Organizer. Cell (in press)
Suzuki, A., Thies, R. S., Yamiji, N., Song, J. J., Wozney, J. M., Murakami,K. and Ueno, N. (1994). A truncated bone morphogenetic protein receptoraffects dorsal-ventral patterning in the early Xenopus embryo. Proc. Natl.Acad. Sci. USA 91, 10255-10259
Thomsen, G. and Melton, D. A. (1993). Processed Vg-1 proteins are axialmesoderm inducers in Xenopus. Cell 74, 433-441
Toyama, R., O’Connell, M. L., Wright, C. V. E., Kuehn, M. R. and Dawid,I. B. (1995). Nodal induces ectopic goosecoid and lim1 expression and axisduplication in zebrafish. Development 121, 383-391
Vassali, A., Matzuk, M. M., Gardner, H. A. R., Lee, K. F. and Jaenisch, R.(1994). Activin/inhibin beta(B) subunit gene disruption leads to defects ineyelid development and female reproduction. Genes Dev. 8, 414-427
von Heijne, G. (1986). A new method for predicting signal sequence cleavagesites. Nucleic Acids Res. 14, 4683-4690
Wittbrodt, J. and Rosa, F. M. (1994). Disruption of mesoderm and axisformation in fish by ectopic expression of activin variants. Genes Dev. 8,1448-1462
Wright, C. V. E., K. W. Y. Cho, J. Hardwicke, R. H. Collins, and E. M.DeRobertis. (1989). Interference with function of a homeobox gene inXenopus embryos produces malformations of the anterior spinal cord. Cell59, 81-93.
Zhou, X., H. Sasaki, L. Lowe, B. L. M. Hogan, and M. R. Kuehn. (1993).Nodal is a novel TGF-β-like gene expressed in the mouse node duringgastrulation. Nature 361, 543-547.
