The origin of the Nieuwkoop center

The major clue in determining how the dorsal blastopore lip obtained its properties came from the experiments of Pieter Nieuwkoop (1969, 1973, 1977). He and his colleagues in the Netherlands demonstrated that the properties of this newly formed mesoderm were induced by the vegetal (presumptive endoderm) cells underlying them. He removed the equatorial cells (i.e., presumptive mesoderm) from a blastula and showed that neither the animal cap (presumptive ectoderm) nor the vegetal cap (presumptive endoderm) produced any mesodermal tissue. However, when the two caps were recombined, the animal cap cells were induced to form mesodermal structures such as notochord, muscles, kidney cells, and blood cells (Figure 10.21). The polarity of this induction (whether the animal cells formed dorsal mesoderm or ventral mesoderm) depended on the dorsal-ventral polarity of the endodermal (vegetal) fragment. While the ventral and lateral vegetal cells (those closer to the side of sperm entry) induced ventral (mesenchyme, blood) and intermediate (muscle, kidney) mesoderm, the dorsalmost vegetal cells specified dorsal mesoderm components (somites, notochord), including those having the properties of the organizer. The dorsalmost vegetal cells of the blastula, which are capable of inducing the organizer, have been called the Nieuwkoop center (Gerhart et al. 1989).

Figure 10.21

Summary of experiments by Nieuwkoop and by Nakamura and Takasaki (1970), showing mesodermal induction by vegetal endoderm. (A) Isolated animal cap cells become a mass of ciliated epidermis, isolated vegetal cells generate gutlike tissue, and isolated (more...)

The Nieuwkoop center was demonstrated in the 32-cell Xenopus embryo by transplantation and recombination experiments. First, Gimlich and Gerhart (Gimlich and Gerhart 1984; Gimlich 1985, 1986) performed an experiment analogous to the Spemann and Mangold studies, except that they used blastulae rather than gastrulae. When they transplanted the dorsalmost vegetal blastomere from one blastula into the ventral vegetal side of another blastula, two embryonic axes were formed (see Figure 10.11B). Second, Dale and Slack (1987) recombined single vegetal blastomeres from a 32-cell Xenopus embryo with the uppermost animal tier of a fluorescently labeled embryo of the same stage. The dorsalmost vegetal cell, as expected, induced the animal pole cells to become dorsal mesoderm. The remaining vegetal cells usually induced the animal cells to produce either intermediate or ventral mesodermal tissues (Figure 10.22). Thus, dorsal vegetal cells can induce animal cells to become dorsal mesodermal tissue.

Figure 10.22

The regional specificity of mesoderm induction can be demonstrated by recombining cells of 32-cell Xenopus embryos. Animal pole cells were labeled with fluorescent polymers so that their descendants could be identified, then combined with individual vegetal (more...)

The Nieuwkoop center is created by the cytoplasmic rotation that occurs during fertilization (see Chapter 7). When this rotation is inhibited by UV light, the resulting embryo will not form dorsal-anterior structures such as the head or neural tube (Vincent and Gerhart 1987). However, these UV-treated embryos can be rescued by transplantation of the dorsalmost vegetal blastomeres from a normal embryo at the 32-cell stage (Dale and Slack 1987; see Figure 10.11A). If eggs are rotated toward the end of the first cell cycle so that the future ventral side is upward, two Nieuwkoop centers are formed, leading to two dorsal blastopore lips and two embryonic axes (see Figure 10.10). Therefore, the specification of the dorsal-ventral axis begins at the moment of sperm entry.

The molecular biology of the Nieuwkoop center

In Xenopus, the endoderm is able to induce the formation of mesoderm by causing the presumptive mesodermal cells to express the Xenopus Brachyury (Xbra) gene. The mechanism of this induction is not well understood (see Harland and Gerhart 1997), but the Xbra protein is a transcription factor that activates the genes that produce mesoderm-specific proteins. While all the vegetal cells appear to be able to induce the overlying marginal cells to become mesoderm, only the dorsalmost vegetal cells can instruct the overlying dorsal marginal cells to become the organizer. The major candidate for the factor that forms the Nieuwkoop center in these dorsalmost vegetal cells is β-catenin.

β-catenin is a multifunctional protein that can act as an anchor for cell membrane cadherins (see Chapter 3) or as a nuclear transcription factor (see Chapter 6). In Xenopus embryos, β-catenin begins to accumulate in the dorsal region of the egg during the cytoplasmic movements of fertilization. β-catenin continues to accumulate preferentially at the dorsal side throughout early cleavage, and this accumulation is seen in the nuclei of the dorsal cells (Figure 10.23A-D; Schneider et al. 1996; Larabell et al. 1997). This region of β-catenin accumulation originally appears to cover both the Nieuwkoop center and organizer regions. During later cleavage, the cells containing β-catenin may reside specifically in the Nieuwkoop center (Heasman et al. 1994a; Guger and Gumbiner 1995).

Figure 10.23

The role of Wnt pathway proteins in dorsal-ventral axis specification. (A-D) Differential translocation of β-catenin into Xenopus blastomere nuclei. (A) Early 2-cell stage of Xenopus, showing β-catenin (orange) predominantly at the dorsal (more...)

β-catenin is necessary for forming the dorsal axis, since experimental depletion of β-catenin transcripts with antisense oligonucleotides results in the lack of dorsal structures (Heasman et al. 1994a). Moreover, the injection of exogenous β-catenin into the ventral side of the embryo produces a secondary axis (Funayama et al. 1995; Guger and Gumbiner 1995). β-catenin is part of the Wnt signal transduction pathway and is negatively regulated by the glycogen synthase kinase 3 (GSK-3; see Chapter 6). GSK-3 also plays a critical role in axis formation by suppressing dorsal fates. Activated GSK-3 blocks axis formation when added to the egg (Pierce and Kimelman 1995; He et al. 1995; Yost et al. 1996). If endogenous GSK-3 is knocked out by a dominant negative protein in the ventral cells of the early embryo, a second axis forms (Figure 10.23E).

So how can β-catenin become localized to the future dorsal cells of the blastula? Labeling experiments (Yost et al. 1996; Larabell et al. 1997) suggest that β-catenin is initially synthesized (from maternal messages) throughout the embryo, but is degraded by GSK-3-mediated phosphorylation specifically in the ventral cells. The critical event for axis determination may be the movement of an inhibitor of GSK-3 to the cytoplasm opposite the point of sperm entry (i.e., to the future dorsal cells). One candidate for this agent is the Disheveled protein. This protein is the normal suppressor of GSK-3 in the Wnt pathway (see Figure 6.23), and it is originally found in the vegetal cortex of the unfertilized Xenopus egg. However, upon fertilization, Disheveled is translocated along the microtubular array to the dorsal side of the embryo (Figure 10.24; Miller et al. 1999). Thus, on the dorsal side of the embryo, β-catenin should be stable, since GSK-3 is not able to degrade it; while in the ventral portion of the embryo, GSK-3 should initiate the degradation of β-catenin.

Figure 10.24

Model of the mechanism by which the Disheveled protein stabilizes β-catenin in the dorsal portion of the amphibian egg. (A) Disheveled (Dsh) associates with a particular set of proteins at the vegetal pole of the unfertilized egg. (B) Upon fertilization, (more...)

β-catenin is a transcription factor that can associate with other transcription factors to give them new properties. It is known that Xenopus β-catenin can combine with a ubiquitous transcription factor known as Tcf3, and that a mutant form of Tcf3 lacking a β-catenin binding domain results in embryos without dorsal axes (Molenaar et al. 1996). The β-catenin/Tcf3 complex appears to bind to the promoters of several genes whose activity is critical for axis formation. One of these genes is siamois, which is expressed in the Nieuwkoop center immediately following the midblastula transition. If this gene is ectopically expressed in the ventral vegetal cells, a secondary axis emerges on the former ventral side of the embryo, and if cortical rotation is prevented, siamois expression is eliminated (Lemaire et al. 1995; Brannon and Kimelman 1996). The Tcf3 protein is thought to inhibit siamois transcription when it binds to that gene's promoters in the absence of β-catenin. However, when the Tcf3/β-catenin complex binds to its promoter, siamois is activated (Figure 10.25; Brannon et al. 1997).

Figure 10.25

Summary of events hypothesized to bring about the induction of the organizer in the dorsal mesoderm. Cortical rotation causes the translocation of Disheveled protein to the dorsal side of the embryo. Dsh binds GSK-3, thereby allowing β-catenin (more...)

The Siamois protein is critical for the expression of organizer-specific genes (Fan and Sokol 1997; Kessler 1997). Siamois protein binds to the promoter of the goosecoid gene and activates its expression (Laurent et al. 1997). The protein product of goosecoid appears to be essential for activating numerous genes in the Spemann organizer. So one could expect that the dorsal side of the embryo would contain β-catenin, that β-catenin would allow this region to express Siamois, and that Siamois would initiate the formation of the organizer. However, Siamois alone is not sufficient for generating the organizer; another protein also appears to be critical in the activation of goosecoid and the formation of the organizer. Recent studies suggest that maximum goosecoid expression occurs when there is synergism between the Siamois protein and a vegetally expressed TGF-β signal (see Chapter 6) (Brannon and Kimelman 1996). While the cortical rotation may activate the β-catenins and allow the expression of siamois in the dorsal region of the embryo, the translation of vegetally localized messages encoding a factor of the TGF-β family may generate a protein that permits the activation of goosecoid best in the cells that will become the organizer. The TGF-β family protein in the Nieuwkoop center could induce the cells in the dorsal marginal zone above them to express some transcription factor that would also bind to the promoter of the goosecoid gene and cooperate with siamois to activate it (see Figure 10.25).

The candidates for this TGF-β factor include Vg1, VegT, and Nodal-related proteins. Each of these proteins is made in the endoderm (see Figure 5.33). Agius and colleagues (2000) have provided evidence that all of these proteins may act in a pathway, and that the critical proteins are the Nodal-related factors. When they repeated the Nieuwkoop animal-vegetal recombination experiments (see Figure 10.21) but included a specific inhibitor of Nodal-related proteins, the induction by the vegetal cells failed to occur. (The inhibitor did not inhibit Vg1, VegT, or activin.) Moreover, they found that during the late blastula stages, three Nodal-related proteins (Xnr1, Xnr2, and Xnr4) are expressed in a dorsal-to-ventral gradient in the endoderm. This gradient is formed by the activation of Xenopus Nodal-related gene expression by the synergistic action of VegT and Vg1 with β-catenin. Agius and his colleagues present a model, shown in Figure 10.26, in which the dorsally located β-catenin and the vegetally located VegT and Vg1 signals interact to create a gradient of Nodal-related proteins (Xnr1, 2, 4) across the endoderm. These Nodal-related proteins specify the mesoderm such that those regions with little or no Nodal-related protein become ventral mesoderm, those regions with some Nodal protein become lateral mesoderm, and those regions with a great deal of Nodal protein become the organizer. These Nodal-related proteins will activate the goosecoid gene, and the specific inhibitor of Nodal-related proteins prevents this activation.

Figure 10.26

Model for mesoderm induction and organizer formation by the interaction of β-catenin and TGF-β proteins. (A) At late blastula stages, Vg1 and VegT are found in the vegetal hemisphere, while β-catenin is located in the dorsal region. (more...)

The diffusible proteins of the organizer I: the BMP inhibitors

Goosecoid protein works in the nucleus. It must activate (either directly or indirectly) those genes encoding the soluble proteins that function to organize the dorsal-ventral and anterior-posterior axis. Early evidence for diffusible signals from the notochord came from several sources. First, Hans Holtfreter (1933) showed that if amphibian embryos are placed in a high salt solution, the mesoderm will evaginate rather than invaginate, and will not underlie the ectoderm. Such ectoderm is not underlain by the notochord, and it does not form neural structures. Further evidence for soluble factors came from the transfilter studies of Finnish investigators (Saxén 1961; Toivonen et al. 1975; Toivonen and Wartiovaara 1976). Newt dorsal lip tissue was placed on one side of a filter thin enough so that no processes could fit through the pores, and competent gastrula ectoderm was placed on the other side. After several hours, neural structures were observed in the ectodermal tissue (Figure 10.28). The identities of the factors diffusing from the organizer, however, took another quarter of a century to identify.

Figure 10.28

Neural structures induced in presumptive ectoderm by newt dorsal lip tissue, separated from the ectoderm by a Nucleopore filter with an average pore diameter of 0.05 mm. Anterior neural tissues are evident, including some induced eyes. (From Toivonen (more...)

Recent studies on induction have resulted in a remarkable and non-obvious conclusion: The ectoderm is actually induced to become epidermal. The agents of this induction are bone morphogenetic proteins (BMPs). The nervous system forms from that region of the ectoderm that is protected from this epidermal induction (Hemmati-Brivanlou and Melton 1997). In other words, (1) the “default fate” of the ectoderm is to become neural; (2) certain parts of the embryo induce the ectoderm to become epidermal tissue, and (3) the organizer tissues act by secreting molecules that block this induction, thereby allowing the ectoderm “protected” by these factors to become neural.

Noggin

In 1992, the first of the soluble organizer molecules was isolated. Smith and Harland (1992) constructed a cDNA plasmid library from dorsalized (lithium chloride-treated) gastrulae. RNAs synthesized from sets of these plasmids were injected into ventralized embryos (having no neural tube) produced by irradiating early embryos with ultraviolet light. Those sets of plasmids whose RNAs rescued the dorsal axis in these embryos were split into smaller sets, and so on, until single-plasmid clones were isolated whose mRNAs were able to restore the dorsal axis in such embryos. One of these clones contained the noggin gene (Figure 10.29). Smith and Harland (1992) have shown that newly transcribed (as opposed to maternal) noggin mRNA is first localized in the dorsal blastopore lip region and then becomes expressed in the notochord (Figure 10.30). Injection of noggin mRNA into 1-cell, UV-irradiated embryos completely rescued the dorsal axis and allowed the formation of a complete embryo. The mRNA sequence for the Noggin protein suggests strongly that it is a secreted protein. In 1993, Smith and his colleagues found that Noggin could accomplish two of the major functions of the Spemann-Mangold organizer: it induced dorsal ectoderm to form neural tissue, and it dorsalized mesoderm cells that would otherwise contribute to the ventral mesoderm. Noggin binds to BMP4 and BMP2 and inhibits their binding to receptors (Zimmerman et al. 1996).

Figure 10.29

Rescue of dorsal structures by Noggin protein. When Xenopus eggs are exposed to ultraviolet radiation, cortical rotation fails to occur, and the embryos lack dorsal structures (top). If such an embryo is injected with noggin mRNA, it develops dorsal structures (more...)

Figure 10.30

Localization of the noggin mRNA in the organizer tissue, shown by in situ hybridization. (A) At gastrulation, noggin message (dark areas) accumulates in the dorsal marginal zone. (B) When cells involute, noggin mRNA is seen in the dorsal blastopore lip. (more...)

Chordin and nodal-related 3

The second organizer protein found was chordin. It was isolated from clones of cDNA whose mRNAs were present in dorsalized, but not in ventralized, embryos (Sasai et al. 1994). These clones were tested by injecting them into ventral blastomeres and seeing whether they induced secondary axes. One of the clones capable of inducing a secondary neural tube contained the chordin gene. The chordin mRNA was found to be localized in the dorsal blastopore lip and later in the dorsal mesoderm of the notochord (Figure 10.31). Like Noggin, chordin binds directly to BMP4 and BMP2 and prevents their complexing with their receptors (Piccolo et al. 1996). In zebrafish, a loss-of-function mutation of chordin (the “chordino” mutant) has a greatly reduced neural plate and an enlarged region of ventral mesoderm (Hammerschmidt et al. 1996). Nodal-related-3 (Xnr-3) is synthesized by the superficial cells of the organizer and is also able to block BMP4 (Smith et al. 1995; Hansen et al. 1997).

Figure 10.31

Chordin mRNA localization. (A) Whole-mount in situ hybridization shows that just prior to gastrulation, chordin message (dark area) is expressed in the region that will become the dorsal blastopore lip. (B) As gastrulation begins, Chordin is expressed (more...)

Follistatin

The fourth organizer-secreted protein, follistatin, was found in the organizer through an unexpected result of an experiment that was looking for something else. Ali Hemmati-Brivanlou and Douglas Melton (1992, 1994) wanted to see whether the protein activin was crucial for mesoderm induction, so they constructed a dominant negative activin receptor and injected it into Xenopus embryos. Remarkably, the ectoderm of these embryos began to express neural-specific proteins. It appeared that the activin receptor (which also binds other structurally similar molecules such as the bone morphogenetic proteins) normally functioned to bind an inhibitor of neurulation. When its function was blocked, all the ectoderm became neural. In 1994, Hemmati-Brivanlou and Melton proposed a “default model of neurulation” whereby the organizer functioned by producing inhibitors of whatever was blocking neurulation. That is to say, the “normal” fate of an ectodermal cell was to become a neuron; it had to be induced to become an epidermal skin cell. The organizer somehow prevented the ectodermal cells from being induced. This model was supported by, and explained, some cell dissociation experiments that had also produced odd results. Three studies, by Grunz and Tacke (1989), Sato and Sargent (1989), and Godsave and Slack (1989) had shown that when whole embryos or their animal caps were dissociated, they formed neural tissue. This result would be explainable if the “default state” of the ectoderm was not epidermal, but neural, and the tissue had to be induced to have an epidermal phenotype. The organizer, then, would block this epidermalizing induction.

Since the naturally occurring protein follistatin binds to and inhibits activin (and other related proteins), it was hypothesized that it might be one of the factors secreted by the organizer. Using in situ hybridization, Hemmati-Brivanlou and Melton (1994) found the mRNA for follistatin in the dorsal blastopore lip and notochord.

So it appeared that there might be a neural default state and an actively induced epidermal fate. This hypothesis was counter to the neural induction model that had preceded it for 70 years. But what proteins were inducing the epidermis, and were they really being blocked by the molecules secreted by the organizer?

The leading candidate for the epidermal inducer is bone morphogenesis protein-4 (BMP4). It was known that there is an antagonistic relationship between BMP4 and the organizer. If the mRNA for BMP4 is injected into Xenopus eggs, all the mesoderm in the embryo becomes ventrolateral mesoderm, and no involution occurs at the blastopore lip (Dale et al. 1992; Jones et al. 1992). Conversely, overexpression of a dominant negative BMP4 receptor resulted in the formation of two dorsal axes (Graff et al. 1994; Suzuki et al. 1994). In 1995, Wilson and Hemmati-Brivanlou demonstrated that BMP4 induced ectodermal cells to become epidermal. By 1996, several laboratories had demonstrated that Noggin, chordin, and follistatin each was secreted by the organizer and that each prevented BMP from binding to the ectoderm and mesoderm near the organizer (Piccolo et al. 1996; Zimmerman et al. 1996; Iemura et al. 1998).

When BMP binds to ectodermal cells, it activates the expression of genes such as msx1, which induce the expression of epidermal-specific genes, while inhibiting those genes that would produce a neural phenotype (Suzuki et al. 1997b). In the mesoderm, BMP4 activates genes such as Xvent1, which give the mesoderm a ventral phenotype. Low doses of BMP4 appear to activate muscle formation; intermediate levels instruct cells to become kidney; and high doses activate those genes that instruct the mesoderm to become blood cells (Hemmati-Brivanlou and Thomsen 1995; Gawantka et al. 1995; Dosch et al. 1997). The varying doses are created by the interaction of BMP4 (coming from the ventral and lateral mesoderm) with the BMP antagonists coming from the organizer^§ (Figure 10.32).

Figure 10.32

Model for the action of the organizer. (A) BMP4 (and certain other molecules) is a powerful ventralizing factor. Organizer proteins such as Chordin, Noggin, and Follistatin block the action of BMP4. The antagonistic effects of these proteins can be seen (more...)

Thus, by 1996, there was a consensus that BMP4 was the active inducer of ventral ectoderm (epidermis) and the ventralizer of the mesoderm (blood cells and connective tissue), and that Noggin, chordin, and follistatin could prevent its function. The organizer worked by secreting inhibitors of BMP4, not by directly inducing neurons.

The diffusible proteins of the organizer II: the Wnt inhibitors

It had been thought that all the neural tissue induced by the organizer was induced to become forebrain, and that the notochord represented the most anterior portion of the organizer. However, the most anterior regions of the head and brain are underlain not by notochord, but by pharyngeal endoderm and head (prechordal) mesoderm. This “endomesoderm” constitutes the leading edge of the dorsal blastopore lip. Recent studies have shown that these cells not only induce the most anterior head structures, but that they do it by blocking the Wnt pathway as well as blocking BMP4.

Cerberus

In 1993, Christian and Moon showed that Xwnt8, a member of the Wnt family of growth and differentiation factors, inhibited neural induction. It was found to be synthesized throughout the marginal mesoderm—except in the region forming the dorsal lip. Thus, in addition to BMP4, there was a second anti-neuralizing protein being secreted from the nonorganizer mesoderm. Were there any proteins in the organizer that countered this activity?

In 1996, Bouwmeester and colleagues showed that the induction of the most anterior head structures could be accomplished by a secreted protein called Cerberus.^¶ Unlike the other proteins secreted by the organizer, Cerberus promotes the formation of the cement gland (the most anterior region of tadpole ectoderm), eyes, and olfactory (nasal) placodes. When cerberus mRNA was injected into a vegetal ventral Xenopus blastomere at the 32-cell stage, ectopic head structures were formed (Figure 10.34). These head structures were made from the injected cell as well as from neighboring cells. The cerberus gene is expressed in the pharyngeal endomesoderm cells that arise from the deep cells of the early dorsal lip, and the Cerberus protein can bind both BMPs and Xwnt8 (Glinka et al. 1997; Piccolo et al. 1999).

Figure 10.34

Cerberus mRNA injected into a single D4 (ventral vegetal) blastomere of a 32-cell Xenopus embryo induces head structures as well as a duplicated heart and liver. The secondary eye (a single cyclopic eye) and olfactory placode can be readily seen. (From (more...)

Frzb and Dickkopf

Shortly after the attributes of Cerebus were demonstrated, two other proteins, Frzb and Dickkopf, were discovered to be synthesized in the involuting endomesoderm. Frzb (pronounced “frisbee”) is a small, soluble form of Frizzled, the Wnt receptor, that is capable of binding Wnt proteins in solution (Figure 10.35; Leyns et al. 1997; Wang et al. 1997). It is synthesized predominantly in the endomesoderm cells beneath the head (Figure 10.36A). If embryos are made to synthesize excess Frzb, Wnt signaling fails to occur, and the embryos lack ventral posterior structures, becoming solely head (Figure 10.36B). The Dickkopf (German; “big head,” “stubborn”) protein also appears to interact directly with Wnt proteins extracellularly. Injection of antibodies against Dickkopf protein into the blastocoel causes the resulting embryos to have small, deformed heads with no forebrain (Glinka et al. 1998).

Figure 10.35

Xwnt8 is capable of ventralizing the mesoderm and preventing anterior head formation in the ectoderm. (A) Frzb is a protein secreted by the anterior region of the organizer. It binds to Xwnt8 before that inducer can bind to its receptor. Frzb resembles (more...)

Figure 10.36

Frzb expression and function. (A) Double in situ hybridization localizing frzb (dark blue) and chordin (brown) messages. frzb mRNA can be seen to be transcribed in the head endomesoderm of the organizer, but not in the notochord (where chordin is expressed). (more...)

Glinka and colleagues (1997) have thus proposed a new model for embryonic induction. The induction of trunk structures may be caused by the blockade of BMP signaling from the notochord. However, to produce a head, both the BMP signal and the Wnt signal must be blocked. This blockade comes from the endomesoderm, now considered the most anterior portion of the organizer.

Conversion of the ectoderm into neural plate cells

So far, we have discussed the factors that prevent the dorsal ectoderm from becoming epidermis. Obviously, once that is accomplished, other genes must transform the ectoderm into neural tissue. The key protein involved in activating the neural phenotype in the ectoderm appears to be neurogenin (Ma et al. 1996). The transcription factors that appear in the ectoderm in the absence of BMP are able to induce the expression of neurogenin, and the transcription factors (such as Msx1) induced in the ectoderm by BMP signals are able to repress neurogenin expression (Figure 10.37; see Sasai 1998). Neurogenin is itself a transcription factor, and it activates a series of genes whose products are responsible for the neural phenotype. One of the genes activated by neurogenin is the gene for NeuroD, a transcription factor that activates the genes producing the structural neural-specific proteins (Lee et al. 1995). In addition, Noggin or Cerberus can induce another transcription factor in the ectoderm. This protein is called Xenopus brain factor-2 (XBF-2), and it appears to repress the epidermal genes (Mariani and Harland 1998). By these pushes and pulls, the dorsal ectoderm is converted into neural plate tissue.

Figure 10.37

Hypothetical pathways differentiating ectoderm into epidermis or neural ectoderm. In the presence of BMP signaling, epidermalizing transcription factors are generated, leading to the activation of the pathway enabling the cell to become an epidermal keratinocyte. (more...)

The determination of regional differences

One of the most fascinating phenomena in neural induction is the regional specificity of the neural structures that are produced. Forebrain, hindbrain, and spinocaudal regions of the neural tube must all be properly organized in an anterior-to-posterior direction. The organizer tissue not only induces the neural tube, but also specifies the regions of the neural tube. This region-specific induction was demonstrated by Hilde Mangold's husband, Otto Mangold (1933). He transplanted four successive regions of the archenteron roof of late-gastrula newt embryos into the blastocoels of early-gastrula embryos (Figure 10.39). The most anterior portion of the archenteron roof induced balancers and portions of the oral apparatus (Figure 10.39A); the next most anterior section induced the formation of various head structures, including nose, eyes, balancers, and otic vesicles (Figure 10.39B); the third section induced the hindbrain structure (Figure 10.39C); and the most posterior section induced the formation of dorsal trunk and tail mesoderm^‖ (Figure 10.39D). Moreover, when dorsal blastopore lips from early salamander gastrulae were transplanted into other early salamander gastrulae, they formed secondary heads. When dorsal lips from later gastrulas were transplanted into early salamander gastrulae, however, they induced the formation of secondary tails (Figure 10.40; Mangold 1933). These results show that the first cells of the organizer to enter the embryo induce the formation of brains and heads, while those cells that form the dorsal lip of later-stage embryos induce the cells above them to become spinal cords and tails.

Figure 10.39

Regional specificity of induction can be demonstrated by implanting different regions (color) of the archenteron roof into early Triturus gastrulae. The resulting embryos develop secondary dorsal structures. (A) Head with balancers. (B) Head with balancers, (more...)

Figure 10.40

Regionally specific inducing action of the dorsal blastopore lip. (A) Young dorsal lips (which will form the anterior portion of the organizer) induce anterior dorsal structures when transplanted into early newt gastrulae. (B) Older dorsal lips transplanted (more...)

Molecular correlates of neural caudalization

In the 1950s, evidence accumulated for the existence of two gradients in amphibian embryos: a dorsal gradient of “neuralizing” activity and a caudal gradient of posteriorizing (“mesodermalizing”) activity (Nieuwkoop 1952; Toivonen and Saxén 1955). The neuralizing activity came from the organizer and induced the ectoderm to be neural. The posteriorizing activity originated in the posterior of the embryo and weakened anteriorly. Recent studies have extended this model and provided candidates for the posteriorizing molecules. As predicted, the two signaling systems—neuralizing and posteriorizing—work independently (Kolm and Sive 1997). Chordin, Noggin, and the other molecules discussed above constitute the neuralizing factors secreted by the organizer. The candidates for the posteriorizing factor include eFGF, retinoic acid, and Wnt3a.

Retinoic acid

Retinoic acid is also likely to play a role in posteriorizing the neural tube. If Xenopus neurulae are treated with nanomolar to micromolar concentrations of retinoic acid (RA), their forebrain and midbrain development is impaired in a concentration-dependent fashion (Figure 10.41; Papalopulu et al. 1991; Sharpe 1991). When lower concentrations are used, the actual induction of neural tissue does not appear to be inhibited, but fewer forebrain messages and structures are produced (Durston et al. 1989; Sive et al. 1990). Retinoic acid appears to affect both the mesoderm and the ectoderm. Ruiz i Altaba and Jessell (1991) found that anterior dorsal mesoderm from RA-treated gastrulae was unable to induce head structures in host embryos, and Sive and Cheng (1991) found that RA-treated ectoderm was unable to respond to the anterior-inducing mesoderm of untreated gastrulae. An RA gradient (tenfold higher in the posterior than in the anterior) has been detected in the dorsal mesoderm of early Xenopus neurulae (Chen et al. 1994). Like eFGF, retinoic acid has been shown to activate the expression of more posterior Hox genes (Cho et al. 1991b; Sive and Cheng 1991; Kolm and Sive 1997).

Figure 10.41

Effects of RA and FGF on the anterior-posterior axis. Late neurula embryos exposed to progressively higher concentrations of retinoic acid (and allowed to grow until the controls reached tadpole stage) lose posterior regions of the embryo (as well as (more...)

Wnt3a

Another candidate for the caudalizing factor is Xenopus Wnt3a (McGrew et al. 1995). This protein is found in the neural ectoderm of the early neurula. When ectoderm is isolated from Xenopus gastrulae but remains connected to the dorsal blastopore lip, the ectoderm develops an anterior-posterior array of neural markers. If the embryo is first injected with Xwnt3a RNA (causing the overexpression of this protein), the anterior markers are lost. The basic model of neural induction, then, looks like the diagram in Figure 10.42.

Figure 10.42

Model for the organizer function and axis specification in Xenopus gastrula. (1) BMP inhibitors from organizer tissue (dorsal mesoderm and pharyngeal mesendoderm) block the formation of epidermis, ventrolateral mesoderm, and ventrolateral endoderm. (2) (more...)

The relationship between these three pathways of posteriorization has yet to be worked out. They may play different roles in different regions of the embryo, and they may work together. Retinoic acid has its major effect on hindbrain patterning, while eFGF is more important in the regionalization of the spinal cord (Blumberg et al. 1997; Kolm et al. 1997; Godsave et al. 1998). Wnt3a may suppress anterior genes and be a permissive factor allowing the other two proteins to function (McGrew et al. 1997).

Specifying the left-right axis

While the developing tadpole looks symmetrical from the outside, there are several internal organs, such as the heart and the gut tube, that are not evenly balanced on the right and left sides. In other words, in addition to its dorsal-ventral and anterior-posterior axes, the embryo has a left-right axis. Somehow, the body must be given clues as to which half is right and which half is left.

In all vertebrates studied so far, the crucial event in left-right axis formation is the expression of a nodal gene in the lateral plate mesoderm on the left side of the embryo. In Xenopus, this gene is Xnr-1 (Xenopus nodal-related-1). If the expression of this gene is also permitted to occur on the right-hand side, the position of the heart (which is normally on the left side) and the coiling of the gut are randomized.

But what causes the expression of Xnr-1 solely on the left-hand side? In Xenopus, it is possible that the first clue is given at fertilization. The microtubules involved in cytoplasmic rotation appear to be crucial, since if their formation is blocked, no left-right axis appears (Yost 1998). Moreover, the Vg1 protein, which appears to be expressed throughout the vegetal hemisphere, seems to be processed into its active form predominantly on the left-hand side of the embryo. Injecting active Vg1 protein into the left vegetal blastomeres has no effect, but adding it to the right vegetal blastomeres leads to the expression of Xnr-1 in both the right and left lateral plate and to the randomization of heart and gut positions (Hyatt et al. 1996). Moreover, when active Vg1 protein is injected into a particular right-side vegetal blastomere (the third vegetal cell to the right of the dorsal midline), the entire left-right axis is inverted. No other TGF-β family member is able to accomplish this reversal (Hyatt and Yost 1998). It is not yet known how the events at fertilization lead to the expression of Xnr-1 in the left lateral plate mesoderm during gastrulation.

The pathway by which the Xnr-1 protein instructs the heart and gut to fold properly is also unknown, but one of the key genes activated by Xnr-1 appears to be pitx2. Since this gene is activated by Xnr-1, it is normally expressed only on the left side of the embryo. However, if the Pitx2 protein is injected into the right side, too, the placement of the heart and the coiling of the gut are randomized (Figure 10.43; Ryan et al. 1998). Pitx2 persists on the left side of the embryo as the heart and gut develop, controlling their respective positions. Pitx2 may be at the “heart of the heart” (Strauss 1998).

Figure 10.43

Pitx2 determines the direction of heart looping and gut coiling. (A) Wild-type (stage 45) Xenopus tadpole viewed from the ventral side, showing rightward heart looping and counterclockwise gut coiling. (B) If an embryo is injected with Pitx2 so that this (more...)

We are finally putting names to the “agents” and “soluble factors” of the experimental embryologists. We are finally delineating the intercellular pathways of paracrine factors and transcription factors that constitute the first steps in the processes of organogenesis. The international research program initiated by Spemann's laboratory in the 1920s is reaching toward a conclusion. But this research has found levels of complexity far deeper than Spemann would have conceived, and just as Spemann's experiments told us how much we didn't know, so today, we are faced with a whole new set of problems generated by our solutions to older ones.

Surveying the field in 1927, Spemann remarked:

We still stand in the presence of riddles, but not without hope of solving them. And riddles with the hope of solution—what more can a scientist desire?

The challenge still remains.