53: Differentiation and Survival of Nerve Cells

• The Proliferation of Neural Progenitor Cells Involves Symmetric and Asymmetric Modes of Cell Division

• Radial Glial Cells Serve As Neural Progenitors and Structural Scaffolds

• The Generation of Neurons or Glial Cells Is Regulated by Delta-Notch Signaling and Basic Helix-Loop-Helix Transcription Factors

• Neuronal Migration Establishes the Layered Organization of the Cerebral Cortex

• Central Neurons Migrate Along Glial Cells and Axons to Reach Their Final Settling Position

  • Glial Cells Serve As a Scaffold in Radial Migration

  • Axon Tracts Serve As a Scaffold for Tangential Migration

  • Neural Crest Cell Migration in the Peripheral Nervous System Does Not Rely on Scaffolding

• The Neurotransmitter Phenotype of a Neuron Is Plastic

  • The Transmitter Phenotype of a Peripheral Neuron Is Influenced by Signals from the Neuronal Target

  • The Transmitter Phenotype of a Central Neuron Is Controlled by Transcription Factors

• The Survival of a Neuron Is Regulated by Neurotrophic Signals from the Neuron's Target

  • The Neurotrophic Factor Hypothesis Was Confirmed by the Discovery of Nerve Growth Factor

  • Neurotrophins Are the Best Studied Neurotrophic Factors

  • Neurotrophic Factors Suppress a Latent Death Program in Cells

• An Overall View

In the preceding chapter we described how local inductive signals pattern the neural tube and establish the early regional subdivisions of the nervous system—the spinal cord, hindbrain, midbrain, and forebrain. Here we turn to the issue of how progenitor cells within these regions differentiate into neurons and glial cells, the two major cell types that populate the nervous system.

We discuss some of the molecules that specify neuronal and glial cell fates and how they are regulated. The basic mechanisms of neurogenesis endow cells with common neuronal properties, features that are largely independent of the region of the nervous system in which they are generated or the specific functions they perform. We also discuss the mechanisms by which developing neurons express neurotransmitters and synaptic receptors.

After the identity and functional properties of the neuron have begun to emerge, additional developmental processes determine whether the neuron will live or die. Remarkably, approximately half of the neurons generated in the mammalian nervous system are lost through programmed cell death. We examine the factors that regulate the survival of neurons and the possible benefits of widespread neuronal loss. Finally, we describe the existence of a core biochemical pathway that programs the death of nerve cells.

The mature brain comprises billions of nerve cells and even more glial cells. Yet its precursor, the neural plate, initially comprises only a few hundred cells. From this simple comparison we infer that regulation of the proliferation of neural cells is a major driving force in shaping brain development. Histologists in the late 19th century showed that neural epithelial cells close to the ventricular lumen of the embryonic brain exhibit features of mitosis, and we now know that the proliferative zones surrounding the ventricles are the major regions involved in the production of neural cells in the cerebral cortex as well as other regions of the central nervous system.

At early stages of embryonic development most progenitor cells in the ventricular zone of the neural tube proliferate rapidly. Many of these early neural progenitors have the properties of stem cells: They can generate additional copies of themselves, a process called self-renewal, and also give rise to differentiated neurons and glial cells (Figure 53–1B). As with other types of stem cells, neural progenitor cells undergo stereotyped programs of cell division.

One mode of cell division is asymmetric: The progenitor produces one differentiated daughter and another daughter that retains its stem cell-like properties. This mode does not permit amplification of the stem cell population. In a second mode neural stem cells divide symmetrically to produce two stem cells, and in this way expand the population of proliferative progenitor cells. Both symmetric and asymmetric modes have been found in the embryonic cerebral cortex in vivo and in cortical cells grown in tissue culture (Figure 53–1C).

The incidence of symmetric and asymmetric cell division is influenced by signals in the local environment of the dividing cell, making it possible to control the probability of self-renewal or differentiation. Environmental factors can influence the outcome of progenitor cell divisions in two fundamental ways. They can act in an "instructive" manner, biasing the outcome of the division process and causing the stem cell to adopt one fate at the expense of others. Or they can act in a "selective" manner, permitting the survival and maturation of only certain cell progeny.

Radial glial cells are the earliest morphologically distinguishable cell type to appear within the primitive neural epithelium. Their cell bodies are located in the ventricular zone and their long process extends to the pial surface. As the brain thickens, the processes of radial glial cells remain attached to the ventricular and pial surfaces. After the generation of neurons is complete, many radial glial cells differentiate into astrocytes, a prominent class of glial cell in the mature brain. The elongated shape of the radial glial cell places it in a favorable position to serve as a scaffold for the migration of neurons that emerge from the ventricular zone (Figure 53–2).

The ventricular zone was once thought to contain two major cell types: radial glial cells and a set of neuroepithelial progenitors, which served as the primary source of neurons. This classical view has changed dramatically in the last few years. Radial glial cells are in fact progenitor cells that generate both neurons and astrocytes in addition to their role in neuronal migration (Figure 53–2). When radial glial cells are selectively labeled with fluorescent dyes or viruses, tracing of their clonal progeny reveals cell clusters that contain both neuronal and radial glial cells. These findings indicate that radial glial cells are able to undergo both asymmetric and self-renewing cell division, and serve as a major source of post-mitotic neurons as well as astrocytes. Radial glial cells may also serve as progenitors of neurons in the adult central nervous system.

How do radial glial cells make the decision to self-renew, generate neurons, or give rise to mature astrocytes? The answer to this question involves an evolutionarily conserved signaling system.

In flies and vertebrates neural fate is regulated by a cell-surface signaling system, comprised of the transmembrane ligand delta and its receptor notch, which regulates a cascade of transcription factors of the basic helix-loop-helix (bHLH) family. This signaling system was revealed in genetic studies in Drosophila. Neurons emerge from within a larger cluster of ectodermal cells, each of which has the potential to generate neurons. This proneural region is defined by expression of proneural genes, which encode transcription factors of the bHLH class. Yet within the proneural region only certain cells form neurons; the others become epidermal support cells.

Initially the ligand delta and its receptor notch are expressed at similar levels by all proneural cells (Figure 53–3A). With time, however, notch activity is enhanced in one cell and suppressed in its neighbor. The cell in which notch activity is highest loses the potential to form a neuron and acquires an alternative fate. The binding of Delta to Notch results in proteolytic cleavage of the Notch cytoplasmic domain, which then enters the nucleus. There it functions as a transcription factor, regulating the activity of bHLH transcription factors that suppress the ability of the cell to become a neuron and reduce the level of expression of the ligand Delta (Figure 53–3B). Through this feedback pathway a minor difference in the initial level of Notch signaling is rapidly amplified to generate an all-or-none difference in the status of Notch activation, and consequently the fates of the two cells. This basic logic of Delta-Notch and bHLH signaling has been conserved in vertebrate and invertebrate neural tissues.

How does Notch signaling regulate neuronal and glial production in mammals? At early stages in the development of the mammalian cortex Notch signaling promotes the generation of radial glial cells by activating members of the Hes family of bHLH transcriptional repressors. Two of these proteins, Hes1 and Hes5, appear to maintain radial glial cell character by activating the expression of an ErbB class tyrosine kinase receptor for neuregulin, a secreted signal that promotes radial glial cell identity. The Notch ligand Delta1 as well as neuregulin are expressed by newly generated cortical neurons; thus the radial glial cells depend on feedback signals from their neuronal progeny for continued production.

At later stages of cortical development Notch signaling continues to activate Hes proteins, but a change in the intracellular response pathway results in astrocyte differentiation. At this stage the Hes proteins work by activating a transcription factor, STAT3, which recruits the serine-threonine kinase JAK2, a potent inducer of astrocyte differentiation. STAT3 also activates expression of astrocyte-specific genes such as the glial-fibrillary acidic protein (GFAP).

The generation of oligodendrocytes, the second major class of glial cells in the central nervous system, follows many of the principles that control neuron and astrocyte production. Notch signaling regulates the expression of two bHLH transcription factors, Olig1 and Olig2, which have essential roles in the production of embryonic and postnatal oligodendrocytes.

The generation of cortical neurons requires that cells avoid exposure to notch signals (Figure 53–4). This task is achieved in part through the expression of numb, a cytoplasmic protein that antagonizes notch signaling. The key role of numb in neurogenesis was first shown in Drosophila, where it determines the neuronal fate of daughter cells of asymmetrically dividing progenitors. In the mammalian cortex numb, as with Notch, is preferentially localized in neuronal daughters and antagonizes notch signaling. As a consequence, loss of numb activity causes progenitor cells to proliferate extensively. The inhibition of notch signaling results in the expression of several proneural bHLH transcription factors, notably Mash1, neurogenin-1, and neurogenin-2. Neurogenins promote neuronal production by activating downstream bHLH proteins such as neuroD, and they block the formation of astrocytes by inhibiting JAK and STAT signaling.

Although delta-notch signaling and bHLH transcription factor activators lie at the heart of the decision to produce neurons or glial cells, several additional transcriptional pathways augment this core molecular program. One important transcription factor, REST/NRSF, is expressed in neural progenitors and glial cells, where it represses the expression of neuronal genes. REST/NRSF rapidly degrades as neurons differentiate, permitting the expression of neurogenic bHLH factors and other neuronal genes. Homeodomain transcription factors of the SoxB class also play an important role in maintaining neural progenitors by blocking neurogenic bHLH protein activity. The differentiation of neurons therefore requires the avoidance of REST/NRSF and SoxB protein activity.

The mammalian cerebral cortex develops in three main stages: a preplate, a cortical plate, and finally the mature pattern of layers. Neural precursors in the ventricular zone of the preplate differentiate into neurons that migrate along radial glial fibers before settling in the cortical plate. These migratory cells divide the preplate into a subplate and a marginal zone (Figure 53–5).

Once within the cortical plate, neurons become organized into well-defined layers. The laminar settling position of a neuron is correlated precisely with its birthday, a term that refers to the time at which a dividing precursor cell undergoes its final round of cell division and gives rise to a postmitotic neuron. Cells that migrate from the ventricular zone and leave the cell cycle at early stages give rise to neurons that settle in the deepest layers of the cortex. Cells that exit the cell cycle at progressively later stages migrate over longer distances and pass earlier-born neurons, before settling in more superficial layers of the cortex. Thus the layering of neurons in the cerebral cortex follows an inside-first, outside-last rule (Figure 53–6A).

Disruption in the migratory and settling programs of cortical neurons underlies much human cortical pathology (Figure 53–6C, D). In lissencephaly (Greek, smooth brain, referring to the characteristic smoothing of the cortical surface in patients with the disorder) neurons leave the ventricular zone but fail to complete their migration into the cortical plate. As a result, the mature cortex is typically reduced from six to four neuronal layers, and the arrangement of neurons in each remaining layer is disordered. Occasionally, lissencephaly is accompanied by the presence of an additional group of neurons in the subcortical white matter. Patients with lissencephalies from mutations in the Lis1 and doublecortin genes often suffer severe mental retardation and intractable epilepsy. The Lis1 and doublecortin proteins have been localized to microtubules, suggesting they are involved in microtubule-dependent nuclear movement, although their precise functions in neuronal migration remain unclear.

Mutations that disrupt the reelin signaling pathway disrupt the final stage of neuronal migration through the cortical subplate. Reelin is an extracellular matrix protein that is secreted from the Cajal-Retzius cells, a class of neurons found in the preplate and marginal zone. Signals from these cells are crucial for the migration of cortical neurons. In mice lacking functional reelin, neurons fail to detach from their radial glial scaffolds and pile up underneath the cortical plate, disobeying the inside-out migratory rule. As a consequence, the normal layering of cell types is inverted and the marginal zone is lost. Reelin acts through cell-surface receptors that include the ApoE receptor 2 and the very low-density lipoprotein receptor. The binding of reelin to these receptors activates an intracellular protein, Dab1, which transduces reelin signals. Cadherin-like adhesion proteins may also be involved in transducing reelin signals. Not surprisingly, the loss of proteins that transduce reelin signals produce similar migratory phenotypes.

The migration of neurons follows one of three major programs, termed radial, tangential, and free migration. With radial migration central neurons move along the long unbranched processes of radial glial cells. With tangential migration central neurons use axonal tracts as their guides. Free migration occurs in the peripheral nervous system without radial glia or axonal tracts.

Glial Cells Serve As a Scaffold in Radial Migration

Classical anatomical studies of cortical development in the primate brain in the 1970s provided evidence that neurons generated in the ventricular zone migrate to their settling position along a pathway of radial glial fibers. Radial glial cells serve as the primary scaffold for radial neuronal migration. Their cell bodies are located close to the ventricular surface and give rise to elongated fibers that span the width of the developing cerebral wall. Each radial glial cell has one basal end-foot in the ventricular zone at the apical surface and processes that terminate in multiple end-feet at the pial surface of the brain (Figure 53–5). Radial glial scaffolds are especially important in the development of the primate cortex, where neurons are required to migrate over long distances as the cerebral wall expands.

Most radial glial cells are transient structures; as we have seen, they give rise to neurons and eventually differentiate into astrocytes. A single radial glial cell scaffold can support the migration of up to 30 generations of cortical neurons.

What forces and molecules power neuronal migration on radial glial cells? After a neuron leaves the cell cycle its leading process wraps around the shaft of the radial glial cell and its nucleus translocates within the cytoplasm of the leading process. Although the leading process of the migrating neuron extends slowly and steadily, the nucleus moves in an intermittent, stepwise manner because of complex rearrangements of the cytoskeleton. A microtubular lattice forms a cage around the nucleus, and movement of the nucleus depends on a centrosome-like structure, termed a basal body, which projects a system of microtubules into the leading process, providing the conduit for nuclear movement (Figure 53–7A). Defects in neuronal migration that result from loss of Lis1 or doublecortin (Figure 53–6) reflect the critical role of these genes in microtubular assembly and function.

Neuronal migration along radial glia also involves adhesive interactions between cells. Adhesive receptors such as integrins promote neuronal extension on radial glial cells. The migration of neurons along glial fibers is nevertheless different from the extension of axons driven by growth cones (see Chapter 54). In neuronal migration the leading process is devoid of the structured actin filaments that typify growth cones and more closely resembles an extending dendrite, an inference made first by Santiago Ramon y Cajal.

Axon Tracts Serve As a Scaffold for Tangential Migration

The second prominent mechanism of neuronal translocation in the developing brain is tangential migration. This form of migration is used to populate distinct regions of the nervous system and may have evolved as a mechanism for increasing the complexity of neuronal circuits. Its major cellular substrate appears to be preexisting axonal tracts that connect regions of neuronal generation with the final settling position of the neurons. In the developing cortex the axons of cortical projection neurons reach the internal capsule just as migratory neurons begin to enter the neocortex; at this intersection immigrating neurons are tightly associated with the bundles of axons that leave the cortex.

Neurons that use tangential migration follow precise routes of navigation and settling. This can best be seen in the ventral telencephalon, which contains two major sites of neuronal production, the medial and lateral ganglionic eminences. Some of the neurons generated in this region are destined to populate the basal ganglia. Tangential migration of neurons from these ventral structures also provides the cerebral cortex, hippocampus, and olfactory bulb with interneurons.

Neurons generated in the medial ganglionic eminence migrate tangentially and settle in the neocortex where they give rise to many interneuron populations, including Cajal-Retzius neurons. Thus cortical neurons originate from two sources: Excitatory neurons from the cortical ventricular zone and interneurons from the medial ganglionic eminence. In contrast, neurons generated in the lateral ganglionic eminence migrate rostrally and contribute the periglomerular and granule interneurons of the olfactory bulb (Figure 53–8). In this rostral migratory stream neurons use neighboring neurons as substrates for migration (chain migration). In the adult brain neurons that follow the rostral migratory stream originate instead in the subventricular zone of the striatum.

Transcription factors control the character of ganglion eminence neurons and ensure their tangential migration. The homeodomain proteins Dlx1 and Dlx2 are expressed by cells in the ganglionic eminences. In mice lacking Dlx1 and Dlx2 activity the perturbation of neuronal migration leads to a profound reduction in the number of GABAergic interneurons in the cortex. Similarly, in mice lacking the gene encoding the homeodomain transcription factor Gsh2 the rostral migratory stream is disrupted and the arrival of interneurons in the olfactory bulb is delayed.

Neural Crest Cell Migration in the Peripheral Nervous System Does Not Rely on Scaffolding

The peripheral nervous system derives from neural crest stem cells, a small group of neuroepithelial cells at the boundary of the neural tube and epidermal ectoderm. Soon after their induction neural crest cells are transformed from epithelial to mesenchymal cells and begin to delaminate from the neural tube. They then migrate to many sites throughout the body. Neural crest cell migration does not rely on scaffolding (ie, radial glial cells or preexisting axon tracts) and thus is called free migration. This form of neuronal migration requires significant cytoarchitectural and cell adhesive changes and differs from most of the migratory events in the central nervous system.

Bone morphogenetic protein (BMP) signaling is critical for neural crest induction (see Chapter 52), as well as neural crest migration. Exposure of neural epithelial cells to BMPs triggers molecular changes that convert epithelial cells to a mesenchymal state, causing them to delaminate from the neural tube and migrate into the periphery. BMPs trigger changes in neural crest cells by inducing expression of transcription factors, notably the zinc finger proteins snail, slug, and twist, which have a conserved role in promoting epithelial-to-mesenchymal transitions. These transcription factors direct expression of proteins that regulate the properties of the cytoskeleton as well as enzymes that degrade extracellular matrix proteins. These enzymes give neural crest cells the ability to break down the basement membrane surrounding the epithelium of the neural tube, permitting them to embark on their migratory journey into the periphery.

As neural crest cells begin to delaminate, their expression of cell adhesion molecules changes. Alterations in expression of adhesive proteins, notably cadherins, permit neural crest cells to loosen their adhesive contacts with neural tube cells and begin the delamination process. Neural crest cells also begin to express integrins, receptors for extracellular matrix proteins such as laminins and collagens that are found along peripheral migratory paths.

The first structures encountered by migrating neural crest cells are somites, epithelial cells that later give rise to muscle and cartilage. Neural crest cells pass through the anterior half of each somite but avoid the posterior half (Figure 53–9). The rostral channeling of migratory neural crest cells is imposed by ephrin B proteins, which are concentrated in the posterior half of each somite. Ephrins provide a repellant signal that interacts with EphB class tyrosine kinase receptors on neural crest cells to prevent their invasion. Neural crest cells that remain within the anterior sclerotome of the somite differentiate into sensory neurons of the dorsal root ganglia; those that migrate around the dorsal region of the somite approach the skin and give rise to melanocytes.

Differentiation of neural crest cells into dorsal root ganglion neurons is initiated at the time the cells emigrate from the neural tube, and depends on early exposure to Wnt signals secreted from the dorsal neural tube and somites as well as expression of neurogenin bHLH factors. In contrast, those neural crest cells that follow a more medial and ventral migratory path are exposed to BMPs secreted from the dorsal aorta and develop as sympathetic neurons that acquire a nonadrenergic phenotype. BMPs promote noradrenergic neuronal differentiation by inducing the expression of a variety of transcription factors that include the bHLH protein Mash1, the homeodomain protein Phox2, and the zinc finger protein Gata2 (Figure 53–10).

Neurons continue to develop after they have migrated to their final position, and no aspect of their later differentiation is more important than the choice of chemical neurotransmitter. The migratory pathway that a neural crest cell pursues to reach its final location exposes the cell to environmental signals that have a critical role in determining its transmitter phenotype.

The Transmitter Phenotype of a Peripheral Neuron Is Influenced by Signals from the Neuronal Target

The transmitter phenotype of autonomic neurons is plastic, and the final transmitter phenotype is determined in part by the end organs they innervate. Most sympathetic neurons use norepinephrine as their primary transmitter. However, those that innervate the exocrine sweat glands in the foot pads use acetylcholine, and even these neurons express norepinephrine at the time they first innervate the sweat glands of the skin. Only after their axons have contacted the sweat glands do they stop synthesizing norepinephrine and start producing acetylcholine.

When the sweat glands from the foot pad of a newborn rat are transplanted into a region that is normally innervated by noradrenergic sympathetic neurons, the synaptic neurons acquire cholinergic transmitter properties, indicating that cells of the sweat gland secrete factors that induce cholinergic properties in sympathetic neurons.

Several secreted factors trigger the switch from a noradrenergic to cholinergic phenotype in sympathetic neurons. The sweat gland secretes a cocktail of interleukin 6-like cytokines, notably cardiotrophin-1, leukemia inhibitory factor, and ciliary neurotrophic factor. Several aspects of neuronal metabolism that are linked to transmitter synthesis and release are controlled by these factors. The neurons stop producing the large dense-core granules characteristic of noradrenergic neurons and start making the small electron-translucent vesicles typical of cholinergic neurons (Figure 53–11).

The Transmitter Phenotype of a Central Neuron Is Controlled by Transcription Factors

Neurons that populate the central nervous system use two major neurotransmitters. The amino acid l-glutamate is the major excitatory transmitter and GABA is the major inhibitory transmitter. Distinct molecular programs are used to establish neurotransmitter phenotype in different brain regions and neuronal classes. We shall illustrate the general strategy for assignment of amino acid neurotransmitter phenotypes by focusing on neurons in the cerebral cortex and cerebellum.

The cerebral cortex contains glutamatergic pyramidal neurons that are generated within the cortical plate and rely on the bHLH factors neurogenin-1 and neurogenin-2 for their differentiation. In contrast, as we discussed earlier in the chapter (see "Axons Tracts Serve As a Scaffold for Tangential Migration"), most GABAergic inhibitory interneurons migrate into the cortex from the ganglionic eminences, and their inhibitory transmitter character is specified by the bHLH protein Mash1 (Figures 53–12A) as well as by the Dlx1 and Dlx2 proteins.

Similarly, the cerebellum contains several different classes of inhibitory neurons (Purkinje, Golgi, basket, and stellate neurons) and two major classes of excitatory neurons (granule neurons and large cerebellar nuclear neurons). These inhibitory and excitatory neurons have different origins; GABAergic neurons derive from the ventricular zone, whereas glutamatergic neurons migrate into the cerebellum from the rhombic lip. The generation of GABAergic and glutamatergic neurons is controlled by two different bHLH transcription factors, Ptf1a for inhibitory and Math-1 for excitatory neurons (Figure 53–12B). These bHLH factors are expressed by neuroepithelial cells but not by mature neurons, implying that differentiation into glutamatergic and GABAergic neurons is initiated prior to neuronal generation.

Not all neurons are destined to survive, and decisions about life and death are therefore aspects of a neuron's fate. The surprising and counterintuitive fact is that cell death is preprogrammed in most animal cells, including neurons.

The Neurotrophic Factor Hypothesis Was Confirmed by the Discovery of Nerve Growth Factor

The target of a neuron is a key source of factors essential for the neuron's survival. The critical role of target cells in neuronal survival was discovered in studies of the dorsal root ganglia.

In the 1930s Samuel Detwiler and Viktor Hamburger discovered that the number of sensory neurons in embryos is increased by transplantation of an additional limb bud into the target field and decreased if the limb target is removed. At the time these findings were thought to reflect an influence of the limb on the proliferation and subsequent differentiation of sensory neuron precursors. In the 1940s, however, Rita Levi-Montalcini made the startling observation that the death of neurons is not simply a consequence of pathology or experimental manipulation, but rather occurs during the normal program of embryonic development. Levi-Montalcini and Hamburger went on to show that removal of a limb leads to the excessive death of sensory neurons rather than a decrease in their production.

These early discoveries on the life and death of sensory neurons were quickly extended to neurons in the central nervous system. Hamburger found that approximately half of all motor neurons generated in the spinal cord die during embryonic development. Moreover, in experiments similar to those performed on sensory ganglia, Hamburger discovered that motor neuron death could be increased by removing a limb and reduced by adding an additional limb (Figure 53–13). These findings indicate that signals from target cells are critical for the survival of neurons within the central as well as peripheral nervous system. We now know that the phenomenon of neuronal overproduction, followed by a phase of neuronal death, occurs in most regions of the vertebrate nervous system. Blocking neuromuscular activity with drugs such as curare also reduces the extent of motor neuron death (Figure 53–13).

The early discoveries of Levi-Montalcini and Hamburger laid the foundations for the neurotrophic factor hypothesis. The core of this hypothesis is that cells at or near the target of a neuron secrete small amounts of an essential nutrient or trophic factor, and that the uptake of this factor by nerve terminals is needed for the survival of the neuron (Figure 53–14). This hypothesis was dramatically confirmed in the 1970s when Levi-Montalcini and Stanley Cohen purified the protein we now know as nerve growth factor (NGF) in the early 1970s and showed that this protein is made by target cells and supports the survival of sensory and sympathetic neurons in vitro. Moreover, neutralizing antibodies directed against NGF were found to cause a profound loss of sympathetic and sensory neurons in vivo.

Neurotrophins Are the Best Studied Neurotrophic Factors

The discovery of NGF prompted a search for additional neurotrophic factors. Today we know of over a dozen secreted factors that promote neuronal survival. The best studied of the neurotrophic factors are related to NGF and are called the neurotrophin family.

There are three main neurotrophins: NGF itself, brain derived neurotrophic factor (BDNF), and neurotrophin-3 (NT-3). Other classes of proteins that promote neuronal survival include members of the transforming growth factor β family, the interleukin 6-related cytokines, fibroblast growth factors, and even certain inductive signals we encountered earlier (BMPs and hedgehogs). Other neurotrophic factors, notably members of the glial cell line-derived neurotrophic factor (GDNF) family, are responsible for the survival of different types of sensory and sympathetic neurons (Figure 53–15).

Neurotrophins interact with two major classes of receptors, the Trk receptors and p75. Neurotrophins promote cell survival and promote cell death through activation of Trk receptors, whereas signaling through the p75 receptor promotes cell death. The Trk family comprises three membrane-spanning tyrosine kinases named TrkA, TrkB, and TrkC, each of which exists as a dimer (Figure 53–16).

As with other tyrosine kinase receptors, the binding of neurotrophins to Trk receptors leads to phosphorylation of an intracellular domain of the receptor, resulting in the dimerization of Trk proteins and phosphorylation of specific tyrosine residues in the activation loop of the kinase domain. Phosphorylation of these residues leads to a conformational change in the receptor and to phosphorylation of tyrosine residues that serve as docking sites for adaptor proteins. Activation of Trk receptors promotes the survival of neurons and also triggers their differentiation. These divergent biological responses involve different intracellular signaling pathways: neuronal differentiation largely by the mitogen-activated kinase (MAPK) enzymatic pathways and survival largely by the phosphatidylinositol-3 kinase pathway (Figure 53–17).

In contrast to the specificity of Trk receptor interactions, all neurotrophins bind the receptor p75 (Figure 53–16). The activation of p75 promotes neuronal survival through a pathway that involves activation of the NF-B enzyme. In the absence of exposure to neurotrophins, p75 receptor activation promotes neuronal death. Receptor p75 is a member of the tumor necrosis factor (TNF) receptor family and promotes cell death by activating proteases of the caspase family, which we discuss below.

Neurotrophin signaling is relayed from the axon terminal to the cell body of the neuron through a process that involves internalization of a complex of neurotrophin bound to Trk receptors. The retrograde transport of this complex occurs in a class of endocytotic vesicles called signaling endosomes. The transport of these vesicles brings activated Trk receptors into cellular compartments able to activate signaling pathways and transcriptional programs essential for neuronal survival.

The picture is more complex for neurons in the central nervous system. The survival of motor neurons, for example, is not dependent on a single neurotrophic factor; different classes of motor neurons require neurotrophins, glial cell line-derived neurotrophic factor (GDNF), and interleukin-6-like proteins expressed by muscles or peripheral glial cells. The survival of these neuronal classes depends on the exposure of axons to local neurotrophic factors.

Neurotrophic Factors Suppress a Latent Death Program in Cells

Neurotrophic factors were once believed to promote the survival of neural cells by stimulating their metabolism in beneficial ways, hence their name. It is now evident, however, that neurotrophic factors suppress a latent death program present in all cells of the body, including neurons.

This biochemical pathway can be considered a suicide program. Once it is activated, cells die by apoptosis (Greek, falling away): They round up, form blebs, condense their chromatin, and fragment their nuclei. Apoptotic cell deaths are distinguishable from necrosis, which typically results from acute traumatic injury and involves rapid lysis of cellular membranes without activation of the cell death program.

The first clue that deprivation of neurotrophic factors kills neurons by unleashing an active biochemical program emerged from studies that assessed neuronal survival after inhibition of RNA and protein synthesis. Exposure of sympathetic neurons to protein synthesis inhibitors was found to prevent the death of sympathetic neurons triggered by removal of NGF. These results sparked the idea that neurons are always capable of synthesizing proteins that are lethal and that NGF prevents their synthesis: Neurotrophins suppress an endogenous death program.

Key insights into the biochemical nature of the endogenous cell death program emerged from genetic studies of the nematode Caenorhabditis elegans. During the development of C. elegans a precise number of cells is generated and a fixed number of these cells die—the same number from embryo to embryo. The findings prompted a screen for genes that block or enhance cell death, which led to the identification of the cell death (ced) genes. Two of these genes, ced-3 and ced-4, are needed for the death of neurons; in their absence every one of the cells destined to die instead survives. A third gene, ced-9, is needed for survival, and works by antagonizing the activities of ced-3 and ced-4 (Figure 53–18). Thus, in the absence of ced-9 many additional cells die, even though these deaths still depend on ced-3 and ced-4 activity.

The cell death pathway in C. elegans has been conserved in mammals. Similar proteins and pathways control the apoptotic death of central and peripheral neurons, indeed of all developing cells. The worm ced-9 gene encodes a protein that is related to members of the mammalian Bcl-2 family, which protect lymphocytes and other cells from apoptotic death. The worm ced-3 gene encodes a protein closely related to a class of mammalian cysteine proteases called caspases. The worm ced-4 gene encodes a protein that is functionally related to a mammalian protein called apoptosis activating factor-1 or Apaf-1.

The mammalian apoptotic cell death pathway works in a way that resembles the worm pathway. The morphological and histochemical changes that accompany the apoptosis of mammalian cells result from the activation of caspases, which cleave specific aspartic acid residues within cellular proteins. Two classes of caspases regulate apoptotic death: the initiator and effector caspases. Initiator caspases (caspase-8, -9 and -10) cleave and activate effector caspases. Effector caspases (caspase-3 and -7) cleave other protein substrates, so triggering the apoptotic process. Perhaps 1% of all proteins in the cell serve as substrates for effector caspases. Their cleavage contributes to neuronal apoptosis through many pathways: by activation of proteolytic cascades, inactivation of repair, DNA cleavage, mitochondrial permeabilization, and initiation of phagocytosis.

The survival of mammalian neurons is determined by the balance between anti-apoptotic and pro-apoptotic members of the Bcl-2 family of proteins. Some Bcl-2- proteins such as BAX and BAK permeabilize mitochondrial outer membranes, causing the release of pro-apoptotic proteins such as cytochrome c into the cytosol. The release of cytochrome c induces Apaf-1 to bind and activate caspase-9, leading to the cleavage and activation of effector caspases. The binding of neurotrophic factors to their tyrosine kinase receptors is thought to lead to the phosphorylation of protein substrates that promote Bcl-2-like activities (Figure 53–19B). Thus withdrawal of neurotrophic factors from neurons changes the balance from anti-apoptotic to pro-apoptotic members of the Bcl-2 family, which triggers the neuron's demise.

The caspase cell death program can also be activated by many cellular insults, including DNA damage and anoxia (Figure 53–19A). The activation of cell-surface death receptors such as Fas by extracellular ligands results in the activation of caspase-8 or -10 as well as the recruitment of death effector proteins such as FADD. Recruitment of an initiator caspase to the Fas-FADD complex then leads to activation of effector caspases. Because many neurodegenerative disorders result in apoptotic death, pharmacological strategies to inhibit caspases are under investigation.

The nervous system generates an overabundance of neurons and superfluous cells. From beginning to end, intercellular signals provide crucial direction to the developing nervous system. After many years of descriptive embryology we now have the first molecular insights into two fundamental issues in neurogenesis: the mechanisms by which cells acquire neuronal and glial identities and those by which certain young neurons and glial cells survive at the expense of others.

Major insights into these aspects of neurogenesis have emerged from genetic studies of two invertebrate organisms: the fruit fly Drosophila melanogaster and the nematode worm Caenorhabditis elegans. This research has shown, once again, the striking phylogenetic conservation of the molecular machinery responsible for animal development. Yet insight into the trophic factors that promote the development and survival of nerve cells came first from studies of vertebrates. Moreover, research on neurotrophic factor signaling and the biochemistry of cell death mechanisms is beginning to be applied to the search for treatment of neurodegenerative disorders.