BARR'S The Human Nervous System: An anatomical viewpoint, 9th Edition

PART 1 - Introduction and Neurohistology

Chapter 2

Cells Of The Nervous System

Important Facts

  • Neurons are cells specialized for rapid communication. Most of the cytoplasm of a neuron is in long processes, the neurites (dendrites and axon, which conduct signals toward and away from the cell body, respectively).
  • In the central nervous system (CNS), neuronal cell bodies and dendrites occur in gray matter. White matter consists largely of axons, most of which have myelin sheaths that serve to increase the velocity of conduction.
  • A neuronal surface membrane has a resting potential of -70 mV, maintained by the sodium pump. This is reversed to +40 mV in an axon during the passage of an action potential.
  • The fastest signals, known as impulses or action potentials, are carried in the surface membrane of the axon. There is rapid (saltatory) conduction in myelinated axons because the ion channels in the axolemma are confined to the nodes.
  • The surface membrane of the perikaryon and dendrites does not conduct impulses. Potential changes move more slowly and are graded. An action potential is initiated when the region of the axonal hillock is depolarized to a threshold level.
  • Neurons communicate with one another at synapses. Chemical transmitters released by axonal terminals evoke changes in the membrane of the postsynaptic cell, which may be either stimulated or inhibited. The effect depends on the transmitter and the type of receptor molecule in the postsynaptic membrane.
  • Local reductions of membrane potential (excitatory postsynaptic potentials or depolarizations) add together and may result in initiation of action potentials. Hyperpolarization (inhibitory postsynaptic potentials) reduces the likelihood of initiation of an impulse.
  • Proteins and other substances are transported within axons at different speeds and in both directions.
  • Much of the cytoplasm of a neuron is removed when the axon is transected. The segment that has been isolated from the cell body degenerates together with its myelin sheath, and the fragments are eventually phagocytosed.
  • The neuronal cell body initially reacts to axotomy with increased protein synthesis, accompanied by structural changes known as the axon reaction, or chromatolysis. In the absence of axonal regeneration, the cell body may later shrink or die. Axons severed in the peripheral nervous system can regrow and reinnervate their targets.
  • In mammals, axons transected within the CNS fail to regenerate effectively. Synaptic rearrangements, however, can occur in partly denervated regions of gray matter, and some recovery of function occurs as a result of recruitment of alternative neuronal circuitry.
  • The neuroglial cells of the normal CNS are astrocytes, oligodendrocytes, ependymal cells (derived from neural tube ectoderm), and microglia (derived from mesoderm). Astrocytes occur throughout the brain and spinal cord. Oligodendrocytes produce myelin and are also found next to the cell bodies of some neurons. Microglial cells become phagocytes when local injury or inflammation is present.
  • The neuroglial cells of the peripheral nervous system are Schwann cells in nerves and satellite cells in ganglia.

Two classes of cells are present in the central nervous system (CNS) in addition to the usual cells found in blood vessel walls. Neurons, or nerve cells, are specialized for nerve impulse conduction and for exchanging signals with other neurons. They are, therefore, responsible for most of the functional characteristics of nervous tissue. Neuroglial cells, collectively known as the neuroglia or simply as glia, have important ancillary functions.

The CNS consists of gray matter and white matter. Gray matter contains the cell bodies of neurons, each with a nucleus, embedded


in a neuropil made up predominantly of delicate neuronal and glial processes. White matter, on the other hand, consists mainly of long processes of neurons, the majority being surrounded by myelin sheaths; nerve cell bodies are lacking. Both the gray and the white matter contain large numbers of neuroglial cells and a network of blood capillaries.


Neurons are cells specialized for sending and receiving chemically mediated electrical signals. The part of the cell that includes the nucleus is called the cell body, and its cytoplasm is known as the perikaryonDendrites are typically short branching processes that receive signals from other neurons. Most neurons of the CNS have several dendrites and are, therefore, multipolar in shape. By reaching out in various directions, dendrites increase the ability of a neuron to receive input from diverse sources. Each cell has a single axon. This process, which varies greatly in length from one type of neuron to another, typically conducts impulses away from the cell body. Some neurons have no axons, and their dendrites conduct signals in both directions. Axons of efferent neurons in the spinal cord and brain are included in spinal and cranial nerves. They end on striated muscle fibers or on nerve cells of autonomic ganglia. The term neurite refers to any neuronal process: axon or dendrite.

The fact that each neuron is a structural and functional unit is known as the neuron doctrine, proposed in the latter part of the 19th century in opposition to the then-prevailing view that nerve cells formed a continuous reticulum or syncytium. The unitary concept, conforming to the cell theory, was advanced by His on the basis of embryological studies, by Forel on the basis of the responses of nerve cells to injury, and by Ramón y Cajal from his histological observations. The neuron doctrine was given wide distribution in a review by Waldeyer of the individuality of nerve cells. The lack of cytoplasmic continuity between neurons at synapses was conclusively demonstrated in the 1950s when it became possible to obtain electron micrographs with sufficient resolution to show the structures of intimately apposed cell membranes.


Although all neurons conform to the general principles already outlined, a wide range of structural diversity exists. The size of the cell body varies from 5 µm across for the smallest cells in complex circuits to 135 µm for the largest motor neurons. Dendritic morphology, especially the pattern of branching, varies greatly and is distinctive for neurons that constitute a particular group of cells. The axon of a local circuit neuron may be as short as 100 µm, less than 1 µm in diameter, and devoid of a myelin covering. On the other hand, the axon of a motor neuron that supplies a muscle in the foot is nearly 1 m long, up to 10 µm in diameter, and encased in a myelin sheath up to 5 µm thick. (Much longer axons are present in large animals such as giraffes and whales.)

Neurons occur in ganglia in the peripheral nervous system and in either laminae (layers) or groups called nuclei in the CNS. The large neurons of a nucleus or comparable region are called Golgi type I or principal cells; their axons carry the encoded output of information from the region containing their cell bodies to other parts of the nervous system. The dendrites of a principal cell are contacted by axonal terminals of several other neurons. These neurons include principal cells of other areas and nearby small neurons. The latter are known variously as Golgi type II, internuncial, or local circuit neurons, or, more simply, as interneurons. In many parts of the brain, these neurons greatly outnumber the principal cells.

Examples of large and small neurons are shown in Figure 2-1, which shows the cells as they might appear in specimens stained by the Golgi method.


Structural features of neurons and neuroglial cells are not well shown in sections prepared by general-purpose staining methods such as the alum-hematoxylin-eosin beloved of pathologists. Specialized staining methods are preferred


for light microscopy. Additional information is obtained with the electron microscope and from studies in which functionally significant chemical compounds are histochemically localized in the cells and parts of cells in which they are synthesized or stored.


FIGURE 2-1 Examples of neurons, showing variations in size, shape, and branching of processes.

Cationic dyes, called “Nissl stains” when applied to nervous tissue, bind to DNA and RNA. Therefore, these stains demonstrate the nuclei of all cells and the cytoplasmic Nissl substance (RNA of rough endoplasmic reticulum) of neurons (Fig. 2-2).

Reduced silver methods produce dark deposits of colloidal silver in various structures, notably the proteinaceous filaments inside axons (Fig. 2-3). Other silver methods are available


for demonstration of different types of neuroglial cells.


FIGURE 2-2 Motor neuron in the spinal cord, stained with cresyl violet to show the Nissl bodies and a prominent nucleolus (X800).

Stains for myelin rely on the affinities of certain dyes for hydrophobic proteins and protein-bound phospholipids. They reveal the major tracts of fibers. Some of the photographs in this book (e.g., in Chapter 7) are of sections stained by Weigert's method for myelin.

The Golgi method, which has many variants, is valuable for the study of neuronal morphology, especially of dendrites. Insoluble salts of silver or mercury are precipitated within the cells in blocks of tissue that are then cut into thick sections. Some neurons, including the finest branches of their dendrites, stand out in black against a clear background (Fig. 2-4). Occasional neuroglial cells are similarly displayed,


but axons (especially if myelinated) are typically unstained. An important feature of these methods is the random staining of only a small proportion of the cells, enabling the resolution of structural details of the dendritic trees of individual neurons.


FIGURE 2-3 Cell body of a neuron in the brain, surrounded by axons. In addition, the nucleolus and a small accessory body of Cajal are seen in the nucleus. (Stained by one of Cajal's silver nitrate methods; X 1,000.)


FIGURE 2-4 Pyramidal cell of the cerebral cortex, stained by the Golgi technique. The cell body is in the lower one third of the picture, and dendrites extend up toward the cortical surface. The axon is not visible. (X90; courtesy of Dr. E. G. Bertram.)

Filling techniques provide pictures similar to those obtained by the Golgi method but for individual neurons that have been studied physiologically. A histochemically demonstrable ion or enzyme or a fluorescent dye is injected into the neuron through a micropipette that has been used for intracellular electrical recording. Some other fluorescent dyes move laterally within cell membranes. These can be applied to fresh or even fixed tissue and used to trace neuronal connections over distances up to 5 mm.

Histochemical and immunohistochemical methods are available for localizing substances contained in specific populations of neurons. These substances include putative neurotransmitters and enzymes involved in their synthesis or degradation. Several previously unrecognized systems of neurons have been identified by the use of these methods. With immunohistochemistry, substances in tissues are detected by the binding of specific antibodies. Immunohistochemical methods for cell-specific proteins have largely replaced the traditional silver methods for staining axons and glial cells.

Electron microscopy reveals the detailed internal structure of neurons and the specializations that exist at synaptic junctions. The necessity of using very thin sections makes it difficult to reconstruct in three dimensions. Electron microscopy may be combined with staining by Golgi methods or with immunohistochemical procedures.

Confocal microscopy allows the examination of thin optical sections within thicker specimens prepared for light (usually fluorescence) microscopy. Resolution is enhanced, and images can be superimposed electronically to make pictures that are in focus for the whole depth of the specimen. In confocal images, immunohistochemical localizations can be combined with tracing based on filling or axonal transport.


The parts of a generalized multipolar neuron are shown in Figure 2-5.

Cell Surface

The surface or limiting membrane of the neuron assumes special importance because of its role in the initiation and transmission of signals. The plasma membrane, or plasmalemma, is a double layer of phospholipid molecules whose hydrophobic hydrocarbon chains are all directed toward the middle of the membrane. Embedded in this structure are protein molecules, many of which pass through the whole thickness. Some transmembrane proteins provide hydrophilic channels through which inorganic ions may enter and leave the cell by diffusion. Each of the common ions (Na+, K+, Ca2+, Cl-) has its own specific type of molecular channel, and there are also mixed ion channels that allow passage of multiple


ions such as Na+ and K+ or Na+, K+, and Ca2+. Some channels are voltage gated, which means that they open and close in response to changes in the electrical potential across the membrane. Other channels open in response to ligand, such as neurotransmitters, binding to specific receptors. Nerve impulses are propagated (conducted) along the cell membrane of the neuronal surface. Pumps are protein molecules of the cell membrane that consume energy (from adenosine


triphosphate [ATP]) as they move ions against concentration gradients. A single pump, known as Na/K-ATPase, transports potassium ions into and sodium ions out of the cell, resulting in a net negative charge within the cell and contributing to the membrane potential. Receptors are protein molecules that respond to specific chemical stimuli, typically by causing the opening of associated channels.


FIGURE 2-5 Components of a neuron, traced from an electron micrograph. Mitochondria are colored green and presynaptic terminals from other neurons yellow. (Modified from Heimer L. The Human Brain and Spinal Cord, 2nd ed. New York: Springer-Verlag, 1995.)

The most abundant ions in extracellular fluid are sodium (Na+) and chloride (Cl-). Inside the cell, potassium (K+) is the main positive ion; it is neutralized by organic anions of amino acids and proteins. Both the extracellular fluid and the cytoplasm are electrically neutral, and each has the same total osmotic pressure. A consequence of these conditions is that there is a potential difference across the membrane: The inside is negative (-70 mV) with respect to the outside when the neuron is not conducting a signal. This resting membrane potential opposes the outward diffusion of K+ and the inward diffusion of Cl- because unlike charges attract and like charges repel one another. The membrane is much less permeable to Na+ because the voltage-gated channels for this cation are closed as a consequence of the resting membrane potential. The cytoplasmic anions are too large to pass through the membrane. The ionic concentrations are maintained by the activity of the sodium pump.

The signals carried by a neuron are changes in the potential difference across the plasmalemma. At rest, the cytoplasm is negative (about -70 mV) with respect to the extracellular fluid. This difference is reversed to about +40 mV inside when an axon is sufficiently stimulated. The reversal, known as an impulse or action potential, propagates along the axon. An action potential is an all-or-none phenomenon. In contrast, the dendrites and the cell body respond to stimuli with graded potential changes. Lowering of the membrane potential to a threshold level of -55 mV at the initial segment of the axon triggers an action potential.

Signaling in Neurons

Nucleus and Cytoplasm

The nucleus of a neuron is usually in the center of the cell body. In large neurons, it is vesicular (with finely dispersed chromatin), but in most small neurons, the chromatin is in coarse clumps. Typically, there is a prominent nucleolus. The sex chromatin (see Fig. 2-5), present only in females, was first described in the large nuclei of motor neurons.

The cytoplasm of the cell body (Fig. 2-6) is dominated by the organelles of protein synthesis (rough endoplasmic reticulum and polyribosomes) and cellular respiration (mitochondria). Also present is a well-developed Golgi apparatus, where carbohydrate side chains are added to protein molecules packaged into membrane-bound vesicles destined to enter or pass through the surface membrane of the cell. In light microscopy, the rough endoplasmic reticulum is conspicuous as striated bodies of Nissl substance (see Fig. 2-2).

Filamentous organelles are most prominent in the neurites. Neurofilaments (diameter, 7.5 to 10 nm) are made of structural proteins similar to those of the intermediate filaments of other types of cells. When gathered into bundles, they form the neurofibrils of light microscopy. Microtubules (external diameter, 25 nm) are involved in the rapid transport of protein molecules and small particles in both directions along axons and dendrites. Microfilaments (4 nm) are molecules of the contractile protein actin. They are present on the inside of the plasmalemma and are particularly numerous in the tips of growing neurites.

Neuronal cytoplasm also contains small numbers of membrane-bound vesicles called lysosomes, which contain enzymes that catalyze the breakdown of unwanted large molecules. Neurons may also contain two types of pigment granules. Lipofuscin is a yellow-brown pigment formed from lysosomes that accumulates with aging. Neuromelanin is a black pigment seen only in neurons that use catecholamines (dopamine or noradrenaline) as neurotransmitters.


Dendrites taper from the cell body and branch in its immediate environs. In some neurons, the smaller branches bear large numbers of minute projections, called dendritic spines, which participate in synapses. The surface of the cell body is also included in the receptive field of the neuron.

The single axon has a uniform diameter throughout its length. In interneurons, it is short and branches terminally to establish synaptic contact with adjacent neurons. Some interneurons have no axon, so they can conduct only graded changes of membrane potential. In principal


cells, the diameter of the axon increases in proportion to its length. Collateral branches may be given off at right angles to the axon. The terminal branches are known astelodendria; they typically end as synaptic terminals (also known as boutons terminaux) in contact with other cells. The cytoplasm of the axon is called axoplasm, and the surface membrane is known as the axolemma. The axoplasm includes neurofilaments, microtubules, scattered mitochondria, and fragments of smooth endoplasmic reticulum.


FIGURE 2-6 Electron micrograph of part of the cell body of a neuron in the preoptic area of a rabbit's brain. The series of membranes, together with the free polyribosomes between the membranes, constitute the Nissl material of light microscopy. M, mitochondria; Memb, membranes of endoplasmic reticulum; PM, plasma membrane at surface of cell. (X36,000; courtesy of Dr. R. Clattenburg.)


The axon of a principal cell is usually surrounded by a myelin sheath, which begins near the origin of the axon and ends short of its terminal branching.


Myelin is laid down by neuroglial cells—Schwann cells in the peripheral nervous system and oligodendrocytes in the CNS. The sheath consists of closely apposed layers of glial plasma membranes. Interruptions called nodes of Ranvier indicate junctions between regions formed by different Schwann cells or oligodendrocytes. The ion movements of impulse conduction in a myelinated axon are confined to the nodes. This arrangement provides for saltatory conduction in which the action potential jumps electrically from one node to the next, so that signaling is much faster in a myelinated than in an unmyelinated axon. A nerve fiber consists of the axon and the surrounding myelin sheath or of the axon only in the case of an unmyelinated fiber. The greater the diameter of a nerve fiber, the faster is the conduction of the nerve impulse.


FIGURE 2-7 (A) The myelin sheath and Schwann cell as they are seen (ideally) by light microscopy. (B-D) Successive stages in the development of the myelin sheath from the plasma membrane of a Schwann cell. (E) Ultrastructure of a node of Ranvier, sectioned longitudinally. (F) Relation of a Schwann cell to several unmyelinated axons.

Myelin sheaths are laid down during the later part of fetal development and during the first postnatal year in the manner shown, for a peripheral fiber, in Figure 2-7. The ultrastructure of the sheath is seen in Figure 2-8. A Schwann cell myelinates only one axon, but in the CNS, each process of a single oligodendrocyte contributes to the myelination of a different axon (Fig. 2-9).

Experiments with peripheral nerves of animals show that all Schwann cells have the potential to


make myelin sheaths and that each neuron determines whether the glial cells around its axon will or will not produce a myelin sheath.


FIGURE 2-8 Ultrastructure of the myelin sheath (M) in a peripheral nerve. The dense and less dense layers alternate, and the latter includes a thin intraperiod line. A, axoplasm; E, endoneurium, with collagen fibers. (Electron micrograph, X 107,500; courtesy of Dr. R. C. Buck.)

Saltatory Conduction in Myelinated Axons

Nerve Fibers

A nerve fiber is an axon together with a myelin sheath, if present, and the ensheathing glial cells. The velocity of conduction of an impulse along a nerve fiber increases with the diameter. The largest axons have the thickest myelin sheaths and, therefore, the greatest external diameters. The axonal diameter is approximately two thirds of the total external diameter of the fiber. The thinnest, most slowly conducting axons are unmyelinated.

Peripheral nerve fibers are classified into groups according to external diameter and conduction velocity (Table 2-1). Axons in the CNS are not as easy to classify; their diameters vary greatly.

Conduction Velocity and the Compound Action Potential


A neuron influences other neurons at junctional points, or synapses. The term synapse, meaning a conjunction or connection, was introduced by Sherrington in 1897. An action potential can be propagated in either direction along the surface of an axon. The direction it follows under physiological conditions is determined by a consistent polarity at most synapses, where transmission is from the axon of one neuron to a dendrite or the perikaryon of another neuron. Consequently, action potentials are initiated at the axonal hillock and are propagated away from the cell body.


FIGURE 2-9 An oligodendrocyte with cytoplasmic extensions forming the myelin sheaths of axons in the central nervous system. (Modified from Bunge MB, Bunge RP, Ris H. Ultrastructural study of remyelinization in an experimental lesion in adult cat spinal cord. J Biophys Biochem Cytol 1961;10:67-94.)


TABLE 2-1 Size and Conduction Velocity of Nerve Fibers

Name and Function
of Type of Fiber

Diameter (µm)

Velocity (m/sec)

Myelinated fibers


Aα or IA




Motor to skeletal muscle; sensory from muscle spindle proprioceptive endings (phasic, annulospiral type)


Aβ or IB




Sensory from tendons (tension); also Ruffini endings in skin


Aβ or II




Sensory from Meissner's and pacinian corpuscles and similar endings in skin and connective tissue; from large hair follicles and tonic proprioceptive endings (flower-spray type) of muscle spindles






Motor to intrafusal fibers of muscle spindles


Aδ or III




Sensory from small hair follicles and from free nerve endings for temperature and pain sensations






Preganglionic autonomic (white rami and cranial nerves 3, 7, 9, and 10)


Unmyelinated fibers


C or IV




Pain and temperature; olfaction; postganglionic autonomic


* Letters are used for any nerve; Roman numerals are used for sensory fibers in dorsal spinal roots.


A point of functional contact between two neurons, or between a neuron and an effector cell, is a synapse. The structural details of synapses can be resolved only by electron microscopy. Most synapses in vertebrate animals are chemical synapses. The surface membranes of the two cells are thickened by deposition of proteins (receptors and ion channels) on their cytoplasmic surfaces. The intervening synaptic cleft contains an electron-dense glycoprotein that is absent from the general extracellular space.

The presynaptic neurite, which is most often a branch of an axon, is known as a synaptic terminal or bouton terminal (“terminal button”; the plural is boutons terminax. This French term recalls the appearance in light microscopy.). A synaptic terminal contains numerous mitochondria and a cluster of synaptic vesicles. The latter are membrane-bound organelles 40 to 150 nm in diameter (Fig. 2-10), which contain chemical neurotransmitters. The vesicles may be spherical (in Gray's type 1 synapses, which are generally excitatory) or ellipsoidal (in Gray's type 2 synapses, which use the inhibitory transmitter gamma-aminobutyrate [GABA]). Type 1 synapses are asymmetric, with deposits of fibrillary material that are conspicuously thicker on the postsynaptic than on the presynaptic membrane.

The postsynaptic structure is typically a dendrite. Often, it bears a pendunculated projection, a dendritic spine, that invaginates the presynaptic neurite. Commonly, synapses are grouped together on a dendrite or an axonal terminal to form a larger structure, known as a synaptic complex or glomerulus. In the CNS, the cytoplasmic processes of protoplasmic astrocytes


intimately invest synaptic complexes, restricting diffusion in the intercellular spaces of released transmitters and inorganic ions such as calcium and potassium. These small molecules and ions are absorbed into the cytoplasm of astrocytes and can then diffuse, by way of gap junctions, to adjacent astrocytes.


FIGURE 2-10 Electron micrograph of an axodendritic Gray's type I (asymmetrical) synapse in a rabbit's hypothalamus. D, dendrite; M, mitochondria; Pre, presynaptic membrane; Post, postsynaptic membrane; SV, synaptic vesicles. (X82,000; courtesy of Dr. R. Clattenburg.)

Some different types of chemical synapse are shown in Figure 2-11. The most common arrangements for transferring signals from one neuron to another are axodendritic and axosomatic synapses. Axoaxonal synapses are strategically placed to interfere either with the initiation of impulses at the initial segments of other axons or with the activities of other synaptic terminals. Dendrodendritic synapses can modify a neuron's responses to input at other synapses.

When the membrane potential of a presynaptic neurite is reversed by the arrival of an action potential (or, in the case of a dendrodendritic synapse, adequately reduced by a graded fluctuation), calcium channels are opened and Ca2+ ions diffuse into the cell because they are present at a much higher concentration in the extracellular fluid than in the cytoplasm. Entry of calcium triggers the fusion of synaptic vesicles to the terminal plasmalemma, thereby releasing neurotransmitters and neuromodulators into the synaptic cleft. A classicalneurotransmitter either stimulates or inhibits the postsynaptic cell. A neuromodulator has other actions, including modifying the responsiveness to transmitters.

Having crossed the synaptic cleft, the transmitter molecules combine with receptors on the postsynaptic cell. If the transmitter-receptor interaction is one that results in excitation, nonspecific cation channels are opened, allowing entry of Na+ and Ca2+ and efflux of K+ at postsynaptic


sites. Inhibition, on the other hand, primarily involves the opening of chloride channels in the postsynaptic membrane, which is transiently hyperpolarized as a consequence of the diffusion of Cl- ions into the cytoplasm. Some inhibition results from the opening of K+ channels, which allows K+ to leave the cell, thereby resulting in a net negative charge inside the neuron, similar to the effect of the entry of Cl- ions. These changes in the membrane potential are additive over the whole receptive surface of the postsynaptic neuron. If the net electrical change reaches a threshold level of depolarization to about -55 mV at the axon hillock, an action potential will be initiated and will travel along the axon. Thus, the sum of the postsynaptic responses in the receptive field of a neuron determines whether, at any given moment, an impulse will be sent along the axon.


FIGURE 2-11 Ultrastructure of various types of synapses. The green areas represent the cytoplasmic processes of astrocytes. A, axons; D, dendrites.

Some neurotransmitters act rapidly (within milliseconds) by combining with ionotropic receptors, which are also the ion channels in the membrane. Other substances, notably the peptides, have more protracted actions (within seconds, minutes, or hours). Slowly acting transmitters or modulators combine with metabotropic receptors associated with G proteins. The latter substances bind guanosine triphosphate and participate in intracellular second-messenger systems in the cytoplasm of the postsynaptic cell. The inhibitory transmitter GABA acts on ionotropic receptors associated with chloride channels and on G protein-associated receptors that induce opening of potassium channels. Glutamate, the most abundant excitatory transmitter, also acts on both ionotropic and metabotropic receptors.

The properties of some neurotransmitters and neuromodulators are summarized in Table 2-2. This table does not include the many peptides that serve as transmitters and modulators throughout the nervous system.


Electrical synapses are common in invertebrates and lower vertebrates and have been observed at a few sites in the mammalian nervous system.


TABLE 2-2 Neurotransmitters and Neuromodulators


Occurrence and Functions

Amino acids



Excitatory neurotransmitter in all parts of the CNS



Inhibitory neurotransmitter in all parts of the CNS



Inhibitory neurotransmitter in spinal cord and brain stem

Amines and related compounds



Excitatory transmitter used by motor neurons and by all preganglionic and some postganglionic autonomic neurons (see Chapter 24). In the CNS, acetylcholine is the transmitter or neuromodulator used by neurons in certain nuclei of the reticular formation (see Chapter 9) and in nuclei in the basal forebrain that project to the cerebral cortex.



Used by neurons in the hypothalamus, substantia nigra, and ventral tegmental area (see Chapters 11, 12, and 18). Modulatory actions in the corpus striatum, limbic system, and prefrontal cortex.


Noradrenaline (norepinephrine)

Transmitter used by most neurons of sympathetic ganglia (see Chapter 24); actions vary with the receptors on the innervated cells.


Noradrenaline-producing neurons in the locus coeruleus and other parts of the reticular formation (see Chapter 9) have neuromodulatory effects throughout the brain and spinal cord.



Excitatory transmitter used by neurons in the tuberomamillary nucleus of the hypothalamus. These neurons have long, branched axons that go to most parts of the brain and are believed to be involved in maintaining consciousness.


Serotonin (5-hydroxytryptamine)

Neuromodulator used by neurons in the midline of the brain stem with long branching axons going to all parts of the CNS. Various actions include involvement in sleep (see Chapter 9), mood (see Chapter 18), and pain (see Chapter 19).

CNS, central nervous system; GABA, gamma-aminobutyrate.

Each consists of a close apposition (2 nm) of presynaptic and postsynaptic membranes, across which the cytoplasms of the two cells are joined by numerous tubules or connexons, formed from transmembrane protein molecules of both cells. Water and small ions and molecules move freely through the connexons. An electrical synapse offers a low-resistance pathway between neurons, and there is no delay because a chemical mediator is not involved. Unlike most chemical synapses, electrical synapses are not polarized, and the direction of transmission fluctuates with the membrane potentials of the connected cells. A cluster of connexons that joins cells is known by the general term gap junction.

Axonal Transport

Proteins, including enzymes, membrane lipoproteins, and cytoplasmic structural proteins, are transported distally within axons from their sites of synthesis in the perikaryon. Two major rates of transport have been identified by studying the distribution of proteins labeled by incorporation of radioactive amino acids. Most of the protein moves distally at a rate of about 1 mm/day. This component consists largely of structural proteins, including the subunits of neurofilaments and microtubules. A smaller proportion is transported much more rapidly at


a mean velocity of 300 mm/day. Transport also occurs simultaneously in the reverse direction, from the synaptic terminals to the cell body. The retrogradely transported material includes proteins imbibed from the extracellular fluid by axonal terminals as well as proteins that reach the axon terminals by fast anterograde transport and are returned to the perikaryon. The rate of retrograde transport is variable, but most of the material moves at about two-thirds the speed of the fast component of the anterograde transport.

The rapid components of axonal transport in both directions involve predominantly particle-bound substances and require the integrity of the microtubules of the axoplasm. Particles move along the outsides of the tubules. It is an amazing feat of biological engineering that different substances can move at different rates and in different directions at the same time within a tube as thin as an axon.

Responses of Neurons to Injury

Neurons may be injured physically or by disease processes such as infarction caused by vascular occlusion. Whereas small interneurons are likely to suffer total destruction, injury to large neurons may result either in destruction of the cell body or transection of the axon with preservation of the cell body. When the cell body is destroyed, the axon is isolated from the synthetic machinery of the cell and soon breaks up into fragments, which are eventually phagocytosed. Similar changes occur distally to the site of an axonal injury. The degeneration of an axon that has been detached from the remainder of the cell is called Wallerian degeneration. This process affects not only the axon but also its myelin sheath, even though the latter is not part of the injured neuron.


Changes in the cell body after axonal transection constitute the axon reaction. They vary according to the type of neuron. Cells in some locations degenerate and disappear. This happens to most neurons when the injury occurs before or soon after birth. Conversely, the proximal portions of some adult neurons are not significantly altered by cutting the axon. In such cells, 24 to 48 hours after interruption of the axon, the normally coarse clumps of Nissl substance are changed to a finely granular dispersion, a change known as chromatolysis (Online Fig. 2-3). The nucleus assumes an eccentric position, and the whole cell body swells. These changes reach a maximum 10 to 20 days after axonal transection, and the closer the injury is to the cell body, the more severe the swelling. In a chromatolytic neuron, accelerated synthesis of RNA and proteins takes place that favors regrowth of the axon when conditions make such regeneration possible. Recovery commonly takes several months, and the cell body is eventually smaller than normal if the axon does not regenerate.

The changes described are most easily seen in motor neurons after transection of a peripheral nerve. In cells confined to the CNS, the axon reaction is conspicuous only in some large neurons. Large cells may exhibit no axon reaction when collateral axonal branches that arise close to the cell body are spared.

Transneuronal degeneration is similar in appearance to the axon reaction, but it occurs in neuronal cell bodies that have been deprived of most of their afferents. For example, transection of the optic tract is followed after several weeks by atrophy of some of the neurons in the lateral geniculate body of the thalamus, which is where most of the optic fibers terminate. The postsynaptic neurons have not been directly injured, and their degeneration is attributed to withdrawal of a trophic substance normally supplied by the presynaptic neurons. Neurons in the CNS of immature animals are particularly susceptible to damage caused by deafferentation.


The axon does not last long when separated from the cell body. Phagocytes remove the residual bits of axon and myelin, and they prepare the nerve to receive any axons that might regenerate into its distal stump. These events constitute Wallerian degeneration.



Simultaneously, throughout its length, on the first day, the axon distal to the lesion becomes irregularly swollen. By the 3rd to 5th day, the axon has broken into fragments. Muscle contraction induced by electrical stimulation of a degenerating motor nerve ceases 2 to 3 days after the nerve is interrupted. The myelin sheath is converted into short ellipsoidal segments during the first few days and gradually undergoes complete disintegration. In the meantime, mononuclear leukocytes emigrate through the walls of blood vessels and accumulate in the cylindrical space within the basal lamina of the column of Schwann cells associated with each nerve fiber. The remains of the axon and its myelin sheath (or the axons only in the case of unmyelinated fibers) are phagocytosed. Thus, the distal stump of a degenerated nerve is filled with tubular formations, known as the bands of von Bungner, that contain phagocytes and Schwann cells.


If the axon of a large neuron is transected halfway along its length, the cell loses more than half of its cytoplasm. This lost part of the neuron can be regrown when the injury occurs within the territory of the peripheral nervous system. The reparative process is known as axonal regeneration. It is important to distinguish between this use of the wordregeneration and the replacement of lost cells by mitosis and reorganization of tissue.

In a severed nerve, the regeneration of axons requires surgical apposition of the cut ends. A crushing injury (or freezing a short length of nerve in a laboratory animal) transects the axons but leaves the connective tissue framework of the nerve intact to guide growing axons to their appropriate destinations.


The following description applies to nerves that have been cleanly cut through and repaired. During the first few days, phagocytes and fibroblasts fill the interval between the apposed ends. Regenerating axons and migrating Schwann cells invade this region by about the 4th day, with each axon dividing into many filamentous branches, each with an enlarged tip known as a growth cone. The rate of axonal growth is slow at first; the growth cones may take up to 3 weeks to traverse the site of transection. Many axons grow into nearby connective tissue, but some find their way into the bands of von Bungner in the distal segment. If


too many axons fail to enter the distal stump, a swelling or neuroma is formed and may become a source of spontaneous pain.


FIGURE 2-12 Longitudinal section of axons regenerating from the proximal stump (A) across the scar at a site of transection and repair (C) and into the distal stump (B) of a peripheral nerve. a, b, c, misdirected regenerating axons; d, coils formed by growing axons that fail to enter the scar; e, branching regenerated axon; fg, axons growing into the peripheral stump. (Adapted from Cajal SR. Degeneration and Regeneration of the Nervous System, vol I. London: Oxford University Press, 1928:243.)

The invasion of a particular tube leading to a specific type of end organ appears to be determined only by chance. After crossing the region of the lesion (Fig. 2-12) and entering the bands of von Bungner, the axonal filaments grow along the clefts between columns of Schwann cells and the surrounding basal laminae. Usually, only one branch of each axon enters a single tube; other sprouts are drawn back into the shaft of the growing axon. The rate of growth within the nerve distal to the lesion is 2 to 4 mm/day.

Regenerating axons eventually reach motor and sensory endings; the proportion of correctly reinnervated endings depends on conditions at the site of the original injury. The time that will elapse between nerve suture and the beginning of functional return may be estimated on the basis of an average regeneration rate of 1.5 mm/day. This value takes into account the time required for the fibers to traverse the lesion and for the peripheral nerve endings to be reinnervated.

In a human limb, axonal regeneration can be monitored by testing for Tinel's sign—that is, when part of a nerve containing regenerating axons is tapped with a small hammer, the patient reports a tingling sensation in the area of skin that normally would be supplied by the nerve.

Each regenerating axon becomes surrounded by the cytoplasm of Schwann cells. For axons that are to be myelinated, the Schwann cells lay down myelin sheaths, starting near the lesion and proceeding distally.

Even years after injury and repair, the fiber diameters, internodal lengths, and conduction velocities are seldom more than 80% of the corresponding normal values. Regenerated motor axons supply more muscle fibers than they formerly did. There is less precise control of reinnervated muscles, and sensory function is also inferior to that mediated by uninjured nerves.


When a substantial length of a nerve has been lost, the deficit can be repaired by inserting a graft taken from a thin cutaneous nerve that is functionally less important than the one to be repaired. Several strands of thin nerve are placed side by side, in the manner of a cable, for grafting into a large nerve. Axonal regeneration in a nerve graft is identical to that in a transected and sutured nerve, but the growing axons have to negotiate two sites of anastomosis. The functional recovery is, therefore, far from perfect. A nerve graft must be an autograft (i.e., taken from the same individual) or an isograft (i.e., taken from an identical twin), or it will be rejected by the immune system.


The simplest lesion to visualize is an incised wound of the brain or spinal cord. The space made by the knife blade fills with blood and later with collagenous connective tissue, which is continuous with the pia mater. The astrocytes in the nervous tissue on each side of the collagenous scar generate longer and more numerous cytoplasmic processes, which form a tangled mass. The number of astrocytes does not increase appreciably, but there is a large increase in the total cell population caused mainly by emigration of monocytes from blood vessels to form phagocytic cells known as reactive microglia. The resting microglia that were present before the injury also transform into phagocytes.

The degeneration of severed central axons and their sheaths is different from the process of Wallerian degeneration in peripheral nerves. Degenerating fragments of myelinated axons are present as extracellular objects for several months after the original injury, and the reactive microglial cells that eventually phagocytose the debris persist in situ for many years, marking the positions of the degenerated fibers.

Axons that have been transected in a nerve regrow vigorously, as described earlier. In contrast, when axons are transected within the brain or spinal cord, their proximal stumps begin to regenerate, sending sprouts into the region of the lesion, but this growth ceases after about 2 weeks. This failure of axonal regeneration is attributed partly to inadequate provision of growth factors, which are proteins that promote survival of neurons and axonal growth. Growth factors are produced by various cell types, including neurons and glial cells. Some proteins inhibit axonal growth; the one best understood is present in oligodendrocytes and myelin.


In a few circumstances, axons regenerate successfully within the mammalian brain. For example, the unmyelinated neurosecretory axons of the pituitary stalk (see Chapter 11) can regenerate effectively in adult mammals. Axons of several kinds can regenerate across lesions made in the brain or spinal cord in newborn rodents or the pouch-young of marsupials. Both newly growing and regenerating axons cross the sites of transection and make appropriate synaptic connections with other neurons. These animals are at developmental stages equivalent to early and midfetal development in humans. Nevertheless, many neurons of immature animals die after axotomy. Central axons can regenerate and accurately reconnect with other neurons in adult fishes and amphibians.


Considerable functional recovery commonly occurs after traumatic or pathological damage to the brain, especially when the lesion is not large. For example, destruction of a small area of cerebral cortex that had a well-defined motor or sensory function is followed by paralysis or loss of sensation, with recovery after several weeks. Similar recovery occurs after partial transection of tracts of fibers. Recovery from paralysis caused by occlusion of blood vessels in the cerebral hemispheres (i.e., stroke) is commonly seen in clinical practice, and functional recovery may even occur after incomplete transverse lesions of the spinal cord.

Functional recovery involves the taking over of the functions of damaged neurons by neurons that remain intact. The reorganization of connections within the brain is known asplasticity. This may be an extension of a normally present adaptability used in the learning of often-repeated tasks. Structural changes accompany the functional plasticity that occurs after injury to the nervous system. Thus, when a group of neurons is deprived of part of its afferent input, preterminal axons that come from quite different places grow new branches that then form synapses at the sites denervated by the original lesion. This event, known as axonal sprouting, may occur within a small group of neurons or over greater distances, as when the axons of intact dorsal root ganglion cells extend their axons for three or four segments up and down the spinal cord after transection of neighboring dorsal roots.


Neurons of the adult CNS die soon after removal from the body, presumably as the result of severance of their axons and dendrites. Axons can, however, grow into and out of fragments or isolated cells taken from the embryonic or fetal brain and transplanted into certain parts of the adult brain. In laboratory animals, transplanted fetal neurons can partly compensate for the effects of injuries and experimentally induced diseases. Many attempts have been made to try such grafts in people with Parkinson's disease (see Chapter 12), but no substantial or lasting benefits to the recipients have been found. Transplantation of fetal neurons into the human brain or spinal cord is unlikely to acquire therapeutic significance because (a) even with multiple fetal donors, the number of grafted neurons is unrealistically small in relation to the corresponding parts of the recipient brain; (b) neurons deposited in what would be the normal locations of their cell bodies are unlikely to generate axons that will grow several centimeters in the right direction through the host brain into appropriate populations of postsynaptic neurons; and (c) neurons deposited in regions their axons might normally innervate will not receive afferent synapses appropriate to the normal locations of their cell bodies.

Current research on transplantation into the brain is focused on stem cells, which may be induced to differentiate into neurons or neuroglial cells and on adult-derived glial cells such as Schwann cells and olfactory ensheathing cells, which can encourage axonal growth in the adult brain. There is also much interest in neural progenitor cells that occur in certain places in the brains of adult animals. These cells can be induced to migrate and differentiate into neurons; exploitation of this property may have therapeutic potential.

Neuroglial Cells

The term neuroglia originally referred only to cells in the CNS. It is now applied also to the nonneuronal cells that are intimately related to


neurons and their processes in peripheral ganglia and nerves. The principal structural features of each type are shown in Figure 2-13. The developmental biology of neuroglial cells is reviewed in Chapter 1.



Astrocytes are variable cells with many cytoplasmic processes. Their cytoplasm contains intermediate filaments composed of glial fibrillary acidic protein (GFAP). Many astrocytic processes are closely applied to capillary blood vessels, where they are known as perivascular end feet. Other end feet are applied to the pia mater at the external surface of the CNS and beneath the single layer of ependymocytes that lines the ventricular system, forming, respectively, the external and the internal glial limiting membranes.


FIGURE 2-13 Neuroglial cells of the central nervous system.

Two extreme types of astrocytes are easily recognized by light or electron microscopy. Fibrous astrocytes occur in white matter and have long processes with coarse bundles of GFAP filaments. Protoplasmic (or velate) astrocytes are found in gray matter, and their processes are greatly branching and flattened to form delicate lamellae around the terminal branches of axons, dendrites, and synapses.


Müller cells (in the retina) and pituicytes (in the neurohypophysis; see Chapter 11) are varieties of protoplasmic astrocytes. Olfactory ensheathing cells occur in the olfactory nerves and in the olfactory bulb of the forebrain. They are derived from the olfactory placode and have properties in common with both astrocytes and Schwann cells.

Synapses and nodes of Ranvier are surrounded by the processes of protoplasmic astrocytes, which bear neurotransmitter-specific transporter molecules on their surfaces. Astrocytes can absorb some neurotransmitters, notably glutamate, thus terminating their actions on the postsynaptic membrane. The absorption of potassium ions by astrocytes around synapses, unmyelinated axons, and nodes of Ranvier restrains the spread of electrical disturbances within bundles of axons and regions of neuropil. The dissipation of potassium ions and other small molecules is further enhanced by the existence of gap junctions between adjacent astrocytes.

Corpora amylacea are spherical bodies 25 to 50 µm in diameter and are seen in the normal brains and spinal cords of most middle-aged and elderly people. The name is from a fancied similarity to starch grains. Most corpora amylacea are formed by accumulation of glycoproteins and lipoproteins within processes of astrocytes, although some contain proteins that are normally present in oligodendrocytes or neurons. Corpora amylacea are often extremely abundant, especially in the white matter of the spinal cord, and it is surprising that they do not interfere with function. At sites of degeneration of the cerebral cortex, a locally increased abundance of corpora amylacea sometimes occurs, but these bodies are not thought to be involved in the causation of disease.


The nuclei of oligodendrocytes are small. A rim of cytoplasm surrounds the nucleus, and the cell has a few long, thin processes. The cytoplasm is conspicuous because of its high electron density and because it contains much granular endoplasmic reticulum and many polyribosomes. Filaments and glycogen are absent, but numerous microtubules are present in the processes. Interfascicular oligodendrocytes occur in rows among myelinated axons, where their cytoplasmic processes form and remain continuous with the myelin sheaths (seeFig. 2-9). This function is equivalent to that of the Schwann cell in peripheral nerves. One oligodendrocyte is connected to several myelinated axons. Satellite oligodendrocytes are closely associated with the cell bodies of some large neurons. Astrocytes are also closely associated with neuronal cell bodies. A third type of oligodendrocyte, which does not form myelin, has cytoplasmic processes that contact the nodes of Ranvier in white matter, alongside processes of astrocytes.


The ependyma is the simple cuboidal-to-columnar epithelium that lines the ventricular system. Three cell types are recognized in the ependyma. Ependymocytes constitute the majority. Their cytoplasm contains all the usual organelles as well as filaments similar to those in astrocytes. Most ependymocytes bear cilia and microvilli on their free or apical surfaces. The bases of the cells have cytoplasmic processes that mingle with the astrocytic end feet of the internal glial limiting membrane. Ependymocytes line the ventricular system and are thus in contact with the cerebrospinal fluid (CSF). These cells are not connected by tight junctions, and molecules of all sizes are freely exchanged between the CSF and the adjacent nervous tissue.

Tanycytes differ from ependymocytes in having long basal processes. Most of these cells occur in the floor of the third ventricle. Their basal processes end on the pia mater and on blood vessels in the median eminence of the hypothalamus (see Chapter 11). It has been suggested that the tanycytes of the ventral hypothalamic region respond to changing levels of blood-derived hormones in the CSF by discharging secretory products into the capillary vessels of the median eminence. Such activity may be involved in the control of the endocrine system by the anterior lobe of the pituitary gland (see Chapter 11).

Choroidal epithelial cells cover the surfaces of the choroid plexuses. They have microvilli at their apical surfaces and invaginations at their basal surfaces, which rest on a basement membrane. Adjacent choroidal epithelial cells are joined by tight junctions, thus preventing the passive movement of plasma proteins into the CSF. These cells are metabolically active in controlling the chemical


composition of the fluid, which is secreted by the choroid plexuses into the cerebral ventricles (see Chapter 26).


About 5% of the total neuroglial population in the CNS is composed of resting microglial cells. These have small, elongated nuclei; scanty cytoplasm; and several short-branched processes with spiny appendages. Resting microglial cells are evenly spaced throughout the gray and white matter, with little overlapping or intertwining of their processes.

Resting microglial cells are equivalent to the resident macrophages of other tissues, and they can acquire phagocytic properties when the CNS is afflicted by injury or disease. They may also be involved in protecting the nervous tissue from viruses, microorganisms, and the formation of tumors.

Abnormal Central Neuroglia

When the brain or spinal cord is injured, the astrocytes near the lesion undergo hypertrophy. The cytoplasmic processes become more numerous and are densely packed with GFAP filaments. There may also be a small increase in the number of the cells caused by mitosis of mature astrocytes. These changes, known as gliosis, occur in many pathological conditions, and sometimes the reactive astrocytes acquire phagocytic properties.

Cells with structural and staining properties similar to those of resting microglial cells appear in large numbers at the sites of injury or inflammatory disease in the CNS. Experimental evidence indicates that some of these pathological cells, known as reactive microglial cells, are formed from resting microglial cells, which retract their processes, divide, exhibit amoeboid movement, and acquire phagocytic properties. The activation of resident microglia occurs immediately after almost any kind of insult. At a later stage, large numbers of monocytes enter the nervous system by passing through the walls of blood vessels, and these also assume the appearance of reactive microglial cells and phagocytose the remains of dead cells, bacteria, and other debris. This function is equivalent to that of macrophages in other parts of the body. Reactive microglial cells that are distended with lipid-rich phagocytosed material are known as gitter cells.


Schwann Cells (Neurolemmocytes)

These tubular cells with elongated nuclei intimately ensheath all axons in all parts of the peripheral nervous system, including nerve roots and peripheral nerves. Each axon is suspended in the cytoplasm of its Schwann cell by a double layer of surface membrane, the mesaxon. The myelin sheaths in peripheral nerves are formed by Schwann cells. A myelinated axon is exposed to extracellular fluid at regular intervals along its length, where there are short gaps between adjacent neurolemmocytes. The gaps are called nodes of Ranvier. One Schwann cell ensheaths either one myelinated axon or several unmyelinated axons. The surface of an unmyelinated axon is in contact with extracellular fluid along its whole length, through the cleft between the layers of its mesaxon. (This cleft is closed off by the formation of a myelin sheath.) On the outside surface of each Schwann cell, a basal lamina is present.

Satellite Cells (Ganglionic Gliocytes)

In sensory and autonomic ganglia, these cells intimately surround the neuronal somata. Ganglia also contain Schwann cells around axons.

The enteric nervous system consists of small ganglia and interconnecting strands of mostly unmyelinated neurites in the wall of the gut (see Chapter 24). The neuroglial cells in this system have structural and chemical features in common with both astrocytes and peripheral gliocytes. No special name has been given to the enteric glial cells.

Suggested Reading

Altman J. Microglia emerge from the fog. Trends Neurosci 1994;17:47-49.

Bahr M, Bonhoeffer F. Perspectives on axonal regeneration in the mammalian CNS. Trends Neurosci 1994;17:473-479.

Borlongan CV, Sanberg PR. Neural transplantation for treatment of Parkinson's disease. Drug Discov Today 2002;7:674-682.

Brecknell JE, Fawcett JW. Axonal regeneration. Biol Rev 1996;71, 227-255.

Bruni JE. Ependymal development, proliferation, and functions: a review. Microsc Res Tech 1998;41:2-13.

Bunge RP. Glial cells and the central myelin sheath. Physiol Rev 1968;48:197-251.

Cajal SR. Degeneration and Regeneration of the Nervous System, vol I. London: Oxford University Press, 1928:243.


Del Bigio MR. The ependyma: a protective barrier between brain and cerebrospinal fluid. Glia 1995;14:1-13.

Jones K, ed. Olfactory ensheathing cells: therapeutic potential for spinal cord regeneration [special issue]. Anat Rec Part B 2003;271:39-85.

Kettenman N, Ransom BR, eds. Neuroglia, 2nd ed. New York: Oxford University Press, 2005.

Landau WM. Artificial intelligence: the brain transplant cure for parkinsonism. Neurology 1990;40:733-740.

Leitch B. Ultrastructure of electrical synapses: review. Electron Microsc Rev 1992;5:311-339.

Nicholls JG, Wallace BG, Fuchs PA, et al. From Neuron to Brain, 4th ed. Sunderland, MS: Sinauer, 2001.

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

Ramsay HJ. Ultrastructure of corpora amylacea. J Neuropathol Exp Neurol 1965;24:25-39.

Schipper HM, Cisse S. Mitochondrial constituents of corpora amylacea and autofluorescent astrocytic inclusions in senescent human brain. Glia 1995;14:55-64.

Shepherd GM. Neurobiology, 3rd ed. New York: Oxford University Press, 1994.

Somjen GG. Nervenkitt: notes on the history of the concept of neuroglia. Glia 1988;1:2-9.

Thored P, Arvidsson A, Cacci E, et al. Persistent production of neurons from adult brain stem cells during recovery after a stroke. Stem Cells 2006;24:739-747.

Weiss S, Dunne C, Hewson J, et al. Multipotent CNS stem cells are present in the adult mammalian spinal cord and ventricular neuroaxis. J Neurosci 1996;16:7599-7609.