Nov. 23, 2022
At vero eos et accusamus et iusto odio dignissimos ducimus qui blanditiis praesentium voluptatum deleniti atque corrupti quos dolores et quas.
Neuroembryology is the basis for understanding both normal and abnormal development of the nervous system. Developmental malformations mainly result from an alteration, suppression, or interruption in 1 or more of the several developmental processes that occur more simultaneously than sequentially. Such an alteration often is due to a mutation or deletion of 1 or more genes that program neural development in a precise temporal and spatial sequence. Such genes often are expressed as gradients along 1 or more of the 3 axes of the neural tube. Acquired lesions of the fetal brain also can produce malformations. Induction is the influence of 1 embryonic tissue on another, and the nervous system is intimately related to surrounding tissues to be both induced by them and also influence their development. Correlations must always be considered between morphogenesis and genetic programming. The nervous system also must be seen as a whole, not narrowly focused on one part, such as the cerebral cortex or cerebellum, because genetic mutations often affect multiple systemic structures even if the clinical manifestations are predominantly referable to one part of the brain. In addition, various structures of the brain are interconnected by synaptic networks and axonal projections as well as functionally.
• Maturation of developmental process within the nervous system, including differentiation and maturation of individual cells, normally occurs in a genetically-programmed synchronous and predictable temporal pattern.
• Developmental processes in the nervous system are mostly simultaneous and overlapping rather than sequential in a series.
• Development and maturational timing are altered not only in genetic disorders, but also by acquired epigenetic disorders affecting the fetus, such as exposure to teratogenic toxins including alcohol, fetal cerebral infarcts, and congenital infections.
• Function does not necessarily require complete neuroanatomical maturation.
• CNS malformations may exhibit altered timing of developmental processes with maturational delay, arrest, or even precociousness.
• In laminated cortices, such as the cerebral and cerebellar, dysmorphic alteration may be focal, as in focal cortical dysplasias, or generalized as in congenital metabolic encephalopathies.
• The persistence of transitory and vestigial structures of the fetal brain may cause neurologic disease.
Historical aspects. Some human brain malformations were recognized in antiquity. Holoprosencephaly was described in ancient Egypt and Greece. In the late 19th century, some large malformations such as hemimegalencephaly were described and occasionally confused with tumors. The first description of hemimegalencephaly by Sims in 1835 was based upon a review of 253 autopsies (121). However, even in that century Wilhelm His and Santiago Ramón y Cajal recognized malformations as disorders of development, though the pathogenesis eluded them. During the last quarter of the 20th century malformations including small lesions were recognized and their development described with the advent of modern neuropathology with immunocytochemistry, neuroimaging to follow serial growth and maturation of malformations in pre- and post-natal life, and genetic advances in the recognition of specific mutations associated with particular malformations. This process of identifying genetic etiologies and fetal neuroanatomical elucidation of pathogenesis in the context of normal developmental processes continues.
Because malformations of the brain involve so many different anatomical types, some generalized and others focal, involving 1 or several developmental processes of the nervous system (eg, disorders of neurulation, cellular lineage, neuroblast migration, or axonal pathfinding) and involving so many different genes, it would appear that generalizations about malformations of the brain are not possible or not useful. Nevertheless, certain common principles might be considered.
A distinction must be made between “pathogenesis” or “mechanism” on the 1 hand and “etiology” on the other. Disorders of neurulation are failures of closure of the neural folds to form the neural tube (pathogenesis); such nonclosure may be due to a deficiency in folic acid, which is essential to promote gene expression that regulates neurulation (etiology). Alobar holoprosencephaly is a lack of bilateral separation or “noncleavage” of the prosencephalon and often of the diencephalon and mesencephalon as well (pathogenesis); it is not “fusion.” Nine genes are now identified in which mutations can produce this same neuroanatomical phenotype of holoprosencephaly (etiology). The pathogenesis of spinal muscular atrophy is progressive degeneration of motor neurons in postnatal life, but the etiology is a failure of arrest of physiological apoptosis of redundant motor neuroblasts due to a genetic defect. Most malformations of the brain are of genetic origin, but acquired lesions of the fetal brain (eg, hemorrhages, infarcts, metabolic insults, and congenital infections) can interrupt developmental processes and create malformations that resemble primary genetic disorders (85).
“Maturation” is a general but useful term that encompasses not only anatomical development but also biochemical, neurophysiological, and functional change from an initial immature state with limited function. A basic principle of neuroembryology is that function does not require complete neuroanatomical maturity. An example is the olfactory system: fetuses of more than 30 weeks gestation can discriminate odorous molecules dissolved in amniotic fluid and reliable neonatal reflexes can be tested with aromatic substances that are not irritants, such as peppermint oil (112; 105).
The corticospinal tract is not anatomically mature until 2 years of age, but infants learn to sit, crawl, walk, run, and climb much sooner.
Advances in prenatal imaging of fetal brains, mainly by ultrasound and magnetic resonance, now enable many diagnoses of brain malformations, even in early stages of development during gestation (76; 50; 55). MRI imaging of normal brain development provide a control of the range of physiological changes over gestational ages that enable the recognition of malformations (72).
Neuropathological examination of both surgical and autopsy brain tissues not only can establish or confirm a diagnosis of a specific malformation and provide details at the microscopic level but also may demonstrate anatomical features that denote risk of epilepsy. For example, in polymicrogyria, there are gaps or discontinuities in the pia mater and glial limitans of adjacent microgyria that enable fusion of their molecular layers of excitatory axodendritic synapses and short-circuitry between gyri that explain why polymicrogyria is highly epileptogenic in most cases; similarly, the progressive loss of dendrites and dendritic spines in the cortical molecular zone in Down syndrome shifts the excitatory/inhibitory synaptic ratio in favor of more inhibition, explaining why most children with Down syndrome do not have seizures despite cortical abnormalities (90). In lobar/semilobar holoprosencephaly with severe abnormalities in cortical organization and lamination that would be epileptogenic, up to 40% of children do not have epilepsy; this lack of seizure can at least partially be attributed to excessive ensheathing of individual axons by keratan sulfate, an insulating extracellular proteoglycan in the developing brain (90).
Neuroembryology and genetic programming of the nervous system. Neuroembryology, in its broadest context, refers to the morphogenesis of the nervous system not only in the embryological period proper (to 6 weeks postconception in the human) but also in the fetal period that follows throughout prenatal gestation and into the postnatal period. Embryology is the essence of development and provides a rational explanation for nearly all malformations of the brain as errors in morphological development. Since the late 1980s, the discovery of many genes that regulate development, interact with other genes, and may be expressed only during a limited time in development or have different functions at later stages of ontogenesis has provided an insight into development not previously even suspected. The mouse is a frequently employed model of genetic programming in nervous system development. The murine genome and molecular architecture of the entire body during maturation are now demonstrated (46). During more advanced stages of fetal development, novel genes are differentially expressed in cortical regions of mammals including humans (59). Many genes essential to neural development also are expressed in other developing organs, such as heart and kidney, but have different functions in those organs.
Embryology explains mechanisms of development, whereas genetics provides an etiology. The comprehensive picture is only appreciated by integrating the 2 aspects rather than regarding each in isolation from the other. In terms of ontogenesis, this is the “phenotype-genotype correlation” that is so essential to comprehending malformations of the nervous system. Several textbooks of classical neuroembryology and developmental neuropathology are available (79; 62) as are anatomical atlases of the human fetal brain (77; 04; 05; 06; 07; 08; 64). Patterns of morphological development may be correlated with data regarding the genetic programming of the nervous system to create an integrated scheme that can explain malformations in terms of both morphogenesis and genetic gradients along the axes of the neural tube and segmentation (108; 83; 27; 99). Holoprosencephaly serves as a prototype disorder that involves all 3 axes and varying degrees of severity and is associated with mutations in any of several known defective genes (33; 94).
Epigenetics. Epigenetics no longer is regarded simply as any insult to the body or brain that is not a primary genetic mutation or chromosomopathy. Some neurotoxins and teratogens of the fetal environment are now documented to be capable of alternating the DNA of fetal progenitor cells and thus cause a genetic mutation as an important part of their pathogenesis. An example is fetal alcohol spectrum disorder, with postnatal clinical correlates of global developmental delay, sometimes autism spectrum disorder, sometimes epilepsy, and often a degree of either spastic or hypotonic diparesis. In this disorder—induced by fetal exposure to maternally ingested alcoholic beverages—fetal DNA methylation is altered in various patterns and affects RNA transcription and protein synthesis (71; 54). Abnormal DNA methylation also forms a basis for classification of brain tumors (15). Adenosine is the principal regulator of DNA methylation that normalizes it and furthermore induces changes in neuronal excitability that helps prevent epilepsy (135). It is likely to be associated with many other fetal neurotoxins and drugs ingested by the mother and to result in postnatal cerebral palsy and other neurologic defects. Fetal effects from teratogenic toxins may be single exposures, as with a 1-time maternal drug ingestion before a mother knew she was pregnant, to exposure over an extended time period of different developmental processes, as with congenital infections or repeated maternal drug abuse throughout gestation.
Fetal exposure to medications taken by the mother and that cross the placental barrier may interfere with development. Antiepileptic drugs taken by mothers with epilepsy are frequently associated with abnormal development of the brain, as documented in a large European registry (129). Changes in the prescribing patterns of such medications before planned pregnancies and during the first trimester in particular, including the temporary use of antiseizure medications with less teratogenicity, already has resulted in lower rates of malformations (130). Valproic acid is one of the most teratogenic of these pharmaceuticals and, in addition to causing brain malformations and neural tube defects, it also is associated with a high risk of childhood autistic spectrum disorder (20).
Genetic/environmental interactions during the critical first 1000 days from conception to postnatal life that can influence neurologic function in later childhood and adult life are increasingly recognized (114).
Embryonic germ layers. The traditional division of all tissues as derivatives originating in 1 of 3 embryonic germ layers (endoderm, mesoderm and ectoderm) remains a useful way of organizing thoughts about early embryonic development. But the advent of molecular genetics in embryology poses doubts about the biological authenticity of this classification and whether these 3 layers are not simply arbitrary and artificial. Most of the important genes known to program the development of the neural tube and used later in cellular differentiation also are expressed in a variety of tissues supposedly derived from all 3 germ layers. Thus, the HOX family of genes is among the most important for segmentation of the early neural tube (ectodermal derivative) but also mediates the development and segmentation of the vertebral column (mesodermal derivative) and plays a role in thyroid maturation (endodermal derivative). The status of the 3 germ layers, thus, is currently uncertain as a true biological phenomenon.
Induction. The term “induction” denotes the influence of 1 embryonic tissue on another, so that the 2 differentiate as different mature tissues. “Neural induction” is the influence of non-neural tissues on the neural tube (eg, the notochord induces the ependymal floor plate of the developing neural tube) or the influence of the neural tube on non-neural tissues (eg, craniofacial development induced by neural crest tissue that emanates from the prosencephalon and mesencephalon). Induction may be between germ layers (examples above) or within the same germ layer (eg, the ectodermal optic cup or future retina induces the ectodermal lens placode and overlying epithelium to form a cornea rather than simply more epidermis).
Principle of redundancy during ontogenesis. A general principle in neural development is redundancy. For example, motor neurons in the ventral horns of the spinal cord are overproduced by at least 50%, and the redundant neurons that do not match with muscle are then deleted by apoptosis. Long axonal projections involve many collaterals that are later retracted with maturation; for example, the corpus callosum at midgestation has many more axons than at term because each axon has numerous collaterals, but most of these collateral axons are later retracted so that in the mature condition there are fewer, but less diffuse and more specific, connections. Synapses are overproduced and then undergo elimination or “synaptic pruning” (73).
Duplication and fusion of structures. In general, the upregulation of a developmental gene results in duplication and hypertrophy of structures, according to the gradient of expression in 1 or more of the 3 axes of the neural tube. A dorsalizing gene, 1 acting in the dorsoventral gradient of the ventral axis of the spinal cord, when upregulated, results in duplication of the dorsal horns but not of the ventral horns; a ventralizing gene, acting in the ventrodorsal gradient of the vertical axis, results in duplication of the central canal or, in more extreme form, segmental diplomyelia. Underexpression in the ventrodorsal gradient, by contrast, results in midline fusion of the 2 ventral horns. The hippocampus begins development as a dorsal structure and, hence, is more influenced by dorsalizing genes. The upregulation of a dorsalizing gene acting at the level of the embryonic forebrain may result in duplication of the hippocampal formation on 1 or both sides of the telencephalon.
Persistence of transitory and vestigial structures. Many structures of the developing brain are transitory, needed during a particular stage of development and not after. They then regress or disappear, but in some cases may persist as vestiges and can cause disease. Examples are the septum that becomes the septum pellucidum and may result in an excessively wide cavum between its vertical leaves, Blake pouch cyst of the fourth ventricle that can cause obstructive hydrocephalus at the outlet of the fourth ventricle, subplate zone of the cerebral cortical plate, micro-columnar architecture of the cortex which is normal only during the first half of gestation and is the neuroanatomical basis of focal cortical dysplasia type 1a, and human tails (90).
The following is a list of the normal developmental processes in which a disturbance can lead to malformations. Though some processes (such as neurulation) are earlier than others (such as synaptogenesis and myelination), one should not presume that this list of processes is a sequence in which as 1 is completed the next is initiated. In reality, the great majority of these processes occur with so much overlap in timing that they are essentially simultaneous. The major processes are here listed:
Gastrulation. This is the first process, at 16 days gestation in the human, that defines the maturation of the nervous system because for the first time in the embryo the neuroepithelium can be distinguished from other primitive tissues; it is the “birthday of the nervous system” according to Wolpert (139). In addition, the time of gastrulation is the time when bilateral symmetry becomes as the body plan of all vertebrates, and the 3 principal axes of the body are identified: longitudinal, vertical, and horizontal. Each axis may have 2 gradients of genetic expression, 1 in each direction. Thus, the longitudinal axis has rostrocaudal and caudorostral gradients; the vertical axis has dorsoventral and ventrodorsal gradients; the horizontal axis has mediolateral and lateromedial gradients. The term cephalization refers to the development of a head end of the longitudinal axis but really describes the rostrocaudal gradient. Dorsal and ventral surfaces are established by the vertical axis and its gradients. The midline is in the longitudinal axis but is the reference point for the horizontal axis. WNT signalling spatially coordinates the formation of the primitive streak from Hensen’s node, which denotes the longitudinal axis of the embryo (09).
Upregulation of some genes expressed early, as at gastrulation, may cause duplication of structures. This is 1 mechanism of conjoined twinning and of diplomyelia.
Neurulation. This process is a genetically programmed bending of the neural placode of either side of the longitudinal axis (midline) to form first neural folds and then a neural tube by fusion in the dorsal midline of the original lateral margin of the neural placode. The closure begins in the cervical region and extends rostrally and caudally along the neural folds. The posterior neuropore closes at 28 days and the anterior neuropore at 24 days, earlier because the distance from the cervical region is shorter. Closure of the neuropores, the last regions to remain open, actually occurs at multiple sites in the vicinity of the dorsum of the lamina terminalis in the case of the anterior neuropore and in the sacral region of the caudal neuropore. With closure of the neural tube, the dorsal surface of the original neural placode becomes the internal surface of the central canal. Ependymal cells of the ventral midline differentiate to form the floor plate, the first neuroepithelial cells to differentiate in the nervous system, followed by the roof plate in the dorsal midline and later the remaining ependyma of the walls of the central canal. The central canal initially is a tall lumen of the neural tube (taller than wide) and only later becomes a smaller, round, tubular structure. A portion of the neural tube remains caudal to the site of posterior neuropore closure and is a solid cylinder of neural tissue in which an ependymal central canal secondarily develops by a growth of ependymal cells within the core of the cylinder; this process is called “secondary neurulation.” In fetuses at midgestation and later, transverse sections through the caudal part of the sacral spinal cord often shows 2 central canals, 1 above the other in the midline, because the central canals from primary and secondary neurulation sometimes do not meet and fuse but, rather, overlap in the vertical midline. This arrangement is a normal anatomical variant--by contrast with 2 central canals in the horizontal plane, side by side, which is always abnormal and usually denotes upregulation of a gene such as Sonic hedgehog, acting in the ventrodorsal gradient of the vertical axis and is a malformation. A more severe or prolonged upregulation of this gene at neurulation can produce not just the mild pathological duplication of the central canal but is another mechanism of focal diplomyelia. Deficiency of Sonic hedgehog production by the notochord, by contrast, may prevent the completion of neurulation with a resultant neural tube defect, as exemplified by sacral agenesis.
Disorders of neurulation produce severe and permanent malformations that result in much disability. The most frequent is the lumbosacral meningomyelocele. Some encephaloceles also are in this category. Prenatal prophylactic administration of folic acid may prevent many of these malformations; folate induces Sonic hedgehog expression.
Migration of neural crest cells. Neural crest precursor cells are identified at the lateral margins of the neural placode. The neural folds carry them to the dorsal midline on closure of the neural tube. Neural crest cells then separate and begin to migrate almost immediately, along predetermined pathways outside the neural tube, to various parts of the body where they differentiate into a variety of tissues including membranous bone, cartilage, blood vessels, peripheral nerve sheaths, melanocytes, and adipocytes (35; 47). There are 3 sites of neural crest production: prosencephalic neural crest arises in the dorsal part of the lamina terminalis and migrates rostrally as a vertical sheet of cells in the midline; mesencephalic neural crest arises from the dorsal midline of the mesencephalon, dorsal to the cerebral aqueduct, and migrates rostrally as parallel streams of cells in the horizontal plane; and rhombencephalic neural crest arises in the hindbrain and future spinal cord and migrates widely in the periphery of the body. The neural crest is so pervasive in inducing development not only of the peripheral nervous system but also of so many other tissues that some authors even propose that its embryological status be elevated to a fourth germ layer (35).
Induction of neural crest cells is an interaction between the neural plate and nonneural ectoderm, but neural plate cells are competent to react to inductive interactions for a shorter time than nonneural ectoderm, hence there are multiple stages of neural crest induction, and the earliest stage occurs at the end of gastrulation (03). The gene Slug is expressed in neural crest formation but is later downregulated unless additional signals maintain it (03). Neural crest forms not only neural structures such as ganglion cells of the myenteric plexus, chromaffin cells of the adrenal medulla, and Schwann cells of peripheral nerves, but many non-neural structures including membranous bone of the cranial vault, vertebral spinous processes and ribs, cartilage, connective tissues, melanocytes, and adipocytes. None of these tissues differentiate until the migratory neural crest cells have reached their destination. Adipose tissue (fat) is innervated by somatosensory nerves, indicating a previously unsuspected role of dorsal root ganglia (134).
Neural crest is the means by which the neural tube induces craniofacial development. Defective prosencephalic neural crest formation accounts for hypertelorism because of failure of the intercanthal ligament to form and keep the orbits together during facial development, for the white forelock in Waardenburg syndromes, and for some frontal and nasal encephaloceles. Defective mesencephalic neural crest causes midfacial hypoplasia with hypotelorism in holoprosencephaly (94). Defective rhombencephalic neural crest formation or migration can explain neurosensory deafness in all 4 types of the Waardenburg syndromes, aganglionic megacolon (Hirschsprung disease), the “lines of Blaschko” in several neurocutaneous syndromes (eg, incontinentia pigmenti) and many hypermelanotic and hypomelanotic cutaneous lesions in many neurocutaneous disorders (97). Brain malformations in neurologic phenotypes of epidermal nevus syndrome, predominantly hemimegalencephaly, may vary despite similar or identical mutations of genetic/metabolic pathways such as mTOR, AKT, and RAS that affect neural crest (29).
Many facial dysmorphisms can be attributed to defective neural crest formation from the prosencephalic and mesencephalic neural crest that induces craniofacial development. Failure to form the intercanthal ligament, derived from prosencephalic neural crest at the dorsal part of the lamina terminals, results in hypertelorism because the orbits are not held near to the midline of the face with lateral growth of the facial bones (17).
Cellular proliferation and apoptosis. The mitotic spindle is composed of microtubules, but the chromatin phase transition prevents microtubule perforation between daughter cells (115). In the cerebrum, most neuroepithelial cellular proliferation occurs in the ventricular zone, at the margin of the ventricle. Because mitotic duplication of cells is exponential, the number of mitotic cycles needed to produce all neurons of the cerebral cortex is about 12 in the mouse and 33 in the human. Disorders that arrest mitotic proliferation (eg, some congenital viral infections or fetal neurotoxins) and cause fewer than the requisite 33 cycles are 1 explanation of microcephaly. In addition to mitoses of the periventricular neuroepithelium before ependymal differentiation, in early (precortical plate) stages extra-ventricular mitoses also occur, similarly in the murine and human brain, and do not yet respond to markers of cellular lineage, though a subset is destined to become microglial cells (16).
Concepts of progenitor cell differentiation in the embryonic neural tube are changing. It is now demonstrated, at least in the zebrafish and probably in all vertebrates, that progenitor “stem” cells as the first neuroepithelial cells in the future spinal cord become motor neurons; the second set of cellular lineage becomes glial, particularly Olig2 cells that differentiate as oligodendrocytes but at times may assume a neuronal lineage (45; 37). The neuronal-to-oligodendrocytic switch in progenitor cells is induced by mitogenic Sonic hedgehog (Shh) signals from the notochord and then from the floor plate ependyma, which produce different lineages. Cells of oligodendroglial hamartomata thus might undergo at least partial neuronal differentiation that potentially is epileptogenic. Shh transcription is continuous, not intermittent, but variations in intensity may mediate different signals for cellular lineage.
In addition to the ventricular zone or “germinal matrix” in the fetal brain, 2 sites of persistent resident progenitor stem cells are found in both the fetal and the mature brain: olfactory bulb (142; 112) and hippocampus. The latter population of bipolar progenitor cells are located in the hilus of the dentate gyrus, just beneath the inner granule cell layer, and is known as the polymorphic zone. One of its polar processes extends through the dentate gyrus. These cells are capable of generating new dentate granule cells in particular and may involve a continuous turnover of granule cells during life including new synaptogenesis, the proliferation increasing in epilepsy arising in the hippocampus (120; 119; 118). A role in the generation of new memory engrams throughout life is speculated but the evidence remains inconclusive.
Apoptosis, or programmed cell death, is a physiological process of elimination of excessive and redundant neurons after overproduction. It occurs in all parts of the CNS. Failure to arrest apoptosis after the predetermined number of neurons is achieved may result in progressive degeneration of remaining cells that are not redundant. An example is spinal muscular atrophy. Apoptosis removes the overproduced abundance of redundant spinal motor neurons, but if this normal process continues into later gestation and postnatally, there is progressive motor neuron degeneration as a spinal muscular atrophy. Some genetic mutations that cause severe infantile epilepsies, such as Dravet syndrome and epileptic (infantile) spasms, also accelerate neuron-specific accelerated apoptosis (78).
Another example of a disorder of apoptosis is found in some cases of agenesis of the corpus callosum, in which callosal axons are formed but cannot cross the midline because of a dense glial barrier, and these axons are diverted into aberrant bundles within the ipsilateral hemisphere; the glial wall at the lamina terminalis, which should undergo physiological apoptotic degeneration to enable callosal axons to cross, fails apoptosis, and the axons encounter a barrier rather than a bridge at the midline (144; 86). The first pioneer axons of the corpus callosum traverse the midline at 10 weeks gestation, but the earliest axons cross in the anterior commissure from the rostral pole of the temporal lobes and from the olfactory bulbs at 7 weeks, hence an absent corpus callosum with an intact anterior commissure means that the timing of the dysgenesis was between 7 and 10 weeks; agenesis of both forebrain commissures indicates that the timing was earlier than 7 weeks (109). Pioneer commissural axons arise from subplate neurons, whereas the permanent callosal and anterior commissural axons are of neurons in layer 3 of the mature cortex.
Ependymal differentiation. Ependymal cells of the floor plate of the future spinal cord are the first neuroepithelial cells to differentiate in the central nervous system, induced by the Sonic hedgehog gene from the notochord. These floor plate cells, unlike other ependymal cells, also express Sonic hedgehog and help induce motor neurons and other cells. The ependyma differentiates in the CNS in a very precise temporal and spatial pattern. The last ependymal cell differentiation, in parts of the lateral ventricles, does not occur until 22 weeks’ gestation. In the fetal brain, the ependyma plays an important role that is very different from its functions in the mature brain. It is advantageous to postpone ependymal differentiation because as soon as it occurs, all mitotic activity of neuroepithelial cells bordering the ventricular lumen ceases. Basal ependymal processes extend into the parenchyma and secrete molecules, such as the glycosaminoglycan keratan sulfate, that repel axonal growth cones and other molecules that attract growth cones, such as nestin and S-100beta protein (80). Ependymal processes are, thus, a major chemotactic guide for the intermediate trajectory of growing axons of long ascending and descending tracts and of axons that cross, including both decussating and commissural axons.
Abnormal ependymal differentiation may contribute to malformations. Too early differentiation arrests the mitotic cycles before the required number of neuroblasts are produced and may result in neuronal deficiency and micrencephaly (81). Such precocious differentiation is, thus, not advantageous and might be induced by viral infections and, in some, metabolic encephalopathies. If the spinal central canal is duplicated in the horizontal axis, roof plate ependymal processes do not appear to form the dorsal median septum separating the developing dorsal columns of the 2 sides, and rostrally growing axons of the dorsal columns may aberrantly decussate and confuse the brain about the laterality of sensory information (123).
Segmentation of the neural tube. After the neural tube is completely closed and neural crest tissue has begun to migrate away from it, there is a series of compartments to limit cellular migration in the longitudinal axis within the neural tube and enable the grouping of similar neurons to form neuroanatomical nuclei. These compartments are called “neuromeres.” Neuromeres of each region have more specific names: those of the spinal cord are “myelomeres,” those of the hindbrain are “rhombomeres,” the midbrain are “mesomeres,” and the forebrain including future diencephalic and telencephalic structures are “prosomeres”. The glial septa that separate the neuromeres are both physical and chemical barriers to cellular movement, and the number of neuroepithelial mitoses at the septa is lower than within the compartments.
The number of neuromeres is a still evolving concept with the recognition that not only are they anatomical compartments but that gene expression often differs between adjacent neuromeres. The spinal cord was previously thought to not be intrinsically segmented and was regarded as part of the final or 8th rhombomere in its entirety. The segmental nerve roots and extramedullary blood vessels are grouped because of segmentation of the somites outside the neural tube, not because of intrinsic segmentation of the dorsal and ventral horns. The notochord, which induces the floor plate epenydma and spinal motor neurons of the developing spinal cord after neurulation, is not a segmented structure. Parasagittal sections of the fetal spinal cord show a column of motor neurons without segmental grouping, except for increased numbers in the cervical and lumbar regions because of the development of the extremities. Nevertheless, there is grouping of autonomic spinal centres with the parasympathetic system confined to the cervical and sacral regions and the sympathetic system in the thoracic and lumbar regions. There also are differences between the spinal cord derived from primary neurulation or folding of the neural placode to form a tube rostral to the posterior neuropore, and the solid core of secondary neurulation that becomes the conus medullaris. The C4 segment of spinal cord is unique because it is the origin of more than 90% of motor neurons of the phrenic nerve to the diaphragm, but also it is the site of origin of neural crest cells that form the spine of the scapula; C4 lateral plate mesoderm contributes to the body of the scapula, and level C4 is the break-point for the transition between muscles attaching the pectoral girdle to the head (post-hyoid strap muscles) and those connecting it with the upper extremity (18). Definitive confirmatory evidence that the spinal cord is segmented is that each level is associated with a slightly different expression of genes from that of adjacent segments (18).
Rarely, an intramedullary lipoma may form within the fetal spinal cord and usually extends the entire length of that structure and appears mainly in the dorsal midline. One likely explanation is that there is a defect in the peripheral migration of rhombencephalic neural crest so that some neural crest cells remain in the dorsal midline of the neural tube at the time of neurulation and become entrapped within the neural tissue. They differentiate into adipose cells, the most frequent fate of neural crest near to or within the neural tube, which also may explain some lipomengomyeloceles.
Several disorders of segmentation of the neural tube are well-documented genetic defects, and others are suspected but not yet proved. Absence of entire neuromeres may occur, for example, agenesis of the mesencephalon and metencephalon (rostral pons) if there is a mutation in WNT or EN1 or EN2 (Engrailed) genes (87). Upregulation of certain genes in 1 neuromere can cause “ectopic expression” in another, wrong neuromere and cause rostral structures to be produced in a neuromere of more caudal structures. (In molecular genetic jargon, the term “heterotopic” does not exist.) Chiari malformations may be a result of this molecular genetic defect of segmentation of the hindbrain and cerebellum (88).
Cellular lineage and cytological maturation. Individual neuroepithelial cells differentiate as neurons, glial cells, and ependymal cells. Further subdifferentiation ensues into specific types of neurons with unique projections, transmitter biosynthesis and membrane receptors, glial differentiation into various types of astrocytes (including specialized forms, such as Bergmann glia of the cerebellum and radial glia of the fetal cerebrum), and ependymal differentiation with specialization of floor plate and roof plate cells. Cellular differentiation from pluripotential stem cells is genetically determined, but acquired lesions of the fetal brain, such as ischaemia or infarction, may redirect cellular lineage by activating different genes.
Some malformations are primary disturbances of cellular lineage, with resulting cells exhibiting poorly regulated growth and shape, having multiple nuclei, and coexpressing both neuronal and glial proteins in their cytoplasm, as well as some abnormal proteins, such as phosphorylated tau, a microtubule-associated protein. Examples are tuberous sclerosis, focal cortical dysplasia type II, and hemimegalencephaly (100; 11). One difference between focal cortical dysplasia type II and hemimegalencephaly is the extent, not the nature, of the dysgenesis, and this is related to timing in the 33 mitotic cycles needed for to produce all cortical neurons, smaller lesions occurring in late cycles and larger lesions occurring in early cycles (100; 109). Secondary disturbances of neuroblast migration may occur in association with primary disorders of cellular lineage because both the neuroblast and the radial glial cell are abnormal (31).
Multipotential “stem cells” are found throughout the periventricular region of the lateral ventricles and the external granular layer of the cerebellum in the fetus and young infant, but there are 2 sites in the mature brain where a population persists of stem cells traditionally thought to be capable of being activated to differentiate as new neurons: the olfactory bulb and the dentate gyrus of the hippocampus (119; 112). However, evidence indicates that hippocampal resident “stem” cells are less able to differentiate pluripotentially with aging and especially in adult life. Embryonic stem cells can create copies of themselves and can mature into almost any type of cell through gene regulators called microRNAs (122). Neural stem cells divide asymmetrically and often amplify the number of their progeny via symmetrically dividing intermediate progenitors (56). The conversion of mature fibroblasts to functional neurons also is possible (133); the transformation of 1 mature tissue into another mature tissue, such as esophageal epithelium to gastric mucosa in Barrett esophagitis, has been known for several decades and is known as “metaplasia.” These topics of stem cells in the CNS and cellular reprogramming to redifferentiate as other cell types are of intense interest in the search for a means of enhancing regeneration of the damaged brain in children and adults by regenerating neurons. No examples are known of direct metaplasia within the mature brain.
Another aspect of abnormal cytological differentiation and lineage is asymmetrical occurrence of this aberration. The prototype example is hemimegalencephaly, with enlargement and dysplasia of only 1 of the cerebral hemispheres, occasionally also involving half of the brainstem and cerebellum on the ipsilateral side, that can be confirmed during life by neuroimaging after suspecting an asymmetrical abnormality because of asymmetrical head shape or facial features of the neonate or by partial seizures emanating from 1 hemisphere (26; 27). Another example is the dysplastic gangliocytoma of the cerebellum (Lhermitte-Duclos disease), which is a hamartomatous nodule of dysplastic cerebellum, also with neurons showing a disturbance of cytological lineage, that occurs in 1 cerebellar hemisphere only and may be confused by imaging with a cerebellar neoplasm (79).
Neuroblast migration. Few neurons of the human brain mature in the site where they differentiate from pluripotential neuroepithelial cells. All primordial neurons must move within the parenchyma of the neural tube to different sites to enable them to create the synaptic connections and circuitry needed for function. This cellular movement in the embryonic and fetal brain is called “migration” and is a precise time-linked and genetically programmed process. Most neuroblasts reach the cortical plate (future cerebral cortex) by radial migration from the subventricular zone. They travel along the outside of a “monorail” of radial glial cells whose cell body is in the subventricular zone and whose long process extends centrifugally through the intermediate zone (white matter) and cortical plate to the pial surface. The earliest arrivals of the waves of radial migration become the deepest layer of the mature cortex, layer 6, because they are displaced downward by the next wave of arrivals, and layer 2 represents the last waves of radial migratory neuroblasts. Radial migratory neuroblasts arise mainly in the dorsal ganglionic eminence and become glutamatergic excitatory neurons as the cortex matures.
About 18% to 20% of neurons in the cerebral cortex arrive by a different route—tangential migration from the medial ganglionic eminence (part of the germinal matrix)—and these cells migrate along preformed axons rather than radial glial fibers (41). They become the GABAergic interneurons of the cerebral cortex. The medial ganglionic eminence also gives origin to neurons in the future corpus striatum and globus pallidus, and these precursor neuroblasts in human early fetuses already express several specific developmental genes, such as DLX, PAX6, and NKX2 before even initiating their migration, thus identifying their intended lineage and fate (67).
In the cerebellar cortex, Bergmann glial cells that occupy the Purkinje cell layer between Purkinje neurons have long centrifugal processes that also extend to the pial surface of the cerebellum and serve as guide fibers for the external granular cells that need to move into the internal granular layer beneath the row of Purkinje cells. Other radial guides are present in the brainstem (eg, the migratory pathway beneath the pia mater from the dorsal rhombic lip at the margin of the 4th ventricle) to carry neuroblasts to the ventral position of the inferior olivary, arcuate, and pontine nuclei.
Adhesion molecules are extracellular matrix molecules secreted by radial glia that attach premigratory neuroblasts and later glioblasts to the radial glial fiber for transport. Aberrations in their molecular structure impair this function and neuroepithelial cells may not attach and may mature in the periventricular region to become periventricular nodular heterotopia. At the other end, disadhesion molecules are needed to detach neuroblasts from the radial glial fiber at the cortical plate, so that the line of neuroblasts behind on the same radial glial fiber may bypass them to reach the surface for the inside-out architecture of the cortical plate. Failure to detach results in dyslamination and disorganization of the cortex.
More genes have been identified that mediate neuroblast migration than for almost any other developmental process. Disorders of these genes may arrest migration at various stages of their migratory trajectory or in the architectural arrangement of the cortical plate. Thus, neurons may never migrate and then mature in their place of birth in the subventricular region (periventricular nodular heterotopia); they may migrate partially and stop within the white matter (subcortical laminar heterotopia or “band heterotopia”), or they may form abnormal cortex without the normal lamination and with many displaced and disoriented neurons. Overmigration may occur if there is a breach in the pial membrane or if the subpial glial layer of Brun is deficient, in which case individual neurons or clusters of neural cells form “glioneuronal heterotopia” within the leptomeninges, usually attached to the brain by a stalk (125). Lissencephaly is 1 of the most recognized cerebral malformations because of its characteristic clinical presentation, clear diagnosis by MRI, and neuropathology, and the associated multiple and well documented genetic mutations or microdeletions. Spatiotemporal gene expression trajectories of radial glia have developmental hierarchies with lineage-specific expression of neurogenic transcription factors and along gradients (63).
Neurons are mature cells that do not and cannot migrate; hence, the often-used term “neuronal migration” is technically incorrect. Postmitotic and undifferentiated neuroepithelial cells that are committed to a neuronal lineage are called “neuroblasts,” and their migration to their mature site within the brain is more properly “neuroblast migration.” Additional confusion arises because the term “blast cell” is used differently in hematology, referring to cells still capable of mitosis and usually neoplastic; hence, erythroblasts and leukoblasts are not parallel to neuroblasts.
If a lesion occurs in the fetal cortical plate, such as an infarct or even a developmental lesion such as a focal dysplasia, subventricular zone neuroblasts emigrate toward that cortical lesion in an attempt to repair damage that can only be done before neuroblast migration is completed (126).
Some malformations are attributed to neuroblast migratory disorders (often erroneously called “disorders of neuronal migration”). However, neurons are mature nerve cells that cannot and do not shift their position without neuropathological evidence that indeed they are not migratory disorders but rather abnormalities of organization of the cortical plate and eventual cerebral cortex. An example is polymicrogyria (90).
Nuclear and cortical architectures. The architecture of the mature vertebrate brain may be classified into 2 major types: nuclear and laminar. Nuclear architecture is represented by the compartmentalized neuronal aggregates in the brainstem that form cranial nerve nuclei. A similar architecture prevails in the thalamus. The aggregates of neurons within nuclei are not randomly disorganized but often follow an orderly, predictable somatotopic pattern. Laminar architecture is layers of neurons of the same type with perpendicular synaptic connections between the layers. Examples of laminar architecture are found in the cerebral neocortex, cerebellar cortex, olfactory bulb, and superior colliculus. Both architectural types may serve equally well from a functional standpoint. The organization of the reptilian and avian forebrain is largely nuclear in regions corresponding to mammalian laminar neocortex, such as the visual striate cortex. Laminar architecture of the human cerebral cortex is preceded in fetal life by an immature radial columnar architecture that is transitory, and the lamination is superimposed after midgestation.
Each layer of the laminae contains neurons of the same type, and synaptic circuitry is mainly between layers. In the hexalaminar cerebral neocortex, layers 3, 5, and 6 are pyramidal neurons with long external projections: axons of layer 3 form the forebrain commissures, corpus callosum, and anterior commissure; axons of layer 5 project axons caudally to form the corticospinal and corticobulbar tracts and the corticopontine tracts to the basis pontis for the corticopontocerebellar pathway; layer 6 forms the short association U-fibers that interconnect sites within the same and immediately adjacent gyri and also projects axons to the thalamus and corpus striatum. Layers 2 and 4 are granular neurons and are sensory for the visual, auditory, somatosensory, and other perceptive systems. Layer 1 is mostly synapses and axons but contains the Cajal-Retzius neurons of the fetal brain that form the first intrinsic cortical circuits by making synapses with neurons of layer 6 and expresses the reelin (RELN) that regulates cortical plate organization (Sarnat and 26b). Prior to the histological layering that becomes evident in the second half of gestation, mRNA studies show that there is layer specificity for specific types of neurons that is not evident histologically (36). Even undifferentiated neuroepithelial cells in the ventricular and subventricular zones show evidence of predestined differentiation as neurons of a specific type in a specific layer of cortex, but these primitive cells remain flexible to change their fate under pathological conditions.
Beneath the developing cortical plate throughout the second and early third trimesters is a transitory “subplate zone” containing large multipolar neurons. Some neurons mostly disappear in early postnatal life but are important in the fetus in organizing cortical connections in the developing cerebrum and in forming interconnections between thalamus and the cortical plate (Sarnat and 26b; 42). Other subplate neurons do not disappear as does the subplate zone, but are incorporated into layer 6, and still others may contribute to the neuronal dispersion in the U-fiber later as the cortex matures. The U-fiber layer develops beginning at midgestation; its neurons, by contrast, are local short connections that do not project subcortically or in commissures (113). The subplate zone is structurally laminated as 2 layers at midgestation, which can be recognized postmortem in 3-Tesla MRI and confirmed neuropathologically; presumably the 2 layers are demonstrable prenatally by fetal MRI (70).
Axonal pathfinding. The axons begin sprouting from the neuronal soma during its migratory journey, before dendritic development is initiated. Both short axons destined for nearby targets and long axons destined for distant targets must be guided during their intermediate trajectory and finally attracted to the neuronal targets on which synaptic connection will occur. The intermediate trajectory is guided largely by chemotaxis from molecules secreted by ependymal processes and in the matrix, molecules that either attract or repel the growing axonal tip, named by Ramón y Cajal “le cône d’accroissement” (“growth cone”). An example is the ascending growth cones of the dorsal column axons of the spinal cord. To ensure that these axons do not meander to the opposite side of the spinal cord and confuse the brain about laterality of sensory impulses, they are prevented from aberrant decussation by the dorsal median septum, a midline structure formed by basal processes of the roof plate ependymal cells. This septum acts not only as a physical barrier but also as a chemical barrier to axonal decussation by secreting the glycosaminoglycan keratin sulfate that strongly repels growth cones (123). A similar functional role is played by the ventral median septum of floor plate ependymal processes. Decussation of longitudinal tracts is prevented, but commissural axons are not repelled and, indeed, actually are facilitated by other attractant molecules, including nestin and S-100β protein.
The growth cone at the tip of the growing axon has constantly retracting and extending points called “filopodia” and a flat membrane between them called “lamellopodia.” Filopodia contain microtubules and microfilaments of actin tightly packed in parallel, which shift to one side if the growth cone is attracted to deviate toward a particular neuronal target; neurofilaments do not enter the growth cone until the target is closely approached and synaptogenesis is about to occur (91). Lamellopodia also have microfilaments of actin, but they are randomly oriented. Other components of growth cones include endocytes that contribute to axonal polarity (25), mitochondria, and growth factors and receptors.
These same proteoglycans are formed in layer 6 of the cerebral cortex and in the underlying U-fiber layer and serve to repel axons from deep white matter heterotopia, thus preventing them from integrating with abnormal epileptic networks (113). The same keratan sulfate that prevents aberrant decussation of ascending axons in the dorsal columns of the spinal cord also is present in the forebrain (89). In the cerebral cortex, it first appears in the molecular zone at midgestation, then has a transitory patchy distribution within the cortical plate and eventually is most concentrated in the deep cortical layers 5 and 6 and in the U-fiber layer, where it prevents axonal penetration from deep white matter heterotopia that mature late and project potentially epileptogenic axons after the keratan sulfate U-fiber barrier has formed. Keratan sulfate is mostly in the intercellular matrix but binds to neuronal somatic membranes, but not to dendritic spines, where it repels glutamatergic axons and facilitates GABAergic axons, explaining why axosomatic synapses are inhibitory and axodendritic synapses are mostly excitatory (89). Furthermore, keratan sulfate envelops axonal fascicles, including long tracts such as the corticospinal and short tracts such as the intrinsic axonal bundles within the globus pallidus and the pencil bundles of Wilson in the corpus striatum. This keratan “insulation” of fascicles and tracts ensures that axons cannot exit until they reach their destination and that axons from grey matter en route cannot enter the fascicles, keeping central axonal fascicles “pure” (89).
Axons crossing in the ventral commissures of the fetal rat spinal cord express L1 and GAD65, similar to axons of GABAergic neurons elsewhere in the CNS (131). Certain molecules residing within the extracellular matrix serve a similar function. Boundary regions between adjacent domains of regulatory gene expression influence where the first axons will extend (136).
Dendritic ramification, spine formation, and synaptogenesis. The dendritic tree usually does not form until the neuron has completed its migration and is in its definitive position, unlike axonal sprouting. The formation not only of dendritic branches but also spines on each of these branches to create an increased surface area for sites of synaptic contact is part of this maturational process. The postsynaptic membrane of the dendrite (as well as synaptic sites on the soma) must have receptors for the specific type of neurotransmitter that its afferent axons release into the synaptic cleft.
Synaptogenesis follows a precise temporal and spatial pattern within the CNS, and the gestational age of a fetus can be determined from its synaptic pattern. Synapse deletion and reformation is a continuous process throughout not only fetal but also postnatal and adult life, and this ongoing change in synapses in the hippocampus, for example, enables the formation of new memories while retaining old memories. This is 1 of the most “plastic” features of the mature brain. Synaptic interactions initially form local circuits or plexi (68) and later extend these plexi to more distant sites in forming networks.
All regions of the nervous system may be divided into those with synaptic stability at maturity, defined as little or no synaptic remodeling, and those with synaptic plasticity, defined as undergoing continuous retraction of synapses and creation of new synapses (Sarnat and 29c). Examples of structures with synaptic stability are autonomic ganglia, spinal cord, brainstem cranial nerve nuclei, basal ganglia, and thalamus. Examples of synaptic plasticity are cerebral cortex, hippocampus, amygdala, retina, and olfactory bulb. Regions with synaptic stability are not epileptogenic and most with synaptic plasticity are capable of generating seizure activity, with or without anatomical malformation. Animal models and even the simplest, most primitive nervous systems of simple species demonstrate principles in evolution that are conserved and apply to human brains. For example, “connectisomes” or synaptic circuits during development in roundworms demonstrate network changes between individual neurons that demonstrate plasticity of maturation (138).
Neurotransmitter biosynthesis and membrane receptors. The features that distinguish a neuron from an immature pre-neuron or neuroblast are an electrically polarized membrane with receptors and ion channels and the biosynthesis of secretory products that, in the case of nerve cells, are neurotransmitters of any type. The initiation of transmitter synthesis is, thus, a landmark developmental milestone in the maturation of neurons. Which specific chemical transmitter is produced by each neuron is a function of neuronal specificity in differentiation.
Flexures, fissures, and sulci. Five fissures and more than 30 sulci are found in the human forebrain. The main difference between fissures and sulci, both grooves in the surface of the brain, is not the features of secondary importance (such as the generally earlier formation of fissures or that many are deeper than sulci), but the manner in which they form. Both are genetically programmed and relatively constant from brain to brain and predictable at any gestational age. Despite their genetic predetermination, both fissures and sulci form from mechanical forces, but fissures result from external forces, including the ventricular system that is enclosed by, but technically outside, the brain; sulci result from internal forces. The 5 fissures of the forebrain are:
(1) Interhemispheric fissure - forms at 4.5 weeks
(2) Choroidal fissure – forms at 5 weeks
(3) Hippocampal fissure – forms at 5 to 6 weeks at the site between the forming dentate gyrus and Ammon’s horn
(4) Sylvian fissure - forms from the telencephalic flexure, the ventral bending at 8 to 9 weeks of the posterior third of the primitive telencephalon so that the posterior pole of the telencephalon of 6 weeks becomes the temporal, not the occipital lobe
(5) Calcarine fissure – forms at 10 to 12 weeks on the medial side of the occipital lobe after the telencephalic flexure has progressed in forming.
The telencephalic flexure forms the dorsal (frontal) and ventral (temporal) lips of the future lateral cerebral (Sylvian) fissure so that both lips represent the ventral surface of the primitive telencephalon. This derivation means that genetic disorders with a ventrodorsal gradient in the vertical axis affect both the frontal and temporal lips, as well as the deeper insula that sometimes is regarded as a third lip of the lateral cerebral (Sylvian) fissure (Sarnat and 29b). The insula is derived from the ventral part of the primitive telencephalon and folds deeply into the operculum, and remains exposed by the separated frontal and temporal lips that are not called the lateral cerebral (Sylvian) fissure until these lips close at near-term to hide the insula from external view. The dorsal part of the primitive telencephalon becomes the parietal and occipital lobes. The ventral bending of the telencephalic flexure also contributes to the posterior “rotation” of the early hippocampus to a more ventral position (Sarnat and 29b).
In alobar and semilobar holoprosencephaly, the telencephalic flexure fails to occur and the hippocampus remains a dorsal structure that is abnormally continuous across the midline. In lissencephalies 1 and 2, the telencephalic flexure may be abnormal. In schizencephaly, the telencephalic flexure is always abnormal, unilaterally or bilaterally.
The formation of sulci and gyri of the cerebral cortex, and of folia of the cerebellar cortex, occur in the second half of gestation in a predictable sequence. Because these cortices are laminated structures histologically, it is advantageous to increase the surface area without a concomitant increase in tissue volume and size. Gyration and foliation are an efficient mechanism to accomplish this purpose. Nature uses the same principle in absorptive and secretory surfaces: the rugae of the gastric and villi of the intestinal mucosa, alveoli of the lungs, and the gills of fish.
The formation of convolutions is abnormal in many malformations, particularly those of neuroblast migration. Lissencephaly is a smooth brain without convolutions, similar to the smooth brain of the fetus before midgestation. Pachygyria refers to gyri that are too wide and too few. Polymicrogyria are gyri that are excessively small and numerous. The excessively small gyri may have normal lamination with a reduced number of neurons or exhibit dyslamination with poorly formed architecture and displaced and disoriented neurons. One of the most striking features of polymicrogyria is that along the lateral surface of the microgyria are multiple gaps or discontinuities of the pia mater so that adjacent microgyria are fused at their molecular layers (125; 40). The fused gyri enable synaptic short-circuiting between them and contribute to epilepsy.
Abnormal convolutions may be generalized and involve the entire cerebral cortex, or they may show a type of gradient with more severe involvement of the frontal lobes than the occipital lobes or predominant severity in the parieto-occipital and posterior temporal lobes, or the abnormality of gyration may be focal and limited to 1 region of the cortex in 1 hemisphere as part of a focal cortical dysplasia.
Myelination. This process is amongst the last to be initiated, beginning at 24 weeks’ gestation in the spinal nerve roots, some cranial nerves, and some longitudinal tracts within the brainstem, such as the medial longitudinal fasciculus (Yakovlev and Lecours 1967; 43). “Myelination cycles,” the interval between the initiation and completion of myelination within a given white matter pathway, vary greatly from short cycles: nerve roots complete myelination in a few days to 2 weeks; the corticospinal tract begins its myelination cycle at 38 weeks gestation and it is not complete until 2 years of age; the cycle of the corpus callosum begins at 2 months postnatally by tissue stains (4 months by MRI detection) and is not complete until mid- to late adolescence; the last pathway to fully acquire myelin within the brain is the ipsilateral fronto-temporal association bundle, at 32 years of age (Yakovlev and Lecours 1967; 14). U-fibers surrounding cortical gyri and consisting of short horizontal connections of the gyrus generally myelinate later than the deep white matter consisting mainly of ascending and descending long tracts. Despite these disparities, the timing of myelination is as precise as all other developmental processes, enabling a diagnosis of delayed myelination by T2-weighted MRI in the living infant and child (32). Communication between axons and oligodendrocytes is another factor regulating myelin formation (37).
Delayed myelination is a feature of many metabolic diseases of the nervous system, in many malformations and in acquired lesions. Germinal matrix hemorrhages in the preterm neonate reduce the proliferation and migration of oligodendrocytes much more than of astrocytes, and a paucity of oligodendrocytes may explain delayed myelination in infants who have suffered such a complication, either prenatally or postnatally. Myelination can continue for many years; hence, an arrest in myelination is much less likely than delay. They also should be distinguished from “dysmyelination” (the formation of myelin with abnormal molecular structure in certain hereditary metabolic diseases, such as the leukodystrophies) and “demyelination” followed by remyelination in multiple sclerosis, neuromyelitis optica, acute demyelinating encephalomyelitis (ADEM), postinfectious demyelination, and other diseases believed to be autoimmune in nature.
The development of the vascular system of the CNS, both the major arteries and venous sinuses outside the brain and the parenchymal vessels within the brain, is as precisely regulated as are other processes of neural development. The fetal circulation is not the same as in the adult. For example, the basilar artery initially flows rostral to caudal and receives blood from the carotid arteries through a series of transitory communicating arteries that later regress and disappear but may persist in pathological situations. The basilar artery begins as a parallel pair of vessels at the base of the rhombencephalon, the longitudinal neural arteries that secondarily fuse to form a single midline basilar artery. The middle and anterior cerebral arteries form from separate primordial neurons and finally fuse with the supraclinoid segment of the internal carotid artery. The microcirculation within the brain also changes with maturation. There are exactly 3 branches between the arteriole and the capillary in brain parenchyma. As with other developmental processes in the brain, genetic regulation of such molecules as the family of vascular endothelial growth factors determines angiogenic sprouting and the formation of capillary networks (128) and development of the blood-brain barrier (38).
Defects in neural development may be caused by vascular malformations that alter hemodynamics in the delivery of oxygen and nutrients to cells within the brain (66). A transitory period of systemic hypotension in fetal life also may cause watershed infarcts not only in the cerebral cortex but also in the brainstem where Möbius syndrome may result and be misinterpreted as agenesis of cranial nerves (83). Supraclinoid internal carotid stenosis and thrombosis or dissecting aneurysm may result in later childhood from imperfect fusion of the middle and internal carotid artery with a defective internal elastic membrane at that site, a congenital rather than an apparent acquired vascular lesion (01).
A few events are brief and affect development only in a limited period of time. Examples are a transient episode of ischemia or of exposure to a teratogenic neurotoxin. Most adverse exogenous effects on the developing neural tube are not so precisely time-limited, however, and continue to act throughout gestation. Examples are continued or repeated exposure to teratogenic drugs or toxins, congenital infections that are active throughout the pregnancy, or genetic defects that remain defective throughout gestation. The timing of malformations, thus, is an attempt to identify the earliest event. An example is agenesis of the forebrain commissures. Agenesis of the corpus callosum, a common malformation, can be timed to 74 gestational days when its first axons cross the midline between the cerebral hemispheres. If the anterior commissure also fails to form, the timing is 3 weeks, at 54 days, when the first axons of this more ventral commissure initially cross. If the anterior commissure is absent but the corpus callosum is present, the event is limited in timing to before 54 days but not extending to 74 days.
Timing in mitotic cycles also determines the extent of hamartomas in focal cortical dysplasia type II and hemimegalencephaly (100; 109). The neuroembryological demonstration that they are merely different timing of onset of expression in the same genetic/metabolic disease has been confirmed by another approach of genetic studies (48; 22; 23). Delayed synaptogenesis or myelination in the nucleus/tractus solitarius, the central pneumotaxic center, may be the basis of apnea of prematurity, central hypoventilation in preterm and term neonates, and some cases of sudden infant death syndrome (SIDS) (102).
A distinction between delay and arrest of maturation depends on the expectation that further development will proceed in the future or that the process is now static and will remain unchanged. Three factors determine this distinction: (1) whether the insult that caused the slow development, whether genetic or acquired, is continuing to be active and influential on further maturation; (2) the nature of the process; and (3) the time interval between the normal phases of development represented by that process and the present time. Some developmental genes are active only for a brief period of ontogenesis, and others continue to act throughout gestation and, indeed, postnatally. Some infections that may be teratogenic are self-limited whereas others continue to actively infect cells. Defective development of some processes, such as neurulation and neuroblast migration, is more likely to be due to maturational arrests with little potential for correction with more time. Others, such as synaptogenesis and myelination, may be immature for a given gestational age but have a great potential for proceeding further to eventually approach normal. The importance of time interval from the normal condition at an earlier gestational age and a present observation may be exemplified by persistent columnar architecture of the cerebral cortex. Radial columnar architecture is the organization of the cortical plate at midgestation, and histological layering or lamination is then superimposed in the second half of gestation. A term neonate whose cerebral cortex shows a strong fetal columnar architecture may have a maturational delay, and the reorganization of the cortex may proceed postnatally to eventually develop normal lamination. The finding of fetal columnar cortical architecture in a 5-year-old child, by contrast, may be considered maturational arrest because if it has not matured further by that age, it is extremely unlikely to do so in future.
Maturational delay and arrest may be applied not only to anatomical developmental processes but also to physiological and functional development. The invariable columnar synaptic architecture of the visual cortex is imposed by on/off circuitry, alterations of which may influence timing and can cause delayed maturation (49). The EEG of preterm infants shows a precise time-linked progression from a discontinuous pattern of semiperiodic bursts of activity to continuous cortical activity; some specialized features do not appear until certain ages, such as sleep spindles that develop at 6 weeks postnatally or the occipital alpha rhythm that develops at 2 years of age. The EEG represents wide fields of cortical synaptic activity, and EEG maturation in the premature infant correlates well with synaptogenesis in the cerebral cortex demonstrated by immunoreactivity of the synaptic vesicle protein synaptophysin in human fetuses and preterm neonates (107). An EEG in early infancy that is immature for conceptional age, thus, may indicate a maturational arrest that will be corrected as cortical synaptogenesis proceeds and is not arrested development for all time. T2-weighted MRI images often define a delay in myelination in particular white matter structures of the brain in infants and children, but myelination may continue, and eventually this delay will be abolished. Another example of physiological developmental delay is a delayed onset of puberty, especially in girls, associated with many brain malformations and fetal brain damage. Less commonly, precocious puberty may appear. Either may be due to abnormalities in the regulation of gonadotropic hormones in the hypothalamus, influenced by other structures of the brain. Fetal brain development may be arrested in some congenital disorders of glycosylation, resulting in microcephaly and severe psychomotor delays postnatally with major neurologic disabilities (60).
Alterations in timing not only can produce maturational delay or arrest, but in some genetic malformations in particular can cause precociousness of synaptogenesis, out of synchrony with other developmental processes of neuronal maturation. An example is holoprosencephaly, in which most fetuses show precocious synapse formation in the cerebral cortex and even the retina of a cyclopean eye before the expected physiological timing of its appearance; this precociousness is not advantageous, on the contrary, it may lead to early development of epileptic circuitry and severe infantile epilepsies postnatally (98; 111). Delay in myelination is a feature shared by many inborn metabolic diseases affecting the nervous system, such as aminoacidurias, organic acidurias, cerebral lipidoses, and leukodystrophies.
Acquired disturbances of fetal life that are not genetic in origin also may result in maturational delay; an example is delayed synaptogenesis of the cerebral cortex in fetal alcohol spectrum disorder (100). Prenatal ischemic encephalopathy, intrauterine growth restriction due to placental insufficiency, and microcephaly due to congenital infections, such as cytomegalovirus and Zika viruses (57; 75) are other examples. Congenital viral infections of the brain usually are associated with multiple infarcts, both large and small, because of endothelial cell involvement. Cerebral infarcts in the fetus due to other causes also can alter the timing of brain maturation and interfere with processes such as neuroblast migration by interrupting radial glial fibers.
The terms maturational delay and arrest also are used to describe functional development. An example is the child with delayed speech development who later learns to talk. Other children with brain damage, particularly involving speech centers in the left hemisphere, never talk and learn to communicate nonverbally unless they are profoundly retarded. A whole pediatric subspecialty of “developmental pediatrics” is predicated on the assessment of functional delays and arrests in various aspects of development in infancy and childhood. Of course, there is always a structural anatomical or physiological basis for such delays, including genetic and chromosomal defects, but the functional assessment is important, regardless of cause, because it helps determine effective treatments, such as speech therapy, to minimize the functional disability and encourage progression in developmental delays.
Focal cortical dysplasias. Focal cortical dysplasias are frequent and represent the most common cause of intractable focal epilepsy in infants and children. The International League Against Epilepsy neuropathological classification scheme (13) is currently under revision (61; 58). Experience with tissue from cortical resections as surgical treatment of focal epilepsy has rapidly expanded since 2011 and many new data on developmental neuroanatomy and genetics are now available that contribute greatly to understanding the pathogenesis of these focal disorders of cortical development, previously also termed microdysgenesis. There is now good agreement between the neuropathology and genetics of focal cortical dysplasia type 2, though the genetics of type 1 is not yet elucidated (10).
Types 1 and 2 focal cortical dysplasia both exhibit abnormal cortical lamination and organization of neurons in various patterns but are distinguished mainly by neurons with normal morphology and size in focal cortical dysplasia 1 and clones of abnormal dysmorphic and megalocytic neurons in type 2 (11; 12). In focal cortical dysplasia subtype 2b and also in hemimegalencephaly and in cortical tubers of tuberous sclerosis complex there are balloon cells in addition. Balloon cells are large globoid cells with coarse processes; they are of mixed cellular lineage, expressing both neuronal and glial proteins as well as primitive filament proteins of progenitor cells, such as nestin and vimentin (31). Cultures of isolated balloon cells show similar reactivities as in tissue sections, and a subpopulation also is labelled with β1 integrin, a marker of progenitor stem cells, interpreted that balloon cells are nonneoplastic stem cells that have failed to differentiate (143). Balloon cells are found in all layers of the cortex including the molecular layer 1 and also in subcortical deep white matter heterotopia (11).
The clinical phenotypes of focal cortical dysplasia 1 and focal cortical dysplasia 2 have focal epilepsy as the principal common feature but there also are differences between them; though focal cortical dysplasia 2 is more severe histopathologically, surgical resections are more likely to be effective because these lesions are smaller and resections are thus more likely to be total rather than subtotal (124). EEG cannot distinguish these 2 types but MRI usually shows a lesion at the base of a gyrus or depth of a sulcus or a transmantle dysplasia in focal cortical dysplasia 2. In focal cortical dysplasia 1 a lesion often is not demonstrated and the MRI is called “nonlesional”, but histopathologically the lesion can be demonstrated and does not show a signal change in MRI because the overall cellular density is unchanged despite the radial architecture.
Type 1a focal cortical dysplasia is characterized microscopically by radial microcolumnar architecture rather than horizontal lamination (21). Not only are neurons in radial columns, but radial synaptic layers occur between the microcolumns of neurons, unlike the horizontal lamination of synaptic activity that occurs in normal fetal cortical maturation (101). This radial pattern is the normal architecture of the cortical plate in the first half of gestation in the human fetal brain, suggesting that its pathogenesis may be a maturational arrest. This same pattern is seen in a generalized distribution throughout all lobes of the cortex and also occurs focally in ischemic cortical zones adjacent to porencephalic cysts classified as focal cortical dysplasia type 3d (101). It was demonstrated that some cases of focal cortical dysplasia 1 are due to genetically defects in glycosylation of the X-linked SLC35A2 gene (137). The radial microcolumns of neurons alternate with abnormal radially oriented laminae of synapses (101) and abnormal radial distribution of keratan sulfate (89).
Type 2 focal cortical dysplasia is a postzygotic somatic mutation involving the mTOR signaling pathway and related pathways including PIK3A, AKT, and GATOR families, each giving a somewhat different clinical phenotype but similar histopathological findings in the cerebral lesions (69; 40; 02). AKT is the genetic basis of hemimegalencephaly associated with several neurocutaneous syndromes, particularly the epidermal nevus syndromes and their neurologic phenotypes including Proteus syndrome (28; 29), which also results in focal somatic overgrowth in the extremities and viscera.
The differentiation of cytomegalic dysmorphic neurons is of primordial importance in distinguishing focal cortical dysplasia type 2 and other disorders of the mTOR signaling pathway. Focal cortical dysplasia type 1, by contrast, has abnormal cortical lamination but individual neurons are normal. The clones of individual neuroepithelial cells with a somatic mutation will all carry the same mutation and will differentiate similarly. Accurate neuropathological diagnosis of mTOR pathway disorders and genetic correlations (132; 53) also is now even more compelling for its therapeutic importance as “targeted therapy” because mTOR inhibitors such as rapamycin are proving effective in delaying, arresting, or even potentially reversing the overgrowth of cells and the mass volume of brain lesions (65; 140). Genomic DNA methylation can distinguish subtypes of human focal cortical dysplasia (44).
Tuberous sclerosis complex is another mTORopathy in which cortical tubers are very similar histopathologically to the lesions of focal cortical dysplasia 2 and hemimegalencephaly. Tuberous sclerosis complex is genetically more complex, however, because it involves both germline and postzygotic somatic mutations. Somatic mutations in the TSC1 and TSC2 genes are implicated not only in cortical tubers in tuberous sclerosis complex, but also in some focal cortical dysplasias without tuberous sclerosis complex, providing a further genetic/metabolic link to explain the morphological similarities in these disorders (34; 52). The mutation-induced hyperactivation of the mTOR pathway was caused by disruption in the formation or function of the TSC1-TSC2 complex germline mutations (34; 145). The relation between the germline mutations and postzygotic somatic mutations in tuberous sclerosis complex has intrigued geneticists, neuropathologists, and epileptologists for years, but now germline mutations also are demonstrated in a minority of patients with focal cortical dysplasias, thus broadening the concept (39). It also provides insight into the etiology of familial cases of focal cortical dysplasia (51). It remains uncertain why some cortical tubers are highly epileptogenic and others are not despite similar size and indistinguishable histopathology.
All genetic/metabolic disorders of the mTOR pathway including PI3K/AKT are associated with upregulation of an abnormally phosphorylated tau protein, a microtubule associated protein of the cytoskeleton that in early neural cell differentiation regulates cellular growth and morphology; dysregulation results in megalocytic dysplastic neurons and glial cells and also produces balloon cells (106; 93; 11; 58). In murine models of mTORopathy, tau reduction decreases the frequency of seizures and also autistic features; this reduction is achieved by disinhibition of phosphatase and tensin homologue deletions on chromosome 10 that affects PTEN, a negative P13K regulator controlled by tau (127). It offers a potential therapeutic application in human patients with brain malformations of mTOR metabolism.
Glial participation in focal cortical dysplasias is another factor that has not received the same attention as the neurons. Cytomegaly of glial cells in focal cortical dysplasia 2 is well documented. Glial-neuronal interactions in focal cortical dysplasia 2 are less well understood, as with dysplastic neurons and balloon cells in terms of their epileptogenic potential (19). In focal cortical dysplasia 2a, oligodendrocyte proliferation is impaired and maturation delayed so that myelination also is delayed (116; 24). This also might be a factor in the pathogenesis of subcortical white matter oligodendroglial hyperplasia (mild malformation of cortical development with oligodendroglial hyperplasia) associated with some minimal focal cortical dysplasias (117).
To understand the problem, one must begin with basic definitions for purposes of clarity of understanding and communication among colleagues. Highly specific terms that denote a precise developmental anomaly should not be degraded into generality by incorrectly using them interchangeably with others when subtle but important differences in meaning do exist. For example, the embryological definition of “ectopic” is in reference to a cell that is displaced outside its organ of origin; “heterotopic”, by contrast, is displacement of a cell within its organ of origin. If the brain is regarded as the organ of origin, a neuron within the leptomeninges is ectopic and a neuron displaced within the white matter is heterotopic. The 2 terms are not identical and interchangeable when used in an anatomical context; molecular genetics, however, does not use the term “heterotopic”; hence, genetic expression in an abnormal site within the neuraxis (in the wrong neuromere) is called “ectopic expression.” “Heterotopia” is already the plural form of this Greek derivative (singular, “heterotopion,” analogous to mitochondrion and mitochondria); hence, “heterotopias” is a redundant plural form and incorrect usage in English. Another example of semantics is the correct reference to “ventral horn cells” rather than the incorrect “anterior horn cells,” the latter having arisen from a change in terms in late 19th century Victorian England by religious fundamentalists who were reacting to Darwin’s theory of evolution by contending that man was exalted and unique amongst animals and, hence, required different terms of reference for his anatomy (82; 84).
Some terms are used differently in neuroembryology than in describing development in other organ systems; an example is neuroblast and hematological blast cells, mentioned above in relation to neuroblast migration. Another common confusion is use of the terms “noncleavage” and “fusion.” A structure that normally divides or separates into symmetrical paired structures (cleavage) but fails to do so in certain malformations is noncleavage, as with the frontal midline of the cerebral cortex in holoprosencephaly. Fusion, by contrast, is the formation of a midline structure from lateral cellular masses that meet in the midline during ontogenesis; 1 example is the cerebellar vermis, and another is the basilar artery that forms from parallel paired neural arteries at the base of the embryonic brainstem.
The diagnosis of abnormal development of the brain or of disorders of cerebral maturation relies during life on mainly clinical clues (such as dysmorphic facies, developmental delay, and epilepsy) that lead to investigations such as EEG and neuroimaging. The EEG reflects large fields of thousands of synaptic activities in the cerebral cortex and, as with other developmental processes, is precisely time-linked in its “electrocerebral maturation” in preterm and term neonates and also postnatally in infancy and childhood. It shows characteristic abnormal patterns in malformations of the brain.
MRI can now be performed in the fetus in utero as well as in the postnatal period, has been a great advance in the diagnosis of malformations of the brain, and also can provide information about myelination sequences, metabolic conditions, and vascular anomalies. The greatest limitation of neuroimaging techniques in general, including MRI, is that lesions must be large enough to detect with the naked eye. Many developmental lesions of the brain are microscopic in size and are detected only by direct tissue examination in the microscope at autopsy or in surgical resections for epilepsy.
Neuropathology has made as many advances in diagnostic techniques as neuroimaging, with the application of immunocytochemical markers of neuronal and glial cell maturation, the demonstration in tissue sections of the distribution and cellular localization of specific neurotransmitters, and the expression of cell-specific proteins. Correlations with neuroimaging help validate and explain findings in MRI and functional imaging studies. Specialized genetic techniques to demonstrate mRNA transcripts in specific cells also may be performed in neurons and glial cells, but this still resides in the domain of research rather than standard diagnostic neuropathology (74).
Nearly 3,000 illustrations, including video clips of neurologic disorders.
Every article is reviewed by our esteemed Editorial Board for accuracy and currency.
Full spectrum of neurology in 1,200 comprehensive articles.
Listen to MedLink on the go with Audio versions of each article.
Nov. 23, 2022
General Child Neurology
Nov. 21, 2022
Behavioral & Cognitive Disorders
Nov. 21, 2022
Nov. 09, 2022
Nov. 06, 2022
Stroke & Vascular Disorders
Nov. 06, 2022
Headache & Pain
Oct. 31, 2022
Headache & Pain
Oct. 31, 2022