Initial dorsal forebrain patterning is a consequence of inductive interactions between the anterior neurectoderm and the adjacent surface ectoderm (Furuta et al., 1997; Liem et al., 1995). Both of these tissues arise during gastrulation (for review see Arkell and Tam 2012) with the origin of the anterior neurectoderm already described above. Prior to gastrulation, precursor cells for the surface ectoderm are found at the future anterior side of the embryo, about half way down the embryonic portion of the epiblast (Figure 1.5A). Like the prospective neurectoderm, these cells do not pass through the primitive streak during gastrulation but differentiate in situ into surface ectoderm. During gastrulation, the cells are arranged an anterior-posterior order, but in a more lateral position than the neurectoderm cells. The progenitors of another non-neural ectoderm derivative, the neural crest cells (which gives rise to the ecto-mesenchyme and cranial ganglia in the head), are juxtaposed between the neurectoderm and surface ectoderm cells at a region known as the neural plate border (Figure 1.5B, 7.5 dpc). The entire ectoderm arises as a contiguous sheet which, in a process called neurulation, folds to form the neural tube and overlying surface ectoderm. Consequently, cells from the medial neural plate take up a ventral position and cells from the lateral neural plate
adopt a dorsal position. Once the process of neurulation is complete, the forebrain neurectoderm at the dorsal midline continues its morphogenesis and, by the combined strategies of low mitosis and high apoptosis, undergoes thinning and invagination to divide the cerebrum into two hemispheres (Figure 1.8A) (Furuta et al., 1997; Groves and LaBonne, 2014). In a secondary phase of dorsal patterning, inductive interactions between the invaginated cells instruct the differentiation of specialised midline dorsal structures. These include the choroid plexus which will secrete cerebrospinal fluid, and the cortical hem which instructs adjacent neurectoderm to differentiate into the hippocampus (Groves and LaBonne, 2014; Hébert et al., 2002).
Much of our knowledge regarding dorsal patterning of the neurectoderm comes mainly from studies of the spinal cord (reviewed in Le Dréau and Martí 2012), but aspects of forebrain neurectoderm patterning involve distinct mechanisms. For example, in the forebrain, dorsal patterning occurs in close proximity to the ANR, a source of morphogenetic signals which sits at the rostral junction of the neural and surface ectoderm (Crossley and Martin, 1995; Shimamura and Rubenstein, 1997). The establishment of forebrain dorsal-ventral pattern therefore intersects and interacts with the anterior-posterior patterning system. Additionally, the forebrain neurectoderm arises earliest of all the neurectoderm subdivisions and dorsally restricted gene expression patterns (such as that of Pax3 [Goulding et al. 1991]) are evident from the time of somite formation. Despite this, the first overt signs of dorsal differentiation in the forebrain (i.e. roof-plate thinning) occur in embryos with approximately 20 somites. In contrast, dorsal development is evident in the hindbrain 24 hours earlier when the neural crest first emerges from the dorsal neurectoderm of the 5-somite embryo (Serbedzija et al., 1990). The influences that promote dorsal differentiation in the hindbrain region must operate in concert with, or very soon after, the signals late in gastrulation that induce the anterior epiblast to adopt a neural fate. Presumably dorsal-ventral patterning of the forebrain also begins during gastrulation and, as in other A-P regions, the early forebrain neurectoderm has a dorsal character which subsequently undergoes modification to direct the secondary differentiation events of midline development. Given the timing of dorsal differentiation, it is possible that the neural crest derived mesenchyme overlying the dorsal forebrain provides instructive signals to drive invagination of the telencephalic midline (Choe et al., 2014), similar to the role of retinoic acid in the chick (Gupta and Sen, 2016). There remains much to learn about the timing and source of the various morphogenetic signals that control dorsal forebrain fate and a proposed model of this multi-step process is shown in Figure 1.8A.
Figure 1.8: Proposed model of dorsal telencephalic midline development and Zic2-associated MIHV HPE.(a) Diagrams of transverse sections of the telencephalon at 8.5 – 11.5 dpc. Step 1: due to the influence of WNT and BMP signals from the border region, the forebrain neurectoderm is initially dorsal in character, as shown by Zic2 expression throughout most of the neurectoderm. SHH signals ventrally from the PrCP to the overlying neurectoderm. Step 2:
Shh expression is established in the rostral-ventral neural midline, which overlays ventral identity information onto the neurectoderm. WNT and BMP signalling, and Zic2 expression become restricted to the dorsal half of the neurectoderm. Step 3: once neurulation is complete and the surface ectoderm overlies the neurectoderm, neural crest cells migrate in from the hindbrain and surround the dorsal neurectoderm. The dorsal-most neurectoderm cells exit the cell cycle. Step 4: The neural crest cells expand under the influence of WNT signalling and the dorsal neurectoderm cells undergo apoptosis, leading to thinning and invagination of the roof- plate. Zic2 expression initiates in the ventral midline. Step 5: BMP and WNT signalling in the invaginated tissues induce the cortical hem, choroid plexus epithelium and the hippocampus. (b) In Zic2kd/kd embryos at 9.0 - 9.5 dpc, WNT expression in the surface ectoderm is delayed.
Consequently, dorsal neurectoderm cells do not exit the cell cycle and neural crest infiltration is reduced. At 10.5 dpc the dorsal cells do not undergo apoptosis and invagination of the roof- plate does not occur. Though there are few neural crest cells, WNT expression is initiated. By 11.5 dpc, dorsal midline structures are absent or hypoplastic, resulting in MIHV. PrCP: prechordal plate, NE: neurectoderm, SE: surface ectoderm, FPl: floor-plate, PhE: pharangeal endoderm, RPl: roof-plate, NCC: neural crest cells, CH: cortical hem, ChPE: choroid plexus epithelium.
The identity of the molecular signals that instruct dorsal patterning has been elusive, perhaps because of redundancy or the iterative use of the same signalling pathway during dorsal patterning and midline structure differentiation (Figure 1.8). Members of the BMP and WNT signalling molecule families are expressed in a spatial-temporal manner consistent with a role in dorsal neural patterning. Five BMP ligands (BMP2, BMP4-7) are expressed at the future dorsal midline in the forebrain neurectoderm and surface ectoderm. Embryos null for either Bmp4 or
Bmp2 die before neurulation is complete (Winnier et al., 1995; Zhang and Bradley, 1996), whilst
Bmp4 conditional mutants develop a phenotypically normal telencephalon (Hébert et al., 2003). Similarly, no neural phenotype is seen in Bmp5, 6 and 7 mutants (Dudley et al., 1995; Kingsley
et al., 1992; Luo et al., 1995; Solloway et al., 1998). A role in the relatively late events of dorsal forebrain patterning is, however, revealed when the BMP-specific receptors are mutated in a time and tissue dependent manner. Animals that are constitutive null for Bmpr1b and in which
Bmpr1a is deleted in the telencephalon at ~ 9.0 dpc do not show the characteristic thinning of the roof-plate at 10.5 dpc that is required for dorsal hemisphere separation. Subsequently, they exhibit MIHV HPE and loss of all dorsal midline cell types (i.e. the choroid plexus and cortical hem fail to form) despite maintenance of Zic2 expression. The specification of ventral and cortical cell types, however, remain unaffected (Fernandes et al., 2007). Once the midline cells invaginate, BMP expression is initiated in the choroid plexus epithelium anlagen (Currle et al., 2005) and overexpression of a constitutively active BMPR1a transforms cortical precursors into choroid plexus cells (Panchision et al., 2001), suggesting BMP signalling induces choroid plexus cell fate. When the roof-plate is ablated prior to differentiation of the choriod plexus and cortical hem, these structures fail to form. The expression of some dorsal midline genes (but not Zic2) can be rescued in tissue explants from roof-plate-ablated embryos, however, by culture in BMP4 (Cheng et al., 2006). These experiments, therefore, establish a primary role for BMP signalling in the prevention of MIHV HPE.
Similarly, Wnt1 and Wnt3a are expressed in the future dorsal neurectoderm along the length of the axis prior to neural tube closure, and in the roof-plate following closure (Megason and McMahon, 2002; Parr et al., 1993). In the canonical WNT signalling pathway (Figure 1.6D), binding of WNT ligand to a cognate receptor complex stimulates a cascade of cytoplasmic events culminating in β-catenin nuclear entry. Nuclear β-catenin associates with transcription factors of the TCF/LEF family and converts target gene repression to activation (reviewed in Arkell et al. 2013). WNTs were primarily considered mitogenic signals for the neurectoderm until elevated WNT signalling was shown to alter progenitor gene expression along the dorsal-ventral axis, promoting the production of dorsal progenitors and suppressing ventral progenitors. WNT signalling acts via TCF/LEF dependent evolutionarily conserved enhancers to establish the dorsal domain of Gli3 expression (Yu et al., 2008). In turn, GLI3, acting as a transcriptional repressor,
inhibits the ventrally produced SHH signal (Alvarez-Medina et al., 2007). This interplay of SHH and WNT signalling may also be relevant for telencephalon patterning as evidenced by the effect of differential regulation of GLI3 upon the ventral and dorsal telencephalic neuronal subtypes generated from human embryonic stem cells (Li et al., 2009). A further role for WNT signalling at later stages of roof-plate development is revealed by conditional deletion of b-catenin in the neural crest cells abutting the telencephalic neurectoderm which causes a failure of neural crest cell expansion and of telencephalic midline invagination (Choe et al., 2014).
In the forebrain, the dorsal-ventral morphogenetic signals of SHH, BMP and WNT intersect those provide by the ANR; a source of FGF ligands (FGF8, FGF15/19, FGF17 and FGF18). When FGF ligands bind to tyrosine kinase receptors (FGFR), multiple signalling pathways are activated such as the RAS/MAPK, PLC-γ, PI3K and STAT pathways (Dailey et al., 2005) (Figure 1.6E). When FGF signalling in the telencephalon is attenuated or abolished via mutation of Fgf8, the FGF receptor
Fgfr1 alone, or other receptor combinations, mice exhibit telencephalic hypoplasia. In embryos lacking Fgfr1 and Fgfr2 in the telencephalon, the decreased size of the forebrain is attributed to reduced cell proliferation in the ventral midline along with increased apoptosis in the dorsal midline (Gutin et al., 2006; Storm et al., 2006). Defects in these animals resemble those defects associated with the ventral forms of HPE. As reviewed by Hoch et al. (2009), there is a complex series of interdependencies between the forebrain signalling centres. For example, the ventral SHH signal maintains rostral midline FGF ligand expression (Hayhurst et al., 2008; Ohkubo et al., 2002) and simultaneously, the dorsal neurectoderm expression of the repressor form of GLI3 (GLI3R) represses FGF ligand dorsally (Theil et al., 1999). Similarly, the rostral source of FGF works via the Foxg1 transcription factor (Gutin et al., 2006; Hébert and Fishell, 2008) to inhibit BMP signalling, thus maintaining rostral proliferation and preventing premature differentiation of neuronal progenitor cells (Dou et al., 2000, 1999; Hanashima et al., 2004; Shimamura and Rubenstein, 1997). Human genetics indicates that anterior FGF signalling impacts dorsal forebrain patterning (Dubourg et al., 2016) but, despite evidence that FGF signalling can control
Zic2 expression in the 9.5 dpc telencephalon (Hayhurst et al., 2008; Okada et al., 2008), it remains unclear whether FGF control of Zic2 expression plays a role in Zic2-associated MIHV. 1.9 Zic2 mutation and dorsal forebrain patterning
Partial loss-of-function alleles of murine Zic2 result in MIHV HPE, suggesting an involvement for ZIC2 protein in dorsal neural patterning and differentiation (Nagai et al. 2000). Late in gastrulation, as the neural plate is forming, Zic2 expression recedes from the posterior embryo proper and becomes restricted to the anterior neuroectoderm. By this time, Zic3 and Zic5
expression is also restricted to this region. The expression of each of these genes then subsides in the medial neural plate, becoming progressively confined to the lateral (future dorsal)
neuroectoderm and the flanking surface ectoderm (Elms et al., 2004; Houtmeyers et al., 2013). After the neural tube closes, high levels of Zic2 expression can be detected along the entire anterior-posterior extent of the neural tube including the dorsal telencephalon (roof-plate and hippocampal primordium) (Cheng et al., 2006; Okada et al., 2008) with extensive overlap in this expression domain of all Zic family members. Zic2 is therefore expressed in a manner consistent with a role in dorsal patterning and indeed, loss-of-function mutations in Zic2 lead to defects in neural crest development, a dorsal cell type of the hindbrain and spinal cord (Elms et al., 2003; Nagai et al., 2000).
Embryos homozygous for a hypomorphic allele of Zic2 in which reduced amounts of the normal transcript are produced (Nagai et al., 2000) lack a telencephalic roof-plate at mid-gestation. After neural tube closure in wild type embryos, the dorsal midline of the telencephalon immediately becomes devoid of mitotic cells and, within 24 hours, apoptosis in this tissue is noticeably higher than in the surrounding tissues (Figure 1.8A). Zic2kd/kd embryos exhibit neither
of these features and, consequently, roof-plate thinning and invagination does not occur. Subsequently, the structures that should be derived from the dorsal midline are either severely hypoplastic or absent (Figure 1.8B). At this stage of development, Wnt3a should be expressed in the dorsal midline of the forebrain and along the spinal cord until the position of the forelimb bud. In Zic2kd/kd embryos this expression is delayed such that it has only just been initiated in the
dorsal forebrain (Nagai et al. 2000). This work firmly connects ZIC2 function to the MIHV form of HPE, but many questions remain unanswered regarding the role of ZIC2 in roof-plate formation. For example, it is not known precisely when the process of dorsal patterning and roof-plate induction fails in Zic2kd/kd embryos. This may reflect a role for ZIC2 in earlier dorsal
patterning events rather than roof-plate induction, per se. It is also not clear whether the expression of Zic2 in the neurectoderm, flanking surface ectoderm or both tissues is required for roof-plate formation. Alternatively, it is possible that the documented role of ZIC2 in neural crest cell development is important since in Zic2kd/kd embryos there is an evident lack of neural
crest cells in the intrahemispheric mesenchyme. Furthermore, ZIC2 is known to be able to physically interact with transcriptional mediators of the WNT, TGF-β and SHH pathways (Fujimi
et al., 2012; Houtmeyers et al., 2016; Koyabu et al., 2001; Pourebrahim et al., 2011) and whether it does so during dorsal patterning is unknown. The experiments on Zic2kd/kd embryos highlight
that roof-plate formation and dorsal midline development are particularly sensitive to loss of ZIC2 levels, with a reduction of Zic2 expression to ~20% sufficient to generate MIHV HPE in mice (Nagai et al., 2000).