Gastrulation

Gastrulation is the birthday of the nervous system. It is not only the time that bilateral symmetry and the three axes are established in the body of all vertebrates but also the time when a neuroepithelium can first be identified and distinguished from primitive germinal tissues. The traditional concept of three germ layers dates from gastrulation as well, but the convenient conception of all mature tissues having been derived from one of the three layers is probably more arbitrary than biologic because the neural crest forms tissues assigned to all three germinal layers and the expression of many families of genes does not respect these germinal boundaries and mediates the development of structures corresponding to all three.

In simple chordates, such as Amphioxus and amphibians, gastrulation is the invagination of a spherical blastula. In birds and mammals, the blastula is collapsed as a flattened, bilayered disk, and gastrulation appears not as an invagination but as a groove between two ridges on one surface of this disk, called the primitive streak on the epiblast. In each embryo, the primitive streak establishes the basic body plan of all vertebrates: a midline axis, bilateral symmetry, rostral and caudal ends, and dorsal and ventral surfaces.

As the primitive streak extends forward, cells aggregate at one end, a collection designated the primitive node or Hensen’s node. Hensen’s node defines rostral. Cells of the epiblast on either side move toward the primitive streak, stream through it, and emerge beneath it to pass into the narrow cavity between the two sheets of cells, with the epiblast above and the hypoblast below; these migratory cells give rise to the mesoderm and endoderm internally, and some then replace the hypoblast.⁵⁴

After extending about halfway across the blastoderm (epiblast), the primitive streak with Hensen’s node reverses the direction of its growth and retreats, moving posteriorly as the head fold and neural plate form anterior to Hensen’s node. As the node regresses, a notochordal process develops in the area rostral to it, and somites begin to form on either side of the notochord, with the more caudal somites differentiating first and successive ones differentiating anterior to the somites already formed. The notochord induces epiblast cells to form neuroectoderm (see “Induction”). Several genes essential in creating the fundamental architecture of the embryo and its nervous system are already expressed in the primitive node,⁵⁵ and many reappear later to influence more advanced stages of ontogenesis.

Induction

Induction refers to the influence of one embryonic tissue on another such that both the inducer and the induced differentiate as different mature tissues. In the case of the nervous system, neural tube development may be defined in terms of gradients of inductive influences. Induction usually occurs between germ layers, as with the notochord (mesoderm) inducing the floor plate of the neural tube (ectoderm), although induction also may occur within a single germ layer. An example is the optic cup (neuroectoderm) inducing the formation of a lens and cornea from the overlying epithelium (surface ectoderm) that otherwise would have differentiated as more epidermis. Neural induction is the differentiation or maturation of neural structures from undifferentiated ectodermal cells as a result of the influence of surrounding embryonic tissues.

Induction was discovered in 1924, when Hans Spemann and Hilde Mangold demonstrated that the dorsal lip of the newt gastrula was capable of inducing the formation of an ectopic second nervous system when transplanted to another site in a host embryo, into another individual of the same species, or to a ventral site of the same embryo.⁵⁶ This dorsal lip of the amphibian gastrula, also called the Spemann organizer, is homologous with the Hensen node of embryonic birds and mammals.

The first gene isolated from the Spemann organizer was Gsc (goosecoid), which encodes a homeodomain protein (see the later section “Transcription Factors and Homeoboxes”) able to recapitulate transplantation of the dorsal lip tissue when injected into an ectopic site. It also normally induces the prechordal mesoderm and contributes to prosencephalic differentiation.^55,57,58 In Hensen’s node in the chick, even before the primitive streak is fully formed, Wnt8c is expressed and is essential for the regulation of axis formation and later for hindbrain patterning in the region of the future rhombomere 4 (r4).⁵⁹ The regulatory gene Cnot, with major domains in the primitive node, notochord, and prenodal and postnodal neural plate, is also involved in the induction of prechordal mesoderm and in formation of the notochord in particular.⁶⁰

The specificity of induction lies not in the inductive molecule but rather in the receptor in the induced cell. This distinction is important because foreign molecules similar in structure to the natural inductor molecule may sometimes be erroneously recognized by the receptor as identical; such foreign molecules may act as teratogens if the embryo is exposed to such a toxin. Induction occurs during a very precise temporal window; the period of responsiveness of the induced cell is designated its competence, and the cell is incapable of responding before or after that precise time.⁶¹

Induction receptors are not necessarily in or on the plasma membrane of the cell but may be in the cytoplasm or in the nucleus. Retinoic acid is an example of a nuclear inducer. In some cases, the stimulus acts exclusively at the plasma membrane of target cells and does not require actual penetration of the cell.^61,62 The receptors that represent the specificity of induction are also genetically programmed. Notch is a particularly important gene in regulating the competence of a cell to respond to inductive cues from within the neural tube and from surrounding embryonic tissues.⁶³ Some mesodermal tissues, such as smooth muscle of the fetal gut, can act as mitogens on the neuroepithelium by increasing the rate of cellular proliferation,^64,65 but this phenomenon is not true neural induction because the proliferating cells do not differentiate or mature. Some organizer and regulatory genes of the nervous system, such as Wnt1, also exhibit mitogenic effects,⁶⁶ and insulin-like growth factor and basic fibroblast growth factor act as mitogens as well.^67–69

Early formation of the neural plate is not accomplished exclusively by mitotic proliferation of neuroepithelial cells; surrounding cells are also converted to a neural fate. In amphibians, a gene known as Xash (achaete-scute) is expressed very early in the dorsal part of the embryo from the time of gastrulation and acts as a molecular switch to change the fate of undifferentiated cells to become neuroepithelium rather than surface ectodermal or mesodermal tissues.⁷⁰ Some cells differentiate as specific types because they are actively inhibited from differentiating into others. All ectodermal cells are preprogrammed to form neuroepithelium, and neuroepithelial cells are preprogrammed to become neurons if not inhibited by genes that direct them along a different lineage, such as epidermal, glial, or ependymal.^71–73

The neural tube induces craniofacial development and mediates it through the neural crest, which migrates rostrally from the prosencephalon, at the dorsal part of the lamina terminalis, and from the dorsal midline of the mesencephalon. The prosencephalic neural crest migrates as a vertical sheet of cells in the midline of the future nose and forehead and forms, among other structures, the intercanthal ligament that hold the orbits together so that the eyes are directed forward in the face instead of being located at the sides of the head. This program is genetically determined in some families of mammals, including primates, felines, canines, bears, and koalas, as well as in one family of birds only, the owls. Other animals have laterally placed eyes, which provide better panoramic, but not stereoscopic vision.

Neurulation

Bending of the neural placode to form the neural tube requires extrinsic and intrinsic mechanical forces in addition to dorsalizing and ventralizing genetic influences, which are discussed in detail later in this chapter.

These forces arise in part from growth of the surrounding mesodermal tissues on either side of the neural tube, the future somites (Table 4-2).⁷⁴ After surgical removal of mesoderm and endoderm from one side of the neuroepithelium in experimental animals, the neural tube still closes, but it is rotated and becomes asymmetric.⁷⁵ The mesoderm appears to be important for orientation but not for closure of the neural tube. Expansion of the surface epithelium of the embryo is the principal extrinsic force for folding of the neuroepithelium to form the neural tube.⁷⁶ Cells of the neural placode are mobile and migrate beneath the surface ectoderm, which causes the lateral margins of the placode to become raised toward the dorsal midline. Growth of the whole embryo itself does not appear to be an important factor because neurulation proceeds equally well in anamniotes (e.g., amphibians), which do not grow during this period, and in amniotes (e.g., mammals), which grow rapidly at this time.⁷⁷

TABLE 4-2 Factors Involved in Closure of Neuroepithelium to Form the Neural Tube

From Menkes JH, Sarnat HB. Child Neurology, 6th ed. Philadelphia: Lippincott Williams & Wilkins; 2000:289.

Among the intrinsic forces of the neuroepithelium, the cells of the floor plate have a wedge shape—narrow at the apex and broad at the base—that facilitates bending.⁷⁸ Although the width of the floor plate is small, its site in the ventral midline is crucial and sufficient to allow a significant influence. It represents yet another aspect of induction of the floor plate by the notochord, apart from its influence on the differentiation of neural cells.⁷⁹ The ependymal cells that form the floor plate are the first neural cells to differentiate, and they induce growth of the parenchyma of the ventral zone more than the dorsal regions.^80,81 This mechanical effect may also facilitate curving of the neural placode. The direction of proliferation of new cells in the mitotic cycle, determined in part by the orientation of the mitotic spindle, becomes another mechanical force shaping the neural tube.^77,78 Adhesion molecules are also probably an important mechanical factor for neurulation. In later stages, the ependymal cell–lined central canal, which is much larger in the fetus than in the newborn, may have a role in exerting a centrifugal force to create the tubular shape, although in early spinal cord development the central canal is a tall, narrow, midline slit and only later in fetal life does it assume a rounded contour as seen in transverse sections.⁸⁰

Neuroepithelial cells of the neural placode or plate downregulate the polarity of their plasma membrane so that the apical and basilar surfaces are not as distinct before neural tube closure. Cell differentiation in general involves such changes in cell polarity.⁸² The rostrocaudal orientation of most mitotic spindles of the neuroepithelium and the direction in which they push by the mass of daughter cells that they form also influence the shape of the neural tube (Fig. 4-2).⁸³

FIGURE 4-2 Primary neurulation: schematic illustration of formation of the neural tube during the third and fourth weeks of gestation.

(From Cowan WM. The development of the brain. Sci Am. 1979;241:113.)

The neural tube closes in the dorsal midline first in the cervical region, with the closure then extending rostrally and caudally such that the anterior neuropore of the human embryo closes at 24 days and the posterior neuropore closes at 28 days, with the distances from the cervical region being unequal. This traditional view of a continuous zipper-like closure is an oversimplification. In the mouse embryo, the neural tube closes in the cranial region at four distinct sites, with the closure proceeding bidirectionally or unidirectionally and in general synchrony with somite formation.^84,85 An intermittent pattern of anterior neural tube closure involving multiple sites has also been described in human embryos.⁸⁶ In this closure pattern, the principal rostral neuropore closes bidirectionally⁸⁷ to form the lamina terminalis, an essential primordium of the forebrain.⁸⁰

Bending of the neural plate to form the neural tube is termed primary neurulation. Failure of the anterior neuropore to close by 24 days results in anencephaly. Because the lamina terminalis does not form, its derivatives (including the basal ganglia and other forebrain structures) do not develop. The lack of forebrain neuroectoderm results in failure of induction of the overlying mesoderm, and the cranium, meninges, and scalp fail to close in the midline.⁸⁸ The term secondary neurulation refers only to the most caudal part of the spinal cord (i.e., conus medullaris), which develops from neuroepithelium caudal to the site of posterior neuropore closure. More details on abnormalities that occur because of problems with secondary neurulation are offered in other chapters in this textbook.

Neural crest cells arise from the dorsal midline of the neural tube at or shortly after the time of closure and migrate extensively along prescribed routes through the embryo to differentiate as the peripheral nervous system. This includes the dorsal root and sympathetic ganglia, adrenal medulla and carotid body chromaffin cells, melanocytes, and a few other cell types of ectodermal and mesodermal origin.^89,90

Segmentation and Regionalization

Segmentation of the neural tube creates intrinsic compartments that restrict the movement of cells by physical and chemical boundaries between adjacent compartments. These embryonic compartments are known as neuromeres. The spinal cord has the appearance of a highly segmented structure; however, it is not intrinsically segmented in the embryo, fetus, or adult but rather corresponds in its entirety to the most caudal of the eight neuromeres that create the hindbrain. The apparent segmentation of the spinal cord results from clustering of nerve roots imposed by true segmentation of surrounding tissues derived from the mesoderm, tissues that form the neural arches of the vertebrae, somites, and associated structures. Neuromeres of the hindbrain are designated rhombomeres.^91–94 The entire cerebellar cortex, vermis, flocculonodular lobe, and lateral hemispheres develop from rhombomere 1 (r1), with a small contribution to the anterior vermis from the mesencephalic neuromere, but the dentate and other deep cerebellar nuclei are formed in rhombomere 2 (r2).^95,96 The rostral end of the neural tube forms a mesencephalic neuromere and probably six forebrain neuromeres (i.e., two diencephalic and four telencephalic prosomeres), although these may be subdivided further.^97–99 The segmentation of the human embryonic brain into neuromeres is summarized in Table 4-3.

TABLE 4-3 Segmentation of the Neural Tube

From Menkes JH, Sarnat HB. Child Neurology, 6th ed. Philadelphia: Lippincott Williams & Wilkins; 2000:280.

The segments of the embryonic neural tube are distinguished by physical barriers formed by processes of early specializing cells that resemble the radial glial cells that appear later in development^100,101 and by chemical barriers from secreted molecules that repel migratory cells. Cell adhesion is increased in the boundary zones between rhombomeres, which also contributes to the creation of barriers against cellular migration in the longitudinal axis. Limited mitotic proliferation of the neuroepithelium occurs in the boundary zones between rhombomeres. Although cells still divide in this zone, their nuclei remain near the ventricle during the mitotic cycle and do not move as far centrifugally within the elongated cell cytoplasm during the interkinetic gap phases as they generally do.¹⁰¹ The rhombomeres of the brainstem may also be visualized as a series of transverse ridges and grooves on the dorsal surface, the future floor of the fourth ventricle; these ridges are gross morphologic markers of the hindbrain compartments.^94,102

The first evidence of segmentation is a boundary that separates the future mesencephalic neuromere from r1 of the hindbrain. More genes play a role in this initial segmentation of the neural tube than in any boundaries that subsequently form to separate other neuromeres. The mesencephalic-metencephalic region appears to develop early as a single, independent unit or “organizer” for other neuromeres rostral and caudal to that zone.^103,104 The organizer genes recognized at the mesencephalic-metencephalic boundary for this earliest segmentation of the neural tube include Pax2, Wnt1, En1, En2, Pax5, Pax8, Otx1, Otx2, Gbx2, Nkx2-2, and Fgf8.

The earliest known gene with regional expression in the mouse is Pax2, and it is expressed even before the neural plate forms. It is the earliest gene recognized in the presumptive region of the midbrain-hindbrain boundary.^105,106 In invertebrates, Pax2 is important for the activation of Wg (wingless) genes; this relationship is relevant because the first gene definitely associated with an identified midbrain-hindbrain boundary in vertebrates is Wnt1, a homologue of Wg. Regulation of Wnt1 may be divided into two phases. In the early phase (1 or 2 somites), the mesencephalon broadly expresses the gene throughout; in the later phase (15 to 20 somites), expression is restricted to the dorsal regions, the roof plate of the caudal diencephalon, the mesencephalon, the myelencephalon, and the spinal cord, but it is also expressed in a ring that extends ventrally just rostral to the midbrain-hindbrain boundary and in the ventral midline of the caudal diencephalon and mesencephalon.^107–109 Wnt1 is essential in activating and preserving the function of the mouse engrailed genes En1 and En2. En1 is coexpressed with Wnt1 at the 1-somite stage in a domain only slightly caudal to Wnt1, which includes the midbrain and r1, the rostral half of the pons, and the cerebellar cortex but excludes the diencephalon.¹¹⁰ Activation of En2 begins at the 4-somite stage, and its function in mesencephalic and r1 development is similar, with differences in some details, particularly their roles in cerebellar development.^111,112 The homeobox gene Otx2 appears early in the initial boundary zone of the midbrain-hindbrain, and as with Wnt1, it appears to be essential for the later expression of En1, En2, and Wnt1.^113,114

The creation of neuromeres allows the development of structures within regions of the brain without the wandering of neuroblasts that form these nuclei to other parts of the neuraxis where they would not be able to later establish their required synaptic relationships. The interaction of genes with one another is a complexity that makes analysis of single-gene expression more difficult in interpreting programmed malformations of the brain.

Patterning of the Neural Tube

Development of the basic characteristics of the body plan is called patterning.⁹¹ These patterns are the anatomic expression of the genetic code within the nuclear DNA of every cell, but they may also result from signals from neighboring cells carried by molecules that are secretory translation products of various families of organizer genes, each in a highly precise and predictable temporal and spatial distribution.

Early development of the CNS in all vertebrates, even before closure of the neural placode or plate to form the neural tube, requires the establishment of a fundamental body plan of bilateral symmetry, with cephalization, or the identity of head and tail ends, and determination of the dorsal and ventral surfaces. These axes of the body itself and the CNS require the expression of genes that impose gradients of differentiation and growth. The genes that determine the polarity and gradients of the anatomic axes are called organizer genes. Many express themselves in the CNS and in other organs and tissues.^54,109 The bilateral symmetry of many organs and programmed asymmetries, probably including neural structures such as the different targets of the left and right vagal nerves and left-right asymmetries in the cerebral cortex, is determined in large part by Pitx2, a gene expressed as early as in the primitive node.¹¹⁵ Some genes function to stimulate or inhibit the expression of others, or there is an antagonism or equilibrium between certain families of genes, as exemplified by those that exert dorsoventral or ventrodorsal gradients. The difference between an organizer gene and a regulator gene is its function, and the same gene often subserves both roles at different stages of development. The definitions and programs of these two groups are summarized in Table 4-4.

TABLE 4-4 Programs of Developmental Genes

Holoprosencephaly

Holoprosencephaly (HPE) results from noncleavage of the midline ventral forebrain at approximately 33 days’ gestation and is classified by the degree of failure of hemispheric separation. This ventral failure of induction also leads to a wide spectrum of craniofacial defects ranging from cyclopia to milder defects such as iris colobomas, cleft palate, and single central incisors.²³⁸ In most cases, the severity of the brain abnormality parallels that of the face.^239,240 Neuropathologic studies suggest that the correlation of severity between brain malformation and midfacial hypoplasia may be due to abnormalities in the expression of key developmental genes along their rostrocaudal gradients, which extend as far as the mesencephalic neuromere. These abnormally expressed genes may interfere with formation, migration, or apoptosis of the mesencephalic neural crest, which forms the membranous bones of the face, orbits, nose, and parts of the eyes.²⁴¹ In alobar HPE, the most severe form, there is a single midline forebrain ventricle and a cerebral holosphere, with no separation at all into cerebral hemispheres. The interhemispheric fissure and corpus callosum are completely absent. In semilobar HPE, the defect is localized more anteriorly, and some portion of the posterior interhemispheric fissure and variable amounts of the splenium of the corpus callosum are present (Fig. 4-4). In lobar HPE, the mildest form, there is separation of most of the cerebral hemispheres, with only the most rostral aspects remaining noncleaved. The corpus callosum is absent in the region affected.

FIGURE 4-4 Holoprosencephaly. A, Midsagittal, T1-weighted magnetic resonance imaging (MRI) of a child with alobar holoprosencephaly shows a pancake of brain anteriorly, with the single, midline ventricle leading into a large, dorsal cyst. The corpus callosum is absent. B, Axial, T1-weighted MRI of the same child shows the lack of any separation of the brain into hemispheres and the crescent-shaped holoventricle leading into the dorsal cyst. C, Axial, T1-weighted MRI of a child with semilobar holoprosencephaly shows a smaller dorsal cyst. Notice the failure of cleavage of the frontal lobes and basal ganglia and the separation into hemispheres posteriorly.

(Courtesy of Joseph Pinter, M.D., Children’s Hospital and Regional Medical Center, Seattle, WA.)

The middle interhemispheric variant, once thought to be rare,²⁴² has in later large studies been identified quite frequently (Fig. 4-5).²⁴³ Radiologic studies have also revealed that HPE is much more variable than the simple classification system described earlier would suggest. There is greater variability in the pattern and degree of noncleavage of the deep gray nuclei than previously appreciated, with almost universal involvement of the hypothalamus.²⁴⁴ Thalamic noncleavage is correlated with the presence of a dorsal cyst, thus supporting the idea that the cyst arises because of blockage of rostral flow of cerebrospinal fluid, which then egresses dorsally.^245,246

FIGURE 4-5 Middle interhemispheric variant of holoprosencephaly. A, Sagittal, T1-weighted magnetic resonance imaging (MRI) shows a grossly normal callosal genu and splenium (arrows) but absence of the midportion of the corpus callosum. B, Coronal, T1-weighted MRI reveals the absence of an interhemispheric fissure and a continuous band of subcortical white matter crossing the midline. The heterotopic gray matter (arrow) at the roof of the ventricle is a common finding in this variant.

(From Barkovich AJ. Pediatric Neuroimaging, 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2000:324.)

Human HPE is often seen as part of trisomy 13, but it has also been mapped to several other chromosomes.²³⁸ SHH was the first gene identified to cause HPE in humans.¹³⁴ The active cleavage product of secreted Shh is covalently bound to cholesterol and then binds to its receptor Ptc (patched), which stops inhibiting the constitutive signaling activity of Smo (smoothened). Binding activates a cascade of molecular events that culminate in the transcription of dorsalizing genes such as Wnt and Bmp, which encode transcription factors. The Shh pathway normally keeps ventralizing and dorsalizing influences in appropriate balance. Interference with this complicated pathway explains why certain toxins cause HPE in animals.²⁴⁷ Mutations in two other genes, SIX3 and TGIF, can cause HPE through distinct signaling pathways.^15,16 Mutations in the ZIC2 gene also cause HPE.¹⁴ ZIC2 promotes embryonic roof plate differentiation in the dorsal midline of the neural tube after closure. In mice, roof plate–specific properties such as an increased apoptotic rate and decreased mitotic rate are critical to formation of the interhemispheric fissure. Compromise of roof plate differentiation could therefore lead to HPE, and one patient with the middle interhemispheric variant has been shown to have this mutation.²⁴³

Because the telencephalic flexure fails to form in HPE, the hippocampus remains a dorsal structure and has the form of a U-shaped cord that extends across the midline and posteriorly to the caudal poles of the malformed forebrain.²³⁷

Neuronogenesis

Normal Proliferation of Neuroblasts

The neuroblast is histologically indistinguishable from other stem cells but is identified by its location in the primitive neural plate. After formation of the neural tube, neurons and glial cells are generated by proliferation of neuroepithelial cells in the ventricular zone with mitoses at the ventricular surface. The rate of division is greatest during the early first trimester in the spinal cord and brainstem and during the late first and early second trimester in the forebrain. Within the ventricular zone of the human fetal telencephalon, 33 mitotic cycles provide the total number of neurons required for the mature cerebral cortex. Most mitotic activity in the neuroepithelium occurs at the ventricular surface, and the orientation of the mitotic spindle determines the subsequent immediate fate of the daughter cells. If the cleavage plane is perpendicular to the ventricular surface, the two daughter cells become equal neuroepithelial cells preparing for further mitosis. If, however, the cleavage is parallel to the ventricular surface, the two daughter cells are unequal (i.e., asymmetric cleavage), and the one at the ventricular surface becomes another neuroepithelial cell, whereas the one away from the ventricular surface is separated from its ventricular attachment and becomes a postmitotic neuroblast ready to migrate to the cortical plate. These asymmetric cleavages are mediated by the products of two proteins that determine cell fate, Notch1 and Numb, which are localized to the basal and apical regions, respectively, of the neuroepithelial cell. With symmetrical cleavage, both daughter cells receive the same amount of each, but with asymmetric cleavage, the cells receive unequal ratios of each, thereby resulting in a migratory neuroblast (Fig. 4-6).^195,197,248

FIGURE 4-6 This schematic diagram of neuroblast proliferation in the ventricular zone shows initiation of migration by asymmetric cleavages mediated by Notch-1 and Numb, proteins that are localized to the basal and apical regions, respectively, of the neuroepithelial cell.

(From Wolpert L, Beddington R, Brocken J, et al. Principles of Development. Oxford: Oxford University Press; 1998:351.)

Neuroblast Migration

Normal Development

No neurons of the mature human brain occupy a site in which they were generated from the neuroepithelium. They are born in the subependymal germinal matrix, the subventricular zone that develops from the lumen of the neural tube, and they migrate to their adult location along tracks of long, specialized fetal astrocytes (Figs. 4-7 and 4-8).

FIGURE 4-7 In the coronal section of the forebrain of a normal fetus of 16 weeks’ gestation, notice the extensive subependymal germinal matrix (G) of neuroblasts and glial precursors that have not yet migrated. Migrating neuroblasts (M) are seen in the subcortical white matter. Sulci are just beginning to develop (arrows). The hemispheres were artifactually separated (hematoxylin-eosin stain). CC, corpus callosum; CN, caudate nucleus; IC, anterior limb of the internal capsule.

(From Sarnat HB, Flores L. Developmental disorders of the nervous system. In: Bradley WG, Daroff RB, Fenichel GM, et al, eds. Neurology in Clinical Practice. Boston: Butterworth-Heinemann; 2000:564.)

FIGURE 4-8 This schematic diagram shows a cortical neuroblast migrating outward from the ventricular proliferative zone to the cortex along a radial glial fiber.

(From Wolpert L, Beddington R, Brocken J, et al. Principles of Development. Oxford: Oxford University Press; 1998:349.)

These radial glial fibers span the entire wall of the fetal cerebral hemisphere, with their cell bodies located in the periventricular region and their end-feet on the limiting pial membrane at the brain’s surface (Fig. 4-8). From about 6 to 34 weeks’ gestation, forebrain neuroblasts migrate in waves, with each successive wave moving past earlier born neurons to add a more superficial layer of cortex in an inside-out fashion. When complete, the cortex has six layers, and each contains specific types of neurons that must have successfully migrated to establish appropriate synaptic contacts with other neurons.

In addition to radial migration to the cerebral cortex, tangential migration also occurs, but the number of neuroblasts is far smaller.^255,256 These migrations perpendicular to the radial fibers probably use axons rather than glial processes as guides for migratory neuroblasts, and this explains why all cells in a given region of cortex are not all from the same clone or vertical column. Neurons arriving at the neocortical plate by tangential migration are GABAergic (secreting γ-aminobutyric acid [GABA]) interneurons and represent about 10% of the total cortical neurons; they are distributed in all layers of the cortex but predominate in sensory granular cell layers 2 and 4 (Fig. 4-9). In contrast, in the cerebellar cortex, neuronal migration differs in that external granule cells first spread over the surface of the cerebellum and then migrate into the folia. They also have a much longer migratory period, which is not complete until after the first year of life.

FIGURE 4-9 Calretinin immunoreactivity in the normal cerebral temporal neocortex of a -year-old girl. This protein is a marker of GABAergic interneurons that arrive in the cortical plate by tangential rather than radial migration. They represent about 10% of the total cortical neurons and are distributed in all layers, but predominantly in sensory layers 2 and 4 (calretinin stain, ×250).

The laminated structure of the mammalian cerebral cortex requires a large cortical surface area to accommodate increasing numbers of migrating neuroblasts and glioblasts. Convolutions provide this large surface area without incurring a concomitant increase in cerebral volume. The formation of gyri and sulci is a direct result of migration. Because most gyri form in the second half of gestation, a period of predominantly gliogenesis and glial cell migration, the proliferation of glia in the cortex and subcortical white matter may be more important than neuroblast migration in the formation of convolutions, but the growth of dendrites and synaptogenesis also probably influence gyration by contributing mass to the neuropil. The timing and sequence of gyral formation in the human brain are as predictable as other aspects of cerebral maturation (Figs. 4-10 and 4-11).

FIGURE 4-10 In this gross specimen of a normal fetal brain at 16 weeks’ gestation, notice the presence of only primary sulci, which gives the brain a smooth, lissencephalic-like appearance.

(From Sarnat HB, Flores L. Developmental disorders of the nervous system. In: Bradley WG, Daroff RB, Fenichel GM, et al, eds. Neurology in Clinical Practice. Boston: Butterworth-Heinemann; 2000:1565.)

FIGURE 4-11 In this schematic overview of brain gyral development and growth during the embryonic and fetal periods, the first five stages are enlarged to show structural detail.

(From Cowan WM. The development of the brain. Sci Am. 1979;241:113.)

Radial glial cells are the first astroglial cells of the human nervous system to differentiate, which occurs as early as 6 weeks’ gestation. On completion of migration of neuroblasts and glioblasts, the radial fiber retracts, and the glial cell is converted into a mature fibrillary astrocyte of the subcortical white matter; some are still present at birth. Mature astrocytes are present throughout the CNS by 15 weeks’ gestation, and several types of glial cells are recognized between 20 and 36 weeks’ gestation. Gliogenesis continues throughout fetal and postnatal life.

Disorders of Neuroblast Migration

Lissencephaly is the condition of a smooth cerebral cortex without convolutions (Fig. 4-12). At midgestation, the brain is normally smooth, with only the interhemispheric, sylvian, and calcarine fissures formed. Gyri and sulci develop between 20 and 36 weeks’ gestation, and the mature pattern of gyration is evident at term. In lissencephaly type 1, the cerebral cortex remains smooth, and the histopathologic pattern is that of a four-layer cortex in which layer 2 corresponds to layers 2 through 6 of normal neocortex, layer 3 represents a persistent fetal subplate zone and is cell sparse, and layer 4 consists of incompletely migrated neurons in the subcortical intermediate zone. In lissencephaly type 2, poorly laminated cortex with disorganized and disoriented neurons is seen histologically, and the gross appearance of the cerebrum is that of a smooth brain or a few, poorly formed sulci. The cerebral mantle may be thin and suggest a disturbance in cell proliferation and neuroblast migration. Malformations of the brainstem and cerebellum are often present. Lissencephaly frequently has a genetic origin, but it may result from nongenetic disturbances in neuroepithelial proliferation or neuroblast migration, including destructive, encephaloclastic processes such as congenital infections during fetal life.

FIGURE 4-12 Lissencephaly. Axial, T1-weighted magnetic resonance imaging shows the thin, outer cortical layer separated from a deep layer of arrested neurons. The sylvian fissures (arrows) are open laterally, thus giving the brain a typical figure-of-eight appearance.

(From Barkovich AJ. Pediatric Neuroimaging, 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2000:303.)

Mutations in two genes, LIS1 and DCX, have been associated with many cases of classic lissencephaly.19–21,33–35 They both appear to disrupt migration by interfering with microtubule function. Neuroimaging studies of gyral patterns in individuals with these mutations have shown distinct radiologic patterns, with more severe malformation posteriorly in individuals with LIS1 mutations and more severe malformation anteriorly in those with DCX mutations.²⁵⁷ DCX mutations result in classic X-linked lissencephaly in boys but cause a syndrome called double cortex or subcortical band heterotopia in girls.³³ Through random X-chromosome inactivation early in embryogenesis, half of an affected girl’s cortical neurons inactivate the mutant allele, and half inactivate the normal allele. This population migrates abnormally and produces band heterotopia (Fig. 4-13).

FIGURE 4-13 Subcortical laminar heterotopia (also known as band heterotopia, double cortex). Coronal, T1-weighted magnetic resonance imaging shows a single band in the parietal lobes (top two white arrows) and two bands (large and small white arrows) in the temporal lobes.

(From Barkovich AJ. Pediatric Neuroimaging, 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2000:311).

Mutations in FCMD, which encodes a putative extracellular matrix molecule called fukutin, cause Fukuyama’s congenital muscular dystrophy, which includes lissencephaly.²² Mutation in RELN, the human gene encoding reelin, has been identified as a cause of lissencephaly with cerebellar hypoplasia.

Periventricular heterotopia (i.e., bilateral periventricular nodular heterotopia) differs from lissencephaly in that migration fails completely rather than being interrupted. The causative gene is FLNA (formerly FLN1), which encodes an actin-binding protein called filamin.²⁶ Further molecular details of these complex disorders can be found in the outstanding review articles by Ross and Walsh²⁵⁸ and by Clark.²⁵⁹

The term pachygyria signifies abnormally large, poorly formed gyri that may be isolated or associated with frankly lissencephalic regions elsewhere. Polymicrogyria refers to excessively numerous and abnormally small gyri that may coexist with pachygyria. Polymicrogyria does not necessarily denote a primary migratory disorder of genetic origin; small and poorly formed gyri may occur in zones of fetal ischemia and regularly surround porencephalic cysts caused by occlusion of the middle cerebral artery in fetal life. Schizencephaly is a unilateral or bilateral deep cleft encompassing the full thickness of the hemispheric wall between the meninges and the lateral ventricle (i.e., pial-ependymal seam). If the cerebral cortical walls on either side of the deep cleft are in contact, the condition is called closed lip, and if a wide subarachnoid space separates the two walls, it is called open lip (Fig. 4-14). Mutations in EMX2, a homeobox gene expressed in the developing forebrain, were initially identified in a small number of patients with severe schizencephaly,³¹ but genetic studies of a large cohort of patients with this malformation did not confirm this genetic defect.²⁶⁰ Because the telencephalic flexure results in both the dorsal and ventral lips of the sylvian fissure being derived from the ventral part of the primitive telencephalon, a genetic defect in schizencephaly would have to follow a ventrodorsal gradient in the vertical axis.²³⁷ Some cases may instead represent a form of porencephaly caused by fetal infarction, and in other cases, the cause of unilateral schizencephaly may be asymmetric mechanical stress in utero without implicating a genetic mechanism.²³⁷

FIGURE 4-14 Schizencephaly. A, Axial, spin-echo 2500/20 magnetic resonance imaging (MRI) shows abnormal gray matter (black arrows) from the cortex to the ventricular surface (white arrow) in the closed-lip form. B, Coronal, T1-weighted MRI shows a large cleft in continuity with the lateral ventricle in the open-lip form. Gray matter lines the entire cleft, and heterotopic gray matter occurs along the roof of the lateral ventricle (arrow).

(From Barkovich AJ. Pediatric Neuroimaging, 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2000:290.)

Anatomic lesions such as periventricular leukomalacia, intracerebral hemorrhages and abscesses, hydrocephalus, and traumatic injuries may disrupt the delicate radial glial guide fibers and prevent normal migration, even though the migrating cell itself may escape the focally destructive lesion (Fig. 4-15).

FIGURE 4-15 Diagram of a coronal section of the cerebral hemispheres of a third-trimester fetus (or preterm infant) indicating three sites where lesions such as infarcts or hemorrhages may disrupt radial glial fibers and interfere with neuroblast migration: (1) the periventricular region, where radial glial cells may be destroyed; (2) the deep hemispheric white matter, where crossing radial glial fibers may be damaged; and (3) the pial surface, where injury may cause retraction of radial glial fibers and the subsequent development of heterotopic collections of neurons.

(From Sarnat HB. Cerebral Dysgenesis. New York: Oxford University Press; 1992:263.)

The result may be the development of heterotopic nodules, with neurons of the various cortical types differentiating without laminar organization and with haphazard orientations of their processes. Postmigration CNS development includes the highly regulated and complex organizational processes of neurite outgrowth and synaptogenesis. Because of limited space, discussion of these nodules is not included in this chapter, but more information may be found in Volpe’s thorough introduction to these processes.²⁶¹

Synaptogenesis

Formation of synapses within the CNS follows as precise a temporal and spatial distribution as other developmental processes do. Neocortical synaptogenesis is indirectly reflected in maturation of the electroencephalogram in preterm infants but represents large fields of thousands of synapses. In tissue sections of fetal and neonatal brain at autopsy, an immunocytochemical marker of the synaptic vesicle protein synaptophysin can be used to demonstrate this process in individual neurons.²⁶² Synaptophysin is nonspecific for the type of synapse—excitatory or inhibitory, axodendritic or axosomatic—or the type of neurotransmitter that the vesicles contain but shows the overall pattern of synapse formation and can be combined with other immunoreactivities to define this specificity. Synaptogenesis requires the terminal axonal projection with mature synaptic vesicles, a postsynaptic membrane with receptors to respond to the chemical stimulus of depolarization, and maturity of the energy-producing Na⁺,K⁺-ATPase system to maintain a resting membrane potential and electrical excitability of the plasma membrane.

Myelination

Myelination is the process of acquiring a specialized myelin membrane around axons, which is elaborated from the plasma membranes of adjacent oligodendrocytes in the CNS and Schwann cells in the peripheral nervous system (Tables 4-6 and 4-7). As with the other processes of nervous system development, programming of differentiation of these myelin-producing cells and myelin formation is under precise genetic regulation.^263,264 Myelination cycles are specific for each tract and occur during precise time windows. The overall period for myelination is very long; it begins as early as 16 weeks’ gestation and continues in some areas past 30 years of age.

TABLE 4-6 Myelination Cycles in the Human Central Nervous System Based on Myelin Tissue Stains

* Myelination as determined by light microscopy using Luxol fast blue and other myelin stains.

† Gestational age is stated in weeks; postnatal age is stated in months and years.

From Sarnat HB. Cerebral Dysgenesis: Embryology and Clinical Expression. New York: Oxford University Press; 1992:61.

TABLE 4-7 Appearance of Myelination by Magnetic Resonance Imaging

* From birth to 3 months, the pattern on T1-weighted imaging (T1WI) is opposite that of the adult brain (gray matter appears brighter than white matter). The adult pattern is achieved by 2 years of age. It is not clear why myelination changes on T1WI and T2-weighted imaging (T2WI) occur at different times. T1WI is most useful for assessing myelination in the first year. Myelination generally progresses from caudad to cephalad, from dorsal to ventral, and from central to peripheral. The changes in signal reflect mainly increased bound water in myelin, which results in shortening of T1 (increased brightness on T1WI).

From Barkovich AJ. Pediatric Neuroimaging, 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2000:39.

Chronic hypoxia in premature infants is probably the most common cause of delayed myelination and contributes to delays in clinical neurological maturation.²⁶⁵ Unlike disorders of neuroblast migration, delay in myelination is not necessarily irreversible, and if the insult is removed, the process may catch up to the appropriate level of maturity over time.