Cell Lineages in Histogenesis of the Central Nervous System

The origins of most cells found in the mature central nervous system can be traced to multipotential stem cells within the early neuroepithelium (Fig. 11.5). These cells undergo numerous mitotic divisions before maturing into bipotential progenitor cells, which give rise to either neuronal or glial progenitor cells. Activation of the proneural genes neurogenin 1 and neurogenin 2 promotes the differentiation of neurons from the bipolar progenitor cells. Glial cells differentiate under the influence of other stimuli. This developmental bifurcation is accompanied by a change in gene expression. Multipotential stem cells express an intermediate filament protein called nestin. Nestin is downregulated as descendants of bipolar progenitor cells separate into neuronal progenitor cells, which express neurofilament protein, and glial progenitor cells, which express glial fibrillary acidic protein.

The neuronal progenitor cells give rise to a series of neuroblasts. The earliest bipolar neuroblasts possess two slender cytoplasmic processes that contact the external limiting membrane and the central luminal border of the neural tube. By retracting the inner process, a bipolar neuroblast loses contact with the inner luminal border in the process of becoming a unipolar neuroblast. The unipolar neuroblasts accumulate large masses of rough endoplasmic reticulum (Nissl substance) in their cytoplasm and begin to send out several cytoplasmic processes. At this point, they are known as multipolar neuroblasts. Their principal developmental activities are to send out axonal and dendritic processes and to make connections with other neurons or end organs.

The other major lineage stemming from the bipotential progenitor cells is the glial line. Glial progenitor cells continue to undergo mitosis, and their progeny split into several lines. One, the O-2A progenitor cell (see Fig. 11.5), is a precursor to two lines of glial cells that ultimately form the oligodendrocytes and type 2 astrocytes. Another glial lineage gives rise to type 1 astrocytes. Human oligodendrocytes arise from progenitor cells located in the ventral ventricular zone (see MN in Fig. 11.10) alongside the floor plate. From there, they spread throughout the brain and spinal cord and ultimately form the myelin coverings around neuronal processes in the white matter. The formation of oligodendrocyte precursors depends on an inductive signal from the notochord (sonic hedgehog [shh]). If the notochord is transplanted alongside dorsal neural tube, oligodendrocyte precursors differentiate there, thus showing that cells with the potential to form oligodendrocytes reside in that area, but normally do not develop because of the lack of an adequate inductive signal.

The third glial lineage has a more complex history. Radial progenitor cells give rise to radial glial cells, which act as guidewires in the brain for the migration of young neurons (see Fig. 11.23). When the neurons are migrating along the radial glial cells during midpregnancy, they inhibit the proliferation of the radial glial cells. After neuronal cell migration, the radial glial cells, now free from the inhibitory influence of the neurons, reenter the mitotic cycle. Their progeny can transform into several cell types. Some can seemingly cross lineage lines and differentiate into type 1 astrocytes (see Fig. 11.5). Other progeny differentiate into various specialized glial cell types, ependymal cells, and even adult neural stem cells. According to some authors, the remaining neuroepithelial cells represent another source of ependymal cells.

Not all cells of the central nervous system originate in the neuroepithelium. Microglial cells, which serve a phagocytic function after damage to the brain, are immigrant cells derived from primitive myeloid precursors (macrophages). Microglia are not found in the developing brain until the brain is penetrated by blood vessels.

Fundamental Cross-Sectional Organization of the Developing Neural Tube

The developing spinal cord is a useful prototype for studying the overall structural and functional features of the central nervous system because it preserves its fundamental organization through much of development. With the beginning of cellular differentiation in the neural tube, the neuroepithelium thickens and appears layered. The layer of cells closest to the lumen (central canal) of the neural tube remains epithelial and is called the ventricular zone (the ependymal zone in older literature). This zone, which still contains mitotic cells, ultimately becomes the ependyma, a columnar epithelium that lines the ventricular system and central canal of the central nervous system (Fig. 11.6). Farther from the ventricular zone is the intermediate (formerly called mantle) zone, which contains the cell bodies of the differentiating postmitotic neuroblasts. As the neuroblasts continue to produce axonal and dendritic processes, the processes form a peripheral marginal zone that contains neuronal processes, but not neuronal cell bodies.

As the spinal cord matures, the intermediate zone becomes the gray matter, in which the cell bodies of the neurons are located. The marginal zone is called the white matter because of the color imparted by the numerous tracts of myelinated nerve fibers in that layer (see Fig. 11.6). During development, the proliferating progenitor cell populations in the ventricular zone become largely exhausted, but it is now known that a subpopulation persists into adulthood as neural stem cells. The remaining cells differentiate into the epithelium of the ependymal layer.

When the basic layers in the spinal cord are established, several important topographical features can be recognized in cross sections of the spinal cord. A sulcus limitans within the central canal divides the spinal cord into a dorsal alar plate and a ventral basal plate on each side of the central canal. The right and left alar plates are connected dorsally over the central canal by a thin roof plate, and the two basal plates are connected ventrally by a floor plate.

The basal plate represents the motor component of the spinal cord. Axons arising from neurons located in the ventral horn of the gray matter exit the spinal cord as ventral motor roots of the spinal nerves (see Fig. 11.15). The gray matter of the alar plate, called the dorsal horn, is associated with sensory functions. Sensory axons from the spinal ganglia (neural crest derivatives) enter the spinal cord as dorsal roots and synapse with neurons in the dorsal horn. A small projection of gray matter between the dorsal and ventral horns at spinal levels T1 to L2 contains cell bodies of autonomic neurons. This projection is called the lateral horn or sometimes the intermediolateral gray column (see Fig. 11.6).

The floor plate is far more than an anatomical connection between the right and left basal plates. Cells of the future floor plate are the first to differentiate in the neural plate after primary induction of the nervous system. Experimental work has shown a specific inductive influence of the notochord on the neuroepithelial cells that overly it. If an extra notochord is grafted along the lateral surface of the neural tube, the neuroepithelial cells closest to it acquire the properties of floor plate cells (Fig. 11.7). Conversely, if a segment of normal notochord is removed, the neuroepithelial cells overlying it do not acquire the properties of floor plate cells. Through its action on the floor plate, the notochord also exerts a profound effect on the organization of the dorsal and ventral roots that enter and leave the spinal cord. If the notochord is absent, the neural tube closes, but recognizable dorsal and ventral roots are absent. Numerous ectopic nerve fibers appear in their place (Fig. 11.8). If the future floor plate is split, the side of the neural tube on which the notochord is located develops normal dorsal and ventral roots, whereas the side lacking these structures gives off ectopic nerves (see Fig. 11.7C).

A molecular basis for cross-sectional pattern formation within the early neural plate and neural tube has been identified (Fig. 11.9). The homeobox-containing transcription factors, Pax-3, Pax-7, Msx-1, and Msx-2, are expressed throughout the early neural plate. Before the neural plate has folded over to become the neural tube, the notochord, which is adherent to the midline neural plate at this stage, releases shh. Local hedgehog signaling stimulates the neural plate cells directly above the notochord to transform into the floor plate. One of the first stages of this transformation is the repression of Pax-3 and Pax-7 expression, which allows the neuroectodermal cells near the midline of the neural plate to adopt a ventral fate (i.e., floor plate or basal plate). Cells of the floor plate itself then become sites of production of shh.

Development of the overall cross-sectional organization of the neural tube involves not only a ventralizing influence from the notochord, but also an opposing dorsalizing influence from the epidermal ectoderm adjacent to the developing neural tube. In the lateral regions of the neural plate (future dorsal region of the neural tube), BMP-4 and BMP-7, expressed by non-neural ectoderm at the ectodermal–lateral neural plate junction, exert a dorsalizing inductive effect on the neuroectodermal cells that results in the formation of the roof plate, which takes shape soon after the last neural crest cells have emigrated from the neural tube. BMP within the roof plate acts as a patterning signal and induces the further dorsalizing molecules Pax-3, Pax-7, Msx-1, and Msx-2 (see Fig. 11.9). Dorsal Wnt signaling promotes the proliferation of neural progenitor cells and also works with BMPs as an overall dorsalizing influence in the dorsoventral patterning of neurons. After closure of the neural tube, signals from the roof plate induce a series of six early and two late generated dorsal interneuronal types in a manner reminiscent of the better defined specification of ventral interneurons (see later).

While the broader regions of the cross section of the spinal cord are being set in place, a tightly controlled molecular grid forms the basis for specification of the major types of neurons found in the ventral part of the spinal cord. Within the basal plate is an array of five types of neurons—motoneurons and four types of interneurons—that are arranged in a well-defined dorsoventral pattern. These classes of neurons are specified by specific combinations of homeodomain transcription factors, whose pattern of expression is set by a gradient of shh emanating from the floor plate as modulated by the activating and repressive properties of the Gl-1 to Gl-3 proteins (Fig. 11.10). Some of these transcription factors (class I) are repressed at various dorsoventral levels by the shh gradient, whereas others (class II) are induced by shh (see Fig. 11.10). The net result is that a different combination of the transcription factors at each dorsoventral level specifies each of the five types of neurons, which are characterized by their own unique molecular signature. One in particular, islet-1, is characteristic of motoneurons. Shortly after the production of motoneurons has ceased, a shift in regulatory factors stimulates the production of glial progenitor cells from the ventral neuroepithelium. This leads to the formation of the oligodendrocytes that become closely associated with the neurons.

In addition to inducing motoneurons, the floor plate plays other roles in the developing nervous system. Many groups of neuronal processes cross from one side of the central nervous system to the other through the floor plate as commissural axons. These axons, leading from neuronal cell bodies located in the dorsal half of the neural tube, are attracted to the floor plate by specific molecules produced in that region (e.g., netrin 1). In mutant animals lacking netrin 1, commissural axons are disorganized and do not cross to the other side through the floor plate. Not only does the floor plate attract certain types of axons, but it also repels others. A specific example is the trochlear nerve (cranial nerve IV), whose axons do not cross to the other side from their cell bodies of origin.

Crossing the midline and proper positioning of axons after crossing are largely controlled by the Slit-Robo system. Slit proteins are axonal repellents, which are located in the ventral midline of the floor plate. Robo proteins (1, 2, and 3) are receptors for the Slit ligands. Robo-1 and Robo-2 combine with Slit to repel axonal extension at the midline. For commissural axons that cross the midline, Robo-3 is expressed as they approach the midline. Robo-3 interferes with the repulsive activity of the Robo-1, Robo-2/slit combination and thereby allows the commissural axons to cross the midline. Once these axons have crossed the floor plate, the expression of Robo-3 is downregulated, and the remaining Robo-1, Robo-2/Slit combinations repel the axons away from the midline that they just crossed.

Patterning in the Hindbrain and Spinal Cord Regions

The rhombomeres (Fig. 11.11), introduced in Chapter 6, are the morphological reflection of a highly segmentally ordered pattern of expression of a variety of developmentally prominent transcription factors (Fig. 11.12). The establishment of the isthmic organizer and the pathways that set up this pattern are discussed in Chapter 6 (see pp. 95-96).

The correspondence between the rhombomeres of the developing hindbrain and other structures of the cranial and pharyngeal arch region is remarkable (see Chapter 14). The cranial nerves, which have a highly ordered pattern by which they supply structures derived from the pharyngeal arches and other structures in the head, have an equally highly ordered origin with respect to the rhombomeres (Fig. 11.13). Cranial nerve V innervates structures derived from the first pharyngeal arch. Cranial nerves VII and IX innervate the second and third arch structures. In embryos of birds, the species studied most extensively, cell bodies of the motor components of cranial nerves V, VII, and IX are initially found exclusively in rhombomeres 2, 4, and 6.

Dye injection studies have shown that the progeny of a single neuroblast remain within the rhombomere containing the injected cell, a finding suggesting that rhombomeres have the properties of cellular compartments. Axons contributing to a cranial nerve extend laterally within the rhombomere and converge on a common exit site in the craniocaudal midpoint of the rhombomere. At a slightly later stage in development, motoneurons originating in the next more posterior rhombomere (3, 5, 7) extend axons laterally. Before the axons reach the margin of the rhombomere, however, they cross into rhombomeres 2, 4, or 6 and converge on the motor axon exit site in the even-numbered rhombomere.

The cell bodies (within the central nervous system, the collection of cell bodies of a single cranial nerve is called a nucleus) of the cranial nerves that innervate the pharyngeal arches arise in register along the craniocaudal axis. The motor nuclei of other cranial nerves that innervate somatic structures (e.g., extraocular muscles or the tongue) arise in a different craniocaudal column along the hindbrain and do not occupy contiguous rhombomeres (see Fig. 11.13).

Direct and indirect evidence indicates that properties of the walls of the rhombomeres prevent axons from straying into inappropriate neighboring rhombomeres. One cellular property, which is also characteristic of regions of somites that restrict the movement of neural crest cells, is the ability of cells of the wall of the rhombomere to bind specific lectins. In apparent contradiction to the compartmentalization just described, processes growing from sensory neuroblasts and from nerves of a tract called the medial longitudinal fasciculus are free to cross rhombomeric boundaries. Blood vessels first enter the hindbrain in the region of the floor plate soon after the emergence of the motor axons and spread within the inter-rhombomeric junctions. The way the vascular branches recognize the boundaries of the rhombomeres is unknown.

In contrast to the hindbrain, the pattern of nerves emanating from the spinal cord does not appear to be determined by craniocaudal compartmentalization within the spinal cord. Rather, the segmented character of the spinal nerves is dictated by the somitic mesoderm along the neural tube. Outgrowing motoneurons from the spinal cord and migrating neural crest cells can easily penetrate the anterior mesoderm of the somite, but they are repulsed by the posterior half of the somite. This situation results in a regular pattern of spinal nerve outgrowth, with one bilateral pair of spinal nerves per body segment. Rotating the early neural tube around its craniocaudal axis does not result in an abnormal pattern of spinal nerves. This further strengthens the viewpoint that the pattern of spinal nerves is not generated within the neural tube itself.

Patterning in the Midbrain Region

A fundamental patterning mechanism in the midbrain region is a molecular signaling center (isthmic organizer) located at the border between the mesencephalon and the metencephalon (see Fig. 6.4). The principal signaling molecule is FGF-8, which is expressed in a narrow ring at the anterior border of the first rhombomere, a subdivision of the metencephalon. Acting with Wnt-1, FGF-8 induces the expression of engrailed genes En-1 and En-2 and Pax-2 and Pax-5, which are expressed in decreasing concentrations at increasing distances from the FGF-8 signaling center (see Fig. 6.4). The main function of Wnt-1 seems to be the stimulation of local cell proliferation, whereas the overall organizing function belongs to FGF-8. The isthmic organizer induces and polarizes the dorsal midbrain region and the cerebellum. Grafts of the isthmic organizer or beads releasing FGF-8 alone into more cranial regions of the forebrain of the avian embryo induce a second tectum (dorsal mesencephalon or colliculi in mammals). Similarly, isthmic grafts into regions of the hindbrain can induce supernumerary cerebellar structures.

One byproduct of these molecular studies is the realization that in some species the boundary between the future midbrain and hindbrain does not correspond to the anatomical constriction between the mesencephalon and rhombencephalon. Instead, it is located cranial to that constriction in a plane marked by the posterior limit of expression of the homeobox-containing gene, Otx-2.

Similar to the spinal cord, the midbrain is also highly patterned along its cross-sectional (dorsoventral) axis, and ventrally secreted shh is the molecular basis for much of this patterning, which is represented by five arc-shaped territories of gene expression. If a point source of shh is introduced into a more dorsal region of the midbrain, an additional ectopic set of five corresponding territorial arcs is produced. In addition to promoting neuronal development in the basal plate of the midbrain, shh restricts ventral expression of molecules, such as Pax-7, which are characteristic of the alar plate.

Cranially, the midbrain is separated from the forebrain (diencephalon) through a different set of molecular interactions. The alar plate of the diencephalon is characterized by the expression of Pax-6, a molecule that, among other things, acts as the master gene underlying eye formation (see Chapter 13). The mesencephalon is a domain of En-1 expression. Through the action of intermediate negative regulators, Pax-6 inhibits En-1 expression, whereas En-1 directly inhibits Pax-6 expression. The craniocaudal level at which each of these molecules inhibits the other becomes a sharp diencephalic-mesencephalic border.

Patterning in the Forebrain Region

Although much less apparent than in the hindbrain, the neuromeric organization of the early brain extends into the forebrain region as a set of three prosomeres, which extend from the midbrain-forebrain boundary through the thalamus (Fig. 11.14). Prosomeres 1 to 3 (p1 to p3), the most posterior prosomeres, become incorporated into the diencephalon, with p2 and p3 forming the dorsal and ventral thalamus, which serves as a major relay station for transmitting neural signals between the cerebral cortex and the body. An earlier representation suggested that an additional set of prosomeres (p4 to p6) formed the organizational basis for the rostral diencephalon (hypothalamus) and telencephalon. A more recent interpretation, however, places these structures in what has been called the secondary rhombencephalon, a developmental field that encompasses the entire prechordal portion of the neural tube. Within this domain, the basal plate develops into the major regions of the hypothalamus, the structure that integrates autonomic nervous functions and controls endocrine release from the pituitary. The alar plate in this domain is the precursor of the cerebral cortex, the basal ganglia (collectively the telencephalon), and the optic vesicles (diencephalic structures that take the lead in formation of the eyes). As development progresses, the secondary rhombencephalon becomes sharply folded beneath p2 to p3, and in humans the enormous outgrowth of the alar plates of the secondary rhombencephalon envelops the other prosomeres as the telencephalic vesicles (future cerebral cortex).

Discrete patterns of gene expression also mirror the basic regional organization of the forebrain. Early in development, FGF-8, secreted by the anterior neural ridge (see Fig. 6.4B), induces the expression of Foxg-1, formerly called BF-1 (brain factor-1), which regulates development of the telencephalon and optic vesicles. Within the forebrain, a thin rim of expression of the transcription factor Nkx 2.2 marks the border between alar and basal plates. Along the dorsoventral axis, the alar plates are characterized by the expression of Emx1 and Emx2 and Pax6, important regulators of regional identity within the cerebral cortex. In an intermediate region of the prosencephalic alar plate, Emx expression drops out, leaving Pax-6 to function as the master gene controlling eye development (see Chapter 13). In the more ventral area that eventually forms the basal ganglia, Dlx (distalless) is a dominant expressed gene.

Similar to the spinal cord, the ventral forebrain is induced and organized by shh, secreted by midline axial structures. In the absence of shh signaling in this area, the tissues of the ventral forebrain are greatly reduced, leading sometimes to midline fusion of the optic vesicles and a general reduction of the growth of the midface region. This situation results in a type of anomaly called holoprosencephaly (see p. 309), which in extreme cases is accompanied by cyclopia.

At the border between the future dorsal (p2) and ventral (p3) thalamus is a narrow band of shh expression called the zona limitans interthalamica. Signals emanating from the zona limitans interthalamica specify aspects of cellular identity and behavior in the diencephalic regions on either side (see Fig. 6.4B). The mechanisms leading to the formation of the zona limitans interthalamica are poorly understood, but this structure arises directly above the anterior tip of the notochord. This finding suggests that the formation of this structure, similar to that of the isthmus, is a reflection of an earlier molecular boundary zone.

Patterns and Mechanisms of Neurite Outgrowth

The outgrowth of neurites (axons or dendrites) involves many factors intrinsic and extrinsic to the neurite. Although similar in many respects, the outgrowth of axons and dendrites differs in fundamental ways.

An actively elongating neurite is capped by a growth cone (Fig. 11.17). Growth cones are characterized by an expanded region of cytoplasm with numerous spikelike projections called filopodia. In vitro and in vivo studies of living nerves show that the morphology of an active growth cone is in a constant state of flux, with filopodia regularly extending and retracting as if testing the local environment. Growth cones contain numerous cytoplasmic organelles, but much of the form and function of the filopodia depends on the large quantities of actin microfilaments that fill these processes. In a growth cone, an equilibrium exists between the extension of actin microfilaments by terminal addition and resorption at the proximal end. Under conditions favorable for growth, the balance tips toward extension, whereas an unfavorable environment results in the resorption of the microfilaments and collapse of the growth cone.

Whether growth cones progress forward, remain static, or change directions depends in large measure on their interactions with the local environment. If the environment is favorable, a filopodium remains extended and adheres to the substrate around it, whereas other filopodia on the same growth cone retract. Depending on the location of the adhering filopodia, the growth cone may lead the neurite to which it is attached straight ahead or may change its direction of outgrowth. This outgrowth seems to be guided by four broad types of environmental influences: chemoattraction, contact attraction, chemorepulsion, and contact repulsion. It now seems that outgrowing nerve processes find themselves in different environments every couple hundred micrometers, and that some environments give them signals to continue extending forward, whereas other environments may act as “stop” signals or “turn” signals. The sensitivity of growth cones to their environment is so great that they may be able to discriminate a concentration difference of as little as one molecule across the surface.

Growth cones can respond to concentration gradients of diffusible substances (e.g., nerve growth factor) or to weak local electrical fields. A major family of chemoattractant molecules is called the netrins. The repulsive counterparts to the netrins are members of a family of secreted proteins called semaphorins. The various growth cone attractants produce their effects by stimulating signal transduction machinery that results in elevated levels of cyclic nucleotide second messengers (cyclic adenosine monophosphate or cyclic guanosine monophosphate), whereas repulsants exert their effects through decreased levels of these second messengers.

Many molecules are involved in the guidance of neuronal outgrowth (Table 11.1). Some of these molecules can exert either attractive or repulsive actions, depending on the character of the neuron, the time in development, or some combination of intrinsic and extrinsic characteristics. To a greater or lesser extent, the nature of the reaction of the growth cone is determined locally, because in vitro studies have shown that such reactions can occur even when the neuronal process is cut off from the cell body.