embryology of the spine and spinal cord - springer · egg cell. the zygote divides iteratively...

53
Embryology of the Spine and Spinal Cord Martin Catala Contents Introduction ....................................................................................... 2 The Development of the Embryo After Fertilization ............................................ 2 Gastrulation ....................................................................................... 3 Mesoderm Induction .......................................................................... 3 Formation of the Primitive Streak and Hensens Node ...................................... 5 Ingression: The Motor of Gastrulation ....................................................... 7 Neural Induction .............................................................................. 7 Cell Movements During Gastrulation ........................................................ 11 Does the Neurenteric Canal Exist? ........................................................... 13 Neurulation ........................................................................................ 15 Primary Neurulation ........................................................................... 15 Secondary Neurulation ........................................................................ 20 Ventrodorsal Polarity of the Neural Tube ........................................................ 31 The Notochord Plays a Major Role in the Establishment of Floor Plate .................... 32 The Neural Crest and the Development of the Peripheral Nervous System .................... 36 Descriptive Embryology ...................................................................... 36 Morphogenesis of the Most Caudal Part of the Spinal Cord ................................ 39 The Development of the Spine ................................................................... 41 Epithelial Somites Dissociate to Form Their Derivatives .................................... 41 The Somite Is Not Polarized According to the Ventrodorsal Axis .......................... 43 Cellular and Molecular Players for Somitic Polarization .................................... 43 The Neural Tube and the Notochord Induce the Formation of the Vertebral Cartilage ..... 43 Conclusion ........................................................................................ 44 References ........................................................................................ 45 M. Catala (*) Fédération de Neurologie, Groupe Hospitalier Pitié-Salpêtrière, Paris, France Laboratoire de biologie du développement, UMR 7622, CNRS and université Pierre et Marie Curie, Paris, France e-mail: [email protected] # Springer-Verlag Berlin Heidelberg 2015 A. Rossi (ed.), Pediatric Neuroradiology , DOI 10.1007/978-3-662-46258-4_70-1 1

Upload: phungduong

Post on 16-Aug-2019

215 views

Category:

Documents


0 download

TRANSCRIPT

Embryology of the Spine and Spinal Cord

Martin Catala

ContentsIntroduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2The Development of the Embryo After Fertilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2Gastrulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Mesoderm Induction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Formation of the Primitive Streak and Hensen’s Node . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Ingression: The Motor of Gastrulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Neural Induction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Cell Movements During Gastrulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Does the Neurenteric Canal Exist? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Neurulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Primary Neurulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Secondary Neurulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Ventrodorsal Polarity of the Neural Tube . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31The Notochord Plays a Major Role in the Establishment of Floor Plate . . . . . . . . . . . . . . . . . . . . 32

The Neural Crest and the Development of the Peripheral Nervous System . . . . . . . . . . . . . . . . . . . . 36Descriptive Embryology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36Morphogenesis of the Most Caudal Part of the Spinal Cord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

The Development of the Spine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Epithelial Somites Dissociate to Form Their Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41The Somite Is Not Polarized According to the Ventrodorsal Axis . . . . . . . . . . . . . . . . . . . . . . . . . . 43Cellular and Molecular Players for Somitic Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43The Neural Tube and the Notochord Induce the Formation of the Vertebral Cartilage . . . . . 43

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

M. Catala (*)Fédération de Neurologie, Groupe Hospitalier Pitié-Salpêtrière, Paris, France

Laboratoire de biologie du développement, UMR 7622, CNRS and université Pierre et Marie Curie,Paris, Francee-mail: [email protected]

# Springer-Verlag Berlin Heidelberg 2015A. Rossi (ed.), Pediatric Neuroradiology,DOI 10.1007/978-3-662-46258-4_70-1

1

AbstractThe first phases of the development of central nervous system begin withgastrulation, which is a very important morphogenetic time ensuring the forma-tion of the three germ layers. Gastrulation takes place according to several steps:mesodermal induction, the formation of the primitive streak and Hensen’s node,and neural induction. At the end of this phase, the neural plate is built up, thisstructure secondarily undergoes movements of neurulation that allow the forma-tion of the neural tube. In mammals and humans in particular, neurulation inrostral regions involves up-folding movements of the neural plate that closes onthe midline. This process is called primary neurulation. The secondary neurula-tion involves cavitation of a solid cylinder at the most caudal part of theembryonic axis. However, the most recent data show that the two processestake place in continuity and the tissues that form the secondary tube are initiallylocated at the embryonic surface. The region between the two types of neurulationundergoes specific movements which have been recently described under theterm junctional neurulation. The neural tube that has been formed whatever theprocess undergoes polarizing influences leading to the formation of multipleventral and dorsal domains. Each of these areas gives rise to specific cell fatesafter this polarization. The most important ventralizing molecule is Sonic hedge-hog while Wnt and BMP play a dorsalizing role. The peripheral nervous system isformed after migration of the so-called neural crest cells that arise from the dorsalpart of the neural tube. At lmast the spine derives from paraxial mesodermorganized into metameric units called somites.

Introduction

In humans, as in all vertebrates, the central nervous system (CNS) emerges from theneuroectoderm, a differentiated tissue deriving from the outermost layer of theembryo. To understand the development of the CNS, it is necessary to return tothe earliest stages of embryogenesis, namely, just after fertilization. I will present afirst general part explaining the phases ensuring the development of theneuroectoderm. These steps are common to the entire CNS and not just for thespinal cord. The second part will be devoted more specifically to the spinal cord andspine. These two organs form a couple of structures whose development is highlycoordinated explaining why an abnormal development of one structure is usuallyassociated with the maldevelopment of the other.

The Development of the Embryo After Fertilization

Fertilization results from the meeting of two haploid gametes produced during theprocess of meiosis. This meeting allows restoration of diploidy that is mandatory forthe subsequent embryonic development and ensures the formation of the zygote or

2 M. Catala

egg cell. The zygote divides iteratively generating two, four, eight, and sixteenidentical cells. At this time, a first cell differentiation process is taking place inhumans; peripheral cells flatten, develop intercellular junctions, and take the appear-ance of an epithelium, the trophectoderm. Inner cells form a cluster apposed at onepole of the trophectoderm. This inner cell mass contains embryonic stem (ES) cells,whose properties interest searchers and physicians for their potential capacities inregenerative medicine. Then, a cavity is built within the inner cell mass that fore-shadows the future amniotic cavity. Amnioblasts form the roof and the walls of theamniotic cavity, whereas its floor is organized in two epithelial layers. The superfi-cial epithelial layer is in contact with the amniotic cavity and is named epiblast,while the deep layer forms the extraembryonic endoderm. Ancient authors mistak-enly believed that these two layers were involved in the formation of the embryo,explaining the name of “didermic stage” once given at this step of development.Nowadays, it is currently known that only epiblast generates embryonic structures,while the extraembryonic endoderm is contributing to the formation of extraembry-onic tissues. It is therefore perfectly obsolete to talk of a didermic stage to charac-terize this developmental step. Furthermore, some authors had described a localizedthickening of the anterior primitive endoderm, a differentiation they calledprochordal plate. Indeed, they emitted the hypothesis that this structure could yielda portion of the axial mesoderm or notochord. It is now shown that this assumption iswrong, and thus the term prochordal plate should be banned from the textbooksnowadays. Such a thickening of the extraembryonic endoderm is also demonstratedin rabbits and mice. It consists of cells expressing the gene Hex that are fated toconstitute the future anterior visceral endoderm, a structure that will play a pivotalrole during gastrulation (Rivera-Pérez et al. 2003).

Gastrulation

Gastrulation is a pivotal step of embryonic morphogenesis that ensures the estab-lishment of the three germ layers. These layers, ectoderm, mesoderm, and endoderm,are the basis for the formation of embryonic structures. This phase of development isa broad process that is involved in the embryonic development of all multicellulartriploblastic organisms; thus, gastrulation is a very ancient phylogenetical process. Inhumans, gastrulation takes place during the third week of development and itconsists of several steps (mesodermal induction, formation of the primitive streakand Hensen’s node, mesendodermal ingression, and neural induction).

Mesoderm Induction

The pioneering works of the Dutch embryologist Peter Nieuwkoop demonstrated theexistence of mesodermal induction in 1969 (Nieuwkoop 1969a, b). Combiningvegetative blastomeres with animal cells of an amphibian blastula, Nieuwkoopshowed that animal cells turned into mesoderm under the action of signals from

Embryology of the Spine and Spinal Cord 3

the vegetative blastomeres. Mesoderm induction precedes neural induction that wasdemonstrated 50 years before by Hans Spemann’s school.

Mesodermal induction has been studied in many vertebrates that are consideredas model organisms in developmental biology: zebrafish, Xenopus, chicken, rabbit,and mouse. No study can be performed in humans because (i) it is impossible to haveaccess to these very early stages of development; and (ii) no experimental study canbe drawn for such a purpose. So, the data that I will present are largely extrapolatedfrom those obtained in these different models.

Four molecular pathways are now recognized as regulating mesoderm inductionin these different vertebrate species: nodal, bone morphogenetic protein (BMP),fibroblast growth factor (FGF), and canonical Wnt (Kimelman 2006). Nodal is afamily of molecules belonging to the transforming growth factor (TGF)-β superfam-ily. The molecules of the nodal family are secreted and act on membrane receptorspresent in the form of heterotetrameric proteins, composed of two type 1 proteinsand two type 2. Ligand binding with its receptor induces phosphorylation ofcytoplasmic proteins, Smad 2 and 3, that associate secondarily with Smad 4 protein.This protein complex translocates to the nucleus and acts as a transcription factor.BMPs are a multiprotein family owned to the TGFβ superfamily. They are secretedand act on membrane heterotetrameric receptors that are molecularly different fromthose that receive the precedent signaling. The activation of these receptors byligands leads to phosphorylation of cytoplasmic proteins Smad 1, 5, and 8 thatcombine secondarily with Smad 4 to act as transcription factor. FGFs act oncytoplasmic receptors present as homodimers and which cause a series of phosphor-ylation of cytoplasmic molecules leading to the activation of MAP kinases whichtranslocate into the nucleus to phosphorylate transcription factors for their activation.Finally, the Wnt canonical pathway leads to stabilization of the beta-catenin thatenters the nucleus and activates transcription in partnership with the Tcf/LEFprotein.

In mice, different molecules have been implicated in mesoderm induction: nodal,GDF3, BMP2, BMP4, BMP7, FGF3, FGF9, Wnt3, Wnt3a, and Wnt8 (Kimelman2006). It is largely outside the scope of this chapter to describe all the availableexperimental data. The central role played by the message mediated by both nodaland BMP families can be demonstrated in mice by knock-outing the gene coding forthe common effector Smad4 in which the mesoderm is not formed (Yang et al. 1998).It is the same for canonical Wnt pathway, the invalidation of the gene coding forbeta-catenin leading to a complete absence of mesoderm in mice (Huelskenet al. 2000). Furthermore, it is interesting to note that certain factors are expressedby the extraembryonic tissues, such as BMP4 being produced by the extraembryonicepithelial tissue in contact with the epiblast in mice and all around this embryonicarea (Lawson et al. 1999). BMP4 cannot act on the future embryonic cephalic poledue to the synthesis of inhibitors by anterior visceral endoderm (Lawson et al. 1999).Other factors like nodal are expressed by the posterior epiblast, where the mesodermwill arise (Conlon et al. 1994).

4 M. Catala

Formation of the Primitive Streak and Hensen’s Node

The first morphological event taking place during gastrulation in amniotes is theformation of the primitive streak, a line orientated rostrocaudally that represents thefuture zone of mesodermal ingression. The rostralmost end of the primitive streak wasfirst described in 1875 by Victor Hensen in rabbit embryos and is consequently knownas Hensen’s node. This development time is a key event in embryonic differentiation,explaining the importance of its study for understanding its construction. However, thedifferent experimental animal models that have been studied show that the strategy tobuild is not unique and varies in chickens, rabbits, and mice. Under these conditions,the exact scenario of development is still largely uncertain in humans.

The formation of the streak has been particularly well studied in birds. Theprimitive streak is induced by the posterior marginal zone (an extraembryonic regionthat is in close contact with the embryonic disk) (Bachvarova et al. 1998). Thisregion induces the formation of the node without contributing to it. This allowsClaudio Stern’s group to propose a homology between the posterior marginal zoneand Nieuwkoop center that is responsible for the induction of the dorsal mesoderm inamphibians. The avian embryonic region expresses two genes coding for secretedfactors Vg1 and Wnt8C. The coexpression of these two genes leads to the inductionof nodal in the epiblast in contact with the posterior marginal zone (Skromme andStern 2002) (Fig. 1). In chicken, the formation of precursor cells of the primitivestreak is restricted to the posterior region by the hypoblast (the avian homologue of

Fig. 1 The embryonic disk is in close contact with the marginal zone (an extraembryonic regionthat expresses Wnt8c). Furthermore, the posterior marginal zone expresses also Vg1. Thecoexpression of these two genes leads to the upregulation of nodal in the overlying epiblast

Embryology of the Spine and Spinal Cord 5

the extraembryonic endoderm). The anterior hypoblast secretes antinodal moleculesand thus protects the future anterior pole of the embryo to mesodermal influences(Bertocchini and Stern 2002). Nodal-expressing cells will adopt a migratory pheno-type, moving from the posterior border of the embryonic disk to its center toeventually form the Hensen’s node. In chicken, intense cell movements take placein the epiblast during the formation of the primitive streak. These movements weredescribed in 1929 by L. Graper under the name “polonaise”movements because theyevoked a traditional dance of this country. Before the beginning of the formation ofthe primitive streak, epiblast cells undergo large movements. The cells located on themidline of the caudal end migrate to the rostral pole. Paramedian cells undergo lateralmovement of divergence that leads to change of the direction of migration when theyreach the lateral embryonic region. There, they migrate in the opposite from therostral to caudal region where they are attracted to the midline. During the beginningof the formation of the primitive streak, lateral epiblast cells are attracted to themidline leading to cellular condensation and increased density (Fig. 2).

In rabbits, the initial movements are quite similar to those described in the chick,but they are more restricted in space (but the embryo is smaller than the avian one)(Viebahn et al. 2002; Halacheva et al. 2011). Lateral cells migrate both caudalwardand toward the midline. Medial cells undergo a forward movement for the generationof the rostral part of the primitive streak. However, in a second step, the median cellsmigrate backward leading to the formation of the caudal primitive streak (Viebahnet al. 2002). In rabbit also, the anterior extraembryonic endoderm restricts theformation of mesoderm to the posterior pole thanks to the secretion of anti-Wntand anti-BMP/nodal proteins (Idkowiak et al. 2004).

Fig. 2 Cell movements during the first stages of gastrulation as described by Gräper and byWetzel.Superficial cells convergence toward the primitive streak and the node. Deep cells diverge laterally(Gräper)

6 M. Catala

In contrast, in mice, the migration movements of gastrulation are practically absent(Williams et al. 2012). The primitive streak is formed in situ by performing a differen-tiation according to a caudo-rostral gradient (Williams et al. 2012). This result, techni-cally challenging and recently obtained, is quite unexpected. It leads to amajor questionas to the mode of gastrulation and formation of the primitive streak in mammals. It isclear that the avian model cannot be extrapolated in mammals. In addition, thedifferences between rabbits and mice lead one to suggest that there is no unambiguousway that could explain these morphogenetic processes in mammals. Because the studyin humans is neither technically feasible nor ethically acceptable, it is imperative toincrease the experimental models (e.g., pig, bovine, but also primates) in an attempt toidentify a strategy that could be common to a group of mammals. Studies of compar-ative embryology could provide insight to the mechanism used in humans. Further-more, in mice as in both chicks and rabbits, the formation of the primitive streak isrestricted to the posterior region by the anterior visceral endoderm via the secretion ofmolecules with antinodal and anti-Wnt action (Perea-Gomez et al. 2002).

Ingression: The Motor of Gastrulation

Cell movements at the primitive streak were called “ingression” by Lucien Vakaet inthe early 1980s. The term ingression implies that epithelial cells lose their characterand migrate alone; it is therefore opposed to that of invagination that implies aninternalization of an epithelial sheet that keeps its histological organization. Ingres-sion takes place in three successive stages. The first is loss of the basal lamina, aprocess which is closely related to mesodermal induction. It is effectively abolishedin mouse mutants Nodal�/� or Wnt3�/� that have both a loss of mesodermalinduction and failure of disintegration of the basal lamina underlying the primitivestreak (Sun et al. 1999). Once this first step is achieved, the second stage is epithelial-mesenchymal transition (EMT) or ingression itself. During this process, a singleepithelial cell located within the epiblast undergoes an apical shrinkage and becomesbottle shaped. Its cell body moves to the inner region, leading to a loss of contactwith the surface to become a mesenchymal cell located between epiblast andextraembryonic endoderm. This step is impaired in the mutant mouse Snail�/� inwhich epiblast cells can undergo EMT but re-epithelization inevitably follows(Carver et al. 2001). Finally, the newly generated mesenchymal cell undergoesmigration provided by FGF signaling (Sun et al. 1999).

Neural Induction

Hans Spemann and the Concept of Neural Induction

Interspecific Chimeras in AmphibiansHans Spemann developed a technique in which a region of an embryo from Tritontaeniatus is grafted into a host from Triton cristatus. The cells coming from these

Embryology of the Spine and Spinal Cord 7

two species of amphibians can be easily recognized since the cytoplasm of onespecies is pigmented whereas it is not in the other (see Spemann (1938) for a review).Using this technique, it is thus possible to follow the fate of a graft from one speciesinto the other one.

Neural Induction in AmphibiansThis technique allowed Hans Spemann and Hilde Mangold to study the effect of thegraft of the dorsal lip of the blastopore into the ventral region of the embryo(experiments realized in 1921 and published in 1924) (Spemann 1938). The blasto-pore is the embryonic region through which cells invaginate during gastrulation inamphibians. After this type of graft, the endogenous axis of the embryo developsnormally, but an extra axis develops in the ventral side of the embryo (see Spemann(1938) for a review). Thanks to the cell specific marker, it is possible to conclude thatthe graft gives rise to notochordal, somitic, and floor plate cells, whereas theremainder of the neural tube derives from the host (Figs. 3 and 4). The conclusionof this experiment is that the dorsal lip of the blastopore is able to recruit ventralectodermal cells to become neural cells. This modification of the fate of the cells ofthe host has been termed neural induction. Furthermore, the dorsal lip of theblastopore is considered as the organizer of the amphibian embryo. In amnioteembryos, the organizer corresponds to the extreme anterior part of the primitivestreak, namely, the Hensen’s node in birds and mammals.

The organizer has two distinct properties: (i) it self-differentiates and gives rise toderivatives that include the notochord and floor plate of the neural tube; and (ii) itinduces neighboring structures into neural tissue. This induction can result from twotheoretical processes: either directly from the organizer itself before its own self-

Fig. 3 The graft of an unpigmented dorsal lip of the blastopore into a pigmented host leads to theformation of a double axis: (a) the endogenous axis derives exclusively from the host (blackarrows); (b) the induced axis (black arrows) is from both origins; the midline derives from thegraft (white arrowheads), whereas the rest of the neural plate arises from the host (From Spemann1938)

8 M. Catala

differentiation or from secondary tissues resulting from its own differentiation. Thisproblem was already present in Spemann’s mind (Spemann 1938).

Johannes Holtfreter, a former student of Spemann, published in 1933 a majorarticle in which he disrupted gastrulation and created what he called “exogastrula.”The latter prevents the invagination of the endoderm and the mesoderm inside thecavity of a blastula, generating a larva in which the ectoderm is separated from theendomesoderm, thereby the differentiation of ectoderm could be studied in theabsence of underlying structures generated during gastrulation. Holtfreter concludedfrom his experiments that neural induction is secondary to the action of the axialmesoderm on the overlying ectoderm. It is interesting to note that Hans Spemannwas not convinced by this demonstration (Spemann 1938, pp. 155–156).

Interpretations of such experiments have been reviewed more than 50 years afterHoltfreter thanks to tissue differentiation markers. The ectoderm of the exogastruladoes not express epidermal keratin, contrary to what is expected by Holtfreter’smodel (Kintner and Melton 1987). In addition, the ectoderm expresses neural

Fig. 4 Histological section of the secondary axis induced by the graft of an unpigmentedblastoporal lip into a pigmented host. The graft gives rise to the floor plate of the neural tube, thenotochord, and a part of the somites. sec. Ch., secondary notochord; sec. Med., secondary neuraltube; sec. Pron., secondary intermediate mesoderm; sec. Uw., secondary somites (l. left; r. right)(From Spemann 1938)

Embryology of the Spine and Spinal Cord 9

markers even in the absence of underlying mesoderm (Doniach et al. 1992). Thesefacts demonstrate that the organizer acts directly on neural induction through signalsthat travel in the plane of the epithelium, explaining the name of “planar induction”given to this phenomenon. In addition, to my knowledge, no experience has shown aprimary neural induction after transplanting a differentiated notochord. Thus, unlikemany data scattered in the reference books of the medical literature, one should notassign neural induction to the notochord of naive ectoderm, but rather incriminatethe node itself.

Molecular Biology of Neural InductionThe neural induction as defined by Spemann and Mangold generated a lot ofexperimental works. However, the identification of molecules possibly involved inthis process only started in 1991, that is, 70 years after the first experiencesperformed by Hilde Mangold. The situation of neural induction has gained a lot ofcomplexities from the discovery of the molecular network involved in this process.

The first molecular data regarding neural induction were obtained by searchingfor genes highly expressed by the amphibian organizer. This approach led the groupof Eddy De Robertis (UCLA) to discoverGoosecoid, a gene encoding a transcriptionfactor belonging to the family of homeobox genes. This gene is highly expressed inthe dorsal lip of the blastopore in Xenopus laevis (Blumberg et al. 1991), and itsventral ectopic expression leads to a duplication of the embryonic axis (Choet al. 1991), reproducing the transplant of the dorsal lip of the blastopore performedby Spemann and Mangold. Subsequently, other genes were identified in this region:Noggin (Smith and Harland 1992), Chordin (Sasai et al. 1994), Follistatin(Hemmati-Brivanlou et al. 1994), and Siamois (Lemaire et al. 1995). All of thesegenes are able to induce a second embryonic axis when expressed ectopically in theamphibian embryo.

These results led to the classic model of neural induction: the organizer expressesa secreted molecule that acts on the epiblast to induce transformation intoneuroectoderm. However, the multiplicity of discovered inducer molecules and theabsence of abnormality of neural induction observed in the knockout mice led to analternative direction.

The blockade of the activin receptor II (a receptor for molecules belonging to thesuperfamily of TGF beta) causes neuralization (Hemmati-Brivanlou and Melton1994). Ectodermal induction is provided by molecules of the TGFβ superfamily,such as BMP4 (Wilson and Hemmati-Brivanlou 1995) or BMP2 and BMP7 (Suzukiet al. 1997). This led to the genesis of the so-called default model. Therein, theepiblast cells that receive a BMP signal become epidermal cells, while those that donot undergo the change into neuroectodermal cells. This process is more plausiblesince several secreted molecules initially described as neural inducers actually playan anti-BMP role (Hemmati-Brivanlou et al. 1994; Piccolo et al. 1996; Zimermanet al. 1996).

However, these models are too simple to explain the complex reality of neuralinduction. Claudio Stern’s group showed that FGF induces, in chicken, the expres-sion of cERNI (early response to neural induction), a gene considered a marker of

10 M. Catala

neural tissue (Streit et al. 2000). The inhibition of FGF by a chemical compound thatis not specific to this pathway leads to a defect of cERNI activation (Streitet al. 2000). However, the role of this gene remains to be established in the processof neural induction, thus weakening the authors’ conclusion that FGF is a (the)neuralizing factor. Subsequently, it was shown that FGF alone cannot induce neuraltissue but requires a concomitant inhibition of Wnt (Wilson et al. 2001). Then,Linker and Stern (2004) showed that factors other than FGF and inhibition of Wntare necessary. Finally, recent experiments suggest that neural induction occurs insteps controlled by different genes (Pinho et al. 2011). Thus, the molecular mech-anism of neural induction is probably not univocal, but this process is achievedthrough gradual and progressive changes that lead to the establishment ofneuroectodermal identity.

The final result of neural induction is the partitioning of the primitive ectoderminto two different derivatives: the neural plate, located in the center of the embryonicdisk, and the lateral ectoderm. The neural plate is the primordium of the centralnervous system, whereas the lateral ectoderm yields the epidermis.

Cell Movements During Gastrulation

The cells of the node and primitive streak will give rise to mesodermal derivatives.Node cells give rise to many types of derivatives (Psychoyos and Stern 1996). Amongthese are the cells fated to generate axial mesoderm. The first axial mesodermal cells todevelop are the cells of the prechordal plate, followed by the cells of the notochord(Psychoyos and Stern 1996). Axial mesodermal cells grow following a single cellstream (Psychoyos and Stern 1996) and not by the meeting of two streams, aspostulated to account for the formation of split notochord. Anyway, it is important tokeep in mind that the formation of the notochord is a complex process that does not usethe same mechanisms according to the anteroposterior axis. At the stage of primitivestreak (namely, at the beginning of gastrulation), precursors of the notochord are notlocalized in a single region of the embryo both in chick (Psychoyos and Stern 1996) andmouse embryos (Kinder et al. 2001). Therefore, the contribution of each primordium issegmental, giving rise to a limited part of the notochord (Psychoyos and Stern 1996). Adisruption of this step of development could account for segmental agenesis of thenotochord thatwould lead to a secondary segmental defect of the spinal cord, as definedin segmental spinal dysgenesis (Tortori-Donati et al. 1999). The notochord formationmechanisms in mice also help to highlight regional differences to distinguish threedifferent areas (head, trunk, and tail) (Yamanaka et al. 2007). All these data show thatthe formation of the notochord during gastrulation is not a uniform phenomenon andsegmental defects of this organ may occur.

In a very schematic way, it should also be noted that the direction of morphoge-netic movements of the notochord changes with time. First, cells originating fromthe node invaginate rostrally to form the axial mesoderm, which is covered by theneural plate (Patten et al. 2003) (Fig. 5). This mode of formation of the notochordaccounts for the development of axial organs, from the telencephalon to the

Embryology of the Spine and Spinal Cord 11

hindbrain (Patten et al. 2003). After this period, the morphogenetic movements thatgive rise to the notochord act on a reverse direction, i.e., from rostral to caudal(Catala et al. 1996) (Fig. 6). An impairment of these axial elongating movementscould explain the occurrence of the so-called caudal agenesis (i.e., caudal regression)syndromes (Catala 2002).

The classic model of notochord formation in humans involves ingression from thenode followed by an integration of the precursors into the endoderm, to give rise tothe so-called notoplate. At last, the cells of the notoplate roll up to form the definitivenotochord. This morphogenesis does not exist in the chick embryo, where thenotochord never undergoes integration movements in the endoderm (Jurand 1962).In mammals, the situation is well known in mice. The initial growth of axialmesoderm precursors takes place immediately inside the endoderm (Yamanakaet al. 2007; Jurand 1974; Sulik et al. 1994). There is no phase of secondary

Fig. 5 Mesodermal movements at the beginning of gastrulation. The axial mesoderm originatesfrom the node and grows from caudal to rostral. The rest of the mesoderm ingresses and divergeslaterally to form the different domains of the mesoderm

12 M. Catala

intercalation in the endoderm as classically described in humans. Such a process hasalso been described in bovine embryos (Haldiman and Gier 1981). The conservationof such movements during morphogenesis in two phylogenetically distant mammalsshould lead one to wonder if such a process is constant in mammals. Therefore, itwould be very interesting to study other mammalian species, including humans, byscanning electron microscopy.

Lateral mesodermal cells invaginate through the primitive streak, diverge laterally,and add to the already formed mesoderm (Figs. 5 and 6). This mode of growth isresponsible for the development of the paraxial, intermediate, and lateral domains of themesoderm. It is interesting to note that the rostrocaudal organization of the primitivestreak foreshadows the future mediolateral mesodermal patterning: the more rostral acell in the primitive streak, the more medial its derivatives (Figs. 5 and 6).

Does the Neurenteric Canal Exist?

A direct communication between the endodermal tract and the neural tube has beendescribed in Amphioxus by Kowalevsky (1877) (Fig. 7). This canal was subse-quently described in amphibians, fishes, and reptiles. In chick (Catala et al. 1996;Jurand 1962), mouse (Jurand 1974; Sulik et al. 1994), and bovine (Haldiman andGier 1981), the neurenteric canal does not exist. In humans, the situation is far from

Fig. 6 Mesodermalmovements from six-somitestage onward. The axialmesoderm derives from thenode and moves from rostralto caudal. The othermesodermal domains set up aspreviously described. NTneural tube, So somite, PAMparaxial mesoderm, IMintermediate mesoderm, LMlateral mesoderm

Embryology of the Spine and Spinal Cord 13

simple. Old analyses of human embryos reported the presence of a neurenteric canal(O’Rahilly et al. 1987); however, the degree of preservation of these specimens wasconsidered poor (O’Rahilly and M€uller 1981). According to O’Rahilly and M€uller(1981), these data cannot be used now and new observations on well-preservedembryos should be made; furthermore, they noticed that the neurenteric canal is farfrom being constant in stage 8 human embryos, since it was absent in 10 among11 cases. We can conclude from these studies that the neurenteric canal is not aprominent feature of human embryos and is possibly only artifact. For this reason,one should be very cautious in explaining human malformations based on a defect ofthe so-called neurenteric canal, such as was proposed in the split notochord syn-drome (Pang et al. 1992). Furthermore, careful analysis of human embryosdisplaying malformations could be very informative. Kirillova et al. (2000) reporteda study of Sonic hedgehog expression in various human embryos presenting neuraltube defects. They described duplication of the notochord in the absence ofdiastematomyelia and diastematomyelia with a single notochord. Thus, the conceptof a duplication of the notochord explaining diastematomyelia is far from being auniversal observation. Other explanations are needed to account for these types ofmalformations.

Fig. 7 Section of anamphioxus embryo. Thelumens of the neural tube (n)and the gut (d ) communicatethrough the neurenteric canal(eo). Five somites are alreadyformed (1–5). Ch notochord,n’ neuropore (FromKowalevsky 1877)

14 M. Catala

Neurulation

The neural plate will undergo a series of morphogenetic movements to generate theneural tube. However, the situation is quite complex for the human spinal cord, sincetwo different basic mechanisms account for neurulation, i.e., primary and secondaryneurulation. Furthermore, a transition program is described in between the twoprevious zones.

Primary Neurulation

I will follow the classic description made by Gary Schoenwolf and his group(Schoenwolf and Smith 1990; Colas and Schoenwolf 2003). After neural induction,the neural plate appears as a sheet of cells located at the dorsal and medial part of theflat embryo. This cellular sheet will undergo a series of morphogenetic movementsthat can be separated into two phases.

Shaping of the Neural PlateDuring shaping, the neural plate will undergo a series of cell modifications allowingthe extension of its medial part. This can be illustrated after marking the early neuralplate and studying its subsequent deformation (Schoenwolf and Sheard 1989)(Fig. 8). It is easy to recognize three basic forces that account for this shaping.

Apicobasal ThickeningThe height of the cells of neural plate increases due to the modification of cellularshape, changing from cuboidal into prismatic.

Transverse NarrowingThis narrowing is due to both transverse narrowing of the cells and cellrearrangement (intercalation). Transverse narrowing is secondary to cell shapemodifications as analyzed previously. Cell rearrangement is explained by migrationof cells that converge to the midline, allowing an extension of the axis (Fig. 9). Suchmovements are a general feature during development in vertebrates. Indeed,convergence-extension is the driving major force of gastrulation in amphibians.

Longitudinal LengtheningThis can be explained by both the previous movements of intercalation and by celldivisions, which are oriented according to the embryonic axis (Schoenwolf andAlvarez 1989) (Fig. 9). This lengthening is accompanied by two rounds ofmediolateral cell rearrangement and two or three rounds of cell division.

All the forces that account for the shaping of the neural plate are intrinsic to theplate itself (Schoenwolf 1991; Moury and Schoenwolf 1995). Convergence-extension has been particularly well studied in amphibians. Its molecular regulationhas been unraveled with the demonstration of the role played by Dishevelled(Wallingford et al. 2000). This leads to the recognition of the importance of the

Embryology of the Spine and Spinal Cord 15

Wnt/PCP pathway in this morphogenesis. Wnt/PCP is a noncanonical pathway forWnt signaling that does not pass through the action of beta-catenin. Wnt/PCP isresponsible for the polarization of epithelial cells at their apical edge explaining itsanteroposterior polarity. Several mouse mutants involving proteins of the PCPpathway display defect in neural tube formation indicating the important role playedby these molecules in this process. VANGL1, gene involved in the Wnt / PCPpathway, is mutated in some forms of human neural tube defects (Kibaret al. 2007a). From these seminal works, many genes controlling this pathwayhave been associated with neurulation impairment in humans (Kibar et al. 2007b;Juriloff and Harris 2012).

Bending of the Neural PlateThis second morphogenetic movement accounts for the formation of the neuralgroove and then of the neural tube. It is classic to separate the bending of the neuralplate into two events.

Fig. 8 The deformation of the avian neural plate during shaping showing both transversenarrowing and longitudinal lengthening (Redrawn from Schoenwolf and Sheard 1989)

16 M. Catala

FurrowingThis movement is explained by the modification of the shape of the cells located inthe midline of the neural plate. These cells adopt a wedge shape forming theso-called medial hinge point (Fig. 10). This morphogenetic process has been clas-sically considered to be due to the action of the underlying axial mesoderm (i.e., thenotochord), since it has been demonstrated that node ablation leads to an impairmentof cell wedging (Smith and Schoenwolf 1989). The cells of the median hinge pointsare the precursors of the floor plate that contributes to the ventrodorsal neural tubebias. However, mouse mutants, in which a floor plate does not differentiate, show nodisturbance of neural tube closure (Chiang et al. 1996; Ding et al. 1998; Matiseet al. 1998; Sasaki and Hogan 1994; Ang and Rossant 1994). Sonic hedgehog(SHH), produced by the notochord, induces the formation of median hinge point(Ybot-Gonzales et al. 2002). However, in the absence of this signaling molecule, theneural plate can close by formation of alternative hinge points (Ybot-Gonzaleset al. 2002).

Fig. 9 Longitudinallengthening can be explainedby (a) orientated mitosis (darkgray cell) and (b) intercalationof lateral cells (light graycells) into the midline

Embryology of the Spine and Spinal Cord 17

FoldingDuring this phase, the neural plate uplifts to form the neural groove. Folding is due tothe formation of two lateral hinge points that resemble the medial hinge pointresponsible for the bending of the neural plate (Fig. 11). This movement is manda-tory for the lateral borders of the neural plate to converge on the midline and toeventually fuse (Figs. 12 and 13). This morphogenetic phase is due to extrinsicforces mainly provided by the surface ectoderm (Alvarez and Schoenwolf 1992).Lateral hinge point formation is inhibited by SHH and favored by antagonism ofBMP signaling (Ybot-Gonzales et al. 2007).

Greater Complexity: Distinct Mechanisms Act at the DifferentAnteroposterior Levels of the Spinal Neural TubeIn the mouse, careful analysis of spinal neurulation leads to the conclusion that themode of formation of the neural tube is different according to the anteroposteriorlevel (Shum and Copp 1996). For rostral levels of the spinal neural tube, the neuralplate folds by forming a single medial hinge point, and no lateral hinge points areevidenced. For intermediate levels, the classic model operates. For caudal levels, allcells of the neural plate bend leading to a part of the neural tube where the lumen iscircular. These different morphological aspects suggest that the molecular regula-tions controlling primary neurulation differ along the anteroposterior axis. Such

Fig. 10 Transverse section ofa quail embryo at four-somitestage. Bending of the neuralplate is secondary to theformation of medial hingepoint (MHP). SE surfaceectoderm, No notochord, PAMparaxial mesoderm, Enendoderm

Fig. 11 Transverse section ofa quail embryo at five-somitestage. Folding is due to theformation of lateral hingepoints (LHP). SE surfaceectoderm, No notochord, PAMparaxial mesoderm, MHPmedial hinge point

18 M. Catala

differences have already been evidenced. Sonic hedgehog is highly expressed byrostral segments, whereas it is weaker in caudal segments in mice (Ybot-Gonzaleset al. 2002). Zic2 has a completely reversed pattern, being highly expressed by thecaudal segments (Ybot-Gonzales et al. 2007). The timing and kinetics of the switchbetween E- and N-cadherin vary according to the anteroposterior axis (Dadyet al. 2012). Genetic models of neural tube defects in mice also show that the geneticcontrol of primary neurulation is heterogeneous. Thus, 70–80 % of mutations resultin isolated exencephaly in mice, contrasting with a rare occurrence of isolated spinabifida (5 % of mutations leading to neural tube defects) (Harris and Juriloff 2007,2010).

The So-Called Multisite Closure ModelIn amniotes, initiation of neural tube closure begins in a discrete zone and closureextends from this starting point. However, the level of closure initiation differsconsiderably with species: somites 1–3 in mice (Golden and Chernoff 1993), somites5–7 in pigs (Straaten et al. 2000), and mesencephalon in chicks (Straaten et al. 1996),whereas three concomitant sites are described in rabbits (Peeters et al. 1998). Inhumans, the neural tube begins to close at the level of somites 2–3 (O’Rahilly and

Fig. 12 Transverse section ofa chick embryo at eight-somite stage. The neural tube(NT) is closing on the dorsalmidline (arrow). SE surfaceectoderm, No notochord, Sosomite, En endoderm

Fig. 13 Transverse section ofa chick embryo at ten-somitestage. The neural tube (NT) isnow closed and liesunderneath the surfaceectoderm (SE). En endoderm,Ao aorta, PAM paraxialmesoderm, IM intermediatemesoderm, LM lateralmesoderm

Embryology of the Spine and Spinal Cord 19

M€uller 2002). However, it is important to note that great variability is observedconcerning the initial level of closure in mice depending on the studied strain (Sakai1989).

The classical model in use in numerous textbooks of developmental biology is theso-called bidirectional closure. According to this model, the neural tube closesinitially at a precise point, and thereafter closure proceeds both rostrally and cau-dally. This process ends with the closure of rostral and caudal neuropores. However,this model was challenged as early as 1976 by Waterman (1976), who observedseparate initiation points of closure in the mouse embryo. This leads to the concept ofmultisite closure, the best model to explain neurulation in amniotes. Nevertheless,the situation is far from simple: in mice, variations are observed according to thestrains that are used (Sakai 1989; Juriloff et al. 1991). It is thus not very surprising tonote that great variations in this scheme are also evidenced in other different species:pig (Straaten et al. 2000), chick (Straaten et al. 1996), and rabbit (Peeters et al. 1998).

What is then the situation in human embryos? Two studies were devoted toaddress this problem, but the results were conflicting. Nakatsu et al. (2000) workedon human embryos belonging to Kyoto Collection; they described an initial point ofclosure at the level of the future neck and second point at the mesencephalic-rhombencephalic boundary. They suggested that a limited fusion proceeds fromrostral to caudal at the level of the rostral neuropore; nevertheless, the rostralneuropore was found to close near the rostral end of the telencephalon. Such afeature is also observed in human embryo atlases (Steding 2009). Furthermore,Nakatsu et al. (2000) did not observe a bidirectional closure of the posteriorneuropore, but rather a rostral-to-caudal sequence. A second study was performedby O’Rahilly and M€uller (1989, 2002) in the Carnegie Collection. These authorsdescribed only two sites of initiation of closure: one located at the level of the firstsomites and a second at the extreme tip of the prosencephalon. It is interesting to notethat the caudal neuropore closes according to a unidirectional process as previouslydescribed. The main difference between the two studies is the presence of a closureinitiation site at the level of the boundary between mesencephalon and rhomben-cephalon. These discrepancies should be borne in mind before attempting to inter-pret neural tube defects in humans. This also implies that the concept of multisiteclosure proposed by Van Allen et al. (1993) based on observations of humanmalformations is still largely speculative.

Secondary Neurulation

Descriptive EmbryologyThe closure of the primary neural tube proceeds according to a cephalocaudalgradient, like the lengthening of the embryonic axis. However, since neurulationproceeds more rapidly than axis elongation, the neural tube closes before completeaxis formation. The last part of the neural plate to close is called “posteriorneuropore” (Fig. 14). After the closure of the posterior neuropore, the tissues locatedcaudally to this structure constitute the so-called tail bud (Fig. 15). Posterior

20 M. Catala

Fig. 14 Dorsal view of achick embryo at the time ofclosure of the posteriorneuropore. The posteriorneuropore is located inbetween the last-formedsomite and the primitivestreak

Fig. 15 Dorsal view of thecaudal part of a 25-somitestage chick embryo showingthe paraxial mesoderm (PAM),the neural tube (NT), and thetail bud (TB)

Embryology of the Spine and Spinal Cord 21

neuropore closure occurs at the tip of the tail in tailed animals (Hughes and Freeman1974). On the contrary, this closure occurs more rostrally in tailless animals (Hughesand Freeman 1974). The primary neural tube closes at 16–22-somite stage in chick(Schoenwolf 1979a), 21-somite stage in hamster (Shedden and Wiley 1987),22-somite stage in rabbit (Peeters et al. 1998), 28-somite stage in pig (Straatenet al. 2000), and 25-somite stage in human (M€uller and O’Rahilly 1987). Thesetiming differences must be considered when analyzing the neural tube defectsaffecting the posterior part of the body, the most frequent ones in humans. Thus,the mechanisms controlling posterior neuropore closure in mice are interesting toestablish, but they are not necessarily perfectly applicable to humans.

The development of the tail bud accounts for the formation of the caudal part ofthe spinal cord, a process generally called secondary neurulation (Catala et al. 1995).The tail bud is organized into cells located between the ectoderm and the endoderm.The morphological aspect of neurulation at this level is radically different fromprimary neurulation. Cells originating from the tail bud (Fig. 16) aggregate medio-dorsally to from a cellular cylinder, the medullary cord (Fig. 17). This cylinderprogressively cavitates with formation of multiple lumens (Fig. 18) that are in closecontinuity with the lumen of the neural tube produced by primary neurulation. Then,

Fig. 16 Transverse section ofa chick tail bud showing thetail bud mesenchyme (TB). SEsurface ectoderm, Enendoderm

Fig. 17 Transverse sectionthrough the medullary cord(MC) of a chick embryo. SEsurface ectoderm, PAMparaxial mesoderm, Enendoderm

22 M. Catala

all the lumens coalesce to form a neural tube with a single lumen (Fig. 19). The nodeis no longer recognizable and constitutes the so-called chordoneural hinge, a regionin contact with the caudal neural tube, caudal notochord, and cells of the tail bud.Secondary neurulation is therefore also known as “neurulation by cavitation.”

The above description applies to the chick embryo (Catala 2002; Catalaet al. 1995; Criley 1969; Schoenwolf 1979b; Schoenwolf and Delongo 1980). Thesecondary neural tube during its formation lies in continuity with the primary one asalready evidenced by reconstructions performed by Schumacher as early as 1927(Schumacher 1927) (Figs. 20 and 21). This feature contradicts the postulate putforward by Dias and Partington (2004) in which the secondary neural tube of thechick is thought to arise from isolated regions that coalesce secondarily. Suchproposal of an original discontinuity should be discarded as it does not representreality. This discrepancy shows, if necessary, that it is always better to interprethuman malformations based on original data.

The situation in mammals is much more complex. In both hamsters and mice, thesecondary neural tube always displays a single lumen in continuity with the lumen ofthe primary tube (Shedden and Wiley 1987; Schoenwolf 1984). In humans, thehistological features are very similar to birds with multiple lumens (Saitsuet al. 2004).

Fig. 18 Transverse sectionthrough the cavitating neuraltube (arrows) of a chickembryo. SE surface ectoderm,PAM paraxial mesoderm, Enendoderm, No notochord

Fig. 19 Transverse sectionafter cavitation is achieved ina chick embryo. SE surfaceectoderm, En endoderm, PAMparaxial mesoderm, Nonotochord, Ao aorta, NTneural tube

Embryology of the Spine and Spinal Cord 23

The Two Classic Concepts of the Tail BudThe histology of the tail bud was described as early as 1911 (Kelsey 1911). Twoauthors proposed radically different models to account for the development of thetail bud.

David Holmdahl [1925] and the Concept of a Caudal BlastemaAccording to David Holmdahl (1925), the tissue constituting the tail bud is made ofundifferentiated cells capable of differentiating into the distinct tissues of the caudalpart of the embryo (notochord, neural tube, endoderm, and mesoderm). This authorconsidered that the tail bud is a sort of blastema whose growth is responsible for themorphogenesis of the caudal part of the body.

Jean Pasteels [1937] and the Concept of a Mosaic of TerritoriesJean Pasteels (1937) considered the tail bud to be the consequence of morphogeneticmovements affecting distinct tissues that are pre-patterned in the tail bud. Accordingto him, the tail bud is constituted by a mosaic of territories whose development is yetsegregated.

A Novel Fate Map of the Tail BudThe discrepancy between these two theories was resolved by constructing a fate mapof the avian tail bud (Figs. 22 and 23) using the quail-chick chimera technique(Catala et al. 1995). The salient features of this fate map are:

Fig. 20 The tail bud of chickembryo as evidenced bySchumacher (1927). Thelumen of the primary neuraltube communicates with thelumen of the secondary one

24 M. Catala

1. The chordoneural hinge (Fig. 24), which is secondary to the development of thenode, gives rise to the notochord and floor plate of the secondary neural tube untilthe tip of the embryonic tail.

2. The rostral part of the tail bud (region 2 on Figs. 22 and 25) yields somiticmesoderm and the rest of the neural tube (i.e., the lateral walls and the roof of theneural tube).

3. The caudal tail bud (region 3 on Figs. 22 and 25) differentiates into somiticderivatives but has no neural potentialities.

4. No epidermal or endodermal cells arise from the mesenchyme of the tail bud.

The total contribution of the tail bud is important in the chick embryo (Fig. 26).The tissues that form the caudal part of the body arise from different regions of thetail bud (Fig. 25). These results are thus in agreement with Pasteels’ model,

Fig. 21 Transverse sectionsof chick embryo showing theaspect of secondaryneurulation and thephenomenon of cavitation(From Schumacher 1927)

Embryology of the Spine and Spinal Cord 25

Fig. 22 Dorsal view of thetail bud indicating theterritories that have beentransplanted. Region1 corresponds to the caudalformed neural tube and itsunderlying chordoneuralhinge. Region 2 represents the2/3 rostral part of the tail bud.Region 3 is the rest of thetail bud

Fig. 23 Sagittal section of a chick embryo at 25-somite stage showing the neural tube,chordoneural hinge, and tail bud

26 M. Catala

suggesting that the tail bud does not function as a blastema but is organizedaccording to different territories whose development is segregated.

This organization is reminiscent of that observed in the tail bud of the amphibianembryo in which secondary neurulation does not take place. The tail bud in Xenopusis organized into discrete regions whose fate is different, revealing a mosaic ofterritories (Gont et al. 1993; Tucker and Slack 1995a, b; Beck and Slack 1998).Furthermore, this heterogeneity is also revealed by gene expression in the amphibiantail bud (Gont et al. 1993; Beck and Slack 1998). Such a genetic heterogeneity is alsoobserved in the mouse tail bud (Gofflot et al. 1997); we have also observed asegregation between the prospective neural territory marked by Sox2 and mesoder-mal one expressing chTbx6L (unpublished personal observations).

Finally, we have recently tested the determination of our regions described in thechick tail bud. If region 2 from a quail embryo is placed at the level of region 3 in achick host, it gives rise to both neural tube and mesodermal derivatives (Fig. 27).The reverse experiment, in which a region 3 from a quail embryo is replaced at thelevel of region 2 in a chick host, shows that the graft is only capable to yieldmesodermal cells (Fig. 28). This unpublished experiment shows that the fate ofthis region is determined and does not change with its new environment.

All the published experiments lead to the conclusion that the tail bud in verte-brates is organized into different territories as postulated by Pasteels (Handrigan2003). Our unpublished results strengthen this conclusion. It is also important toreinforce the problem of pluripotentiality of the tail bud. The adjective“pluripotential” must be applied to a single cell to describe its derivative potentials.

Fig. 24 Contribution of thechordoneural hinge to theformation of the caudal part ofthe embryo. Quail cells arerecognized by condensedheterochromatin. The quailgraft yields the notochord(gray arrow) and the floorplate (the region in betweenthe two black arrows)

Embryology of the Spine and Spinal Cord 27

A cell is called pluripotent if it is capable to generate cells coming from the threegerm layers. In developmental biology, “pluripotent” cannot be applied to a tissuemade of different cells. To the best of my knowledge, no study addresses thequestion of pluripotency of the cells of the tail bud. All the already published papersconcern the fate of regions of the tail bud. So, strictly speaking, one should avoidusing the term pluripotent to describe the tail bud in vertebrates.

Fig. 25 Contribution of thetail bud in avian embryo. The2/3 rostral part of the tail bud(light gray) gives rise to thecaudal spinal cord (except forits floor plate) and participatein the formation of the caudalvertebrae. The rest of the tailbud (squared) contributeswith the previous region to theformation of the caudalvertebrae

Fig. 26 Contribution of thetail bud in the chick embryo.The cells of the tail bud havebeen labeled by a defectiveretrovirus leading to theincorporation of the lacZ genein the cells. Afterdevelopment, the grafted cellsare revealed by their bluecolor (here in black). Thearrow points the tip of theembryonic tail. HL hindlimb

28 M. Catala

Continuity Between Primary and Secondary NeurulationAfter revealing the heterogeneity of the avian tail bud and the homology between themorphogenesis of this region and the movements of gastrulation, the relationshipsbetween primary and secondary neurulation should be explored. In order to address

Fig. 27 Contribution of agraft of a part of a quail rostraltail bud transplanted into thecaudal region of the tail bud ina chick host. The quail tissuesgive rise to mesodermalderivatives (gray arrows) andparticipate into the formationof the neural tube (blackarrow). Not: notochord. Notethat the developmentalpotentials of the graft are notmodified by this newenvironment

Fig. 28 Contribution of agraft of a part of a quail caudaltail bud transplanted into therostral region of the tail bud ina chick host. The quail tissuesgive rise to mesodermalderivatives (black arrows) anddo not participate into theformation of the neural tube.Note that the developmentalpotentials of the graft are notmodified by this newenvironment

Embryology of the Spine and Spinal Cord 29

this question, we constructed a fate map of the avian embryo at six-somite stage(Catala et al. 1996). This led to a series of important results.

First, grafting the node allowed us to show that this region regresses until the tipof the tail, generating both notochord and floor plate along its wake (Catalaet al. 1996). The dorsal endoderm that is produced during this morphogeneticmovement is explained by the presence of an endodermal layer beneath the nodeitself. It is important to note that the node gives rise to these derivatives underlyingboth primary and secondary neural tubes. This means that the formation of noto-chord and floor plate is continuous from six-somite stage onward, evidencing that nodiscontinuity between primary and secondary body formation could be established.This continuity is observed in amphibian development (Gont et al. 1993; Tucker andSlack 1995a) but also in mouse development (Wilson and Beddington 1996; Cam-bray and Wilson 2002). All these results show that this mode of formation is highlyconserved in vertebrates and thus is also likely to apply to humans.

An impairment of caudal regression of the node would lead to a neural tubedevoid of both notochord and floor plate. Such a pattern would lead to massiveapoptosis of the caudal region, generating a caudal truncation of the embryo(Charrier et al. 1999). This could explain some of the so-called caudal dysgenesis(“caudal regression” in humans (Catala 2002)).

The second important result was gained after grafting the caudal region of theneural plate. This region generates the secondary neural tube (Catala et al. 1996;Catala 2002). This proves that the tissues that eventually form the secondary neuraltube are located in the superficial layer of the embryo at the very beginning of theirformation. This important result for interpretation of human malformations has beenconfirmed recently (Shimokita and Takahashi 2011). These fate maps confirm thatprogenitors of secondary neural tube are superficially located early during develop-ment (Fig. 29). In conclusion, defect in their internalization would lead to an openneural tube defect contrarily to what is generally thought.

The Junction Between Primary and Secondary NeurulationIn humans, some cases of neural tube defects present very peculiar features. In twocases (Van Allen et al. 1993; Dady et al. 2014), the neural tissue appears as an openlesion at its rostral extremity, whereas it is closed at its caudal moiety. In other cases,the neural tube is open dorsally and closed ventrally (Saitsu et al. 2007). We thinkthat these features could be explained by peculiarities at the junction betweenprimary and secondary neurulation. In mice (Schoenwolf 1984) and hamsters(Shedden and Wiley 1987), these two regions are continuous, without a junctionbetween them. In contrast, in chick, primary and secondary neural tubes overlapalong the dorsoventral axis (Criley 1969; Schoenwolf 1979b; Schoenwolf andDelongo 1980): the primary neural tube ends dorsally whereas the secondary onebegins ventrally. Such is also the case for humans (Saitsu et al. 2004). This promptedus to specifically study this region in the chick embryo (Dady et al. 2014). Afterstaining cells of the superficial neural plate, we demonstrated that some of these cellsundergo an epithelial-mesenchymal transition to ingress and secondarily condensate

30 M. Catala

to form the junction between the two types of neurulation. Furthermore, we con-firmed that all the cells of the secondary neural tube come from the superficial neuralplate. Finally, we perturbed selectively this process by preventing the production ofPrickle 1, a protein involved in PCP pathway; this led to a malformation thatrecapitulates the histological aspect of the human lesions already described.

Ventrodorsal Polarity of the Neural Tube

The neural tube is organized according to a ventrodorsal polarity (Fig. 30). Thispolarity is reminiscent of the former mediolateral polarity of the neural plate beforeneurulation. The medial part of the neural plate will become the floor plate, while thelateral parts will fuse to form the roof plate. The walls of the neural tube aresubdivided into the basal (ventral) plate and the alar (dorsal) plate. The cells of thefloor plate are easily recognized by their shape, and they can also be evidenced by thegenes that are expressed, such as the transcription factor FOXA2 (formerly HNF3β)(Fig. 31) and the secreted proteins Sonic hedgehog, chordin, and noggin.

Fig. 29 Dorsal view of a chick embryo at six-somite stage (left panel). If the caudal part of thequail neural plate (white triangle) is grafted into a chick host, it contributes to the formation of thesecondary neural tube (arrows in right panel). This demonstrates that the tissues that are fated togenerate the secondary neural tube are initially superficial

Embryology of the Spine and Spinal Cord 31

The Notochord Plays a Major Role in the Establishment of Floor Plate

Grafting an Extra Notochord or an Extra Floor Plate Leadsto the Formation of a New Floor PlateThe graft of an extra notochord close to the lateral wall of a neural tube in a chickembryo leads to the formation of an additional floor plate facing the graftednotochord (Smith and Schoenwolf 1989; Straaten et al. 1985; Placzek et al. 1990;Yamada et al. 1991). The same results are also observed in case of graft of an extrafloor plate (Hirano et al. 1991).

Ablation of the Notochord Prevents the Formation of the Floor PlateIf the notochord is extirpated by microsurgery in the chick embryo, the neural tubedevelops without a floor plate (Smith and Schoenwolf 1989; Yamada et al. 1991;Straaten and Hekking 1991). The two sets of experiments (induction of an extra floorplate and inhibition of its differentiation by removing the notochord) lead to theconcept of a necessary induction of the floor plate by the notochord. This model hasbeen challenged by Teillet et al. (1998). These authors propose that the adhesionsbetween notochord and floor plate during development are so tight that it is techni-cally impossible to remove the notochord without ablating the floor plate. Themolecules involved in this induction are important to study in order to try solvingthese discrepancies.

Sonic Hedgehog Can Generate an Extra Floor PlateThe induction of an extra floor plate can be reproduced by expressing sonichedgehog ectopically in the neural tube (Echelard et al. 1993) or by placing sonichedgehog secreting cells in contact with the neural tube (Roelink et al. 1994). Micein which the Sonic hedgehog gene is knocked out are normal in the heterozygousstate (Chiang et al. 1996). This result contrasts with the observation of SHHheterozygous mutants in humans suffering an autosomal dominant form ofholoprosencephaly (Roessler et al. 1996). These comparisons indicate once again

Basal plate

Floor plate

Notochord

Neuraltube

DORSAL

VENTRAL

Alar plate

Roof plate

Fig. 30 The neural tube is organized according to a ventrodorsal polarity

32 M. Catala

that one must avoid transferring information obtained from mouse to the humansituation. These homozygous mice do not express Sonic hedgehog; they develop anormal Hensen’s node, and the notochord is differentiating but degenerates second-arily. The floor plate is never observed in the neural tube as well as motor neurons.Finally, the neural tube is highly abnormal with excessive dorsalization (Chianget al. 1996). All these data point to the central role played by Sonic hedgehog incontrolling the ventrodorsal polarization of the neural tube.

In order to better understand this phenomenon, it is important to describe theSonic hedgehog pathway. Sonic hedgehog is produced as a big precursor thatundergoes a proteolytic cleavage generating two secreted proteins. The inductiveactivity is carried by the N-terminal protein that should be modified to be active.These posttranslational modifications are an addition of a cholesterol moiety and apalmitoyl group. These modifications are mandatory for the inductive action of sonichedgehog, explaining the phenotype of such conditions as the Smith-Lemli-Opitzsyndrome, in which cholesterol addition is perturbed. Once secreted, the N-terminalfragment of Sonic hedgehog binds to a membrane receptor, Patched. Patchednormally inhibits another membrane receptor called Smoothened. The binding ofsonic hedgehog leads to an inhibition of Patched, allowing Smoothened to be active.Activation of Smoothened leads to a cascade of molecular events whose eventualconsequence is the modification of the GLI proteins. These proteins are transcriptionfactors belonging to the family of zinc finger proteins. Mammals synthesize threeGLI proteins, namely, GLI1, GLI2, and GLI3. Gli1 is only transcribed in presence ofHedgehog signaling. Gli2 is expressed by ventral tissues of the neural tube. Themajor role of GLI2 is to active transcription; a minor repressor activity is alsodescribed. Gli3 is transcribed in more dorsal domains of the neural tube. In theabsence of hedgehog signaling, GLI3 is cleaved in GLI3R that acts as a repressor oftranscription. When hedgehog signaling is present, GLI3 is activated into GLI3Athat plays an active role on transcription. Several modulators of this pathway modifythis signaling by acting either at the membrane level or at the cytoplasmic one.

Dissecting the Role of Sonic Hedgehog on the Spinal Neural TubeThe subject of ventralization of the neural tube and of its genetic regulation is ahighly active domain in research. Sonic hedgehog acts on different genes; some of

Fig. 31 In situ hybridizationusing a probe detecting Hnf3βgene expression. The gene isexpressed by the floor plate(arrows) of the neural tube(NT) and the endoderm (En).No notochord, SE surfaceectoderm, So somite, IMintermediate mesoderm, LMlateral mesoderm

Embryology of the Spine and Spinal Cord 33

them encode transcription factors, suggesting a fundamental role in the control ofpatterning of the spinal neural tube. Class I genes are repressed by the hedgehogpathway whereas class II are activated (Jacob and Briscoe 2003). Before closure ofthe neural tube, the neuroectoderm displays a dorsal identity with expression of Pax3and Pax7 (Le Dréau and Martí 2012). In the presence of Sonic hedgehog, the neuraltube ventralizes by expressing class I genes and repressing class II genes. After themodifications induced by hedgehog signaling, couples of genes display mutuallyrepressive activity. Within one couple, the class I gene represses the expression of theclass II gene and vice versa. This leads to the formation of boundaries of geneticexpression. The combination of genes defines a domain that is fated to form aspecific type of derivatives. Five different domains are defined by their specificgenetic expression (p3, pMN, p2, p1, and p0) (Fig. 32). For example, the p3 domainis characterized by the coexpression of Nkx6.1 and Nkx2.2 and will contribute to thegeneration of the so-called V3 interneurons (Borowska et al. 2013). The pMNdomain gives rise to all spinal motoneurons. The situation is really complex evenfor the sole spinal cord since variations in the distribution of these interneurons isobserved along the rostrocaudal axis (Francius et al. 2013).

By comparing different mutants in mice, it is possible to try to construct a likelyscenario that accounts for the ventralization of the neural tube. It is obviously beyondthe scope of this chapter to give a complete account for this question. However, I willfocus on a few questions that could be important for medical applications.

1. The effect on ventralization of the spinal neural tube of invalidation of Smooth-ened in mouse (Wijgerde et al. 2002) is more important than that of sonic

Fig. 32 The ventral neural tube can be subdivided into sub-domains: dP1 to 6 in the alar plate; p0to 3 in the basal plate. Each domain is characterized by a combination of gene expression (e.g., thep2 sub-domain expresses Pax6, Irx3, Nkx6.1, and Nkx6.2). Each of these sub-domains willeventually give rise to specific neurons

34 M. Catala

hedgehog (Chiang et al. 1996). This can be explained by the presence of otherhedgehogs, like Indian hedgehog, in the vicinity of the neural tube.

2. After invalidation of Gli2 and Gli3, no hedgehog signaling is observed (Baiet al. 2004). However, a reduced number of motoneurons are observed at thecervical, forelimb, and thoracic regions, but not at more caudal levels (Baiet al. 2004). This results shows that motoneurons can be generated in the absenceof hedgehog signaling in certain regions of the spinal cord.

3. The invalidation of Gli2 leads to a defect of formation of floor plate and V3interneurons, whereas other ventral domains of the spinal cord are correctlyorganized (Ding et al. 1998; Matise et al. 1998). This mutant demonstrates thatthe development of the floor plate needs hedgehog signaling and allows bettercharacterizing the relationships between notochord and floor plate.

4. The size of the spinal cord in Sonic hedgehog mutants is dramatically reduced(Chiang et al. 1996). In the double mutant Shh�/�/Gli3�/�, the size of the spinalcord is restored (Ruiz i Altaba et al. 2003) indicating that the ratio of GLI3R/Aplays a role in this trophic activity. This result could help to explain the so-calledsegmental spinal dysgenesis. If a segment of notochord is deficient and leads to animpairment of Sonic hedgehog secretion, the overlying neural tube will developpoorly, generating a segmental spinal cord dysgenesis.

Dorsalization of the Neural Tube Involves the TGFβ Family of SecretedMoleculesThe dorsal neural tube, as the ventral one, is divided into six domains (dP1 the mostdorsal to dP6 the most ventral). Each of these domains is characterized by acombination of gene expression (Fig. 32). Several members of the TGFβ familyare able to dorsalize the cells of the neural tube in vitro (Liem et al. 1995, 1997).Blocking the BMP pathway leads to a reduction of the most dorsal domains in thespinal cord (Le Dréau andMartí 2013). This effect is mediated by BMP receptor typeI (Stottmann and Klingesmith 2011; Perron and Dodd 2011) through Smad 5 (Hazenet al. 2012). GDF7 seems to play a restricted role on a subregion of dP1 (Leeet al. 1998), suggesting that other factors from this family of secreted moleculesare needed to account for the complete role of dorsalization (Le Dréau and Martí2013). Furthermore, BMP signaling is antagonized by anti-BMPs produced by axialorgans of the embryo (Liem et al. 2000).

The Wnt canonical pathway acts through the nuclear translocation of beta-catenin. In the dorsal neural tube, two Wnts are produced and activate the canonicalpathway, Wnt1 and Wnt3a. The knockout of each of these factors does not induce animpairment of the dorsal neural tube, whereas their doubling affects dorsalization(Ikeya et al. 1997). Wnt3 pathway induces repression of Sonic hedgehog (Robertsonet al. 2004) due to activation of expression of Gli3 (Alvarez-Medina et al. 2008). Thecurrent model (Ulloa and Martí 2010) is that Wnt induces the production of Gli3 inthe most dorsal region of the neural tube, the level of Sonic hedgehog is notsufficient, and Gli3R represses ventralization. In contrast, where Sonic hedgehogis present at high concentration, Gli3 is converted into Gli3A and participates inventralization.

Embryology of the Spine and Spinal Cord 35

In conclusion, the ventrodorsal polarization of the spinal neural tube results fromthe action of antagonistic systems: Shh/Gli, canonical Wnts, and BMPs (Fig. 33).

The Neural Crest and the Development of the Peripheral NervousSystem

The neural crest is a transient population of cells originating from the dorsal neuraltube and characterized by extensive migratory behavior and numerous cell deriva-tives. In the trunk of the embryo, neural crest cells give rise to the whole peripheralnervous system. Neural crest cells have been extensively studied by numerouslaboratories, and it is impossible for me to give here a thorough presentation ofthis cell population. Interested readers can find an extensive analysis of neural crestcells in Le Douarin and Kalcheim (1999).

Descriptive Embryology

After closure of the neural tube, cells located in the dorsal roof of the neural tubeleave the neural tube to form the neural crest. These cells are highly mobile and willmigrate extensively to eventually give rise to numerous derivatives. In birds andmammals, unlike amphibians, neural crest cells arising from the spinal levels of theneural tube leave the tube only after the achievement of neurulation.

Fig. 33 Molecular pathways are involved in the ventrodorsal polarization of the neural tube. Sonichedgehog (SHH) is expressed by both the floor plate and the notochord and is responsible for theventralization of the tube via activation of Gli2 and Gli3A. Canonical Wnts favors the formation ofGli3R that inhibits SHH. The BMPs are secreted by dorsal tissues and are responsible for partialdorsalization of the tube. Furthermore, floor plate and notochord cells secrete anti-BMPs that inhibitthe dorsalization of BMPs

36 M. Catala

Migration of Neural Crest Cells

The Three Streams of Migration of the Trunk Neural CrestThis migration is due to loss of epithelial characteristics by cells that will thenmigrate isolated. These cells follow three streams of migration (Fig. 34): (i) dorsal,between the dermomyotome and the surface ectoderm; (ii) ventrolateral, within therostral moiety of the somite (which is permeable to neural crest cells migrationcontrarily to the caudal somite) (Fig. 35); and (iii) ventromedial, between the neuraltube and the somite (Le Douarin and Teillet 1974; Rickmann et al. 1985; Bronner-Fraser 1986; Teillet et al. 1987; Serbedzija et al. 1989; Erickson et al. 1989;Newgreen et al. 1990). Cells following the dorsal pathway will differentiate intomelanocytes. Cells migrating along the ventrolateral pathway will give riseto Schwann cells, neurons, and satellite cells of the dorsal root ganglia. Lastly,cells migrating along the ventromedial stream will give rise to sympatheticneurons, parasympathetic neurons, Schwann cells, and cells populating themedullo-adrenals.

The Formation of the Roots and the Plexus

The Formation of the Nerve Roots and Dorsal Root Ganglia Is Explained bythe Intrinsic Properties of the SomitesSince only the rostral half of the somite is permeable to neural crest cell migration(Fig. 35), the continuous stream of neural crest cells arising from the neural tube isthen segmented into discontinuous streams. If the succession of rostral and caudalhemi-somites is perturbed (e.g., by grafting only rostral hemi-somites), the formationof dorsal root ganglia is severely impaired. For example, if the somites are onlyformed by rostral halves, neural crest cells migrate in a continuous stream and willfrom a mega-ganglion without overt sign of segmentation (Kalcheim and Teillet1989).

The Plexus Shape Is Imposed by the Lateral MesodermThe nerves at cervical, lumbar, and sacral levels form plexuses whose shape isspecific for each anatomical level. How are these morphological features deter-mined? By grafting extra wings in a chick embryo, it is possible to show that thedevelopment of the plexus depends on the mesenchymal tissues of the wing, whichderive from the somatopleure (a derivative of the lateral mesodermal domain) andnot from neural cells (Narayanan 1964; Lance-Jones 1988).

The Molecular Control of Neural Crest MigrationThe subject of molecular control of neural crest migration has gained a dramaticimportance. The most fascinating results in recent years reveal that migration ofneural crest cells does not involve a unique pattern for the three migration routes butis selective allowing cell sorting very early after neural tube emigration.

Embryology of the Spine and Spinal Cord 37

Melanoblast and Dorsolateral PathwayFOXD3 is a transcription factor required for the emergence of the neural crest (Koset al. 2001) and inhibits melanogenesis (Kos et al. 2001; Thomas and Erickson2009). Its downregulation allows the expression of melanocytic factors such MITF(Thomas and Erickson 2009; Nitzan et al. 2013a). Melanoblasts express EphB2(Harris et al. 2008; Santiago and Erickson 2002), which is permissive for migrationin the dorsolateral path through ephrin-B ligands. Other factors are involved inmigration in this domain: EDNRB2 (Harris et al. 2008) is repressed by FOXD3(Nitzan et al. 2013a) and c-Kit in mammals (Alexeev and Yoon 2006) but not inbirds (Harris et al. 2008). The situation is more complex because a second mode of

Fig. 34 Transverse section through a chick embryo at the 25-somite stage. Neural crest cellsarising from the roof of the neural tube (NT) migrate according to three migratory routes: 1 thedorsal stream between the surface ectoderm (SE) and the dermomyotome (DM), 2 the ventrolateralstream within the dorsal moiety of the somite, and 3 the ventromedial streal between the NT and thesclerotome (Scl). No notochord

Fig. 35 Neural crest cells(arrows) originating from thedorsal part of the neural tubepenetrate the sole rostralhemi-somite

38 M. Catala

melanocyte differentiation exists. Melanocytes can differentiate from Schwann cellsand not directly from the neural crest (Nitzan et al. 2013b). Such differentiationcould account for the occurrence of melanotic schwannomas.

Extracellular Matrix and Intrasomitic PathwayExtracellular matrix plays a major role for defining the pathways of migrationfollowed by neural crest cells, whatever their migratory stream. Fibronectin and itsreceptor, integrins, were the first molecules linked with such a mechanism(Newgreen and Thiéry 1980; Thiéry et al. 1982; Duband et al. 1986). Subsequentresearches have focused on the composition of the extracellular matrix and itsvariations between rostral and caudal somites. For example, F-spondin (Debby-Brafman et al. 1999) and collagen IX (Ring et al. 1996) are exclusively present inthe caudal half of the somite and could be repulsive for neural crest cell migration. Incontrast, thrombospondin is concentrated in the rostral half of the somite and couldpromote neural crest cell migration (Tucker et al. 1999). Fibronectin is present bothin rostral and caudal moieties of the somite, but its activity is antagonized by thepresence of versican in the sole caudal part (Landolt et al. 1995).

The Sequential Role of Neuropilins and Semaphorins 3Semaphorins 3 are secreted molecules present in vertebrate embryos that bind to afamily of membrane receptors belonging to neuropilins. The couple neuropilin 1 andits ligand semaphorin 3A favors the migration of neural crest cells into the somiticderivatives (Schwarz et al. 2009a, b). In the absence of neuropilin 1 expressed,neural crest cells follow the medial stream of migration. The restriction of migrationinto the sole rostral domain of the sclerotome is favored by neuropilin 2 andsemaphorin 3 F (Schwarz et al. 2009b; Gammill et al. 2006).

Morphogenesis of the Most Caudal Part of the Spinal Cord

The Concept of Normal Caudal RegressionThe development of the most caudal part of the spinal cord in humans was studied inthe Carnegie Institution by Kunitomo (1918) and Streeter (1919). Kunitomo (1918)described two processes that could account for the development of the caudal part ofthe spinal cord. Firstly, he stated that the most caudal neural tube has a limitedcapacity of growth. Secondly, by analyzing the number of dorsal root ganglia, heconcluded that some coccygeal ganglia vanish; this process is accompanied with aregression of the already formed caudal spinal cord. Streeter (1919), 1 year afterKunitomo, reinforced this process by speaking of “retrogressive differentiation” tohighlight the observed features. However, one may wonder if these authors hadreliable benchmarks for assessing the anatomical levels of the spinal cord. Indeed, inFig. 1 from Streeter (1919), nine coccygeal somites are represented, whereasKunitomo mentioned only four. Thus, I should encourage readers to be very cautiousconcerning these very old papers.

Embryology of the Spine and Spinal Cord 39

Experimental Approach to Caudal Spinal Cord DevelopmentThe spinal cord is a highly conserved structure of the central nervous system invertebrates. It is possible to recognize three basic patterns of spinal cord. The firsttype is an ancestral one, characterized by a neural cord occupying the whole lengthof the vertebral canal, with motor and sensory nerves arising from each metamere;this type of spinal cord is encountered in reptiles, fishes, and amphibians beforemetamorphosis. The second type is characterized by a spinal cord occupying theentire length of the vertebral canal but presenting a caudal extremity devoid of anymotor and sensory nerves; this type is encountered in birds. Lastly, the third type isthe mammalian type with a spinal cord ending before the end of the vertebral canaland prolonging into the filum terminale (which can be considered as a rudimentaryneural tube devoid of motor and sensory nerves). This anatomical description showsthat the caudal part of the avian neural tube can be considered as a homologue of thecaudal part of the mammalian one.

The developmental potentials of the neural crest cells arising from the most caudallevels of the neural tube of the avian embryo have been studied (Catala et al. 2000). Inthe chick embryo, the total number of somites is 53. Caudally to the 53rd somite, themesoderm does not segment and does not give rise to somites. The neural tube caudalto somite 53 is unable to give rise to neural crest cells, whereas it is capable of doingso between the levels of somites 48 and 53. However, these cells are not able todifferentiate into neurons. Rostrally to somite 47, the neural tube gives rise to neuralcrest cells that can differentiate into Schwann cells, melanocytes, and neurons. If thecaudal neural tube devoid of neuronogenic neural crest cells is transplanted in morerostral position, it cannot produce neurons of the peripheral nervous system. Thisstudy shows that the absence of sensory nerves is explained by an intrinsic loss ofpotentialities of the neural crest cells arising from these levels of the neural tube. Thegenetic control for generation of neural crest cells is the same for both primary andsecondary chick neural tube (Osório et al. 2009a). We tested the molecular cause forthe absence of neuronogenic neural crest cells at the level of somites 48–53 (Osórioet al. 2009b). We found that noggin expression is maintained at this level leading to amassive apoptosis. This could be mimicked at more levels by electroporating nogginin the neural tube (Osório et al. 2009b). Furthermore, inhibition of noggin in the mostcaudal neural tube leads to an increase of Bmp4-Wnt1 signaling and an induction ofproneurogenic markers (Osório et al. 2009b). Thus, the persistence of noggin expres-sion in the caudal neural tube of the chick embryo accounts for the lack of sensoryganglia at these very caudal levels. Is this phenomenon relevant for mammals? Thisquestion remains to be established in mouse.

The absence of motor nerves at these very caudal levels of the chick neural tube isexplained by an impairment of the ventrodorsal patterning of the neural tube (Afonsoand Catala 2005). However, this peculiar feature of the caudal neural tube is reversedafter heterotopic transplantation in more rostral positions (Afonso and Catala 2005).Indeed, when heterotopically transplanted, the neural tube is now capable of gener-ating motoneurons and motor nerves, showing that this is not an intrinsic property ofcaudal neural tube in the chick embryo but is modulated by its environment (Afonsoand Catala 2005). Such a problem has been assessed in the mouse embryo recently

40 M. Catala

(Shum et al. 2010). The molecular patterns that are observed in the mouse are slightlydifferent from that of the chick. However, the results are very similar between thesetwo species. In particular, neither regression of dorsal root ganglia nor retrogressivedifferentiation of the caudal spinal cord is observed. It should therefore be veryinteresting to study such features in other mammals, in particular tailless ones.Furthermore, to extend the important work byNievelstein et al. (1993), other analyseson human embryos should also be performed on well-preserved specimens.

The Development of the Spine

The spine is formed of repetitive anatomical parts called vertebrae that constitutemetameric units. Each vertebra has ventrodorsal characteristics with a ventral body, avertebral arch around the spinal cord, and a dorsal spinous process. In addition, onecan observe an anteroposterior (or rostrocaudal) regionalization that allows todescribe large regions: cervical, thoracic, lumbar, sacral, and coccygeal. The verte-brae derive from metameric units of paraxial mesoderm, namely, the somites, formedduring gastrulation.

Epithelial Somites Dissociate to Form Their Derivatives

During gastrulation, cells located in the primitive streak ingress and migrate laterallyto form the different domains of the mesoderm (Bellairs 1986; Tam et al. 1993). Theparaxial mesodermal domain derives from cells located at the rostral third of theprimitive streak (Psychoyos and Stern 1996; Nicolet 1970; Tam and Beddington1987). These cells form the cephalic mesoderm at the level of the head and thesomitic mesoderm in the trunk. At first, the paraxial trunk mesoderm isunsegmented; as development proceeds, epithelial spheres, called somites, areformed in a cephalocaudal gradient (Christ and Ordahl 1995). An epithelial somiteis constituted by an epithelial wall surrounding a mesenchymal core (Fig. 36a).These two regions are not true compartments. Cells of the epithelial wall are able toremain in the epithelium or to add to the mesenchyme of the core (Wong et al. 1993).In contrast, cells of the core remain in this location (Wong et al. 1993). The epithelialsomite matures during development, and this maturation proceeds according to acephalocaudal gradient, leading to a dissociation of the epithelial somite that even-tually forms the dermatome (dorsal), the myotome (intermediate), and thesclerotome (ventral) (Fig. 36b). The dermatome is located underneath the surfaceectoderm. It will give rise to dermal cells for the dorsal moiety of the body (theventral dermal cells originate from the somatopleure). The myotome generates allthe striated muscle fibers of the trunk. The sclerotomal cells differentiate intocartilaginous cells of the vertebrae, cells of the intervertebral disks and ligaments,and cells of the spinal meninges. Furthermore, the somite gives rise to endothelialcells. The sclerotome forms after an epithelial-mesenchymal transition affecting theventralmost part of the somite. It is important to note that the sclerotome is first

Embryology of the Spine and Spinal Cord 41

Fig. 36 Temporal evolution of the somite during dissociation. (a) Before dissociation, the somite iscomposed of both a wall and a mesenchymal core. (b) After dissociation, the somite gives rise to thesclerotome, the myotome, and the dermatome

42 M. Catala

located ventrally, and then it spreads to enwrap the entire neural tube forming at itsdorsal face the so-called dorsal mesoderm. This dorsal mesoderm is a late-appearingstructure because the neural tube and the surface ectoderm are tightly apposed justafter neurulation, the interepithelial space being permeable late after the achievementof neurulation (Martins-Green 1988).

The Somite Is Not Polarized According to the Ventrodorsal Axis

The ventral part of the somite differentiates into the sclerotome, whereas its dorsalpart will give rise to dermatome and myotome. This difference between the fates ofthe hemi-somite is not fixed when the somite segments. Indeed, if one rotates thelast-formed somite according to the ventrodorsal axis, the new ventral hemi-somite(which was initially dorsal) gives rise to sclerotome (Aoyama and Asamoto 1988).The new dorsal hemi-somite (formerly ventral) differentiates into dermatome andmyotome (Aoyama and Asamoto 1988). This shows that the fate of the hemi-somitesis not fixed and can be changed by the environment. This plasticity is not yet possibleif the rotation of the somite is made three hours after its generation (Brand-Saberiet al. 1993), demonstrating that cellular interaction responsible for sclerotomeformation is a transient process rapidly occurring after somitic epithelialization.

Cellular and Molecular Players for Somitic Polarization

The ventralization of the somite is promoted by the notochord and floor plate(Pourquié et al. 1993; Fan and Tessier-Lavigne 1994), as is the case for neuraltube ventralization. The dorsalization of the somite is promoted by surface ectodermand by dorsal neural tube (Fan and Tessier-Lavigne 1994; Spence et al. 1996).

Sonic hedgehog, secreted by the notochord and floor plate, is able to induce theformation of the sclerotome (Fan and Tessier-Lavigne 1994; Johnson et al. 1994).The Shh�/� mouse does not develop any vertebral structures, since the sclerotomesfail to form (Chiang et al. 1996). The secreted molecules Wnts induce the dorsaltissues of the somite (i.e., the dermatome and the myotome) (Ikeya and Takada1998). In the segmented somite, Sonic hedgehog (produced by both notochord andfloor plate) acts to induce the expression of Gli1 by ventral cells, whereas Wnts(produced by both surface ectoderm and dorsal neural tube) act to induce theexpression of Gli2 and Gli3 by dorsal cells (Borycki et al. 2000).

The Neural Tube and the Notochord Induce the Formationof the Vertebral Cartilage

It is well known from classic studies of tissue recombination that the neural tube andthe notochord can induce (together or isolated) the formation of cartilaginous cellsfrom the somite (Holtzer and Detwiler 1953; Lash et al. 1957). It is important to note

Embryology of the Spine and Spinal Cord 43

that notochord and neural tube behave differently according to cartilage induction.The induced cartilage is contiguous with the notochord, whereas it is always at somedistance from the neural tissue (Lash et al. 1957). It can be postulated that the neuraltube induces meningeal tissue in close contact and cartilage at distance.

One striking result about vertebral formation is that three domains can berecognized within the vertebrae (Ding et al. 1998; Monsoro-Burq et al. 1994; Moet al. 1997; Watanabe et al. 1998; Monsoro-Burq and Le Douarin 2000). The ventraldomain is responsible for the formation of the vertebral body and is controlled byGli2 (Ding et al. 1998; Mo et al. 1997). Within this domain, mutations of Bapx1 inthe mouse lead to a defect in chondrogenesis of the vertebral body, whereas the restof the vertebra is normal (Tribioli and Lufkin 1999). Such is also the case for thedouble knockout Pax1/Pax9 (Peters et al. 1999), a result that seems very logicalsince these two genes cooperate to induce Bapx1 expression in the ventralsclerotome. As these three genes are expressed by mesodermal derivatives, thesedefects can be attributed to an impairment of the formation of the vertebral primor-dium. The Brachyury gene is important for the maintenance of notochordal cells(Pennimpede et al. 2012). So, the ventral vertebral defects observed in humanmutants of Brachyury (Postma et al. 2014) could be better explained by a defectiveinductive signaling on sclerotomal cells than by a putative persistence of thenotochordal canal.

The neural arch of the vertebra forms the second domain and is under the controlof Gli3 (Mo et al. 1997). Lastly, the dorsal domain (which derives from the so-calleddorsal mesoderm) differentiates into the spinous process. This dorsal domaindepends on the dorsal neural tube, on surface ectoderm, and on BMP4 (Monsoro-Burq and Le Douarin 2000). These different sub-domains in the development of thevertebrae may explain some human spine malformations affecting selectively thedorsal domain and sparing the ventrolateral one, such as the case of spina bifida withlipoma (Catala 1997).

Conclusion

The development of the spinal cord and vertebrae is intimately connected. Themechanisms that govern their formation are beginning to be identified by the useof animal models. These results give major data and could explain some aspects ofhuman malformations. Nevertheless, it appears that the control processes vary alongrostrocaudal axis, and it is desirable to pay attention to the excessive generalizationof data when they are actually established only for a given anatomical region.Likewise, experimental models do not always summarize what happens in humans.It is therefore important to multiply these models and still be cautious beforeapplying the results to humans. However, developmental biology is a fascinatingdiscipline that obtains very useful results for physicians.

44 M. Catala

References

Afonso ND, Catala M. Sonic hedgehog and retinoic acid are not sufficient to induce motoneurongeneration in the avian caudal neural tube. Dev Biol. 2005;279:356–67.

Alexeev V, Yoon K. Distinctive role of the cKit receptor tyrosine kinase signaling in mammalianmelanocytes. J Invest Dermatol. 2006;126:1102–10.

Alvarez IS, Schoenwolf GC. Expansion of the surface epithelium provides the major extrinsic forcefor bending of the neural plate. J Exp Zool. 1992;261:340–8.

Alvarez-Medina R, Cayuso J, Okubo T, Takada S, Martí E. Wnt canonical pathway restricts gradedShh/Gli patterning activity through the regulation of Gli3 expression. Development.2008;135:237–47.

Ang SL, Rossant J. HNF-3beta is essential for node and notochord formation in mouse develop-ment. Cell. 1994;78:561–74.

Aoyama H, Asamoto K. Determination of somite cells: independence of cell differentiation andmorphogenesis. Development. 1988;104:15–28.

Bachvarova RF, Skromme I, Stern CD. Induction of primitive streak and Hensen’s node by theposterior marginal zone in the early chick embryo. Development. 1998;125:3521–34.

Bai CB, Stephen D, Joyner AL. All mouse ventral spinal cord patterning by hedgehog is Glidependent and involves an activator function of Gli3. Dev Cell. 2004;6:103–15.

Beck CW, Slack JMW. Analysis of the developing Xenopus tail bud reveals separate phases of geneexpression during determination and outgrowth. Mech Dev. 1998;72:41–52.

Bellairs R. The primitive streak. Anat Embryol. 1986;174:1–14.Bertocchini F, Stern CD. The hypoblast of the chick embryo positions the primitive streak by

antagonizing nodal signalling. Dev Cell. 2002;3:735–44.Blumberg B, Wright CVE, De Robertis EM, Cho KWY. Organizer-specific homeobox genes in

Xenopus laevis embryos. Science. 1991;253:194–6.Borowska J, Jones CT, Zhang H, Blacklaws J, Goulding M, Zhang Y. Functional subpopulations of

V3 interneurons in the mature mouse spinal cord. J Neurosci. 2013;33:18553–65.Borycki A-G, Brown AMC, Emerson CP. Shh and Wnt signalling pathways converge to control Gli

gene activation in avian somites. Development. 2000;127:2075–87.Brand-Saberi B, Ebensperger C, Wilting J, Balling R, Christ B. The ventralizing effect of the

notochord on somite differentiation in the chick embryos. Anat Embryol. 1993;188:239–45.Bronner-Fraser M. Analysis of the early stages of trunk neural crest migration in avian embryos

using monoclonal antibody HNK-1. Dev Biol. 1986;115:44–55.Cambray N, Wilson V. Axial progenitors with extensive potency are localised to the mouse

chordoneural hinge. Development. 2002;129:4855–66.Carver EA, Jiang R, Lan Y, Oram KF, Gridley T. The mouse Snail gene encodes a key regulator of

the epithelia-mesenchymal transition. Mol Cell Biol. 2001;21:8184–8.Catala M. Embryogenesis. Why do we need a new explanation for the emergence of spina bifida

with lipoma? Childs Nerv Syst. 1997;13:33–6340.Catala M. Genetic control of caudal development. Clin Genet. 2002;61:89–96.Catala M, Teillet M-A, Le Douarin NM. Organization and development of the tail bud analysed

with the quail-chick chimaera system. Mech Dev. 1995;51:51–65.Catala M, Teillet M-A, De Robertis EM, Le Douarin NM. A spinal cord fate map in the avian

embryo: while regressing, Hensen’s node lays down the notochord and floor plate thus joiningthe spinal cord lateral walls. Development. 1996;122:2599–610.

Catala M, Ziller C, Lapointe F, Le Douarin NM. The developmental potentials of the caudalmostpart of the neural crest are restricted to melanocytes and glia. Mech Dev. 2000;95:77–87.

Charrier J-B, Teillet M-A, Lapointe F, Le Douarin NM. Defining subregions of Hensen’s nodeessential for caudalward movement, midline development and cell survival. Development.1999;126:4771–83.

ChiangC, LitingtungY, LeeE,YoungKE, Corden JL,Westphal H,BeachyPA. Cyclopia and defectiveaxial patterning in mice lacking Sonic hedgehog gene function. Nature. 1996;383:407–13.

Embryology of the Spine and Spinal Cord 45

Cho KW, Blumberg B, Steinbeisser H, De Robertis EM. Molecular nature of Spemann’s organizer:the role of the Xenopus gene goosecoid. Cell. 1991;67:111–1120.

Christ B, Ordahl CP. Early stages of chick somite development. Anat Embryol. 1995;191:381–96.Colas J-F, Schoenwolf GC. Towards a cellular and molecular understanding of neurulation. Dev

Dyn. 2003;221:117–45.Conlon FL, Lyons KM, Takaesu N, Barth KS, Kispert A, Herrmann B, Roberston EJ. A primary

requirement for nodal in the formation and maintenance of the primitive streak in the mouse.Development. 1994;120:1919–28.

Criley BB. Analysis of the embryonic sources and mechanisms of development of posterior levelsof chick neural tubes. J Morphol. 1969;128:465–502.

Dady A, Blavet C, Duband J-L. Timing and kinetics of E- to N-cadherin switch during neurulationin the avian embryo. Dev Dyn. 2012;241:1333–49.

Dady A, Havis E, Escriou V, Catala M, Duband J-L. Junctional neurulation: a unique developmen-tal program shaping a discrete region of the spinal cord highly susceptible to neural tube defects.J Neurosci. 2014;34:13208–21.

Debby-Brafman A, Burstyn-Cohen T, Klar A, Kalcheim C. F-spondin, expressed in somite regionsavoided by neural crest cells, mediates inhibition of distinct somite domains to neural crestmigration. Neuron. 1999;22:475–88.

Dias MS, Partington M. Embryology of myelomeningocele and anencephaly. Neurosurg Focus.2004;16(2):E1.

Ding Q, Motoyama J, Gasca S, Mo R, Sasaki H, Rossant J. Diminished Sonic hedgehog signalingand lack of floor plate differentiation in Gli2 mutant mice. Development. 1998;125:2533–43.

Doniach T, Phillips CR, Gerhart JC. Planar induction of anteroposterior pattern in the developingcentral nervous system of Xenopus laevis. Science. 1992;257:542–5.

Duband JL, Rocher S, Chen WT, Yamada KM, Thiéry JP. Cell adhesion and migration in the earlyvertebrate embryo: location and possible role of the putative fibronectin receptor complex. J CellBiol. 1986;102:160–78.

Echelard Y, Epstein DJ, St-Jacques B, Shen L, Mohler J, Mc Mahon JA, Mc Mahon AP. Sonichedgehog, a member of a family of putative signaling molecules, is implicated in the regulationof CNS polarity. Cell. 1993;75:1417–30.

Erickson CA, Loring JF, Lester SM. Migratory pathways of HNK-1-immunoreactive neural crestcells in the rat embryo. Dev Biol. 1989;134:112–8.

Fan C-M, Tessier-Lavigne M. Patterning of mammalian somites by surface ectoderm and noto-chord: evidence for sclerotome induction by a hedgehog homolog. Cell. 1994;79:1175–86.

Francius C, Harris A, Rucchin V, Hendricks TJ, Stam FJ, Barber M, Kurek D, Grosveld FG,Pierani A, Goulding M, Clotman F. Identification of multiple subsets of ventral interneurons anddifferential distribution along the rostrocaudal axis of the developing spinal cord. PLoS One.2013;8(8):e70325. doi:10.1371/journal.pone.0070325. eCollection 2013.

Gammill LS, Gonzalez C, Gu C, Bronner-Fraser M. Guidance of trunk neural crest migrationrequires neuropilin2/semaphoring 3F signaling. Development. 2006;133:99–106.

Gofflot F, Hall M, Morriss-Kay G. Genetic patterning of the developing mouse tail at the time ofposterior neuropore closure. Dev Dyn. 1997;210:431–45.

Golden JA, Chernoff GF. Intermittent pattern of neural tube closure in two strains of mice.Teratology. 1993;47:73–80.

Gont LK, Steinbeisser H, Blumberg B, De Robertis EM. Tail formation as a continuation ofgastrulation: the multiple cell populations of the Xenopus tailbud derive from the late blastoporelip. Development. 1993;119:991–1004.

Halacheva V, Fuchs M, Dönitz J, Reupke T, P€uschel B, Viebahn C. Planar cell movements andoriented cell division during early primitive streak formation in the mammalian embryo. DevDyn. 2011;240:1905–16.

Haldiman JT, Gier HT. Bovine notochord origin and development. Anat Histol Embryol.1981;10:1–14.

46 M. Catala

Handrigan GR. Concordia discors: duality in the origin of the vertebrate tail. J Anat.2003;202:255–67.

Harris MJ, Juriloff DM. Mouse mutants with neural tube closure defects and their role inunderstanding human neural tube defects. Birth Defects Res A Clin Mol Teratol.2007;79:187–210.

Harris MJ, Juriloff DM. An update to the list of mouse mutants with neural tube closure defects andadvances toward a complete genetic perspective of neural tube closure. Birth Defects Res A ClinMol Teratol. 2010;88:653–69.

Harris ML, Hall R, Erickson CA. Directing pathfinding along the dorsolateral path – the role ofEDNRB2 and EphB2 on overcoming inhibition. Development. 2008;135:4113–22.

Hazen VM, Andrews MG, Umans L, Crenshaw EB, Zwijsen A, Butler SJ. BMP receptor-activatedSmads confer diverse functions during the development of the dorsal spinal cord. Dev Biol.2012;367:216–27.

Hemmati-Brivanlou A, Melton DA. Inhibition of activin receptor signaling promotes neuralizationin Xenopus. Cell. 1994;77:278–81.

Hemmati-Brivanlou A, Kelly OG, Melton DA. Follistatin, an antagonist of activin, is expressed inthe Spemann organizer and displays direct neutralizing activity. Cell. 1994;77:283–95.

Hirano S, Fuse S, Sohal GS. The effect of the floor plate on pattern and polarity in the developingcentral nervous system. Science. 1991;251:310–3.

Holmdahl DE. Experimentelle Untersuchungen €uber die lage der Grenze zwischen primärer undsekundärer Körperentwicklung beim Huhn. Anat Anz. 1925;59:393–6.

Holtzer H, Detwiler SR. An experimental analysis of the development of the spinal column. III.Induction of skeletogenous cells. J Exp Zool. 1953;123:335–69.

Huelsken J, Vogel R, Brinkmann V, Erdmann B, Birchmeier C, Birchmeier W. Requirement forbeta-catenin in anterior-posterior axis formation in mice. J Cell Biol. 2000;148:567–78.

Hughes AF, Freeman RB. Comparative remarks on the development of the tail among highervertebrates. J Embryol Exp Morphol. 1974;32:355–63.

Idkowiak J, Weisheit G, Plitzner J, Viebhan C. Hypoblast controls mesoderm generation and axialpatterning in the gastrulating rabbit embryo. Dev Genes Evol. 2004;214:591–605.

Ikeya M, Takada S. Wnt signaling from the dorsal neural tube is required for the formation of themedial dermomyotome. Development. 1998;125:4969–76.

Ikeya M, Lee SM, Johnson JE, McMahon AP, Takada S. Wnt signalling required for expansion ofneural crest and CNS progenitors. Nature. 1997;389:966–70.

Jacob J, Briscoe J. Gli proteins and the control of spinal-cord patterning. EMBO Rep.2003;4:761–5.

Johnson RL, Laufer E, Riddle RD, Tabin C. Ectopic expression of Sonic hedgehog alters dorsal–ventral patterning of somites. Cell. 1994;79:1165–73.

Jurand A. The development of the notochord in chick embryos. J Embryol Exp Morphol.1962;10:602–21.

Jurand A. Some aspects of the development of the notochord in mouse embryos. J Embryol ExpMorphol. 1974;32:1–33.

Juriloff DM, Harris MJ. A consideration of the evidence that genetic defects in planar cell polaritycontribute to the etiology of human neural tube defects. Birth Defects Res A Clin Mol Teratol.2012;94:824–40.

Juriloff DM, Harris MJ, Tom C, MacDonald KB. Normal mouse strains differ in the site of initiationof closure of the cranial neural tube. Teratology. 1991;44:225–33.

Kalcheim C, Teillet M-A. Consequences of somite manipulation on the pattern of dorsal rootganglion development. Development. 1989;106:85–93.

Kelsey H. Subdivision of the spinal canal in the lumbar region of chick embryos. Proc Roy SocVictoria. 1911;24:152–5.

Kibar Z, Torban E, McDearmid JR, Reynolds A, Berghout J, Mathieu M, Kirillova I, De Marco P,Merello E, Hayes JM, Wallingford JB, Drapeau P, Capra V, Gros P. Mutations in VANGL1associated with neural-tube defects. N Engl J Med. 2007a;356:1432–7.

Embryology of the Spine and Spinal Cord 47

Kibar Z, Capra V, Gros P. Toward understanding the genetic basis of neural tube defects. ClinGenet. 2007b;71:295–310.

Kimelman D. Mesoderm induction: from caps to chips. Nat Rev Genet. 2006;7:360–72.Kinder SJ, Tsang TE, Wakamiya M, Sasaki H, Behringer RR, Nagy A, Tam PP. The organizer of the

mouse gastrula is composed of a dynamic population of progenitor cells for the axial mesoderm.Development. 2001;128:3623–34.

Kintner CR, Melton DA. Expression of Xenopus N-CAM RNA in ectoderm is an early response toneural induction. Development. 1987;99:311–25.

Kirillova I, Novikova I, Augé J, Audollent S, Esnault D, Encha-Razavi F, Lazjuk G, Attié-Bitach T,Vekemans M. Expression of the Sonic Hedgehog gene in human embryos with neural tubedefects. Teratology. 2000;61:347–54.

Kos R, Reedy MV, Johnson RL, Erickson CA. The winged-helix transcription factor FoxD3 isimportant for establishing the neural crest lineage and repressing melanogenesis in avianembryos. Development. 2001;128:1467–79.

Kowalevsky A. Weitere Studien €uber die Entwickelungsgeschichte des Amphioxus lanceolatus,nebst einem Beitrage zur Homologie des Nervensystems der W€urmer und Wirbelthiere. ArchMikrok Anat. 1877;13:181–204.

Kunitomo K. The development and reduction of the tail and of the caudal end of the spinal cord.Contrib Embryol Carnegie Inst. 1918;8:161–98.

Lance-Jones C. The effect of somite manipulation on the development of motoneuron projectionpatterns in the embryonic chick hindlimb. Dev Biol. 1988;126:408–19.

Landolt RM, Vaughan L, Winterhalter KH, Zimmermann DR. Versican is selectively expressed inembryonic tissues that act as barriers to neural crest cell migration and axon outgrowth.Development. 1995;121:2303–12.

Lash J, Holtzer S, Holtzer H. An experimental analysis of the development of the spinal column.VI. Aspects of cartilage induction. Exp Cell Res. 1957;13:292–303.

Lawson KA, Dunn NR, Roelen BAJ, Zeinstra LM, Davis AM, Wright CVE, Korving JPWFM,Hogan BLM. Bmp4 is required for the generation of primordial germ cells in the mouse embryo.Genes Dev. 1999;13:424–36.

Le Douarin NM, Kalcheim C. The neural crest. 2nd ed. Cambridge: Cambridge University Press;1999.

Le Douarin NM, Teillet M-A. Experimental analysis of the migration and differentiation ofneuroblasts of the autonomic nervous system and of neuroectodermal mesenchymal derivatives,using a biological cell marking technique. Dev Biol. 1974;41:163–84.

Le Dréau G, Martí E. Dorso-ventral patterning of the neural tube: a tale of three signals. DevNeurobiol. 2012;72:1471–8.

Le Dréau G, Martí E. The multiple activities of BMPs during spinal cord development. Cell MolLife Sci. 2013;70:4293–305.

Lee KJ, Mendelsohn M, Jessell TM. Neuronal patterning by BMPs: a requirement for GDF7 in thegeneration of a discrete class of commissural interneurons in the mouse spinal cord. Genes Dev.1998;12:3394–407.

Lemaire P, Garrett N, Gurdon JB. Expression cloning of Siamois, a Xenopus homeobox geneexpressed in dorsal-vegetal cells of blastulae and able to induce a complete secondary axis. Cell.1995;81:85–94.

Liem KF, Tremml G, Roelink H, Jessell TM. Dorsal differentiation of neural plate cells induced byBMP-mediated signals from epidermal ectoderm. Cell. 1995;82:969–79.

Liem KF, Tremml G, Jessell TM. A role for the roof plate and its resident TGFbeta-related proteinsin neuronal patterning in the dorsal spinal cord. Cell. 1997;91:127–38.

Liem KF, Jessell TM, Briscoe J. Regulation of the neural patterning activity of sonic hedgehog bysecreted BMP inhibitors expressed by notochord and somites. Development. 2000;127:4855–66.

Linker C, Stern CD. Neural induction requires BMP inhibition only as a late step, and involvessignals other than FGF and Wnt antagonists. Development. 2004;131:5671–81.

48 M. Catala

Martins-Green M. Origin of the dorsal surface of the neural tube by progressive delamination ofepidermal ectoderm and neuroepithelium: implications for neurulation and neural tube defects.Development. 1988;103:687–706.

Matise MP, Epstein DJ, Park HL, Platt KA, Joyner AL. Gli2 is required for induction of floor plateand adjacent cells, but not most ventral neurons in the mouse central nervous system. Devel-opment. 1998;125:2759–70.

Mo R, Freer AM, Zinyk DL, Crackower MA, Michaud J, Heng HHQ, Chi KW, Shi XM, Tsui LC,Cheng SH, Joyner AL, Hui CC. Specific and redundant functions of Gli2 and Gli3 zinc fingergenes in skeletal patterning and development. Development. 1997;124:113–23.

Monsoro-Burq A-H, Le Douarin NM. Duality of molecular signaling involved in vertebralchondrogenesis. Curr Top Dev Biol. 2000;48:43–75.

Monsoro-Burq A-H, Bontoux M, Teillet M-A, Le Douarin NM. Heterogeneity in the developmentof the vertebra. Proc Natl Acad Sci U S A. 1994;91:10435–9.

Moury JD, Schoenwolf GC. Cooperative model of epithelial shaping and bending during avianneurulation: autonomousmovements of the neural plate, autonomousmovements of the epidermis,and interactions in the neural plate/epidermis transition zone. Dev Dyn. 1995;204:323–37.

M€uller F, O’Rahilly R. The development of the human brain, the closure of the caudal neuropore,and the beginning of secondary neurulation at stage 12. Anat Embryol. 1987;176:413–30.

Nakatsu T, Uwabe C, Shiota K. Neural tube closure in humans initiates at multiple sites: evidencefrom human embryos and implications for the pathogenesis of neural tube defects. AnatEmbryol. 2000;201:455–66.

Narayanan CH. An experimental analysis of peripheral nerve pattern development in the chick. JExp Zool. 1964;156:49–60.

Newgreen D, Thiéry JP. Fibronectin in early avian embryos: synthesis and distribution along themigration pathways of neural crest cells. Cell Tissue Res. 1980;211:269–91.

Newgreen DF, Powell ME, Moser B. Spatiotemporal changes in HNK-1/L2 glycoconjugates onavian embryo somite and neural crest cells. Dev Biol. 1990;139:100–20.

Nicolet G. Analyse autoradiographique de la localisation des différentes ébauches présomptivesdans la ligne primitive de l’embryon de poulet. J Embryol Exp Morphol. 1970;23:70–108.

Nieuwkoop PD. The formation of the mesoderm in urodelean amphibians. I. The induction by theendoderm. Wilhelm Roux Arch Entw Mech Org. 1969a;162:341–71.

Nieuwkoop PD. The formation of the mesoderm in urodelean amphibians. II. The origin of thedorso-ventral polarity of the mesoderm. Wilhelm Roux Arch Entw Mech Org.1969b;163:298–315.

Nievelstein RAJ, Hartwig NG, Vermeij-Keers C, Valk J. Embryonic development of the mamma-lian caudal neural tube. Teratology. 1993;48:21–31.

Nitzan E, Krispin S, Pfaltzgraff ER, Klar A, Labosky PA, Kalcheim C. A dynamic code of dorsalneural tube regulates the segregation between neurogenic and melanogenic neural crest cells.Development. 2013a;140:2269–79.

Nitzan E, Pfaltzgraff ER, Labosky PA, Kalcheim C. Neural crest and Schwann cell progenitor-derived melanocytes are two spatially segregated populations similarly regulated by Foxd3.Proc Natl Acad Sci U S A. 2013b;110:12709–14.

O’Rahilly R, M€uller F. The first appearance of the human nervous system at stage 8. Anat Embryol.1981;163:1–13.

O’Rahilly R, M€uller F. Developmental stages in human embryos, including a revision of Steeter’s“Horizons” and a survey of the Carnegie Collection. Carnegie Institution of Washington,Publication 637. 1987; 306p.

O’Rahilly R, M€uller F. Bidirectional closure of the rostral neuropore in the human embryo. Am JAnat. 1989;184:259–68.

O’Rahilly R, M€uller F. The two sites of fusion of the neural folds and the two neuropores in thehuman embryo. Teratology. 2002;65:162–70.

Osório L, Teillet M-A, Palmeirim I, Catala M. Neural crest ontogeny during secondary neurulation:a gene expression pattern study in the chick embryo. Int J Dev Biol. 2009a;53:641–8.

Embryology of the Spine and Spinal Cord 49

Osório L, Teillet M-A, Catala M. Role of noggin as an upstream signal in the lack of neuronalderivatives found in the avian caudal-most neural crest. Development. 2009b;136:1717–26.

Pang D, Dias MS, Ahab-Barmada M. Split cord malformations: part I: a unified theory ofembryogenesis for double spinal cord malformations. Neurosurgery. 1992;31:451–80.

Pasteels J. Etudes sur la gastrulation des vertébrés méroblastiques. III. Oiseaux. IV. Conclusionsgénérales. Arch Biol. 1937;48:381–488.

Patten I, Kulesa P, Shen MM, Fraser S, Placzek M. Distinct modes of floor plate induction in thechick embryo. Development. 2003;130:4809–21.

Peeters MC, Viebahn C, Hekking JW, van Straaten HW. Neurulation in the rabbit embryo. AnatEmbryol. 1998;197:167–75.

Pennimpede T, Proske J, König A, Vidigal JA, Morkel M, Bramsen JB, Herrmann BG, Wittler L. Invivo knockdown of Brachyury in skeletal defects results in skeletal defects and urorectalmalformations resembling caudal regression syndrome. Dev Biol. 2012;372:55–67.

Perea-Gomez A, Vella FD, Shawlot W, Oulad-Abdelghani M, Chazaud C, Meno C, Pfister V,Chen L, Robertson E, Hamada H, Behringer RR, Ang S-L. Nodal antagonists in the anteriorvisceral endoderm prevent the formation of multiple primitive streaks. Dev Cell.2002;3:745–56.

Perron JC, Dodd J. Inductive specification and axonal orientation of spinal neurons mediated bydivergent bone morphogenetic protein signaling pathways. Neural Dev. 2011;6:36.

Peters H, Wilm B, Sakai N, Imai K, Maas R, Balling R. Pax1 and Pax9 synergistically regulatevertebrate column development. Development. 1999;126:5399–408.

Piccolo S, Sasai Y, Lu B, De Robertis EM. Dorsoventral patterning in Xenopus: inhibition of ventralsignals by direct binding of chordin to BMP-4. Cell. 1996;86:589–98.

Pinho S, Simonsson PR, Trevers KE, Stower MJ, Sherlock WT, Khan M, Streit A, Sheng G, SternCD. Distinct steps of neural induction revealed by Asterix, Obelix and TrkC, genes induced bydifferent signals from the organizer. PLoS One. 2011;6(4):e19157. doi:10.1371/journal.pone.0019157.

Placzek M, Tessier-Lavigne M, Yamada T, Jessell T, Dodd J. Mesodermal control of neural cellidentity: floor plate induction by the notochord. Science. 1990;250:985–8.

Postma AV, Alders M, Sylva M, Bilardo CM, Pajkrt E, van Rijn RR, Schulte-Merker S, Bulk S,Stefanovic S, Ilgun A, Barnett P, Mannens MMAM, Moorman AFM, Oostra RJ, van MaarleMC. Mutations in the T (brachyury) gene cause a novel syndrome consisting of sacral agenesis,abnormal ossification of the vertebral bodies and a persistent notochordal canal. J Med Genet.2014;51:90–7.

Pourquié O, Coltey M, Teillet M-A, Ordahl C, Le Douarin NM. Control of dorsoventral patterningof somatic derivatives by notochord and floor plate. Proc Natl Acad Sci U S A. 1993;90:5242–6.

Psychoyos D, Stern CD. Fates and migratory routes of primitive streak cells in the chick embryo.Development. 1996;122:1523–34.

Rickmann M, Fawcett JW, Keynes RJ. The migration of neural crest cells and the growth of motoraxons through the rostral half of the chick somite. J Embryol Exp Morphol. 1985;90:437–55.

Ring C, Hassell H, Hafter W. Expression pattern of collagen IX and potential role in segmentationof the peripheral nervous system. Dev Biol. 1996;180:41–53.

Rivera-Pérez JA, Mager J, Magnuson T. Dynamic morphogenetic events characterize the mousevisceral endoderm. Dev Biol. 2003;261:470–87.

Robertson CP, Braun MM, Roelink H. Sonic hedgehog patterning in chick neural plate is antago-nized by a Wnt3-like signal. Dev Dyn. 2004;229:510–9.

Roelink H, Augsburger A, Heemskerk J, Korzh V, Norlin S, Ruiz i Altaba A, Tanabe Y, Placzek M,Edlund T, Jessell TM, Dodd J. Floor plate and motor neuron induction by vhh-1, a vertebratehomolog of hedgehog expressed by the notochord. Cell. 1994;76:761–75.

Roessler E, Belloni E, Gaudenz K, Jay P, Berta P, Scherer SW, Tsui L-C, Muenke M. Mutations inthe human Sonic Hedgehog gene cause holoprosencephaly. Nat Genet. 1996;14:357–60.

Ruiz i Altaba A, Nguyên V, Palma V. The emergent design of the neural tube: prepattern, SHHmorphogen and GLI code. Curr Opin Genet Dev. 2003;13:513–21.

50 M. Catala

Saitsu H, Yamada S, Uwabe C, Ishibashi M, Shiota K. Development of the posterior neural tube inhuman embryos. Anat Embryol. 2004;209:107–17.

Saitsu H, Yamada S, Uwabe C, Ishibashi M, Shiota K. Aberrant differentiation of the axiallycondensed tail bud mesenchyme in human embryos with lumbosacral myeloschisis. Anat Rec.2007;290:251–8.

Sakai Y. Neurulation in the mouse: manner and timing of neural tube closure. Anat Rec.1989;223:194–203.

Santiago A, Erickson CA. Ephrin-B ligands play a dual role in the control of neural crest cellmigration. Development. 2002;129:3621–32.

Sasai Y, Lu B, Steinbeisser H, Geissert D, Gont LK, De Robertis EM. Xenopus chordin: a noveldorsalizing factor activated by organizer-specific homeobox genes. Cell. 1994;79:779–90.

Sasaki H, Hogan BLM. HNF-3β as a regulator of floor plate development. Cell. 1994;76:103–15.Schoenwolf GC. Observations on closure of the neuropores in the chick embryo. Am J Anat.

1979a;155:445–66.Schoenwolf GC. Histological and ultrastructural observations of tail bud formation in the chick

embryo. Anat Rec. 1979b;193:131–48.Schoenwolf GC. Histological and ultrastructural studies of secondary neurulation in mouse

embryos. Am J Anat. 1984;169:361–76.Schoenwolf GC. Cell movements driving neurulation in avian embryos. Development. 1991;

S2:157–68.Schoenwolf GC, Alvarez IS. Roles of the neuroepithelial cell rearrangement and division in shaping

of the avian neural plate. Development. 1989;106:427–39.Schoenwolf GC, Delongo J. Ultrastructure of secondary neurulation in the chick embryo. Am J

Anat. 1980;158:43–63.Schoenwolf GC, Sheard P. Shaping and bending of the avian neural plate as analysed with a

fluorescent-histochemical marker. Development. 1989;105:17–25.Schoenwolf GC, Smith JL. Mechanisms of neurulation: traditional viewpoint and recent advances.

Development. 1990;109:243–70.Schumacher S. Über die sogenannte Vervielfachung des Medullarrohres (bzw. des Canalis centralis)

bei Embryonen. Z Mikrosk Anat Forsch. 1927;10:75–109.Schwarz Q, Maden CH, Vieira JM, Ruhrberg C. Neuropilin 1 signaling guides neural crest cells to

coordinate pathway choice with cell specification. Proc Natl Acad Sci U SA. 2009a;106:6164–9.

Schwarz Q, Maden CH, Davidson K, Ruhrberg C. Neuropilin-mediated neural crest cell guidance isessential to organise sensory neurons into segmented dorsal root ganglia. Development.2009b;136:1785–9.

Serbedzija GN, Bronner-Fraser M, Fraser SE. A vital dye analysis of the timing and pathways ofavian trunk neural crest cell migration. Development. 1989;106:809–16.

Shedden PM, Wiley MJ. Early stages of development in the caudal neural tube of the golden Syrianhamster (Mesocricetus auratus). Anat Rec. 1987;219:180–5.

Shimokita E, Takahashi Y. Secondary neurulation: fate-mapping and gene manipulation of theneural tube in tail bud. Dev Growth Differ. 2011;53:401–10.

Shum AS, Copp AJ. Regional differences in morphogenesis of the neuroepitheliumsuggest multiple mechanisms of spinal neurulation in the mouse. Anat Embryol. 1996;194:65–73.

Shum ASW, Tang LSC, Copp AJ, Roelink H. Lack of motor neuron differentiation is an intrinsicproperty of the mouse secondary neural tube. Dev Dyn. 2010;239:3192–203.

Skromme I, Stern CD. A hierarchy of gene expression accompanying induction of the primitivestreak by Vg1 in the chick embryo. Mech Dev. 2002;114:115–8.

Smith WC, Harland RM. Expression cloning of noggin, a new dorsalizing factor localized to theSpemann organizer in Xenopus embryos. Cell. 1992;70:829–40.

Smith JL, Schoenwolf GC. Notochordal induction of cell wedging in the chick neural plate and itsrole in neural tube formation. J Exp Zool. 1989;250:49–62.

Embryology of the Spine and Spinal Cord 51

Spemann H. Embryonic development and induction. New Haven: Yale University Press; 1938.401p.

Spence MS, Yip J, Erickson CA. The dorsal neural tube organizes the dermamyotome and inducesaxial myocytes in the avian embryo. Development. 1996;122:231–41.

Steding G. The anatomy of the human embryo. A scanning electron-microscopic atlas. Basel:Karger; 2009.

Stottmann RW, Klingesmith J. Bone morphogenetic protein signaling is required in the dorsalneural folds before neurulation for the induction of spinal neural crest cells and dorsal neurons.Dev Dyn. 2011;240:755–65.

Streeter GL. Factors involved in the formation of the filum terminale. Am J Anat. 1919;25:1–11.Streit A, Berliner AJ, Papanayotou C, Sirulnik A, Stern CD. Initiation of neural induction by FGF

signalling before gastrulation. Nature. 2000;406:74–8.Sulik K, Dehart DB, Inagaki T, Carson JL, Vrablic T, Gesteland K, Schoenwolf GC. Morphogenesis

of the murine node and notochordal plate. Dev Dyn. 1994;201:260–78.Sun X, Meyers EN, Lewandoski M, Martin GR. Targeted disruption of Fgf8 causes failure of cell

migration in the gastrulating mouse embryo. Genes Dev. 1999;13:1834–46.Suzuki A, Kaneko E, Ueno N, Hemmati-Brivanlou A. Regulation of epidermal induction by BMP2

and BMP7 signaling. Dev Biol. 1997;189:112–22.Tam PPL, Beddington RSP. The formation of mesodermal tissues in the mouse embryo during

gastrulation and early organogenesis. Development. 1987;99:109–26.Tam PP, Williams EA, Chan WY. Gastrulation in the mouse embryo: ultrastructural and molecular

aspects of germ layer morphogenesis. Microsc Res Tech. 1993;26:301–28.Teillet M-A, Kalcheim C, Le Douarin NM. Formation of the dorsal root ganglia in the avian

embryo: segmental origin and migratory behavior of neural crest progenitor cells. Dev Biol.1987;120:329–47.

Teillet M-A, Lapointe F, Le Douarin NM. The relationships between notochord and floor plate invertebrate development revisited. Proc Natl Acad Sci U S A. 1998;95:11733–8.

Thiéry JP, Duband JL, Delouvée A. Pathways and mechanisms of avian trunk neural crest cellmigration and localization. Dev Biol. 1982;93:324–43.

Thomas AJ, Erickson CA. FOXD3 regulates the lineage switch between neural crest-derived glialcells and pigment cells by repressing MITF through a non-canonical mechanism. Development.2009;136:1849–58.

Tortori-Donati P, Fondelli MP, Rossi A, Raybaud CA, Cama A, Capra V. Segmental spinaldysgenesis: neuroradiologic findings with clinical and embryologic correlation. Am JNeuroradiol. 1999;20:445–56.

Tribioli C, Lufkin T. The murine Bapx1 homeobox gene plays a critical role in embryonicdevelopment of the axial skeleton and spleen. Development. 1999;126:5699–711.

Tucker AS, Slack JMW. The Xenopus tail forming region. Development. 1995a;121:249–62.Tucker AS, Slack JMW. Tail bud determination in the vertebrate embryo. Curr Biol.

1995b;5:807–13.Tucker RP, Hagios C, Chiquet-Ehrismann R, Lawler J, Hall RJ, Erickson CA. Thrombospondin-1

and neural crest cell migration. Dev Dyn. 1999;214:312–22.Ulloa F, Martí E. Wnt won the war: antagonistic role of Wnt over Shh controls dorso-ventral

patterning of the vertebrate neural tube. Dev Dyn. 2010;239:69–76.Van Allen MI, Kalousek DK, Chernoff GF, Juriloff D, Harris M, McGillivray BC, Yong S-L,

Langlois S, MacLeod PM, Chitayat D, Friedman JM, Wilson RD, McFadden D, Pantzar J,Richtie S, Hall JG. Evidence for multi-site closure of the neural tube in humans. Am J MedGenet. 1993;47:723–43.

van Straaten HWM, Hekking JWM. Development of floor plate, neurons and axonal outgrowthpattern in the early spinal cord of the notochord-deficient chick embryo. Anat Embryol.1991;184:55–63.

52 M. Catala

van Straaten HWM, Thors F, Wiertz-Hoessels L, Hekking DJ. Effect of a notochordal implant onthe early morphogenesis of the neural tube and neuroblasts: histometrical and histologicalresults. Dev Biol. 1985;110:247–54.

van Straaten HW, Janssen HC, Peeters MC, Copp AJ, Hekking JW. Neural tube closure in the chickembryo is multiphasic. Dev Dyn. 1996;207:309–18.

van Straaten HW, Peeters MC, Hekking JW, van der Lende T. Neurulation in the pig embryo. AnatEmbryol. 2000;202:75–84.

Viebahn C, Stortz C, Mitchell SA, Blum M. Low proliferative and high mobility activity in the areaof Brachyury expressing mesoderm progenitor cells in the gastrulating rabbit embryo. Devel-opment. 2002;129:2355–65.

Wallingford JB, Rowning BA, Vogell KM, Rothbächer U, Fraser SE, Harland RM. Dishevelledcontrols cell polarity during Xenopus gastrulation. Nature. 2000;405:81–5.

Watanabe Y, Duprez D, Monsoro-Burq A-H, Vincent C, Le Douarin NM. Two domains in vertebraldevelopment: antagonistic regulation by SHH and BMP4 proteins. Development.1998;125:2631–9.

Waterman RE. Topographical changes along the neural fold associated with neurulation in thehamster and mouse. Am J Anat. 1976;146:151–71.

Wijgerde M, McMahon JA, Rule M, McMahon AP. A direct requirement for Hedgehog signalingfor normal specification of all ventral progenitor domains in the presumptive spinal cord. GenesDev. 2002;16:2849–64.

Williams M, Burdsal C, Periasamy A, Lewandoski M, Sutherland A. Mouse primitive streak formsin situ by initiation of epithelial to mesenchymal transition without migration of a cell popula-tion. Dev Dyn. 2012;241:270–83.

Wilson V, Beddington RS. Cell fate and morphogenetic movement in the late mouse primitivestreak. Mech Dev. 1996;55:79–89.

Wilson PA, Hemmati-Brivanlou A. Induction of epidermis and inhibition of neural fate by BMP-4.Nature. 1995;376:331–3.

Wilson SI, Rydström A, Trimborn T, Willert K, Nusse R, Jessell TM, Edlund T. The status of Wntsignalling regulates neural and epidermal fates in the chick embryo. Nature. 2001;411:325–30.

Wong GK, Bagnall KM, Berdan RC. The immediate fate of cells in the epithelial somite of the chickembryo. Anat Embryol. 1993;188:441–7.

Yamada Y, Placzek M, Tanaka H, Dodd J, Jessell TM. Control of cell pattern in the developingnervous system: polarizing activity of the floor plate and notochord. Cell. 1991;64:635–47.

Yamanaka Y, Tamplin OJ, Beckers A, Gossler A, Rossant J. Live imaging and genetic analysis ofmouse notochord formation reveals regional morphogenetic mechanisms. Dev Cell.2007;13:884–96.

Yang Y, Li C, Xu X, Deng C. The tumor suppressor SMAD4/DPC4 is essential for epiblastproliferation and mesoderm induction in mice. Proc Natl Acad Sci U S A. 1998;95:3667–72.

Ybot-Gonzales P, Cogram P, Gerrelli D, Copp AJ. Sonic hedgehog and the molecular regulation ofmouse neural tube closure. Development. 2002;129:2507–17.

Ybot-Gonzales P, Gaston-Massuet C, Girdler G, Klingensmith J, Arkell R, Greene NDE, CoppAJ. Neural plate morphogenesis during mouse neurulation is regulated by antagonism of Bmpsignalling. Development. 2007;134:3203–11.

Zimerman LB, De Jesus-Escobar JM, Harland RM. The Spemann organizer signal noggin bindsand inactivates bone morphogenetic protein 4. Cell. 1996;86:599–606.

Embryology of the Spine and Spinal Cord 53