Establishment of mesodermal tissues in the amphibian body involves a series of inductive interactions probably elicited by a variety of peptide growth factors. Results reported here suggest that mesodermal patterning involves an array of signalling molecules including DVR-4, a TGF-β-like molecule. We show that ectopic expression of DVR-4 causes embryos to develop with an overall posterior and/or ventral character, and that DVR-4 induces ventral types of mesoderm in animal cap explants. Moreover, DVR-4 overrides the dorsalizing effects of activin. DVR-4 is therefore the first molecule reported both to induce posteroventral mesoderm and to counteract dorsalizing signals such as activin. Possible interactions between these molecules resulting in establishment of the embryonic body plan are discussed.

Mechanisms by which the embryonic body plan is established are being studied in species as diverse as the fruit fly, frog and mouse. Drosophila and murine systems emphasize genetic manipulation to study these problems, whereas the frog is amenable to dissection and biochemical studies. One of the best understood systems of embryonic patterning is the process of anterior axis specification in Drosophila. A gradient of bicoid protein initiates expression of genes such as hunchback, thereby establishing domains of anterior development. By contrast, specification of posterior regions of the fly embryo (the abdomen) requires the action of nanos, which seems to repress expression of hunchback, thereby allowing expression of posterior gap genes such as knirps and giant (reviewed by St. Johnston and Nusslein-Volhard, 1992). Therefore, it appears that in Drosophila an initial step in posterior development is the active repression of genes involved in anterior specification.

In contrast, molecular interactions leading to establishment of the vertebrate body plan are poorly understood. Fate mapping (Dale and Slack, 1987a) and tissue explantation experiments (Dale and Slack, 1987b) have demonstrated that mesoderm from different regions of the early Xenopus embryo gives rise to different tissue types. One model put forward to account for these observations is the “three signal hypothesis” of Slack and colleagues (Dale and Slack, 1987b; Smith, 1989). This theory proposes that two states of mesoderm are induced by signals originating in the vegetal pole and acting on the marginal zone of the amphibian embryo. The dorsovegetal signal induces extreme dorsal mesoderm, the future Spemann organizer, while the ventrovegetal signal induces ventral mesoderm, predominantly blood, mesenchyme and mesothelium. The third signal arises from newly induced dorsal tissue and dorsalizes the adjacent ventral mesoderm, thereby directing formation of the majority of the muscle in the amphibian embryo (Dale and Slack, 1987a). Since the formulation of this model, a great deal of effort has focused on identifying molecules that function as these signals. A major finding of these efforts is that these signalling molecules are peptide growth factors (PGF). The ventrovegetal signal is probably related to fibroblast growth factor (FGF) (Slack et al., 1987; reviewed by Smith, 1989). Moreover, recent studies involving microinjection of wnt-1 and wnt-8 RNAs show that a molecule of this class is capable of inducing a functional Spemann organizer (Sokol et al., 1991; Smith and Harland, 1991), and therefore is probably related to the endogenous dorsovegetal signal. The TGF-0 family of growth factors is the third class of signalling molecules that has been implicated in axis specification. Isolated animal cap cells exhibit several thresholds of differential response to activin (Green and Smith, 1990), and consequently differentiate into either dorsal (high activin concentrations) or more ventral (lower concentrations) mesodermal cell types. These studies demonstrate that activin can function as a morphogen, a property that would be expected of a dorsalizing signal originating from the organizer region and acting on adjacent mesoderm. Of course, in vivo localization of these proteins and their receptors is absolutely required before firm conclusions can be reached.

One class of TGF-β-related molecules, which have expression patterns consistent with a role in many inductive interactions during vertebrate development, is the DVR family (originally known as Bone Morphogenetic Proteins; reviewed in Lyons et al., 1991). In situ hybridization studies suggest that two members of the family, DVR-2 and -4, act as signalling molecules during murine development. For example, expression patterns in the myocardium of the early embryonic heart, the apical ectodermal ridge and mesenchyme of the limb buds, and the mesenchyme and ectodermal placodes of developing hair follicles are consistent with a role in epithelial-mesenchymal interactions (Lyons et al., 1990; Jones et al., 1991). In addition, DVR-4 transcripts are localized during gastrulation, first to the allantois at 7.5 days p.c., and by 8.5 days p.c., to posterior and ventral tissues in the primitive streak region (Jones et al., 1991). This led us to hypothesize that DVR-4 specifies posterior and ventral mesodermal cell types during late gastrulation stages of mouse development. Further support for this hypothesis came from reports that DVR-4 transcripts are present throughout early Xenopus development and that DVR-4 conditioned medium induces animal cap tissue to differentiate into mesoderm, albeit in a low percentage of the cases (Koster et al., 1991).

In this report, we use the Xenopus model system to test our hypothesis that DVR-4 specifies posterior or ventral mesodermal cell types. We use a combination of RNA microinjection, tissue transplantation and animal cap induction assays to test this idea. Taken together, our results suggest that DVR-4 counteracts the dorsalizing effects of activin, and by itself induces formation of mesoderm with posteroventral characteristics.

Xenopus embryo manipulation

Eggs were obtained from hormonally stimulated Xenopus laevis females and were fertilized in vitro using testis homogenates. Dejellied embryos were allowed to develop as described (Cho et al., 1988) and staged according to Nieuwkoop and Faber (1967). Animal caps were isolated at stage 8, treated with the various factors in 0.5 ×MMR(l × MMR =100 mM NaCl, 2 mM KC1, 1 mM MgSO4, 2 mM CaCh, 5 mM HEPES, pH 7.6, 0.1 mM EDTA) for several hours until rounded, and transferred to half-strength NAM (Asashima and Grunz, 1983) for long-term culture. Tissues were fixed in Bouin’s fixative, paraffin embedded, sectioned at 7 μm, and stained with hematoxylin and eosin.

Preparation of RNA and microinjection

A human DVR-4 cDNA (kindly provided by John Wozney, Genetics Institute) was amplified through PCR so that an Ncol site was introduced over the initiation codon (5’GACACCATGGTTCCTG-GTAACC3’) and an Xbal site introduced in the 3’ region, approximately 60 bp 3’ of the stop codon. The amplified product was then cloned into Ncol/Xbai restricted pSP64-X βM (Krieg and Melton, 1984). The β-globin 5’ leader in this construct has been shown to direct efficient production of proteins when injected into oocytes and embryos (Wright et al., 1989). XZxtl-linearized template was used to produce 5’-capped RNA using SP6 RNA polymerase (Promega) and standard methods (Krieg and Melton, 1987). RNA from three separate transcription reactions was used and each produced a protein of expected size when translated in a rabbit reticulocyte lysate (Promega). Injections were carried out with a Singer micromanipulator and a gas-driven injector (Narashige). Approximately 0.5 –1.0 ng RNA was injected over a 5 second period for one-celled embryo injections and approximately 1-3 pg RNA was delivered to single blastomeres for dorsal and ventral injections at cleavage stages (32 –64 cell stage). Presumptive dorsal and ventral sides of embryos were identified by pigmentation (Nieuwkoop and Faber, 1967), and embryos with regularly tiered cleavage patterns were chosen. A sample of siblings was always left to develop normally to ensure that the pigmentation criterion was accurate. Control injections used only buffer (Wright et al., 1989), RNA encoding p-globin, or RNA encoding a mutant form of DVR-6, which has no effect on development. Einsteck assays were performed by transplanting half an animal cap into the blastocoel of stage 10-10.5 host embryos, and pushing the transplanted tissue to the ventral side. Surgery was performed in 1 × MBS (Gurdon, 1976) using an eyebrow hair knife, and the embryos were allowed to heal for approximately 30 minutes, transferred back to 0.199 x MBS, and allowed to develop.

RNAse protection

RNA production and RNAse protection analyses for different molecular markers were performed as described (Cho and De Robertis, 1990) on a batch of RNA from a single injection experiment. All assays were performed at least twice on RNA from separate injection experiments. Input RNA for cardiac actin and EF-1 a was half an embryo equivalent, while for XBra, wnt-8, DVR-4, and Xhox 3, three embryo equivalents (approximately 12 μg total RNA) were used. The XBra cDNA was provided by Dr. J. C. Smith, the Xenopus DVR-4 probe by Dr. N. Ueno, the Xenopus wnt-8 probe by Dr. A. McMahon, and the Xhox 3 probe by Drs. M. Saha and R. Grainger.

DVR-4 RNA injection into one-celled embryos

Our initial experiments involved injecting synthetic human DVR-4 RNA into one-celled Xenopus embryos. This results in the ectopic expression of DVR –4 in all cells of the developing embryo. Development of the injected embryos proceeds normally through cleavage and blastula stages, but by neurula stages it is evident that no dorsal structures form. DVR-4-injected embryos develop through the formation of a normal dorsal lip (stage 10) and yolk plug (stage 11 –12), and apparent closure of the blastopore (Fig. 1). However, histological examination of the DVR-4 injected embryos reveals that gastrulation movements are inhibited (approx. 99% of injected embryos show this phenotype). No mesoderm is seen below the blastocoel roof, but a clear delineation is evident between more darkly staining tissue, tentatively identified as mesoderm (molecular evidence presented below), and yolky endoderm (Fig. 2).

Fig. 1.

Effect of DVR-4 RNA injections on embryonic development. Dorsal views of stage 18 control embryo (A) and DVR-4-injected embryo (D) reveal absence of dorsal structures in the injected embryo. Comparing control (B,C) and injected (E,F) embryos at earlier stages reveals formation of a normal dorsal blastopore lip (E) and yolk plug (F) in DVR-4-injected embryos. Scale bar = 300 mm.

Fig. 1.

Effect of DVR-4 RNA injections on embryonic development. Dorsal views of stage 18 control embryo (A) and DVR-4-injected embryo (D) reveal absence of dorsal structures in the injected embryo. Comparing control (B,C) and injected (E,F) embryos at earlier stages reveals formation of a normal dorsal blastopore lip (E) and yolk plug (F) in DVR-4-injected embryos. Scale bar = 300 mm.

Fig. 2.

Histological examination of DVR-4-injected embryos and loaded animal caps. (A) Examination of injected embryo (sibling stage 18) shows that gastrulation movements are inhibited but tissue resembling mesoderm develops, as judged by cell size and staining, (delimited from yolky endoderm cells by arrowheads). The collapsed blastocoel roof emphasizes absence of supporting mesoderm. Caps isolated from stage 8 DVR-4-injected embryos and then treated with activin (B) form structures (93%, n =57) containing a large mass of tissue resembling the mesoderm-like tissue shown in A. Caps isolated from DVR-4-injected embryos and cultured without further treatment (C) form vesicles with mesothelium and blood-like cells (90%, n =31). Scale bars = 100 μm (A) or 50 μm (B,C).

Fig. 2.

Histological examination of DVR-4-injected embryos and loaded animal caps. (A) Examination of injected embryo (sibling stage 18) shows that gastrulation movements are inhibited but tissue resembling mesoderm develops, as judged by cell size and staining, (delimited from yolky endoderm cells by arrowheads). The collapsed blastocoel roof emphasizes absence of supporting mesoderm. Caps isolated from stage 8 DVR-4-injected embryos and then treated with activin (B) form structures (93%, n =57) containing a large mass of tissue resembling the mesoderm-like tissue shown in A. Caps isolated from DVR-4-injected embryos and cultured without further treatment (C) form vesicles with mesothelium and blood-like cells (90%, n =31). Scale bars = 100 μm (A) or 50 μm (B,C).

To assess the mesoderm-inducing activity of the DVR-4 protein made in vivo, we explanted animal caps from injected embryos and cultured them for approximately two days (sibling stage 32) before processing them for histology. Histological examination reveals that the DVR-4 loaded caps form vesicular structures, characteristic of ventral mesodermal differentiation in this assay (Fig. 2). In addition, we treated the loaded caps with activin at concentrations (10, 20 and 50 ng/ml) known to induce dorsal types of mesoderm (striated muscle and notochord) in animal caps. DVR-4 loaded caps treated in this way do not form structures recognizable as dorsal mesoderm, but instead develop into vesicular structures filled with a solid mass of tissue (Fig. 2). It therefore appears that DVR-4 not only induces mesoderm of a ventral type, but also overrides the dorsalizing effects of activin.

These animal cap assays suggest that DVR-4 specifies ventral mesoderm. Evidence that DVR-4 can direct formation of posterior tissues is illustrated in Fig. 3, using the Einsteck procedure, in which tissue is transplanted into the blastocoel of host embryos. The majority of DVR-4-loaded animal caps transplanted into the blastocoel of stage 10 host embryos direct formation of a secondary posterior axis (73%, n =22). In addition to external appearance, histological examination shows that these secondary axes contain both notochord and somite arrays, indicative of true secondary axes. In control Einsteck experiments, animal caps from uninjected sibling embryos produce no secondary axes, while uninjected caps treated with 20 ng/ml activin induce extensive secondary axes containing head structures (Fig. 3). DVR-4-loaded caps treated with activin (10 –50 ng/ml) never induce secondary’ anterior axes, consistent with the overriding of activin effects in explanted caps. In a few cases (10%), structures resembling a secondary posterior axis were formed, but in the majority of cases, the implanted tissue remained as an enlarged bump, with no external axial morphology (data not shown).

Fig. 3.

Effect of injecting DVR-4 RNA into dorsal and ventral blastomeres and implanting animal caps into blastocoel of gastrula embryos (Einsteck procedure). (A) DVR-4 RNA injected into presumptive dorsal blastomeres of 16 – to 32-cell embryos inhibits the formation of anterior structures (lower embryo), while RNA injected into prospective ventral blastomeres has no discernable effect on development (upper embryo). (B,C) Results of Einsteck experiments with animal caps from DVR-4-injected embryos (B) or animal caps from uninjected sibling embryos treated with 20 ng/ml activin (C). The DVR-4-loaded caps induce secondary posterior axes, whereas the activin-treated caps induce extensive secondary axes containing anterior structures.

Fig. 3.

Effect of injecting DVR-4 RNA into dorsal and ventral blastomeres and implanting animal caps into blastocoel of gastrula embryos (Einsteck procedure). (A) DVR-4 RNA injected into presumptive dorsal blastomeres of 16 – to 32-cell embryos inhibits the formation of anterior structures (lower embryo), while RNA injected into prospective ventral blastomeres has no discernable effect on development (upper embryo). (B,C) Results of Einsteck experiments with animal caps from DVR-4-injected embryos (B) or animal caps from uninjected sibling embryos treated with 20 ng/ml activin (C). The DVR-4-loaded caps induce secondary posterior axes, whereas the activin-treated caps induce extensive secondary axes containing anterior structures.

Analysis of gene expression in DVR-4 injected embryos

A variety of markers for different embryonic tissue types were used to assess the effects of DVR-4 at the level of gene expression. We used cardiac actin which is specific for muscle, and Brachyury (XBra; Smith et al., 1991), which is expressed initially in the marginal zone, and later around the blastopore and in the notochord, wnt-8 (expressed in ventral mesoderm), and Xenopus DVR-4. In addition, we chose the homeobox gene Xhox 3 as a marker of posteroventral mesoderm. Xhox 3 is first expressed predominantly in mesoderm, and has a graded distribution in early development, with higher levels of expression in posterior than in anterior regions (Ruiz i Altaba and Melton, 1989a).

In DVR-4 injected embryos, XBra transcripts increase transiently relative to controls after zygotic transcription begins, reaching a peak approximately 4-fold above control levels by stage 10 (Fig. 4A and data not shown). Subsequently, XBra RNA levels decline. Since XBra expression becomes restricted to the notochord after mesodermal invagination (Smith et al., 1991), and is also expressed around the blastopore during neurula stages, the absence of expression in injected embryos at these stages probably reflects the fact that morphogenesis and the progression of dorsal mesodermal differentiation is halted in the DVR-4 injected embryos. Likewise, cardiac actin RNA levels are undetectable in the injected embryos (Fig. 4A), consistent with the absence of striated muscle. In contrast, a dramatic increase in the level of Xhox 3 is seen in the injected embryos, first evident at the midblastula transition (data not shown) and reaching high levels by stages 10 and 13 (Fig. 4B). The Xhox 3 expression suggests that mesoderm induction occurs in the injected embryos, (Ruiz i Altaba and Melton, 1989a), and that the mesoderm is most likely a posterior or ventral type, since-it does not express markers of dorsal mesoderm.

Fig. 4.

(A) RNAse protection analysis of gene expression in control (−) and DVR-4-injected (+) embryos. Expression profiles of the molecular markers used suggest that development proceeds normally in the DVR-4-injected embryos until gastrulation stages, when expression of the dorsal mesodermal markers, Brachyury (XBra) and cardiac actin, is dramatically reduced, while wnt-8 and DVR-4 continue to be expressed. Human DVR-4 autoinduces expression of endogenous Xenopus DVR-4 (2-fold) shortly after the midblastula transition (Stage 9). The protection pattern for the DVR-4 probe results from nucleotide variations in DVR-4 transcripts from the tetrapioid Xenopus genome. Densitométrie quantitation shows that XBra expression also increases 2-fold over control levels shortly after stage 9. (B) Xhox 3 expression is also seen to dramatically increase in the DVR-4-injected embryos, reaching peak levels in late gastrula embryos, remaining elevated through stage 17 (data not shown), but returning to levels comparable to control by late neurula stages. Absence of a Xhox 3 band in the stage 10 control embryo lane is due to exposure time and low yield of RNA. EF-1 p is used to standardize amounts of RNA in each assay.

Fig. 4.

(A) RNAse protection analysis of gene expression in control (−) and DVR-4-injected (+) embryos. Expression profiles of the molecular markers used suggest that development proceeds normally in the DVR-4-injected embryos until gastrulation stages, when expression of the dorsal mesodermal markers, Brachyury (XBra) and cardiac actin, is dramatically reduced, while wnt-8 and DVR-4 continue to be expressed. Human DVR-4 autoinduces expression of endogenous Xenopus DVR-4 (2-fold) shortly after the midblastula transition (Stage 9). The protection pattern for the DVR-4 probe results from nucleotide variations in DVR-4 transcripts from the tetrapioid Xenopus genome. Densitométrie quantitation shows that XBra expression also increases 2-fold over control levels shortly after stage 9. (B) Xhox 3 expression is also seen to dramatically increase in the DVR-4-injected embryos, reaching peak levels in late gastrula embryos, remaining elevated through stage 17 (data not shown), but returning to levels comparable to control by late neurula stages. Absence of a Xhox 3 band in the stage 10 control embryo lane is due to exposure time and low yield of RNA. EF-1 p is used to standardize amounts of RNA in each assay.

RNAse protection analysis also reveals that injection of human DVR-4 induces expression of the endogenous Xenopus DVR-4 genes (Fig. 4A). Endogenous DVR-4 levels are approximately 2-fold more than control levels shortly after the midblastula transition, and rise to a 7-fold difference by sibling stage 18. This 7-fold difference is maintained at least until siblings reach stage 35 (data not shown). Thus, DVR-4 may upregulate expression of the endogenous Xenopus DVR-4 genes and/or affect decay of its mRNA. Finally, wnt-8 transcripts remain at control levels until at least sibling stage 35 (Fig. 4A and data not shown).

Localized misexpression of DVR-4 in cleavage stage embryos

The results described above, in particular the dramatic increase in Xhox 3 expression in injected embryos, suggest that DVR-4 inhibits the formation of anterior-dorsal mesoderm, and instead, specifies mesoderm of a posterior-ventral character. Since injection of DVR-4 into one-celled embryos causes an inhibition of gastrulation, only limited information can be gained as to the role DVR-4 plays in overall patterning of the embryonic mesoderm. We therefore attempted to find an assay allowing more localized misexpression of the protein. In particular, we wanted to test the effects of overexpressing the protein in presumptive dorsoanterior regions of the embryo. DVR-4 RNA injected into two prospective dorsal or ventral blastomeres of 4 – and 8 –cell embryos consistently arrests gastrulation in injected embryos. However, if relatively small amounts of RNA (approximately 1-3 pg) are injected into 1 or 2 blastomeres of 32-to 64-cell embryos, the injected embryos can proceed past gastrulation stages. Fig. 3 shows that misexpression of DVR-4 on the presumptive dorsal side of the embryo results in an inhibition of anterior structures (82% of dorsally injected embryos). Embryos injected in this manner have a much more defined anteroposterior axis than one-cell injected embryos, but lack anterior structures (heads). This is strikingly similar to the phenotype resulting from injection of Xhox 3 RNA into a similar location (Ruiz i Altaba and Melton, 1989b). This phenotype is consistent with the hypothesis that DVR-4 specifies more posterior structures or inhibits dorsoanterior differentiation. Furthermore, if the same amount of DVR-4 RNA is injected into presumptive ventral blastomeres, the embryos are morphologically indistinguishable from uninjected sibling controls at least until swimming tadpole stages.

Animal cap assays using purified DVR-4, activin and Bfgf

A reasonable hypothesis from the experiments described above is that DVR-4 and activin could interact to establish an ordered anterior-posterior pattern in the Xenopus embryo. We therefore used purified factors in animal cap induction assays in an attempt to quantitate interactions between the two molecules. From the above experiments, one prediction is that purified DVR-4 should induce a ventral type of mesoderm, and 10 –50 ng/ml activin more dorsal mesoderm, while a combination of the two factors should induce ventral and/or intermediate types of mesoderm. When we treated animal caps with DVR-4 protein alone (10-200 ng/ml), in seven separate experiments, less than 5% of treated caps form recognizable mesodermal tissues. The majority differentiate, like controls, into ciliated atypical epidermis, when characterized histologically (Table 1, Fig. 5, and data not shown). Consistent with other reports, caps treated with activin alone develop into elongated structures containing notochord, muscle and neural tissue. However, when the two factors are added in combination, the majority of caps develop into vesicular structures, representative of ventral mesodermal differentiation (Fig. 5, panel D). Moreover, the caps treated with a combination of DVR-4 and activin express higher levels of Xhox 3 than do those treated with activin alone (Fig. 6A), again suggesting that the mesoderm induced in these caps has a posterior and/or ventral character.

Table 1.

Mesodermal differentiation induced by purified, recombinant human activin and DVR-4

Mesodermal differentiation induced by purified, recombinant human activin and DVR-4
Mesodermal differentiation induced by purified, recombinant human activin and DVR-4
Fig. 5.

Effects on animal caps of purified DVR-4 protein (B), activin A (C), or a combination of the two factors (D). Based on histological examination, control (A) and DVR-4-treated (B; 100 ng/ml) caps form ciliated atypical epidermis, while activin (C; 10 ng/ml) induces striated muscle (m), notochord (no), and neural tissue (nt). A combination of activin and DVR-4 at varying concentrations (Table 1) induces mainly vesicles containing mesenchyme, mesothelium (arrowhead), and presumptive blood cells (D). Scale bars = 50 μm.

Fig. 5.

Effects on animal caps of purified DVR-4 protein (B), activin A (C), or a combination of the two factors (D). Based on histological examination, control (A) and DVR-4-treated (B; 100 ng/ml) caps form ciliated atypical epidermis, while activin (C; 10 ng/ml) induces striated muscle (m), notochord (no), and neural tissue (nt). A combination of activin and DVR-4 at varying concentrations (Table 1) induces mainly vesicles containing mesenchyme, mesothelium (arrowhead), and presumptive blood cells (D). Scale bars = 50 μm.

Fig. 6.

RNAse protection analysis of Xhox 3 expression in animal caps treated with activin A, DVR-4, bFGF, or combinations. (A) Xhox 3 expression in caps treated with activin and DVR-4 compared with caps isolated from injected embryos. Xhox 3 is induced to highest levels in DVR-4-loaded caps, while activin in combination with DVR-4 protein induces higher levels than either activin or DVR-4 alone (a faint band, not photographically reproducible, is present in the DVR-4 lane; see Fig. 6B). (B) Xhox 3 induction in explants treated with bFGF and DVR-4. When standardized to EF-1P, Xhox 3 levels are comparable in bFGF and bFGF + DVR-4-treated caps, and both treatments induce the same type of (ventral) mesodermal morphology (data not shown). Additionally, purified DVR-4, while not inducing mesodermal tissue recognizable by histology, induces a very low level of Xhox 3 expression, and control caps are negative. In other experiments (e.g. Fig. 6A), where RNA yields were higher, a protected Xhox 3 fragment in control caps is never detected, and is consistent with published reports (Ruiz i Altaba and Melton, 1989c). Numbers in parentheses are concentrations in ng/ml.

Fig. 6.

RNAse protection analysis of Xhox 3 expression in animal caps treated with activin A, DVR-4, bFGF, or combinations. (A) Xhox 3 expression in caps treated with activin and DVR-4 compared with caps isolated from injected embryos. Xhox 3 is induced to highest levels in DVR-4-loaded caps, while activin in combination with DVR-4 protein induces higher levels than either activin or DVR-4 alone (a faint band, not photographically reproducible, is present in the DVR-4 lane; see Fig. 6B). (B) Xhox 3 induction in explants treated with bFGF and DVR-4. When standardized to EF-1P, Xhox 3 levels are comparable in bFGF and bFGF + DVR-4-treated caps, and both treatments induce the same type of (ventral) mesodermal morphology (data not shown). Additionally, purified DVR-4, while not inducing mesodermal tissue recognizable by histology, induces a very low level of Xhox 3 expression, and control caps are negative. In other experiments (e.g. Fig. 6A), where RNA yields were higher, a protected Xhox 3 fragment in control caps is never detected, and is consistent with published reports (Ruiz i Altaba and Melton, 1989c). Numbers in parentheses are concentrations in ng/ml.

To further assess the actions of purified DVR-4 in animal cap mesoderm induction assays, and as a control for toxic effects of our purified protein, we treated isolated explants with either bFGF alone, or bFGF in combination with DVR-4. Animal caps treated under both conditions are histologically indistinguishable (data not shown), and Xhox 3 expression levels are comparable in both groups (Fig. 6B). Xhox 3 levels are however, substantially higher in caps treated with bFGF in combination with DVR-4 than in caps treated with DVR-4 alone, suggesting that DVR-4 does not non-specifically inhibit the response to all mesoderm inducers. These results argue that the effects seen in combination with activin are not due to toxic contaminants in the DVR-4 protein preparation.

RNA microinjection suggests a role for DVR-4 in mesodermal patterning

In order to ectopically express DVR-4 during early stages of development, we injected RNA encoding human DVR-4 into one-celled Xenopus embryos. The 100% amino acid sequence identity in the presumed mature region (last 100 amino acids) between the human and frog proteins makes differences in actions of the two proteins unlikely, and overexpression of the human transcript allows us to distinguish between injected and endogenous RNA. The first result observed in our DVR-4 injection experiments is the absence of dorsal structures such as the neural tube by the time sibling embryos reach neurula stages (Fig. 1). Moreover, sectioning of the injected embryos reveals that gastrulation movements are inhibited. Mesodermal tissue appears to be present in the injected embryos, as judged by the clear delineation between darkly staining smaller cells (mesoderm) and endoderm as shown in Fig. 2. Confirmation that mesodermal tissue is present in the injected embryos comes from molecular analysis of gene expression. While markers of dorsal mesoderm derivatives, such as cardiac actin and Xbra, are absent in later stage injected embryos, genes such as wnt-8 (expressed in ventral mesoderm; Christian et al., 1991) and Xhox 3 (first detected in posterior mesoderm; Ruiz i Altaba and Melton, 1989a) are expressed in the injected embryos. Moreover, Xhox 3 is induced to extremely high levels in DVR-4-injected embryos, with peak expression occurring during early gastrula stages (Fig. 4B), and persisting throughout gastrulation. Since the normal expression of Xhox 3 first peaks at approximately stage 15 (neurula), it appears that ectopic DVR-4 expression alters the timing of Xhox 3 expression, precociously increasing transcript levels earlier in development. Because of the graded distribution of Xhox 3transcripts in normal embryos (higher in posterior regions; Ruiz i Altaba and Melton, 1989a), and Xhox 3 mRNA injection being incompatible with anterior differentiation, one feasible explanation for the phenotype caused by DVR-4 injection is that the embryo develops a uniform posteroventral character, with gastrulation being inhibited because of the absence of dorsal mesoderm, which through convergent extension normally drives gastrulation.

It has been reported that wnt-8 is expressed in ventral mesoderm, but wnt-8 transcript levels are unchanged in UV ventralized embryos, and its normal role during early development is unclear (Christian et al., 1991). If, as has been suggested, wnt-8 is a true marker of ventral mesoderm, the fact that wnt-8 expression is unchanged, but Xhox 3 is highly induced in DVR-4-injected embryos, argues more for a posteriorizing effect than a ventralizing effect. However, it should be noted that there is no clear distinction between these two positional values during early amphibian development (reviewed by Smith, 1989).

To test the mesoderm-inducing ability of DVR-4, we explanted and cultured animal caps from injected embryos. These explants develop into ventral mesoderm (Fig. 2) and, consistent with this phenotype, express high levels of Xhox 3 (Fig. 6A). When the same, loaded caps are exposed to concentrations of activin that normally induce striated muscle and notochord, they form vesicular structures almost filled with a mass of tissue perhaps resembling the intermediate type of mesoderm induced by high concentrations of FGF (Fig. 2). Caps treated with activin in combination with DVR-4 express lower levels of Xhox 3 than DVR-4 loaded caps, but higher levels than caps treated with activin alone (Fig. 6A). Therefore, it appears that DVR-4 acts to divert animal cap cells from their normal response to activin, so that they differentiate instead into mesoderm with a more ventral or posterior character. DVR-4 is therefore the first molecule reported both to induce ventral mesoderm, and to act dominantly over the dorsalizing effects of activin.

Further support for our hypothesis that DVR-4 actively specifies posteroventral mesoderm is obtained when DVR-4-loaded animal caps are used in Einsteck assays. The caps from injected embryos instruct the formation of secondary posterior axes when transplanted into the host blastocoel (Fig. 3). This is similar to the results reported for bFGF-treated (Ruiz i Altaba and Melton, 1989c) and XlHbox 6-loaded caps (Cho et al., 1991). Moreover, DVR-4-induced secondary axes contain notochord and aligned somites extending throughout the secondary structures (data not shown), and are therefore more complete than those reported for XlHbox 6 (Cho et al., 1991). While an organized neural tube has not been observed in our secondary axes, they do contain melanocytes, a neural crest derivative, but the origin of these cells in the secondary axes is unknown, and they could represent cells that have migrated in from the primary axis. Nevertheless, our results suggest that DVR-4 alone initiates a cascade of signals that establishes an extensive posterior axis in the Xenopus embryo. Furthermore, localized overexpression of DVR-4 on the presumptive dorsal side of the embryo inhibits formation of dorsoanterior structures, while the same amount of RNA injected into future ventral blastomeres has no apparent effect on development (Fig. 3). These results are consistent with the hypothesis that DVR-4 induces posteroventral mesoderm and counteracts the effects of dorsalizing signals such as activin (see below).

DVR-4 loaded animal caps used in Einsteck assays interact with naive ventral tissues to induce secondary tails, but DVR-4 mRNA injected into prospective ventral blastomeres has no apparent effect on development. These data suggest that the loaded caps act as an organizing center to direct production of secondary posterior structures, a situation different from injection of mRNA into similar ventral regions.

Exogenous DVR-4 counteracts the effects of activin, and acts in establishing ventral mesoderm

To better characterize interactions between DVR-4, activin and bFGF, the purified factors were used in animal cap assays for mesoderm induction. Surprisingly, purified DVR-4 is inefficient at inducing mesoderm when applied to animal cap explants at concentrations between 10 and 200 ng/ml (0%, Table 1; no more than 5%, unpublished observations). The fact that animal caps respond to DVR-4 produced in vivo from injected RNA by forming ventral mesoderm, but not to purified exogenous protein suggests that the caps must be exposed to high local concentrations of the factor for prolonged periods of time in order to differentiate into ventral mesodermal tissues. This is a condition achieved by RNA injection but not by incubation in the purified factor. Alternatively, the DVR-4 produced from injected mRNA by embryonic cells could be qualitatively different from the purified protein (e.g. post-translational modifications). Nevertheless, the purified protein preparation does counteract the effects of activin when the two factors are applied in combination.

We have considered the possibility that DVR-4 could attenuate the actions of activin by blocking access to the activin receptor. While it is theoretically possible that this mechanism could in fact be used by the embryo, recent reports show only very low levels of competition between activin and DVR-4 (BMP-4) for several different splice variants of the activin receptor in vitro (Artisano et al., 1992), thereby arguing against this being the sole mechanism for DVR-4 inhibiting the effects of activin. If DVR-4 does not block the activin receptor, it must be eliciting some molecular change in the animal cap cells that alters their response to activin. While the majority of animal caps treated with purified DVR-4 were histologically indistinguishable from control caps (forming atypical epidermis), we found that Xhox 3 was induced, albeit to very low levels, in the DVR-4-treated caps (Fig. 6B). It therefore seems that the purified factor (at least as applied here, for a limited time to whole animal caps) induces a program of gene expression that is not sufficient by itself to induce mesoderm differentiation, but is still able to direct mesoderm induced by “high concentrations” of activin toward more posteroventral development. Clearly, more information is needed concerning receptors, protein-binding specificities, and the intracellular signalling molecules they generate, before precise mechanisms can be established.

Another PGF widely studied in amphibian mesoderm induction is bFGF. Since bFGF is thought to induce mesoderm of a ventral or intermediate type (Slack et al., 1987), its interactions with DVR-4 were also studied. When caps are cultured in bFGF at concentrations that normally result in the formation of intermediate types of mesoderm (50-150 ng/ml), no histological differences are detectable between caps treated with bFGF or bFGF in combination with DVR-4 (data not shown), and Xhox 3 expression is comparable in both groups (Fig. 6B). The fact that bFGF and DVR-4 seem to induce similar tissue types does not make this result surprising. However, it is possible that the two factors induce tissue types that have different molecular characteristics. In support of this hypothesis is the one important difference between the effects of adding bFGF and DVR-4. Experiments show that tissue induced by bFGF can be dorsalized when exposed to PGFs such as activin (i.e. activin is dominant over bFGF; Cooke, 1989). However, our experiments using DVR-4-loaded caps treated with activin, and purified activin in combination with exogenous DVR-4, suggest that tissue exposed to DVR-4 cannot express a dorsalized phenotype (i.e. that the effect of DVR – 4 is dominant over that of activin).

A revised model for mesoderm patterning

The data reported here regarding inductive properties of DVR-4 could warrant a revision of the “three signal model” for mesoderm patterning (Dale and Slack, 1987b). DVR-4 is capable of inducing mesoderm with ventral and posterior characteristics, as assayed by animal cap induction and Einsteck assays, respectively. Furthermore, DVR-4 produced in vivo through RNA injection, and purified DVR-4 added to isolated animal caps, are both capable of overriding the dorsalizing effects of activin. We therefore propose that the ventrovegetal signal of induction may involve not only FGF-like molecules, but also DVR-4.

Our hypothesis would place DVR-4 as an inducer and/or specifier of extreme ventroposterior mesoderm, while FGF, or FGF in combination with DVR-4, would induce mesoderm with more intermediate or lateral characteristics. Moreover, mesoderm induced by FGF is capable of responding to dorsalizing signals from the organizer. DVR-4 and low concentrations of activin or bFGF induce indistinguishable phenotypes in animal cap induction assays. However, at high concentrations, both activin and FGF can induce dorsal and intermediate mesodermal tissues, respectively, while this is not the case for DVR-4. In addition, DVR-4 actions are dominant over those of activin while FGF effects are not.

In addition to DVR-4 actively inducing ventroposterior mesoderm in vivo, we propose that the protein also maintains tissue in a state of ventroposterior differentiation. In support of this idea, we note that injection of DVR-4 autoinduces expression of the endogenous genes and that this induction is maintained for an extended period of time (Fig. 4A and data not shown). In light of these results, we suggest that endogenous DVR-4 might actively oppose the dorsalizing signal (possibly activin), thereby inhibiting gene expression leading to differentiation of dorsoanterior phenotypes, a situation analogous to the posterior determination system in the Drosophila embryo (see Introduction). In addition, DVR-4 induces genes such as Xhox 3, which may set up a cascade of gene expression that directs posteroventral development, whereby dorsalizing signals are disregarded. In this way, establishing the posterior and ventral regions of the amphibian embryo is probably an active process and not simply a default program due to absence of dorsalizing signals.

We thank Anthony Mason (Genentech) for supplying purified activin A, Elizabeth Wang and loannis Moutsalsus (Genetics Institute) for purified DVR-4, and Lorene Batts for excellent technical assistance. This work was supported by NIH grant RO 125580 to B.L.M.H. and HD28062 to C.V.E.W.

Artisano
,
L.
,
Wrana
,
J. L.
,
Cheifetz
,
S.
and
Massague
,
J.
(
1992
).
Novel activin receptors: distinct genes and alternative mRNA splicing generate a repertoire of serine/threonine kinase receptors
.
Cell
68
,
97
108
.
Asashima
,
M.
and
Grunz
,
H.
(
1983
).
Effects of inducers on inner and outer gastrula ectoderm layers of Xenopus laevis
.
Differentiation
23
,
206
212
.
Cho
,
K. W. Y.
, and
De Robertis
,
E. M.
(
1990
).
Differential activation of Xenopus homeobox genes by mesoderm-inducing growth factors and retinoic acid
.
Genes Dev
.
4
,
1910
1916
.
Cho
,
K. W. Y.
,
Goetz
,
J.
Wright
,
C. V. E.
,
Fritz
,
AM
Hardwicke
,
J.
and
De Robertis
,
E. M.
(
1988
).
Differential utilization of the same reading frame in a Xenopus homeobox gene encodes two related proteins sharing the same DNA-binding specificity
.
EMBOJ
.
7
,
2139
2149
.
Cho
,
K. W. Y.
,
Morita
,
E. A.
,
Wright
,
C. V. E.
and
De Robertis
,
E. M.
(
1991
).
Overexpression of a homeodomain protein confers axis-forming activity to uncommitted Xenopus embryonic cells
.
Cell
65
,
55
64
.
Christian
,
J. L.
,
McMahon
,
J. A.
,
McMahon
,
A. P.
and
Moon
,
R.T.
(
1991
).
Xwnt-8, a Xenopus Wnt-l/int-1-related gene responsive to mesoderm-inducing growth factors, may play a role in ventral mesodermal patterning during embryogenesis
.
Development
111
,
1045
1055
.
Cooke
,
J.
(
1989
).
Mesoderm-inducing factors and Spemann’s organiser phenomenon in amphibian development
.
Development
107
,
229
241
.
Dale
,
L.
and
Slack
,
J. M. W.
(
1987a
).
Fate map of the 32-cell stage of Xenopus laevis
.
Development
99
,
527
551
.
Dale
,
L.
and
Slack
,
J. M. W.
(
1987b
).
Regional specification within the mesoderm of early embryos of Xenopus laevis
.
Development
100
,
279
295
.
Green
,
J. B. A.
and
Smith
,
J. C.
(
1990
).
Graded changes in dose of a Xenopus activin A homologue elicit stepwise transitions in embryonic cell fate
.
Nature
347
,
391
394
.
Gurdon
,
J. B.
(
1976
).
Injected nuclei in frog oocytes: fate, enlargement, and chromatin dispersal
.
J. Embryol. Exp. Morphol
.
36
,
523
540
.
Jones
,
C. M.
,
Lyons
,
K. M.
and
Hogan
,
B. L. M.
(
1991
).
Involvement of Bone Morphogenetic Protein 4 (BMP-4) and Vgr-1 in morphogenesis and neurogenesis in the mouse
.
Development
111
,
531
542
.
Koster
,
M.
,
Plessow
,
S.
,
Clement
,
J. H.
,
Lorenz
,
A.
Tiedemann
,
H.
and
Knochel
,
W.
(
1991
).
Bone morphogenetic protein 4 (BMPt), a member of the TGF-fJ family, in early embryos of Xenopus laevis: analysis of mesoderm inducing activity
.
Meeh. Dev
.
33
,
191
200
Krieg
,
P. A.
and
Melton
,
D. A.
(
1984
).
Functional messenger RNAs are produced by SP6 in vitro transcription of cloned cDNAs
.
Nucl. Acids Res
.
12
,
7057
7070
.
Krieg
,
P. A.
and
Melton
,
D. A.
(
1987
).
In vitro RNA synthesis with SP6 RNA polymerase
.
Meth. Enzymol
.
155
,
397
415
.
Lyons
,
K. M.
,
Jones
,
C. M.
and
Hogan
,
B. L. M.
(
1991
).
The DVR gene family in embryonic development
.
Trends Genet
.
7
,
408
412
.
Lyons
,
K. M.
,
Pelton
,
R. W.
and
Hogan
,
B. L. M.
(
1990
).
Organogenesis and pattern formation in the mouse: RNA distribution patterns suggest a role for Bone Morphogenetic Protein-2A (BMP-2A)
.
Development
109
,
833
844
.
Nieuwkoop
,
P.
and
Faber
,
J.
(
1967
).
Normal Table of Xenopus laevis Daudin, North-Holland
,
Amsterdam
Ruiz i Altaba
,
A.
and
Melton
,
D. A.
(
1989a
).
Bimodal and graded expression of the Xenopus homeobox gene Xhox 3 during embryonic development
Development
106
,
173
183
.
Ruiz i Altaba
,
A.
and
Melton
,
D. A.
(
1989b
).
Involvement of the Xenopus homeobox gene Xhox 3 in pattern formation along the anterior-posterior axis
.
Cell
57
,
317
326
.
Ruiz i Altaba
,
A.
and
Melton
,
D. A.
(
1989c
).
Interaction between peptide growth factors and homeobox genes in the establishment of antero-poste-rior polarity in embryos
.
Nature
341
,
33
38
.
Slack
,
J. M. W.
,
Darlington
,
B. G.
,
Heath
,
J. K.
and
Godsave
,
S. F.
(
1987
).
Mesoderm induction in early Xenopus embryos by heparin-binding growth factors
.
Nature
326
,
197
200
.
Smith
,
J. C.
(
1989
).
InducUon and early amphibian development
.
Current Opinion in Cell Biology
1
,
1061
1070
.
Smith
,
J. C.
,
Price
,
B. M. JM
Green
,
J. B. A.
,
Weigel
,
D.
and
Herrmann
,
B G.
(
1991
).
Expression of a Xenopus homolog of Brachyury (T) is an immediate-early response to mesoderm induction
.
Cell
67
,
79
87
.
Smith
,
W. C.
and
Harland
,
R. M.
(
1991
).
Injected Xwnt-8 RNA acts early in Xenopus embryos to promote formation of a vegetal dorsalizing center
.
Cell
67
,
753
765
.
Sokol
,
SM
Christian
,
J.
,
Moon
,
R. T.
and
Melton
,
D. A.
(
1991
).
Injected Wnt RNA induces a complete body axis in Xenopus embryos
.
Cell
67
,
741
752
.
St Johnston
,
D.
and
Nusslein-Volhard
,
C.
(
1992
).
The origin of pattern and polarity in the Drosophila embryo
.
Cell
68
,
201
219
.
Wright
,
C .V. E.
,
Cho
,
K. W. Y.
,
Hardwicke
,
J.
,
Collins
,
R. H.
and
De Robertis
,
E. M.
(
1989
).
Interference with function of a homeobox gene in Xenopus Embryos produces malformations of the anterior spinal cord
.
Cell
59
93.