Neuronal Function

The emphasis in the previous section was on the diversity of nerve cell types. Discussion of the functioning of the adult neuron, however, requires a focus on commonalities. The goal of this section is not to describe the biochemical and biophysical mechanisms that the cell uses to create, store, and transmit electrical signals; this information is provided in Chapter 49. Rather, this section follows a packet of information as it moves through a typical nerve cell in the brain, with the goal of introducing the biology of the cell. Figure 6-2A is a diagram of one typical neuron. Synaptic input occurs on all synaptic spines, as shown in the enlarged spines at the far left. Synaptic spines can receive input from a single (Fig. 6-2B) or multiple (Fig. 6-2C) axons.

FIGURE 6-2 The main domains of a neuron. A, Schematic view of a neuron. A typical neuronal impulse would travel from the dendrite at the left through the cell body to the axon hillock (initial segment), where an action potential would be generated, and end at the synaptic terminal. B and C, Electron micrographs illustrating synaptic structure. In B, a simple Purkinje cell dendritic spine (s) exhibits the characteristic features of a thin neck, bulbous head, and smooth endoplasmic reticulum. It forms a synapse with a granule cell axon (a). In C, nine presynaptic terminals (numbered) synapse on the dendritic spine.

(A, From Hof P, Trapp BD, de Vellis J, et al. The cellular components of nervous tissue. In: Zigmond M, Bloom FE, Landis SC, et al, eds. Fundamental Neuroscience. San Diego, CA: Academic Press; 1999:41-69; B, from Palay S, Chan-Palay V. Cerebellar Cortex: Cytology and Organization. Berlin: Springer-Verlag; 1974; C, from Peters A, Palay SL, Webster HF. The Fine Structure of the Nervous System: Neurons and Their Supporting Cells, 3rd ed. New York: Oxford University Press; 1991.)

The information arrives at the nerve cell at a highly specialized structure known as a synapse. At this site, a process from the previous neuron in the circuit approaches to within a few hundred angstroms of the next cell but does not touch it. The gap between the cells remains continuous with the extracellular space. The presence of the gap requires a specialized mechanism for transferring an electrical signal from the presynaptic to the postsynaptic cell. This transfer is achieved by way of the presynaptic cell secreting a chemical transmitter substance into the gap. The transmitter is usually a small molecule such as acetylcholine or noradrenaline, but it can also be a peptide such as substance P or vasointestinal peptide. Diffusion carries this pulse of chemicals across the gap to the membrane of the postsynaptic cell, which is covered with receptor proteins. These receptors recognize the secreted chemicals and transform the information from the chemical pulse into an electrical event that can be propagated down the neuron. This mechanism is discussed in more detail later. A synapse can occur virtually anywhere on the postsynaptic cell, but the most common location is on the neuronal dendrite. The dendrite is usually the first part of the neuron to initiate an electrical response to a signal from the presynaptic neuron. On many neurons, the dendrites have specialized adornments known as spines, and it is with these structures that the presynaptic neuron forms a synapse. Spines can vary in size and shape, but they are generally no more than a few micrometers in length, with a bulb-like shape at the end of a tapered shaft (see Fig. 6-2). An electrical signal that initiates in a spine travels to the dendritic branch on which it occurs and moves down the dendrite toward the cell body. Although many neuritic processes look similar (especially in culture), a dendrite can usually be distinguished from an axon because it is tapered—decreasing in diameter as distance from the cell body increases. A nerve cell can have a single dendritic shaft emanating from its cell body (as in the Purkinje cell in Fig. 6-1A), or it can have several. A typical pyramidal cell in the cerebral cortex, for example, has a single apical dendrite but also several basal dendrites; a cerebellar granule cell has four to six short dendrites in a star-shaped configuration around the cell body.

The nerve cell body is the most prominent feature of the neuron in most basophilic stain preparations (e.g., hematoxylin-eosin). If one considers size alone, however, the cell soma is usually considerably smaller than the dendrite in both surface area and volume.² This relatively small soma-to-process volume ratio represents a logistic problem for the cell body: the genetic machinery is located exclusively in the nucleus, yet the products of this machinery must be transported to sites on the cell that are up to a meter away.

The electrical signal reaching the cell body from the dendrite travels to the point on the cell where the axon emerges. The axon of a neuron is the structure that captures the summed electrical information in the dendrites and cell body and routes it to the next neuron (or target organ) in the circuit. Morphologically, the axon can be distinguished from the dendrite by the fact that it has a typically constant caliber over its entire length. Soon after the axon leaves the cell body, a specialized region known as the initial segment (or axon hillock) can be identified. This part of the cell is the biochemical boundary of the axon and the point of initiation of the action potential. Up to this point, the information packets from the dendrites and cell body travel primarily by electrotonic spread, with different packets of electrical activity coming together and summing in a graded (analog) manner. By contrast, the axon transmits information in a strictly digital fashion. If the combined electrogenic signal reaching the axon from the cell body is sufficiently strong, the axon will fire and pass the information along; otherwise, the signal stops and proceeds no farther in the circuit. If the decision is a “go,” the axon transforms the information into a self-propagating electrical wave known as an action potential that travels undiminished down the axon to its end. The action potential is an electrical signal that results from the coordinated functioning of sodium and potassium channels (see Chapter 49 for details), usually in collaboration with glial cells (see later).

The synaptic terminal is the site where the information packet leaves the neuron for the next cell in the circuit. The morphology of the terminal is topologically complementary to the postsynaptic site to which it will transfer its information. The regular caliber of the axon and its linear array of constituent microtubules end, the axon swells in diameter, and a new collection of vesicles and mitochondria are found. When the action potential reaches the presynaptic terminal, a series of biochemical events are initiated that result in the secretion of a burst of neurotransmitter; this enables the information packet to pass across the acellular gap of the synapse and initiates an electrical response in the next cell.

Although the preceding description applies directly to many neurons, it is inadequate to fully describe the large number of cells that function as neurons in the human body. Even though each neuron transmits information from one part of its cell body to another, the nature of the information transferred is often quite different from one cell to the next. This diversity of functions is paralleled by an equal diversity of morphologies and biochemistries.

Sensory Neurons

The fundamental function of the nervous system is to enable an organism to respond quickly to its environment. To this end, a wide variety of cell types have evolved that efficiently transform information about the environment into electrical impulses that can be integrated and translated into a behavioral response. These cells are neurons, but rather than receiving information from a preceding neuron in a circuit, their input comes in the form of signals in the environment. These signals can be grouped into three basic modalities: mechanical, chemical, and physical.

Mechanical Receptors

The simplest receptor cells of this type are used to receive information about touch and pain. As might be expected, most are located in the skin and other integuments. Receptors for light touch are illustrated in Figure 6-3. These receptor endings have different precisions and sensitivities and are specialized to receive different types of stimuli.³ Each receptor represents a specialized ending of a sensory ganglion axon. These include Ruffini’s endings or pacinian corpuscles (for deep receptors) and Merkel cells or Meissner’s corpuscles (for more superficial receptors). Sometimes the neurite of a mechanoreceptor is wrapped around the interfusal muscle fibers of one of the striated muscles. The cellular deformation associated with movement of the axonal membrane activates a series of stretch-sensitive ion channels. The ionic current through these channels initiates the electrical activity that signals a sensory stimulus to the organism.

FIGURE 6-3 Touch receptors. The diversity of neuronal form occurs not only in the overall shape of the cell but also in the fine structure. These four different morphologies serve specific receptor functions, yet each is associated with a single neuronal cell type: the dorsal root ganglion neuron.

(Modified from Kandel ER, Schwartz JH, Jessell TM. Principles of Neural Science. New York: Elsevier; 1991.)

A subset of mechanical receptors has evolved to serve the auditory and vestibular systems. The principal function of these cells is the same as for light touch receptors: deformation of a hair opens a number of specialized ion channels, which results in the generation of an electrical signal. The difference in the acousticovestibular system is that the “hairs” are actually cilia on the basal surface of the cell and hence are part of the receptor cell itself. Indeed, the ion channels that are opened in response to movement of the hair are located at the tip of the cilium. A diagram of this cell type is shown in Figure 6-4. In the auditory system, the vibration of sound waves is transduced into the vibration of fluid in the cochlea. Receptor cells at different positions in the cochlear spiral respond to different auditory frequencies and transmit both pitch and volume information to the auditory system. In the vestibular system, a morphologically similar configuration of receptors is found in the semicircular canals. Movement of the head in any of the three orthogonal planes leads to movement of fluid in the canals. This movement displaces the cilia of the vestibular receptors and initiates an electrical signal that is transmitted through the eighth nerve to the brain, where the vestibular system interprets the information to determine the orientation and movement of the organism in space. In each of these systems, however, the common feature of the receptor is that deflection of a hair leads to deformation of the cellular membrane, which in turn results in opening of a specific set of ion channels. The resulting change in conductance leads to an electrical “packet” of information that moves along the neuron to the rest of the brain.

FIGURE 6-4 Typical hair cell of the ear. Movement of fluid causes the stereocilia to become displaced and opens a dedicated class of ion channel that depolarizes the cell and signals through the synaptic body to the afferent nerve fiber. The response of the cell can be modified centrally through depolarizations relayed through a reciprocal efferent connection.

(Modified from Hof P, Trapp BD, de Vellis J, et al. The cellular components of nervous tissue. In: Zigmond M, Bloom FE, Landis SC, et al, eds. Fundamental Neuroscience. San Diego, CA: Academic Press; 1999:41-69.)

Effector Neurons

By far the most common target of a neuron is another neuron. Thus, most axons terminate with synapses on the dendritic spines or shafts of other neurons. Cell bodies and even presynaptic terminals also serve as common axonal targets. The networks of neurons formed by these interconnections are what allow an organism to perform complex behavioral responses to environmental stimuli. The axon of a neuron can have other targets, however; usually, the function of these cells is to transmit the calculations of the nervous system to a non-nervous cell or structure and effect a change in organism behavior.

Neuronal Organization

When several neurons link to form a circuit, sophisticated computational feats can be accomplished and useful work performed. Because of this higher level of organization, neurons can be categorized on the basis of not only their own shape and size but also the role that they play in a circuit. For example, neurons are often divided into two classes based on the distance of the target from the cell body. Projection neurons send long axons to other neurons in distant regions of the nervous system. Thalamocortical neurons (thalamus to cerebral cortex) and dentatorubral neurons (cerebellar dentate nucleus to red nucleus) are two examples of this type of cell. Local circuit interneurons, by contrast, have short ramified axons; they participate mainly in local computational processes.

Neurons can also be classified according to the function that they perform within a circuit. In the simple descriptions of the previous sections, the firing of a neuron was assumed to stimulate the next neuron to fire as well. In fact, neurotransmitters can also inhibit the electrical activity of other neurons. The function of an inhibitory neuron, therefore, is to suppress the action of the downstream target, thus making it less likely to fire. The role of an inhibitory neuron is crucial to the more sophisticated functions of a neural circuit. It allows temporal or spatial sharpening of a response by preventing neighboring cells from firing or dampening the runaway tendency of a circuit. Epileptic activity may well be due to failure of such inhibitory action.

Neuronal Structure

Visualization of Neurons

Neurons, like most other cells, are pale, clear, and difficult to see in living tissue. As a consequence, many of the significant advances in the study of the nervous system were made possible by improved methods of visualizing the nerve cell and its processes. The basophilic dyes are the oldest and still most widely used method of staining nervous tissue. Common stains of this class include hematoxylin and cresyl violet. These dyes bind avidly to RNA and DNA and thus highlight the heterochromatin of the nucleus and the rough endoplasmic reticulum of the cytoplasm (Fig. 6-5A). Basophilic stains are most commonly used on sections with a thickness ranging from 2 to 20 µm. Examination of nervous tissue stained with such reagents results in a clear picture of the cell body and proximal dendrites; the axons and distal dendrites are usually invisible, so white matter and neuropil are generally clear of stain.

FIGURE 6-5 Different staining methods reveal different features of neurons. A, Cresyl violet staining. This image is from the cerebellar cortex of an adult mouse. Note the strong staining of nuclear heterochromatin (especially in the cell-dense granule cell layer) and the rough endoplasmic reticulum (Nissl substance) of the Purkinje cell. Neuropil is unstained. B, Golgi impregnation. These mouse hindbrain neurons have been filled with heavy metal salts. Note the fine dendritic processes that emanate from the cell body. C, Immunocytochemistry. Human cortical neurons are stained with antibodies specific for neurofilament proteins, a major cytoskeletal component of neurons, dendrites, and axons. The dark precipitate was formed by subsequent incubation with an enzyme-linked antibody against rabbit immunoglobulin, followed by reaction with chromogenic enzyme substrates. D, In situ hybridization for messenger RNA encoding the orphan nuclear receptor RORα. A field from the mouse cerebellar cortex, similar to A, is shown. Digoxigenin-substituted nucleotides in the oligonucleotide probe are revealed with antidigoxigenin antibody.

During the 1800s, silver salts were found to have a special avidity for nervous tissue. Because of their binding to neuron-specific classes of intermediate filaments, a variety of protocols were developed that revealed the neuronal axon with great clarity. Among this class of stains, the Bodian and Bielschowsky stains are still commonly used. A special class of silver stain is the Golgi impregnation method. In this procedure, pieces of tissue are incubated for many weeks in heavy metal salts. During this time, a small number of cells take up the salts, with their entire intracellular spaces being filled. The tissue is then embedded and sectioned. When the sections are “developed” in reducing agents, an opaque black precipitate is formed that fills the impregnated cells completely. For unknown reasons, only 1% to 2% of the cells react in this fashion (seemingly at random). The details of an individual cell can be seen against a clear background (Fig. 6-5B). Although the technique reveals the finest details of dendritic structure, axons are more resistant to filling and are commonly invisible in Golgi preparations. The technique is used to best advantage in sections ranging from 80 to 120 µm thick.

In the second half of the 20th century, new technologies dramatically expanded our ability to visualize and analyze the nerve cells of the brain. Beginning in the 1950s and 1960s, the transmission electron microscope led to a quantum leap in the ability to resolve the details of nerve cellular structure. In this method, small pieces of tissue (typically 2 to 3 mm wide) are embedded in plastic and cut with a glass or diamond blade in sections ranging in thickness from tens to hundreds of nanometers. Before embedding, the tissues are usually stained with uranyl acetate, lead citrate, and osmium tetroxide, lipophilic dyes that reveal membrane structure with a high degree of clarity. Phosphotungstic acid is a frequently used stain that has a particular affinity for synapses. The resolution afforded by the electron microscope allows the fine structure of the cell to be revealed, and the organelles of the cell body can be analyzed. The unique advantages of using electron microscopy to view the nervous system include the ability to resolve synaptic structures (see Fig. 6-2B and C) such as synaptic vesicles, details of the presynaptic and postsynaptic membranes, and the material of the synaptic cleft. Axon and dendrite morphology, with their unique collections and arrangements of filaments, can also be seen.

In the 1970s and 1980s, serum antibodies were developed as highly specific stains by using the techniques of immunocytochemistry. Lightly fixed tissue is exposed to an antiserum or monoclonal antibody raised against a particular neuronal epitope. The antibody selectively binds to the neural structure that contains that epitope. Most frequently, this primary antibody is revealed through the application of a secondary antibody that is derivatized to carry a detection molecule—either a fluorescent compound, such as fluorescein or rhodamine, or an enzyme, such as horseradish peroxidase or alkaline phosphatase. In the former case, the location of the antibody is determined by examining the tissue with a fluorescence microscope. In the latter instance, the marker enzyme is localized through the use of a specific substrate whose action deposits an insoluble, chromagenic product. Immunocytochemistry is most commonly used to reveal proteins (e.g., tyrosine hydroxylase, microtubule-associated protein 2 [MAP2], the α₆ receptor of γ-aminobutyric acid [GABA_A]), but the location of carbohydrates (e.g., polysialic acid, gangliosides, chondroitin sulfate proteoglycan) can also be determined. Immunocytochemistry can be applied at the electron microscopic level as well, where peroxidase or gold particles are used to reveal the location of the secondary antibody. Figure 6-5C presents an example of immunocytochemistry using an antibody against a neurofilament protein, one of the major cytoskeletal components of neuronal cytoplasm.

In the 1980s and 1990s, advances in molecular biology enabled the detection of messenger RNA for specific proteins through a technique known as in situ hybridization (see Chapter 3). Medium- to high-abundance messages are localized to the cell body transcribing them by exposing lightly fixed tissue to labeled nucleotide probes that are complementary to the message sequence. The labeled probe is detected either because of its radioactivity (³⁵S-labeled nucleotide precursors are commonly used) or because of the presence of other derivatives (such as biotin or digoxigenin attached to nucleotides). In either case, the tissue is treated with proteinase to remove the bound proteins and then hybridized at temperatures that ensure the specificity of the probe-message interaction. If radioactivity is used, the location of the hybridized message is determined by apposing the section to x-ray film or by dipping the section in liquid photographic emulsion, which forms a thin coating and is subsequently fogged wherever the labeled nucleotide is found. If nonradioactive methods are used in the detection protocol, their location is revealed with a secondary probe (derivatized avidin in the case of biotin-labeled probes or antibody to digoxigenin). A digoxigenin-labeled probe hybridized to the message for the Purkinje cell–specific nuclear receptor RORα is shown in Figure 6-5D. An important aspect of the interpretation of such images is that the location of the message marks the cell body where the message is synthesized, not the place where the protein is found. For example, the image shown in Figure 6-5D could represent messenger RNA encoding RORα (a nuclear protein), the δ₂-glutamate receptor (found predominantly in dendritic spines), or synaptophysin (found principally in the presynaptic axon terminal).

Synaptic Structure

The physiology of synaptic function is covered in detail in Chapter 49. Underlying the details of ionic fluxes, however, are a number of crucial elements in the cell biologic structure of the neuron. On the presynaptic side, the axon terminates in a highly specialized presynaptic structure (see Fig. 6-2B and C). The shaft of the axon broadens in diameter and assumes a shallow cup-like form. Although microtubules are a prominent feature of the central axon domain, few extend into the terminal area. Instead, the major structural elements of the presynaptic terminal include neurofilaments and actin filaments. Mitochondria are more abundant than in the axon shaft, and a collection of small vesicles appears near the synaptic cleft itself. These vesicles contain the neurotransmitter substances that will be released on invasion of the terminal by an action potential. The vesicles are polymorphic in appearance. Those at excitatory synapses tend to be round, whereas those at inhibitory synapses are more ovoid or flattened. Synapses that release peptides contain larger vesicles with electron-dense material in their centers. These dense-core vesicles are typical of neurosecretory cells. Well in excess of a dozen proteins have been identified as being crucial to the process, which involves filling of the vesicle with transmitter, docking, fusion of the vesicle with cellular membranes, release, and finally, recycling.

Neurotransmitters are packaged and released from synaptic vesicles clustered in the vicinity of the synapse, where modern electron microscopic techniques suggest that they are interconnected with a meshwork of fine filaments. These vesicles must first be filled with neurotransmitter, a function accomplished by a variety of transport proteins. In an active synapse, the filled vesicles are then moved to a region near the active zone where release occurs—a region of about 0.1 µm². As conditions demand, the vesicles dock on the plasma membrane of the active zone. An adenosine triphosphate (ATP)-dependent process then “primes” the vesicle. In this state, the influx of Ca⁺² ions that accompanies invasion of the action potential into the nerve terminal triggers a process of vesicle-plasma membrane fusion that releases the contents of the vesicle into the synaptic cleft. Vesicle recycling then occurs, although the precise mechanism of this process is still debated and may be dependent on the specific synapse. In general, an endocytic process recycles the fused vesicular membrane back into the cell. The vesicular and plasma membrane proteins that accomplish this process have been well studied, but a full listing of their identities and functions is beyond the scope of this volume.

The Cell Biology of Neuronal Death

Included among the important physiologic functions of a brain cell is its program of cell death. Cell death was once viewed as a passive process, but our understanding of its biology has advanced to the point where we now appreciate that most cell loss is an active process that involves a well-orchestrated program of gene expression, proteolysis, and rearrangement of cellular organelles.^10–13 The programs of cell death are generally divided into the more active form of cell death known as apoptosis and the more passive form known as necrosis. This duality is oversimplified, however; a more nuanced and probably more accurate characterization is that there are many processes that contribute to cell death. It is the summation of these processes along with the internal protective mechanisms that usually come into play in a neuron under stress that ultimately determines whether a neuron will live or die.

Apoptosis is also referred to as programmed cell death or nuclear cell death. During this process, a sequence of genetic transcription and translation is initiated. At a morphologic level, apoptosis is a process that appears to package the cell for removal without initiating an inflammatory response. The cell membrane blebs as the cell shrinks in size. The nucleus condenses and fragments, and the cell dumps its water as it shrinks, with consequent darkening of the cytoplasm. In the end, there remains only a small, round, membrane-bound corpse that is easily ingested by a local macrophage. Different pathways have been found that produce apoptosis. The intrinsic pathway begins with the breakdown of mitochondrial integrity and massive release of cytochrome C from the mitochondria into the cytoplasm. This catalyzes formation of the protein complex known as the apoptosome, which in turn activates one or more members of a group of cysteine-aspartate proteases known as caspases by proteolytically cleaving their “pro” form to generate the activated form of the caspase. There are currently 11 known caspase enzymes. Their substrates range from other caspases to many important cellular targets, where cleavage can either activate a zymogen or destroy a crucial component of cellular homeostasis. A second apoptotic pathway is known as the extrinsic pathway. It is triggered by binding of a ligand to any of several death receptors such as tumor necrosis factor (TNF) or Fas. The receptors trimerize and position pro-forms of caspase-8 or caspase-10 to be proteolytically activated. In virtually every situation that has been studied, adult neurons that are lost to cell death cannot be replaced from endogenous stem cells or other sources. It is not surprising, therefore, that the progress of apoptosis—both the intrinsic and extrinsic pathways—is regulated by several means. Mitochondrial integrity is regulated by a family of small peptides known as the Bcl2 family, which includes Bcl2, Bax, Bad, and others. These proteins bind to each other, and the dimers can either promote or inhibit cell death. Other factors that work against completion of the final apoptotic pathway include such proteins as inhibitors of apoptosis (IAPs).

Necrosis is often referred to as cytoplasmic cell death and is the opposite of apoptosis. The process is believed to be a largely passive one that is not dependent on participation of the synthetic processes of the cell such as transcription or translation. The ATP stores of the cell drop, thereby starving the cell for the energy that it needs for normal homeostatisis. As a result, the cell takes on water and swells, as do the constituent organelles. The mitochondria swell, and the cytoplasm takes on a dilute appearance. Eventually, the surface membrane of the cell loses its integrity, and the cellular contents are dumped into the extracellular space. This process is most common in the immediate aftermath of brain trauma or during certain types of metabolic imbalance.

Recently, autophagy, a process of bulk cellular waste removal, has been proposed as a distinct mechanism of cell death. The normal function of autophagy is to enable the cell to remove large particles of debris such as aggregates of misfolded proteins from the cytoplasm. It is now suspected that this process can become overactive, overwhelm the protective devices of the cell, and thus cause it to literally eat itself. The morphology of autophagic cell death includes swelling of organelles, loss of cytoplasmic membrane integrity, and vacuolization of the cytoplasm. A role for this form of cell death has been proposed in neurodegenerative disease.

The triggers for these cell death programs are diverse. A common example is that of excitotoxic cell death. If a neuron becomes hyperactive, one unavoidable consequence is the accumulation of abnormally high concentrations of intracellular calcium. This calcium activates a variety of calcium-activated proteases, channels, and pumps that will ultimately initiate a caspase cascade leading to an apoptotic crisis. This type of nerve cell death is common after a seizure (hyperactivity of a neuronal network) or a vascular insult (local depolarization inducing concentrations of various ions). Although excitotoxic cell death appears to largely be apoptotic in nature, it has been suggested that it may in fact represent a fourth uniquely neuronal form of cell death.

Another emerging association with neuronal cell death is loss of cell cycle regulation. This represents an odd situation in that most adult neurons exit the cell cycle during embryogenesis, never to return (see Chapter 4). Nonetheless, in a variety of neurodegenerative conditions, including Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis, and ataxia-telangiectasia, as well as stroke and human immunodeficiency virus–associated dementia, neurons that are known to be at risk for death are found to re-express proteins that are normally found only in cells that are actively engaged in a cell cycle. These proteins include such cell cycle components as Ki-67, cdk4, cyclin D, p16, proliferating cell nuclear antigen, cyclin B, and cdc2.^14–17 Curiously, the cycle is not a productive one; it is never completed. DNA replication is involved, as studies with DNA hybridization techniques have shown in both Alzheimer’s and Parkinson’s disease, and chromosome copy number increases, as would be expected during a normal S phase of the cell cycle. The actual mitotic process, however, does not occur; chromosomes do not condense and mitotic spindles are not observed, nor are any forms of cytokinesis. Studies in tissue culture suggest that cell cycle initiation is an early part of the death pathway inasmuch as blocking the cycle blocks the death. In the adult brain, however, the linkage between the two processes is not direct; it is estimated that it can be many months to a year from the time that a neuron intitates a cycle until it finally dies. Whether this is a new type of cell death or merely a precursor or trigger to apoptosis or necrosis is not known.

The idea that entrance into a cell cycle might be lethal for a neuron even while it is restorative for other cells in other tissues highlights the fact that neurons, because of their highly specialized form and function, are vulnerable to a variety of insults. Thus, because they are postmitotic and highly differentiated, they cannot tolerate reinitiation of the cell cycle.¹⁸ Similarly, because they are in a nonmitotic state, repair of DNA damage is a crucial neuroprotective activity. Syndromes that involve disruptions of genes that encode elements of the DNA repair pathways nearly always include neurodegeneration as part of their phenotype.¹⁹ Excitotoxicity also reflects unique neuronal vulnerability. Because its internal calcium concentration is exquisitely sensitive to neurotransmitters, it is highly susceptible to damage from overactive synaptic activity. Oxidative damage is also a common trigger for neurodegenerative disease, and failure of antioxidant strategies of the cell can result in death. In truth, both the triggers of neuronal cell death and the death pathways themselves probably have significant overlap and may represent a continuum rather than distinct and wholly separate events. For example, a regular target of oxidative damage is DNA, a lesion that would trigger both DNA repair and antioxidant responses. Similarly, the processes usually associated with apoptosis can be found in cells that have lost their ATP stores and thus would be considered to be undergoing necrosis.

Neurodegenerative Diseases

There are many types of neurodegenerative disorders that afflict humans in which these processes of neuronal cell death run amok. These disorders vary in their age at onset and the specificity of the affected cell populations. Alzheimer’s disease is an example of these conditions. Its disease process not only highlights several features of neuronal cell death but also emphasizes the importance of interaction among the various nervous system cell types. Alzheimer’s disease is a progressive dementia that affects millions of individuals in the United States alone. Clinically, the disease is manifested as loss of short-term memory, failure of executive function, and a variety of behavioral disorders such as depression and apathy. The disease is defined by the presence of two pathologic features: extracellular deposits of a peptide fragment known as β-amyloid in a largely insoluble aggregate known as a plaque and twisted configurations of fibrils (made up largely of neurofilaments and filament-associated proteins) known as a neuritic tangle. The course of the illness involves not only neurons but also several non-neuronal cells. In the vicinity of the plaque are reactive astrocytes, as well as activated microglial cells^20,21 (discussed later in this chapter). The astrocytes appear to surround the plaque, as though trying to wall it off from the brain. The microglial cells appear to invade the plaque, as though trying to digest it. Associated with the dementia is loss of the majority of the neurons in several discrete neuronal populations: the hippocampus and entorhinal cortex, the basal nucleus of Meynert, the dorsal raphe, and the locus caeruleus. As the disease progresses, a more diffuse loss of neurons is observed that affects many regions of the frontal cortex and leads to significant atrophy of the brain in the later stages. As of this writing, the exact mechanism of neuronal death is controversial. Because the disease process itself can be many years in duration, only a tiny fraction of 1% of the cells in even the most affected population would be expected to be undergoing cell death at any one time. This slow loss of neurons makes it difficult to find the precise cells that are actively engaged in the process of dying. Perhaps as a consequence of this protracted disease course, no consistent evidence of either apoptosis or necrosis can be found in any of the affected populations. Like many other diseases, it is a condition that affects all the cells of the brain—neurons and non-neurons alike. For example, one probable scenario is that the aggregates of the β-amyloid peptide trigger a reaction in the astrocytes and microglial cells. This produces a complex inflammatory cascade that induces the release of numerous cytokines into the brain environment.^22,23 Vascular deposits of amyloid are also found during the disease, and a role of vascular endothelial cells has additionally been proposed. Thus, this common neurodegenerative condition of the elderly highlights both the severe consequences for the individual of losing neurons in large numbers and emphasizes the integrated nature of the cellular functions of the brain. Neurons may be the long-distance carriers of information in the brain, but in both health and disease, the brain works as an ensemble of multiple different cell types.

Astrocytes

Astrocytes are the most abundant cell within the CNS. It has been estimated that astrocytes constitute 20% to 50% of the volume of many brain areas. Astrocytes mediate many functions through physical contact between their processes and other CNS components. Therefore, astrocyte distribution, shape, and function reflect in part the distribution of the CNS components that they contact. Many astrocyte functions occur ubiquitously throughout the CNS, whereas others are restricted in location. Astrocyte morphologies range from bipolar to stellate. The major functions of astrocytes are summarized in Figure 6-6 and are related to astrocyte distribution and morphology as presented in the following paragraphs.

FIGURE 6-6 Relationships between astrocytes and other central nervous system cells. Astrocyte processes surround blood vessels (BV), synapses, nodes of Ranvier, neuronal cell bodies (blue), and groups of myelinated axons (A). They form the glial limitans at the pial and ependymal surfaces (not shown). The larger processes contain high concentrations of intermediate filaments. m, myelin; N, nucleus; basal lamina in orange.

How many types of astrocytes are there? Most textbooks identify three major astrocyte subtypes: radial glial, fibrous astrocytes of white matter, and protoplasmic astrocytes of gray matter. Astrocytes could be classified into dozens of subtypes (most found in gray matter) based on the differential expression of a variety of molecules, including ion channels, neurotransmitters, neurotrophin, and other cell surface receptors. Common to all astrocytes is a unique population of intermediate filaments enriched in glial fibrillary acidic protein. Bundles of 10-nm-thick intermediate filaments are a characteristic ultrastructural feature of astrocytes, as is an abundance of glycogen granules.²⁹ The latter is a reflection of the important role that astrocytes play in brain energy metabolism.

Oligodendrocytes

Rapid electrical communication between the 10¹¹ neurons in the human brain controls and integrates the sophisticated mental and motor functions that set us apart from other species. Consider, for example, a complex task such as a 6-ft human dunking a basketball. The motor, sensory, and decision-making circuitry of the brain and peripheral nerves must integrate and coordinate jumping, manipulating the torso, handling the ball, avoiding defenders, and putting the ball in the basket. Millions of coordinated nerve impulses govern this action, and many must travel more than a meter in a fraction of a second. Without rapid nerve conduction, the 10¹¹ neurons in the human brain would not be an advantage for function or survival. Two mechanisms have evolved to permit rapid conduction of nerve impulses. In invertebrates, axonal conduction velocity is related to the diameter of the axon.³⁶ Large-diameter axons conduct at a much faster rate than do small-diameter axons. Although this situation is sufficient for regulating neural function in smaller, less sophisticated organisms, the CNS of humans cannot accommodate the increase in axonal volume required for rapid nerve conduction. For example, the large-diameter motor axons conduct at a velocity of approximately 40 m/sec. If this conduction velocity were regulated solely by axonal diameter, the diameter of this axon would be several millimeters. Multiplying this by the millions of axons in the spinal cord would result in a spinal cord as wide as a telephone pole. Therefore, an additional mechanism evolved to accommodate rapid nerve conduction in the vertebrate brain. Analogous to the conduction of electrical wire, the mammalian nervous system increases the resistance and decreases the capacitance of axonal membrane potentials by surrounding axons with a multilamellar, tightly compacted membrane insulation called myelin (Fig. 6-7). Myelin is a specialized extension of the plasma membrane of the oligodendrocyte in the CNS³⁷ and of the Schwann cell in the PNS.³⁸ The length of individual myelin segments or myelin internodes varies between several hundred micrometers and 2 mm, with larger diameter axons having longer and thicker (more spiral wraps) myelin internodes. Along each axon, individual myelin internodes are separated from their neighbors by a node of Ranvier, a specialized unmyelinated axonal segment (1 to 5 µm in length) enriched in sodium channels and analogous to the axon hillock or initial axonal segment. Enrichment of voltage-sensitive sodium channels at the node generates ionic impulses only at the node (see Fig. 6-7). The action potential “jumps” from node to node by a process called saltatory conduction (see Chapter 49). Propagation of the action potential is also an energy-dependent process. Saltatory conduction consumes less energy than nonsaltatory conduction does. Therefore, myelin has three general advantages for the function of the human CNS: fast axonal conduction, energy conservation, and space conservation. Myelin is essential for normal neurological function, and a variety of inherited, metabolic, and immune-mediated myelin diseases occur in humans. Axonal degeneration is a consistent and neurologically important phenotype in the brains of individuals with myelin disease,³⁹ an observation indicating that myelin provides extrinsic trophic support for the survival of axons.^40,41 Similarly, when myelin fails to form developmentally, axons fail to mature. Myelination is therefore essential for the maturation and survival of axons and should be considered an integral part of nerve regeneration paradigms.

FIGURE 6-7 Oligodendrocytes (OL) in the central nervous system myelinate multiple axons (A). One myelin internode is “unrolled” to depict the continuity between oligodendrocytes and myelin internodes. The compact nature of myelin (m) is shown in the electron micrograph in the upper right. Myelin internodes end in paranodal loops at the nodes of Ranvier, depicted in the cut-away view. Myelination supports saltatory conduction, in which ion exchange occurs only at the nodes of Ranvier (lower schematic).

(From Peters A, Palay SL, Webster HF. The Fine Structure of the Nervous System: Neurons and Their Supporting Cells. 3rd ed. New York: Oxford University Press; 1991.)

The only known function of oligodendrocytes is myelination and axonal support. Consequently, the distribution of oligodendrocytes correlates with the density of axons requiring myelin. Oligodendrocyte-myelin ratios, however, are not directly proportional because different oligodendrocytes can myelinate different numbers of axons.⁴² Oligodendrocytes have small, round cell bodies, extend single processes to each myelin internode, and can be identified in tissue sections with a variety of myelin-specific antibodies. As expected, oligodendrocyte density is high in white matter tracts, where the cells often occur in rows oriented parallel to axons. Oligodendrocytes are also present at significant densities in gray matter because many axons terminating on and originating from neurons are myelinated. Oligodendrocyte cell bodies can be found close to neuronal cell bodies in gray matter or close to blood vessels in both gray and white matter. It is not known whether these cells perform functions other than myelination.

Schwann Cells

Myelin is formed by Schwann cells in the PNS, where it serves identical functions as CNS myelin. Schwann cells, however, differ from oligodendrocytes in that they can promote axonal regeneration. Schwann cells form single myelin internodes and surround these internodes with a specialized connective tissue or basal lamina (Fig. 6-8). The basal lamina forms a continuous tube around the entire length of each myelinated fiber. When a PNS axon is transected, the distal axonal segment rapidly degenerates, the myelin is removed, and Schwann cells multiply and form a continuum within each basal lamina tube (Fig. 6-9).⁴³ The proximal end of the axon, still connected to the neuronal cell body, does not degenerate and can regrow into the Schwann cell tube, which provides a substrate and often direct continuity for regeneration and reinnervation of appropriate targets. Denervated peripheral nerve segments, as well as Schwann cells isolated and expanded in vitro, have been experimentally transplanted into the CNS or PNS to promote axonal regeneration.^44,45 Although this is a common and often successful method of PNS regeneration in humans, Schwann cell transplantation as a therapy for human spinal cord injury is still under experimental development. Myelination is not the only function of Schwann cells. They also surround the perikaryon of PNS neurons, small-diameter axons, and portions of the neuromuscular junction (see Fig. 6-8). These Schwann cells have a molecular phenotype that differs significantly from myelinating Schwann cells.

FIGURE 6-8 Schwann cells form single myelin internodes in the peripheral nervous system (PNS). The upper left schematic depicts an unrolled myelin internode; the stippled area represents compact myelin (m) shown in the electron micrograph. The outer surface of PNS myelin internodes is surrounded by basal lamina (orange). Schwann cells (SC) also surround multiple unmyelinated axons (A), neuromuscular junctions, and neuronal perikarya (N) in PNS ganglia. Basal lamina also surrounds these Schwann cells.

FIGURE 6-9 Axonal regeneration after nerve transection in the peripheral nervous system (PNS). After transection of PNS axons (A), the distal axon and myelin degenerate (B). The proximal axon sprouts and uses the basal lamina (orange) as a substrate for regeneration (C). Regenerated axons are remyelinated (D) by Schwann cells.

Distribution of Microglia and Oligodendrocyte Progenitor Cells

Although the functions of microglia and OPCs are not known to be intimately related, their distribution and morphology are compared here because of their striking similarity. Both cell types are ubiquitous in the normal adult CNS, they project multiple processes from relatively small cell bodies, and each cell occupies distinct domains with little overlap between the processes of neighboring cells (Fig. 6-10). The processes of neighboring microglial cells do not touch; the ends of OPC processes are often closely apposed, and some project to blood vessels. Microglia and OPCs can be distinguished molecularly by immunocytochemistry and to the experienced observer by morphology. Regional variation in the density and shape of each cell type (e.g., white versus gray matter) occurs, however, indicative of the fact that their distribution and shape are regulated by local environments. Based on these morphologic characteristics, both cell types form a lattice-like network that covers most of the brain parenchyma. They are therefore appropriately positioned to function as homeostatic regulators of normal brain function and as guards ready to respond to brain pathology or neural dysfunction. Although this hypothesis is clearly established for microglia, little is actually known about OPC function in the adult brain.

FIGURE 6-10 Comparison of the distribution of oligodendrocyte progenitor cells (A) and microglia (B) in human brain sections. Both cell types establish a lattice-like network that covers much of the brain. Oligodendrocyte progenitor cells have more radially distributed processes than do microglia. Scale bars = 40 µm.

Function of Microglia

Microglia function as resident immune cells and phagocytes in the CNS. A lineal relationship between microglia and monocytes is supported by the following observations: all antibodies thus far raised against microglia also react with macrophages in peripheral tissues; microglia express activated monocyte markers, including major histocompatibility complex (MHC) class I and II; and microglia can process and present antigens to activated T cells. Microglia also display features not shared by peripheral macrophages and monocytes. Microglia proliferate spontaneously⁴⁶ and express a unique ion channel pattern that directs large, inwardly rectifying currents dependent on potassium.⁴⁷ Microglia can therefore be considered monocyte-derived cells that have evolved specialized features during their adaptation to the CNS. Although their functions in the normal adult brain are not fully understood, microglia have important roles in CNS development and disease. On the basis of current knowledge, it appears that most microglia originate from bone marrow–derived monocytes that enter the brain during early development. Initially ameboid in shape, these cells help phagocytose degenerating neurons and oligodendrocytes that undergo programmed cell death as part of normal development. They are mitotically active and secrete cytokines and growth factors that may help regulate gliogenesis and angiogenesis. After the early stages of brain development, ameboid microglial cells transform into ramified resident microglial cells that persist throughout adulthood.

Ramified microglia constitute 5% to 20% of the glial cells in the adult CNS.⁴⁸ They are more abundant in gray than in white matter, and phylogenetically newer regions of the CNS (cerebral cortex, hippocampus) have more microglia than do older regions (cerebellum, brainstem). Regional variation in the number and shape of microglia suggests that microglial distribution and morphology are regulated by local environments and that microglia play a role in tissue homeostasis. Although the nature of this homeostasis remains to be elucidated, microglia respond quickly and dramatically to all forms of brain pathology.

“Reactive” microglia can be distinguished from resting microglia by two criteria: change in morphology (Fig. 6-11) and upregulation of monocyte-macrophage molecules. Reactive microglia generally have larger cell bodies and shorter, thicker processes than do resting microglia. Reactive microglia also appear to be motile because they can surround or target structures such as dying cells, demyelinated axons, synapses, or as described later, amyloid plaque in Alzheimer’s disease. Reactive microglia express a variety of cell surface molecules that are not usually detectable in resting microglia, including MHC class II and CD68.⁴⁶ In response to frank pathology, resident microglia follow a stereotyped pattern of first becoming “activated” and then phagocytic. Phagocytic CNS macrophages can also originate from circulating monocytes and from neighboring microglia that proliferate and migrate to the site of injury.

FIGURE 6-11 Microglial activation in tissue sections from human brain. Resting microglia in normal brain (A) extend thin processes. Activated microglia in diseased cerebral cortex (B) have thicker processes, and some may have a rod-like shape (arrow). In regions of frank pathology (C), microglia transform into phagocytic macrophages, which can also develop from monocytes that enter the brain from the circulation. Scale bar = 40 µm.

(From Hof P, Trapp BD, de Vellis J, et al. The cellular components of nervous tissue. In: Zigmond M, Bloom FE, Landis SC, et al, eds. Fundamental Neuroscience. San Diego, CA: Academic Press; 1999:41-69.)

In addition to their role as phagocytic macrophages, resident microglia can respond to remote lesions and conditions lacking frank necrotic pathology or activation of the peripheral immune system. After transection of the PNS portion of the facial nerve, for example, resident microglia in the facial nerve nucleus become activated and displace afferent synaptic terminals from the injured facial motoneurons by a process called synaptic stripping.⁴⁹ These cells become phagocytic only when the neurons die. As part of an early CNS immune defense mechanism, microglia play a primary role in CNS disease. Microglia release several potentially cytotoxic substances, including nitric oxide, TNF, proteases, cytokines, and glutamate.⁵⁰ One of the more intriguing responses of microglia occurs in the brains of individuals with Alzheimer’s disease. Alzheimer’s disease is often referred to as an inflammatory disease of the CNS, but the inflammation in the brain does not consist of the classic perivascular accumulation and diffuse parenchymal distribution of blood-borne monocytes and lymphocytes in the cerebral cortex. It involves primary microglial activation. Most prominent among these responses is the migration to and infiltration of amyloid plaque by microglia (Fig. 6-12).^20,21 It is likely that these microglia try, but fail to phagocytose amyloid, especially that associated with the center of the core plaque. As part of this activated microglial response to amyloid, neurites may become injured or dystrophic as a bystander effect of the secretion of microglial cytotoxic substance. Axons may also be transected by a similar mechanism in the lesions of multiple sclerosis (MS).⁴⁰ Amyloid plaque is a good example of how microglia and astrocytes may coordinate reactions to brain pathology (see Fig. 6-12). Astrocytes near the core plaque reorient and send processes into the plaque.²¹ Although the functions of these astrocytic processes are unclear, they express high levels of proteoglycans⁵¹ and amyloid precursor protein, which is the source of amyloid. This is just one example of how microglia and astrocyte activation may affect brain function. Although glial cell reactivity is designed to be protective, it is possible (as proposed in Alzheimer’s disease) that attempts to remove amyloid may destroy axons and dendrites and release cytokines that drive susceptible neurons toward cell death. This and other examples have led to the concepts of good inflammation and bad inflammation. Understanding and regulating good and bad inflammation and other glial reactions is a major challenge to future neurosurgical interventions, including nerve regeneration, transplantation of stem or progenitor cells into the brains of patients with MS, and stroke therapies.

FIGURE 6-12 Glial reaction to amyloid plaque. Microglial cells become activated and enter the amyloid plaque. Astrocytes also send processes into the amyloid plaque. Dystrophic or damaged axons or dendrites, or both, are commonly found around the core of amyloid plaque.

Turnover of Microglia

Because microglia originate from blood-borne monocytes, the possibility has been raised that genetically engineered monocytes or bone marrow transplants may be an effective mechanism of gene therapy for CNS disease. Based on bone marrow chimera experiments in normal adult rodents, ramified microglia have very low (<1%) turnover.⁵² In contrast, 20% to 70% of the monocytes and macrophages in the leptomeninges, choroid plexus, and perivascular spaces are replaced by cells from the bone marrow. Blood-borne cells appear to have a much higher CNS entry rate during brain development or in pathologic environments in the mature CNS. In this regard, bone marrow transplants have significantly increased the life span and quality of life of some children with leukodystrophies.⁵³ Although the cellular mechanisms underlying the improved brain function are unknown, recent studies suggest that bone marrow also contains a stromal stem cell that can differentiate into microglia, astrocytes,⁵⁴ and possibly oligodendrocytes and neurons. In the future, bone marrow transplants may prove to be a useful therapy for other CNS diseases.

Oligodendrocyte Progenitor Cells

Although little is known about OPC function in the adult brain, a significant literature describes the in vitro properties of OPCs isolated from neonatal rodent CNS. Because these cells give rise to oligodendrocytes in serum-free media and to type II astrocytes in serum-containing media, they were initially referred to as O2A cells.⁵⁵ They can be identified in vitro by the expression of A2B5 gangliosides. Characterization of O2A cells in vivo was initially hampered by the ubiquitous expression of A2B5. This changed when O2A cells were demonstrated to express two polypeptide antigens: platelet-derived growth factor receptor α (PDGFαR) and the sulfated proteoglycan NG2. With the use of these markers, substantial evidence now exists that OPCs can give rise to oligodendrocytes in the developing brain.²⁸ The potential for OPCs to give rise to astrocytes in vivo, however, remains to be established.

Oligodendrocyte Progenitor Cells in Development

Because few studies have characterized OPCs during human brain development, the following description is based primarily on studies of rodent brain development. All available evidence suggests that gliogenesis is remarkably similar in human and rodent brains. OPCs originate from discrete regions of the subventricular zone during early brain development. In the spinal cord, OPCs originate from the ventral subventricular zone, dorsal to the floor plate, and their development depends on expression of sonic hedgehog⁵⁶ and neuregulin.⁵⁷ In the cerebrum, OPCs originate from the subventricular zone as PDGFαR-positive cells.²⁸ These cells migrate throughout the brain and, shortly after leaving the subventricular zone, express detectable levels of NG2. These bipolar cells multiply as they migrate and eventually establish a network of stellate cells throughout the entire brain (Fig. 6-13; also see Fig. 6-10). The timing of oligodendrocyte production varies in different brain regions. Signals that regulate oligodendrocyte production are poorly understood, but PDGF-driven proliferation of OPCs appears to be one requirement.⁵⁸ Stellate OPCs are highly mitotic (≈30% by short BrdU labeling) during early brain development,⁵⁹ and these cells produce oligodendrocytes and new OPCs by symmetrical and asymmetric division. The oligodendrocyte initially displays a premyelinating phenotype by extending multiple myelin protein–positive processes that do not immediately ensheathe and myelinate axons (see Fig. 6-13).⁶⁰ These cells have a limited life span and either go on to myelinate or die by programmed cell death.

FIGURE 6-13 Oligodendrocyte progenitor cells originate from the subventricular zone and colonize the brain during early development. The early progenitor transforms into a stellate resident progenitor that produces premyelinating oligodendrocytes or remains as an adult progenitor. Premyelinating oligodendrocytes have a limited life span and either go on to myelinate axons or die by programmed cell death. Adult progenitors may have the potential to produce the oligodendrocytes needed for remyelination.

Ependymal Cells

Ependymal cells line the ventricular spaces of the adult brain. They have a simple ciliated cuboidal morphology.³¹ The cilia project into the ventricular space and spinal canal, oscillate approximately 200 times per minute, and are thought to assist in the rostrocaudal flow of cerebrospinal fluid. Ependymal cells are connected by apicolateral gap and zona occludens junctions. The lack of tight junctions between ependymal cells permits free exchange between the extracellular space of the brain and cerebrospinal fluid. Ependymal cells are mitotically active in the adult brain and thus are susceptible to genetic changes associated with tumorigenesis.