Nodal signaling patterns the organizer

Spemann’s organizer, which resides on the dorsal side of the vertebrate gastrula, has the ability to induce a secondary axis when transplanted to the ventral side of a host embryo. Organizer cells also contribute to the dorsal mesoderm of the axial midline. This axial mesoderm is regionalized along the anteroposterior axis, with the prechordal plate located anteriorly and the notochord forming more posteriorly (Spemann and Mangold, 1924; Harland and Gerhart, 1997; Schier and Talbot, 1998). Both the prechordal plate and the notochord have important roles in patterning surrounding structures such as the somites and the overlying neural plate. The prechordal plate later differentiates into head and eye muscles, pharyngeal endoderm and the hatching gland in fish (Adelmann, 1932; Keller, 1976; Jacob et al., 1984; Wachtler et al., 1984; Thisse et al., 1994; Kimmel et al., 1995), whereas the notochord is a mesodermal rod, which provides structural support to the embryo and later contributes to the development of vertebral bodies.

In the zebrafish and Xenopus blastula, notochord and prechordal plate precursors are both located in the dorsal equatorial region (Keller, 1976; Kimmel et al., 1990; Vodicka and Gerhart, 1995; Melby et al., 1996; Shih and Fraser, 1996; Zoltiewics and Gerhart, 1997; Harland and Gerhart, 1997; Warga and Nüsslein-Volhard, 1999). During early gastrula stages, anteroposterior polarity is already apparent, with prechordal plate progenitors located in the anterior portion of the involuting axial mesoderm and notochord precursors located more posteriorly (Keller, 1976; Melby et al., 1996; Vodicka and Gerhart, 1995; Zoltiewics and Gerhart, 1997; Harland and Gerhart, 1997). At these stages, anterior and posterior regions of the organizer also have different specification profiles, distinct inductive activities and express different marker genes (Vodicka and Gerhart, 1995; Melby et al., 1996; Zoltiewics and Gerhart, 1997; Harland and Gerhart, 1997). For instance, the expression of the homeobox genes goosecoid (gsc) (Cho et al., 1991; Stachel et al., 1993) and floating head (flh) (Talbot et al., 1995), the zebrafish homologue of Xnot (von Dassow et al., 1993), marks prechordal plate progenitors and notochord progenitors, respectively (Melby et al., 1996; Cho et al., 1991; Stachel et al., 1993; Talbot et al., 1995). These observations suggest that the notochord and prechordal plate are specified before the end of gastrulation. It has not been determined, however, whether the segregation of these fates is initiated at blastula stages, or if cells become allocated to separate territories only during gastrulation.

Development of the notochord and prechordal plate requires the Nodal family of TGFβ ligands (Conlon et al.,1994; Zhou et al., 1993; Feldman et al., 1998; Sampath et al., 1998; Rebagliati et al., 1998b; reviewed in Schier and Shen, 2000). Double mutants for the zebrafish nodal-related genes cyclops (cyc) and squint (sqt) lack all axial mesoderm (Feldman et al., 1998). Similarly, maternal-zygotic mutants for the EGF-CFC gene one-eyed pinhead (oep) are missing the notochord and prechordal plate (Zhang et al., 1998; Gritsman et al., 1999). Oep is an essential cofactor for Nodal signaling in all processes thought to involve Nodal signals, including germ layer formation, floor plate development and left-right patterning (Schier et al., 1997; Gritsman et al., 1999; Yan et al., 1999). In addition, members of the Lefty family of divergent TGFβ factors, such as zebrafish Antivin (Thisse and Thisse, 1999; also called Lefty 1 (Bisgrove et al., 1999)) antagonize Nodal signaling (Meno et al., 1999; Bisgrove et al., 1999). Overexpression of Antivin can inhibit axial mesoderm formation, resulting in a phenotype that closely resembles that of cyc;sqt and maternal-zygotic oep mutants (Thisse and Thisse, 1999; Meno et al., 1999; Bisgrove et al., 1999). Together, these results suggest that Nodal signals and Oep promote axial mesoderm formation, while Lefty antagonizes it.

While Nodal signaling is required for the development of all axial mesoderm, three observations suggested that it might also be involved in specifying the difference between prechordal plate and notochord fates. First, nodal-related genes have a dose-dependent effect on the induction of dorsal mesodermal genes in Xenopus animal caps. Low levels of Nodal signaling induce the posterior mesoderm and notochord marker Xbra, while higher levels are needed to induce the anterior mesodermal marker gsc (Jones et al., 1995; Lustig et al., 1996; Erter et al., 1998). Second, zygotic oep mutants, which have attenuated Nodal signaling, develop a notochord but not a prechordal plate (Schier et al., 1997). Third, a partial block of Nodal signaling by overexpression of Antivin can result in a phenotype similar to that of zygotic oep mutants (Thisse and Thisse, 1999). Based on these results, we hypothesized that different levels of Nodal signaling regionalize the axial mesoderm during blastula or gastrula stages. Using a combination of fate mapping, marker gene expression analysis and modulation of Nodal signaling, we show here that differential Nodal signaling patterns the organizer before the onset of gastrulation, leading to the formation of prechordal plate anteriorly and notochord more posteriorly.

Fate mapping using caged fluorescein-dextran was performed as previously described in Drosophila (Vincent and O’Farrell, 1992) and zebrafish (Serbedzija et al., 1998; Kozlowski et al., 1997). DMNB-caged biotinylated lysine-fixable fluorescein dextran dye (Molecular Probes) was dissolved at 2.5% in 5 mg/ml Phenol red in 0.2 M KCl and centrifuged for 5 minutes. The supernatant was microinjected into embryos (0.1-0.3 nl per embryo) from an oep^tz57 heterozygous intercross at the 1-to 4-cell stage. Embryos were staged according to Kimmel et al. (1995). At blastula stages, embryos were mounted in 3% methylcellulose. Uncaging was performed with a 10 second pulse of a 375 nm pulsed nitrogen laser (MicroPoint Laser System, Photonic Instruments) focused through the 40× water-immersion objective of a Zeiss Axioplan 2 microscope onto a single cell at a time. Fluorescent and Nomarski images were acquired sequentially using a cooled CCD camera (Pentamax, Princeton Instruments) and shutters (Uniblitz) controlled by the Metamorph 3.0 Imaging System (Universal Imaging). Corresponding fluorescence and Nomarski images were individually adjusted for optimal brightness and contrast and then overlayed.

The dorsoventral position of labeled cells was assigned at the onset of gastrulation, when the thickening of the shield on the dorsal side demarcates the organizer (Kimmel et al., 1990). For dorsal clones, fates were determined at tailbud stage, when the prechordal plate has the morphology of a knob formed by the most anterior hypoblast (Xu et al., 1994) and the notochord forms a thin rod just posterior to it. Fates were confirmed at 24 hours and 5 days postfertilization by fluorescence and by anti-fluorescein antibody staining as follows. Embryos were fixed in 4% paraformaldehyde in PBS at 4°C for 1 day, rehydrated in a graded series of methanol and PBT (PBS + 0.1% Tween 20), and blocked for 1 hour in 2 mg/ml BSA and 5% sheep serum in PBT at room temperature. Embryos were heat treated at 65°C for 10 minutes in blocking solution and incubated for 2 hours at room temperature with preadsorbed anti-fluorescein alkaline phosphatase Fab fragments (Boehringer Mannheim) at 1:2000 dilution in 2 mg/ml BSA in PBT. After eight 15-minute washes at room temperature, embryos were pre-equilibrated with three 5-minute washes with NTMT (0.1 M Tris-HCl pH 9.5, 50 mM MgCl₂, 0.1 M NaCl, 0.1% Tween 20) and stained with the NBT-BCIP substrate (Boehringer Mannheim).

The following plasmids were linearized and antisense RNA was synthesized by in vitro transcription: pBSgsc (EcoRI, T7; Stachel et al., 1993), pBSflh (EcoRI, T7; Talbot et al., 1995), pCS2+sqt (EcoRI, T7; Feldman et al., 1998) and pBScyclops (NotI, T7; Rebagliati et al., 1998b). In situ hybridization with a single probe was performed as described (Schier et al., 1997) using the NBT-BCIP substrate (Boehringer Mannheim). Bicolor in situ hybridization was performed as described in Jowett and Yan (1996). The gsc-digoxigenin and flh-fluorescein probes were added to the hybridization mix simultaneously. The flh probe was detected using the anti-fluorescein-AP Fab fragments and developed with the NBT-BCIP substrate. After inactivation of the alkaline phosphatase, the gsc probe was detected with the anti-digoxigenin-AP Fab fragments and developed with the INT-BCIP substrate (Boehringer Mannheim). More background staining was observed when INT-BCIP was used as a substrate. Embryos were mounted in glycerol.

For plastic sections, embryos from an oep^tz57 heterozygous intercross were first processed by whole-mount bicolor in situ hybridization or by hybridization with the gsc probe alone, and mutant embryos were identified by the absence of gsc expression. Stained embryos were dehydrated in a graded series of ethanol, which dissolves the INT-BCIP substrate, and embedded in Polybed resin (Ted Pella, Inc.) in trapezoid molds (Pelco, Inc.). 5 ×m sagittal sections were cut on a Leica Ultracut UCT microtome.

The following plasmids were linearized and sense capped mRNA was synthesized using the mMESSAGE mMACHINE system (Ambion): pcDNA3oep^m134 (NotI, T7; Zhang et al., 1998), pCS2+sqt (NotI, SP6; Feldman et al., 1998), pCS2+cyc (NotI, SP6; Rebagliati et al., 1998), pCS2+lacZ (NotI, SP6), or pCS2+antivin (NotI, SP6; Thisse and Thisse, 1999). In vitro synthesized mRNA was injected into the yolk of 1-to 8-cell-stage wild-type embryos and embryos were fixed at 50% epiboly. For oep^m134 injections, 100 pg of mRNA was microinjected into the yolk syncitial layer (YSL) of embryos from an oep^tz57 homozygous intercross at blastula stages. Embryos were phenotypically analyzed and documented at 24-30 hours after fertilization. Genotyping was not required, as 100% of the progeny were maternal-zygotic oep mutants.

To determine when distinct fate domains arise for the prechordal plate and notochord, we performed fate mapping in wild-type blastulae. 1-to 4-cell-stage embryos were filled with caged fluorescein-dextran dye and left to develop to blastula stage when distinct cellular tiers are discernible from vegetal to animal positions, with tier 1 corresponding to the first row of cells at the blastoderm margin and tiers 2 and higher residing further from the margin.

We labelled rows of 15-20 cells within two tiers by laser-assisted uncaging and monitored their position and fate at later stages. We found that at dome stage (about 90 minutes before gastrulation) and 30% epiboly (about 60 minutes before gastrulation), both prechordal plate and notochord precursors can be found at the dorsal margin (data not shown). In contrast, distinct fate domains for the prechordal plate and notochord become apparent at 40% epiboly (about 45 minutes before gastrulation; Table 1, Fig. 1). Prechordal plate, but not trunk notochord, cells arise from tiers 1-2, and are later found in the hatching gland, head mesoderm and pharyngeal endoderm (Table 1; Fig. 1A-F). In contrast, notochord but not prechordal plate cells arise from tiers 6 and above (Table 1; Fig. 1G-L). Both prechordal plate and notochord cells can arise from tiers 3-5 (Table 1). When tiers 6 and above were labeled, cells began to contribute to neural fates as well as to the notochord, suggesting that the mesodermal-neural boundary in our fate map is located around the level of tier 6 (Table 1, Fig. 1L). Our data indicate that prechordal plate progenitors can be located within tiers 1-5, while notochord precursors can be found within tiers 3 and above at 40% epiboly (Table 1). These results establish that cells at the dorsal blastoderm margin become allocated to partially separate prechordal plate and notochord fate domains within the relatively short time window between dome stage and 40% epiboly.

To determine how cell fates correlate with molecular specification within the organizer, we examined the expression of gsc and flh. At dome stage, both gsc and flh are expressed at the dorsal margin in cell tiers 1-5, with flh expression extending to tier 6 (Fig. 3C,D). In contrast, at 40% epiboly flh expression becomes weaker near the margin and is predominantly found within tiers 3-9 (Fig. 2C,I), while gsc expression is still found within tiers 1-5 (Fig. 2A,I). By 50% epiboly, flh is no longer expressed at the margin and is found only within tiers 3-9 (Fig. 2G,K), while gsc is expressed in tiers 1-5 (Fig. 2E,K). Therefore, starting at 40% epiboly, the prechordal plate fate domain overlaps with the region that expresses gsc and the notochord fate domain overlaps with the region that expresses flh. Together, these data establish that anteroposterior regionalization and fate allocation are already apparent at 40% epiboly, and thus initiated by cues operating before gastrulation.

The Nodal signaling pathway has been implicated in organizer development in vertebrates (Zhou et al., 1993; Conlon et al., 1994; Feldman et al., 1998; Sampath et al., 1998; Rebagliati et al., 1998; Osada and Wright, 1999; Gritsman et al., 1999; Meno et al., 1999). In particular, the nodal-related genes cyc and sqt, and the Nodal cofactor oep, are required for prechordal plate and notochord development in zebrafish (Feldman et al., 1998; Sampath et al., 1998; Rebagliati et al., 1998b; Gritsman et al., 1999). Furthermore, cyc and sqt can both induce gsc upon overexpression in zebrafish embryos or in Xenopus animal caps (Feldman et al., 1998; Rebagliati et al., 1998a; Sampath et al., 1998; Erter et al., 1998). These observations suggested that Nodal signaling might be involved in regionalization of the organizer.

Both sqt and cyc are expressed in dorsal mesodermal progenitors at blastula stages (Feldman et al., 1998; Rebagliati et al., 1998a, Erter et al., 1998; Sampath et al., 1998). To test whether cyc and sqt are expressed in the right place at the right time to control anteroposterior pattern at the dorsal margin, we examined their expression in wild-type embryos at dome stage, when gsc and flh expression still overlaps within tiers 1-5 (Fig. 3C,D). At this stage, cyc and sqt are only expressed within tiers 1-2 (Fig. 3A,B). By 40% epiboly, flh expression becomes repressed within tiers 1-2 (Fig. 2C,I), while gsc expression is maintained in tiers 1-5 (Fig. 2A,I). Therefore, the blastoderm margin is a local source of Nodal-related signals before the gsc and flh expression domains become distinct.

It is conceivable that different levels of Nodal signaling may lead to differential regulation of gsc and flh at the dorsal margin. As a first test of this idea, we asked how different doses of exogenous cyc or sqt would affect gsc and flh expression. mRNA encoding either cyc or sqt was microinjected into wild-type embryos at the 1-to 8-cell stage, and the injected embryos were examined for expression of gsc and flh at 50% epiboly. We found that 0.05 pg of cyc induced ectopic expression of flh but not of gsc (Fig. 4B,E,H,K). However, at 1 pg of cyc, ectopic expression of both flh and gsc was induced (Fig. 4C,F,I,L). Similar results were obtained with sqt (data not shown). These results indicate that lower levels of Nodal signaling are sufficient to induce ectopic flh, while higher levels are required to induce ectopic gsc. The fact that overexpression of cyc or sqt shows different dose-response effects on gsc and flh expression further suggests that Nodal-related signals may play a role in anteroposterior pattern formation in the organizer.

To futher understand the role of Nodal in organizer patterning, we wanted to modulate the level of Nodal signaling in vivo. Based on the dose-dependent induction of flh and gsc by different levels of Nodal signaling, we predicted that decreasing or attenuating Nodal signaling would result in the loss of gsc expression at intermediate levels of Nodal signaling and in the absence of flh expression at low levels of Nodal signaling. We used three strategies to test this prediction. First, we decreased Nodal signaling using the inhibitor Antivin. Second, we studied the effects of attenuated Nodal signaling in zygotic oep mutants, which can respond to Nodal early but not late. Third, in the reverse experiment, we reconstituted Nodal signaling at different time points by adding Oep to maternal-zygotic oep mutants at different stages.

In order to decrease the level of Nodal signaling, we first took advantage of Antivin, which is known to block Nodal and Activin signaling when overexpressed in zebrafish (Meno et al., 1999; Bisgrove et al., 1999; Thisse and Thisse, 1999; reviewed in Schier and Shen, 2000). To inhibit Nodal signaling to different extents, we microinjected increasing doses of antivin mRNA into wild-type embryos at the 1-to 8-cell stage, and then assessed gsc and flh expression at 50% epiboly. We found that decreasing Nodal signaling to intermediate levels with 2 pg of antivin efficiently blocks gsc expression, but not flh expression (Fig. 5B,E). In this case, flh is expressed ectopically in tiers 1-2 at the margin (data not shown). When Nodal signaling is decreased to very low levels with 10 pg of antivin, expression of both gsc and flh was inhibited (Fig. 5C,F). These results indicate that lower levels of Nodal signaling are required for flh expression than for gsc expression at the onset of gastrulation.

As a second strategy to modulate Nodal signaling in vivo, we analyzed the phenotype of zygotic oep mutants. Loss of zygotic oep expression in the presence of maternal Oep attenuates Nodal signaling at blastula stages (Zhang et al., 1998, Gritsman et al., 1999). This results in mutants that form a notochord, but display cyclopia and lack a prechordal plate (Schier et al., 1997; Strähle et al., 1997).

To determine whether Oep functions in anteroposterior specification at the dorsal margin, we first studied the prechordal plate and notochord fate domains by analyzing marker gene expression. In zygotic oep mutant embryos at 40% epiboly, gsc expression begins to fade in intensity but occupies the same domain as in wild-type embryos (Schier et al., 1997; Fig. 2B). In contrast to wild-type embryos, however, the flh expression domain is expanded toward the margin in zygotic oep mutants, occupying tiers 1-9 (Fig. 2D,J). At 50% epiboly and at the beginning of gastrulation, gsc is no longer expressed in zygotic oep mutants (Schier et al., 1997; Fig. 2F,O), while flh is expressed ectopically at the margin (Fig. 2H,J,L) and in the involuting hypoblast (Fig.2P). These results indicate that cells that normally express gsc and become prechordal plate in wild-type embryos express the notochord marker flh in zygotic oep mutant embryos.

To test if the attenuation of Nodal signaling in zygotic oep mutants might lead to the transformation of gsc-expressing prechordal plate precursors into flh-expressing notochord progenitors, we performed fate mapping in zygotic oep mutants at 40% epiboly (Table 1; Fig. 6). As in wild-type embryos, dorsal cells in zygotic oep mutants arising from tiers 6 and above were found in the epiblast of the early shield (Fig. 6A-C) and then contributed to the notochord (Table 1; Fig. 6D-F). Labeling within tiers 6 and above resulted in cells also adopting neural fates (Table 1; Fig. 6F), suggesting that the overall size of the region occupied by dorsal mesodermal precursors is unchanged in zygotic oep mutants. Similar to wild-type embryos, cells labeled in marginal tiers in zygotic oep mutants were present at the leading edge of the involuting hypoblast (Fig. 6G-I). However, instead of migrating anteriorly and forming a prechordal plate, these cells remained posteriorly and contributed to the notochord (Table 1; Fig. 6J-L). Cells from tiers 3-5 also contributed to the notochord (Table 1). These results establish that the organizer fate map is changed as a result of attenuation of Nodal signaling in zygotic oep mutant embryos. In particular, cells located in the prechordal plate domain contribute to the notochord. The failure of oep mutant dorsal marginal cells to adopt the correct anteroposterior pattern further supports the idea that Nodal is a key modulator of fate decisions within the presumptive organizer.

Our analysis of zygotic oep mutants indicates that attenuated Nodal signaling is sufficient to specify posterior fates near the dorsal margin, but anterior fate specification requires full activation of the Nodal signaling pathway. The extent of activation of this pathway may be determined partly by the period of time that dorsal marginal cells can respond to Nodal signaling, which depends upon the availability of Oep (Gritsman et al., 1999). Our third strategy to modulate Nodal signaling in vivo was to administer Oep at different times to maternal-zygotic oep mutants, to enable them to respond to Nodal signaling for different durations.

We previously showed that injection of mRNA encoding the secreted version of Oep termed oep^m134 (Zhang et al., 1998) can rescue maternal-zygotic oep mutants by injection into the extraembryonic yolk syncytial layer (YSL) at early blastula stages (Gritsman et al., 1999). To modulate the response of marginal cells to Nodal signaling, we injected oep^m134 mRNA at different times into the YSL of maternal-zygotic oep mutants and asked how this would affect anteroposterior patterning at the dorsal margin. First, we assessed the phenotypes of live embryos at 30 hours. Only the earliest injections at oblong stage resulted in embryos with extensive rescue (class I phenotype; Fig. 8). These embryos looked almost wild type, with two eyes and a hatching gland, suggesting that the prechordal plate was rescued (Fig. 7B,G). Later injections at sphere stage (1 hour later than oblong), dome stage (2 hours later than oblong), or 30% epiboly (2.5 hours later than oblong) resulted in partially rescued embryos with class II or class III phenotypes (Fig. 8). Class II embryos had cyclopia with no hatching gland, but had rescue of the notochord and trunk muscle (Fig. 7C,H). Class III embryos had cyclopia with no hatching gland or notochord, but had rescue of the trunk somites and blood (Fig. 7D,I,E,J). The cyclopia and lack of hatching gland in class II and III embryos suggests that the prechordal plate was not rescued in these cases.

We also analyzed gsc and flh expression in similarly injected maternal-zygotic oep embryos at 50% epiboly. Uninjected maternal-zygotic oep mutants do not express gsc (Fig. 7K) and express very reduced levels of flh (Fig. 7N) at 50% epiboly. The earliest injection of oep^m134 mRNA at the 1K stage rescued gsc expression to levels slightly lower than wild-type levels in 13/16 embryos (Fig. 7L), and rescued flh in 19/20 embryos (Fig. 7O). These embryos expressed flh throughout tiers 1-9 (data not shown), suggesting that rescue was incomplete. Injection 1 hour later at sphere stage failed to rescue gsc in 18/20 embryos (Fig. 7M), but could still rescue flh in 14/19 embryos (Fig. 7P). In this case flh was also expressed throughout tiers 1-9 (data not shown). In summary, we observed more complete rescue when we injected oep^m134 mRNA earlier, suggesting that, at the concentration we examined, sustained Nodal signaling is required for gsc expression and prechordal plate formation, whereas transient Nodal signaling is sufficient for flh expression and notochord development (Figs 7, 8).

In previous zebrafish fate maps (Kimmel et al., 1990; Shih and Fraser, 1995; Melby et al., 1996; Warga et al., 1999), the notochord and structures derived from the prechordal plate were both found to arise from the dorsal blastoderm margin, the region of the organizer. It has been unclear, however, if these two fates are already segregated before gastrulation. To locate a specific fate domain for the prechordal plate, we labeled cells before the onset of gastrulation. Fates were assigned at the end of gastrulation, when the prechordal plate is visible as a coherent structure anteriorly, and the notochord is seen more posteriorly. By labeling tiers of cells along the animal-vegetal axis at varying distances from the dorsal blastoderm margin, we were able to locate distinct but partially overlapping anterior and posterior fate domains starting at 40% epiboly, about 45 minutes before involution begins. As we never saw trunk notochord cells arising from tiers 1-2 at this stage, we can exclude the possibility that prechordal plate and notochord precursors are intermingled at the margin. Likewise, there is no intermingling between these two fates in tiers 6 and above, which contain notochord precursors but no prechordal plate precursors. There may be common progenitors for the prechordal plate and notochord within tiers 3-5, where we see an overlap between these fates. Alternatively, prechordal plate and notochord progenitors may be intermingled within this region. The overlap between these two fates within tiers 3-5 suggests that there is further refinement of the anteroposterior pattern after 40% epiboly.

The pattern that we have observed by fate mapping correlates closely with the expression of the homeobox genes gsc and flh. Current resolution does not allow us to determine overlap at the single cell level, but at 40-50% epiboly gsc is expressed closer to the blastoderm margin where the prechordal plate precursors are located, while flh is expressed farther from the margin, in the region occupied by notochord progenitors. Indeed, there might be a causal relationship between the expression of these organizer genes and the fate of cells: flh is an essential determinant of notochord development (Talbot et al., 1995, Halpern et al., 1995; Melby et al., 1996; Gont et al., 1996) and gsc can induce migratory behavior reminiscent of prechordal plate cells (Niehrs et al., 1993). Moreover, we found that flh-expressing cells at the margin of zygotic oep mutants acquire a notochord fate instead of prechordal plate fate.

In summary, our results suggest three different stages of anteroposterior pattern formation in the axial mesoderm. Initially, there is a mixed population of cells that can contribute to either prechordal plate or notochord and express both flh and gsc. Beginning at 40% epiboly, dorsal marginal cells segregate into prechordal plate and notochord fate domains and gsc-and flh-expressing regions. The complete allocation of prechordal plate and notochord cells occurs during gastrulation, resulting at the end of gastrulation in non-overlapping territories of gsc-expressing prechordal plate cells anteriorly and flh-expressing notochord cells posteriorly. Importantly, the occurrence of anteroposterior pattern prior to gastrulation demonstrates that regionalization within the organizer is an active process already initiated before gastrulation by early inductive cues, rather than a passive process enacted exclusively by gastrulation movements.

Previous studies have shown that Nodal signaling is required for the development of the entire axial mesoderm (Feldman et al., 1998; Gritsman et al., 1999; Thisse and Thisse, 1999; Bisgrove et al., 1999; Meno et al., 1999). The results presented here indicate an additional role for Nodal signaling in patterning axial mesoderm precursors in the organizer along the anteroposterior axis by differential activation of the Nodal signaling pathway. This conclusion is based on three different experimental approaches to modulate Nodal signaling in vivo. First, we overexpressed Antivin/Lefty, which acts as an inhibitor of Nodal signaling, at increasing levels in wild-type embryos (Thisse and Thisse, 1999; Bisgrove et al., 1999; Meno et al., 1999). Antivin overexpression causes cyclopia and loss of prechordal plate at low doses and eliminates all axial mesoderm at higher doses (Thisse and Thisse, 1999). We found that decreasing Nodal signaling with low doses of Antivin can abolish gsc expression at 50% epiboly, while higher doses of Antivin are needed to inhibit flh expression. These results indicate that anterior organizer fates require higher levels of Nodal signaling than posterior fates.

Second, we analyzed the fate of dorsal marginal cells in zygotic oep mutants, which attenuate Nodal signaling during blastula stages. We have shown that cells within the prechordal plate fate domain lack expression of gsc, misexpress flh and adopt a notochord fate. This early perturbation in marker gene expression indicates that the subsequent changes in cell migration and differentiation are likely to be a consequence of the misspecification of these marginal cells as notochord progenitors. To test the possibilty that the prechordal plate precursors are simply deleted and then replaced by notochord progenitors, we labeled cells within the notochord fate domain in zygotic oep mutants. We found that the notochord fate domain, as well as the boundary between mesodermal and neural progenitors, appear unchanged in zygotic oep mutants. These results show that zygotic oep mutants have the same approximate number of mesodermal progenitors as wild-type embryos, but all cells within this region express flh and can contribute to the notochord. Therefore, the dorsal marginal cells undergo a posterior fate transformation upon attenuation of Nodal signaling in the absence of zygotic Oep. This suggests that early Nodal signaling in zygotic oep mutants is sufficient for expression of flh at the dorsal margin and maintenance of notochord fates, but not sufficient to maintain prechordal plate fates and gsc expression at the margin.

Third, we modulated the activation of the Nodal signaling pathway during blastula stages by adding Oep to maternal-zygotic oep mutants at different times. We found that longer exposure to a given level of Oep can rescue both prechordal plate and notochord formation and both gsc and flh expression in these mutants, while shorter exposure can only rescue flh expression and notochord development. This result might be due to exposure to Nodal signals for different times and could also be influenced by the degree of Nodal accumulation.

Taken together, our analysis of marker gene expression and fate specification in the context of reduced Nodal signaling indicates that differential Nodal signaling is required for the specification of prechordal plate and notochord fates and for anteroposterior patterning in the organizer. In particular, our results suggest that anterior fates close to the organizer margin require high levels of Nodal signaling, whereas more posterior fates at a distance from the margin require lower levels of Nodal signaling.

How might the Nodal signaling pathway be activated differentially within the organizer to generate anteroposterior pattern? Our results can be interpreted in light of recent studies on the patterning activities of the TGFβ ligand Activin. Activin has been studied extensively as an example of a morphogen that can pattern the mesoderm (Green et al., 1992, 1997; Gurdon et al., 1994, 1995, 1996, 1998; Jones et al., 1996; McDowell et al., 1997; Dyson and Gurdon, 1998). In Xenopus animal caps, Activin can induce gsc at high concentrations and Brachyury (Xbra), a marker for the notochord and posterior mesoderm, at lower concentrations (Green et al., 1997). In addition, gsc is induced closer to the source of Activin than Xbra, suggesting that Activin protein can form a gradient by diffusion (Gurdon et al., 1994). Furthermore, it has been shown that a longer period of exposure to Activin is required to induce gsc than Xbra (Gurdon et al., 1995; Dyson and Gurdon, 1998). Detailed quantitative studies have revealed that downstream gene activation depends on the absolute number of receptors occupied by Activin per cell (Dyson and Gurdon, 1998). Importantly, higher occupancy can be achieved either by exposure to a higher concentration of Activin, or by receiving a constant amount of Activin for a longer period of time (Dyson and Gurdon, 1998). While the in vivo role of Activin is unclear (Harland and Gerhart, 1997; Matzuk et al., 1995; Schulte-Merker et al., 1994), these results suggest a model wherein a localized source of a TGFβ ligand could pattern surrounding tissue in a concentration-and time-dependent fashion.

Several observations suggest that a similar mechanism may explain the cell fate decisions in response to Nodal signaling. First, Nodal-related ligands are thought to signal via a similar intracellular pathway as Activin (Oh and Li, 1997; Whitman, 1998; Waldrip et al., 1998; Weinstein et al., 1998; Nomura and Li, 1998; Gu et al., 1998; Gritsman et al., 1999; Meno et al., 1999; Song et al., 1999; reviewed in Schier and Shen, 2000). Second, ectopic expression of nodal-related genes can cause similar phenotypes and dose responses as Activin in Xenopus and zebrafish embryos (Jones et al., 1995; Joseph and Melton, 1997; Rebagliati et al., 1998a; Erter et al., 1998; Sampath et al., 1998). Third, we have shown that overexpression of the zebrafish nodal-related genes cyc and sqt has a dose-dependent effect on axial mesoderm markers: ectopic flh is induced at low levels of cyc or sqt expression, while ectopic gsc is induced only at higher levels. Fourth, cyc and sqt are both expressed at the blastoderm margin at dome stage, before the expression domains of gsc and flh become distinct. This localized expression could allow cells close to the blastoderm margin to experience higher concentrations of Nodal-related protein. In the simplest model, these results and the morphogen-like properties of Activin suggest that during normal development high levels of Nodal signals close to the margin allow expression of gsc, whereas lower levels in higher tiers allow for maintenance of flh expression. In this model, loss of Nodal signals would lead to a loss of gsc and flh expression (as observed in cyc;sqt mutants and maternal-zygotic oep mutants), whereas a decrease in Nodal signaling would block gsc but not flh expression (as observed in zygotic oep mutants and upon expression of low doses of Antivin).

While our results are consistent with the idea that Cyc and Sqt could act as morphogens to pattern the organizer along the anteroposterior axis, the range of action of Cyc and Sqt in vivo has yet to be determined. For instance, it has been shown that the Nodal-related protein Xnr2 can only act at short range in Xenopus ectodermal explants (Jones et al., 1996). It is thus still possible that the differential regulation of gsc and flh is not mediated directly by different concentrations of Nodal signals, but indirectly by relay factors that are regulated by Nodal signaling.

Interestingly, our results suggest that organizer patterning might not only be dependent on different concentrations of Nodal signals, but also on the length of exposure to these signals. In particular, the attenuation of Nodal signaling at blastula stages in zygotic oep mutants causes a transformation of prechordal plate precursors into notochord progenitors. Furthermore, in our partial rescue experiment using maternal-zygotic oep mutants, transient exposure to Nodal signaling can only rescue the notochord, while sustained exposure is required to rescue the prechordal plate. These results suggest that cell fate decisions are influenced by the duration of signaling.

Consistent with this observation, different exposure times to Activin can also result in different cellular responses (Gurdon et al., 1995; Dyson and Gurdon, 1998). Furthermore, it has been shown in the case of tyrosine kinase signaling that transient versus sustained activation of this signaling pathway leads to distinct cellular responses (reviewed in Marshall, 1995). We thus propose that both Nodal activity at a given time and the time of exposure to Nodal signaling are critical for fate decisions in dorsal marginal cells. The extent of activation of the Nodal signaling pathway within a cell would thus depend upon several factors: the concentration of Nodal protein, the distance of responding cells from the source of Nodal, the length of time exposed to Nodal signals, the concentration of inhibitory proteins such as Antivin, and the availabilty of Oep. The interplay of these factors would result in high Nodal signaling close to the margin, causing specification of prechordal plate progenitors, and lower Nodal signaling in higher tiers, resulting in specification of notochord progenitors.

We have shown that differential Nodal signaling plays an important role in animal-vegetal patterning on the dorsal side of the embryo. In the zebrafish and Xenopus blastula, there is also an animal-vegetal pattern in lateral and ventral regions, with endoderm progenitors located vegetally and most mesoderm precursors located more animally (Dale and Slack, 1987; Warga and Nusslein-Volhard, 1999). In zebrafish, Nodal signaling is required for endoderm formation (Schier et al., 1997; Strahle et al., 1997; Feldman et al., 1998; Gritsman et al., 1999; Rodaway et al., 1999; Alexander and Stainier, 1999) and evidence from Xenopus animal cap assays suggests that higher levels of Nodal signaling are needed to induce endodermal markers than mesodermal markers (Piccolo et al., 1999; Clements et al., 1999; Yasuo and Lemaire, 1999; Erter et al., 1998; Zorn et al., 1999). Interestingly, zygotic oep mutants, which have attenuated Nodal signaling, form most mesoderm, but lack both prechordal plate and endoderm (Schier et al., 1997; Strähle et al., 1997). These data suggest that, similar to prechordal plate formation at the dorsal margin, high levels of Nodal signaling are essential for endoderm specification of marginal cells. Therefore, the essential role of Nodal signaling in anteroposterior patterning of the organizer may be just one example of a more general function for Nodal signaling in patterning the blastula along the animal-vegetal axis.

We thank M. Chao, J. Schlessinger, E. Skolnik, A. Whitty and the members of the Schier and Talbot laboratories for helpful discussions, B. Feldman and G. Fishell for help with fate mapping and microscopy, S. Burden, R. Burdine, Y. Chen, R. Lehmann and H. Sirotkin for comments on the manuscript, and S. McManus and R. Feeney for fish maintenance. This work was supported by NIH RO1GM57825 (W. S. T.), the Pew Scholars Program (W. S. T.) and NIH RO1GM56211 (A. F. S.). A. F. S. is a Scholar of the McKnight Endowment Fund for Neuroscience.