branchless

EVOLUTIONARY HOMOLOGS

Table of contents

FGFs and neural induction

During neural induction, the 'organizer' of the vertebrate embryo instructs neighboring ectodermal cells to become nervous system rather than epidermis. This process is generally thought to occur around the mid-gastrula stage of embryogenesis. The isolation of ERNI, an early response gene to signals from the organizer (Hensen's node), is reported in this study. Using ERNI as a marker, evidence is provided that neural induction begins before gastrulation -- much earlier in development than previously thought. The organizer and some of its precursor cells produce a fibroblast growth factor signal, which can initiate, and is required for, neural induction (Streit, 2000).

The prevailing model for neural induction suggests that cells differentiate into neural fates by default, but are normally inhibited by bone morphogenetic proteins (BMPs). The organizer, by emitting BMP antagonists, allows cells in its vicinity to execute their default neural program. However, other work suggests a more complex mechanism. In the chick embryo, naive epiblast cells do not respond to BMP antagonists unless previously exposed to organizer signals for five hours. A differential screen has been designed for genes that are induced in the epiblast by a grafted organizer within this time period. Among the complementary DNAs isolated is the gene ERNI (for early response to neural induction); it is not homologous to any known sequence and contains a putative coiled-coil domain and tyrosine phosphorylation site. When transfected into COS cells, ERNI protein is found throughout the cytoplasm in most cells, but is restricted to the nucleus in about 10% of cells, which are invariably smaller and fibroblast-like. The predicted structure and subcellular localization suggest that ERNI could be part of a protein complex that travels from the cytoplasm to the nucleus (Streit, 2000).

Induction is defined as "an interaction between two tissues, as a result of which the responding cells change their fate." To ensure that the responses to the grafted organizer are due to induction, rather than recruitment of cells from the neural plate, the screen was designed using the extra-embryonic region, which does not normally contribute to the nervous system. Therefore any gene identified in this screen that is relevant to neural induction should be expressed at some stage in the prospective neural plate of the normal embryo. Indeed, cERNI begins to be expressed at pre-primitive streak stages, throughout the region that will contribute to the nervous system; by streak stages, its distribution coincides with the known limits of the prospective neural plate. Shortly thereafter, expression clears from the center of the neural plate and becomes confined to its border; transcripts disappear by early somite stages. A quail Hensen's node induces cERNI expression in chick extra-embryonic epiblast in as little as 1-2 h. By 5h cERNI induction is most intense, and by 8h it begins to clear from the center, forming a ring resembling its normal expression at the border of the neural plate. These findings make cERNI the earliest known marker for a response to organizer signals, even earlier than Sox3 (whose early expression it resembles and which is induced by the node in 3 h (Streit, 2000).

Which signaling molecules from the organizer are responsible for inducing cERNI? One approach to identifying candidate factors is to assess the ability of embryonic tissues to induce cERNI, to map the distribution of inducing factors. Head process, notochord, prechordal mesendoderm (stage 5; 14/14) and presomitic mesoderm all induce cERNI, whereas other tissues tested have either reduced or no ability to generate ectopic cERNI expression. This distribution of inducing ability is reminiscent of sites of fibroblast growth factor (FGF) activity. Indeed, FGF8-coated beads induce cERNI expression as strongly and as quickly as does the node, within 1-2 h, without inducing the mesodermal markers brachyury or Tbx6L. In contrast, ectopic expression of cERNI is never observed after misexpression of the BMP antagonists chordin, noggin or cerberus, or of BMP4 or hepatocypte growth factor/scatter factor (HGF/SF). FGF8 also strongly induces the expression of Sox3, but not the later neural marker Sox-2. Together, these findings implicate FGFs as possible early signals in the neural induction cascade. Of the members of this family, FGF8 is the best candidate endogenous inducer because at primitive streak stages it is expressed in the anterior part of the streak including the node, and is downregulated as the node begins to lose neural inducing ability (Streit, 2000).

Is FGF expression in Hensen's node required to induce cERNI and  Sox3? Two different loss-of-function approaches have been used: the FGF-receptor inhibitor SU5402, which specifically interferes with the FGF signaling pathway, and cells secreting the extracellular portion of the FGF receptor. SU5402 greatly reduced the frequency of induction by a grafted node of Sox3 and of cERNI. Moreover, cells secreting chimaeric FGF receptor markedly reduced induction by the node of Sox3 and of cERNI. A marked reduction in the level of expression of both cERNI and Sox3 occurs in the normal neural domain of the host embryo in the presence of FGF inhibitors (Streit, 2000).

These findings suggest that neural induction is initiated before the beginning of gastrulation by FGF emanating from a population of organizer precursors at the posterior margin of the embryo (perhaps reinforced by the spreading hypoblast). The coming together of this cell population with a second precursor population in the epiblast generates a fully functional organizer that provides the remaining signals in a cascade, including BMP antagonists. These results provide an explanation for the hitherto unexplained finding that in Xenopus, BMP antagonists do not induce neural tissue in the presence of dominant-negative FGF receptors and for controversial reports of FGF as a direct neural inducer. FGF signals are clearly not sufficient to generate a complete nervous system. But are they sufficient to sensitize the epiblast to BMP antagonists and to generate expression of later neural markers? The finding that msx1 is upregulated by FGF8 raises the possibility that this is part of a mechanism that leads to self-maintaining activation of BMP signaling, which would be a required first step if the BMPs are later to be inhibited. In contrast, neither FGF nor 5 h of signals from the node followed by BMP antagonists is sufficient to generate induction of Sox2 or later neural markers. It is proposed that neural induction is a multi-step process of considerable complexity. FGF mimics the first 5 h of signals from the organizer, but further steps remain to be discovered (Streit, 2000).

The expression patterns of region-specific neuroectodermal genes and fate-map analyses in zebrafish gastrulae suggest that posterior neural development is initiated by nonaxial signals, distinct from organizer-derived secreted bone morphogenetic protein (BMP) antagonists. This notion is further supported by the misexpression of a constitutively active form of zebrafish BMP type IA receptor (CA-BRIA) in the zebrafish embryos. It effectively suppressed the anterior neural marker, otx2, but not the posterior marker, hoxb1b. Furthermore, the cells in the presumptive posterior neural region lose their neural fate only when CA-BRIA and Xenopus dominant-negative fibroblast growth factor (FGF) receptors (XFD) are coexpressed. The indications are that FGF signaling is involved in the formation of the posterior neural region, counteracting the BMP signaling pathway within the target cells. The functions of Fgf3 in posterior neural development were examined. Zebrafish fgf3 is expressed in the correct place (dorsolateral margin) and at the correct time (late blastula to early gastrula stages), the same point at which the most precocious posterior neural marker, hoxb1b, is first activated. Unlike other members of the FGF family, Fgf3 has little mesoderm-inducing activity. When ectopically expressed, Fgf3 expands the neural region with suppression of anterior neural fate. However, this effect is mediated by Chordino (zebrafish Chordin), because Fgf3 induces chordino expression in the epiblast and Fgf3-induced neural expansion is substantially suppressed in dino mutants with mutated chordino genes. The results obtained in the present study reveal multiple actions of the FGF signal on neural development: it antagonizes BMP signaling within posterior neural cells, induces the expression of secreted BMP antagonists, and suppresses anterior neural fate (Koshida, 2002).

Inhibition of phosphatidylinositol (PI) 3-kinase severely attenuates the activation of extracellular signal-regulated kinase (Erk) following engagement of integrin/fibronectin receptors and Raf is the critical target of PI 3-kinase regulation. To investigate how PI 3-kinase regulates Raf, sites on Raf1 required for regulation by PI 3-kinase were examined and the mechanisms involved in this regulation were explored. Serine 338 (Ser338), which 1s critical for fibronectin stimulation of Raf1, is phosphorylated in a PI 3-kinase-dependent manner following engagement of fibronectin receptors. In addition, fibronectin activation of a Raf1 mutant containing a phospho-mimic mutation (S338D) is independent of PI 3-kinase. Furthermore, integrin-induced activation of the serine/threonine kinase Pak-1, which has been shown to phosphorylate Raf1 Ser338, is also dependent on PI 3-kinase activity, and expression of a kinase-inactive Pak-1 mutant blocks phosphorylation of Raf1 Ser338. These results indicate that PI 3-kinase regulates phosphorylation of Raf1 Ser338 through the serine/threonine kinase Pak. Thus, phosphorylation of Raf1 Ser338 through PI 3-kinase and Pak provides a co-stimulatory signal which together with Ras leads to strong activation of Raf1 kinase activity by integrins (Chaudhary, 2000).

In Xenopus embryos, fibroblast growth factors (FGFs) and secreted inhibitors of bone morphogenetic protein (BMP)-mediated signalling have been implicated in neural induction. The precise roles, if any, that these factors play in neural induction in amniotes remains to be established. To monitor the initial steps of neural induction in the chick embryo, an in vitro assay of neural differentiation in epiblast cells was developed. Using this assay, evidence was found that neural cell fate is specified in utero, before the generation of the primitive streak or Hensen's node. Early epiblast cells express both Bmp4 and Bmp7, but the expression of both genes is downregulated as cells acquired neural fate. During prestreak and gastrula stages, exposure of epiblast cells to BMP4 activity in vitro is sufficient to block the acquisition of neural fate and to promote the generation of epidermal cells. Fgf3 is expressed in the early epiblast, and ongoing FGF signalling in epiblast cells is required for acquisition of neural fate and for the suppression of Bmp4 and Bmp7 expression. It is concluded that the onset of neural differentiation in the chick embryo occurs in utero, before the generation of Hensen's node. Fgf3, Bmp4 and Bmp7 are each expressed in prospective neural cells, and FGF signalling appears to be required for the repression of Bmp expression and for the acquisition of neural fate. Subsequent exposure of epiblast cells to BMPs, however, can prevent the generation of neural tissue and induce cells of epidermal character (Wilson, 2000).

Pax6 activation occurs in phase with somitogenesis in the spinal cord. The presomitic mesoderm exerts an inhibitory activity on Pax6 expression. This repressive effect is mediated by the FGF signaling pathway. The presomitic mesoderm displays a decreasing caudorostral gradient of FGF8, and grafting FGF8-soaked beads at the level of the neural tube abolishes Pax6 activation. Conversely, when FGF signaling is disrupted, Pax6 is prematurely activated in the neural plate. It is proposed that the progression of Pax6 activation in the neural tube is controlled by the caudal regression of the anterior limit of FGF activity. Hence, as part of its posteriorizing activity, FGF8 downregulation acts as a switch from early (posterior) to a later (anterior) state of neural epithelial development (Bertrand, 2000).

FGF is capable of inducing Xenopus gastrula ectoderm cells in culture to express position-specific neural markers along the anteroposterior axis in a dose-dependent manner. However, conflicting results have been obtained concerning involvement of FGF signaling in the anterior neural induction in vivo using the same dominant-negative construct of Xenopus FGF receptor type-1 (delta XFGFR-1 or XFD) as was used in the in vitro study. This issue has been explored by employing in addition a similar construct of another receptor, XFGFR-4a, since expression of XFGFR-4a is seen to peak between gastrula and neurula stages, when the neural induction and patterning take place, whereas expression of XFGFR-1 does not have a distinct peak during that period. Further, these two FGFRs are distantly related within the Xenopus FGFR family. When mRNA of a dominant-negative version of XFGFR-4a (delta XFGFR-4a) is injected into eight animal pole blastomeres at the 32-cell stage, anterior defects including loss of normal structure in telencephalon and eye regions become prominent as examined morphologically or by in situ hybridization. Overexpression of delta XFGFR-1 appears far less effective than that of delta XFGFR-4a. Requirement of FGF signaling in ectoderm for anterior neural development was further confirmed in culture: when ectoderm cells that overexpress delta XFGFR-4a are cocultured with intact organizer cells from either early or late gastrula embryos, expression is inhibited of anterior and posterior neural markers, respectively. delta XFGFR-4 strongly suppresses autonomous neuralization of the anterior-type observed in ectoderm cells that have been subjected to prolonged dissociation. This suppression is even stronger than that exerted by delta XFGFR-1. It is thus indicated that FGF signaling in ectoderm, mainly through XFGFR-4, is required for the anterior neural induction by organizer (Hongo, 1999).

The present results seem, at first glance, inconsistent with the neural default model, recently proposed for the molecular mechanism of neural induction. This model features central roles for BMP signaling within ectoderm and for BMP antagonists, such as noggin and chordin, secreted by the organizer. BMP signaling alone induces ectoderm to form epidermis and suppresses ectoderm neuralization, whereas noggin and chordin work by locally antagonizing the BMP signaling; they act by directly binding BMP4 to prevent it from activating its receptor; this allows dorsal ectoderm to follow its default neural fate. It is argued that neural fate, specifically anterior neural fate, is the default fate of gastrula ectoderm in the sense that the neural induction, at least its initial step, requires only the absence of epidermal-inducing signals. However, the present observations indicate that the presence of FGF signaling in ectoderm is also required for neural induction. One basis for the neural default model is the finding that ectodermal cells, subjected to prolonged dissociated culture during gastrula stages, form histologically recognizable neural tissue after reaggregation. It has also been found that ectodermal explants from gastrula embryos are neuralized by expression of a dominant-negative version of a BMP receptor. In both cases, ectoderm adopts an anterior neural fate in the absence of neural inducing signals from the organizer. Thus it has been postulated that deprivation of endogenous neural-inhibiting signaling (such as BMP signaling) in ectoderm cells as a result of prolonged dissociation or overexpression of a dominant-negative receptor, is enough to cause their neuralization. However, the present studies show that autonomous neuralization in dissociated ectoderm cells requires FGF signaling. Interestingly, Xenopus gastrula ectoderm cells have been shown to express several members of the FGF family in addition to BMPs, though the level of their expression is considerably lower than that in the organizer region. It should be noted that these FGF family members have the common property of binding to components of extracellular matrix, such as heparin; these are characteristics that would make them not readily released from the cell surface, as compared to BMPs even in prolonged dissociated culture. More interesting is the fact that gastrula ectoderm cells contain novel types of ligands for the FGF receptor: FRL1 and FRL2. These ligands have an N-terminal signal sequence and can be anchored to the cell membrane by their C-termini. It is possible that some ligands of the two FGF receptors listed above support constitutive FGF signaling in ectoderm cells, contributing their neuralization without signals from the organizer. The neural default model and the present data can be reconciled by simply postulating that the default state of ectoderm is endowed with constitutive FGF signaling. According to this idea, both BMP and FGF signaling are working constitutively in intact ectoderm in either autocrine or paracrine manner (Hongo, 1999 and references therein).

Vertebrate neural development is initiated during gastrulation by the inductive action of the dorsal mesoderm (Spemann's organizer in amphibians) on neighbouring ectoderm, which eventually gives rise to the central nervous system from forebrain to spinal cord. bFGF can mimic the organizer action by inducing Xenopus ectoderm cells in culture to express four position-specific neural markers (XeNK-2, En-2, XIHbox1 and XIHbox6) along the anteroposterior axis. bFGF also induces the expression of a general neural marker NCAM but not the expression of immediate-early mesoderm markers (goosecoid, noggin, Xbra and Xwnt-8), suggesting that bFGF directly neuralizes ectoderm cells without forming mesodermal cells. The bFGF dose required to induce the position-specific markers is correlated with the anteroposterior location of their expression in vivo, with lower doses eliciting more anterior markers and higher doses more posterior markers. These data indicate that bFGF or its homolog is a promising candidate for a neural morphogen for anteroposterior patterning in Xenopus. Further, the ability of ectoderm cells to express the anterior markers in response to bFGF is lost by mid-gastrula, before the organizer mesoderm completely underlies the anterior dorsal ectoderm. Thus, an endogenous FGF-like molecule released from the involuting organizer may initiate the formation of the anteroposterior axis of the central nervous system during the early stages of gastrulation by forming a concentration gradient within the plane of dorsal ectoderm (Kengaku, 1995).

The embryonic cerebral cortex contains a population of stem-like founder cells capable of generating large, mixed clones of neurons and glia in vitro. The default state of early cortical stem cells is neuronal, and stem cells are heterogeneous in the number of neurons that they generate. In low growth factor (FGF2) concentrations, most maintain this specification, generating solely neuronal progeny. Oligodendroglial production within these clones is stimulated by a higher, threshold level of FGF2, and astrocyte production requires additional environmental factors. Because most cortical neurons are born before glia in vivo, these data support a model in which the scheduled production of cortical cells involves an intrinsic neuronal program in the early stem cells and exposure to envoronmental, glia-inducing signals. Interestingly, increased levels of FGF2 are made by differentiating cortical neurons. This could provide a feedback mechanism whereby early neuron production raises FGF2 levels, which in turn could stimulate glia generation from responsive stem cells (Qian, 1997).

The subventricular zone (SVZ) of the adult mammalian forebrain contains kinetically distinct precursor populations that contribute new neurons to the olfactory bulb. Because among forebrain precursors there are stem-like cells that can be cultured in the presence of mitogens such as epidermal growth factor (EGF) and fibroblast growth factor 2 (FGF2), it was asked whether distinct subsets of stem-like cells coexist within the SVZ or whether the proliferation of a single type of SVZ stem-like cell is controlled by several GFs. The latter is shown to be the case. Thus cells isolated from the SVZ coexpress the EGF and FGF receptors; by quantitative analysis, the number of stem-like cells isolated from the SVZ by either FGF2 or EGF is the same, whereas no additive effect occurs when these factors are used together. Furthermore, short-term administration of high-dose [3H]thymidine in vivo depletes both the EGF- and FGF2-responsive stem-like cell populations equally, showing they possess closely similar proliferation kinetics and likely belong to the constitutively proliferating SVZ compartment. By subcloning and population analysis, it has been demonstrated that responsiveness to more than one GF endows SVZ cells with an essential stem cell feature: the ability to vary self-renewal, which has been until now undocumented in CNS stem-like cells. The multipotent stem cell-like population that expands slowly in the presence of FGF2 in culture switches to a faster growth mode when exposed to EGF alone and expands even faster when exposed to both GFs together. Analogous responses are observed when the GFs are used in the reverse order, and furthermore, these growth rate modifications are fully reversible (Gritti, 1999).

FGF9 is induced during retinoic acid (RA)-induced neuronal differentiation of murine embryonal carcinoma P19 cells. FGF9 was originally isolated from a culture medium of a human glioma cell line as a growth-promoting factor for glial cells. Upon induction of neuronal differentiation by forming cell aggregates with 10(-6) M RA, the gene expression of FGF9 increases biphasically during the first 96 hours when cells are aggregating and from 168 hours to 192 hours followed by plating onto a tissue culture dish as glia-like cells proliferate. Neither undifferentiated P19 cells, nor the cells aggregated without RA which remain undifferentiated, express FGF9. This indicates that RA regulates the gene expression of FGF9 and play an important role in neuronal differentiation in both early and late stages of the developmental process (Seo, 1995).

FGF inhibits neurite outgrowth over monolayers of astrocytes and fibroblasts. FGF increases the production of arachidonic acid (AA) in cerebellar neurons, and when added directly to cultures or generated endogenously via activation of phospholipase A2, this second messenger mimics the inhibitory effect of FGF. FGF and AA can also specifically inhibit neurite outgrowth stimulated by three cell adhesion molecules (NCAM, N-cadherin and L1) expressed in transfected fibroblasts (or, in the case of L1, that binds to a tissue culture substratum). These data demonstrate that in certain cellular contexts, FGF can act as an inhibitory cue for axonal growth and that arachidonic acid is the second messenger responsible for this activity. Arachidonic acid may inhibit neurite outgrowth by desensitizing the second messenger pathway underlying neuronal responsiveness to cell adhesion molecules (Williams, 1995).

During eye development in chick embryos, optic vesicles evaginate laterally from the neural tube and develop into two bilayered eye cups, composed of an outer pigment epithelium layer and an inner neural retina layer. The anterior part of the optic vesicle normally forms the neural retina, while the posterior part of the optic vesicle gives rise to the pigment epithelium. Despite their similar embryonic origin, the pigment epithelium and neural retina differentiate into two very distinct tissues. Previous studies have demonstrated that the developmental potential of the pigmented epithelial cells is not completely restricted; until embryonic day 4.5 the cells are able to switch their phenotype and differentiate into neural retina when treated with fibroblast growth factors (FGF). FGF has gbeen found to be necessary for neural retina differentiation during the initial stages of eye cup development. Optic vesicles from embryonic day 1.5 chick were cultured for 24 hours as explants in the presence of FGF or neutralizing antibodies to FGF2. The cultured optic vesicles formed eye cups that contained a lens vesicle, neural retina and pigmented epithelium, based on morphology and expression of neural and pigmented epithelium-specific antigens. Addition of FGF to the optic vesicles causes the presumptive pigmented epithelium to undergo neuronal differentiation and, as a consequence, a double retina is formed. By contrast, neutralizing antibodies to FGF2 blocks neural differentiation in the presumptive neural retina, without affecting pigmented epithelial cell differentiation. These data, along with evidence for expression of several FGF family members and their receptors in the developing eye, indicate that members of the FGF family may be required for establishing the distinction between the neural retina and pigmented epithelium in the optic vesicle. It is hypothesized that a localized source of FGF synthesized in ectoderm overlying the optic vesicle plays a direct role in the induction of the neural retina (Pittack, 1997).

To evaluate the role of mitogen-activated protein (MAP) kinase and other signaling pathways in neuronal cell differentiation by basic fibroblast-derived growth factor (bFGF), a conditionally immortalized cell line was used from rat hippocampal neurons (H19-7). Activation of MAP kinase kinase (MEK) is insufficient to induce neuronal differentiation of H19-7 cells. To test the requirement for MEK and MAP kinase (ERK1 and ERK2), H19-7 cells were treated with the MEK inhibitor PD098059. Although the MEK inhibitor blocks the induction of differentiation by constitutively activated Raf, the H19-7 cells still undergoes differentiation by bFGF. These results suggest that an alternative pathway is utilized by bFGF for differentiation of the hippocampal neuronal cells. Expression in the H19-7 cells of a dominant-negative Ras (N17-Ras) or Raf (C4-Raf) blocks differentiation by bFGF; this suggests that Ras and probably Raf are required (See Drosophila Ras85D). Expression of dominant-negative Src (pcSrc295Arg) or microinjection of an anti-Src antibody blocks differentiation by bFGF in H19-7 cells, indicating that bFGF also signals through an Src kinase-mediated pathway. Although neither constitutively activated MEK (MEK-2E) nor v-Src is sufficient individually to differentiate the H19-7 cells, coexpression of constitutively activated MEK and v-Src induces neurite outgrowth. These results suggest that (1) activation of MAP kinase (ERK1 and ERK2) is neither necessary nor sufficient for differentiation by bFGF; (2) activation of Src kinases is necessary but not sufficient for differentiation by bFGF; and (3) differentiation of H19-7 neuronal cells by bFGF requires at least two signaling pathways activated by Ras and Src (Kuo, 1997).

The vertebrate central nervous system (CNS) contains a small group (~24,000 in human, ~3,200 in rodent, and ~7-10 in zebrafish) of evolutionary conserved noradrenergic (NA) neurons known as the locus coeruleus (LC). These neurons reside in the ventro-lateral region of the first hindbrain rhombomere and project to regions throughout the CNS. Their degeneration is associated with Parkinson's and Alzheimer's disease, whereas their abnormal function is thought to play a role in depression, sleep disorders, and schizophrenia. The zebrafish mutation soulless, in which the development of locus coeruleus noradrenergic neurons fails to occur, disrupts the homeodomain protein Phox2a. Phox2a is not only necessary but also sufficient to induce Phox2b+ dopamine-beta-hydroxylase+ and tyrosine hydroxylase+ NA neurons in ectopic locations. Phox2a is first detected in LC progenitors in the dorsal anterior hindbrain, and its expression there is dependent on FGF8 from the mid/hindbrain boundary and on optimal concentrations of BMP signal from the epidermal ectoderm/future dorsal neural plate junction. These findings suggest that Phox2a coordinates the specification of LC in part through the induction of Phox2b and in response to cooperating signals that operate along the mediolateral and anteroposterior axes of the neural plate (Guo, 1999).

The role of fibroblast growth factors (FGFs) in neural induction is controversial. Although FGF signaling has been implicated in early neural induction, a late role for FGFs in neural development is not well established. Indeed, it is thought that FGFs induce a precursor cell fate but are not able to induce neuronal differentiation or late neural markers. It is also not known whether the same or distinct FGFs and FGF receptors (FGFRs) mediate the effects on mesoderm and neural development. Xenopus embryos expressing ectopic FGF-8 develop an abundance of ectopic neurons that extend to the ventral, non-neural ectoderm, but show no ectopic or enhanced notochord or somitic markers. FGF-8 inhibits the expression of an early mesoderm marker, Xbra, in contrast to eFGF, which induces ectopic Xbra robustly and neuronal differentiation weakly. The effect of FGF-8 on neurogenesis is blocked by dominant-negative FGFR-4a (deltaXFGFR-4a). Endogenous neurogenesis is also blocked by deltaXFGFR-4a and less efficiently by dominant-negative FGFR-1 (XFD), suggesting that it depends preferentially on signaling through FGFR-4a. The results suggest that FGF-8 and FGFR-4a signaling promotes neurogenesis and, unlike other FGFs, FGF-8 interferes with mesoderm induction. Thus, different FGFs show specificity for mesoderm induction versus neurogenesis and this may be mediated, at least in part, by the use of distinct receptors (Hardcastle, 2000).

Cortical progenitor cells give rise to neurons during embryonic development and to glia after birth. While lineage studies indicate that multipotent progenitor cells are capable of generating both neurons and glia, the role of extracellular signals in regulating the sequential differentiation of these cells is poorly understood. To investigate how factors in the developing cortex might influence cell fate, a cortical slice overlay assay was developed in which cortical progenitor cells are cultured over cortical slices from different developmental stages. Embryonic cortical progenitors cultured over embryonic cortical slices differentiate into neurons and those cultured over postnatal cortical slices differentiate into glia, suggesting that the fate of embryonic progenitors can be influenced by developmentally regulated signals. In contrast, postnatal progenitor cells differentiate into glial cells when cultured over either embryonic or postnatal cortical slices. Clonal analysis indicates that the postnatal cortex produces a diffusible factor that induces progenitor cells to adopt glial fates at the expense of neuronal fates. The effects of the postnatal cortical signals on glial cell differentiation are mimicked by FGF2 and CNTF, which induce glial fate specification and terminal glial differentiation respectively. These observations indicate that cell fate specification and terminal differentiation can be independently regulated and suggest that the sequential generation of neurons and glia in the cortex is regulated by a developmental increase in gliogenic signals (Morrow, 2001).

Before the nervous system establishes its complex array of cell types and connections, multipotent cells are instructed to adopt a neural fate and an anterior-posterior pattern is established. Smad10, a Medea related member of the Smad family of intracellular transducers of TGFß signaling, is required for formation of the nervous system. In addition, two types of molecules proposed as key to neural induction and patterning, bone morphogenetic protein (BMP) antagonists and fibroblast growth factor (FGF), require Smad10 for these activities. These data suggest that Smad10 may be a central mediator of the development of the frog nervous system (LeSueur, 2002).

Two separate classes of organizer-derived secreted molecules -- FGFs and BMP antagonists -- are thought to be key to neural induction and anterior-posterior patterning. In Xenopus, FGFs induce posterior neural tissue, and FGFs are required for neural induction in the chick and spinal cord formation in the frog. Since Smad10 is necessary for formation of neural structures, including the spinal cord, it follows that FGF might require Smad10 for its neuralizing properties. To test this notion, embryos were injected with control morpholino-modified oligonucleotide (morpholinos or MO), Smad10 MO, DNS10 mRNA, or ß-galactosidase mRNA and explanted animal caps. During gastrulation, the caps were cultured in the presence of FGF under conditions that induce formation of neural tissue and then the caps were analyzed for expression of neural markers. Smad10 does not block all FGF activities, but rather is required specifically for neural induction by FGF (LeSueur, 2002).

If Smad10 transduces an RTK signal, what ligand might activate the cascade? One plausible candidate is FGF. FGFs signal via an RTK pathway that involves phosphorylation and activation of Erks. This FGF pathway has similar biological functions to Smad10; both induce posterior neural fates. Furthermore, FGF requires Smad10 for this activity. These data, coupled with the in vitro phosphorylation results and the inability of the Smad10-PXAPx3 mutant to induce spinal cord formation, are consistent with the idea that FGF initiates an RTK pathway that leads to activation of Erk, subsequent phosphorylation of Smad10, and induction of posterior neural fates. Additional biochemical experiments will be required to test this hypothesis (LeSueur, 2002).

The data suggest that a RTK pathway may regulate the function of Smad10 and may do so in a direct biochemical sense. Smad10 contains Erk consensus phosphorylation sites, Erk2 directly phosphorylates Smad10 in vitro, and the PX(S/T)P consensus phosphorylation sites on Smad10 are required for this Erk-dependent phosphorylation. Of note, when the Erk consensus sites are mutated to alanine, Smad10 remains functional and generates anterior neural fates; however, the mutant no longer produces posterior neural fates. This suggests that the nonphosphorylated and phosphorylated forms of Smad10 might interact with different subsets of transcription factors to generate distinct cell fates. Erks often phosphorylate and activate transcription factors that regulate gene expression. Smad10 may be another example of such a transcription factor. Taken together, these data suggest that an RTK signal, rather than a TGFß signal, might control Smad10's biological function in anterior versus posterior neural development (LeSueur, 2002).

In chordates, formation of neural tissue from ectodermal cells requires an induction. The molecular nature of the inducer remains controversial in vertebrates. Using the early neural marker Otx as an entry point, the neural induction pathway in the simple embryos of Ciona intestinalis was dissected. The regulatory element driving Otx expression in the prospective neural tissue was isolated; this element directly responds to FGF signaling and FGF9/16/20 acts as an endogenous neural inducer. Binding site analysis and gene loss of function established that FGF9/16/20 induces neural tissue in the ectoderm via a synergy between two maternal response factors. Ets1/2 mediates general FGF responsiveness, while the restricted activity of GATAa targets the neural program to the ectoderm. Thus, this study identifies an endogenous FGF neural inducer and its early downstream gene cascade. It also reveals a role for GATA factors in FGF signaling (Bertrand, 2003).

Otx expression starts in the animal a6.5 pair of blastomeres as they become restricted to anterior neural fate, at the onset of the neural induction process. At this stage, Otx is also activated in the animal b6.5 pair of blastomeres (precursors of the posterior dorsal neural tube and of the dorsal midline which constitutes a neurogenic region and in some vegetal B-line blastomeres (precursors of the posterior mesendoderm). Interestingly, Otx activation in b6.5, as in a6.5, requires an induction from vegetal blastomeres (Bertrand, 2003 and references therein).

The region in Otx located between -1541 and -1417 is required for expression in the a6.5 lineage, and is referred to as the a-element. Consistent with the simultaneous induction of Otx in a6.5 and b6.5 by vegetal cells, deletion of the a-element also reduces the activity in the b6.5 lineage. Finally, regions located between positions -1417 to -1133, and -706 to -271 are required for expression in A-line, and B/b-lines respectively (Bertrand, 2003).

Otx activation in the a6.5 neural precursors requires an interaction with the anterior vegetal blastomeres (A-line). Thus, the inducing FGF should be expressed in A-line blastomeres, before the onset of Otx expression at the 32-cell stage. The Ciona intestinalis genome contains 6 members of the FGF family. By in situ hybridization, only detect one FGF, FGF9/16/20, could be detected that was expressed at the right time and place to be the inducer. Its expression starts at the 16 cell-stage in the A-line and some B-line cells. Expression is stronger in the A-line than in the B-line, and this difference is further enhanced at the early 32-cell stage. This expression pattern is similar to that of the Ciona savignyi ortholog and is consistent with a role for FGF9/16/20 as endogenous neural inducer (Bertrand, 2003).

By both gene loss of function and binding sites analysis it has been determined that cooperation between the maternal transcription factors, Ets1/2 and GATAa, mediates the initial transcriptional response to FGF. Ets transcription factors have already been shown to act in the FGF pathway in vertebrates, and the members of the Ets1/2 subfamily can be directly phosphorylated and activated by Erk. A role for GATAa in this process was more unexpected, since GATA factors have so far not been implicated in the FGF pathway. However, the fact that multimerized GATA binding sites mediate FGF responsiveness indicate that, in this system, GATA does not act solely to modify or enhance Ets activity but functions as an FGF-activated transcription factor. Consistent with the proposal of a direct involvement of GATA factors in the FGF pathway in vivo, it has recently been shown, in vitro, that vertebrate GATA4 can be directly phosphorylated and activated by Erk (Bertrand, 2003).

Could members of the Ets1/2 and GATA families also play a role in neural induction in vertebrates? Ets2 messenger is present maternally in Xenopus eggs and has recently been shown to be required for the induction of Brachyury by FGF in mesodermal cells. It will be interesting to test whether it also acts in the neural induction pathway. Vertebrate GATA factors are thought to antagonize rather than promote neural tissue formation; GATA1/2/3 family members are expressed during gastrulation in the nonneural ectoderm in zebrafish, Xenopus, and chick and GATA1 has an antineuralizing activity when overexpressed in Xenopus. However, GATA2 has no antineuralizing activity, showing that this is not a general property of GATA factors. GATA2 and GATA5 are present in Xenopus eggs but the early role of these maternal GATA factors has not been studied, leaving open the possibility of an involvement in neural induction. Finally, it is proposed that, in ascidians, the use of different response factors accounts for the activation of different target genes in neuroectoderm and mesoderm. It will be interesting to test whether the same logic is used in vertebrates or whether the increase in gene number has led to the recruitment of different FGF inducers or receptors in these two lineages (Bertrand, 2003 and references therein).

FGFs and neural patterning

Three inductive interactions result in the regionalization of the mouse forebrain: (1) medial (ventral) patterning signals originating from the notochord and the more anterior precordal plate induce the primordia of the basal plate; (2) local signals arising from the anterior neural ridge (ANR), including Fgf8, induce expression of BF1, which regulates the development of specific forebrain structures such as telencephalic and optic vesicles, and (3) lateral (dorsal) patterning signals (BMPs) that arise from the non-neural ectoderm flanking the neural plate induce expression of Msx1 and patterning of the alar plate. This paper deals with the first two of these inductive interactions. Molecular properties of the medial neural plate are regulated by signals originating from the prechordal plate perhaps through the action of Sonic Hedgehog. Sonic induces homeobox gene Nkx2.1 (a homolog of Drosophila vnd) in the medial part of the mouse prosencephalic neural plate as early as the 3-somite stage, and Pax6 is expressed more laterally at similar or slighty later times. HNF3ß and not Nkx2.1 is expressed in posterior parts of explants, demonstrating that this tissue responses to Sonic and is not competent to express Nkx2.1. This suggests that the forebrain employs the same medial-lateral (ventral-dorsal) patterning mechanisms present in the rest of the central nervous system (Shimamura, 1997).

In the chick embryo, neural cells acquire midbrain, hindbrain, and spinal cord character over a ~6 hr period during gastrulation. The convergent actions of four signals appear to specify caudal neural character. Fibroblast growth factors (FGFs) and a paraxial mesoderm-caudalizing (PMC) activity are involved, but neither signal is sufficient to induce any single region. FGFs act indirectly by inducing mesoderm that expresses PMC and retinoid activity and also directly on prospective neural cells, in combination with PMC activity and a rostralizing signal, to induce midbrain character. Hindbrain character emerges from cells that possess the potential to acquire midbrain character upon exposure to higher levels of PMC activity. Induction of spinal cord character appears to involve PMC and retinoid activities (Muhr, 1999).

Are FGF and PMC activities in combination able to generate cells of midbrain and hindbrain character? On its own, stage 4 anterior primitive streak (itself a source of FGFs and PMC activity), as well as stage 4 caudal paraxial mesoderm in the presence of FGF2, induce Otx2+/En1/2+ cells and Krox20+ cells in stage 3 R explants. Both stage 4 streak and paraxial mesoderm killed by freeze/thawing, in combination with FGF2, are sufficient to induce Otx2+/En1/2+ and Krox20+ cells, suggesting that PMC activity is expressed by cells in the anterior streak by stage 4. These results suggest that PMC activity and FGF, in combination, act directly on epiblast cells to induce midbrain and hindbrain character in stage 3 R explants. They also support the idea that the differential response of stage 3 C and 3 R explants to stage 4 caudal paraxial mesoderm reflects the prior exposure of C cells to FGFs. The effects of SU5402, an inhibitor of FGF receptor signaling, were examined on the induction of midbrain and hindbrain character in caudal epiblast cells. SU5402 blocks the FGF2-mediated induction of midbrain and hindbrain character in stage 3 R epiblast explants grown with stage 4 paraxial mesoderm but does not block the induction of Isl1+/HB9+ motor neurons in stage 4 C epiblast explants exposed to Shh-N. Therefore, it was asked whether SU5402 inhibits the induction of cells of midbrain and hindbrain character in stage 3 C explants grown together with stage 4 caudal paraxial mesoderm. SU5402 completely blocks the generation of Otx2+/En1/2+ and Krox20+ cells in these conjugates, and no expression of Hoxb8 is detected. These results provide evidence that FGF signaling in caudal epiblast cells is required to induce cells of midbrain and hindbrain character. They also suggest that FGF signaling is still ongoing in stage 3 C explants grown in vitro. However, in stage 4 C explants exposed to stage 4 caudal paraxial mesoderm in the presence of SU5402, Krox20+ cells are generated, indicating that a brief period of FGF signaling is sufficient for the generation of hindbrain character. In contrast, the generation of Otx2/En1/2+ cells is still inhibited, suggesting a requirement for a more prolonged period of FGF signaling in the generation of cells of midbrain character. Stage 7 paraxial mesoderm still induces Hoxb8+ expression in stage 3 C and 3 R explants in the presence of SU5402, suggesting that the induction of spinal cord identity by paraxial mesoderm is independent of FGF signaling (Muhr, 1999).

These findings suggest that at stages 2 and 3, prospective caudal neural plate cells are specified as cells of anterior prosencephalic-like character and begin to be exposed to FGFs derived from the primitive streak. By stage 3, caudal epiblast cells that migrate through the primitive streak may be exposed to high levels of FGF and differentiate into caudal paraxial mesodermal cells, which express PMC activity. The specification of caudal neural cells appears to be initiated at ~stage 3+ when caudal epiblast cells are first exposed to PMC activity that derives from the nascent paraxial mesoderm. At this stage, prospective caudal neural plate cells acquire predominantly midbrain character, yet 2 hr later, they have also acquired the potential to generate cells of hindbrain character. The differentiation of the epiblast into cells with these two regional characters coincides with the temporal increase in the level of PMC activity and with the gradual restriction of a rostralizing (midbrain-promoting) activity to the anterior tip of the primitive streak. Changes in the level of exposure of caudal epiblast cells to these two activities between stages 3 and 4 may, therefore, contribute to the initial differentiation of cells of midbrain and hindbrain character. After another ~2-3 hr, the caudal paraxial mesoderm starts to express high levels of RALDH2 and thus acquires retinoid synthetic capacity. he combined actions of PMC activity and retinoids provided by the paraxial mesoderm appear to induce spinal cord character in caudal epiblast cells, at the expense of midbrain and hindbrain character (Muhr, 1999).

Fibroblast growth factors (Fgfs) form a large family of secreted signalling proteins that have a wide variety of roles during embryonic development. Within the central nervous system (CNS) Fgf8 is implicated in patterning neural tissue adjacent to the midbrain-hindbrain boundary. However, the roles of Fgfs in CNS tissue rostral to the midbrain are less clear. The patterning of the forebrain was examined in zebrafish embryos that lack functional Fgf8/Ace. Ace is required for the development of midline structures in the forebrain. In the absence of Ace activity, midline cells fail to adopt their normal morphology and exhibit altered patterns of gene expression. This disruption to midline tissue leads to severe commissural axon pathway defects, including misprojections from the eye to ectopic ipsilateral and contralateral targets. Ace is also required for the differentiation of the basal telencephalon and several populations of putative telencephalic neurons but not for overall regional patterning of forebrain derivatives. ace expression co-localizes with anterior neural plate cells that have previously been shown to have forebrain patterning activity. Removal of these cells leads to a failure in induction of ace expression indicating that loss of Ace activity may contribute to the phenotypes observed when anterior neural plate cells are ablated. However, since ace mutant neural plate cells still retain at least some inductive activity, then other signals must also be produced by the anterior margin of the neural plate. In ace minus embryos, there are severe defects in the establishment of both the postoptic commissure and the anterior commissure such that both commissures are usually initially absent and fused at later stages. This indicates that Ace has a crucial role in the development of commissural neuroepithelium. In support of this, alterations in expression of midline genes encoding proteins that are likely to directly influence axon extension (netrin1, sema3D), and in genes more likely to indirectly influence guidance cues (no-isthmus embryos that carry mutations in the pax2.1 gene, six3 and Tiggywinkle hedgehog) (Shanmugalingam, 2000).

Fibroblast growth factor (FGF) has been proposed to be involved in the specification and patterning of the developing vertebrate nervous system. There is conflicting evidence, however, concerning the requirement for FGF signaling in these processes. To provide insight into the signaling mechanisms that are important for neural induction and anterior-posterior neural patterning, the dominant negative Ras mutant, N17Ras, was employed, in addition to a truncated FGF receptor (XFD). Both N17Ras and XFD, when expressed in Xenopus laevis animal cap ectoderm, inhibit the ability of FGF to generate neural pattern. They also block induction of posterior neural tissue by XBF2 and XMeis3. However, neither XFD nor N17Ras inhibits noggin, neurogenin, or XBF2 induction of anterior neural markers. MAP kinase activation has been proposed to be necessary for neural induction, yet N17Ras inhibits the phosphorylation of MAP kinase that usually follows explantation of explants. In whole embryos, Ras-mediated FGF signaling is critical for the formation of posterior neural tissues but is dispensable for neural induction (Ribisi, 2000).

Posterior mesoderm tissue induces midbrain and hindbrain fates from prospective forebrain, an activity that is mimicked in explant culture by bFGF. Treatment of early gastrula age animal cap ectoderm with bFGF protein induces the expression of the spinal cord marker hoxB9. Late gastrula-age (stage 11) animal cap ectoderm treated with bFGF expresses the midbrain and hindbrain marker genes en2 and krox20, in addition to hoxB9, and the forebrain marker otx2 is not induced. The combination of somite tissue with animal caps of gastrula age (stage 10.5) induces the expression of hindbrain-specific genes and low levels of spinal cord-specific genes in the animal cap tissue and this induction is partially sensitive to XFD. Keller explants faithfully recapitulate the A-P distribution of neural markers observed in the whole embryo: blocking FGF signaling using XFD eliminates posterior neural development in Keller explants. The claim that FGF signaling is required for the formation of posterior neural tissue is supported by the results of explant assays. The induction of posterior neural markers requires FGF and Ras signaling. Animal caps do not express posterior neural markers in response to either XBF2 or XMeis3 when either FGF or Ras signaling is blocked. In addition, when MAPK activation is directly inhibited by MAP kinase phosphatase, the ability of XMeis3 to induce the expression of posterior neural markers is greatly curtailed (Ribisi, 2000 and references therein).

Several molecules of the FGF family have been implicated in the development of the vertebrate brain, but the effectors of these molecules remain largely unknown. Erm and Pea3 (both are most closely related to Drosophila (CG6892) are two ETS domain transcription factors, and their expression correlates closely with the domains of fgf8 and fgf3 expression. In situ hybridization analysis in wild-type and acerebellar (ace) mutant embryos defective for fgf8 demonstrate a requirement of Fgf8 for normal expression levels of erm and pea3 transcripts in and close to various domains of Fgf8 action, including the prospective midbrain-hindbrain region, the somites, the neural crest, the forebrain, and developing eyes. Morpholino-oligomer-assisted gene knock-down experiments targeted against fgf8 and fgf3 suggest that Fgf3 and Fgf8 are co-regulators of these genes in the early forebrain anlage. Furthermore, inhibition of Fgf signaling by overexpression of sprouty4 or application of the Fgf inhibitor SU5402, leads to a loss of all erm and pea3 expression domains. Conversely, ectopically provided fgf3 mRNA or implanted beads coated with Fgf8 elicit ectopic transcription of erm and pea3. Both activation and loss of transcripts can be observed within short time frames. It is concluded that both the transcriptional onset and maintenance of these factors are tightly coupled to Fgf signaling and it is proposed that erm and pea3 transcription is a direct readout of cells to Fgf levels. Given the accumulated knowledge on the posttranslational control of ETS domain factors and their combinatorial interactions with other transcription factors, it is suggested that the close coupling of erm and pea3 transcription to Fgf signaling might serve to integrate Fgf signaling with other signals to establish refined patterns in embryonic development (Raible, 2001).

The telencephalon is formed in the most anterior part of the central nervous system (CNS) and is organized into ventral subpallial and dorsal pallial domains. In mice, it has been demonstrated that Fgf signaling has an important role in induction and patterning of the telencephalon. However, the precise role of Fgf signaling is still unclear, owing to overlapping functions of Fgf family genes. To address this in zebrafish embryos, the activation of Ras/mitogen-activated protein kinase (MAPK), one of the major downstream targets of Fgf signaling, has been examined. Immunohistochemical analysis reveals that an extracellular signal-regulated kinase (ERK), a vertebrate MAPK, is activated in the anterior neural boundary (ANB) of the developing CNS at early segmentation stages. Experiments with Fgf inhibitors reveal that ERK activation at this stage is totally dependent on Fgf signaling. Interestingly, a substantial amount of ERK activation is observed in ace mutants in which fgf8 gene is mutated. The function of Fgf signaling in telencephalic development was analyzed by use of several inhibitors to Fgf signaling cascade, including dominant-negative forms of Ras (RasN17) and the Fgf receptor (Fgfr), and a chemical inhibitor of Fgfr, SU5402. In treated embryos, the induction of telencephalic territory normally proceeds but the development of the subpallial telencephalon is suppressed, indicating that Fgf signaling is required for the regionalization within the telencephalon. Finally, antisense experiments with morpholino-modified oligonucleotides suggest that zebrafish fgf3, which is also expressed in the ANB, co-operates with fgf8 in subpallial development (Shinya, 2001).

Early neural patterning in vertebrates involves signals that inhibit anterior (A) and promote posterior (P) positional values within the nascent neural plate. In this study, the contributions of, and interactions between, retinoic acid (RA), Fgf and Wnt signals have been investigated in the promotion of posterior fates in the ectoderm. Expression and function of cyp26/P450RAI, a gene that encodes retinoic acid 4-hydroxylase, has been examined as a tool for investigating these events. Cyp26 is first expressed in the presumptive anterior neural ectoderm and the blastoderm margin at the late blastula. When the posterior neural gene hoxb1b is expressed during gastrulation, it shows a strikingly complementary pattern to cyp26. Using these two genes, as well as otx2 and meis3 as anterior and posterior markers, it has been shown that Fgf and Wnt signals suppress expression of anterior genes, including cyp26. Overexpression of cyp26 suppresses posterior genes, suggesting that the anterior expression of cyp26 is important for restricting the expression of posterior genes. Consistent with this, knock-down of cyp26 by morpholino oligonucleotides leads to the anterior expansion of posterior genes. Fgf- and Wnt-dependent activation of posterior genes is mediated by RA, whereas suppression of anterior genes does not depend on RA signaling. Fgf and Wnt signals suppress cyp26 expression, while Cyp26, an enzyme that degrades RA, limits the range of RA-mediated posteriorization in the embryo by suppressing the RA signal. Thus, cyp26 has an important role in linking the Fgf, Wnt and RA signals to regulate AP patterning of the neural ectoderm in the late blastula to gastrula embryo in zebrafish (Kudoh, 2002).

Gene expression in the antero-lateral neural plate (the anterior neural ridge is the junction between the anterior neural plate and anterior non-neural ectoderm) is regulated by non-neural ectoderm and bone morphogenetic proteins. BF1 (a Forkhead related protein in a distinct subfamily, more distantly related to Drosophila Forkhead than is HNF3ß) expression is first detectable as early as the 3-somite-stage in the non-neural ectoderm underlying the anterior margin of the neural plate. By the 8-somite-stage, the expression is also detectable in the anterolateral neural plate. BF-1 expression in the developing brain is restricted to the telencephalic neuroepithelium and the nasal half of the retina and optic stalk. Its expression domain is adjacent to that of BF-2, which is restricted to the rostral diencephalon and the temporal half of the retina and optic stalk. Thus, the anterior neural ridge regulates patterning of the anterior neural plate, through a mechanism that is distinct from those that regulate general medial-lateral patterning. The anterior neural ridge is essential for expression of BF1; this neural ridge expresses Fgf8. Recombinant FGF8 protein is capable of inducing BF1, suggesting that FGF8 regulates the development of anterolateral neural plate derivatives (Shimamura, 1997).

The neural plate is subdivided into distinct anterior-posterior domains that have different responses to inductive signals from the prechordal plate, Sonic Hedgehog, the anterior neural ridge and FGF8. For example, Engrailed 2 is induced by beads placed more posteriorly than those that induce BF1. The induced BF1-expression domain is delineated posteriorly by a sharp boundary, which may be orthogonal to the long axis of the explants. The posterior boundary of BF1 and the anterior boundary of En2 are nearly adjacent. In sum, these results suggest that regionalization of the forebrain primordia is established by several distinct patterning mechanisms: (1) anterior-posterior patterning creates transverse zones with differential competence within the neural plate; (2) patterning along the medial-lateral axis generates longitudinally aligned domains and (3) local inductive interactions, such as a signal(s) from the anterior neural ridge, further define the regional organization (Shimamura, 1997).

Initiation of Hox genes requires interactions between numerous factors and signaling pathways in order to establish their precise domain boundaries in the developing nervous system. There are distinct differences in the expression and regulation of members of the Hox gene family within a complex, suggesting that multiple competing mechanisms are used to initiate Hox gene expression domains in early embryogenesis. In this study, by analyzing the response of HoxB genes to both RA and FGF signaling in neural tissue during early chick embryogenesis (HH stages 7-15), two distinct groups of Hox genes have been defined based on their reciprocal sensitivity to RA or FGF during this developmental period. The sharp reciprocal transition from RA to FGF responsiveness in moving from the 3' (Hoxb1 to Hoxb5) to the 5' (Hoxb6-Hoxb9) Hox genes is surprising. In mouse the 3' Hox genes do not respond uniformly to RA treatment, since there is a progressive temporal shift in their competence or ability to respond to RA during gastrulation, such that successively more 5' genes respond in later time windows. Hence, it had been suggested that the most posterior 5' Hox genes might also be progressively sensitive to RA in later stages at the end of or after gastrulation. The expression domain of 5' members from the HoxB complex (Hoxb6-Hoxb9) can be expanded anteriorly in the chick neural tube up to the level of the otic vesicle following FGF treatment and these same genes are refractory to RA treatment at these stages (Bel-Vialar, 2002).

The chick caudal-related genes, cdxA and cdxB, are also responsive to FGF signaling in neural tissue and their anterior expansion is also limited to the level of the otic vesicle. Using a dominant negative form of a Xenopus Cdx gene (XcadEnR) it has been found that the effect of FGF treatment on 5' HoxB genes is mediated in part through the activation and function of CDX activity. Conversely, the 3' HoxB genes (Hoxb1 and Hoxb3-Hoxb5) are sensitive to RA but not FGF treatments at these stages. In ovo electroporation of a dominant negative retinoid receptor construct (dnRAR) shows that retinoid signaling is required to initiate expression. Elevating CDX activity by ectopic expression of an activated form of a Xenopus Cdx gene (XcadVP16) in the hindbrain ectopically activates and anteriorly expands Hoxb4 expression. In a similar manner, when ectopic expression of XcadVP16 is combined with FGF treatment, it was found that Hoxb9 expression expands anteriorly into the hindbrain region. These findings suggest a model whereby, over the window of early development examined, all HoxB genes are actually competent to interpret an FGF signal via a CDX-dependent pathway. However, mechanisms that axially restrict the Cdx domains of expression, serve to prevent 3' genes from responding to FGF signaling in the hindbrain. FGF may have a dual role in both modulating the accessibility of the HoxB complex along the axis and in activating the expression of Cdx genes. The position of the shift in RA or FGF responsiveness of Hox genes may be time dependent. Hence, the specific Hox genes in each of these complementary groups may vary in later stages of development or other tissues. These results highlight the key role of Cdx genes in integrating the input of multiple signaling pathways, such as FGFs and RA, in controlling initiation of Hox expression during development and the importance of understanding regulatory events/mechanisms that modulate Cdx expression (Bel-Vialar, 2002).

The valentino (val) mutation in zebrafish perturbs hindbrain patterning and, as a secondary consequence, also alters development of the inner ear. The relationship between these defects and expression of fgf3 and fgf8 in the hindbrain were examined. The otic vesicle in val/val mutants is smaller than normal, yet produces nearly twice the normal number of hair cells, and some hair cells are produced ectopically between the anterior and posterior maculae. Anterior markers pax5 and nkx5.1 are expressed in expanded domains that include the entire otic epithelium juxtaposed to the hindbrain, and the posterior marker zp23 is not expressed. In the mutant hindbrain, expression of fgf8 is normal, whereas the domain of fgf3 expression expands to include rhombomere 4 through rhombomere X (an aberrant segment that forms in lieu of rhombomeres 5 and 6). Depletion of fgf3 by injection of antisense morpholino (fgf3-MO) suppresses the ear patterning defects in val/val embryos: excess and ectopic hair cells are eliminated, expression of anterior otic markers is reduced or ablated, and zp23 is expressed throughout the medial wall of the otic vesicle. By contrast, disruption of fgf8 does not suppress the val/val phenotype but instead interacts additively, indicating that these genes affect distinct developmental pathways. Thus, the inner ear defects observed in val/val mutants appear to result from ectopic expression of fgf3 in the hindbrain. These data also indicate that val normally represses fgf3 expression in r5 and r6, an interpretation further supported by the effects of misexpressing val in wild-type embryos. This is in sharp contrast to the mouse, in which fgf3 is normally expressed in r5 and r6 because of positive regulation by kreisler, the mouse ortholog of val. Implications for co-evolution of the hindbrain and inner ear are discussed (Kwak, 2002).

The neocortex is divided into multiple areas with specific architecture, molecular identity and pattern of connectivity with the dorsal thalamus. Gradients of transcription factor expression in the cortical primordium regulate molecular regionalization and potentially the patterning of thalamic projections. Reduction of Fgf8 levels in hypomorphic mouse mutants shifts early gradients of gene expression rostrally, thereby modifying the molecular identity of rostral cortical progenitors. This shift correlates with a reduction in the size of a molecularly defined rostral neocortical domain and a corresponding rostral expansion of more caudal regions. Despite these molecular changes, the topography of projections between the dorsal thalamus and rostral neocortex in mutant neonates appears the same as the topography of wild-type littermates. Overall, this study demonstrates the role of endogenous Fgf8 in regulating early gradients of transcription factors in cortical progenitor cells and in molecular regionalization of the cortical plate (Garel, 2003).

Complex spatiotemporal expression patterns of fgf3 and fgf8 within the developing zebrafish forebrain suggest their involvement in its regionalization and early development. These factors have unique and combinatorial roles during development of more posterior brain regions, and similar findings have been made for the developing forebrain. Fgf8 and Fgf3 regulate different aspects of telencephalic development, and Fgf3 alone is required for the expression of several telencephalic markers. Within the diencephalon, Fgf3 and Fgf8 act synergistically to pattern the ventral thalamus, and are implicated in the regulation of optic stalk formation, whereas loss of Fgf3 alone results in defects in zona limitans intrathalamica development. Forebrain commissure formation is abnormal in the absence of either Fgf3 or Fgf8; however, most severe defects are observed in the absence of both. Defects are observed in patterning of both the midline territory, within which the commissures normally form, and neuronal populations, whose axons comprise the commissures. Analysis of embryos treated with an FGFR inhibitor suggests that continuous FGF signalling is required from gastrulation stages for normal forebrain patterning, and identifies additional requirements for FGFR activity (Walshe, 2003).

Vertebrate body axis extension involves progressive generation and subsequent differentiation of new cells derived from a caudal stem zone; however, molecular mechanisms that preserve caudal progenitors and coordinate differentiation are poorly understood. FGF maintains caudal progenitors and its attenuation is required for neuronal and mesodermal differentiation and to position segment boundaries. Furthermore, somitic mesoderm promotes neuronal differentiation in part by downregulating Fgf8. retinoic acid (RA) has been identified as this somitic signal; retinoid and FGF pathways have opposing actions. FGF is a general repressor of differentiation, including ventral neural patterning, while RA attenuates Fgf8 in neuroepithelium and paraxial mesoderm, where it controls somite boundary position. RA is further required for neuronal differentiation and expression of key ventral neural patterning genes. These data demonstrate that FGF and RA pathways are mutually inhibitory and suggest that their opposing actions provide a global mechanism that controls differentiation during axis extension (Diez del Corral, 2003).

FGF can maintain an undifferentiated cell state, and retinoids can drive differentiation in many different contexts; for example, mouse ES cells form neural precursors in vitro under the influence of FGF signaling, while exposure to RA promotes neuronal differentiation. In the mouse embryo, excess RA, due to mutation of the RA-metabolizing enzyme CYP26, has been shown to repress Fgf8 expression in the tail bud, and RA downregulates Fgf8 in both neural and mesodermal tissues in vitro. Although Fgf8 expression perdures in caudal regions of Vitamin A-deficient (VAD) quails, it is still eventually lost from the presomitic mesoderm and neuroepithelium. This suggests either that somites provide another signal that can repress Fgf8 or that Fgf8 transcripts normally decay and that RA acts to accelerate this process. In the presomitic mesoderm, Fgf8 can be induced by FGF8 and so RA could effect Fgf8 reduction by interfering with the FGF signaling pathway. Conversely, FGF8 controls RA synthesis by inhibiting onset of Raldh2 in the paraxial mesoderm. Furthermore, exposure to FGF also blocks neuronal differentiation in explants of neural tube that do not express Raldh2, suggesting that FGF can also oppose RA activity in the neuroepithelium. It is proposed that during normal extension of the axis, a slight decline in Fgf8 transcripts (facilitated by regression of the primitive streak that expresses FGFs able to induce Fgf8) is sufficient for Raldh2 onset. As presomitic mesoderm begins to synthesize RA, retinoid signaling then accelerates Fgf8 downregulation in both the paraxial mesoderm and adjacent preneural tube. This mutual opposition of FGF and RA pathways thus helps to ensure the coordinated differentiation of mesodermal and neural tissues (Diez del Corral, 2003).

The level of FGF signaling in the presomitic mesoderm controls where a somite boundary will form, and the ability of RA to attenuate Fgf8 in the paraxial mesoderm identifies a role for the retinoid pathway in this process. According to the current model, somite size is determined by two components: the period of oscillation of transiently expressed mRNAs associated with Notch signaling such as Hairy1 (the segmentation clock) and the speed of FGF decline in the presomitic mesoderm (the maturation wavefront). Changes in FGF signaling do not alter the period of oscillation: resulting segmentation defects are due to alteration in the speed at which FGF levels fall below a threshold. Since RA downregulates Fgf8 in the presomitic mesoderm, it must set the rate of maturation wavefront progression and thereby influence somite size. Further, since FGF and RA pathways are mutually inhibitory, this could create a sharp transition in cell signaling in the presomitic mesoderm, and one possibility is that this change precisely defines the future somite border. Consistent with this, in VAD embryos where Fgf8 expression is prolonged and wavefront progression is slowed, initial somite size is smaller (Diez del Corral, 2003).

Finally, opposition of FGF and RA pathways may be a conserved mechanism for controlling differentiation and maintaining progenitor pools in the developing embryo. A striking analogy can be drawn with the proximo-distal extension of the limb in which distal FGF signaling provided by the apical ridge restricts RA synthesis and RARß receptor expression to the proximal limb. FGF signaling also opposes RA in the forming hindbrain, preserving rhombomere1 as a site of FGF activity that undergoes extensive proliferation to generate the cerebellum. The data suggest that mutual inhibition and opposing activities of FGF and RA pathways act to maintain a critical balance between preservation of the progenitor pool/stem zone and the progressive differentiation of neural and mesodermal tissues during the extension of the embryonic axis (Diez del Corral, 2003).

Fibroblast growth factor (Fgf) and retinoic acid (RA) signals control the formation and anteroposterior patterning of posterior hindbrain. They are also involved in development processes in other regions of the embryo. Therefore, responsiveness to Fgf and RA signals must be controlled in a context-dependent manner. Inhibiting the caudal-related genes cdx1a and cdx4 in zebrafish embryos caused ectopic expression of genes that are normally expressed in the posterior hindbrain and anterior spinal cord, and ectopic formation of the hindbrain motor and commissure neurons in the posteriormost neural tissue. Combinational marker analyses suggest mirror-image duplication in the Cdx1a/4-defective embryos, and cell transplantation analysis further revealed that Cdx1a and Cdx4 repress a posterior hindbrain-specific gene expression cell-autonomously in the posterior neural tissue. Expression of fgfs and retinaldehyde dehydrogenase 2 suggested that in the Cdx1a/4-defective embryos, the Fgf and RA signaling activities overlap in the posterior body and display opposing gradients, compared with those in the hindbrain region. Fgf and RA signals were required for ectopic expression. Expression of the posterior hox genes hoxb7a, hoxa9a or hoxb9a, which function downstream of Cdx1a/4, or activator fusion genes of hoxa9a or hoxb9a (VP16-hoxa9a, VP16-hoxb9a) suppressed this loss-of-function phenotype. These data suggest that Cdx suppresses the posterior hindbrain fate through regulation of the posterior hox genes; the posterior Hox proteins function as transcriptional activators and indirectly repress the ectopic expression of the posterior hindbrain genes in the posterior neural tissue. These results indicate that the Cdx-Hox code modifies tissue competence to respond to Fgfs and RA in neural tissue (Shumizu, 2006).

FGFs and the midbrain-hindbrain organizer

The regionalization of the neural tube along the anteroposterior axis is established through the action of patterning signals from the endo-mesoderm, including the organizer. These signals set up a pre-pattern which is subsequently refined through local patterning events. The midbrain-hindbrain junction, or isthmus, is endowed with such an organizing activity. It is able to induce graded expression of the Engrailed protein in the adjacent mesencephalon and rhombencephalon, and subsequently elicits the development of tectal and cerebellar structures. Ectopically grafted isthmus has also been shown to induce Engrailed expression in diencephalon and otic and pre-otic rhombencephalon. Fgf8 is a signaling protein that is produced by the isthmus and which is able to mimic most isthmic properties. The isthmus, when transposed to the level of either rhombomere 8 or the spinal cord, loses its ability to induce Engrailed and cerebellar development in adjacent tissues. This is accompanied by the down-regulation of fgf8 expression in the grafted isthmus and by the up-regulation of a marker of the recipient site, Hoxb-4. Moreover, these changes in gene activity in the transplant are followed by a transformation of the fate of the grafted cells, which adjust to their novel environment. These results show that the fate of the isthmus is not determined at the 10-somite stage and that the molecular loop of isthmic maintenance can be disrupted by exogenous signals (Grapin-Botton, 1999).

Beads containing recombinant FGF8 (FGF8-beads) were implanted in the prospective caudal diencephalon or midbrain of chick embryos at stages 9-12. This induces the neuroepithelium rostral and caudal to the FGF8-bead to form two ectopic, mirror-image midbrains. Furthermore, cells in direct contact with the bead form an outgrowth that protruded laterally from the neural tube. Tissue within such lateral outgrowths developed proximally into isthmic nuclei and distally into a cerebellum-like structure. These morphogenetic effects are apparently due to FGF8-mediated changes in gene expression in the vicinity of the bead, including a repressive effect on Otx2 and an inductive effect on En1, Fgf8 and Wnt1 expression. The ectopic Fgf8 and Wnt1 expression domains form nearly complete concentric rings around the FGF8-bead, with the Wnt1 ring outermost. These observations suggest that FGF8 induces the formation of a ring-like ectopic signaling center (organizer) in the lateral wall of the brain, similar to the one that normally encircles the neural tube at the isthmic constriction, which is located at the boundary between the prospective midbrain and hindbrain. This ectopic isthmic organizer apparently sends long-range patterning signals both rostrally and caudally, resulting in the development of the two ectopic midbrains. Interestingly, the data suggest that these inductive signals spread readily in a caudal direction, but are inhibited from spreading rostrally across diencephalic neuromere boundaries. These results provide insights into the mechanism by which FGF8 induces an ectopic organizer and suggest that a negative feedback loop between Fgf8 and Otx2 plays a key role in patterning the midbrain and anterior hindbrain (Martinez, 1999).

The patterns of the Gbx2, Pax2, Wnt1, and Fgf8 gene expression were analyzed in the chick with respect to the caudal limit of the Otx2 anterior domain, taken as a landmark of the midbrain/hindbrain (MH) boundary. The Gbx2 anterior boundary is always concomitant with the Otx2 posterior boundary. The ring of Wnt1 expression is included within the Otx2 domain and Fgf8 transcripts included within the Gbx2 neuroepithelium. Pax2 expression is centered on the MH boundary with a double decreasing gradient. A new nomenclature is proposed to differentiate the vesicles and constrictions observed in the avian MH domain at stage HH10 and HH20, based on the localization of the Gbx2/Otx2 common boundary (Hidalgo-Snachez, 1999).

The mid/hindbrain junction region, which expresses Fgf8, can act as an organizer to transform caudal forebrain or hindbrain tissue into midbrain or cerebellar structures, respectively. FGF8-soaked beads placed in the chick forebrain can similarly induce ectopic expression of mid/hindbrain genes and development of midbrain structures. In contrast, ectopic expression of Fgf8a in the mouse midbrain and caudal forebrain using a Wnt1 regulatory element produces no apparent patterning defects in the embryos examined. FGF8b-soaked beads can not only induce expression of the mid/hindbrain genes En1, En2 and Pax5 in mouse embryonic day 9.5 (E9.5) caudal forebrain explants, but also can induce the hindbrain gene Gbx2 and alter the expression of Wnt1 in both midbrain and caudal forebrain explants. FGF8b-soaked beads can repress Otx2 in midbrain explants. Furthermore, Wnt1-Fgf8b transgenic embryos in which the same Wnt1 regulatory element is used to express Fgf8b, have ectopic expression of En1, En2, Pax5 and Gbx2 in the dorsal hindbrain and spinal cord at E10.5, as well as exencephaly and abnormal spinal cord morphology. More strikingly, Fgf8b expression in more rostral brain regions appears to transform the midbrain and caudal forebrain into an anterior hindbrain fate through expansion of the Gbx2 domain and repression of Otx2 as early as the 7-somite stage. These findings suggest that normal Fgf8 expression in the anterior hindbrain not only functions to maintain development of the entire mid/hindbrain by regulating genes like En1, En2 and Pax5, but also might function to maintain a metencephalic identity by regulating Gbx2 and Otx2 expression (Liu, 1999).

It is interesting that the phenotype observed in early Wnt1-Fgf8b transgenics is similar to that seen in Otx1+/-Otx2+/- or Otx1-/-Otx2+/- double mutants; an early induction of Gbx2 and repression of Otx2 in the midbrain and caudal forebrain. In Otx1-/-;Otx2+/- embryos, an anterior expansion of Fgf8 expression precedes an anterior shift of Wnt1 and En1 expression and an anterior retraction of Otx2 expression. The Otx mutant studies suggest a certain level of Otx2 expression is necessary to repress expression of Fgf8 in the midbrain and forebrain, and these results suggest that, in addition, expanded Fgf8 expression could contribute to repression of Otx2 expression in the midbrain. A reciprocal negative regulation between Otx2 and Fgf8 might therefore normally contribute to maintaining the Otx2 caudal boundary and positioning the organizer (Liu, 1999 and references therein).

Fgf8, which is expressed at the embryonic mid/hindbrain junction, is required for and sufficient to induce the formation of midbrain and cerebellar structures. To address the genetic pathways through which FGF8 acts, the epistatic relationships of mid/hindbrain genes that respond to FGF8 were examined, using a novel mouse brain explant culture system. En2 and Gbx2 are the first genes to be induced by FGF8 in wild-type E9.5 diencephalic and midbrain explants treated with FGF8-soaked beads. By examining gene expression in En1/2 double mutant mouse embryos, it was found that Fgf8, Wnt1 and Pax5 do not require the En genes for initiation of expression, but do for their maintenance, and Pax6 expression is expanded caudally into the midbrain in the absence of EN function. Since E9.5 En1/2 double mutants lack the mid/hindbrain region, forebrain mutant explants were treated with FGF8 and, significantly, the EN transcription factors were found to be required for induction of Pax5. Thus, FGF8-regulated expression of Pax5 is dependent on EN proteins, and a factor other than FGF8 could be involved in initiating normal Pax5 expression in the mesencephalon/metencephalon. The En genes also play an important, but not absolute, role in repression of Pax6 in forebrain explants by FGF8. Gbx2 gain-of-function studies have shown that misexpression of Gbx2 in the midbrain can lead to repression of Otx2. However, in the absence of Gbx2, FGF8 can nevertheless repress Otx2 expression in midbrain explants. In contrast, Wnt1 is initially broadly induced in Gbx2 mutant explants, as in wild-type explants, but not subsequently repressed in cells near FGF8 that normally express Gbx2. Thus GBX2 acts upstream of, or parallel to, FGF8 in repressing Otx2, and acts downstream of FGF8 in repression of Wnt1. This is the first such epistatic study performed in mouse that combines gain-of-function and loss-of-function approaches to reveal aspects of mouse gene regulation in the mesencephalon/metencephalon that have been difficult to address using either approach alone (Liu, 2001).

Fibroblast growth factor (FGF) has been implicated in a variety of developmental processes including posterior mesoderm and neural patterning. Previous work has led to contradictory roles for FGF in neural induction and anteroposterior neural patterning. A loss-of-function assay was used to examine whether FGF is required for neural patterning in three experimental situations: (1) in Xenopus early embryos, (2) in embryonic explants consisting of presumptive dorsal mesoderm and neurectoderm (Keller explants), and (3) in explants of dorsal ectoderm and posterior mesoderm in which FGF signaling is specifically blocked in the ectoderm. When cultured until tailbud stages, Keller explants develop neural tissue with normal anteroposterior pattern. Overexpression of the dominant-negative FGF receptor (XFD) in Keller explants inhibits the posterior neural markers En-2, Krox-20, and HoxB9, but not the panneural marker nrp-1 and the anterior neurectodermal markers XAG-1 and Xotx-2. Similar results are seen in whole embryos, but only when XFD RNA is targeted to both the dorsal and lateral regions. In contrast, addition of FGF to Keller explants results in a shift of the midbrain-hindbrain boundary marker En-2 to a more anterior position normally fated to become cement gland. It was also determined whether FGF is required specifically by the neurectoderm for anteroposterior neural patterning. Recombinants of dorsal ectoderm and posterior mesoderm were made in which FGF was specifically blocked in the ectoderm. Spinal cord and hindbrain markers are inhibited in these recombinants, whereas anterior markers and cement gland development are enhanced. These results demonstrate that FGF is important for posterior development in both mesoderm and neurectoderm and that neural induction and posteriorization represent separable developmental events (Holowacz, 1999).

The midbrain-hindbrain (MHB) junction has the properties of an organizer that patterns the MHB region early in vertebrate development. Classical transplantation experiments demonstrate that MHB tissue can induce midbrain structures when transplanted into the diencephalon and cerebellar tissue when transplanted to the myelencephalon. Fgf8 is thought to mediate this organizer function. In addition to Fgf8, Fgf17 and Fgf18 are also expressed in the MHB junction. Fgf17 is expressed later and broader than either Fgf8 or Fgf18. Disrupting the Fgf17 gene in the mouse decreases precursor cell proliferation in the medial cerebellar (vermis) anlage after E11.5. Loss of an additional copy of Fgf8 enhances the phenotype and accelerates its onset, demonstrating that both molecules cooperate to regulate the size of the precursor pool of cells that develop into the cerebellar vermis. However, expression patterns of Wnt1, En2, Pax5 and Otx2 are not altered, suggesting that specification and patterning of MHB tissue is not perturbed and that these FGFs are not required to pattern the vermis at this stage of development. The consequence of this developmental defect is a progressive, dose-dependent loss of the most anterior lobe of the vermis in mice lacking Fgf17 and in mice lacking Fgf17 and one copy of Fgf8. Significantly, the differentiation of anterior vermis neuroepithelium is shifted rostrally and medially, demonstrating that FGF also regulates the polarized progression of differentiation in the vermis anlage. Finally, this developmental defect results in an ataxic gait in some mice (Xu, 2000).

Genes encoding fibroblast growth factors (FGFs) are expressed in early Xenopus neurulae in the prospective midbrain--hindbrain boundary (MHB) region of the neural plate. These expression domains overlap those of XWnt-1 and XEn-2, raising the question of the role of FGF signaling in the regulation of these genes, and more generally about the function of FGF during Xenopus midbrain development. Explants from the prospective MHB grafted into the anterior neural plate in midneurula stage embryos induce XWnt-1 expression and, at a lower frequency, XEn-2 expression in the vicinity of the graft. Such a process is likely to involve FGF signaling. Implantation of FGF4- or FGF8-soaked beads in the prospective forebrain at neurula and tailbud stages causes the up-regulation of XWnt-1 and XEn-2 in the dorsal and lateral region of the anterior midbrain. This effect is not relayed by endogenous FGF genes since exogenous FGFs inhibit the expression of endogenous XFGF3 or XFGF8. However, consequences of grafting MHB or implanting FGF4 or FGF8 beads on tadpole brain development are different. MHB grafts induce ectopic mesencephalic structures, strongly suggesting that a region homologous to the isthmic organizer of amniotes is specified as early as the midneurula stage. In contrast, exogenous FGFs do not cause the formation of ectopic mesencephalic structures but an overgrowth of mesencephalon and diencephalon. It is proposed that FGF signals from the prospective MHB play a crucial role in the spatial regulation of XWnt-1 and XEn-2 expression in the posterior midbrain, but that the full organizing activity of the MHB involves other factors in combination with FGF (Riou, 1998).

The most studied secondary neural organizer is the isthmic organizer, which is localized at the mid-hindbrain transition of the neural tube and controls the anterior hindbrain and midbrain regionalization. Otx2 and Gbx2 expressions are fundamental for positioning the organizer and the establishment of molecular interactions that induce Fgf8. Evidence in this study demonstrates that Otx2 and Gbx2 have an overlapping expression in the isthmic region. This area is the transversal domain where expression of Fgf8 is induced. The Fgf8 protein produced in the isthmus stabilizes and up-regulates Gbx2 expression, which, in turn, down-regulates Otx2 expression. The inductive effect of the Gbx2/Otx2 limit keeps Fgf8 expression stable and thus maintains its positive role in the expression of Pax2, En1,2 and Wnt1 (Garda, 2001).

Whether Gbx2 is required after embryonic day 9 (E9) to repress Otx2 in the cerebellar anlage and position the midbrain/hindbrain organizer was examined. In contrast to Gbx2 null mutants, mice lacking Gbx2 in rhombomere 1 (r1) after E9 (Gbx2-CKO) are viable and develop a cerebellum. A Gbx2-independent pathway can repress Otx2 in r1 after E9. Mid/hindbrain organizer gene expression, however, continues to be dependent on Gbx2. Fgf8 expression normally correlates with the isthmus where cells undergo low proliferation and in Gbx2-CKO mutants this domain is expanded. It is proposed that Fgf8 permits lateral cerebellar development through repression of Otx2 and also suppresses medial cerebellar growth in Gbx2-CKO embryos. This work has uncovered distinct requirements for Gbx2 during cerebellum formation and provides a model for how a transcription factor can play multiple roles during development (Li, 2002).

In Gbx2-CKO embryos, the juxtaposition of the Wnt1 and Fgf8 expression domains is present at the 8 somite stage, but, consistent with previous studies showing that an interaction between Otx2 and Gbx2 positions the mid/hindbrain organizer, the border is shifted posteriorly to the new Otx2/Gbx2 border. In contrast, at E9.5 when Gbx2 transcripts are no longer detected in r1, Wnt1 and Fgf8 were broadly coexpressed in the alar plate of r1. The derepression of Wnt1 in the alar plate of r1 where Gbx2 is normally expressed demonstrates a cell-autonomous requirement for Gbx2 in repression of Wnt1 expression after E9.5, in agreement with previous studies. Since ectopic expression of Wnt1 in r1 can induce Fgf8 in chick embryos, derepression of Wnt1 in r1 cells in Gbx2-CKO embryos could contribute to the expansion of Fgf8 expression in this region. Furthermore, the expression domain of Pax2 in the isthmus is expanded posteriorly in Gbx2-CKO embryos from E9.5 and largely overlaps with that of Fgf8, consistent with the observation that Pax2 is essential for induction of Fgf8. These experiments show that Gbx2 is required from E8.5 onward to repress Wnt1 expression in r1 and maintain the normal relative expression domains of Wnt1 and Fgf8 (Li, 2002).

Development of the CNS involves highly combinatorial actions of transcription factors. Gbx2 is initially required to repress Otx2 before E8.5 to allow specification of the cerebellar primordium. After E8.5, Gbx2 is not essential for the repression of Otx2 because a second pathway is induced that can repress Otx2. Gbx2 is nevertheless still required for maintenance of normal expression of Wnt1 and Fgf8. The temporal changing requirement for Gbx2 during cerebellar development demonstrated in this work provides a different paradigm for how the same transcription factor can control sequential events during a single developmental process (Li, 2002).

Numerous studies have demonstrated that the midbrain and cerebellum develop from a region of the early neural tube comprising two distinct territories known as the mesencephalon (mes) and rostral metencephalon (met; rhombomere1), respectively. Development of the mes and met is thought to be regulated by molecules produced by a signaling center, termed the isthmic organizer (IsO), which is localized at the boundary between them. FGF8 and WNT1 have been implicated as key components of IsO signaling activity, and previous studies have shown that in Wnt1^-/- embryos, the mes/met is deleted by the 30 somite stage. The function of FGF8 in mouse mes/met development has been studied using a conditional gene inactivation approach. In mutant embryos, Fgf8 expression is transiently detected, but then is eliminated in the mes/met by the 10 somite stage. This results in a failure to maintain expression of Wnt1 as well as Fgf17, Fgf18, and Gbx2 in the mes/met at early somite stages, and in the absence of the midbrain and cerebellum at E17.5. A major cause of the deletion of these structures is ectopic cell death in the mes/met between the 7 and 30 somite stages. Interestingly, the prospective midbrain is deleted at an earlier stage than the prospective cerebellum. A remarkably similar pattern of cell death is observed in Wnt1 null homozygotes, and also ectopic mes/met cell death in En1 is detected in null homozygotes. These data show that Fgf8 is part of a complex gene regulatory network that is essential for cell survival in the mes/met (Chi, 2003).

The midbrain-hindbrain domain (MH) of the vertebrate embryonic neural tube develops in response to the isthmic organizer (IsO), located at the midbrain-hindbrain boundary (MHB). MH derivatives are largely missing in mutants affected in IsO activity; however, the potentialities and fate of MH precursors in these conditions have not been directly determined. To follow the dynamics of MH maintenance in vivo, artificial chromosome transgenesis was used in zebrafish to construct lines where egfp transcription is driven by the complete set of regulatory elements of her5, the first known gene expressed in the MH area. In these lines, egfp transcription faithfully recapitulates her5 expression from its induction phase onwards. Using the stability of GFP protein as lineage tracer, her5, first demonstrated at gastrulation, is a selective marker of MH precursor fate. By comparing GFP protein and her5 transcription, the spatiotemporal dynamics of her5 expression that conditions neurogenesis progression towards the MHB over time was further revealed. The molecular identity of GFP-positive cells was traced in the acerebellar (ace) and no-isthmus (noi) mutant backgrounds to analyze directly fgf8 and pax2.1 mutant gene activities for their ultimate effect on cell fate. Most MH precursors are maintained in both mutants but express abnormal identities, in a manner that strikingly differs between the ace and noi contexts. These observations directly support a role for Fgf8 in protecting anterior tectal and metencephalic precursors from acquiring anterior identities, while Pax2.1 controls the choice of MH identity as a whole. Together, these results suggest a model where an ordered MH pro-domain is identified at gastrulation, and where cell identity choices within this domain are subsequently differentially controlled by Fgf8 and Pax2.1 functions (Tallafuß, 2003).

Specification of the forebrain, midbrain and hindbrain primordia occurs during gastrulation in response to signals that pattern the gastrula embryo. Following establishment of the primordia, each brain part is thought to develop largely independently from the others under the influence of local organizing centers like the midbrain-hindbrain boundary (MHB, or isthmic) organizer. Mechanisms that maintain the integrity of brain subdivisions at later stages are not yet known. To examine such mechanisms in the anterior neural tube, the establishment and maintenance of the diencephalic-mesencephalic boundary (DMB) was studied. Maintenance of the DMB requires both the presence of a specified midbrain and a functional MHB organizer. Expression of pax6.1, a key regulator of forebrain development, is posteriorly suppressed by the Engrailed proteins, Eng2 and Eng3. Mis-expression of eng3 in the forebrain primordium causes downregulation of pax6.1, and forebrain cells correspondingly change their fate and acquire midbrain identity. Conversely, in embryos lacking both eng2 and eng3, the DMB shifts caudally into the midbrain territory. However, a patch of midbrain tissue remains between the forebrain and the hindbrain primordia in such embryos. This suggests that an additional factor maintains midbrain cell fate. Fgf8 is a candidate for this signal, because it is both necessary and sufficient to repress pax6.1 and hence to shift the DMB to the anterior, independent of the expression status of eng2/eng3. By examining small cell clones that are unable to receive an Fgf signal, it has been shown that cells in the presumptive midbrain neural plate require an Fgf signal to keep them from following a forebrain fate. Combined loss of both Eng2/Eng3 and Fgf8 leads to complete loss of midbrain identity, resulting in fusion of the forebrain and the hindbrain primordia. Thus, Eng2/Eng3 and Fgf8 are necessary to maintain midbrain identity in the neural plate and thereby position the DMB. This provides an example of a mechanism needed to maintain the subdivision of the anterior neural plate into forebrain and midbrain (Scholpp, 2003).

Early patterning of the vertebrate midbrain and cerebellum is regulated by a mid/hindbrain organizer that produces three fibroblast growth factors (FGF8, FGF17 and FGF18). The mechanism by which each FGF contributes to patterning the midbrain, and induces a cerebellum in rhombomere 1 (r1) is not clear. FGF8b can transform the midbrain into a cerebellum fate, whereas FGF8a can promote midbrain development. A chick electroporation assay and in vitro mouse brain explant experiments have been used to compare the activity of FGF17b and FGF18 to FGF8a and FGF8b. (1) FGF8b is the only protein that can induce the r1 gene Gbx2 and strongly activate the pathway inhibitors Spry1/2, as well as repress the midbrain gene Otx2. Consistent with previous studies that indicated high level FGF signaling is required to induce these gene expression changes, electroporation of activated FGFRs produce similar gene expression changes to FGF8b. (2) FGF8b extends the organizer along the junction between the induced Gbx2 domain and the remaining Otx2 region in the midbrain, correlating with cerebellum development. By contrast, FGF17b and FGF18 mimic FGF8a by causing expansion of the midbrain and upregulating midbrain gene expression. This result is consistent with Fgf17 and Fgf18 being expressed in the midbrain and not just in r1 as is Fgf8. (3) Analysis of gene expression in mouse brain explants with beads soaked in FGF8b or FGF17b shows that the distinct activities of FGF17b and FGF8b are not due to differences in the amount of FGF17b protein produced in vivo. Finally, brain explants were used to define a positive feedback loop involving FGF8b mediated upregulation of Fgf18, and two negative feedback loops that include repression of Fgfr2/3 and direct induction of Spry1/2. Since Fgf17 and Fgf18 are co-expressed with Fgf8 in many tissues, these studies have broad implications for how these FGFs differentially control development (Liu, 2003).

The following steps in midbrain and cerebellum development in mouse are proposed. At the four-somite stage, Fgf8 is induced in the presumptive r1 territory by an unknown factor. Pax2 is required for this induction and OTX2 inhibits Fgf8 from being induced in the midbrain. FGF8b then induces Fgf18 in the surrounding cells, producing a larger domain and gradient of Fgf mRNA that extends into the midbrain. FGF8b also maintains two negative feedback loops by inducing Spry1 and Spry2 expression and inhibiting Fgfr2 and Fgfr3. Fgf17 is then induced by an unknown mechanism that is dependent on Fgf8 in a broader domain than Fgf18, further extending the gradient of Fgf mRNA expression. FGF17 and FGF18 protein, and possibly FGF8a and a low level of FGF8b, then regulate proliferation of the midbrain and cerebellum and En expression. The narrow domain where Fgf8 is expressed becomes the isthmus because of the activity of FGF8b, and the adjacent Otx2-negative r1 cells become the cerebellum. By the 15-somite stage Gbx2 is not required in r1 for cerebellum development, but is required earlier to specify r1. Thus, once Fgf8 expression in r1 is stabilized, perhaps by a secreted factor from the midbrain, a key function of high level signaling by FGF8b is to maintain a cascade of gene expression in the midbrain/r1 that maintains an Otx2-negative domain in r1 in which the cerebellum develops (Liu, 2003).

The organizer at the midbrain-hindbrain boundary (MHB) forms at the interface between Otx2 and Gbx2 expressing cell populations, but how these gene expression domains are set up and integrated with the remaining machinery controlling MHB development is unclear. The isolation, mapping, chromosomal synteny and spatiotemporal expression of gbx1 and gbx2 in zebrafish is reported. Focus was placed on the expression of these genes during development of the midbrain-hindbrain territory. The results suggest that these genes function in this area in a complex fashion, as evidenced by their highly dynamic expression patterns and relation to Fgf signaling. Analysis of gbx1 and gbx2 expression during formation of the MHB in mutant embryos for pax2.1, fgf8 and pou2 (noi, ace, spg), as well as Fgf-inhibition experiments, show that gbx1 acts upstream of these genes in MHB development. In contrast, gbx2 activation requires ace (fgf8) function, and in the hindbrain primordium, also spg (pou2). It is proposed that in zebrafish, gbx genes act repeatedly in MHB development, with gbx1 acting during the positioning period of the MHB at gastrula stages, and gbx2 functioning after initial formation of the MHB, from late gastrulation stages onwards. Transplantation studies furthermore reveal that at the gastrula stage, Fgf8 signals from the hindbrain primordium into the underlying mesendoderm. Apart from the general involvement of gbx genes in MHB development reported also in other vertebrates, these results emphasize that early MHB development can be divided into multiple steps with different genetic requirements with respect to gbx gene function and Fgf signaling. Moreover, these results provide an example for switching of a specific gene function of gbx1 versus gbx2 between orthologous genes in zebrafish and mammals (Rhinn, 2003).

FGF homologs and segmentation of the hindbrain

Neural tissue in developing Xenopus embryos is induced by signals from the dorsal mesoderm. Induction of anterior neural tissue could be mediated by noggin, a secreted polypeptide found in dorsal mesoderm. bFGF, a known mesoderm inducer of blastula staged ectoderm, induces neural tissue from gastrula stage ectoderm. The type of neural tissue induced by bFGF from stage 10.25 ectoderm is posterior, as marked by Hox B9 expression. When bFGF and noggin are combined on early gastrula stage ectoderm, a more complete neural pattern is generated and no mesodermal tissue is detected. Explants treated with noggin and bFGF elongate and display distinct anterior and posterior ends marked by otx2 and Hox B9 expression, respectively. Furthermore, treatment of early gastrula ectoderm with noggin and bFGF results in the induction of En-2, a marker of the midbrain-hindbrain junction, and Krox 20, a marker of the third and fifth rhombomeres of the hindbrain. Neither of these genes is induced by noggin alone or bFGF alone at this stage, suggesting a synergy in anterior-posterior neural patterning. The response of later gastrula (stage 11-12) ectoderm to bFGF changes so that Krox 20 and En-2 are induced by bFGF alone, while induction of more posterior tissue marked by Hox B9 is eliminated. The dose of bFGF affects the amount of neural tissue induced, but has little effect on the anterior-posterior character; rather the age of the ectoderm treated is the determinant of the response. Thus, an FGF signal may account for posterior neural induction; anterior-posterior neural patterning could be partly explained by the actions of noggin and FGF, together with the changing response of the ectoderm to these factors (Lamb, 1995).

Krox20 and mafB/kreisler are regulatory genes involved in hindbrain segmentation and anteroposterior (AP) patterning. They are expressed in rhombomeres (r) r3/r5 and r5/r6 respectively, as well as in the r5/r6 neural crest. Since several members of the fibroblast growth factor (FGF) family are expressed in the otic/preotic region (r2-r6), their possible involvement in the regulation of Krox20 and mafB/kr was investigated. Application of exogenous FGFs to the neural tube of 4- to 7-somite chick embryos leads to ectopic expression in the neural crest of the somitic hindbrain (r7 and r8) and to the extension of the Krox20- or mafB/kr-positive areas in the neuroepithelium. Application of an inhibitor of FGF signaling leads to severe and specific downregulation of Krox20 and mafB/kr in the hindbrain neuroepithelium and neural crest. These data indicate that FGFs are involved in the control of regional induction and/or maintenance of Krox20 and mafB/kr expression, thus identifying a novel function for these factors in hindbrain development, in addition to their proposed more general role in early neural caudalization (Marin, 2000).

The segmentation of the vertebrate hindbrain into rhombomeres is highly conserved, but how early hindbrain patterning is established is not well understood. Rhombomere 4 (r4) functions as an early-differentiating signaling center in the zebrafish hindbrain. Time-lapse analyses of zebrafish hindbrain development show that r4 forms first and hindbrain neuronal differentiation occurs first in r4. Two signaling molecules, FGF3 and FGF8, which are both expressed early in r4, are together required for the development of rhombomeres adjacent to r4, particularly r5 and r6. Transplantation of r4 cells can induce expression of r5/r6 markers, as can misexpression of either FGF3 or FGF8. Genetic mosaic analyses also support a role for FGF signaling acting from r4. Taken together, these findings demonstrate a crucial role for FGF-mediated inter-rhombomere signaling in promoting early hindbrain patterning and underscore the significance of organizing centers in patterning the vertebrate neural plate (Maves, 2002).

Current evidence suggests that the anterior segment of the vertebrate hindbrain, rhombomere 1, gives rise to the entire cerebellum. It is situated where two distinct developmental patterning mechanisms converge: graded signaling from an organizing center (the isthmus) located where the midbrain/hindbrain boundary confronts segmentation of the hindbrain. The unique developmental fate of rhombomere 1 is reflected by its being the only hindbrain segment in which no Hox genes are expressed. Ectopic FGF8 protein, a candidate for the isthmic organizing activity, is able to induce and repress gene expression within the hindbrain in a manner appropriate to rhombomere 1. Using a heterotopic, heterospecific grafting strategy it has been demonstrated that rhombomere 1 is able to express Hox genes but that both isthmic tissue and FGF8 inhibit their expression. Inhibition of FGF8 function in vivo shows that it is responsible for defining the anterior limit of Hox gene expression within the developing brain and thereby specifies the extent of the r1 territory. A retinoid morphogen gradient determines the axial limit of expression of individual Hox genes within the hindbrain. A model is proposed wherein activation by retinoids is antagonized by FGF8, acting as an inhibitor in the anterior hindbrain, to set aside the territory from which the cerebellum will develop (Irving, 2000).

Specification and polarization of the midbrain and anterior hindbrain involves planar signals originating from the isthmus. Current evidence suggests that FGF8, expressed at the isthmus, provides this patterning influence. In this study, novel genes were sought that are involved in the process by which regional identity is imparted to midbrain and anterior hindbrain (rhombomere 1). An enhanced differential display reverse transcription method was used to clone cDNAs derived from transcripts expressed specifically in either rhombomere 1 or midbrain during the period of isthmic patterning activity. This gene expression screen has identified 28 differentially expressed cDNAs. A clone upregulated in cDNA derived from rhombomere 1 tissue shows a 91% identity at the nucleotide level to the putative human receptor tyrosine kinase antagonist: sprouty2. In situ hybridization on whole chick embryos shows chick sprouty2 to be expressed initially within the isthmus and rhombomere 1, spatially and temporally coincident with Fgf8 expression. However, at later stages this domain is more extensive than that of Fgf8. Introduction of ligand-coated beads into either midbrain or hindbrain region reveal that sprouty2 can be rapidly induced by FGF8. These data suggest that sprouty2 participates in a negative feedback regulatory loop to modulate the patterning activity of FGF8 at the isthmus (Chambers, 2000a).

In a differential display analysis to identify genes involved in patterning the mid/hindbrain region of the chick neural tube, a sprouty ortholog, sprouty2, has been identified. In the developing chick embryo there is a close correlation with known sites of FGF activity but little correlation with expression patterns of members of the EGF family. Initially, transcripts are associated with the primitive streak. During the period of neural tube patterning expression is detected in the anterior neuropore, in the isthmic region and in neural plate and posterior spinal cord. Transcripts are also detected in the otic placode, tail bud, mesoderm of the branchial arches, somitic myotome, retina, limb buds and gut mesenchyme; all are known sites of FGF action (Chambers, 2000b).

Hindbrain (brainstem) segmentation ultimately serves to organize the development of neuronal populations and their projections, and regional diversity is achieved through each segment having its own identity -- the latter being established through differential expression of a hierarchy of transcription factors, including Hox genes, Krox20, and Kreisler/Valentino. A novel signaling center has been identified in the zebrafish embryo that arises prior to establishment of segmental patterning; it is located centrally within the hindbrain territory in a region that corresponds to the presumptive rhombomere 4. Signaling from this region by two members of the FGF family of secreted proteins, FGF3 and FGF8, is required to establish correct segmental identity throughout the hindbrain and for subsequent neuronal development. Spatiotemporal studies of Fgf expression suggest that this patterning mechanism is conserved during hindbrain development in other vertebrate classes (Walshe, 2002).

Previous work on signals that regulate establishment of hindbrain segmental identity has mostly focused on the role of retinoic acid (RA) released from paraxial mesoderm at a distance from the hindbrain primordium. While RA clearly directly regulates expression of Hox genes, another role may be to position the Fgf signaling center within the presumptive hindbrain. Ectopic RA application to zebrafish embryos results in respecification of r2 to an r4 phenotype, and Fgf3 expression is induced in r2. By contrast, inhibition of RA function results in posterior expansion of anterior hindbrain such that the Fgf3 domain lies at somitic levels (Walshe, 2002).

These studies identify the presumptive r4 territory as a source of planar signals that serve to pattern the neural plate. Other such signaling centers are the isthmus, where Fgf8 functions at later stages to pattern midbrain and r1, and the anterior neural ridge, which patterns the telencephalon, in part by Fgf3 and Fgf8 signaling. Thus, the common feature of all three is that Fgf provides a planar signal to pattern adjacent neural territories, indicating that Fgf signaling has been coopted to impart regional identity multiple times during evolution of the vertebrate brain (Walshe, 2002).

Vertebrate hindbrain segmentation is a highly conserved process but the mechanism of rhombomere determination is not well understood. Recent work in the zebrafish has shown a requirement for fibroblast growth factor (Fgf) signaling and for the transcription factor variant hepatocyte nuclear factor 1 (vhnf1) in specification of rhombomeres 5 and 6 (r5+r6). vhnf1 functions in two ways to subdivide the zebrafish caudal hindbrain domain (r4-r7) into individual rhombomeres: (1) vhnf1 promotes r5+r6 identity through an obligate synergy with Fgf signals to activate valentino and krox20 expression; (2) vhnf1 functions independently of Fgf signals to repress hoxb1a expression. Although vhnf1 is expressed in a broad posterior domain during gastrulation, it promotes the specification of individual rhombomeres. This is achieved in part because vhnf1 gives cellular competence to respond to Fgf signals in a caudal hindbrain-specific manner (Wiellette, 2003).

Table of contents

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.