Xvent-1 mediates BMP-4-induced suppression of the dorsal-lip-specific early response gene XFD-1' in Xenopus embryos
1998; Springer Nature; Volume: 17; Issue: 8 Linguagem: Inglês
10.1093/emboj/17.8.2298
ISSN1460-2075
Autores Tópico(s)Genomics and Chromatin Dynamics
ResumoArticle15 April 1998free access Xvent-1 mediates BMP-4-induced suppression of the dorsal-lip-specific early response gene XFD-1′ in Xenopus embryos Henner Friedle Henner Friedle Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Sepand Rastegar Sepand Rastegar Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Hubert Paul Hubert Paul Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Eckhard Kaufmann Eckhard Kaufmann Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Walter Knöchel Corresponding Author Walter Knöchel Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Henner Friedle Henner Friedle Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Sepand Rastegar Sepand Rastegar Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Hubert Paul Hubert Paul Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Eckhard Kaufmann Eckhard Kaufmann Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Walter Knöchel Corresponding Author Walter Knöchel Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany Search for more papers by this author Author Information Henner Friedle1, Sepand Rastegar1, Hubert Paul1, Eckhard Kaufmann1 and Walter Knöchel 1 1Abteilung Biochemie, Universität Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany *Corresponding author. E-mail: [email protected] The EMBO Journal (1998)17:2298-2307https://doi.org/10.1093/emboj/17.8.2298 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Ectopic expression of the ventralizing morphogen BMP-4 (bone morphogenetic protein-4) in the dorsal lip (Spemann organizer) of Xenopus embryos blocks transcription of dorsal-lip-specific early response genes. We investigated the molecular mechanism underlying the BMP-4-induced inhibition of the fork head gene XFD-1′. The promoter of this gene contains a BMP-triggered inhibitory element (BIE) which prevents activation of this gene at the ventral/vegetal side of the embryo in vivo. In the present study, we show that BMP-4-induced inhibition is not direct but indirect, and is mediated by Xvent homeobox proteins. Micro-injections of Xvent-1 RNA and XFD-1′ promoter deletion mutants demonstrate that Xvent-1 mimics the effect of BMP-4 signalling not only by suppression of the XFD-1′ gene, but also by utilizing the BIE. Suppression could be reverted using a dominant-negative Xvent-1 mutant. The repressor domain was localized to the N-terminal region of the protein. Gel-shift and footprint analyses prove that Xvent-1 binds to the BIE. Moreover, PCR-based target-site selection for the Xvent-1 homeodomain confirms distinct motifs within the BIE as preferential binding sites. Thus, biological and molecular data suggest that Xvent-1 acts as direct repressor for XFD-1′ transcription and mediates BMP-4-induced inhibition. Introduction Mesoderm formation and dorsal/ventral patterning are crucial events in the embryogenesis of all vertebrate organisms. Numerous studies, mainly performed with amphibian embryos, have shown that different parts of this germ layer are induced by distinct members of different growth factor families, like the FGF, the TGF-β or the Wnt families (for recent reviews, see Kessler and Melton, 1994; Dawid, 1994; Slack, 1994; Smith, 1995; Tiedemann et al., 1996). For example, members of the TGF-β family have been identified in the animal cap assay and/or in vivo to be potent inducers of dorsal or ventral mesoderm formation. The most dorsal part, the dorsal blastopore lip, acquires organizing properties. Spemann and Mangold discovered in the early 1920s that transplantation of this structure into the ventral/vegetal region of a host embryo recruits the neighbouring cells in the organization of a complete secondary body axis (Spemann and Mangold, 1924). Thus, the Spemann organizer had been thought to be a local source of activating signals which led to dorsalization and neural induction. However, in the light of recent findings, this model has been revised (for review, see Graff, 1997). In fact, one of the primary functions of the organizer is to antagonize ventralization. BMP-4 as a ventralizing morphogen is bound and sequestered by organizer-secreted molecules, such as chordin (Piccolo et al., 1996), noggin (Zimmerman et al., 1996) or follistatin (Fainsod et al., 1997); dorsal and neural structures only develop in the absence of BMP-4 signalling. This concept is supported by previous findings that inhibition of BMP signalling by a dominant-negative receptor leads to dorsalized embryos often having a second axis (Graff et al., 1994; Suzuki et al., 1994), whereas ectopic expression of BMP-4 on the dorsal side yields completely ventralized embryos (Dale et al., 1992; Jones et al., 1992). While the lack of dorsal structures implies, and is known to correlate with, a repression of dorsallip-specific early response genes, like goosecoid (Fainsod et al., 1994) or XFD-1/XFD-1′ (Clement et al., 1995; Re'em-Kalma et al., 1995; Jones et al., 1996; Kaufmann et al., 1996), the underlying mechanism is still unknown. An important aspect in understanding the organizer's role is to gain insight into the regulation of genes and genetic cascades which are activated in this specific embryonic structure. Of particular interest are those genes which can be activated in the animal cap assay by the dorsal inducer activin A in the presence of cycloheximide, like goosecoid (Cho et al., 1991) or XFD-1′. XFD-1′/XFD-1 (XFKH1/pintallavis) (Dirksen and Jamrich, 1992; Knöchel et al., 1992; Ruiz i Altaba and Jessell, 1992) are pseudo-allelic fork head/winged helix transcription factors that are specifically activated within the organizer after midblastula transition, the onset of zygotic transcription. As in the case of another activin A-induced gene, the Xmix.2 gene (Huang et al., 1995), the promoters of goosecoid (Watabe et al., 1995) and XFD-1′ (Kaufmann et al., 1996; Howell and Hill, 1997) genes contain distinct activin responsive elements mediating activin signalling. An important finding in these studies was the discovery that particular promoter deletion mutants enabled transcription of reporter genes not only in the dorsal, but also in the ventral/vegetal region of the embryo, which provided an argument for an endogenous inducer being present throughout the vegetal half. However, an analysis of the XFD-1′ promoter has demonstrated the existence of an additional BMP inhibitory element (BIE) located further upstream, which responds to BMP-2/4 signalling and which might be responsible for the suppression of this gene in the ventral/vegetal region in vivo (Kaufmann et al., 1996). The inhibition mediated by the BIE is much more effective than the activation induced by activin A or an activin homologue. This observation is fully compatible with earlier results obtained from animal cap assays suggesting that BMP-4 signalling overrides the inducing activity of activin A (Dale et al., 1992; Jones et al., 1992). Therefore, the local expression of the dorsal-lip-specific XFD-1′ gene is not achieved by direct local activation, but rather results as a consequence of the lack of inhibition by BMP-4. In the present study, we have analysed the underlying mechanism for the inhibition caused by BMP-4. By investigating the regulation of the XFD-1′ gene, we have found that repression is not based upon a direct inhibition by BMP signalling but involves expression of another transcriptional repressor as a mediator. This repressor was found to be Xvent-1 (Gawantka et al., 1995), a member of a recently discovered group of Xenopus homeobox proteins expressed by BMP signalling (Ladher et al., 1996, Onichtchouk et al., 1996; Papalopulu and Kintner 1996; Schmidt et al., 1996; Tidman-Ault et al., 1996). Based upon a common characteristic feature, an isoleucine to threonine exchange within the third helix of the homeodomain, it has been suggested that these proteins are related to the Bar family of homeobox proteins (Papalopulu and Kintner, 1996). Ectopic expression studies by Xvent-1 RNA injections, in combination with the investigation of XFD-1′ promoter deletion mutants, showed the previously identified BIE to be responsible for down-regulation of XFD-1′ transcription in the ventral/vegetal region of the embryo in vivo. Since gel-shift assays and footprint analyses provide direct proof of the binding of Xvent-1 to the BIE, we conclude that BMP-4-induced inhibition of the dorsal-lip-specific early response gene XFD-1′ in the ventral/vegetal region of the embryo is directly caused by the transcriptional repressor Xvent-1. Results BMP-4 signalling does not directly inhibit transcription of XFD-1′ By testing various combinations of activin A, BMP-4 and cycloheximide in the animal cap assay, we could demonstrate that the inhibitory effect of BMP-4 signalling on XFD-1′ gene expression is not direct but indirect and requires de novo protein synthesis. It is known that transcriptional activation of XFD-1′ by activin A is independent of cycloheximide (Dirksen and Jamrich, 1992; Ruiz i Altaba and Jessell, 1992), but is strongly reduced in the presence of BMP-4 (Clement et al., 1995; Re'em-Kalma et al., 1995; Jones et al., 1996; Kaufmann et al., 1996). As shown in Figure 1, in the presence of cycloheximide, BMP-4 is no longer able to override activin A. Interestingly, treatment of caps with cycloheximide alone results in a slight autoinduction; moreover, incubation with activin A and cycloheximide leads to a superinduction of XFD-1′ transcription. Similar results have previously been achieved with the goosecoid (Tadano et al., 1993) and the Xnot gene (von Dassow et al., 1993) which might reflect a common feature for certain dorsal-lip-specific genes. However, this effect does not interfere with our finding that cycloheximide prevents the BMP-4-induced inhibition of XFD-1′ transcription. Therefore, we conclude that BMP signalling in vivo leads to the expression of another factor which, as a transcriptional suppressor, subsequently down-regulates the XFD-1′ gene. Such intermediate synthesis gives rise to the question of whether there might be a temporal delay for the inhibitory mechanism in vivo, allowing an initial transcriptional activation of the XFD-1′ gene. Transient activation of the goosecoid gene after midblastula transition in BMP-4 RNA injected embryos has recently been described (Jones et al., 1996). However, whole mount in situ hybridizations for XFD-1′ transcripts in wild-type and BMP-4 RNA injected embryos from blastula stage until the end of gastrulation (Figure 2A) revealed no evidence for such a delay in the inhibition of XFD-1′ transcription. From blastula stage until the end of gastrulation, we were unable to detect XFD-1′ transcripts in BMP-4 RNA injected embryos (the staining observed in some embryos most probably reflects inefficient translation of BMP-4 RNA). This suggests that the inhibitor is rapidly expressed after midblastula transition and is present in sufficient amounts to repress XFD-1′ gene activation. Figure 1.RT–PCR of XFD-1′ transcripts in animal caps. RNA was extracted from animal caps, which had been treated with growth factors in the absence or presence of cycloheximide, and assayed for XFD-1′ transcripts by RT–PCR. Transcription of the ubiquitous histone H4 gene served as an internal reference. From left to right: P, products of the RT–PCR in the presence of primers but in the absence of template RNA; C, RT–PCR of RNA from untreated caps; the remaining lanes show transcription of XFD-1′ and histone H4 in the presence of activin A, BMP and cycloheximide in the indicated combinations. Download figure Download PowerPoint Figure 2.Transcription of the XFD-1′ gene. (A) BMP-4 inhibits transcriptional activation of XFD-1′. 1 ng BMP-4 mRNA was injected into both dorsal blastomeres of four-cell-stage Xenopus embryos. Injected embryos (lower) and uninjected controls (upper) were fixed at various developmental stages (from left to right: early gastrula, middle/late gastrula, early neurula) and submitted to whole mount in situ hybridization for XFD-1′ transcripts. Note that injected embryos show a slightly retarded development as compared with uninjected controls. (B) XFD-1′ transcription is blocked by BMP-4, Smad1 and Xvent-1. Four-cell-stage embryos were injected into the marginal zone of both dorsal blastomeres with 1 ng BMP-4, Smad1 or Xvent-1 mRNA. Whole mount in situ hybridizations for Xvent-1 (top) and XFD-1′ (middle) were carried out at gastrula stage. The resulting phenotype (bottom) is shown for embryos at stage 28. Download figure Download PowerPoint Smad1 and Xvent-1 mimic BMP-4-induced inhibition of the XFD-1′ gene In search of candidate factors acting as repressors, we decided to investigate whether Xvent homeobox proteins could serve as such mediators for the following reasons: (i) the previously identified BIE sequence contains nucleotide motifs similar to homeobox binding sites already described, it contains a canonical Oct-1 target site (Kaufmann et al., 1996); (ii) Xvents are downstream of BMP-4, Xvent proteins mimic BMP-4 signalling and can rescue the phenotype created by dominant-negative BMP-4 receptor; (iii) Xvent genes are directly activated by BMP-4, transcripts co-localize on the ventral side of the embryo; and (iv) ectopic Xvent expression leads to a down-regulation of goosecoid (for review, see Lemaire, 1996). Thus, this newly described subclass of homeobox proteins fulfils all requirements with respect to their spatial expression and biological function to serve as mediators of BMP-4 action. Indeed, micro-injection experiments performed with BMP-4, Smad1 as a transducer of BMP signalling (Graff et al., 1996; Thomsen, 1996) or Xvent-1 RNA revealed that ectopic overexpression of BMP-4 or Smad1 activates Xvent-1, and that all injections, including that of Xvent-1 RNA, lead to an inhibition of XFD-1′ transcription (Figure 2B). The results suggest that suppression of the dorsal-lip-specific XFD-1′ gene in the ventral/vegetal region in vivo is mediated by the following cascade: BMP-4 signalling including the signal transducer Smad1 activates Xvent-1 which subsequently represses XFD-1′. Xvent-1 as XFD-1′ repressor requires the BMP-triggered inhibitory element (BIE) To investigate whether this inhibition correlates with an interaction with the previously identified BIE, which has been mapped between positions −289 and −188 upstream of the transcription start site of the XFD-1′ gene (Kaufmann et al., 1996), we tested various promoter deletion mutants fused to the luciferase reporter gene for their ability to be repressed upon coinjection with Xvent-1 RNA into the dorsal blastomeres of four-cell-stage embryos. Figure 3A demonstrates that reporter gene activities of all 5′-deletions down to position −233 are strongly inhibited by Xvent-1 injection, whereas more extensive deletions become refractory and the −180 deletion mutant regains full activity. This implies that Xvent proteins act as repressors on the BIE which, with consideration of our previous findings (see above), can now be narrowed to the region between −233 and −188. We have also investigated the localization of the repressor domain within the Xvent-1 protein. Co-injections of the −257 promoter/luciferase construct with Xvent-1 RNA mutants containing the DNA binding domain, but deleted at either the N- or at the C-terminal coding regions, revealed that the repressor function requires the N-terminus of the protein (Figure 3B). In line with this finding, injection of the ΔC mutant containing the coding region of the N-terminus led to ventralized embryos, whereas injection of the ΔN mutant did not alter the normal phenotype (data not shown). It may be argued that failure of the ΔN mutant to inhibit reporter gene activity does not result from the lack of a repressor domain but is due to inefficient translation or removal of a nuclear localization signal (NLS). Therefore, we have fused the ΔN mutant to a myc-tag and an NLS-containing vector (Rupp et al., 1994). Following injection of this RNA, the corresponding protein could be visualized on Western blots of embryonic protein extracts by immunostaining with anti-myc antibodies (data not shown). Also, co-injection of the mutant RNA with the −257 promoter/luciferase construct did not lead to a significant reduction of reporter gene activity (Figure 3C). We conclude, therefore, that the N-terminal domain is required for repressor function. Figure 3.Xvent-1 down-regulates the XFD-1′ promoter in vivo. (A) Four-cell-stage embryos were injected into the marginal zone of both dorsal blastomeres with 20 pg of indicated XFD-1′ promoter deletion/luciferase constructs, or co-injected with 0.5 ng Xvent-1 mRNA. Luciferase activities were determined at stage 11. Results obtained from DNA injections in the absence of Xvent-1 mRNA were set as 100%. (B) 0.3 ng of Xvent-1 RNA or indicated Xvent-1 RNA mutants (ΔC: truncated at the 3′ end; ΔN: truncated at the 5′ end; homeobox: truncated at the 5′ and the 3′ ends) were co-injected with 20 pg of the −257 promoter deletion mutant/luciferase construct (control) into both dorsal blastomeres of four-cell-stage embryos. The luciferase activity was determined at stage 11. (C) 20 pg of −257 promoter deletion/luciferase DNA (control) were co-injected with 0.2 ng ΔN Xvent-1/myc, 0.3 ng Xvent-1 or indicated amounts of VP16/Xvent-1 RNAs, respectively, into both dorsal blastomeres of four-cell-stage embryos. Note that reversal of promoter activity, as shown in the last column, was achieved by co-injection with 5 pg VP16/Xvent-1 RNA. The luciferase activity was determined at stage 11. Download figure Download PowerPoint XFD-1′ activity is rescued by a dominant-negative Xvent-1 mutant Initial loss-of-function studies, i.e. to rescue for normal phenotypes after ventralization and to revert inhibition of XFD-1′ activity, were carried out with the myc-tagged ΔN mutant. While this mutant led to neither ventralization nor to a significant inhibition of the −257 promoter/reporter construct, it was rather inefficient in competing for the ventralizing and inhibitory effect of the wild-type protein. Upon co-injection of Xvent-1 and ΔN mutant RNAs, we observed neither the rescue of ventralized phenotypes, nor a reversal of promoter activity of the XFD-1′ gene (Figure 3C). Also, at higher ratios of mutant to wild type Xvent-1, we did not observe a significant increase of reporter gene activity (data not shown). We therefore used a recently described dominant-negative Xvent-1 mutant in which the N-terminal domain had been replaced by the VP16 activating domain (Onichtchouk et al., 1998). This VP16/Xvent-1 mutant exhibits dorsalizing activity when injected at the ventral side and is able to rescue for normal phenotypes after ventralization by Xvent-1. We show here that this mutant is also able to revert the inhibitory effect of wild-type Xvent-1 on the BIE and to restore reporter gene activity to normal values (Figure 3C). Additionally, upon injection of this mutant we observe a significant enhancement of XFD-1′ promoter activity above normal levels in a dose-dependant manner, which can readily be explained by replacement of the repressor to a strong activating domain. Thus, results obtained with the dominant-negative VP16/Xvent-1 mutant support our notion that Xvent-1 acts on the BIE and regulates XFD-1′ activity. Xvent-1/BIE complex formation All results presented so far do not exclude the possibility that Xvent-1 might only act as an intermediate factor, being required for the activation of another gene whose product would bind to the BIE. Therefore, we have checked the BIE-binding properties of Xvent-1 by gel-shift and footprint analyses. First, gel-shift analysis shows that bacterially expressed Xvent-1 homeodomain binds to the labelled BIE fragment in a concentration-dependent manner, with higher protein concentrations leading to the formation of oligomers (Figure 4A). This binding is independent of the presence of an unspecific competitor DNA, but is significantly reduced by the addition of unlabelled BIE as specific competitor. Further fragmentation of the BIE localized the binding site between positions −233 and −188 (Figure 4B). Figure 4.Xvent-1 homeodomain binds to the BIE. (A) Mobility shift assays were performed with bacterially expressed Xvent-1 homeodomain and labelled −256/−188 BIE fragment of the XFD-1′ promoter. Lane a shows free DNA, while the amounts of protein added to the other reactions were 50 ng (lane b), 100 ng (lane c), 200 ng (lane d) and 400 ng (lanes e–i) (denoted by the black bar at the top of the figure). White triangles denote addition of unlabelled, unspecific competitor DNA (lanes f and g) and unlabelled BIE as specific competitor (lanes h and i) at 10- or 100-fold excess, respectively. (B) The binding site of Xvent-1 homeodomain protein is located between −233 and −188 of the XFD-1′ promoter. The scheme shows the results from four mobility shift assays carried out with promoter fragments as shown; (+) indicates a shift with Xvent-1 homeodomain, (−) no shift detected. Download figure Download PowerPoint This result is strongly supported by DNase I footprint experiments which were performed on both strands (Figure 5). For each strand, we can demonstrate two protected regions of different lengths (PR-1 and PR-2). The positions of these regions overlap for both strands, and are located between −232 and −211 (PR-1), and −207 and −200 (PR-2). Nucleotide sequence comparison between PR-1 and PR-2 revealed a common 5′-CTATT(T/C)G-3′ consensus motif (reverse complementary in PR-2). Interestingly, this motif is part of a canonical Oct-1 target site (5′-ATTTGCAT-3′; Singh et al., 1986) present within PR-1, and displays a distant relationship to homeobox target sites (Locker, 1996) described previously. However, as noted above, Xvent proteins are related to the Bar family of homeobox proteins, which are characterized by an isoleucine to threonine exchange within the third helix. Therefore, they might exhibit different target site specificity and bind to related, but not to identical, sequence motifs as has been found for other classes of homeobox proteins. Figure 5.DNase I footprinting specifies DNA-binding sites of Xvent-1. Both strands of the BIE (−256/−188) were used for DNase I protection assays. 3′-end labelled templates were incubated with increasing amounts (10, 20 and 40 ng; see triangle) of Xvent-1 homeodomain and subsequently digested with DNase I. G/A: Maxam–Gilbert reaction specific for dG and dA residues. Sequences and protected regions of the sense and antisense strands are shown on the left or, as a double strand, at the bottom. Protected regions (PR-1 and PR-2) are denoted by lines. Download figure Download PowerPoint Xvent-1 target site selection To gain additional information on these binding motifs, we have used the bacterially expressed Xvent-1 homeodomain to perform a PCR-based target site selection from a mixture of 18-fold degenerated oligonucleotides. After seven rounds of selection/amplification, the targets were cloned. A computer-aided comparison of 70 resulting sequences with the PR-1 and PR-2 sequences led to the finding that 95% of all target sequences can be aligned to three distinct types of motifs, all being present within PR-1. Figure 6 shows a selection of 15 sequences. A major group (∼60%) is characterized by the consensus 5′-CTATTTG-3′, which had already been detected in footprint studies (see above), a second group (25%) by 5′-TGCATTTTG-3′ and a minor group (10%) by 5′-TTGATC-3′. Although we are aware that these sites share an overlap of three or four nucleotides, we cannot yet explain why these different types of targets are selected. Gel-shift assays with randomly selected targets from each of the three groups revealed highest affinities for targets of the major group (data not shown). A more detailed characterization of individual sequences for their binding properties, including determination of KD values, and mutational analyses are currently under investigation. However, the fact that the major group of selected sequences shares extensive homologies with the PR-1, as well as with the PR-2 motifs, strongly suggests that these sites represent a preferential target site for the Xvent-1 homeodomain and thereby also supports our findings from the footprint experiments. Figure 6.PCR-based target site selection. A comparison of five consecutive matches between protected regions from the footprint (PR-1 and PR-2) and 15 sequences (from a total of 70) obtained by target site selection after seven rounds of amplification is shown. Based upon their homology to the PR-1 sequence, the targets are subdivided into three categories. 60% of all 70 sequences fall into the first category (consensus: 5′-CTATTTG-3′), 25% into the second (consensus: 5′-TGCATTTTG-3′) and 10% into the third category (consensus: 5′-TTGATC-3′). Note that the PR-2 motif matches the first category. Download figure Download PowerPoint Discussion The ventralizing morphogen BMP-4 is regarded as a key molecule in patterning the mesoderm of vertebrate embryos. The molecular basis for the dorsalizing and neuralizing activities of the organizer are secreted proteins, such as chordin (Piccolo et al., 1996), noggin (Zimmerman et al., 1996) and follistatin (Fainsod et al., 1997), which bind and sequester the BMP-4 protein. Expression of these antagonizing proteins leads to an elimination of BMP-4 signalling at the dorsal side of the gastrula stage embryo which, in turn, is a pre-requisite for formation of dorsal and neural structures. A primary step in organizer formation is the activation of early response genes, which are induced by activin or activin-like signalling even in the absence of de novo protein synthesis. Distinct but different activin responsive elements have been described for the Xmix.2, goosecoid, XFD-1′ and Xlim-1 genes (Huang et al., 1995; Watabe et al., 1995; Kaufmann et al., 1996; Howell and Hill, 1997; Rebbert and Dawid, 1997). Remarkably, certain promoter deletions of goosecoid and XFD-1′ genes render expression of reporter genes also in the ventral/vegetal region of the embryo, which provides an argument for an ubiquitous distribution of dorsalizing activity within the vegetal half. Thus, there is demand for a mechanism which prevents transcription of dorsal-lip-specific genes at the ventral side of the embryo in vivo. Since BMP-2/4 are known to inhibit transcription of XFD-1/XFD-1′ (Clement et al., 1995; Re'em-Kalma et al., 1995; Jones et al., 1996), we have focused our interest on the molecular mechanism underlying the inhibitory action of BMPs on dorsal-lip-specific genes. Our hypothesis of BMPs serving as natural inhibitors is also based upon the distribution of BMP-4 transcripts in the ventral/lateral mesoderm and their absence from the dorsal lip (Fainsod et al., 1994; Hemmati-Brivanlou and Thomsen, 1995; Schmidt et al., 1995). In a previous study, we have shown that the XFD-1′ promoter contains a BMP-4-triggered inhibitory element (BIE) leading to a ventral inactivation of reporter genes in whole embryos (Kaufmann et al., 1996), but it was uncertain whether this element directly responded to BMP signalling. From the results of our present study, it is now clear that inhibition by BMP-4 signalling is not direct but indirect, and that Xvent homeobox proteins serve as mediators in the BMP-4-induced inhibition of the dorsal-lip-specific XFD-1′ gene. Based upon co-injection of Xvent-1 mRNA and XFD-1′ promoter deletion mutants, we can show that the BIE is required for down-regulation of reporter gene activity. Furthermore, gel-shift and footprint experiments performed in vitro provide evidence of the physical interaction between the Xvent-1 homeodomain and distinct sequence motifs within the BIE. These findings are further supported by results from PCR-based target site selection. Although it is known that homeodomains bind DNA in vitro with relatively low sequence specificity, and this approach in the case of the Xvent-1 homeodomain also did not reveal a strictly conserved target motif, a major group of binding sequences can be aligned to the motif 5′-CTATT(TC)G-3′, also present within the BIE (PR-1 and PR-2). Thus, all the experimental evidence suggests that Xvent-1 serves as an intermediate and as a transcriptional repressor in BMP-4-induced inhibition of XFD-1′ gene activation. The repressor function of Xvent-1 resides in the N-terminal part of the protein. While a mutant lacking the N-terminus did not suppress −257 promoter/reporter activity, it was not, on the other hand, sufficient by itself to compensate for the effects obtained with wild-type Xvent-1. However, ventralized phenotypes have recently been reported to be rescued using a dominant-negative mutant in which the N-terminus had been replaced by the VP16 activating domain (Oni
Referência(s)