Direct and indirect regulation of derrière, a Xenopus mesoderm-inducing factor, by VegT

The mesoderm of the amphibian embryo is formed through an inductive interaction in which cells of the vegetal hemisphere of the late blastula act on overlying equatorial cells (Sive, 1993; Heasman, 1997). The strongest candidates for natural mesoderm-inducing signals include members of the transforming growth factor β (TGFβ) family, and particularly the nodal-related genes Xnr1 and –2 and –4 to –6 and derrière (Jones et al., 1995; Joseph and Melton, 1997; Sun et al., 1999; Takahashi et al., 2000). These genes are expressed at the right times and in the right places in the embryo to act as endogenous inducing factors, and inhibition of their effects, by various means, causes defects in mesodermal and endodermal differentiation (Osada and Wright, 1999; Sun et al., 1999; Agius et al., 2000; Onuma et al., 2002).

Although these factors play important roles in the early embryo, rather little is known about the transcriptional regulation of the nodal-related genes or of derrière. One of the most significant observations in this respect concerns the requirement for VegT in their activation. VegT is a maternally expressed member of the T-box family whose transcripts are restricted to the vegetal hemisphere of the Xenopus egg and early embryo (Lustig et al., 1996; Stennard et al., 1996; Zhang and King, 1996; Horb and Thomsen, 1997). Ablation of maternal VegT transcripts causes loss of endodermal and mesodermal structures, and expression of the nodal-related genes and of derrière does not occur (Zhang et al., 1998; Kofron et al., 1999; Takahashi et al., 2000). VegT is a transcriptional activator, and it is possible that the nodal-related genes and derrière number among its direct targets. However, only Xnr1 has been shown to contain T-box binding sites in its promoter, and these sites appear not to be required for the vegetal expression of a reporter gene driven by the Xnr1 promoter (Hyde and Old, 2000).

We have investigated the transcriptional regulation of derrière. This gene has been little-studied compared with the nodal-related genes, yet is a strong candidate for an endogenous inducing agent: it is expressed in the right cells, inhibition of its function impairs mesoderm development (Sun et al., 1999), and in this paper we show that derrière, unlike, for example, Xnr2, is able to exert long-range effects in the developing embryo. Our results show that derrière is subject to complex regulation involving not only VegT but also members of the TGFβ family and perhaps FGF family members. The combined effects of these different signals, acting in a series of autoregulative loops, may help ensure the rapid activation of mesoderm-inducing agents at the mid-blastula transition.

Xenopus embryos were obtained by artificial fertilisation and maintained in 10% normal amphibian medium (NAM) (Slack, 1984). They were staged according to Nieuwkoop and Faber (Nieuwkoop and Faber, 1975). Embryos at the one- or two-cell stage were injected with RNA dissolved in 10 nl water as described by Smith (Smith, 1993). For animal cap assays, embryos were dissected and cultured in 75% NAM. Dexamethasone was dissolved in ethanol to a concentration of 2 mM and then diluted to a final concentration of 1 μM in 75% NAM containing 0.1% bovine serum albumin. Protein synthesis inhibition experiments were performed as described by Smith et al. (Smith et al., 1991), except that cycloheximide was applied continuously at a concentration of 10 μM. This treatment was sufficient to reduce incorporation of [³⁵S]methionine into isolated animal pole regions by over 90%.

Animal cap recombinants were performed as described previously (Jones et al., 1996). Conjugates were fixed when sibling embryos reached stage 10.5 and expression of Xbra (Smith et al., 1991) was analysed by in situ hybridisation.

pSP64T.VegT-GR was produced by fusing the ligand-binding domain of the human glucocorticoid receptor (hGR) to the C terminus of VegT, thereby creating a construct similar to the previously published Xbra-GR (Tada et al., 1997). Details are available on request. A mouse activin A cDNA (Albano et al., 1993) was cloned into pSP64T to create pSP64T.mactivinA. pCS2-derrière (Sun et al., 1999), pSP64T.Xnr2 (Jones et al., 1995) pΔXAR (Hemmati-Brivanlou and Melton, 1992), pXFD/Xss (Amaya et al., 1991), pd50 (Amaya et al., 1991) and placZ (Kolm et al., 1997) were as described. Capped RNAs were synthesised using SP6 RNA polymerase.

To obtain derrière promoter sequence, a Xenopus genomic library, prepared in the vector λFIXII (Stratagene), was screened using a probe corresponding to the first 356 base pairs of the derrière open reading frame. Restriction digestion and Southern blotting identified a 2 kb XbaI fragment, which was sub-cloned into pBluescriptII (SK) and sequenced. This fragment (d1.2.1) consists of the 247 base pair exon 1 of derrière preceded by 850 base pairs of upstream sequence and 844 base pairs of intron (see Fig. 2A). The transcription start site was mapped by RNAase protection and is located approximately 30 base pairs downstream of a TATA box.

For luciferase assays, the 5′ 850 base pairs of d1.2.1 were amplified by PCR and sub-cloned into the MluI and BglII sites of the vector pGL3-basic (Promega). For transgenesis, the same fragment was sub-cloned into pGL3-GFP, a modified version of pGL3-basic in which the luciferase coding region is replaced by the GFP coding region (Casey et al., 1999). Deletion constructs were produced by PCR. In vitro mutagenesis was carried out by a PCR-based approach. Details are available on request.

Electrophoretic mobility shift assays (EMSAs) were performed as described by Trindade and colleagues (Trindade et al., 1999). Proteins used in EMSAs were produced by in vitro translation of synthetic RNA. pSP64TBX.VegT-HA (Conlon et al., 2001) was linearised with SalI and transcribed with SP6 RNA polymerase and pFTX9-XlFast1 (Howell et al., 1999) was linearised with XbaI and transcribed with T7 RNA polymerase.

Luciferase assays were performed using the Promega Dual-Luciferase assay kit. Embryos were injected with 20 pg pGL3-basic containing the appropriate promoter fragment, 20 pg pRL-SV40/TK, and an appropriate amount of RNA encoding either β-galactosidase or the inducing agent being tested. Animals caps were dissected at stage 8.5 and cultured in 75% NAM for the desired period. They were then suspended in 10 μl of 1× Passive Lysis Buffer per cap and after centrifugation 5 μl was taken for assay. Oocytes were suspended in 20 μl 1× Passive Lysis Buffer per oocyte and 20 μl was taken for assay. All values are expressed as Relative Luciferase Units (Firefly luciferase activity/Renilla luciferase activity).

Transgenic Xenopus embryos were created as described by Sparrow and colleagues (Sparrow et al., 2000), itself a modified version of the original protocol of Kroll and Amaya (Kroll and Amaya, 1996).

In situ hybridisations were carried out essentially as described previously (Harland, 1994), except that BM purple was used as a substrate. derrière (Sun et al., 1999) and VegT (Zhang and King, 1996) probes were as described. A GFP construct (mgfp5) (Zernicka-Goetz et al., 1996) in pBluescriptII(SK) was linearised with NcoI and a Xenopus Brachyury construct (pXT1) (Smith et al., 1991) was linearised with BglII. Both were transcribed with T7 RNA polymerase in the presence of digoxigenin-11-UTP (Roche). β-galactosidase staining was performed as described previously (Kolm and Sive, 1995).

RNA was prepared from pooled animal caps using the acid guanidinium thiocyanate-phenol-chloroform method (Chomczynski and Sacchi, 1987). RNAase protection analysis was carried out essentially as described by Jones and colleagues (Jones et al., 1992), using RNAase T1 alone for all probes. A derrière probe was made by cloning a PCR fragment of derrière (nucleotides 775-939) into pBluescriptII, linearising with NotI and transcribing with T7 RNA polymerase. Probes for Bix4 (Casey et al., 1999) and ornithine decarboxylase (Isaacs et al., 1992) were as described.

derrière is a member of the TGFβ superfamily that is expressed in the vegetal hemisphere of the Xenopus embryo from the late blastula stage. Mis-expression of derrière can induce isolated animal pole regions to form mesoderm, and the factor is therefore a strong candidate for an endogenous mesoderm inducing agent (Sun et al., 1999). Consistent with this idea, derrière is expressed in the same domain as VegT (Fig. 1), a maternal transcript that is necessary both for mesoderm induction and for zygotic activation of derrière.

The phenomenon of mesoderm induction was discovered in an assay in which animal pole tissue was induced to form mesoderm following juxtaposition with vegetal blastomeres (Nieuwkoop, 1969). According to this assay, a mesoderm inducing factor should be capable of acting in a non cell-autonomous fashion. That is, the signal should be able to activate mesoderm-specific gene expression some cell diameters away from its site of production.

The signalling range of derrière was tested by the technique employed by Jones and colleagues (Jones et al., 1996). Animal pole regions derived from embryos injected with RNA encoding derrière were juxtaposed with animal caps dissected from embryos injected with the lineage label fluorescein-lysine-dextran (FLDx). As controls, caps were derived from embryos injected with RNA encoding activin, which is known to exert long-range effects, or with RNA encoding Xnr2, which in this assay acts essentially cell-autonomously (Jones et al., 1996). The conjugates were cultured for 3 hours and then fixed and analysed for expression of Xbra by in situ hybridisation. Fig. 2 shows that derrière, like activin and unlike Xnr2, can activate expression of Xbra in FLDx-labelled cells, indicating that it can exert long-range effects.

Interestingly, the pattern of Xbra activation in derrière-expressing conjugates differs from that in activin-expressing conjugates. In the former, Xbra is activated throughout the derrière-expressing animal cap and appears to spread from there into the FLDx-labelled tissue. In the latter, and as described by others, Xbra is most strongly induced in the FLDx-labelled animal cap, in a domain that presumably corresponds to a particular concentration of activin. We are now investigating whether this difference is due to different diffusion properties of the two factors (see Ohkawara et al., 2002) or to their different concentration-dependent effects. Whatever the explanation, the results show clearly, at least in this over-expression system, that derrière is capable of exerting long-range effects in the Xenopus embryo.

Ablation of mRNA encoding the vegetally localised transcription factor VegT prevents zygotic expression of derrière in the Xenopus embryo, indicating that VegT is essential for expression of derrière during late gastrula stages of development (Kofron et al., 1999). Is derrière a direct target of VegT? To answer this question we constructed a hormone-inducible version of VegT termed VegT-GR, in which the ligand-binding domain of the glucocorticoid receptor is fused to the C terminus of VegT. This construct is inactive unless the steroid hormone dexamethasone (DEX) is added to the embryo culture medium (data not shown). In combination with the protein synthesis inhibitor cycloheximide (CHX), this construct allows one to examine the ability of VegT to activate derrière directly.

RNA encoding VegT-GR was injected into Xenopus embryos at the 1-cell stage. Animal pole regions were then dissected at the late blastula stage, and groups of 15-20 animal caps were incubated in the absence of factors, in DEX or CHX alone, or in both reagents. Derrière proved to be activated by dexamethasone even in the presence of cycloheximide (Fig. 3), indicating that the effects of VegT do not require intervening protein synthesis. This observation is consistent with the suggestion that VegT acts directly to induce expression of derrière. We note, however, that the level of activation of derrière by VegT-GR is reduced by cycloheximide, and this does not occur with the induction of Bix4 (Fig. 3). As we discuss below, one explanation of this observation is that optimal activation of derrière by VegT involves some indirect effects.

As a first step towards confirming the idea that VegT acts directly on the derrière promoter, we isolated a genomic fragment comprising the whole of the first exon of derrière flanked by 851 nucleotides of 5′ sequence and 844 nucleotides of intron 1 (Fig. 4A; GenBank accession number AF527059). Consistent with the idea that VegT activates expression of derrière directly, we noticed two sites resembling a T-box ‘half site’ (Kispert and Herrmann, 1993; Conlon et al., 2001) positioned 90 (Tbs1) and 169 (Tbs2) nucleotides 5′ of the transcription start site (Fig. 3B,C). We also noted the presence of two Fast binding sites at 194 and 209 nucleotides 5′ of the transcription start site.

To determine whether sequence 5′ of the derrière transcription start site can drive correct expression of a reporter gene, we placed this region upstream of green fluorescent protein (GFP) and made transgenic Xenopus embryos (Fig. 5A). Expression of the reporter was not strong enough to be observed by GFP fluorescence, but it was readily detectable by in situ hybridisation using a GFP probe. As is observed with the endogenous gene (Fig. 1A-C), expression of the reporter construct is strongest in the marginal zone, but it is also weakly detectable in the vegetal hemisphere (Table 1; Fig. 5B,C and data not shown). One difference from the endogenous expression pattern, however, is that expression of the transgene persists in involuted mesoderm (Fig. 5D). This might reflect perdurance of GFP RNA; alternatively, our construct may lack an element that is responsible for down-regulation of derrière following involution.

Electrophoretic mobility shift assays were used to test the possibility that VegT regulates expression of derrière through direct interaction with Tbs1 or Tbs2. VegT proved to interact with the Tbs1 site (Fig. 6), but poorly, if at all, with Tbs2 (data not shown). These observations are consistent with the fact that Tbs1 is a better match to the consensus T box binding sequence derived from binding site selection experiments (Kispert and Herrmann, 1993; Conlon et al., 2001) than is Tbs2 (Fig. 4C). Results described below, however, suggest that Tbs2 may nevertheless play a role, albeit minor, in the regulation of derrière.

The results described above suggest that VegT regulates expression of derrière through direct interaction with the its 5′ regulatory region. Consistent with this idea, we observe that VegT can drive expression of a luciferase reporter gene placed downstream of the derrière promoter. Constructs were injected into Xenopus embryos at the 1- to 2-cell stage in the presence or absence of VegT RNA. Animal caps then were dissected at the mid blastula stage and cultured to the early gastrula stage when they were assayed for luciferase activity. VegT caused a 25- to 30-fold induction of luciferase expression (Fig. 7A,B). Xbra and Eomes are also able to activate this construct (data not shown) and therefore may also contribute to derrière regulation, although this has yet to be investigated.

To investigate the roles of Tbs1 and Tbs2 in the activation of derrière reporter constructs, the sequences were mutated so as to disrupt T box binding (Fig. 7A). Single mutations, or mutation of both binding sites, proved to have no effect on the ability of VegT to activate the construct (Fig. 7B). Furthermore, when such constructs were introduced into Xenopus embryos by transgenesis, expression still occurred in endoderm and mesoderm (Table 1; Fig. 7C-F). Thus, although VegT might be capable of activating expression of derrière directly, it also appears to be able to induce transcription indirectly.

VegT activates the expression of the nodal-related genes Xnr1-4 as well as that of derrière itself (Clements et al., 1999; Kofron et al., 1999; Sun et al., 1999; Yasuo and Lemaire, 1999). One possibility, therefore, is that the indirect induction of derrière reporter constructs by VegT occurs through activation of TGFβ family members such as these. Consistent with this idea we have identified two potential Fast sites in the derrière 5′ regulatory region (Fig. 4), and electrophoretic mobility shift assays confirm that these do indeed bind Fast-1 (Fig. 8).

Further experiments demonstrated that the TGFβ family member activin can activate a derrière reporter construct in isolated animal pole regions, and results obtained with a deletion series of the derrière 5′ regulatory region are consistent with the idea that this occurs through the Fast sites (Fig. 9A). We also note, however, that mutation of the Fast sites does not abolish activin responsiveness completely (Fig. 9B), and one possibility is that activation of the reporter construct by TGFβ family members can also occur indirectly, through the induction of T box family members such as Xbra (Smith et al., 1991), eomesodermin (Ryan et al., 1996) and Antipodean (Stennard et al., 1996). Indeed, mutation of the T box sites in addition to the Fast sites prevents activin-induced activation of luciferase activity (Fig. 9B).

The above results suggest that activation of the derrière promoter by VegT can occur indirectly, through the induction of TGFβ family members, which in turn act through the Fast sites we identified (Fig. 4). One prediction of this model is that if one were to analyse reporter gene activation at an earlier stage, before the indirect effects had occurred, mutation of the T box sites should abolish the response to VegT. Another prediction is that if one were to analyse reporter gene activation in oocytes rather than in animal caps, thereby preventing TGFβ-mediated indirect effects, mutation of the T box sites should again abolish the response to VegT. Fig. 10 shows that these predictions are realised. Analysis of reporter gene expression 2 hours after dissection of animal pole regions reveals that mutation of the two T box sites reduces significantly the ability of VegT to induce luciferase activity, while dissection at 3 hours reveals no difference between the wild-type and mutated constructs (Fig. 10A). Similarly, mutation of Tbs1 and Tbs2 prevents the response of a derrière reporter construct to VegT in oocytes (Fig. 10B). Mutation of Tbs1 in this assay has a more dramatic effect than does mutation of Tbs2, suggesting that much of the induction is mediated by Tbs1 (Fig. 10B).

This scheme suggests that the initial activation of derrière occurs in response to VegT alone, and that amplification and maintenance of expression (including activation in the marginal zone) involves signalling by members of the TGFβ family and perhaps other factors (see Clements et al., 1999; Yasuo and Lemaire, 1999). To investigate the importance of such factors in the maintenance of derrière expression during normal development, RNA encoding the dominant-negative activin receptor ΔXAR (Hemmati-Brivanlou and Melton, 1992), or the truncated FGF receptor XFD (Amaya et al., 1991), or the control construct d50 (Amaya et al., 1991) was injected into one blastomere of embryos at the two-cell stage along with RNA encoding β-galactosidase as a lineage marker (Fig. 11). Embryos were fixed at late blastula stage 9.5 or mid-gastrula stage 11 and examined for expression of derrière by in situ hybridisation. The dominant-negative receptors only weakly inhibited activation of derrière at stage 9.5, with expression occurring in 75% of XFD-expressing embryos (n=32) and in 81% of ΔXAR-expressing embryos (n=37; Fig. 11A-C). However, they completely prevented maintenance of expression to stage 11 (n=26 for XFD-expressing embryos and n=19 for ΔXAR-expressing embryos; Fig. 11D-F).

Although it is clear that members of the TGFβ family play important roles in mesoderm induction in the Xenopus embryo, little is known about the different functions of the many TGFβs that are expressed in the early embryo, or about the regulation of their expression. In this paper we turn our attention to derrière. Previous work using a dominant-negative derrière construct has demonstrated that the function of this gene is required for posterior mesoderm formation (Sun et al., 1999), and our own preliminary experiments using an antisense derrière morpholino oligonucleotide are consistent with this suggestion, although the phenotype is milder than is observed with the dominant-negative version of the protein (R. J. W. and J. C. S., data not shown). It is possible that there are additional copies of derrière in the Xenopus genome that are inhibited by the dominant-negative construct but not by the antisense oligonucleotide. Alternatively, the dominant-negative derrière construct may not be completely specific, perhaps inhibiting to some extent the effects of members of the nodal-related family (Sun et al., 1999).

One noteworthy feature of derrière is that it is able to exert long-range effects in our animal cap assay, whereas Xnr2, for example, is not. In this regard, we note that although the N-terminal region of derrière contains a group of four basic amino acids (KKRR), this is not as extensive a domain as the basic region in BMP4 that interacts with heparan sulphate proteoglycans and thereby restricts its signalling range (Ohkawara et al., 2002). The ability of derrière to exert long-range effects in the Xenopus embryo, together with results obtained following inhibition of its activity (see above) mark this protein out as a strong candidate for an endogenous mesoderm-inducing factor, and makes analysis of its regulation all the more significant.

Experiments using a hormone-inducible version of VegT and cycloheximide suggest that derrière is a direct target of VegT (Fig. 3). Our results indicate, however, that cycloheximide does cause some reduction in the level of activation of derrière by VegT-GR, a phenomenon we do not observe with induction of Bix4 (Fig. 3). This reduction is consistent with the conclusion of this paper that the activation of derrière by VegT also involves indirect effects, mediated in part by the induction of TGFβ family members.

The direct action of VegT on the 5′ regulatory region of derrière is likely to occur through the T box sites highlighted in Fig. 4. These sites are required for the early activation of a derrière reporter construct in isolated animal pole regions (Fig. 10A) and for the activation of such constructs in Xenopus oocytes, in which indirect activation does not occur (Fig. 10B). Consistent with binding site selection experiments (Kispert and Herrmann, 1993; Conlon et al., 2001), the site designated Tbs1 (TGACACCT) proved to interact more strongly with VegT in electrophoretic mobility shift assays than did Tbs2 (AGACACCT) (Fig. 6). We note that the nucleotide that differs between the two sites (the first ‘T’ in Tbs1 becomes an ‘A’ in Tbs2) is directly contacted by Xbra in the crystal structure of that protein (Muller and Herrmann, 1997).

Mutation of Tbs1 did not, however, completely prevent reporter gene activation in response to VegT in Xenopus oocytes (Fig. 10B), indicating that Tbs2 may also play a role in the activation of derrière by VegT. It is possible that the two sites cooperate to ensure the normal regulation of the gene, in a manner resembling that suggested to occur in the regulation of Ci-trop in Ciona intestinalis (Di Gregorio and Levine, 1999). This is under investigation.

Although the T box sites in the derrière promoter are required for the early response to VegT and for activation in a system such as the Xenopus oocyte where indirect effects do not occur, they are not required for later responses. Thus, mutation of the T box sites does not prevent the vegetal activation of derrière reporter constructs in transgenic Xenopus embryos (Fig. 7C-F), and nor does it prevent induction by VegT of similar constructs in animal pole regions following culture for 3, rather than 2 hours (Fig. 7B). Our results suggest that the activation of derrière reporter constructs under these conditions occurs indirectly, due, at least in part, to the activation of members of the TGFβ family such as the nodal-related genes and perhaps even derrière itself. Thus activin, a member of the TGFβ family, can induce expression of endogenous derrière and can activate the expression of derrière reporter constructs. This inducing activity is likely to occur through the Fast sites, as shown in Fig. 4; deletion analysis demonstrates that loss of the Fast sites reduces significantly the ability of activin to induce reporter gene expression (Fig. 9A), although there may be indirect effects at work here too, because there remains some residual activity that is substantially abolished by deletion of the T box sites (Fig. 9B). It is possible that while VegT can exert indirect effects through the activation of TGFβ family members, TGFβ family members can exert indirect effects through the activation of T box genes. These might include Xbra (Smith et al., 1991) and eomesodermin (Ryan et al., 1996), as well as Antipodean (Stennard et al., 1996), itself an alternatively spliced isoform of VegT (Stennard et al., 1999).

Investigation of this indirect pathway will require carefully timed experiments that make use of specific inhibitors of particular members of the TGFβ family. Our initial experiments along these lines indicate that a truncated activin receptor does not prevent the initial activation of derrière but does inhibit its maintenance (Fig. 11). Interestingly, a recent report making using of dominant-negative versions of Xnr5 and Xnr6 found no evidence of down-regulation of derrière at any stage, even though the Xnr5 construct also inhibited the functions of Xnr2, Xnr4, Xnr6, derrière itself and BVg1 (Onuma et al., 2002). This suggests that maintenance of derrière expression can occur through Xnr1 or activin signalling.

The data described so far suggest that VegT and members of the TGFβ family participate in a network of autoregulatory loops (see Fig. 12). But VegT and members of TGFβ family are unlikely to be the only members of this network, and we may not have identified all the regulatory elements in the derrière promoter. For example, expression of our reporter construct persists in involuted mesoderm (Fig. 5D); as discussed above, this might reflect perdurance of GFP RNA, but it may be that our constructs lack an element that is responsible for the down regulation of derrière following involution. Furthermore, we find that the late response to VegT, and the expression of reporter constructs in transgenic embryos, is not abolished (and if anything is slightly elevated) even if both the Fast sites and the T-box sites are mutated (data not shown). This suggests that there may be another gene ‘X’ that is regulated by VegT and which acts on the derrière promoter to enhance and maintain its expression (Fig. 12). Candidates for such a gene include members of the FGF family; inhibition of FGF function does not inhibit the initial expression of derrière but does prevent its continued expression (Fig. 11).

Why should derrière be subject to such complicated regulation? One possibility is that a network of autoregulatory interactions of this sort might facilitate the rapid changes in gene expression that occur during early Xenopus development. Another idea is that the network might act to restrict later expression of derrière to the mesoderm rather than to the mesendoderm (Sun et al., 1999). If this were the case, gene X may be represented by eFGF (Isaacs et al., 1992), whose function might be to maintain expression of derrière in the marginal zone.

This work was supported by the Medical Research Council, the Human Frontiers Science Program and the Wellcome Trust. We thank members of our laboratories for their helpful comments and discussion throughout the course of this work. We are also grateful to Caroline Hill for the XlFast-1 construct and for much helpful advice.