Chapter 1 Development and maturation of the spinal cord: implications of molecular and genetic defects GREGORY W.J. HAWRYLUK 1, 2 , CRYSTAL A. RUFF 1 , AND MICHAEL G. FEHLINGS 1, 2, 3 * 1 Division of Genetics and Development, Toronto Western Research Institute, Institute of Medical Science, University of Toronto, Toronto, Canada 2 Division of Neurosurgery, University of Toronto, Toronto, Canada 3 Spinal Program, University Health Network, Toronto Western Hospital, Toronto, Canada GROSS EMBRYOLOGY Overview A human term pregnancy lasts approximately 40 weeks, and the most dramatic and complex developmental pro- cesses are completed in the embryonic period spanning the first 8 weeks. In the embryonic period, critical devel- opmental milestones include establishment of the midline and anteroposterior axis, formation of the three germ layers through gastrulation, and organogenesis. The sub- sequent fetal period is comparatively simple, wherein the developing human predominantly grows in size. Fertilization to gastrulation Fertilization characteristically takes place in the ampul- lary region of the fallopian tube. The fertilized egg then undergoes a number of mitotic divisions, eventually forming a 16-cell morula 3 days after fertilization (Fig. 1.1). Around the time the morula enters the uterus, it becomes known as a blastocyst and develops a cystic cavity known as a blastocele. By this time, the inner and outer cell masses have formed, which give rise to the em- bryo proper and the placenta respectively. The outer cell mass, also known as the trophoblast, secretes proteolytic enzymes which facilitate implantation in the endome- trium, which occurs about 1 week following fertilization. In the second week of development, the inner cell mass, now known as the embryoblast, separates into two distinct cell layers, the hypoblast and the epiblast, which form the endoderm and ectoderm respectively. A second cystic cavity then develops adjacent to the epiblast. These layers thus form a bilaminar disc sand- wiched between two cavities; the hypoblast lines the blastocyst cavity (primitive yolk sac) while the epiblast lines the developing amniotic cavity. Gastrulation and Hensen’s node In the second week gastrulation occurs, which establishes the third germ layer, mesoderm (Fig. 1.2). Gastrulation be- gins with formation of the primitive streak in the caudal region of the epiblast. The cranial end of the primitive streak forms a thickening known variously as the primi- tive knot, the primitive node, or Hensen’s node. The prim- itive pit forms immediately posterior to the node and cells from the epiblast migrate here, invaginate, and then form intraembryonic endoderm and mesoderm. The primitive node migrates caudally as gastrulation progresses, and although it typically regresses and forms the caudal eminence or end bud after migration to the sacrococcygeal area, it is deserving of some fur- ther discussion. Hensen’s node secretes morphogens such as fibroblast growth factor (FGF), sonic hedgehog (Shh) and retinoic acid (RA), playing key roles in neural induction and patterning which will be discussed in detail. In this fashion, Hensen’s node establishes the longitudinal axis, polarity and right–left sidedness within the embryo. It also participates in rostrocaudal specification along with paraxial mesoderm. Failure of Hensen’s node to regress can lead to formation of a sacrococcygeal teratoma. *Correspondence to: Michael G. Fehlings MD, PhD, FRCSC, FACS, Krembil Chair, Neural Repair and Regeneration, Head, Spinal Program, University Health Network, Toronto Western Hospital, McLaughlin Pavilion, 12th floor Rm. 407, 399 Bathurst Street, Toronto, Ontario, Canada M5T 2S8. Tel: þ 1-416-603-5627, Fax: þ 1-416-603-5298, E-mail: [email protected]Handbook of Clinical Neurology, Vol. 109 (3rd series) Spinal Cord Injury J. Verhaagen and J.W. McDonald III, Editors # 2012 Elsevier B.V. All rights reserved
28
Embed
Development and maturation of the spinal cord ... · PDF fileDevelopment and maturation of the spinal cord: implications of molecular and genetic ... bryoproperandthe placenta ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Handbook of Clinical Neurology, Vol. 109 (3rd series)Spinal Cord InjuryJ. Verhaagen and J.W. McDonald III, Editors# 2012 Elsevier B.V. All rights reserved
Chapter 1
Development and maturation of the spinal cord: implications
of molecular and genetic defects
GREGORY W.J. HAWRYLUK1, 2, CRYSTAL A. RUFF 1, AND MICHAEL G. FEHLINGS 1, 2, 3*1Division of Genetics and Development, Toronto Western Research Institute, Institute of Medical Science,
University of Toronto, Toronto, Canada2Division of Neurosurgery, University of Toronto, Toronto, Canada
3Spinal Program, University Health Network, Toronto Western Hospital, Toronto, Canada
GROSS EMBRYOLOGY
Overview
A human term pregnancy lasts approximately 40 weeks,and the most dramatic and complex developmental pro-cesses are completed in the embryonic period spanningthe first 8 weeks. In the embryonic period, critical devel-opmental milestones include establishment of the midlineand anteroposterior axis, formation of the three germlayers through gastrulation, and organogenesis. The sub-sequent fetal period is comparatively simple, wherein thedeveloping human predominantly grows in size.
Fertilization to gastrulation
Fertilization characteristically takes place in the ampul-lary region of the fallopian tube. The fertilized egg thenundergoes a number of mitotic divisions, eventuallyforming a 16-cell morula 3 days after fertilization(Fig. 1.1). Around the time the morula enters the uterus,it becomes known as a blastocyst and develops a cysticcavity known as a blastocele. By this time, the inner andouter cell masses have formed, which give rise to the em-bryo proper and the placenta respectively. The outer cellmass, also known as the trophoblast, secretes proteolyticenzymes which facilitate implantation in the endome-trium, which occurs about 1 week following fertilization.
In the second week of development, the inner cellmass, now known as the embryoblast, separates intotwo distinct cell layers, the hypoblast and the epiblast,which form the endoderm and ectoderm respectively.
*Correspondence to: Michael G. FehlingsMD, PhD, FRCSC, FACSProgram, University Health Network, Toronto Western Hospital,
A second cystic cavity then develops adjacent to theepiblast. These layers thus form a bilaminar disc sand-wiched between two cavities; the hypoblast lines theblastocyst cavity (primitive yolk sac) while the epiblastlines the developing amniotic cavity.
Gastrulation and Hensen’s node
In the second week gastrulation occurs, which establishesthe third germ layer, mesoderm (Fig. 1.2). Gastrulation be-gins with formation of the primitive streak in the caudalregion of the epiblast. The cranial end of the primitivestreak forms a thickening known variously as the primi-tive knot, the primitive node, or Hensen’s node. The prim-itive pit forms immediately posterior to the node and cellsfrom the epiblast migrate here, invaginate, and then formintraembryonic endoderm and mesoderm.
The primitive node migrates caudally as gastrulationprogresses, and although it typically regresses andforms the caudal eminence or end bud after migrationto the sacrococcygeal area, it is deserving of some fur-ther discussion. Hensen’s node secretes morphogenssuch as fibroblast growth factor (FGF), sonic hedgehog(Shh) and retinoic acid (RA), playing key roles in neuralinduction and patterning which will be discussed indetail. In this fashion, Hensen’s node establishes thelongitudinal axis, polarity and right–left sidednesswithin the embryo. It also participates in rostrocaudalspecification along with paraxial mesoderm. Failure ofHensen’s node to regress can lead to formation of asacrococcygeal teratoma.
Fig. 1.1. Development of the bilaminar disc. The two-cell stage (A) is reached approximately 30 h after fertilization and the
zygote eventually forms a 16-cell morula 2 days later (B). Inner and outer cell masses form at this time during a process referred
to as compaction. The inner cell mass goes on to form the embryo, while the outer cell mass or trophoblast forms the placenta and
extra-embryonic membranes. These masses become more apparent when the morula becomes a blastocyst 4.5 days after fertil-
ization (C) and the blastocyst cavity develops. With further development the inner cell mass is known as the embryoblast. The
blastocyst typically implants in the uterine mucosa 5–6 days after fertilization. The bilaminar disc forms within the blastocyst
during the second week of development when the amniotic cavity develops within the epiblast (D). The constituent layers of
the bilaminar disc are the epiblast (primitive ectoderm, lining the amniotic cavity) and the hypoblast (primitive endoderm, lining
the primitive yolk sac).
4 G.W.J. HAWRYLUK ET AL.
Formation of the notochord
Another critical event occurring in the second week isthe formation of the notochord (see Fig. 1.2). The noto-chord is a cylindrical structure derived from mesoder-mal cells which specifies the midline of the embryo, inaddition to forming a rigid axis around which theembryo can develop. It also secretes inductive signalscritical to the formation of the nervous system fromthe overlying ectoderm.
Prenotochordal cells which form the notochord mi-grate in from the primitive streak, and move rostrallytoward the prechordal plate (future buccopharyngealmembrane) to form the notochordal process, a precur-sor of the notochord. The notochordal process initiallyintercalates with the hypoblast to form the notochordalplate. At this time an important transitory communica-tion between the amniotic cavity and yolk sac formswhich is known as the neurenteric canal. This canal isof great significance to spine and spinal cord maldeve-lopment, as it is currently believed to play a critical rolein numerous malformations such as neurenteric cystsand split cord malformations (Pang and Dias, 1992),as will be discussed.
Almost immediately after the notochordal plate formsand directly contacts the yolk sac, it separates from the en-doderm, moves slightly dorsally and re-forms a cord ofcells running along the rostrocaudal axis of the embryo’smidline. Despite its embryological significance, few noto-chordremnantspersist in theadult.Thesecellsmakeupthenucleus pulposus at the center of the intervertebral discand notochord remnants are also believed to give rise tochordomas and notochordal rests (Kyriakos et al., 2003).
Primary neurulation
The central nervous system begins to develop in the thirdweek postfertilization and the process begins withneurulation (Figs 1.3, 1.4). At the outset of neurulation,the notochord induces a subset of ectodermal cells todifferentiate into neural precursor cells, forming acolumnar epithelium referred to as the neural plate.Primary neurulation occurs when the neural plate foldsand closes to form the neural tube.
Neural tube closure is mechanistically complex.Central to this process is the medial hinge point, orfloor plate, which forms in the ventral midline, fromnotochord-derived cells under the inductive influence
Fig. 1.2. Gastrulation and development of the notochord. At the end of the second week of development a thickening of cells
forms in the caudal midline of the bilaminar germ disc, referred to as the primitive streak (A). The prechordal plate is visible
at the rostral end of the disc and eventually develops into the buccopharyngeal membrane. (B) and (C) show coronal views through
the bilaminar disc. Epiblast cells invaginate at the primitive pit and primitive streak creating the cells of the definitive endoderm as
well as the mesoderm through the process of gastrulation. Prenotochordal cells invaginate during this process and migrate as far
rostral as the prechordal plate. Initially they intercalate with the hypoblast forming the notochordal plate (E). The notochordalplate then detaches from the endoderm, and forms a tube referred to as the definitive notochord (F). (E) and (F) are coronal viewslooking rostral from planes Ro and Ca shown in (D), which is a mid-sagittal section through the embryo at 17d postfertilization.
The neurenteric canal is a temporary communication between the amniotic cavity and yolk sac believed to play a central role in
many malformations of the spine and spinal cord.
DEVELOPMENT AND MATURATION OF THE SPINAL CORD 5
of the notochord. The floor plate is the key attachmentpoint around which the neural folds can elevate as itscells acquire a pyramidal shape due to contraction of ac-tin-like microfilaments at their apices. Closure of theneural tube takes 4–6 days. The folds initially meet inthe region of the fourth somite, at the junction of whatbecomes the hindbrain and spinal cord. Further fusionthen proceeds in both rostral and caudal directions withcaudal closure occurring alongside newly developedsomites. Harris and Juriloff have noted that at least inmice, the precise mechanics of neural tube closure aredistinct at different rostrocaudal sites (Harris and Juril-off, 1999). They described four rostrocaudal zones(A–D) based on differences in the regions of initialcontact and sequence of tissue fusion (Fig. 1.5). For in-stance, in the most rostral zone, A, fusion of neuro-epithelium precedes that of surface ectoderm, while in
zone B, contact and fusion of these layers are concur-rent. In zone C, fusion of surface ectoderm occurs priorto the neuroepithelium (Geele and Langman, 1977;Harris et al., 1994; Shum and Copp, 1996). Notably, cellsof the neural plate switch from expression of E-cadherinto N-cadherin intercellular adhesion molecules toprevent the neural folds from fusing inappropriatelywith the epidermis (Takeichi et al., 1990; Edelman andJones, 1998).
During closure, the as yet unfused rostral and caudalregions are referred to as neuropores. The cranial andcaudal neuropores remain open to the amniotic fluid un-til closure at approximately 25 and 27 days respectively.Failure of the cranial and caudal neuropores to close re-sults in anencephaly and spina bifida respectively, whilemore extensive failure of neural tube closure is referredto as craniorachischisis (anencephaly with contiguous
Fig. 1.3. Primary neurulation. The neural tube and cells of the
neural crest are derived from surface ectoderm, forming a co-
lumnar epithelium referred to as the neural plate, as a result of
induction by the notochord. Neural crest cells initially reside
lateral to those that will form the neural tube (A). Folding in-
ternalizes these cells (B). Medial and lateral hinge points serve
to anchor the neural tube, facilitating this folding (C). The me-
dial hinge point is also known as the floor plate. The neural
crest cells separate and form a mass dorsal to the neural tube
(D). They later migrate to form dorsal root ganglia and many
important cells types within the embryo (E).
Fig. 1.4. Closure of the neural tube. Dorsal views of the embryo ar
with a prominent neural plate is about to undergo primary neurula
and the neural folds begin to meet and fuse; this process proceeds
and 27th day respectively). Until fusion is completed, neural tissue
to as the rostral and caudal neuropores. (C) and (D) are dorsal vi
6 G.W.J. HAWRY
spinal defect involving at least the cervical spine).Prolonged exposure to amniotic fluid is destructiveto neural tissue, though in anencephaly, some neuraltissue typically remains including the adenohypophysisand brainstem which are covered with vascular tissuereferred to as the area cerebrovasculosa.
Secondary neurulation
Unlike more rostral elements of the spinal cord, theconus medullaris and the filum terminale form throughthe process of secondary neurulation, which remainspoorly understood. Neural folds do not form or fusehere: instead cells from the neural ectoderm as well assome from the endoderm condense to form the medul-lary cord which later canalizes and connects with therostral neural tube. Exogenous RA administrationprevents this connection (Yasuda et al., 1990).
Development of the spinal cord
Upon closure of the neural tube, its constituent neuro-epithelial cells give rise to primitive nerve cells known asneuroblasts. These neuroblasts form a new layer withinthe developing spinal cord, known as the mantle layer,which becomes the gray matter of the spinal cord. An ad-ditional, outer layer forms from axons proceeding in andout of the mantle layer. This outer layer is referred to asthe marginal layer and it becomes the white matter of thespinal cord.
As development progresses, additional neuroblastsare added to the mantle layer resulting in dorsal and ven-tral thickenings called the basal plates and alar plates re-spectively. The basal plates form gray matter whichsubserves motor function while the alar plates subservesensory function. An additional lateral swelling forms in
LUK ET AL.
e shown with the amnion removed. (A) An 18-day-old embryo
tion. (B) At 20 days post-fertilization, somites begin to appear
bidirectionally. The rostral and caudal extremes fuse last (25th
is exposed to amniotic fluid. The unfused extremes are referred
ews at days 22 and 23 respectively.
Elevation ZonesDay 8 to 9
Fusion Initiation SitesDay 8 to 9
PNP
A B C Fate of Elevation ZonesDay 14
sb
exB
B
2
A
A
3
DD
1
C
C
4
Fig. 1.5. Regions of neural tube with distinct closure mechanisms in mouse. Four regions of the mouse embryo have been iden-
tified with unique mechanisms of neural fold elevation and fusion (regions A–D). In (A) triangles denote sites of initial fusion.Lines with arrows denote the subsequent direction of fusion. PNP, posterior neuropore; ex and sb indicate locations of open neural
folds in exencephaly and spina bifida aperta respectively. (From Harris and Juriloff, 1999. Used with permission.)
DEVELOPMENT AND MATURATION OF THE SPINAL CORD 7
the gray matter of thoracic and upper lumbar regionsof the spinal cord and is known as the intermediatehorn which contains nerve cells of the sympatheticnervous system.
Early in development, the spinal cord extendsthroughout the entire length of the embryo, and spinalnerves pass through the immediately adjacent neural fo-ramina. Beginning in the third fetal month, the vertebralcolumn and dura lengthen more quickly than the spinalcord. In the adult, the conus thus lies at the level of theL1–L2 intervertebral disc and nerve roots must descendto leave from their once-adjacent neural foramina.
Dorsal structures and neural crest cells
While the neural tube is closing, the cutaneous ectodermseparates from the neuroectoderm to form the overlyingskin while the lateral mesoderm migrates between thesetwo layers to form the posterior vertebral arches. Failureof this mesodermal migration leads to spina bifidaocculta, which is common and generally asymptomatic.Neural crest cells form from ectoderm just lateral to theregionwhich gives rise to the neural plate. These cells takean intermediate position between the neural tube and sur-face ectoderm (see Fig. 1.3) and then divide into twogroups: those that go dorsally to become melanocytes,and a group migrating ventrolaterally to form the dorsalroot ganglia, Schwann cells, odontoblasts, meninges,sympathetic and adrenal ganglia, and mesenchyme of
the pharyngeal arches. N-cadherin is expressed in highlevels before and after, but not during, dorsoventral mi-gration (Pla et al., 2001). Shh is believed to mediate thismigration by suppressing the cell adhesion moleculeintegrin (Testaz and Duband, 2001).
CELLULAR ANDMOLECULAREMBRYOLOGYOF THE SPINE
Overview/general principles
Innumerable molecules appear to play a role in spinalcord development; however, a handful of key regulatorshave been discovered, as have general paradigms sur-rounding their activities. Regulators known as mor-phogens coordinate complex developmental processesfollowing their secretion from regions known as orga-nizing centers. These molecules are known to have tem-porally and spatially distinct effects – a single moleculecan have different influences on spatially separated tis-sues simultaneously and can additionally have differenteffects on the same cells and tissues at distinct develop-mental time points.
The notion of “competence” is key to understandinghow this occurs. Competence refers to a cell or tissue’sability to respond to a signal and largely relates to itsreceptor expression profile. Different cells and tissuesoften express variable receptors and receptor isoformsat the same developmental time points. Likewise, the
Y
same cells and tissues may express different receptorsand receptor isoforms at distinct developmental timepoints. This facilitates responsiveness to alternate sig-nals, or changed responses to the same signals. Anotherkey mechanism responsible for divergent developmentis the ability of cells to respond differentially to slightdifferences in the concentration of morphogens, whichform concentration gradients as they diffuse throughembryological tissue.
A final key property of this system is that the differ-ential responses are mediated ultimately by expression ofunique transcription factors directing distinct cells andtissues to their distinct developmental fates. In the spine,these transcription factors are generally Hox and Limhomeodomain proteins that have been grouped intoclasses I and II based on their response to morphogens(Jungbluth et al., 1999; Dawid and Chitnis, 2001).
8 G.W.J. HAWR
Neural induction
Work by Spemann and others demonstrated that the CNSmust be induced to form from ectoderm (Holtfreter,1988). Ablation and ectopic transplantation experimentsshowed that the notochord is responsible for this neuralinduction (Placzek, 1995). It has become clear that bonemorphogenic protein 4 (BMP4), a TGFb family member,plays a central role in this process. BMP4, produced by theectoderm, inhibits formation of neural tissue, triggeringectodermal cells to differentiate into epidermis. Neural in-duction thus requires inhibition of BMP4, which is accom-plished by chordin, noggin and follistatin produced by thenotochord and paraxial mesoderm (Lamb et al., 1993).
Regulation of stem cell and progenitorcell migration
Like cell fate decisions, complex intrinsic and extrinsiccues govern cell migration. In neurons, induction ofcyclin-dependent kinase inhibitor p27Kip1 is requiredfor radial migration of cortical neurons, just as it isimportant for stem cell differentiation. Ngn2 then in-duces transcription of genes involved in cell migration(Gensert and Goldman, 1997). Rho A is an important sig-naling molecule in the process of neuronal migration, asit is downstream of both p27Kip1 and the semaphorin/plexin/neuropilin system (Deng et al., 2007; Nguyenet al., 2007). Semaphorin action leads to activationof transmembrane receptors called plexins and theirco-receptors, the neuropilins. After binding and activa-tion by semaphorin 4 C, receptor plexin-B2 can regulateproliferation and migration of granule cell precursors inthe developing dentate gyrus, olfactory bulb, and cerebel-lum, through ErbB-2 and RhoA associated mechanisms(Deng et al., 2007).
Oligodendrocytes migrate to follow axons (de Castroand Bribian, 2005). Oligodendrocyte precursor cells(OPCs) originate in multiple but discrete foci alongthe neural tube. In many cases, the germinal foci forOPCs overlap territories that give rise to different typesof neurons. Shh signaling appears to play an evenmore important role in specifying oligodendrocytes thanneurons, possibly acting through basic helix-loop-helixproteins such as Olig1 and 2 (de Castro and Bribian,2005). Various contact-dependent cues support or pre-vent OPC migration, targeting these cells to the appropri-ate location. For instance, fibronectin and merosinpromotemigration while tenascin-C inhibits themigrationof some (but not all) oligodendrocytes. During late em-bryonic and early postnatal stages, both polysialylatedneural cell adhesion molecule (PSA-NCAM) and polysia-lic acid contribute to the migration of OPCs (Wang et al.,1994). In addition, avb-integrins are key mediators ofOPC migration in vitro (Milner et al., 1997).
In the forebrain, some glial precursors derived fromthe subventricular zone undergo radial migration in thefirst postnatal week (Kakita and Goldman, 1999; Mallonet al., 2002). This observation supports the hypothesisthat radial glia would be involved in the migration notonly of neurons but also of OPCs. This hypothesiswas advanced after OPCs in the spinal cord had beenobserved in close association with radial glia (Hiranoand Goldman, 1988).
Secreted factors are also important for the migrationof OPCs. Secreted growth factors include basic fibro-blastic growth factor (FGF-2, also known as FGF-b),platelet-derived growth factor (PDGF), and epidermalgrowth factor (EGF). These molecules have been impli-cated in oligodendroglial proliferation, migration, dif-ferentiation, and survival (Calver et al., 1998; Fortinet al., 2005; Gonzalez-Perez et al., 2009). PDGF exertsa chemoattractive effect on migrating OPCs (Zhanget al., 2004); myelin defects in PDGF-A knockout ani-mals are most severe in the regions most distant fromthe periventricular germinal zones of the neural tube,suggesting that PDGF might, in fact, act as a long-rangestimulator of the migration of OPCs (Fruttiger et al.,1999b). EGF is involved in the proliferation and migra-tion of subventricular zone progenitors to produce oli-godendrocytes (Gonzalez-Perez et al., 2009) and FGFreceptor (FGFR) in neurons has been linked to N-cadherin signaling, which can enhance neuronal motilityand is involved in axon guidance (Derycke and Bracke,2004). Gene deletion studies show that FGF signaling iscrucial for OPCs to acquire a motile phenotype and, sub-sequently, for their migration (Osterhout et al., 1997).
The second group of secreted molecules reported toplay a role in the migration of OPCs are the chemotropic(or chemotactic) molecules. During the initial dispersion
LUK ET AL.
A
of OPCs from the ventral ventricular zone in the spinalcord, netrin-1 is a chemorepulsive mediator (Tsai et al.,2003). Analysis of netrin-1 and deleted in colorectal can-cer (DCC) deletion mutants confirm that the migrationof OPCs in the spinal cord from their ventral origin totheir final destinations is impaired (Jarjour et al.,2003). A clear stop signal for migrating OPCs in the spi-nal cord is the chemokine CXCL1 (Tsai et al., 2002).CXCL1 is expressed by white matter astrocytes duringthe period in which OPCs invade this structure (Milleret al., 1997; Robinson et al., 1998). This effect is rapidand reversible and is mediated via its receptor CXCR2,which is expressed by 85% of OPCs. In CXCR2 knock-out mice, where ventral-to-dorsal OPC migration is dis-rupted, even though axonal development is normal,spinal cord myelin is present only at the periphery ofthe structure, mirroring the distribution of OPCs andstrongly supporting the notion that the developmentof neurons and oligodendroglia would reflect twoindependent events (Tsai et al., 2002). The number of as-trocytes is also normal in these mutants, suggesting thatdefects in CXCL1/CXCR2 signaling selectively affectoligodendrocytes but not other cell populations in thenervous system.
DEVELOPMENT AND MATUR
Axonal pathfinding
There are several key regulators of axonal pathfinding.These include growth promoters, such as neurotrophinsand FGF, chemoattractants such as netrins 1 and 2, Ephreceptor tyrosine kinases and their ligands, extraneuro-nal adhesion molecules, such as laminin, fibronectin,tenascin, and some N-CAMs and neuronally expressedfactors such as N-cadherin and other N-CAMs. Whileseveral of these systems are also involved in cell migra-tion and have been discussed already, the major cell ad-hesion molecules have yet to be explored.
Cell adhesionmolecules arepresent inboth the extracel-lular matrix (ECM) and on neurons; through heterophilicand homophilic interactions, they communicate to guidegrowing axons along the proper developmental path.
Tenascin, laminin, and fibronectin are glycoproteinspresent in the ECM that are important for growth coneguidance during development. While tenascin acts as anorientation signal, laminin and fibronectin provide asubstrate upon which axonal elongation and NPC migra-tion can occur. Tenascins participate in different cellularprocesses, including cell adhesion and migration, andinterestingly, can either stimulate or inhibit cell migrationdepending on the isoform expressed, presumably viaepigenetic regulation. These large ECM glycoproteinsconsist of N-terminal cysteine-rich regions, followed byEGF-like segments, fibronectin-type III repeats and aC-terminal fibrinogen-like region (Chiquet-Ehrismann
and Chiquet, 2003). EGF-like segments have been shownto have axonal repulsive properties and splice variantsof the fibronectin-type III domain demonstrate bothgrowth repulsive and promoting qualities, dependingon the isoform expressed (tenascin-C contains distinctadhesive, antiadhesive, and neurite outgrowth promot-ing sites for neurons). Laminin has some involvementwith netrin-4 (Schneiders et al., 2007); along with fibro-nectin, it is often used as a substrate in cell culture ex-periments. Interestingly, a7 has been shown to mediatecell adhesion and migration via direct interaction withlaminin (Yao et al., 1996) and mice deficient in a7 showcompensatory upregulation of binding partner b1 integ-rin, with concomitant increases in nerve sprouting butdecreases in successful reinnervation following injury,indicating the close relationship between these factorsduring axonal adherence and pathfinding (Werneret al., 2000; Makwana et al., 2009).
The major receptors involved in contact-dependentaxonal growth are the b1 integrins, which recognizeN-CAM, N-cadherin and the L1 glycoprotein (Bixbyand Harris, 1991; Walsh and Doherty, 1997) and are thesurface receptors for fibronectin and laminin (Horwitzet al., 1985). Neurite outgrowth stimulated by N-CAM,N-cadherin and L1 is also dependent on the tyrosinekinase activity of the FGF receptor (FGFR) in neurons(Skaper et al., 2001), and interactions can be homophilic(Bixby andZhang, 1990) or heterophilic (Kuhn et al., 1991).
N-CAM is an immunoglobulin superfamily memberthat is expressed ubiquitously by nearly all neurons;its various isoforms are regulated through post-translational modification and splicing variation. Thereare three primary NCAM isoforms, NCAM 120, NCAM140 and NCAM 180, based on their different molecularweights (Goridis et al., 1983) and other post-translationalmodifications, such as polysialylation, can also affectNCAM function. Each isoform has a unique spatialand temporal organization that reflects differential de-velopmental patterning. NCAM 120 is primarily foundon glial cells. Through direct interaction with FGFR, itactivates the phospholipase C-g (PLCg) pathway, whichresults in a Ca2þ influx into neurons and an activationof protein kinase C (PKC) or the activation of themitogen-activated protein (MAP) kinases, extracellularsignal-regulated kinase (ERK)1 and 2 (Doherty andWalsh, 1996). NCAM 140 is present in both pre- andpostsynaptic growth cones, as well as muscle cells,and mediates axonal outgrowth via cell signalinginteractions (B€uttner et al., 2005). Lastly, NCAM 180is nervous system-specific and appears later in develop-ment, after neuronal cell migration. Expressed by post-synaptic membranes, NCAM 180 stabilizes cell–cellcontacts by association with the cytoskeleton linkerprotein spectrin (Pollerberg et al., 1985).
TION OF THE SPINAL CORD 9
Y
Neural or N-cadherin is a calcium-dependent adhe-sion molecule found in neurons. During embryogenesis,it is the key molecule involved in gastrulation and neuralcrest development and can activate pathways suchas FGFR-mediated tyrosine kinase signaling or RhoGTPase (Derycke and Bracke, 2004). N-cadherin hasbeen implicated in axonal growth, guidance, and plastic-ity (Doherty and Walsh, 1996). N-cadherin-mediatedaxonal outgrowth can be inhibited by FGFR blockingagents, such as dominant negative FGFR (Williamset al., 1994). Downstream of N-cadherin, Rho GTPases,such as RhoA, TC10, Cdc42 and Rac have been shown toregulate cytoskeletal structure, as well as cell adhesionand dynamics (Aepfelbacher et al., 1997; Murphyet al., 1999; Aspenstr€om et al., 2004; Begum et al.,2004; Coisy-Quivy et al., 2006; Benarroch, 2007), andthe slit-robo GTPase-activating protein 2 is a receptor-linked adaptor molecule for Rho GTPase (Maduraet al., 2004; Lin et al., 2005). Following nerve injury,these four proteins are upregulated, with TC10 showingthe greatest change (Tanabe et al., 2000). Inhibition ofRhoA, as well as overexpression of cyclin-dependentkinase inhibitor p21 (Cip1/WAF1), which directly inhibitsRho-kinase function by complex formation, have bothbeen shown to enhance CNS regeneration and axonalpathfinding both in vitro and in vivo (Dergham et al.,2002; Ellezam et al., 2002; Tanaka et al., 2002). Addi-tionally, studies show that enhanced expression of
10 G.W.J. HAWR
Fig. 1.6. Dorsal and ventral patterning. (A) Organizing centers imthe neural tube. Morphogen sources (arrow origins) and targets (ar
arrows. Shh is necessary and sufficient for ventral patterning, show
represses the transcription of class I homeodomain proteins while
induced by notch, and FGF, BMP and RA modulates its effects. FG
and the anterior neural ridge represses neuronal differentiation, pre
signals prematurely. RA released from paraxial mesoderm inhibi
differentiation. Additionally, RA released from motor neurons as
nearby and this plays a role in generating sharp spatial cutoffs in n
ulates neurite outgrowth and is a key factor in rostrocaudal specif
plate to form, and also induce Wnt which controls proliferation, s
rons. BMP inhibitors noggin, follistatin and chordin also play an
protein; FGF, fibroblast growth factor; RA, retinoic acid; Shh, so
p21, accomplished by upregulating nuclear localizedprotein p311, also increases regeneration and target rein-nervation followingperipheral injury (Fujitani et al., 2004).
Ventral patterning
Ventral patterning precedes dorsal patterning and is theprocess whereby specific motor neuron subtypes areproduced in localized progenitor domains of the ventralcord to form longitudinally oriented columns (Fig. 1.6).There is evidence that this is temporally regulated –motor neurons destined for the medial portion of thelateral motor column exit progenitor pools before motorneurons destined for the lateral portion of the samemotor column (Hollyday and Hamburger, 1977). Insightinto the mechanisms underlying ventral patterning camefrom the observation that animals with notochord dupli-cations formed floor plates in the region of both noto-chords while the floor plate failed to form when thenotochord was removed (Placzek, 1995). Indeed, furtherstudies demonstrated that the notochord and the floorplate it generates specify the identity and location ofmotor neuron cells via secretion of diffusible factors(Roelink et al., 1994).
Ectopic expression studies have demonstrated thatSonic hedgehog (Shh), the vertebrate ortholog of theDrosophila segment polarity gene hedgehog, is secretedfrom the notochord and is critical for induction of boththe floor plate and motor neuron specification (Roelink
LUK ET AL.
portant in dorsal and ventral patterning are shown in relation to
row heads) are denoted. Inhibition is denoted with blunt-ended
n in (B). Shh is released from the notochord and floor plates and
inducing transcription of class II proteins. Shh appears to be
F (in particular FGF3 and FGF8) emitted from Hensen’s node
venting progenitors from responding to dorsoventral patterning
ts this FGF-mediated repression and is required for neuronal
they form influences the specification of subsequent neurons
euron type. RA also inhibits neural crest cell migration, stim-
ication. In dorsal patterning, (C) BMP 4 and 7 induce the roof
pecification, migration, and axon guidance of dorsal interneu-
important role in dorsal patterning. BMP, bone morphogenic
nic hedgehog.
A
et al., 1994). The floor plate also secretes Shh after it isformed. Shh thus forms a mere 2–3-fold concentrationgradient across the developing spinal cord which is suf-ficient to generate five subclasses of ventral neurons.In general, Shh represses transcription of class I home-odomain transcription factors while inducing those ofclass II. Shh must act in two phases: firstly it primesthe ventral neural tube, committing it to a ventral fate(evidenced by Pax7 expression). Motor neuron produc-tion requires a second exposure to a 10-fold higher con-centration of Shh (Ericson et al., 1996).
Downstream, homeodomain proteins such as Nkx2.2,Nkx6.1, and Irx3 play an important role in generatingmotor neurons in the basal plates (Briscoe et al., 2000;Sander et al., 2000). Nkx6.1 induces the expression oftranscription factors essential for motor neuron specifi-cation, including Olig2 and MNR2, while Nkx2.2 andIrx3 ensure that motor neuron induction is not initiatedoutside of appropriate ventral and dorsal regions respec-tively (Tanabe et al., 1998; Novitch et al., 2001). Olig2,which is induced by RA, represses Irx3 in order to main-tain the motor neuron potential of progenitors. MNR2is a transcription factor that acts as a dedicated determi-nant of motor neuron identity.
At around e9.5 in mice, shortly after their generation,spinal motor neurons express transcriptional regulatorssuch as Hb9, Lhx3, Isl1, and Isl2, which aid in furthersubtype diversification, as well as axonal projection out-side the spinal cord and acetylcholine release (Pfaffet al., 1996; Sharma et al., 1998; Arber et al., 1999; Thaleret al., 1999, 2004; Dasen and Jessell, 2009).
DEVELOPMENT AND MATUR
Dorsal patterning
Just as the floor plate is important in ventral patterning,the roof plate serves as an important organizing centerfor dorsal patterning, along with the dorsal epidermis.The roof plate forms in the dorsal midline of thedeveloping spinal cord shortly after neural tube closureas a result of induction byBMP4 andBMP7, both secretedby the dorsal epidermis. Roof plate progenitors as well asmature roof plate cells express the LIM-homeodomaintranscription factors Lmx1a/b, which is necessary and suf-ficient for roof plate induction (Chizhikov and Millen,2004; Nakatani et al., 2010). In an autocrine, positive feed-back loop, BMP4 and BMP7 trigger their own productionfrom the roof plate and also stimulate production ofwingless-related mouse mammary tumor virus integra-tion site proteins (Wnt) that are important in controllingproliferation, specification,migration, and axonguidanceof adjacent dorsal interneurons.
Ablation studies suggest that the roof plate organiz-ing center is important for specification of the threedorsal-most types of interneurons. It produces BMPr1a
which promotes cellular proliferation and also inducesexpression of BMPr1b which triggers differentiation.Subsequently BMP7 expressed by the roof plate playsa role in axon guidance via repulsion (Chizhikov andMillen, 2005). Not surprisingly, the BMP inhibitors nog-gin and follistatin previously discussed also play an im-portant role in dorsal patterning; evidence also suggeststhat there are additional morphogens yet to be described.
TION OF THE SPINAL CORD 11
Longitudinal column formation
An important aspect of patterning is the grouping ofneurons with common functions into longitudinal col-umns. Signals from the paraxial mesoderm, in particularRA and FGF, appear critical for inducing appropriatecolumns at appropriate rostrocaudal cord levels. For in-stance, RA signals are responsible for the localization ofnuclei subserving autonomic function being localized tothe thoracic and upper lumbar regions, and for restrict-ing those supplying the limbs to cervical and lumbarregions. Graded FGF signaling is involved in brachial,thoracic, and lumbar motor neuron patterning.
A number of columns and subcolumns thus form,which can be identified by their unique patterns of home-odomain transcription factor expression – specificallyIsl-2, Isl-1, Lim-1 and Lim-3. Axial muscles are innervatedby neurons located in a medial motor column subcolumn,called the MMCm, which express Isl-1, Isl-2 and Lim-3.Neurons that project to body wall muscles are foundmore laterally in the MMCl and express both Isl-2and Lim-1. Neurons that project to limb muscles arefound in the lateral motor column (LMC), express Isl-2and Isl-1, and are subdivided into the LMCl and LMCmwhich project to dorsal and ventral limb muscles respec-tively. As a result of rostrocaudal patterning, and asexpected, LMC motor neurons are present only at limblevels, andMMCl motor neurons present only at thoraciclevels. Additionally, preganglionic autonomic motor neu-rons of the column of Terni express the Isl-1 transcriptionfactor and are present only at thoracic levels.
Rostrocaudal formation is highly affected by Hoxtranscription factor regulation. Retinoids, FGF, Wnt,and TGFb superfamily members act in concert to regu-late Hox gene expression (Liu et al., 2001; Bel-Vialaret al., 2002; Diez del Corral and Storey, 2004; Liu,2006; Nordstr€om et al., 2006), which has been linked tospinal cord patterning within tightly regulated 30–50
Hox gene clusters. Differential expression of FGFfrom high to low down a 50–30 concentration gradientis involved in the initial Hox gene induction at brachial,thoracic and lumbar spinal cord levels (Liu et al., 2001;Bel-Vialar et al., 2002; Dasen et al., 2003), with Hox4–8 paralogs expressed at brachial levels, Hox 8–9
Y
expressed at thoracic, and Hox 10–13 expressed in thelumbar spinal cord (Dasen and Jessell, 2009).
In contrast to FGF-mediated initiation of Hox geneexpression, within-cluster regulation of Hox genes is ac-complished by inhibitory RA signaling. By an unknownmechanism, retinoids are known to antagonize the FGFgradient, particularly at brachial levels (Liu et al., 2001;Diez del Corral and Storey, 2004). At more caudal levels,the TGFb superfamily member GDF11 can regulate Hox8–10 gene expression in the thoracic and lumbar spinalcord (McPherron et al., 1999; Liu, 2006).
12 G.W.J. HAWR
Cellular differentiation
In mammals, the neural tube initially consists of a singlelayer of multipotent cells lining the central canal. Thislayer, referred to as the ventricular zone, gives rise toboth neurons and glia, typically in an inside-out se-quence. Neuroblasts, which give rise to neurons, arethe first cell type to form. Once neuroblasts are formedthey lose their ability to divide. Glial cells are formedfrom glioblasts after the formation of neuroblasts hasceased. They migrate from the neuroepithelial layer tothe mantle andmarginal layers. Once production of glio-blasts has ceased, the neuroepithelial layer differentiatesinto the ependymal cells which line the central canal.
Important exogenous switches from neurogenesis togliogenesis appear to be cytokines from the IL-6 familywhich activate the JAK STAT3 pathway, BMP2/4, andcardiotrophin-1 (Koblar et al., 1998; Barnabe-Heideret al., 2005). Intrinsically, the switch from neurogenesisto gliogenesis involves both the attenuation of neuro-genic genes such as the neurogenin transcription factors(Sun et al., 2001) and the activation of pro-glial genes,such as the nuclear factors 1A and 1B (NFIA/B)(Deneen et al., 2006). NFIA appears to be a key mole-cule, as it inhibits neurogenesis and induces the forma-tion of astrocytes. NFIA expression is induced by notchsignaling from neurons and its inhibition appears to becritical for the generation of oligodendrocytes. How-ever, there is a direct interaction between NFIA andOlig2, a key transcription factor in the oligodendrocytelineage (Okano and Temple, 2009). It is also noteworthythat BMP can shift oligodendrocyte precursors orpre-progenitors into the astrocyte lineage (Mabieet al., 1997; Grinspan et al., 2000).
Epigenetic modifications (chromatin remodeling andDNA methylation of glia-specific genes) also appear tobe critical in cell fate specification, particularly in termsof specifying competence to respond to extracellularsignals (Takizawa et al., 2001; Namihira et al., 2008).For instance, in the brain it has been shown that methyl-ation of the astrocyte-specific promoter GFAP is criticalfor astrocytic differentiation. As well, NFIA has been
shown to demethylate astrocyte promoters (Mizutaniet al., 2007; Namihira et al., 2009). Likewise, NCAMis involved in epigenetic regulation of oligodendroglialcell fate in the developing nervous system. Axonsexpress the polysialylated (PSA) form of NCAM duringdevelopment, prior to myelination (Jakovcevski et al.,2007). Sialylation at multiple sites reduces homophilicinteractions between NCAMs, due to the negativecharge and/or hydration volume of the PSA (Kleeneand Schachner, 2004). PSA-NCAM also mediates het-erophilic interactions with other glycans, such as theheparan sulfate proteoglycans expressed on OPCs(Winkler et al., 2002). As a negative regulator of myelinformation, PSA-NCAM levels decline at the onsetof myelination and myelination only occurs on PSA-NCAM axons. Its premature removal from neuronsin vitro enhances differentiation and myelination by4–5-fold, while preserving oligodendrocyte cell number(Charles et al., 2000), further illustrating its inhibitoryrole in myelin formation by epigenetic regulation.
LUK ET AL.
Oligodendrocytes and myelinationpatterning
Myelination of the spinal cord is critical to its function. Inthe central nervous system, oligodendrocytes myelinatenumerous axons; this is in contrast to the peripheral ner-vous system, where Schwann cells only myelinate a singleaxon. Oligodendrocyte progenitors in spinal cord arise inthe ventral ventricular zone adjacent to the floor plate(Ono et al., 1995) in response to Shh, which inducesneuroepithelial cells to express the specifying transcrip-tion factors Olig1 and Olig2 (Poncet et al., 1996; Orentaset al., 1999).
Olig1 and 2 are helix-loop-helix (HLH) transcriptionfactors, expressed in developmental and adult OPCs,as well as mature oligodendrocytes (Zhou et al., 2000;Ligon et al., 2006). Olig2 is primarily responsible for celldevelopment into the oligodendroglial lineage and inhi-bition of motor neuron lineage (Lu et al., 2002), whereasOlig1 functions later in development, during oligoden-droglial maturation from OPCs (Xin et al., 2005). Olig1regulates OPC differentiation by upregulating myelin-associated genes, such as proteolipid protein (PLP)and myelin basic protein (MBP) and myelin-associatedglycoprotein (MAG), as well as by suppressing glialfibrillary acidic protein (GFAP), an astrocytic marker(Xin et al., 2005; Li et al., 2007). Olig2 regulatesSox10 and NKX2.2 in sequence, during a dosage-dependent developmental process (Liu et al., 2007).GPR17, a G protein-coupled orphan receptor, is the firstcharacterized negative regulator of oligodendrocyte dif-ferentiation (Chen et al., 2009). Expressed during earlyoligodendrocyte lineage, GPR17 is downregulated in
A
mature myelinating oligodendrocytes and upregulatedduring demyelination in both murine and human modelsof multiple sclerosis (MS) and it is activated by nucleo-tides or inflammatory stimuli (Ciana et al., 2006).Once specified, oligodendrocytes progress through fivedevelopmental stages (pre-progenitor, precursor, pro-oligodendroblast, immature oligodendrocyte, and matureoligodendrocyte) before they can myelinate (Fig. 1.7).
To form mature myelinating oligodendrocytes, pro-genitor cells must be specified, then proliferate, and mi-grate from the germinal zone into the white matter wherethey differentiate. The maturation of OPCs into function-ally myelinating oligodendrocytes is multivariably andtemporally regulated.
DEVELOPMENT AND MATUR
Pre-progenitor Precursor Pro-Oligodendrobl
04A2B5A2B5E–NCAM
VimentinNestinGD3NG2
A
B
C
D
PDGFR µVimentin
Mar
kers
Pro
cess
es
fact
ors
Tra
nscr
iptio
n
Nestin
Specification–Shh
Proliferation
Differ
Survival
Olig1, Olig2
Sox10
MyT1
PDGF, FGF, neurequlin, IGF-1
PDGF, FGF, Neurequlin, IGF-1, NT-3, CNTF, LIF
PDGFR µGD3NG2
Fig. 1.7. Oligodendrocyte development – key stages and molec
stages from pre-progenitor to mature oligodendrocyte (A). Distiby the cells at each stage facilitating their identification (B). ProceMolecules which have been shown to play key roles in oligodend
are shown in (C) in relation to the developmental stages at which
processes are shown in (D). (From Grinspan, 2002. Used with pe
PDGF-A is an important factor in this process. It issecreted by astrocytes and neurons and acts as an oligo-dendroglial mitogen. When used in vitro, in combinationwith thyroid hormone, OPC progeny of a single clonegrown separately showed similar cycles of cell divisionbefore terminal differentiation – usually 6–8 (Templeand Raff, 1986; Durand and Raff, 2000). This was initiallyattributed to intrinsicmonitoring of cell cycle numbers, butfurther study suggests OPCsmonitor time and not numberof cell cycles (Gao and Raff, 1997). Although the internalmechanismbywhichOPCsmonitor progenitor duration ispoorly understood, it involves thyroid hormone receptor-a1 activation, and several cyclin-dependent kinases,inhibitors, andmitotic regulatory proteins influence both
TION OF THE SPINAL CORD 13
astImmatureOligodendrocyte
MatureOligodendrocyte
GaIC04CNP
GaIC04CNP
entiation
Nkx2.2
GTX
TH, Notch-jagged, suppression of Ids
MBP, PLP, MAG, MOG
ules. Oligodendrocytes progress through five developmental
nct patterns of immunohistochemical markers are expressed
ss length and complexity increase as oligodendrocytes mature.
rocyte specification, proliferation, differentiation and survival
they occur. Putative transcription factors important for these
rmission.)
Y
proliferation and differentiation (Durand et al., 1998;Tokumoto et al., 2001, 2002; Dugas et al., 2007). PDGFalso enhances differentiation andmigration of oligoden-drocyte progenitors, and increases their survival(Grinspan and Franceschini, 1995; Fruttiger et al.,1999a; Grinspan, 2002; Klinghoffer et al., 2002). Thisis evidenced by mice engineered to overexpress PDGF-A in neurons which showed a 7-fold increase in OP num-bers (Calver et al., 1998), aswell as studies demonstratingmarked oligodendrocyte loss in the absence of PDGF-A(Barres et al., 1992; Fruttiger et al., 1999a). Further fac-tors, such as insulin-like growth factor 1 (IGF-1), can alsoenhance proliferation and oligodendrocyte differentia-tion from OPCs in vitro (Mozell and McMorris, 1991).
Mature oligodendrocytes express abundant myelinproteins such as PLP and MBP. PLP provides structuralsupport for the myelin membrane, but has additionallydemonstrated a critical role in oligodendrocyte differen-tiation (Yang and Skoff, 1997). Mutations in PLP areoften lethal and result not only in lack of myelinationbut also premature apoptosis of oligodendrocytes(Grinspan et al., 1998). MBP plays a major role in centralmyelin compaction and mutations result in severe dys-myelination in the central nervous system (Shine, 1992).The shiverer mouse mutant has an autosomal recessivemutation leading to loss of all MBP isoforms and almosttotally lacks central myelin. These animals have tremors(shivering), convulsions, and early death; they are anexcellent model organism for cell replacement therapieswhich aim to remyelinate.
In humans spinal cord myelination begins in thefourth fetal month; however, some tracts are not fullymyelinated until the first year of postnatal life. Thecorticospinal tract is slowest to myelinate; it is not fullymyelinated until 2–3 years after birth.
Motor neuron pools
Within columns, motor neurons which innervate thesame muscle form clusters known as motor neuronpools. Neurons within a single pool are electricallycoupled, presumably by gap junctions. Relatively littleis known about how these pools form, though evidencecurrently suggests that motor neuron pools are specifiedprior to muscle innervation and that neurotrophins, inparticular glial cell line-derived neurotrophic factor(GDNF), play a supportive role in the formation of thesepools (Price and Briscoe, 2004).
Development of descending pathways
Although the corticospinal tract is of proven importancefor motor functions, it is often ascribed an importancewhich exceeds its proven role, especially in animalmodels (Midha et al., 1987). Nonetheless, it is the best
14 G.W.J. HAWR
described descending tract, and much is known aboutits development. In humans, it is one of the latest des-cending pathways to fully develop. It has recently beenshown that each neuron’s pattern of homeodomain pro-tein expression programs a neuron to respond appropri-ately to environmental signals required to reach theirdistinct muscle targets (Price and Briscoe, 2004; tenDonkelaar et al., 2004).
Attractive and repulsive signals are required to directthe growth cones of corticospinal axons through the inter-nal capsule, cerebral peduncle, pons and medulla to reachtheir caudal targets. These signals are both diffusible andcontact-mediated. As well, guideposts, or intermediatetargets along the axon’s path, are of critical importanceto correct targeting.The corticospinal tract originates fromneurons in cortical layer 5 and initial outgrowth of its axonsinto thewhitematter appears to be initiatedby semaphorinswhile the chemoattractant netrin-1 directs axons towardsthe internal capsule (tenDonkelaaret al., 2004). Slit2guidesaxons into the cerebral peduncle (ten Donkelaar et al.,2004). Upon growth into the spinal cord, it is believed thatrepulsive cues in the cuneate and gracile fascicles, perhapsfrom their already present myelin, keep the corticospinaltract appropriately located and compacted (Price andBriscoe, 2004; ten Donkelaar et al., 2004).
The formation of the pyramidal decussation is aninteresting aspect of corticospinal tract development.A vimentin raphe forms a barrier in the midline of thespinal cord and hindbrain but is absent in the regionof the pyramidal decussation facilitating fiber crossing.Here L1CAM and netrin-1 also appear to play an impor-tant role in guiding axons through the decussation.Ephrin-B3 expressed in the midline repels axons whichexpress the EphA4 receptor and appears to prevent cor-ticospinal fibers from re-crossing, evidenced by ephrin-B3 mutants which have bilateral innervation of distaltargets. It is thus interesting that animals with ephrin-B3 mutations do not demonstrate normal left-rightalternation in their gait, but instead hop like a rabbit(Coonan et al., 2001; Kullander et al., 2001; Kiehn andButt, 2003), as a result of developmental reconfigurationof the locomotor central pattern generator (Kullanderet al., 2003). Despite these measures to ensure contralat-eral innervation, electrophysiological studies demonstratethat human muscles receive bilateral innervation for atime after birth, however the ipsilateral connections arelargely pruned (Dottori et al., 1998; Eyre, 2007). Nonethe-less, some bilateral innervation in humans is not unusual.Anomalies of the decussation of the pyramidal tract arefrequently found in posterior fossa malformations suchas occipital encephaloceles, the Dandy–Walker malfor-mation, Joubert syndrome, and in cases with extensivemalformations of the brainstem such as M€obius syn-drome (Lagger, 1979).
LUK ET AL.
A
At birth, too many corticospinal neurons exist – eventhe occipital lobes contribute fibers. Furthermore, axonsthat participate in the corticospinal tract typically elabo-rate a subcortical branch in addition to their spinalprojection. The supernumerary axons (in particular,those that have failed to make appropriate connections)and the subcortical branch are also pruned over the first3 postnatal months. Otx1 appears to play an importantrole in the pruning process, evidenced by reeler(D’Arcangelo et al., 1995) and yotari mice (Yoneshimaet al., 1997), which are mutants of this gene.
DEVELOPMENT AND MATUR
Spinal nerves
Ultimately, to innervate muscles and glands, impulsesfrom descending tracts must synapse on lower motorneurons and exit the spinal cord. Signals from the somitestrigger the formation of peripheral nerves in the fourthweek of development. Ablation studies have demon-strated that the dorsal aspect of the somite plays animportant role in segmental formation of dorsal rootganglia while spinal axon segmentation is disrupted withablation of the anterior somite; however, the inductivesignals are not known at this time. Peripheral nerve rootspreferentially traverse the rostral half of each sclerotomeand fail to sprout if this region is removed (Keynes andStern, 1984).
After leaving the spinal cord, axons innervate theirdistal targets by making a series of binary choices atguideposts on the way to their target muscle; the appro-priate response to these binary choices is encoded ineach neuron’s homeodomain expression pattern (Priceand Briscoe, 2004; ten Donkelaar et al., 2004). The firstbinary choice sees motor axons project either dorsally,towards axial musculature, or ventrally, towards mus-cles of the body wall and limbs. In LMC axons whichproject to the limbs, the next binary choice is at the base
Fig. 1.8. Spina bifida variants. Spina bidifa is a general term appli
subtle as in spina bifida occulta, where the posterior vertebral arch i
may be overlying hypertrichosis (A). In meningocele (B) and myel
the arachnoid herniates out the defect. Neural tissue is also herni
herniation of these structures is also described. In rachischisis (D),tissue is not covered by skin.
of the limb, where they must project to either dorsally orventrally derived limb muscles. The lateral LMC (LMCl)subcolumn motor neurons are directed to dorsallyderived (largely extensor) muscles while those of themedial subcolumn (LMCm) motor neurons project toventrally derived (largely flexor) muscles. It appearsthat EphA receptor tyrosine kinases and their ligands,the ephrin As, may have a repulsive effect importantfor directing appropriate axons to the dorsal limbmesenchyme (Price and Briscoe, 2004).
TION OF THE SPINAL CORD 15
SPINAL CORDMALDEVELOPMENT
Just as recent advances in molecular biology are helpingto define the molecular underpinnings of normal devel-opment, so too are the causes of maldevelopment beingelucidated. This active area of research is challenging tostudy and remains poorly understood. The ontogeny istypically complex and mutant animals are often of mar-ginal viability (Harris and Juriloff, 1999). Here too,though, patterns and general principles are emerging.Interestingly, while similar etiologies now seem to ex-plain malformations that were thought to be unrelated,others that were suspected to have a common origin (inparticular spina bifida aperta and occulta) now appear tohave distinct causes. It is also fascinating to considerthat many abnormalities likely occur that do not leadto functional deficits or imaging findings and thus goundetected and undescribed.
Neural tube defects
Neural tube defects (NTDs) are malformations involv-ing the posterior vertebrae and possibly the neural ele-ments (Fig. 1.8). NTDs are classified as open or closedand are most commonly seen in the lumbosacral regionof the spine. Closed defects, termed spina bifida occulta
ed to malformations of the posterior spine. The defects may be
s missing and the only clue to the existence of such an anomaly
omeningocele (C), the overlying skin remains intact; however,
ated in myelomeningocele but not in a meningocele. Ventral
a more severe anomaly, the neural tube fails to fold and neural
Y
(SBO), are common, seen in perhaps 20–30% of thepopulation (Warder, 2001). As its name implies, SBO istypically asymptomatic and usually involves a defect ofthe vertebral arches without involvement of the underly-ing neural tissue. NTDs are often associated with cutane-ous stigmata such as an overlying hairy patch. Moresevere open defects, interchangeably termed spina bifidacystica or spina bifida aperta (SBA), involve exposure ofneural elements to damaging amniotic fluid in utero andparalysis and infection risk postpartum. Open NTDs arenot only an important cause of mortality and morbidity,but are also an important public health problem. The med-ical costs to society are about $200 million per year withtotal costs of about $250000 per person per lifetime(CDC, 1989). Fortunately open defects are comparativelyrare, seen in only 1:1000 newborns; however, the incidencevaries among different populations. The incidence is highin Ireland (1:200) while rates as low as 1:10000 are reportedin those of African descent. Of interest, different popula-tions exhibit propensities for different rostrocaudal levelsof involvement and exhibit differential effects on cogni-tive development. As well, while anencephaly has a femalepredominance, spina bifida has an equal sex distribution.Besides folic acid deficiency, numerous other conditionshave been associated with NTDs, and are presented inTable 1.1. It is important to note, however, that NTDsare multifactorial in origin (Mitchell et al., 2004).
Two subclasses of spina bifida include meningoceleand myelomeningocele. In a meningocele, the least fre-quent neural tube defect, only fluid-filled meninges pene-trate the defect and patients are typically neurologicallyintact. When neural tissue is included in the defect it is
16 G.W.J. HAWR
Table 1.1
Risk factors for neural tube defects
Established risk factor Relative risk
History of previous affected pregnancywith same partner
30
Failure to consume folic acid supplements/inadequate maternal intake of folic acid
referred to as myelomeningocele. A more severe defect,rachischisis, occurs when the neural tube fails to fold.
LUK ET AL.
NTDS AND FOLIC ACID
NTD prevention is unquestionably superior to surgicalrepair and we have been fortunate to have seen a revo-lution in this regard. While the possibility of maternalfolate deficiency as a cause of NTDs was hypothesizedas early as 1964 (Hibbard, 1964; Ausman and Slavin,1995), it took the British Medical Research CouncilVitamin Study Research Group to cement this associa-tion with a double-blind randomized trial, which demon-strated a remarkable 72% reduction in the occurrence ofNTDs with folic acid supplementation (MRC VitaminStudy Research Group, 1991). This was subsequentlyreproduced in a number of other trials (Laurenceet al., 1981; Smithells et al., 1983; Mulinare et al., 1988;Bower and Stanley, 1989; Milunsky et al., 1989; Vergelet al., 1990). This has led to the recommendation, firstput forth in 1992 (CDC, 1992), that women should re-ceive supplementary folic acid (0.4 mg daily) 1 monthbefore conception through 3 months post-conception.It is recommended that women with a previously af-fected pregnancy instead ingest 10� that dose (4 mgdaily) as they are at 4–5� higher risk of this defect infuture pregnancies (CDC, 2004).
It is still unclear, however, how folic acid contributesto this problem (Ausman and Slavin, 1995). Folic acid isalso known as vitamin B9 and its metabolically activeform, folinic acid (Harris and Juriloff, 1999), plays animportant role in nucleic acid synthesis as well as
Substances reducing the incidence of neural tube defects
when provided as maternal dietary supplementation
Cause Mutation or mutant
Methionine AxdFolic acid or thymidine SpInositol Curly tail
(Adapted from Harris and Juriloff, 1999)
ATION OF THE SPINAL CORD 17
methylation reactions by contributing to the generationof S-adenosylmethionine (SAM) from homocysteine(Blom and Shaw, 2006). As an intermediate step in thisconversion, methyltetrahydrofolate, which requiresfolic acid as a cofactor, turns homocysteine into methi-onine that can be used to generate SAM. SAM can beused to methylate DNA, RNA and proteins; methylatedDNA is generally silenced. Vitamin B12 has been identi-fied as another cofactor whose deficiency is associatedwith NTDs (Mitchell et al., 2004) which is not surprisinggiven that it is a cofactor in the synthesis of both methi-onine and folic acid (Harris and Juriloff, 1999). Interest-ingly, in splotch mutant mice methionine increasessusceptibility to spina bifida (Harris and Juriloff, 1999)and in the Folr1-null mouse, folate supplementationmerely serves to move some embryos from the early le-thal phenotype to survivors with NTDs, both of whichhighlight the complexity of this pathophysiology. As well,high homocysteine levels which may occur with folate de-ficiency may be directly toxic, leading to NTDs (Millset al., 1996).
Increasing evidence suggests that folic acid deficiencyleads to NTDs at least in part because of deficient meth-ylation. Here it is critical to note that with the exception ofthe Folr1 mouse (Spiegelstein et al., 2004), null mutantsfor genes involved in the folate pathway (Cbs, Folr2,Mthfr, Mtr) do not have NTDs (Piedrahita et al., 1999;Chen et al., 2001; Swanson et al., 2001) while mutantsof several genes that contribute to methylation of thegenome are associated with this risk, including: Cecr2,Dnmt3b, Dnmt3l, Gtf2i, Hdac4, Sirt1, Smarca4, andSmarcc1.Dnmt3b, needed for de novoDNAmethylationin the elevating cranial neural folds, causes exencephalyin almost all embryos (Okano et al., 1999). Further sup-port for this hypothesis comes from the female predom-inance of anencephaly, which some have postulatedrelates to their increased demand for methyl groups forsilencing X-chromosomes (for example, the formationof Barr bodies).
It is also interesting, however, that methionine con-trols localization of actin and ab-tubulin in neuroepithe-lial cells, and when it is deficient the cells become roundrather than columnar (Moephuli et al., 1997). This couldprevent neuroepithelial cells from forming the pyrami-dal shape important to neural fold elevation.
Despite the importance of folic acid in human NTDs,of the mouse NTD models tested for folic acid responsefollowing maternal supplementation, only five show adecrease in NTD frequency (Cart1, Cited2, and Folr1null mutants, the Cd mutant of Lrp6, and the Sp andSp2H mutants of Pax3). In these animals, the dose offolate appears to be important (Piedrahita et al., 1999;Spiegelstein et al., 2004). Five mutants (Axd, ct,Grhl3-null, Efna5-null, and Map3k4-null) and the
DEVELOPMENT AND MATUR
SELH/Bc strain (which exhibits high frequency of non-syndromic genetically multifactorial exencephaly)showed no folate response, though three of these didshow a strong response to other maternal nutrientsupplementation including methionine in Axd models(Essien, 1992), inositol in ct mutants (Greene andCopp, 1997; Cogram et al., 2002), and an uncertain com-ponent of the Purina 5015 diet in SELH/Bc strains(Harris and Juriloff, 2005), raising hope that otherdietary supplements may have benefit in humans. Thesedata are illustrated in Table 1.2.
MOUSE MODELS OF NTDS
Important progress is being made in our understandingof genetic causes of NTDs. Nearly 200 mutant mousestrains that develop neural tube defects are now avail-able (Harris and Juriloff, 1999, 2007). Analysis of thesemutants has made it clear that a simple, or even moder-ately complex genetic explanation of NTDs will not bepossible (Harris and Juriloff, 2007). First, it is apparentthat any marginally viable embryo can develop NTDs inthe absence of specific defects that directly impair neu-ral tube closure. As well, some mutant mice only de-velop defects in isolated regions of the neural tubewhile others seem at risk for deficits in any region(Harris and Juriloff, 2007). Furthermore, even differentalterations in the same gene can lead to distinct patternsof abnormality. For instance, Lrp6 hypomorphs lead toonly spina bifida, while hypermorphs exhibit only exen-cephaly (Kokubu et al., 2004; Carter et al., 2005). Evenwhen the function of disrupted genes is known it can beextremely challenging to understand why NTDs occur(Harris and Juriloff, 1999), and furthermore there ap-pear to be important interspecies differences in genefunction. For instance gene defects associated with onlyspina bifida in humans cause only exencephaly in mice(Harris and Juriloff, 2007), which likely relates at leastin part to anatomical differences between the species.
Despite the extreme complexity of this picture, anumber of common themes are emerging amongsingle-gene defects recognized as contributing to NTDs.
Y
Many of these defects can be grouped into those leadinginterfering with cell polarity, apoptosis, actin regulation,cell–cell interaction, intracellular protein transport, andsignaling or transcription (Harris and Juriloff, 2007).These data are illustrated in Tables 1.3 and 1.4. Indeed,most mutations alter the anatomy of the neural tube,resulting in abnormal neural fold elevation rather thanfusion failure, with a few notable exceptions such asthe ephrin mutant Efna which fails to fuse and theFkbp8 mutant whose neural tube opens around gesta-tional day 18 after initial closure 8 days earlier(Holmberg et al., 2000; Finnell et al., 2006).
It is understandable that planar cell polarity gene mu-tations have a high rate of NTDs. These genes are re-sponsible for convergent extension and elongation ofthe neural plate. In homozygous mutants for thesegenes, the neural plate remains broad and the neuralfolds are unable to meet and consequently fail to fuse,often along the entire length of the neural tube (Coppet al., 2003; Doudney and Stanier, 2005). With the
18 G.W.J. HAWR
Table 1.3
Tissue causes for failure of neural fold elevation
CauseMutation ormutant
Slow growth of adjacent, tethered tissue Curly tailDefective forebrain mesenchyme Cart1, twistDefective basal lamina in surface ectoderm Lama5Excessive breadth of floor plate andnotochord
Lp
Abnormal neuroepithelium ApoB, Sp,Tcfap2a
Morphological deformation of theneural folds
jmj
Abnormal neuroepithelial and neural crestcell gap-junction communication
Gja1
Incomplete compensation for a defective
step in neural fold elevation
SELH/Bc
(Adapted from Harris and Juriloff, 1999)
Table 1.4
Biochemical causes for failure of neural fold elevation
Cause Mutation or mutant
Faulty regulation of apoptosis Trp53, p300Premature differentiation Hes1Disruption of actin function Macs, MlpAbnormal telomerase complex TercFaulty pyrimidine synthesis Sp
(Adapted from Harris and Juriloff, 1999)
exception of the Ptch1mutant which affects Shh signal-ing, all mouse craniorachischisis known to date is causedby cell polarity mutations (Greene et al., 1998; Milenk-ovic et al., 1999).
Numerous genes involved with apoptosis and the cellcycle have cranial NTDs but only two of these (Traf4 andfog of the Apaf1 pathway) have caudal NTDs (Harris andJuriloff, 2007). NTDs can be associated with either exces-sive or deficient apoptosis and it is easy to understandhoweither could alter the ability of the neural folds tomeet.
Normal actin function appears to be required for cra-nial (but not caudal) neural tube closure in mice (Copp,2005). This likely relates to actin’s role in contracting theapices of cells in the neural fold to make them pyrami-dal; however, it is unclear why mutations lead to defectsonly in the cranial region.
It is logical to screen human NTD cases for mutationsin genes complementing those of mouse models(Doudney and Stanier, 2005). Surprisingly, only threemurine NTD mutations appear to cause NTDs inhumans: p53 mutation homozygotes, curly tail (ct) andSELH/Bc have human counterparts (Harris and Juriloff,1999; Boyles et al., 2005). The validity of the mousemutants as models of human NTDs is thus uncertain,although their analysis has been very informative.
LUK ET AL.
NTD INHERITANCE
Most of the NTDs described above are loss of functionsingle-gene mutations. A number of other patterns ofinheritance of genetic NTDs have been described. Ofnote, while not yet described in humans, gain of func-tion mutations have been described in the mouse(Zhang et al., 2006). These gene products are thus re-ferred to as genotoxins. Another model, referred to asoligogenic, notes stepwise increase in risk for NTD withmutations in four genes, namely Exen1—4 mutations(Juriloff and Hoscheit, 2006). A similar but distinctmechanism involves modifier mutations. These havebeen described in relation to the ctmutant, where muta-tions in three other genes are unable to cause defects ontheir own, but are each capable of exacerbating ct muta-tions (Neumann et al., 1994; Letts et al., 1995). Most be-lieve that multifactorial inheritance is the mechanismunderlying the majority of human NTDs. Here it is be-lieved that numerous insults combine, but only lead toa defect if a “threshold” is reached. Indeed, mathematicalanalysis supports the multifactorial threshold hypothesisin humans (Carter, 1969, 1974) over the alternate possibil-ity of single-gene mutations with low penetrance(Lippman-Hand et al., 1978; Toriello and Higgins, 1983;Czeizel and Metneki, 1984; Hunter, 1984; Koch andFuhrmann, 1984), given the large drop-off in risk forNTDs from first- to second-degree relatives of probands
A
witha largerdecrease inriskfor third-degreerelativeswhichis still elevated from population baseline (Carter, 1969).
Spina bifida occulta
SBO is characterized by deficiency of the posterior verte-bral arch and arises after neural tube closure (Harrisand Juriloff, 1999) as a result of deficient mesodermalmigration making its ontogeny distinct from SBA (Payneet al., 1997). Indeed, the few SBO mouse mutants de-scribed (Hollander, 1976; Park et al., 1989; Payne et al.,1997) do not give rise to offspring with SBA with fewexceptions (Harris and Juriloff, 2007), reviewed inTable 1.5. Inhumansfamilial occurrencehasbeendescribed(ThompsonandMcKay, 1986). In addition to its associationwith cutaneous anomalies suchashairy patches, subcutane-ous lipomas and capillary hemangiomas (Guggisberg et al.,2004), SBO may also be associated with polythelia(Panigrahi et al., 2008), testicular cancer (Agostini et al.,1991), and idiopathic or symptomatic epilepsy (Klepel andFreitag, 1992).
Lipomyelomeningocele
Lipomyelomeningocele refers to spinal dysraphism withlipoma, and is a common form of dysraphism (Fig. 1.9C).It isaconcernprincipallybecause it formsa tethering lesion.
T (Tc/þ mutant) Transcription �Traf4 Apoptosis þZic1 Transcription �
(Harris and Juriloff, 2007)
AE, anencephaly; SBA, spina bifida aperta.
Unlike more usual intradural lipomas, those associatedwith incomplete neurulation blend with the spinal cord,making their surgical management challenging. LikeSBO, these lesions are often associated with cutaneousstigmata.
The embryology underlying these lesions is poorly un-derstood, though one theory posits that in these defectsthe cutaneous ectoderm closes slightly before the under-lying neural tube and paraxial mesenchyme migrates intothe gap in the neural tube. The reason this mesenchymeforms primarily fat is believed to be because of inductiveinfluences from the ependymal surface of the placode(Park and Scott, 2003a). It is interesting, however, thatfolic acid supplementation does not appear to reducethe rate of lipomyelomeningocele, suggesting that its eti-ology may differ from that of other NTDs (McNeely andHowes, 2004). Familial forms of lipomyelomeningoceleare reported but are rare (Seeds and Powers, 1988); thiscondition appears to be polygenic and exhibits autosomalrecessive inheritance (Kannu et al., 2005).
TION OF THE SPINAL CORD 19
SPLIT CORD MALFORMATIONS AND ANOMALIES OF THE
NEURENTERIC CANAL
The neurenteric canal, which transiently connects the yolksac with the amniotic cavity (see Fig. 1.2), is currently
h Other defects
Reference
Hydrocephalus; skull,mandible, heart, trachea,
limb, somites, ribs
Kume et al. (1998)
Cleft palate: skull, middle ear,heart, somites
Iida et al. (1997)
Somites, tail Kokubu et al. (2004)
Broad limbs, short body and
tail
Stottmann et al. (2006)
Cleft face, blebs; skull, ribs Payne et al. (1997)Cleft palate Ding et al. (2004)Edema; kidney, pancreas Lu et al. (2001)
Cleft palate; skull, mandible,middle ear, limb
Martin et al. (1995)
Short snout Hollander (1976)
Cleft palate; skull, mandible,heart, limb
Sanford et al. (1997)
Tail Park et al. (1989)
Ribs, trachea, tail Regnier et al. (2002)Ribs Aruga et al. (1999)
Fig. 1.9. Split cord malformations (SCMs). Split cord malformations are believed to arise from accessory endomesenchymal
tracts that develop from abnormal ecto-endoderm adhesions (A). This accessory tract may be present to a variable extent ven-
trodorsally and may split the notochord and neural tube (B). SCMs are classified into type I and type II lesions based on the clas-
sification scheme of Pang. Type I was often previously referred to as diastematomyelia (C, E) while type II malformations (D, F)were commonly referred to as diplomyelia. Though both malformations are now believed to result from a common developmental
anomaly (differential healing around the endomesenchymal tract as inC, D) they have different features. Type I SCMs are typified
by a hypertrophic neural arch and separate spinal cords enclosed within separate dura, and separated by a midline bony septum
(E). In contrast, in type II SCMs these features are absent and the two spinal cords are housed within the same dura. (Adapted from
Pang and Dias, 1992, with permission.)
20 G.W.J. HAWRYLUK ET AL.
believed to play an important role in the genesis of a va-riety of spinal malformations. In particular, duplicationsof this structure are believed to be responsible for splitcord malformations (SCMs).
SCMs have been recognized for over 100 years; how-ever, our understanding of their ontogeny continues toevolve. Study of these lesions has been impaired byinconsistencies in the nomenclature applied to the twomain variants. In 1906 Bruce used the term “diastemato-myelia” to refer to two hemicords, each surrounded bytheir own dural tubes, and separated by a midline bodyspur, while “diplomyelia” was used to refer to two hemi-cords within a single dural sac (Bruce and Mc’Donald,1906). Unfortunately, over the last century both termswere applied inconsistently (Ugarte et al., 1980). Alsoproblematic has been the notion, held for decades, thatboth lesions had a distinct ontogeny. Herren and Edwardsproposed that diplomyelia resulted from neural tubes“over-rolling” during neurulation; however, this theorydoes not account for more extensive anteroposterior mid-line defects associated with these anomalies (Pang andDias, 1992). It was also believed that diastematomyeliaresulted from induction of two neural tubes, perhaps asa result of notochord duplication. It is clear, however, thatduplication of the notochord does not induce two neuraltubes, but instead two floor plates, which results in a tri-angular-shaped cord.
Analysis of 41 human cases led Pang et al. to propose anew classification system in 1992 (Pang and Dias, 1992).
In this system, a type I split cord malformation refers totwo hemicords, each within their own dural tubes, sepa-rated by an osseocartilaginous median septum, previ-ously inconsistently referred to as diastematomyelia.Type II split cord malformations involve two hemicordswithin a single dural tube, generally referred to in the pastas diplomyelia (Fig. 1.10). Pang also suggested that bothSCMs were tethering lesions, not just type I lesions ashad traditionally been thought (Pang and Dias, 1992).
Pang additionally postulated that a common embry-ological malformation can lead to either type of SCM.Central to this theory is the notion that accessory neur-enteric canals form as a result of abnormal adhesionsbetween ectoderm and endoderm. Variable healing orresolution of this anomaly is thought to explain the en-tire spectrum of split cord malformation, as well as per-sistent neurenteric fistulas and cysts, lipomas anddermoids (see Fig. 1.9). The appearance of both typesof SCMs in the same patient supports the notion of acommon ontogeny (Pang and Dias, 1992). It is also inter-esting to note that this theory suggests that the embryo-logical defect must occur at the neural plate stage as thehemicords of both lesions have central canals which area product of neurulation (Pang and Dias, 1992).
Cells of the meninx primitiva generate precursor cellswhich form the meninges and vertebrae; Pang’s theoryholds that they are the critical determinant of the typeof split cord malformation. The cells initially form ven-tral to the neural tube. They then migrate dorsally and
Fig. 1.10. Other malformations related to accessory endomesenchymal tracts. In addition to split cordmalformations, a number of
additional malformations are believed to be related to accessory endomesenchymal tracts. Anomalies of the overlying skin such as
hypertrichosis (A, C) may signify the presence of an underlying anomaly. (A) Abnormal connections with the overlying skin can
be seen, referred to as dermal sinus tracts. Similarly abnormal connections can arise anteriorly, connecting the spine with the
lumen of the gut or respiratory tract. Myelomeningocele manque (B) is a similar anomaly in which blood vessels and nervous
cells form a midline septum. Several tumours are also believed related to this abnormal development (C). Dermoid cysts and
lipomas are two such tumours, and they are frequently found posterior to the spinal cord and may be associated with a split cord
malformation, as in this case.
DEVELOPMENT AND MATURATION OF THE SPINAL CORD 21
Y
encircle the neural tube and line the variably healed rem-nant of the accessory endomesenchymal tract. Variationsin the extent of meninx invasion between the neuraltubes is believed to determine which SCM forms – typeII malformations are believed to result from a failureof precursor cells to be recruited from the meninxbecause it is believed that the endomesenchymal tractforms prior to the appearance of meninx cells.
A large number of additional midline anomalies havebeen attributed to the accessory endomesenchymal tract.The abnormal tract can traverse vertebra, “splitting”them, and the mesenchymal cells which condense aroundit can form bone, fibroblasts, cartilage, blood vessels, fat,and myoblasts. Abnormal bone formation can lead to hy-pertrophic neural arches and fusion of adjacent vertebrae.Endoderm can invade abnormal fistula forming an endo-mesenchymal tract lined by respiratory or gastrointestinalepithelia, connecting the spinal cord with the lumen of therespiratory or gastrointestinal tracts (with the latter beingreferred to as a spinal-enteric tract). This communicationcan tether the gut and lead to malrotation. As well, it isbelieved that neurenteric cysts form from rests of endo-derm at dorsal aspects of the endomesenchymal tractsuch that they abut the anterior aspect of the cord. Mid-line lipomas between the split cords or just dorsal to themare quite common. Myelomeningocele manque is a spe-cial situation in which the abnormal midline structure iscomposed of nerve roots, fibrous bands, and blood ves-sels attaching to the dorsal dura. From a surgical perspec-tive, it is important to remember that tethering lesions arealways oriented caudally and oblique as a result of ascentof the spinal cord with development.
Genetic defects leading to SCM are just beginning tobe elucidated. Tubbs and Oakes report SCM associatedwith lumbosacral agenesis, lipomyelomeningocele, sin-gle central maxillary incisor along with a deficit of17q (Tubbs and Oakes, 2004a). Associations with spon-dylocostal dysostosis and Angelman syndrome havealso been reported (Mastroyianni and Kontopoulos,2002; Etus et al., 2003). Perhaps the strongest associa-tion described as yet is with Klippel–Feil syndrome(David et al., 1996; Tubbs et al., 2003), though it is im-portant to remember that bony overgrowth and vertebralfusion are recognized as being a part of SCM; the pres-ence of additional stigmata of Klippel–Feil syndromesuch as Sprengel’s deformity will be critical to establish-ing a true link between these conditions. It is also note-worthy that all reported siblings with SCM have beenfemale, although there is currently no known explana-tion for this sex predilection (Ersahin et al., 2002). Itis also interesting to note that despite an apparentgenetic contribution to these lesions, current animalmodels of SCM rely upon iatrogenic injury (Emuraet al., 2000; Ersahin et al., 2002) inducing a fistula
22 G.W.J. HAWR
analogous to a neurenteric canal. We are unaware ofany mutant animal models of SCM.
DERMAL SINUS TRACT
When the endomesenchymal tract persists posteriorly itcan lead to a dermal sinus tract. These are stratifiedsquamous epithelium-lined tracts in communicationwith the skin typically located at the extreme rostraland caudal ends of the nervous system, with caudal be-ing more common and seen in 1–2% of neonates. Thesetracts may terminate superficially, or can extend to thedural tube, perhaps through a defect in the vertebralarches. These can be important pathways for infectionto enter the CNS, and contents of the tract can also leadto aseptic meningitis. Furthermore, these may be tether-ing lesions. Epidermoid cysts (stratified squamous epi-thelium containing keratin) and dermoid cysts (dermiswith dermal appendages) can form along this tract. As-sociation with other lesions is believed to result from thefact that a dorsal endomesenchymal tract can interferewith neurulation and lead to myelomeningocele and fail-ure of mesodermal cells to migrate over the cord (Pangand Dias, 1992).
Recent evidence suggests that anomalies of FGF3,FGF4, FGF19 and ORAOV1 may be associated withdermoid sinus development based on analysis of Rhode-sian ridgeback dogs which are predisposed to theselesions (Salmon Hillbertz et al., 2007). In particular,these animals, which have an abnormal dorsal hair ridge,demonstrate a 133 kB duplication of one or more ofthese genes. The inheritance of these lesions is believedto be autosomal dominant (Hillbertz and Andersson,2006). The report of a mother and child with occipitaldermoid sinus may support the notion of a genetic basisfor these lesions in humans (Ansari et al., 2006).
Tethered spinal cord
In the absence of other tethering lesions, a tethered cord isassociated with a low-lying conus medullaris and a shortthickened filum terminale, measuring 2 mm or more indiameter (Hoffman et al., 1976; Lee et al., 2006). Tetheredcord is believed to be a disorder of secondary neurulationand the regulatory genes for this process are thought to bedistinct from those of primary neurulation (Bassuk et al.,2005). Development of a tight filum may arise from de-fective retrogressive differentiation of the caudal neuraltube (Park and Scott, 2003b).
There is a developing literature regarding possible ge-netic causes of tethered cord syndrome (Bassuk et al.,2005). Mice with mutation of the homeobox geneHOXB13 are typified by a low-lying spinal cord(Economides et al., 2003). In humans it has been notedthat tethered cord runs in some families, lending
LUK ET AL.
ATION OF THE SPINAL CORD 23
credence to the possibility of a genetic association(Bassuk et al., 2005). Indeed, several human case reportshave been published beginning in the 1990s (Roumeet al., 1990; de Toni et al., 1993; Motohashi et al.,1993; Helali et al., 1996; Salihu et al., 1997; Grahamet al., 1998; Nowaczyk et al., 1998; Ragan et al., 1999;Campbell et al., 2002; Tiranti and D’Adamo, 2004; Tubbsand Oakes, 2004b). In humans mutations of the homeo-boxHLXB9 gene have been noted in patientswith tetheredcord syndrome seen in conjunction with the Currarinotriad: anal abnormalities, anteriormeningocele, and sacralabnormalities (Bassuk et al., 2005). However, Bassuket al. were unable to delineate contributions of mutationsin eitherHLXB9 orHOXB13 in a prospective screen of 33patients (Bassuk et al., 2005). In a retrospective study thesame group identified 26 patients and concluded thatanomalies of chromosomes 21 or 22 may be associated.On chromosome 22, the locus 22q11.2 seemed important,and the geneTBX1 from this region has been suggested asa candidate gene (Jerome and Papaioannou, 2001; Lindsayet al., 2001; Merscher et al., 2001; Vitelli et al., 2002).Trisomy of 21q also appears to be associated with in-creased risk for tethered cord syndrome, and the ETS2gene residing on 21q has been described as deserving fur-ther scrutiny given its role in skeletal anomalies. Tetheredcord has also been observed in patients with trisomy12q32, trisomy 8, NF1, as well as Klippel–Trenauny–Weber syndrome, Klippel–Feil anomaly, Dandy–Walkeranomaly, and Fuhrmann syndrome (Bassuk et al.,2005). The FG syndrome (Opitz andRauch, 2001) has alsobeen suggested to bear a close association with tetheredcord syndrome (Opitz, 2005).
DEVELOPMENT AND MATUR
CONCLUSION
Althoughmuch remains to be learned, there has been ap-preciable recent progress in understanding molecularmechanisms underlying normal and abnormal develop-ment of the spinal cord. We may never fully appreciatethese processes in their full complexity, but thanks toemerging research, novel therapies may soon join folicacid supplementation in our armamentarium of agentspreventing spinal cord malformations. It is also highlyprobable that many of these insights will enhance ourability to repair the injured CNS, as neural repair reca-pitulates development, to some extent. Indeed, these ad-vances may offer new hope to many in great need of it.
REFERENCES
Aepfelbacher M, Essler M, Huber E et al. (1997). Bacterial
toxins block endothelial wound repair. Evidence that
Rho GTPases control cytoskeletal rearrangements in