Several of the genes listed in Table 1.1 can be linked into a genetic pathway that is involved in neural tube closure. The genes Notch3, RBP-Jk and Hesl form part of the Notch signal transduction cascade. This has been most extensively studied in
Drosophila (Artavanis-Tsakonas et a l, 1995). Binding of ligand (Delta or Serrate) to the Notch receptor is thought to induce proteolytic processing of Notch, allowing
translocation of the intracellular fragment into the cell nucleus. There, it associates with Suppressor of Hairless (Su(H)), which then activates transcription of particular target genes, including the Enhancer of split (B(spl)) complex. NotchS is one of four mammalian homologues of Drosophila Notch, while RBP-Jk protein (recombination signal sequence binding protein for immunoglobulin Jk genes) is homologous to Su(H), and the Hes genes are homologues of the E(spl) complex.
Signalling through the Notch pathway acts to maintain cells in an undifferentiated state, in both Drosophila and vertebrates (Artavanis-Tsakonas et a l, 1995). Ectopic
signal, thereby inhibiting neural differentiation (Lardelli et a l, 1996). This generates an increase in the number of cells that remain in the ventricular zone, and if this defect is severe enough, it results in a failure of cranial neural tube closure. Conversely, disruption of the pathway leads to premature neurogenesis, as clearly seen in the Hesl
knockout, which exhibits premature expression of a number of markers of postmitotic neurons (Ishibashi et al, 1995). Accelerated neurogenesis in the cranial region leads to a significant depletion of the population of dividing neuronal progenitors, causing an overall reduction in proliferation, and this leads to a cranial neural tube defect.
Disruption of the RBP-Jk gene is predicted also to block Notch signalling, and indeed, the expression of a number of markers of postmitotic neurons is increased in this mutant, reflecting premature neurogenesis (De la Pompa et a l, 1997).
Three other genes {Shh, Patched and GH3) are linked as components of the Shh signal transduction pathway, which is thought to be similar to the Hedgehog pathway in
Drosophila. Hedgehog (HH) binds to the transmembrane receptor Patched (Ptc) activating the transmembrane protein Smoothened which then causes alterations in a cytoplasmic complex that involves Fused, Costal-2 and Cubitus Interruptus (Cl)
(reviewed recently by Tabin and McMahon, 1997). In the absence of HH signalling. Cl is processed to a truncated N-terminal form, which acts as a transcriptional repressor of downstream targets. When HH binds to Ptc, the proteolytic cleavage of Cl is blocked; this stabilizes the full-length form of Cl, which acts as a transcriptional activator. Shh and Patched are vertebrate homologues of Drosophila HH and Patched, and the Gli family represent homologues of CL In mice, loss of the Patched protein or
overexpression of Shh should both result in enhanced expression of the normal downstream target genes. While GUI exhibits a similar function to the full-length Cl protein, mediating transcriptional activation, Gli3 has a function more similar to the N- terminal fragment of CL Gli3 acts to repress target genes, including Shh expression, and is itself repressed by Shh signalling (Marigo er a/., 1996; Masuya gf a/., 1997). The
GU3 knockout therefore mediates a similar effect to that produced by Shh
overexpression. Stimulation of the Shh pathway acts to induce ventralization of the neural tube (see section 1.4.4.1), and is also an inhibitor of dorsal neural fold bending (Ybot-Gonzalez & Copp, unpublished). Excessive stimulation of the Shh pathway may prevent neural tube closure by inhibiting normal bending of the neural folds (see below).
Several other genes from Table 1.1 can be linked, albeit somewhat tentatively, into a bone morphogenetic protein (BMP) signalling pathway. BMPs are expressed in the surface ectoderm adjacent to the neuroepithelium prior to neural tube closure (see also section 1.4.4.3) and BMP-signalling may mediate cell cycle modulation involved in hinge point formation (Ybot-Gonzalez & Copp, unpublished). The downstream effectors of BMP signalling include the SMAD proteins and the retinoblastoma protein (Mehler et a l, 1997). Jumonji is a retinoblastoma binding protein homologue
(Takeuchi et al, 1995), so may play a role in the BMP pathway. Mf-3 is a winged helix transcription factor, and the Xenopus winged helix protein Fast-1 interacts with Smad2 proteins, to mediate gene activation in response to activin (Chen et a l, 1997). It is possible that Mf-3 may interact with Smad proteins in the mouse, therefore acting downstream in a BMP signalling pathway. AP-2 and Pax3 are upregulated in the dorsal neural tube by BMP signalling, and are therefore perhaps also part of this signalling cascade.
Figure 1.3 illustrates a model for the effects of Shh and BMP signalling in neural tube closure. In this model, BMP signalling from the surface ectoderm stimulates bending of the neural folds, while Shh signalling from the notochord has an inhibitory effect. The balance of BMP and Shh signals determines whether bending occurs. Indeed,
observations of the level of Shh expression at different rostro-caudal levels of the spinal neural tube correlate with the three modes of neurulation (Ybot-Gonzalez & Copp, unpublished). In the anterior spinal cord, Shh expression in the notochord is high, dorsolateral hinge point formation is inhibited, and the folds close in a V-shape, described as Mode 1. Further posteriorly, Shh is less strongly expressed, and it is thought that the BMP signal is able to induce DLHP formation, creating Mode 2. In caudal regions of the embryo, Shh expression is weak; BMP signalling causes bending of the entire neuroepithelium, resulting in the rounded morphology of Mode 3.
Increased signalling through the Shh pathway (as in overexpression of Shh or mutation of Ptc or Gli3) or decreased signalling through the BMP pathway (in the jumonji or Mf3 null mutants) therefore mediate similar effects, resulting in failure of fold fusion owing to decreased bending of the neural folds. However, the defects observed in these mutants only affect cranial neural tube closure, although the BMP/Shh signals are thought to act throughout the neural tube. Disruption of Shh/BMP signalling in the
trunk does not result in a failure in neural tube closure since inhibition of fold bending merely converts Mode 2 or 3 of closure to Mode 1. However, in the cranial region, bending is required to convert the biconvex neural folds into the concave morphology, and an inhibition of this bending causes failure of closure.
BMP
DLHP
Shh
Figure 1.3 Schematic diagram illustrating a model for hinge point formation in the neuroepithelium. This model is based on the unpublished work of Ybot-Gonzalez and Copp. BMPs from the surface ectoderm stimulate bending of the neural folds by modulating the cell cycle. This action is inhibited by Shh secreted from the notochord. The balance of positive and negative signals determines whether hinge point formation occurs, and overexpression of Shh or underexpression of BMPs are predicted to cause a defect in neural fold closure, by lack of fold bending.
The noggin mutant appears to generate a neural tube defect by causing ectopic signalling through the BMP pathway. Noggin antagonises BMP function, and noggin expression in the notochord is thought to block BMP signalling, to allow Shh-mediated
ventralisation of the neural tube. Embryos lacking noggin exhibit dorsalisation of the ventral midline of the neural tube, with severity increasing more posteriorly, and have ectopic BMP4 expression in the caudal region of the notochord and floor plate
(McMahon et al, 1998). It is thought that ventral midline cells receiving both Shh and BMP signals undergo cell death, causing the NTD.
A number of the remaining genes in Table 1.1 exert effects on cell proliferation and cell death, although they cannot obviously be linked into a signalling pathway. Loss of the
zinc-finger transcription factor Brcal appears to cause increased cell proliferation and excessive cell death, although this effect has not been quantitated (Gowen et al, 1996). The tumour suppressor p53 is also likely to act by changing cell proliferation, while loss of ski is associated with an increase in apoptosis (Berk et a l, 1997).
Disruptions of the actin cytoskeleton may be a common effect of three further NTD mutants. Abl interacts with actin, and the Abl/Arg double mutant displays alterations in the actin cytoskeleton (Koleske et a l, 1998). MARCKS and MRP also contain an actin binding site and have the ability to crosslink actin. Both genes are expressed in the ventricular and subventricular cells of the neuroepithelium, as well as in many other tissues, and it is possible that they mediate changes in the actin cytoskeleton following changes in protein kinase C activity (Stumpo et al, 1995; Wu et a l, 1996). However, the actin filament cytoarchitecture has not been examined in either mutant.
Finally, the ApoB knockout and RARa/ydouhXt mutant can be linked functionally, at least superficially, since both cause embryonic vitamin deficiency. Apolipoprotein B encodes a major component of the plasma lipoprotein particles, which transport
cholesterol, lipids and vitamin E in the circulation (Homanics et a l, 1995). Absence of ApoB may lead to NTD through an effect on the metabolism of fat soluble vitamins, and indeed, vitamin E deficiency in rodents causes exencephaly. The RARs are receptors for both ^W-trans retinoic acid and 9-cis retinoic acid (RA), the metabolites of vitamin A (retinol). Fetuses from dams fed a vitamin A-deficient diet develop multiple congenital abnormalities, strikingly similar to the compound RAR mutants. The abnormalities observed are likely to result from alterations in a number of developmental processes, and will undoubtedly involve a large number of downstream genes; indeed, numerous gene promoters contain potential retinoic acid response elements (RAREs) and are implicated in some of the processes listed above. These include transcription factors, cell adhesion molecules, growth factors, and growth factor receptors (Mendelsohn et al,
1994). Interestingly, the list of retinoic acid responsive genes includes PDGFRa and
Hox-al, and disruptions of these genes also have an effect on neural tube closure (Lufkin et al, 1991; Soriano, 1997).