Independent roles for retinoic acid in segmentation and neuronal differentiation in the zebrafish hindbrain

Segmentation of the vertebrate hindbrain into rhombomeres is essential for the anterior-posterior patterning of cranial motor nuclei and their associated nerves. The vitamin A derivative, retinoic acid (RA), is an early embryonic signal that specifies rhombomeres, but its roles in neuronal differentiation within the hindbrain remain unclear. Here we have analyzed the formation of primary and secondary hindbrain neurons in the zebrafish mutant neckless (nls), which disrupts retinaldehyde dehydrogenase 2 (raldh2), and in embryos treated with retinoid receptor (RAR) antagonists. Mutation of nls disrupts secondary, branchiomotor neurons of the facial and vagal nerves, but not the segmental pattern of primary, reticulospinal neurons, suggesting that RA acts on branchiomotor neurons independent of its role in hindbrain segmentation. Very few vagal motor neurons form in nls mutants and many facial motor neurons do not migrate out of rhombomere 4 into more posterior segments. When embryos are treated with RAR antagonists during gastrulation, we observe more severe patterning defects than seen in nls. These include duplicated reticulospinal neurons and posterior expansions of rhombomere 4, as well as defects in branchiomotor neurons. However, later antagonist treatments after rhombomeres are established still disrupt branchiomotor development, suggesting that requirements for RARs in these neurons occur later and independent of segmental patterning. We also show that RA produced by the paraxial mesoderm controls branchiomotor differentiation, since we can rescue the entire motor innervation pattern by transplanting wild-type cells into the somites of nls mutants. Thus, in addition to its role in determining rhombomere identities, RA plays a more direct role in the differentiation of subsets of branchiomotor neurons within the hindbrain.     

Novel retinoic acid generating activities in the neural tube and heart identified by conditional rescue of Raldh2 null mutant mice

Retinoid control of vertebrate development depends upon tissue-specific metabolism of retinol to retinoic acid (RA). The RA biosynthetic enzyme RALDH2 catalyzes much, but not all, RA production in mouse embryos, as revealed here with Raldh2 null mutants carrying an RA-responsive transgene. Targeted disruption of Raldh2 arrests development at midgestation and eliminates all RA synthesis except that associated with Raldh3 expression in the surface ectoderm of the eye field. Conditional rescue of Raldh2(-/-) embryos by limited maternal RA administration allows development to proceed and results in the establishment of additional sites of RA synthesis linked to Raldh1 expression in the dorsal retina and to Raldh3 expression in the ventral retina, olfactory pit and urinary tract. Unexpectedly, conditionally rescued Raldh2(-/-) embryos also possess novel sites of RA synthesis in the neural tube and heart that do not correspond to expression of Raldh1-3. RA synthesis in the mutant neural tube was localized in the spinal cord, posterior hindbrain and portions of the midbrain and forebrain, whereas activity in the mutant heart was localized in the conotruncus and sinus venosa. In the posterior hindbrain, this novel RA-generating activity was expressed during establishment of rhombomeric boundaries. In the spinal cord, the novel activity was localized in the floorplate plus in the intermediate region where retinoid-dependent interneurons develop. These novel RA-generating activities in the neural tube and heart fill gaps in our knowledge of how RA is generated spatiotemporally and may, along with Raldh1 and Raldh3, contribute to rescue of Raldh2(-/-) embryos by producing RA locally.     

The zebrafish detour gene is essential for cranial but not spinal motor neuron induction.

The zebrafish detour (dtr) mutation generates a novel neuronal phenotype. In dtr mutants, most cranial motor neurons, especially the branchiomotor, are missing. However, spinal motor neurons are generated normally. The loss of cranial motor neurons is not due to aberrant hindbrain patterning, failure of neurogenesis, increased cell death or absence of hh expression. Furthermore, activation of the Hh pathway, which normally induces branchiomotor neurons, fails to induce motor neurons in the dtr hindbrain. Despite this, not all Hh-mediated regulation of hindbrain development is abolished since the regulation of a neural gene by Hh is intact in the dtr hindbrain. Finally, dtr can function cell autonomously to induce branchiomotor neurons. These results suggest that detour encodes a component of the Hh signaling pathway that is essential for the induction of motor neurons in the hindbrain but not in the spinal cord and that dtr function is required for the induction of only a subset of Hh-mediated events in the hindbrain.     

Retinoic acid signalling in the zebrafish embryo is necessary during pre-segmentation stages to pattern the anterior-posterior axis of the CNS and to induce a pectoral fin bud

A number of studies have suggested that retinoic acid (RA) is an important signal for patterning the hindbrain, the branchial arches and the limb bud. Retinoic acid is thought to act on the posterior hindbrain and the limb buds at somitogenesis stages in chick and mouse embryos. Here we report a much earlier requirement for RA signalling during pre-segmentation stages for proper development of these structures in zebrafish. We present evidence that a RA signal is necessary during pre-segmentation stages for proper expression of the spinal cord markers hoxb5a and hoxb6b, suggesting an influence of RA on anteroposterior patterning of the neural plate posterior to the hindbrain. We report the identification and expression pattern of the zebrafish retinaldehyde dehydrogenase2 (raldh2/aldh1a2) gene. Raldh2 synthesises retinoic acid (RA) from its immediate precursor retinal. It is expressed in a highly ordered spatial and temporal fashion during gastrulation in the involuting mesoderm and during later embryogenesis in paraxial mesoderm, branchial arches, eyes and fin buds, suggesting the involvement of RA at different times of development in different functional contexts. Mapping of the raldh2 gene reveals close linkage to no-fin (nof), a newly discovered mutant lacking pectoral fins and cartilaginous gill arches. Cloning and functional tests of the wild-type and nof alleles of raldh2 reveal that nof is a raldh2 mutant. By treating nof mutants with RA during different time windows and by making use of a retinoic acid receptor antagonist, we show that RA signalling during pre-segmentation stages is necessary for anteroposterior patterning in the CNS and for fin induction to occur.     

Retinoic acid-metabolizing enzyme Cyp26a1 is essential for determining territories of hindbrain and spinal cord in zebrafish

Retinoic acid (RA) plays a critical role in neural patterning and organogenesis in the vertebrate embryo. Here we characterize a mutant of the zebrafish named giraffe (gir) in which the gene for the RA-degrading enzyme Cyp26a1 is mutated. The gir mutant displayed patterning defects in multiple organs including the common cardinal vein, pectoral fin, tail, hindbrain, and spinal cord. Analyses of molecular markers suggested that the lateral plate mesoderm is posteriorized in the gir mutant, which is likely to cause the defects of the common cardinal vein and pectoral fin. The cyp26a1 expression in the rostral spinal cord was strongly upregulated in the gir mutant, suggesting a strong feedback control of its expression by RA signaling. We also found that the rostral spinal cord territory was expanded at the expense of the hindbrain territory in the gir mutant. Such a phenotype is the opposite of that of the mutant for Raldh2, an enzyme that synthesizes RA. We propose a model in which Cyp26a1 attenuates RA signaling in the prospective rostral spinal cord to limit the expression of hox genes and to determine the hindbrain-spinal cord boundary.     

Retinoic acid-dependent signaling pathways and lineage events in the developing mouse spinal cord

Studies in avian models have demonstrated an involvement of retinoid signaling in early neural tube patterning. The roles of this signaling pathway at later stages of spinal cord development are only partly characterized. Here we use Raldh2-null mouse mutants rescued from early embryonic lethality to study the consequences of lack of endogenous retinoic acid (RA) in the differentiating spinal cord. Mid-gestation RA deficiency produces prominent structural and molecular deficiencies in dorsal regions of the spinal cord. While targets of Wnt signaling in the dorsal neuronal lineage are unaltered, reductions in Fibroblast Growth Factor (FGF) and Notch signaling are clearly observed. We further provide evidence that endogenous RA is capable of driving stem cell differentiation. Raldh2 deficiency results in a decreased number of spinal cord derived neurospheres, which exhibit a reduced differentiation potential. Raldh2-null neurospheres have a decreased number of cells expressing the neuronal marker β-III-tubulin, while the nestin-positive cell population is increased. Hence, in vivo retinoid deficiency impaired neural stem cell growth. We propose that RA has separable functions in the developing spinal cord to (i) maintain high levels of FGF and Notch signaling and (ii) drive stem cell differentiation, thus restricting both the numbers and the pluripotent character of neural stem cells.     

Requirement of mesodermal retinoic acid generated by Raldh2 for posterior neural transformation

Studies in amphibian embryos have suggested that retinoic acid (RA) may function as a signal that stimulates posterior differentiation of the nervous system as postulated by the activation-transformation model for anteroposterior patterning of the nervous system. We have tested this hypothesis in retinaldehyde dehydrogenase-2 (Raldh2) null mutant mice lacking RA synthesis in the somitic mesoderm. Raldh2^−/− embryos exhibited neural induction (activation) as evidenced by expression of Sox1 and Sox2 along the neural plate, but differentiation of spinal cord neuroectodermal progenitor cells (posterior transformation) did not occur as demonstrated by a loss of Pax6 and Olig2 expression along the posterior neural plate. Spinal cord differentiation in Raldh2^−/− embryos was rescued by maternal RA administration, and during the rescue RA was found to act directly in the neuroectoderm but not the somitic mesoderm. RA generated by Raldh2 in the somitic mesoderm was found to normally travel as a signal throughout the mesoderm and neuroectoderm of the trunk and into tailbud neuroectoderm, but not into tailbud mesoderm. Raldh2^−/− embryos also exhibited increased Fgf8 expression in the tailbud, and decreased cell proliferation in tailbud neuroectoderm. Our findings demonstrate that RA synthesized in the somitic mesoderm is necessary for posterior neural transformation in the mouse and that Raldh2 provides the only source of RA for posterior development. An important concept to emerge from our studies is that the somitic mesodermal RA signal acts in the neuroectoderm but not mesoderm to generate a spinal cord fate.     

Analysis of Wnt8 for neural posteriorizing factor by identifying Frizzled 8c and Frizzled 9 as functional receptors for Wnt8

The dorsal ectoderm of vertebrate gastrula is first specified into anterior fate by an activation signal and posteriorized by a graded transforming signal, leading to the formation of forebrain, midbrain, hindbrain and spinal cord along the anteroposterior (A-P) axis. Transplanted non-axial mesoderm rather than axial mesoderm has an ability to transform prospective anterior neural tissue into more posterior fates in zebrafish. Wnt8 is a secreted factor that is expressed in non-axial mesoderm. To investigate whether Wnt8 is the neural posteriorizing factor that acts upon neuroectoderm, we first assigned Frizzled 8c and Frizzled 9 to be functional receptors for Wnt8. We then, transplanted non-axial mesoderm into the embryos in which Wnt8 signaling is cell-autonomously blocked by the dominant-negative form of Wnt8 receptors. Non-axial mesodermal transplants in embryos in which Wnt8 signaling is cell-autonomously blocked induced the posterior neural markers as efficiently as in wild-type embryos, suggesting that Wnt8 signaling is not required in neuroectoderm for posteriorization by non-axial mesoderm. Furthermore, Wnt8 signaling, detected by nuclear localization of beta-catenin, was not activated in the posterior neuroectoderm but confined in marginal non-axial mesoderm. Finally, ubiquitous over-expression of Wnt8 does not expand neural ectoderm of posterior character in the absence of mesoderm or Nodal-dependent co-factors. We thus conclude that other factors from non-axial mesoderm may be required for patterning neuroectoderm along the A-P axis.     

Cloning and embryonic expression of zebrafish neuropilin genes

Neuropilin (Nrp), a cell surface receptor for class 3 semaphorins and for certain heparin forms of vascular endothelial growth factors, functions in many biological processes including axon guidance, neural cell migration and angiogenesis in the development of the nervous system and the cardiovascular system. To understand the role of neuropilins in zebrafish embryogenesis, we have cloned three zebrafish neuropilin homologues, nrp1b, nrp2a and nrp2b. Based on synteny, zebrafish nrp1b and the previously cloned nrp1a are orthologous to human nrp1, and zebrafish nrp2a and 2b orthologous to human nrp2. We have characterized the expression patterns of these four zebrafish neuropilin genes in wild type embryos from the beginning of somitogenesis to 48 h post-fertilization. Zebrafish nrp1a is expressed in the neural tube including telencephalon, epithalamus, cells along the axonal trajectory of the posterior commissure and the medial longitudinal fascicle, hindbrain neurons, vagus motor neurons and spinal motoneurons. Zebrafish nrp1b is expressed in the nose, the cranial neural crest cell (NCC) derived tissue underlying the hypothalamus, endothelial precursors and the trunk and tail vasculature. Zebrafish nrp2a is expressed in telencephalon, anterior pituitary, oculomotor and trochlear motor neurons, cells along the supra-optic and posterior commissures, hindbrain rhombomere 1, hindbrain neurons, cranial NCCs and sclerotome. Zebrafish nrp2b is expressed in telencephalon, thalamus, hypothalamus, epiphysis, cells along the anterior and posterior commissures, post-optic and supra-optic commissures and the olfactory axonal trajectory, hindbrain neurons, cranial NCCs, somites and spinal cord neurons.     

Segmental development of reticulospinal and branchiomotor neurons in lamprey: insights into the evolution of the vertebrate hindbrain

During development, the vertebrate hindbrain is subdivided along its anteroposterior axis into a series of segmental bulges called rhombomeres. These segments in turn generate a repeated pattern of rhombomere-specific neurons, including reticular and branchiomotor neurons. In amphioxus (Cephalochordata), the sister group of the vertebrates, a bona fide segmented hindbrain is lacking, although the embryonic brain vesicle shows molecular anteroposterior regionalization. Therefore, evaluation of the segmental patterning of the central nervous system of agnathan embryos is relevant to our understanding of the origin of the developmental plan of the vertebrate hindbrain. To investigate the neuronal organization of the hindbrain of the Japanese lamprey, Lethenteron japonicum, we retrogradely labeled the reticulospinal and branchial motoneurons. By combining this analysis with a study of the expression patterns of genes identifying specific rhombomeric territories such as LjKrox20, LjPax6, LjEphC and LjHox3, we found that the reticular neurons in the lamprey hindbrain, including isthmic, bulbar and Mauthner cells, develop in conserved rhombomere-specific positions, similar to those in the zebrafish. By contrast, lamprey trigeminal and facial motor nuclei are not in register with rhombomere boundaries, unlike those of gnathostomes. The trigeminal-facial boundary corresponds to the rostral border of LjHox3 expression in the middle of rhombomere 4. Exogenous application of retinoic acid (RA) induced a rostral shift of both the LjHox3 expression domain and branchiomotor nuclei with no obvious repatterning of rhombomeric segmentation and reticular neurons. Therefore, whereas subtype variations of motoneuron identity along the anteroposterior axis may rely on Hox-dependent positional values, as in gnathostomes, such variations in the lamprey are not constrained by hindbrain segmentation. We hypothesize that the registering of hindbrain segmentation and neuronal patterning may have been acquired through successive and independent stepwise patterning changes during evolution.     

Development of noradrenergic neurons in the zebrafish hindbrain requires BMP, FGF8, and the homeodomain protein soulless/Phox2a

We report that the zebrafish mutation soulless, in which the development of locus coeruleus (LC) noradrenergic (NA) neurons failed 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.     

Nodal-related signals establish mesendodermal fate and trunk neural identity in zebrafish

The vertebrate body plan arises during gastrulation, when morphogenetic movements form the ectoderm, mesoderm, and endoderm. In zebrafish, mesoderm and endoderm derive from the marginal region of the late blastula, and cells located nearer the animal pole form the ectoderm [1]. Analysis in mouse, Xenopus, and zebrafish has demonstrated that Nodal-related proteins, a subclass of the TGF-beta superfamily, are essential for mesendoderm development [2], but previous mutational studies have not established whether Nodal-related signals control fate specification, morphogenetic movements, or survival of mesendodermal precursors. Here, we report that Nodal-related signals are required to allocate marginal cells to mesendodermal fates in the zebrafish embryo. In double mutants for the zebrafish nodal-related genes squint (sqt) and cyclops (cyc) [3] [4] [5], dorsal marginal cells adopt neural fates, whereas in wild-type embryos, cells at this position form endoderm and axial mesoderm. Involution movements characteristic of developing mesendoderm are also blocked in the absence of Nodal signaling. Because it has been proposed [6] that inhibition of Nodal-related signals promotes the development of anterior neural fates, we also examined anteroposterior organization of the neural tube in sqt;cyc mutants. Anterior trunk spinal cord is absent in sqt;cyc mutants, despite the presence of more anterior and posterior neural fates. These results demonstrate that nodal-related genes are required for the allocation of dorsal marginal cells to mesendodermal fates and for anteroposterior patterning of the neural tube.     

parachute/n-cadherin is required for morphogenesis and maintained integrity of the zebrafish neural tube

N-cadherin (Ncad) is a classical cadherin that is implicated in several aspects of vertebrate embryonic development, including somitogenesis, heart morphogenesis, neural tube formation and establishment of left-right asymmetry. However, genetic in vivo analyses of its role during neural development have been rather limited. We report the isolation and characterization of the zebrafish parachute (pac) mutations. By mapping and candidate gene analysis, we demonstrate that pac corresponds to a zebrafish n-cadherin (ncad) homolog. Three mutant alleles were sequenced and each is likely to encode a non-functional Ncad protein. All result in a similar neural tube phenotype that is most prominent in the midbrain, hindbrain and the posterior spinal cord. Neuroectodermal cell adhesion is altered, and convergent cell movements during neurulation are severely compromised. In addition, many neurons become progressively displaced along the dorsoventral and the anteroposterior axes. At the cellular level, loss of Ncad affects beta-catenin stabilization/localization and causes mispositioned and increased mitoses in the dorsal midbrain and hindbrain, a phenotype later correlated with enhanced apoptosis and the appearance of ectopic neurons in these areas. Our results thus highlight novel and crucial in vivo roles for Ncad in the control of cell convergence, maintenance of neuronal positioning and dorsal cell proliferation during vertebrate neural tube development.     

Role of retinoic acid during forebrain development begins late when Raldh3 generates retinoic acid in the ventral subventricular zone

Retinoic acid (RA) synthesized by Raldh3 in the frontonasal surface ectoderm of chick embryos has been suggested to function in early forebrain patterning by regulating Fgf8, Shh, and Meis2 expression. Similar expression of Raldh3 exists in E8.75 mouse embryos, but Raldh2 is also expressed in the optic vesicle at this stage suggesting that both genes may play a role in early forebrain patterning. Furthermore, Raldh3 is expressed later in the forebrain itself (lateral ganglionic eminence; LGE) starting at E12.5, suggesting a later role in forebrain neurogenesis. Here we have analyzed mouse embryos carrying single or double null mutations in Raldh2 and Raldh3 for defects in forebrain development. Raldh2(-/-);Raldh3(-/-) embryos completely lacked RA signaling activity in the early forebrain, but exhibited relatively normal expression of Fgf8, Shh, and Meis2 in the forebrain. Thus, we find no clear requirement for RA in controlling expression of these important forebrain patterning genes, but Raldh3 expression in the frontonasal surface ectoderm was found to be needed for normal Fgf8 expression in the olfactory pit. Our studies revealed that later expression of Raldh3 in the subventricular zone of the LGE is required for RA signaling activity in the ventral forebrain. Importantly, expression of dopamine receptor D2 in E18.5 Raldh3(-/-) embryos was essentially eliminated in the developing nucleus accumbens, a tissue lying close to the source of RA provided by Raldh3. Our results suggest that the role of RA during forebrain development begins late when Raldh3 expression initiates in the ventral subventricular zone.     

Retinoic acid synthesis and hindbrain patterning in the mouse embryo

Targeted disruption of the murine retinaldehyde dehydrogenase 2 (Raldh2) gene precludes embryonic retinoic acid (RA) synthesis, leading to midgestational lethality (Niederreither, K., Subbarayan, V., Dolle, P. and Chambon, P. (1999). Nature Genet. 21, 444-448). We describe here the effects of this RA deficiency on the development of the hindbrain and associated neural crest. Morphological segmentation is impaired throughout the hindbrain of Raldh2-/- embryos, but its caudal portion becomes preferentially reduced in size during development. Specification of the midbrain region and of the rostralmost rhombomeres is apparently normal in the absence of RA synthesis. In contrast, marked alterations are seen throughout the caudal hindbrain of mutant embryos. Instead of being expressed in two alternate rhombomeres (r3 and r5), Krox20 is expressed in a single broad domain, correlating with an abnormal expansion of the r2-r3 marker Meis2. Instead of forming a defined r4, Hoxb1- and Wnt8A-expressing cells are scattered throughout the caudal hindbrain, whereas r5/r8 markers such as kreisler or group 3/4 Hox genes are undetectable or markedly downregulated. Lack of alternate Eph receptor gene expression could explain the failure to establish rhombomere boundaries. Increased apoptosis and altered migratory pathways of the posterior rhombencephalic neural crest cells are associated with impaired branchial arch morphogenesis in mutant embryos. We conclude that RA produced by the embryo is required to generate posterior cell fates in the developing mouse hindbrain, its absence leading to an abnormal r3 (and, to a lesser extent, r4) identity of the caudal hindbrain cells.     

Gli2 and Gli3 play distinct roles in the dorsoventral patterning of the mouse hindbrain

Sonic Hedgehog (Shh) signaling plays a critical role during dorsoventral (DV) patterning of the developing neural tube by modulating the expression of neural patterning genes. Overlapping activator functions of Gli2 and Gli3 have been shown to be required for motoneuron development and correct neural patterning in the ventral spinal cord. However, the role of Gli2 and Gli3 in ventral hindbrain development is unclear. In this paper, we have examined DV patterning of the hindbrain of Shh(-/-), Gli2(-/-) and Gli3(-/-) embryos, and found that the respective role of Gli2 and Gli3 is not only different between the hindbrain and spinal cord, but also at distinct rostrocaudal levels of the hindbrain. Remarkably, the anterior hindbrain of Gli2(-/-) embryos displays ventral patterning defects as severe as those observed in Shh(-/-) embryos suggesting that, unlike in the spinal cord and posterior hindbrain, Gli3 cannot compensate for the loss of Gli2 activator function in Shh-dependent ventral patterning of the anterior hindbrain. Loss of Gli3 also results in a distinct patterning defect in the anterior hindbrain, including dorsal expansion of Nkx6.1 expression. Furthermore, we demonstrate that ventral patterning of rhombomere 4 is less affected by loss of Gli2 function revealing a different requirement for Gli proteins in this rhombomere. Taken together, these observations indicate that Gli2 and Gli3 perform rhombomere-specific function during DV patterning of the hindbrain.     

Retinaldehyde dehydrogenase 2 and Hoxc8 are required in the murine brachial spinal cord for the specification of Lim1+ motoneurons and the correct distribution of Islet1+ motoneurons

Retinoic acid (RA) activity plays sequential roles during the development of the ventral spinal cord. Here, we have investigated the functions of local RA synthesis in the process of motoneuron specification and early differentiation using a conditional knockout strategy that ablates the function of the retinaldehyde dehydrogenase 2 (Raldh2) synthesizing enzyme essentially in brachial motoneurons, and later in mesenchymal cells at the base of the forelimb. Mutant (Raldh2L-/-) embryos display an early embryonic loss of a subset of Lim1+ brachial motoneurons, a mispositioning of Islet1+ neurons and inappropriate axonal projections of one of the nerves innervating extensor limb muscles, which lead to an adult forepaw neuromuscular defect. The molecular basis of the Raldh2L-/- phenotype relies in part on the deregulation of Hoxc8, which in turn regulates the RA receptor RARbeta. We further show that Hoxc8 mutant mice, which exhibit a similar congenital forepaw defect, display at embryonic stages molecular defects that phenocopy the Raldh2L-/- motoneuron abnormalities. Thus, interdependent RA signaling and Hox gene functions are required for the specification of brachial motoneurons in the mouse.