The role of Mixer in patterning the early Xenopusembryo

In vertebrate embryos, the spatially restricted expression of specific zygotic genes presages the movements of gastrulation and the establishment of the three germ layers. In Xenopus, the first step in this process occurs at the mid-blastula transition (MBT), when the vegetally localized T-box transcription factor VegT activates the expression of transcription factors and TGFβ signaling molecules in vegetal cells(Kofron et al., 1999; Xanthos et al., 2002; Xanthos et al., 2001; Zhang et al., 1996; Zhang et al.,1998). Within 2 hours of MBT, VegT activates members of the GATA,homeobox and HMG box families of transcription factors, as well as FGF and TGFβ class signaling molecules. As a result, the equatorial region of the late blastula stage embryo becomes specified toward mesodermal fates, while vegetal cells become specified for endodermal fates(Heasman et al., 1984; Nieuwkoop, 1969; Smith et al., 1985). One question that remains to be understood about this process is how the boundary between prospective mesoderm and endoderm is established and maintained.

The homeodomain transcription factor Mixer is a good candidate for a role in boundary formation in the Xenopus gastrula(Henry and Melton, 1998). Its expression is dependent on maternal VegT and nodal related proteins and is confined to an 8-hour period during gastrulation(Henry and Melton, 1998; Xanthos et al., 2002; Xanthos et al., 2001). Mixer mRNA is expressed in the prospective endoderm cells of the vegetal hemisphere in the early gastrula stage embryo, but is particularly concentrated at the boundary between the equatorial, prospective mesoderm and the vegetal, prospective endoderm regions(Engleka et al., 2001; Henry and Melton, 1998). Previous functional studies show that Mixer is required for endoderm development. Overexpression in animal caps leads to the ectopic induction of endodermal molecular markers including endodermin, cerberus, DKK-1and Xsox17 and their expression is blocked by a dominant inhibitory form of Mixer, Mixer-ENR (Henry and Melton, 1998). Mixer-ENR mRNA-injected embryos develop with gastrulation defects, as well as gut and head abnormalities, and vegetal explants from such embryos lack the expression of late endodermal markers IFABP and Xlhbox8(Henry and Melton, 1998). However, the expression of Mixer mRNA in maternal VegT-depleted embryos rescues the expression of Xsox17 to only a limited extent,suggesting that its inducing activity relies on other VegT-dependent co-factors (Xanthos et al.,2001). Mixer interacts with phospho-Smad2 through a Smad interaction motif and acts as a transducer of Xnr signals(Germain et al., 2000).

There are six other Paired-like homeobox family members expressed at the same time as Mixer in Xenopus gastrulae, including Mix.1, the founding member, together with Mix.2,Bix1/Mix4, Bix2, Bix3/Milk and Bix4(Casey et al., 1999; Ecochard et al., 1998; Henry and Melton, 1998; Latinkic and Smith, 1999; Mead et al., 1996; Rosa, 1989). All have roles in endoderm formation and gastrulation, and Mix.1 can also repress the mesodermal marker Xbra (Latinkic and Smith, 1999). Two zebrafish Mix.1-related genes, bon and mezzo have been described(Kikuchi et al., 2000; Poulain and Lepage, 2002). Although neither are homologs of Xenopus Mixer, they have been shown to act redundantly with each other in the formation of endoderm, as well as in prechordal plate mesoderm formation(Poulain and Lepage, 2002). Recently, Bon was also demonstrated to function in precursors of the axial mesoderm to regulate anterior neural patterning(Trinh et al., 2003). In mice and humans, only one family member, Mixl1 has been identified(Pearce and Evans, 1999; Robb et al., 2000). Mouse Mixl1^-/- mutants have defects in gastrulation and axial mesoderm patterning, including increased expression of the mesodermal gene Brachyury, and of the nodal signaling molecule(Hart et al., 2002). Relatively normal early expression of endodermal markers Sox17 and Cer1occurs, but the embryos die at the early somite stage (8.5 dpc), without forming a heart or gut tube (Hart et al.,2002).

To establish the roles of Mixer in endoderm and mesoderm specification in Xenopus, we used antisense morpholino oligos(MixerMO) to block the translation of Mixer protein throughout the entire period of Mixer expression during gastrulation. We find that Mixer-depleted embryos develop with severe abnormalities of the head and gut,which are partially rescued by the expression of a non-complementary Mixer mRNA. The pattern of early zygotic gene expression is reproducibly altered at the gastrula stage, and the effects of Mixer-depletion are gene specific rather than germ layer specific. We show that Mixerloss-of-function results in overexpression of the mesodermal markers eomesodermin, Bix3 and Fgf8 in their usual equatorial location, as well as the spread of their expression into deeper endodermal territory. We also show that Mixer acts cell autonomously, and represses some genes (Bix3, Xnr5, Xnr1 and Fgf8) while activating others(cerberus, Xsox17). In functional assays of Mixer-depleted vegetal cells, we show that a major biological role of Mixer is to control the degree to which cells induce the formation of mesoderm.

Eggs were obtained by injecting Xenopus laevis with 1000 U of chorionic gonadotropin (Sigma Aldrich, Milwaukee, WI), and fertilized using a sperm suspension in 1×MMR. Embryos were maintained in 0.1×Marc's Modified Ringer's Solution (MMR), and dejellied using 2% cysteine (Sigma Aldrich, Milwaukee, WI) at pH 7.8. Staging was according to Nieuwkoop and Faber (Nieuwkoop and Faber,1967). For injections of morpholino oligos or mRNA, embryos were transferred to 2% Ficoll (Amersham Bioscience, Uppsala, Sweden) in 0.5×MMR, and then maintained in 0.1×MMR after the blastula stage. The sites of MO and mRNA injection are described in the text.

Mid-blastula (stage 8) wild-type and Mixer MO-injected embryos were devitellined and dissected with tungsten needles on agar coated dishes in 1×MMR. After washing away dead cells, vegetal and animal pieces were placed alone (Fig. 4A) or together in the combinations described in Fig. 6 and cultured on agar in OCM for 2 hours. The recombinants were separated using tungsten needles and stray vegetal cells were identified by their larger size and whitish, opaque color and removed from the explants. Animal caps were then cultured in fresh OCM on agar until sibling uninjected embryos were stage 11 and frozen in batches of 10 caps for analysis

Vegetal cells were transplanted to host blastulae as described previously(Heasman, 1984). Briefly,vegetal masses were dissected from late-blastula stage embryos, dissociated in 67 mM phosphate buffer, washed once with 1×Ca/Mg-free MMR, transferred to Ca/Mg-free 1×MMR with 10 μg/ml TRITC (Sigma T-3163) and labeled for 5 minutes. Cells were washed twice in 1×Ca/Mg-free MMR and transplanted to recipient blastulae. Embryos were cultured to the tailbud stage and were fixed in 2% TCA for 2 hours at room temperature. Embryos were embedded in wax, mounted, and cut into 20 μm sections and mounted on positively charged slides. Slides were dewaxed and stained with 12/101 antibody at 1:500 dilution in PBS/Tween 0.1% for 3 hours at room temperature. Slides were washed three times for 20 minutes at room temperature. Secondary antibody was 1 hour 1:250 GAM Cy5 conjugate (Jackson Immuno). Slides were washed, mounted in 90% glycerol and imaged on LSM 510 confocal microscope(Zeiss).

Three overlapping morpholino oligos were designed complementary to the junction between the Mixer 5′UTR and the open reading frame. Fig. 1A shows the reverse complement of each oligo sequence aligned with Mixer, a Mixer pseudo-allele from the EST sequencing project, the related genes Mix.1 and Mix.2, and a morpholino-resistant mRNA(Mixer MO-R) designed for the rescue experiments. Mixer MO-1 was most efficient in blocking the in vitro translation of Mixer mRNA, but did not block translation of Mix.1 mRNA(Fig. 1B and data not shown). However, it also caused cell disaggregation and apoptosis at the late gastrula stage in doses above 15 ng. To determine whether this was a specific effect,the oligo was injected into the four animal blastomeres of eight-cell stage embryos to target cells that do not express Mixer mRNA. MO-1 caused disaggregation and apoptosis of epidermal cells at the neurula stage, as evidenced by TUNEL staining, indicating that disaggregation was not due to blocking Mixer translation (data not shown). MO-2 was shifted eight nucleotides 5′ of the region targeted by Mixer MO-1 and caused more severe apoptosis beginning at gastrula stage. Mixer MO-3 targeted a region four nucleotides 3′ of Mixer MO-1. MO-3 caused no disaggregation effects (data not shown) and blocked translation of Mixer mRNA, although with less efficiency than MO-1(Fig. 1B). We confirmed by in vitro translation that neither oligo blocked the translation of the most closely related family member Mix.1 (data not shown). The experiments described here were performed using both MO-1 and MO-3 oligos.

A morpholino resistant mRNA (Mixer MO-R mRNA) was synthesized by PCR amplification of the Mixer-coding region using a 5′ primer designed to introduce conserved substitutions into this sequence. The PCR amplification was performed using the proofreading Advantage-2 HF polymerase(Clontech). The PCR amplified product was subcloned into pCS2+ and was sequenced to ensure no PCR induced mutations were introduced. These base changes introduce 10 mismatches for Mixer MO-1, 18 mismatches for Mixer MO-2, and six mismatches for Mixer MO-3.

Wild-type or Mixer MO-R mRNA was used in a Biotin in vitro Translation kit (Roche). Reactions were assembled on ice and then incubated at 30°C for 1 hour. Aliquots of the reactions were separated in a 10%Tris-HCl electrophoresis gel and blotted to PVDF. The membranes were blocked then incubated with streptavidin-POD and developed with Chemiluminescence substrate (Roche). Membranes were then exposed to X-ray film.

Whole-mount in situ hybridization was performed on pigmented or albino embryos as described (Harland,1991) using BM Purple as substrate (Roche). After satisfactory color development, embryos were fixed in MEMFA for 1 hour at room temperature,washed and stored in 100% methanol. Embryos for half-mount in situ hybridization were prepared by fixing whole gastrulae for 1 hour in MEMFA,bisecting the embryos along the dorsoventral axis with a scalpel blade, fixing 1 additional hour in MEMFA, washing and storing in 100% ethanol. The in situ hybridization were performed exactly as above with the exception of reducing the proteinase K treatment to 5 μg/ml for 10 minutes to reduce damage to embryos.

Total RNA was prepared from oocytes, embryos and explants using proteinase K and then treated with RNase-free DNase as described(Zhang et al., 1998). Approximately 0.5 μg RNA was used for cDNA synthesis with oligo (dT)primers followed by real-time RT-PCR and quantitation using the `LightCycler'System (Roche) as described previously(Kofron et al., 1999). The primers and cycling conditions used are listed in Table 1. Relative expression values were calculated by comparison to a standard curve generated by serial dilution of uninjected control cDNA. Samples were normalized to levels of ornithine decarboxylase (ODC), which was used as a loading control. Samples of water alone or controls lacking reverse transcriptase in the cDNA synthesis reaction failed to give specific products in all cases.

To block the translation of endogenous Mixer mRNA, which is expressed in the vegetal hemisphere of embryos at the gastrula stage, Mixer MO-1 or MO-3 was injected into the vegetal region of two-cell stage embryos. Embryos injected with Mixer MO-1 or MO-3 developed normally to the gastrula stage. There was a dose-dependent delay in the appearance of the blastopore and gastrulation was slower than in controls,although neural folds eventually formed normally(Fig. 2A,B). At the tailbud stage, embryos failed to elongate, had curved body axes and reduced head size compared with uninjected controls (Fig. 2D; Table 2). These effects were due specifically to the depletion of Mixer mRNA because they were significantly rescued by the introduction of Mixer MO-R mRNA at the eight-cell stage (Fig. 2C,D; Table 2). These results show that Mixer has a non-redundant function in early development in Xenopus.

As Mixer mRNA is particularly concentrated at the boundary between mesoderm and endoderm at the gastrula stage(Henry and Melton, 1998), we asked whether its depletion affected the expression of early zygotic genes expressed in this region. Uninjected control and Mixer-depleted embryos were frozen at the early and mid-gastrula stages and the mRNA expression levels of early zygotic genes were compared using real-time RT-PCR(Fig. 3A; data not shown).

Zygotic gene expression was altered in a highly reproducible fashion in these embryos (7/7 experiments). Molecular markers could be placed into three categories according to whether they were downregulated, upregulated or not altered by Mixer depletion. The downregulated category included the general endodermal markers Xsox 17, Gata5, endodermin(Hudson et al., 1997; Sasai et al., 1994; Weber et al., 2000), the anterior endodermal marker cerberus(Bouwmeester et al., 1996), and the mesodermal markers, Xbra, antipodean, eFGF(Slack, 1994; Smith et al., 1991; Stennard et al., 1999). Surprisingly, several mesodermal markers that are normally expressed in the same domain as Xbra, including eomesodermin, Fgf3, Fgf8,Xnot and Gata2 (Ryan et al.,1996; Christen and Slack,1997; Lombardo et al.,1998; Yasuo and Lemaire,2001; Zon et al.,1991) were upregulated by Mixer depletion. Similarly, the homeobox genes Bix1 and Bix4(Casey et al., 1999; Tada et al., 1998), which are normally expressed in both prospective mesoderm and endoderm, were overexpressed in Mixer-depleted embryos, as were the nodal related genes Xnr1 and Xnr5 (Jones et al., 1995; Takahashi et al.,2000). By comparison, the expression of the anterior endodermal gene Xlim1 was not affected by Mixer depletion(Fig. 3A). This indicates that Mixer regulates the level of expression of many zygotic genes in a complex fashion, and is not simply an endodermal determinant.

To confirm that these changes in gene expression were specific, we analyzed the effects of the introduction of the non-complementary Mixer mRNA(Mixer MO-R mRNA) into Mixer-depleted embryos. Fig. 3B shows that both upregulation and downregulation of target genes caused by the Mixer depletion were rescued by subsequent Mixer MO-R mRNA injection into the four vegetal blastomeres at the eight-cell stage.

As several mesodermal genes were upregulated in Mixer-depleted embryos,this suggested that a normal function of Mixer may be to keep these genes off in the developing endoderm. To test this, control and Mixer-depleted embryos were dissected at the mid-blastula stage into animal, equatorial and vegetal parts, which were cultured separately until the mid-gastrula stage and analyzed for the expression of eomesodermin, Fgf8, Bix1, Bix4 and Xsox17. Fig. 4A shows that Mixer depletion caused an increase in the expression of the mesodermal markers in vegetal explants, indicating that Mixer normally inhibits the ectopic expression of these mesodermal genes. By contrast, the amount of expression of Xsox17 is reduced in vegetal explants depleted of Mixer compared to control expression levels. This suggests that a normal role of Mixer is to repress mesodermal genes and activate endodermal genes in vegetal cells.

In order to study the effect of depleting Mixer on the regional expression of mesodermal and endodermal genes, wild-type and Mixer-depleted embryos were fixed and bisected at the early and mid-gastrula stages and the location of eomesodermin, Bix3, Fgf8, Xbra and cerberus mRNA was examined by in situ hybridization (Fig. 4B and data not shown). Although cerberus and Xbra expression was clearly reduced in Mixer-depleted embryos at both stages, the expression of eomesodermin, Bix 3 and Fgf8 was increased in equatorial cells and expanded into deeper cells. To examine the expansion of the domain of eomesodermin expression more closely,Mixer MO-1 was injected into single vegetal cells of eight-cell stage embryos,to block Mixer translation over only one quarter of the vegetal mass at the gastrula stage. In comparison with the control side, where eomesodermin mRNA is expressed in equatorial cells only, Fig. 4C shows an expansion of eomesodermin expression into the vegetal territory in the Mixer-depleted area (arrow). These data show clearly that a normal role of Mixer is to repress the expression of mesodermal genes in the endoderm.

Both gain- and loss-of-function experiments show that Mixer is required for the increased expression of some genes and the decreased expression of others,both in the mesoderm and endoderm. What is the biological role of Mixer? To test this, we used recombination experiments(Nieuwkoop, 1969) in which wild-type animal caps from mid-blastula stage embryos were placed in contact with vegetal explants from uninjected control or Mixer-depleted late blastula stage embryos. After a 2 hour co-culture period, animal caps were peeled off the vegetal masses, stripped of any adherent vegetal cells, cultured until sibling embryos had reached the mid-gastrula stage and analyzed for mesodermal markers, including Xbra, Xnr1, Fgf8, eomesodermin and the endodermal marker Xsox 17. Fig. 5indicates that Mixer-depleted vegetal masses induced higher levels of expression of all the mesodermal markers in animal caps than did control vegetal masses. By comparison, the endodermal marker Xsox 17 was not significantly induced in caps by either control or Mixer-depleted vegetal masses. The experiment was repeated with a similar result (data not shown). This suggests that Mixer normally functions to control the amount of mesoderm-inducing activity in the vegetal mass.

We next tested the effect of Mixer-depletion on the ability of vegetal cells to induce mesoderm in vivo. To do this, we disaggregated vegetal masses from control and Mixer-depleted late blastulae and transplanted small groups(12-15) of cells to the blastocoel cavities of untreated host embryos. Previous studies have shown that transplanted vegetal cells from late blastulae give rise to descendants mostly in endoderm and occasionally in the lateral plate mesoderm (Wylie et al.,1987). We found that descendants of the Mixer-depleted groups were found only in the endoderm (11/16 cases), or in both lateral plate mesoderm and endoderm (5/16 cases) (Fig. 6B). In three out of 16 cases, an ectopic somite was induced near to the descendant cells (Fig. 6D,E) and identified using the somite-specific 12.101 antibody. Induction of an ectopic somite was not seen when control vegetal cells were transplanted (Fig. 6C, 0/13 cases). These data, together with the Nieuwkoop recombinant experiments show that Mixer-depleted vegetal cells have an increased capacity to induce mesoderm.

Mixer, unlike the other six paired homeodomain proteins of the Mix/Bix family expressed during gastrulation in Xenopus, is most strongly expressed at the boundary between prospective mesoderm and endoderm(Henry and Melton, 1998). In order to test the essential roles for Mixer in normal development, we have depleted it and assayed the effects first on gene expression and second on biological function of the Mixer-expressing cells. The morpholino antisense oligos chosen for these studies, are complementary to all isoforms of Mixer mRNA but have at least four mismatches with other family members, and do not block translation of Mix.1. We confirm that the loss of function is Mixer-specific by rescuing both phenotypically and in terms of molecular markers with non-complementary Mixer mRNA.

These studies show that the roles of Mixer are more complex than was suggested previously. We find that Mixer negatively regulates several early zygotic genes both in the mesoderm and endoderm, while positively regulating others. Further experiments are required to determine whether these effects of Mixer are direct or indirect. One possible scenario is that Mixer activates genes such as cerberus directly and these then block Xnr5 signaling activity, and thus the Xn5 autoregulatory loop reducing Xnr5 mRNA expression. This seems unlikely because of the timing of expression of these two genes. Xnr5 mRNA is already overexpressed in Mixer-depleted embryos at the early gastrula stage when cerberus mRNA expression is only just beginning [Fig. 3A; see Xanthos et al. (Xanthos et al., 2002)for a more extensive temporal series of cerberus mRNA expression]. Another possibility is that Mixer activates some genes and repress others directly. As one dose of oligo upregulates some target genes, and downregulates others in the same area, it seems unlikely that the type of activity is concentration dependent. More likely, the behavior of Mixer is context dependent and gene specific, its combinatorial activity with other transcription factors, as well as with co-repressors and co-activators,determining the final outcome of Mixer binding. Detailed analysis of the regulatory regions of these target genes, and the identification of Mixer-associated co-repressors and co-activators is necessary to determine if Mixer acts as a transcriptional activator and repressor.

It is interesting that, although the T-box transcription factors eomesodermin, Xbra and Antipodean are all recognized as pan-mesodermal markers and have very similar expression patterns, they are not regulated in the same way by Mixer, as eomesodermin expression increases, while the other genes are downregulated after Mixer depletion. Thus, similarity of expression patterns is not necessarily indicative of similar mechanisms of gene regulation, and Mixer cannot be viewed as a simple mesodermal repressor.

Xnr1, Xnr5, Fgf3 and Fgf8, which are repressed by Mixer, encode secreted proteins that act as mesoderm inducing signals in Xenopus embryos. Xnr1 and Xnr5 have been shown to rescue mesoderm and axis formation in VegT-depleted embryos(Kofron et al., 1999; Takahashi et al., 2000), while FGF proteins are implicated in muscle cell precursor formation(Standley et al., 2001). We show in Nieuwkoop assays and by carrying out cell transplantation experiments,that that the loss of Mixer activity results in excessive production of mesoderm inducing signals. This suggests that one important role for Mixer is in limiting the signaling that specifies mesodermal fates, and thus in dictating the size of the mesodermal territory. One interesting aspect of this possible role for Mixer is the timing of its expression. The temporal expression of Mixer mRNA lags considerably behind that of Xnr and Fgf genes. We and others have shown that Xnr1, Xnr2, Xnr4, Xnr5 and Xnr6 are strongly expressed downstream of VegT at the late blastula stage (Takahashi et al., 2000; Xanthos et al., 2002), while Mixer expression begins at this stage but peaks 4 hours later at the gastrula stage (Xanthos et al.,2002). The onset of Mixer activity thus corresponds to the time of `loss of competence' of the vegetal mass, as defined by its ability to induce mesoderm in animal caps(Jones and Woodland, 1987),and is preceded by a 3-4 hour time window when mesoderm induction is uninhibited. It will be important to understand what, besides VegT, controls the temporal pattern of expression of Mixer.

Interestingly, transplantation experiments suggest that Mixer expression is not essential for endodermal fate specification. All individually transplanted and most small groups of Mixer-depleted vegetal cells transplanted onto the blastocoel floor of host embryos continued to develop in the endoderm germ layer and did not redistribute to mesoderm, suggesting that Mixer-depletion is insufficient to cause their relocation to mesoderm. Previous studies have shown that other transcription factors including GATA5 and other Mix family members have important roles and may act in parallel with Mixer in determining endodermal fate (Casey et al.,1999; Weber et al.,2000). It is also possible that the surrounding wild-type vegetal cells release inductive signals that maintain the endodermal fate of the Mixer-depleted cells.

This and previous work suggest the following model for formation of boundaries of expression of several VegT target genes(Fig. 7). After MBT, VegT activates the expression of Xnr genes, Xsox17, Gata4, Gata5 and Gata6, and Fgf genes in vegetal cells. The secreted mesoderm-inducing molecules induce adjacent marginal cells, which do not express VegT, to express more of themselves as well as the presumptive mesodermal genes including eomesodermin and Xbra. At the early gastrula stage, Mixer is cell autonomously induced by VegT-expressing cells. It represses further expression of Xnr1, Xnr5, Fgf3 and Fgf8, and therefore prevents further mesoderm inducing signals being released. Mixer also activates Xsox17 and cerberusexpression, consolidating endodermal fates for vegetal cells and further limiting the range of signaling molecule activity by the antagonistic activity of cerberus.

We are grateful to Stephanie Lang for assistance with in situ hybridization analysis and to Aaron M. Zorn for useful discussions regarding this manuscript. We also thank J. Summerton and Gene Tools for their assistance in choosing the morpholinos used in this study. This work was supported by NIH RO1 HD38272 to J.H. The 12/101 monoclonal antibody developed by Jeremy P. Brockes was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa,Department of Biological Sciences, Iowa City, IA 52242.