In early embryo of the vertebrates the development of the nervous system from the ectoderm as a consequence of the signal from the underlying mesoderm called the neural induction. Dorsal lip of the blastopore (DLB) in amphibian is an organizer that directs the formation of body axis and it secretes organizing molecules that induces neural tissues in ectoderm cells (1). This is evidence by the Spemann and Mangold experiment in which they used two differently pigmented amphibians (Triturus) to transplant DLB (Figure: - 1 A) from one species to the ventral side of other species (1, 4). During gastrulation the host developed second axis in the future belly region of the host. When they cut section through the second axis they have observed that the pigmented transplanted tissue (graft) was in the somites and small amount in the floor plate other than that all the rest of it just mesodermal tissue not nervous system as a results this second nervous system thought to be derived from the host embryo not from the graft (1, 4). Using the pigmented species helped them to clearly identify from which specie the nervous system formed from. This suggests that the organizer's signal of the graft lead to the induction of the nervous system (1, 4). As the DLB in the amphibian, the Henson node in amniotes and the shield of teleosts are said to be the organizers (Figure: - 1 A) which induce complete nervous system (1, 2, 4). Here I will provide evidence of what are the signaling molecules and how they lead to the neural induction.
Evidence for and against the default model
Following the search for neural inducers for almost 70 years it was discovered that when animal cap of Xenopus (frog) dissociated in to single cells and putting back together would form neural tissue. Not separating the cell would give epidermis (Figure: - 1 B). Activin (TDF-β) found to be the inducing factor for mesoderm (1, 3). Therefore the dominant negative type II version of activin blocks the formation of the mesoderm but get neural marker even without dissociation. This suggests that expression of blocking agent (antigen) could induce neural fate just like dissociation (1, 3). It later turned out that so called activin receptor at that time is a receptor for bone morphogenetic protein 4 (BMP4) which is another member of family.
The study of A. Hemmati-Brivanlou and D. Melton (1997) showed that the BMP4 can be used to change the outcome of the dissociation (3). As we already seen that if the intact animal cap was cultured its self would form epidermis and dissociating it and then reaggregating it would give neural tissue (3, 4). However if dissociating the animal cap then culturing and washing those cells in the presence of BMP4 which is member of TDF-β family would get back epidermis not neural (1, 3, 4). Therefore this suggest that BMP4 able to block the effect of dissociation and retain the epidermis. BMP4 is expressed in the entire ectoderm before gastrulation (3). After gastrulation it begins to clears from the dorsal side the region where the neural plate and the organizer will then form. When the organizer is formed, it secretes neural inducers that repress the expression of BMP4 in the future neural plate (3). Neural inducers also inhibit the signals produced by type II activin receptor at the dorsal side of embryo and form neural tissue, thus activin molecule said to be the neural inhibitor (3).
In 1992 Harland et al were looking again for mesoderm and its pattern, they have done expression screen at molecule that are expressing the organizer that can dorsalise mesoderm (10). Thus molecule that can change the pattern of the mesoderm instead of lateral mesoderm gives somites to make muscle (that's what they were looking for). In that expression they isolated another gene that encoded protein that called noggin. When they injected noggin RNA to the animal cap it produced signal for neural to form (10). Therefore they concluded that noggin mimics the effect of dissociates. Almost immediately after the discovery of noggin it turned out that there were several other molecules such as chordin, follistatin, cerebus and Xnr3 had practically the same effect of noggin (1, 4). These molecules are gene encoding proteins that can induce neural tissue (when provided in the form of RNA) in an intact animal cap by switching on the neural markers that present on the ectoderm (1, 3, 4).
All the neural inducers are expressed and secreted from the organizer and they bind to BMP4 protein preventing BMP4 from binding to its receptor in another word they inhibit the BMP4 signaling which is a neural inhibitor (3, 5). BMP4 normally binds to activin receptors. There are two types of activin receptors (type 1 & type 2) these two receptors dimerise and cross phosphorylate, once they phosphorylate they recruits smads proteins that are used for signaling (1, 5). The inhibitors work by binding to BMP4 in the intracellular medium and as a consequences prevent BMP from binding to receptor and activating it (5). The BMP4 suggested inducing epidermal formation as well as neural inhibition which occurs at the same time. In another word blocking the activity of BMP lead to the down regulation of the expression of BMP which then clears from the neural plate and that eventually gives rise to the neural fate (5). All of these findings are well-matched to the default model of neural induction which suggests that neural inducers released from the organizer could work by antagonizing the signal to from neural tissue (4).
Smads are downstream effect of BMP4; they have the same effect as BMP4 of the animal cap. Smads are messenger inside the cells that transmit the signal to the BMP bound receptor (5). BMP4 normally binds to receptor and that phosphorylate smad 1. Smad 1 then recruits co-smad which is called smad 4 (1, 5). Activating the target of BMP signaling is enough to prevent the effect of dissociation thereby mimic BMP even though extracellular domain was unaltered.
In later discovery it was found out that none of these finding fit any model. The first two inhibitors such as noggin and chordin were discovered. In 1996 it was found out that the noggin in fact not expressed in the organizer of the amphibian at all at any stage (6). It expressed in the organizer dissenters of the notochord and only starting at the stage when the node its self starts to lose its capacity to induce, thereby the other stage when the node can induce there is nothing is expressed (1, 6). Thereby the noggin is expressed in the descending of node (not in the node its self) when the node had already started to induce. Chordin is expressed at the stage when the organizer is active but is also expressed when the organizer not active (8). Thereby none of these antagonists are same as what have been described; as a result this does not fit the default model.
The experiment that was carried out in chick epiblast provides strong proof against the neural default model of BMP inhibition (9). This experiment shows that inhibition of BMP through the use of chordin or noggin is not adequate to induce neural tissue in the epidermal cells. In other word these inhibitors not enough to block the BMP signaling (they partially block) which lead to the mesoderm dorsalization as a consequences the neural induction does not take place (1, 4, 9). The very same thing thought to happen in the chick epidermal differentiation, thus the expression of the BMP2 and BMP7 in chick do not inhibit neural plate development (4). These finding in the chick lead to the dispute that BMP inhibition are not sufficient for the neural induction because of the BMP inhibitors not expressed in the correct time. Later it was discovered that FGF (Fibroblast Growth Factor) expression in chick plays role in down regulating the BMP transcript which in turn inhibits BMP for the period of the neural induction (1, 4). This findings lead to the idea that when FGF signaling is blocked in the embryo both noggin and chordin could not induce neural tissue (1, 6). Also the BMP4 inhibition is not sufficient for neural induction. As mutant mouse for chordin and noggin have nervous system, BMP4 do not inhibit neural induction therefore the patterns of expression of BMP4 does not fit the model (1, 9).
It later turn out that the timing of the induction marker is important for neural cell formation. Graft of the organizer from quail to chick showed that 5 hours exposure is required in order to produce signals from the organizer (Henson node) is required for cells to be sensitive to chordin (1). Also 13 hours exposure is need for neurons to be formed (1). After a number of years a screening was carried out for first 5 hours of grafting to identify the special actions have effect during this time (1, 7). Five hours after graft, 15 genes were identified 12 of them were novel gene. Two of the genes identified are ERNI (Early Response to Neural Induction) which was induced after 1 hour which is induced by the expression of FGF8 and Churchill which was induced after 4 - 5 hours (1, 7).
These two genes are expressed in neural plate and are induced by FGF but not by BMP antagonist. FGF8 was identified as the signal required for induction of other signals early in development (8). BMP, BMP antagonist and wnt antagonist were not identified at this time. FGF8 induces 14 genes out of the 15 genes that were identified. Unlike BMP antagonist FGF8 is expressed exactly the right time. These results indicate that FGF could be the early signals which induce early markers for the neural induction (7, 8). Also the Catherine Launay el al (1996) experiment via truncated FGF type I receptor (XFD) put forward that FGF signaling is essential for the formation of the neural tissue and also for the neutralization of animal caps to BMP antagonist (6). However recent evidence suggests that neural induction by FGFs is independent of BMP antagonism (6, 7, 8).
Addition of SU5402 which is a FGF inhibitor results in absence of neural induction as a consequences the epidermal fate is re-established which evidence that the FGF is required for neural induction (1, 4, 7). When the noggin or chordin is added to this it would give rise to neural fate (7). This in fact indicates that in chick BMP signaling is necessary for ectodermal fate. Also the dominant negative version of FGF receptor (these are soluble protein that blocks the signaling) would gives the same result as SU5402 (4). These two experimental evidences suggest that just blocking the FGF signaling is enough to abolish induction of everything by the node. However another recent study which also tested whether FGF signaling required for the formation of neural fate by adding SU5402 with noggin mRNA and they discovered that not all neural expression is dependent on the expression FGF as small amount of definitive neural marker nrp was expressed even when the FGF inhibitor was present (8).
FGF8 is expressed in the organizer (anterior part of primitive streak not in the posterior part) and it induces early neural markers such as ERNI, Churchill and SOX3 (induced at 3 hours) in the future neural plate of the graft (1, 7). ERNI and SOX3 are expressed before gastrulation at that time FGF8 is also expressed in the tissue underneath called hypoblast (7). This suggests that neural induction could also begin before gastrulation. ERNI expressed early but clears from neural plate just before SOX2 expression (1, 7).
Recent evidence proposes that β-catenin morphants express sox2 in the region of the border of blastopore but they do not express organizer, therefore sox 2 is expressed in the absence of organizer (8). However Sox2 only said to mark pre-neural state not neural tissue (8). Because FGF signals induce neural gene and it is necessary for neural development and thus expression of sox2 in the blastopore is also controlled by FGF signals (8). The experiment carried out by Andrea E. Wills (2009) using SU5402 to incubate β-catenin evidenced that the increasing level of β-catenin blocked the sox2 expression which suggest that sox2 expression is dependent on FGF signaling (8).
The default model suggests that neural fate in the ectoderm is inhibited by BMP signaling (1). Therefore the organizer release BMP antagonist (such as noggin, chordin, follistatin, cerberus and DNA) and that blocks BMP signaling in adjacent cells this as a result induce neural tissue (9). This default model is supported by the evidence of Xenopus embryo which shows that the dissociated and reaggregated animal cap gives rise to neurons which could be blocked by the injection of BMP4 protein (1, 8, 9). This default model has been accepted world wide as a device for neural induction.
However the observation in the chick embryo rose as the first evidence against the default model. This showed that BMP4 and its antagonists (follistatin, noggin and chordin) were not expressed in the organizer to fit the default model. This was evidenced by misexpressing the noggin or chordin to epiblast via grafts which as a result showed that the secreting cells did not induce neural tissue (1, 7, 8, 9). Experiment of chick epiblast also suggested that 5 hours time point in exposure to organizer is critical time for the development of the neural tissues (1, 9). ERNI and Churchill are first two identified genes which are expressed with the five hour time. The use of ERNI as a neural marker leads to the discovery of FGF which expresses all the neural markers that are expressed during the five hours (1, 7, 9). The FGF is now thought to express even before gastrulation and it is one of the main neural inducing signal that have been identified (1, 8, 9). As the sox2 and sox3 are found to be expressed in the stem cells and in pluripotent cells, these two are recently been suggested as final neural markers that has been discovered (8). As BMP antagonist and FGF expression are required for sox expression they are also suggested to be the important candidate for neural induction.
1. Claudio D. Stern (2005). Neural induction: old problem, new findings, yet more questions. Development 132, 2007-2021.
2. C. H. Waddington (1932). Induction by the primitive streak and its derivatives in the chick.
3. A. Hemmati-Brivanlou & D. Melton (1997). Vertebrate neural induction. Neurosci. 1997. 20:43-60.
4. Paul A. Wilson and Ali Hemmati-Brivanlou (1997). Vertebrate Neural Induction: Inducers, Inhibitors, and a New Synthesis. Neuron, Vol. 18, 699-710.7.
5. Ignacio Muñoz-Sanjuán and Ali H. Brivanlou (2002). Neural induction, the default model and embryonic stem cells.
6. Catherine Launay, Valérie Fromentoux, De-Li Shi and Jean-Claude Boucaut* (1996). A truncated FGF receptor blocks neural induction by endogenous Xenopus inducers. Development 122, 869-880.
7. Andrea Streit*, Alyson J. Berliner², Costis Papanayotou*, AndreÂs Sirulnik* & Claudio D. Stern*² (2000). Initiation of neural induction by FGF signalling before gastrulation. NATURE | VOL 406.
8. Andrea E. Wills a, Vivian M. Choi b, Margaux J. Bennett a, Mustafa K. Khokha c, Richard M. Harland a (2009). BMP antagonists and FGF signaling contribute to different domains of the neural plate in Xenopus. Developmental Biology 337 (2010) 335-350.
9. Claudia Linker and Claudio D. Stern (2004). Neural induction requires BMP inhibition only as a late step, and involves signals. Development 131, 5671-5681
10. William C. Smith and Richard M. Harland (1992). Expression cloning of noggin, a New Dorsalizing Factor Localized to the Spemann Organizer in Xenopus Embryos. Cell, Vol. 70, 829-840.