The Notch-target gene hairy2a impedes the involution of notochordal cells by promoting floor plate fates in Xenopus embryos

The floor plate (FP) is an epithelial structure located on the ventral midline of the vertebrate neural tube, extending from the midbrain to the tail region, and constituting an important source of signals involved in the induction of motor neurons and guidance for axonal pathfinding(Tanabe and Jessell, 1996; Colamarino and Tessier-Lavigne,1995; Stoeckli and Landmesser,1998). Two cell populations contribute to this structure: the medial FP (MFP) in the midline and the flanking lateral FP (LFP). In zebrafish and mouse embryos, foxa2 [formerly known as hnf3b; for unified nomenclature of the winged helix/forkhead transcription factors see Kaestner et al. (Kaestner et al.,2000)] is found in both populations, while sonic hedgehog(shh) is only expressed by the MFP. In chicken, medial and lateral cells initially express both markers, but foxa2 becomes later restricted to the MFP, while some shh expression remains in the LFP(reviewed by Strähle et al.,2004).

The origin of the FP has been subject of a recent controversy, which can be synthesized in three models, based on those described by Strähle and colleagues (Strähle et al.,2004). (1) The `induction' model postulates that the FP derives from the neural ectoderm and is induced by secreted Shh from the notochord(Placzek et al., 2000; Tanabe and Jessell, 1996; Chiang et al., 1996). (2) The`allocation' model postulates a common origin for notochord and MFP in the vertebrate's organiser, with MFP precursors invading the midline of the overlying neural plate (Spemann and Mangold, 1924; Selleck and Stern, 1991; Gont et al.,1993; Wilson and Beddington,1996; Catala et al.,1995; Catala et al.,1996; Shih and Fraser,1995; Melby et al.,1996; Teillet et al.,1998; Amacher et al.,2002; Latimer et al.,2002) (see also Le Douarin and Halpern, 2000). In this model, Shh is regarded as a factor necessary for survival of FP cells and for maintenance of their phenotype, and as an inducer of anterior and lateral FP(Le Douarin and Halpern, 2000; Thibert et al., 2003; Charrier et al., 2002; Patten et al., 2003; Rebagliati et al., 1998; Sampath et al., 1998; Schauerte et al., 1998; Odenthal et al., 2000). (3)The `induction and allocation' model tends to reconcile the experimental evidence from the first two models by proposing that the inductive step for the MFP [involving Shh and/or Nodal signalling, which varies among vertebrate species (see Strähle et al.,2004)] takes place before the segregation of MFP and notochord precursors emerging from the organiser. Then, specified MFP precursors populate the midline of the neural plate.

In Xenopus embryos the term notoplate designates the midline neural plate cells that later become the FP of the neural tube. It was reported that the notoplate arises from the dorsal non-involuting marginal zone (DNIMZ), a region of ectoderm located just above the dorsal involuting marginal zone (DIMZ) or dorsal lip. The latter contains mesodermal cells that enter through the blastopore during gastrulation and extend along the anterior-posterior (AP) axis, giving rise to the prechordal plate and notochord. The notoplate precursors remain in the ectodermal layer but,together with the notochord precursors, undergo convergent-extension movements during gastrulation, resulting in the midline extension of the future FP along the AP axis (Jacobson, 1981; Keller et al., 1985; Keller and Danilchik,1988).

We have recently described that before mid-gastrula Notch executes a binary cell-fate switch that favours FP development at the expense of the notochord,leading to the specification of the different cell populations that contribute to the dorsal midline (DML) in Xenopus(López et al., 2003). As a corollary, we proposed that the early organiser indeed contains cells that have the potential to develop either as notochord or FP, but the question of whether they constitute a mixed population or occupy different compartments within the organiser remained unanswered. In addition, we described that Notch signalling activates shh expression, and secreted Shh would amplify the effects of the binary decision by inhibiting notochord specification. By this means, Shh would refine the segregation of both cell-populations initially started by Notch. This mechanism could in part underlie the role of shh as a FP inducer during the early event proposed by the third model of FP formation.

The main goal of the present work is to understand the molecular and cellular mechanisms that govern the development of the DML structures,beginning from their precursors in the Spemann's organiser. In particular, we wanted to answer the following questions: (1) Which is the ligand(s) for the Notch receptor that triggers the cell-fate switch FP vs. notochord? (2) Which is the Notch-target gene(s) that executes this switch? (3) How do cells from the Spemann's organiser give rise to the FP?

Albino Xenopus laevis embryos were obtained using standard methods(Ruiz i Altaba, 1993) and staged according to Nieuwkoop and Faber(Nieuwkoop and Faber, 1994). Synthetic capped mRNAs for microinjection were obtained as described previously (Franco et al.,1999). The hairy2a antisense oligodeoxynucleotide (AMOh)used was a 3′-carboxyfluorescein-tagged 25-mer morpholino oligonucleotide (Gene Tools, LLC) with the base composition 5′-ATGGTATCTGCGGGCATGTTCAGTT-3′, complementing the hairy2a sequence from -8 to +17 relative to the initiation codon. The templates for X-notch^ICD and X-su(H)^DBM mRNA synthesis were described by Wettstein et al. (Wettstein et al., 1997),and those for X-delta1 and X-delta1^STU, by Chitnis et al. (Chitnis et al.,1995). The full-length X-hairy2a cDNA construct in pCS2+,kindly provided by Dave Turner, was digested with NotI and transcribed with SP6 RNA polymerase.

Samples were injected as previously described(López et al., 2003). The amounts of synthetic mRNAs and morpholino injected are indicated in the figures. Some injections included 0.5 ng of nuc-lacZ mRNA as tracer.

X-gal staining, preparation of digoxigenin-labelled antisense RNA probes and whole-mount in situ hybridisation were performed as described previously(Franco et al., 1999; Pizard et al., 2004), except that the proteinase K step was omitted in in situ hybridization. The hairy2a template (Turner and Weintraub, 1994) was digested with BamHI and transcribed with T7 RNA polymerase. For double in situ hybridization, fluorescein-labelled antisense RNA probes were prepared with fluorescein-12-UTP (Amersham). Embryos were hybridised with digoxigenin and fluorescein-labelled probes simultaneously, washed and blocked according to the standard protocol and incubated first with one of the antibodies conjugated with alkaline phosphatase (AP) (1/2000 of anti-digoxigenin-AP or 1/5000 of anti-fluorescein-AP, Fab fragments; Amersham). The corresponding probe was revealed with 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) or Magenta Phos(Sigma). Inactivation of the AP was carried out (65°C 30 minutes in methanol, two 5-minute washes in methanol at room temperature), followed by two 5-minute washes in MAB buffer and 15 minutes incubation in blocking reagent before adding the second AP-conjugated antibody, which was revealed with NBT+BCIP or BCIP alone (Sigma).

The c-myc epitope harboured by the notch^ICD or su(H)^DBM constructs used for mRNAs injections and the fluorescein tag of AMOh were detected by immunohistochemistry. For this purpose, after in situ hybridization we performed the AP-inactivation, washing and blocking steps as described above. Then, embryos were incubated for 4 hours at room temperature with mouse 9E10 anti-Myc monoclonal antibody (Santa Cruz) diluted 1/500 or with anti-fluorescein-AP (Fab fragments, Amersham)diluted 1/5000, both in blocking reagent (the same as for in situ hybridization). The unbound antibodies were washed three times with MAB, 10 minutes each, at room temperature, and overnight at 4°C. The following day, embryos injected with AMOh were revealed with BCIP or Magenta Phos, as in the in situ hybridization protocol, and embryos injected with notch^ICD or su(H)^DBM were incubated for 4 hours at room temperature with anti-mouse IgG-AP (Santa Cruz) diluted 1/500 or with anti-mouse IgG-HRP (Dako) diluted 1/100 in blocking reagent. After washing the excess of antibody as before, the anti-mouse IgG-AP was revealed with BCIP or Magenta Phos, as in the in situ hybridization protocol. Embryos incubated with the HRP-conjugated antibody were washed twice with TBS(10 mM Tris-HCl pH 6.5, 150 mM NaCl), 5 minutes each, equilibrated 30 minutes at room temperature with DAB solution (0.5 mg/ml of 3,3′-diaminobenzidine (Sigma) in 10 mM Tris-HCl pH 6.5) and revealed with 0.009% of H₂O₂ in DAB solution.

For histology, 50 μm sections were taken in an Oxford Vibratome and mounted onto gelatine coated slides, as described by Hollemann et al.(Hollemann et al., 1996).

We searched for Notch ligands and direct target genes whose expression patterns in Xenopus could suggest their participation in the DML binary switch. The presence of delta1 transcripts in the dorsal blastopore lip at stage 10.5 has been reported(Ma et al., 1996). Among the bHLH-O transcriptional repressors that mediate Notch signalling isolated from Xenopus (Davis and Turner,2001), hairy2a was very interesting because, apart from bordering the neural ectoderm, it is also present along the DML in open neural plate-stage embryos (Turner and Weintraub,1994). A more precise study of the early distribution of both transcripts would help to determine whether they are present at the right time and place to be involved in the proposed cell-fate switch. Therefore, we analysed the distribution of delta1 and hairy2a transcripts in more detail by in situ hybridization and compared their expression patterns.

In the dorsal marginal zone, evident transcription of both genes appears in the organiser region around stage 10.5(Fig. 1). delta1-positive (delta1+) cells are scattered in the dorsal lip (asterisks, Fig. 1B,B′), and strongest expression is found in the rest of the marginal zone (arrow, Fig. 1B,B′). Sagittal sections of these embryos show that, in the organiser, delta1 transcripts are only found in the dorsal mesoderm that has not yet involuted (arrow, Fig. 1B″), whereas involuted axial mesoderm does not express this gene (asterisk, Fig. 1B″). At the same time, hairy2a+ cells are distributed in an arc on the DNIMZ (arrowheads, Fig. 1D-F″). This arc gradually accumulates more hairy2a+ cells and ultimately converges and extends along the AP axis, forming the notoplate (prospective FP) (G-K,M). Interestingly, at early and mid-gastrula stages several hairy2a-expressing cells are scattered on the DIMZ (asterisks, Fig. 1F-F″), where scattered delta1 cells are also found(Fig. 1L). Later, at neural plate stage, hairy2a is strongly expressed in the prospective FP,flanked by bilateral stripes of delta1 corresponding to the proneural domains of ventral motor neurons. Among the involuted DML cells, hairy2a is only found in the head mesoderm, while the notochord is devoid of transcripts (Fig. 1G-M).

Next, we analysed the expression of hairy2a in the context of other markers of notochord and FP fates. Chordin (chd)transcripts are normally present in the notochord and the prechordal mesoderm(Sasai et al., 1994). Double in situ hybridization of chd and hairy2a revealed that their territories are mutually exclusive. While chd is expressed in notochordal cells, hairy2a is only found in FP precursors(Fig. 2A,A′). The winged-helix transcription factor foxa4a (formerly known as pintallavis) was reported to participate in FP development(Lee et al., 1997). FoxA4a mRNA is first detected in the dorsal marginal zone of late blastulae and persists during gastrulation in the DML cells that undergo convergent-extension movements. At the early neurula stage, transcripts are distributed throughout the DML in the three germ layers, i.e. the prospective FP, the notochord and the dorsal endodermal cells lining the archenteron(Ruiz i Altaba and Jessell,1992). At gastrula stages (around st. 11), foxa4a and hairy2a have partially overlapping domains(Fig. 2B,B′). In the DNIMZ, the outer limit of foxa4a coincides with the hairy2aarc that marks the prospective notoplate (asterisks). While the entire DIMZ in the organiser expresses foxa4a, only some scattered cells are hairy2a+ (arrow). At this stage, single in situ hybridization of foxa4a distinguishes a superficial population of cells with strong expression of foxa4a, which overlaps the hairy2a arc that demarcates the prospective notoplate (asterisks, Fig. 2C,C′). Double in situ hybridization of chd and foxa4a reveals that these genes have partially overlapping domains(Fig. 2D-E″). In early gastrulae, while both are expressed in the DIMZ and in the involuted notochordal cells, only foxa4a is found in the DNIMZ, in an area corresponding to the hairy2a+ arc that demarcates the future notoplate (asterisk, Fig. 2D,D′). In late gastrulae foxa4a transcripts are found in the prospective FP, which is devoid of chd transcripts, as chd is only expressed by the notochord(Fig. 2E″). Expression of shh in the dorsal marginal zone starts later than that of chd and foxa4a, and shh transcripts are ultimately found in FP and notochord cells (Ekker et al., 1995; López et al., 2003). Double in situ hybridization of shh and hairy2a show that both genes are co-expressed in the prospective FP,but the latter is excluded from the notochord, where shh transcripts are also found (Fig. 2F-G′).

In conclusion, hairy2a is an interesting marker for the FP precursors and, as a Notch-target gene, it is a good candidate for mediating some of the molecular changes involved in DML cell-fate decisions. The scattered delta1+ cells in the organiser may be the source of the ligand that triggers the Notch pathway on the surrounding cells, leading them to activate hairy2a expression. In early gastrulae, an arc of hairy2a+, foxa4a+, chd-negative (chd-) cells located in the DNIMZ demarcates the prospective notoplate. During the course of the gastrulation, this group of cells also becomes shh+,re-accommodates along the AP axis by convergent-extension movements and will form the FP of the neural tube.

In order to determine whether hairy2a is a target of Notch in the DML precursors, we activated or prevented Notch signalling with notch^ICD or su(H)^DBM mRNA,respectively, and analysed the expression of hairy2a. To identify those cells that inherited and translated the injected mRNAs, we revealed by immunohistochemistry the c-myc epitope fused as a tag to the notch^ICD and su(H)^DBM fragments. Injection of notch^ICD mRNA, which encodes a constitutively active form of the receptor independent of ligand binding, produced an enlargement of the hairy2a domain on the injected side, both in gastrulae and neurulae (85%, n=122, Fig. 3A,A′,C-C″,green asterisks). To corroborate whether endogenous Notch activity was indeed involved in this modulation, we injected su(H)^DBM mRNA,which encodes a dominant negative variant of the Notch transducer Su(H). We observed a decrease of hairy2a+ cells both in gastrulae and neurulae(70%, n=114, Fig. 3B,B′,D-D″, red asterisks). Transverse sections of neural-plate stage embryos confirmed that these changes occurred in the FP precursors (Fig. 3C″,green asterisk; D″, red asterisk). Therefore, hairy2a, whose midline expression at the trunk level exclusively marks FP cells, corroborates that Notch signalling increases the FP size, as previously shown with other markers that are co-expressed by FP and notochord(López et al.,2003).

To test whether Delta1 could be the ligand that triggers the binary switch executed by Notch in the DML precursors, we overexpressed delta1 or blocked the ligand function with the antimorph delta1^STU(Chitnis et al., 1995) and looked for changes in the expression patterns of chd and hairy2a (as notochord and FP markers, respectively). Overexpression of delta1 reduced the number of chd+ cells in the organiser(77% of injected embryos, n=13, Fig. 3E,E′, red asterisk)and increased the number of the hairy2a+ ones (49%, n=49, Fig. 3G,G′, green asterisk). Injection of delta1^STU mRNA resulted in the opposite effects (increase of chd+ cells: 76%, n=38, Fig. 3F,F′, green asterisk; decrease of hairy2a+ cells: 63%, n=16, Fig. 3H,H′, red asterisk).

From these results, we conclude that hairy2a behaves as a Notch target in the DML precursors and that Delta1 is the ligand capable of switching on the binary decision executed by Notch in the DML precursors.

To assess whether hairy2a is able to execute the cell-fate switch triggered by Notch in the DML, we firstly analysed the effects of overexpressing or blocking hairy2a on notochord development by looking at the expression of two notochordal markers: chd and brachyury (bra). When we injected 1 ng of hairy2amRNA, chd was drastically repressed in the Spemann's organiser after dorsal injections (94%, n=33, Fig. 4A, red asterisk). Paradoxically, ventral or lateral injections resulted in ectopic chd transcription in the rest of the marginal zone (100%, n=27; arrowhead, Fig. 4B). When sibling control embryos reached the neural plate stage,all dorsally injected embryos were arrested at gastrulation. Chd+cells were reduced in number, they could not migrate anteriorly and remained mostly in the outer layer encircling the blastopore, which was unable to complete its closing (100%, n=23; Fig. 4C, red asterisk). Interestingly, at the neurula stage, ventrally injected embryos developed a normal dorsal axis (white arrow, Fig. 4D), but the ectopic chd+ cells remained close to the ventral blastopore lip and were unable to migrate and extend anteriorly (100%, n=20, arrowhead, Fig. 4D). Lower doses of hairy2a mRNA also repressed dorsal chd expression and resulted in ectopic chd+ cells on ventral or lateral mesodermal locations in a similar proportion of injected embryos(e.g. for chd repression in dorsally injected embryos: 86% with 0.5 ng of hairy2a, n=14; 95% with 0.25 ng of hairy2a, n=20) but qualitatively, the effects were gradually weaker (not shown). However, with 0.25 ng injections, gastrulation could better proceed, allowing us to examine other markers at neural plate stages (see below). To block hairy2afunction we injected an antisense morpholino oligonucleotide complementary to a sequence of hairy2a comprising the initiation codon (AMOh). 10 ng of AMOh increased the number of chd+ cells on the injected side, both in gastrulae (Fig. 4E-E‴, green asterisks) and neurulae (not shown) (82%, n=45). When co-injected with 0.5 ng of hairy2a mRNA, 10 ng of AMOh reversed the down-regulation of chd that produces hairy2a mRNA alone, demonstrating that this antisense morpholino specifically interferes with the translation of hairy2a transcripts[chd+ cells decreased in only 10% of the dorsally injected embryos,the remaining ones evidenced an increase of chd+ cells (67%) or were unaffected (23%), n=30; Fig. 4F,F′, green asterisk].

In early gastrula, bra expression is normally observed in a ring demarcating the entire marginal zone, broader at the ventral side when the dorsal cells begin to involute at the blastopore(Fig. 4G,G′). Injection of hairy2a mRNA represses bra expression in all locations,including the organiser (91% with 0.25 ng of hairy2a mRNA, n=44, Fig. 4H-I, red asterisks; 100% with 1 ng, n=18, not shown; the effects are qualitatively stronger with higher doses of mRNA). However, dorsal injection of 10 ng of AMOh increases bra+ cells, both in gastrulae and neurulae(74%, n=43, Fig. 4J-K′, green asterisks).

Together, these results show that hairy2a, like Notch signalling,represses bra and chd in the Spemann's organiser. Therefore, hairy2a may be the mediator of Notch in the repression of the notochordal fate that takes place during the DML cell-fate switch. In addition, an excess of hairy2a activity interferes with the normal movement of chd+ (notochordal) cells, blocking their involution.

We next addressed the question of whether hairy2a was able to promote the FP fate as does Notch signalling. Therefore, we overexpressed or blocked hairy2a and looked at the expression of two FP markers: foxa4a and shh. However, these genes are also expressed in the notochord progenitors. Although at stage 11 it is possible to discern the contribution of foxa4a to the arc of the notoplate, the low expression of shh at this time does not allow the determination of the FP and notochord components that contribute to the shh domain. Thus, we were interested in analysing the effects of hairy2a at neural plate stages, when both components can be clearly distinguished. Although 1 ng of hairy2a mRNA had the strongest effects on mesodermal markers at gastrula stages and also promoted ectopic expression of foxa4a on ventral locations (arrowhead, Fig. 5A″), this dose severely interfered with gastrulation movements, and we were unable to distinguish between the FP and the notochordal components of shhexpression in these extremely affected embryos. Thus, we decided to lower the dose until we were able to analyse shh in neurulae, but preserving the effects, although milder, on notochordal markers. This compromise could be reached with 0.25 ng of hairy2a mRNA, as described above. In these conditions, overexpression of hairy2a increased foxa4a+cells in the notoplate precursors (80%, n=15, Fig. 5A,A′, green asterisk) and increased shh+ cells in the prospective FP (67%, n=27, Fig. 5D-D″, green asterisks). In contrast, blocking hairy2a with 10 ng of AMOh decreased foxa4a+ cells in the notoplate in gastrulae and in the prospective FP in neurulae (54%, n=44, Fig. 5B-C″, red asterisks) and shh+ cells in the prospective FP in neural plate stage embryos (58%, n=19, Fig. 5E-E″, red asterisks). In conclusion, hairy2a, like Notch signalling, increases foxa4a+ and shh+ cells within the prospective FP domain. These results suggest that hairy2a is a mediator of Notch in the promotion of FP specification that takes place during the DML cell-fate switch. Overall, hairy2a is able to promote the FP fate at the expense of the notochord, and this was confirmed by double in situ hybridization of chd and foxa4a in injected embryos: 0.25 ng of hairy2a mRNA increase the domain of chd-foxa4a+ cells (FP precursors, green asterisk) and concomitantly decrease the domain of chd+foxa4a+ cells (notochord precursors, red asterisk) (79% and 92%,respectively; n=14) (Fig. 5F).

If hairy2a is required for the cell-fate switch induced by Notch,then, blocking hairy2a activity would impede the effects of notch^ICD on DML development. Therefore, we performed co-injections of 10 ng of AMOh plus 1 ng of notch^ICD mRNA and compared the effects on chd, foxa4a and shh expression with those obtained after injecting embryos with 1 ng of notch^ICD mRNA alone or 10 ng of AMOh alone(Table 1, Fig. 6). As previously shown(López et al., 2003), notch^ICD alone decreased the number of notochordal precursors (chd+ cells) in the organiser(Fig. 6A, red asterisk) and increased the number of foxa4a+ and shh+ cells in the FP domain (Fig. 6D,D′,G,G′, green asterisks). The opposite results were observed in embryos injected with AMOh alone(Fig. 6B, green asterisk, chd+ cells increase; E,E′,H,H′, red asterisks, foxa4a and shh repression) or co-injected with notch^ICD plus AMOh(Fig. 6C, green asterisk, chd increase; F,F′,I,I′, red asterisks, foxa4aand shh repression). Thus, AMOh could reverse the effects of notch^ICD on DML markers, as expected. We conclude that Notch signalling represses the notochordal fate and promotes FP specification through the activation of hairy2a.

We have previously proposed that Notch signalling may be executing a binary cell-fate decision in the Xenopus organiser within a bipotential cell population: when active, it promotes FP specification at the expense of the notochord (López et al.,2003). However, the question of whether these precursors are mixed or if they occupy different compartments within the organiser remained unanswered.

We present evidence that Delta1 is the ligand that triggers this cell-fate switch, and that hairy2a is the Notch target that mediates the repression of the notochordal fate and the promotion of FP development. We found that hairy2a is able to repress genes that are involved in dorsal axial mesoderm development, such as chd and bra(Chesley, 1935; Smith et al., 1991; Cunliffe and Smith, 1992; Cunliffe and Smith, 1994; Halpern et al., 1993; Sasai et al., 1994; Sasai et al., 1995; O'Reilly et al., 1995; Piccolo et al., 1996; Piccolo et al., 1997; Hammerschmidt et al., 1996; Conlon et al., 1996; Schulte-Merker et al., 1997),and to promote the expression of genes related to FP specification, such as shh and foxa4a (Ruiz i Altaba et al., 1993; Ruiz i Altaba et al., 1995; Roelink et al., 1994; Chiang et al.,1996; Chang et al.,1997; Sasaki et al.,1997; Epstein et al.,1999; Müller et al.,1999). Since hairy2a is a transcriptional repressor, at least two alternative explanations for its role in DML development arise. (1)A permissive role for FP development, which implies that hairy2arepresses genes that specify the notochordal fate and allows the development of FP identity through some default mechanism. In this context, it is intriguing that markers of FP specification are also expressed by the notochord (e.g. shh, foxa4a), the only exception being hairy2a itself, whereas notochordal markers seem to be exclusively present in the notochord (e.g. bra, chd). hairy2a may thus deplete the DML precursors of molecules required for the specification of notochord, allowing FP to develop. (2) An instructive role, which implies that, apart from the repression of genes required for notochord development, hairy2a may indirectly promote FP specification by repressing a negative regulator of genes that specify FP fate (e.g. shh,foxa4a).

Bra behaves as a transcriptional activator(Conlon et al., 1996), and the zebrafish homologue is no tail (ntl). Notably, while the notochord does not differentiate in ntl mutant embryos, the FP is widened (Halpern et al.,1997). This suggests that Bra activity antagonises FP development while promoting notochord formation. Therefore, bra may constitute a key target in the Notch-dependent binary switch, and it may be directly repressed by hairy2a. Further experimentation will be needed to elucidate whether a hierarchical relationship links hairy2a and bra in this switch in Xenopus. Interestingly, although hairy2a is able to activate chd ectopically in ventral or lateral mesoderm, it represses bra in all locations. Since chd+bra- cells are unable to involute and migrate properly and brais involved in convergent-extension movements(Conlon and Smith, 1999; Kwan and Kirschner, 2003), it is conceivable that hairy2a interferes with them by repressing bra. We suggest that the specification of FP or notochord fates is intimately linked to cell movements during gastrulation and that hairy2a constitutes a master gene operating on both events.

It was recently described that the zinc finger transcriptional activator Churchill (ChCh) stops the ingression of cells through the primitive streak in chicken embryos (Sheng et al.,2003). The authors postulated that during normal embryogenesis a decision between paraxial mesodermal and neural fates is made by establishing the boundary that restricts cell ingression during gastrulation. We propose that an analogous mechanism may take place during DML development, with hairy2a stopping the ingression of notoplate cells that, otherwise,would have been incorporated into the notochord. This specialised neural vs. mesodermal switch for DML precursors is envisaged through the observation that the Xenopus FP retains the potential to differentiate into neurons,since the ectopic expression of the proneural gene X-ngnr1 in the FP precursors turns on N-tubulin expression, and the bHLH-O transcriptional repressor XHRT1, which is expressed in the prospective FP, represses neurogenesis(Taelman et al., 2004). Interestingly, XHRT1 is able to heterodimerize with Hairy proteins,suggesting that they may be biologically relevant partners. Although XHRT1 responds to Notch and can inhibit chd and bra, it is unlikely to be involved in the notochord vs. FP switch because it appears after mid-gastrula, later than hairy2a (this work)and hairy2b (Taelman et al.,2004). However, the three transcripts ultimately co-localise in the FP precursors. This fact, together with the presence of notch1transcripts in FP at neural plate stages(López et al., 2003)and of delta1 transcripts flanking the hairy2a FP domain(Fig. 1M of this work) support the idea of XHRT1 having a role in preventing FP cells from adopting a neuronal fate (Taelman et al.,2004), perhaps in collaboration with hairy2a/b.

Thus, the FP fate would be specified and/or maintained in the course of at least two binary decisions involving Notch signalling: (1) The early notochord/FP decision, which appears to take place before mid-gastrula(López et al., 2003). Two parallel mechanisms seem to contribute to stop this switch: (a) notch1 and delta1 transcripts disappear once the notochordal cells involute (Wittenberger et al.,1999; López et al.,2003) (this work), suggesting that they become refractory to divert to FP in response to Notch ligands, which, these cells do not produce anymore; (b) the competence of the binary switch for responding to active Notch decreases throughout gastrulation(López et al., 2003).(2) The neuron/FP decision. In this event, Delta1 flanking the midline at neural plate stages would activate the Notch receptor in the neighbouring FP cells, which then would turn on hairy2/XHRT1 bHLH-O genes, thus repressing the neuronal fate to maintain the FP phenotype.

Fig. 7 (left column) shows the dynamics of hairy2a expression in notch^ICD-injected embryos in a developmental series from early to late gastrula stages. Supernumerary hairy2a+ cells(asterisks) first appear on the DIMZ on the injected-side and gradually incorporate into the hairy2a-expressing arc on the DNIMZ that demarcates the prospective notoplate. We propose that, in normal embryogenesis, the early Xenopus organiser contains cells that have the potential to develop either as notochord or FP(Fig. 7 right column). Delta1 expression starts at early gastrula in scattered cells on the organiser and interacts with the Notch receptor in the surrounding cells,leading to the activation of hairy2a and the repression of chd and bra. hairy2a, in turn, impedes the movement of involution, and hairy2a+ cells gradually incorporate into the growing arc on the DNIMZ. This arc ultimately converges and extends along the AP axis(white arrows), forming the notoplate (prospective FP). By this mechanism involving notch and hairy2a, Delta signalling executes a binary cell-fate switch that favours FP development at the expense of the notochord, leading to the specification of the different cell populations that contribute to the DML. This model reconciles the findings that the notoplate arises from the DNIMZ with the hypothesis that FP cells arise from the organiser (DIMZ). However, we cannot rule-out additional contributions to FP development from the neural ectoderm (more specifically, to the anterior and lateral FP) involving Shh as inducer, as it has been described for chicken and zebrafish embryos (Le Douarin and Halpern,2000; Thibert et al.,2003; Charrier et al.,2002; Patten et al.,2003; Rebagliati et al.,1998; Sampath et al.,1998; Schauerte et al.,1998; Odenthal et al.,2000). Interestingly, the Axolotl homologue of foxa4a,unlike its Xenopus counterpart, has only a superficial expression in the early organiser and later, it is only detected in the FP but never in the notochord, thus resembling the expression of hairy2a in Xenopus (Whiteley et al.,1997). Whiteley et al. provide two possible explanations: (a) the superficial foxa4a+ cells in the axolotl organiser are future neural FP cells, programmed very early at gastrulation; (b) the expressing cells are a mixture of notochord and FP precursors, but later, foxa4a only persists in FP cells. We presume that both explanations may not be mutually exclusive and, as with Xenopus hairy2a, some cells may be representing anterior FP cells, programmed at very early stages of gastrulation, as those described in chicken(Patten et al., 2003), and other cells may be posterior FP precursors being specified from a bipotential population, which is also able to give rise to notochordal cells.

In this scenario, our proposal fits better with the induction and allocation model (see Introduction), where the inductive event (in charge of Delta-Notch signalling) takes place before the segregation of notochord and FP precursors (this work), with Shh expression (enhanced by Notch signalling in FP precursors) contributing to repress the notochordal fate and amplifying the switch (López et al.,2003). It remains to be elucidated which are the molecules that pattern the scattered expression of delta1 in the organiser, which initially defines the distribution of FP and notochord precursors according to our model.

We wish to acknowledge the following colleagues for providing us with the constructs for making synthetic mRNA: Dave Turner for CS2+ hairy2a,Chris Kintner for X-su(H)^DBM, Tomas Pieler for X-notch^ICD, X-delta1 and X-delta1^STU,and Richard Harland for nuc-lacZ. We are also grateful to the following researchers for providing us with the constructs for making in situ hybridization probes: Ariel Ruiz i Altaba for foxa4a, Eric Bellefroid for X-delta1, Abraham Fainsod for X-brachyury, Eddy De Robertis for chordin and again to Dave Turner for hairy2a.We thank Paula Raimondi for her effort and collaboration. S.L.L., A.R.P. and A.E.C. are from CONICET. M.V.R.S. was supported by fellowships Ramón Carrillo-Arturo Oñativia 2001 and from ANPCyT, and P.G.F. by a fellowship from ANPCyT. This paper was supported by grants to A.E.C. from CONICET (PIP 0848/98), ANPCyT (BID802/OC-AR PICT 3410, BID1201/OC-AR PICT 11219) and Beca Ramón Carrillo-Arturo Oñativia 2000-2001, and to S.L.L. from CONICET (PEI 6356/03).