SOMA

The plasma membrane of the soma is unmyelinated and contacted by inhibitory and excitatory axosomatic synapses: very occasionally, somasomatic and dendrosomatic contacts may be made. The non-synaptic surface is covered by either astrocytic or satellite oligodendrocyte processes.

The cytoplasm of a typical soma (Fig. 3.2) is rich in rough and smooth endoplasmic reticulum and free polyribosomes, indicating a high level of protein synthetic activity. Free polyribosomes often congregate in large groups associated with the rough endoplasmic reticulum. These aggregates of RNA-rich structures are visible by light microscopy as basophilic Nissl bodies or granules. They are more obvious in large, highly active cells, such as spinal motor neurones (Fig. 3.4), which contain large stacks of rough endoplasmic reticulum and polyribosome aggregates. Maintenance and turnover of cytoplasmic and membranous components are necessary in all cells: the huge total volume of cytoplasm within the soma and processes of many neurones requires a considerable commitment of protein synthetic machinery. Neurones also synthesize other proteins (enzyme systems, etc.) involved in the production of neurotransmitters and in the reception and transduction of incoming stimuli. Various transmembrane channel proteins and enzymes are located at the surfaces of neurones where they are associated with movements of ions. The apparatus for protein synthesis (including RNA and ribosomes) occupies the soma and dendrites, but is usually absent from axons.

Fig. 3.4 Spinal motor neurones (toluidine blue stained resin section, rat tissue) showing a group of cell bodies (somata, S), some with the proximal parts of axonal and dendritic processes (P) visible. Their nuclei (N) typically have prominent, deeply staining nucleoli, indicative of metabolically highly active cells; two are visible in the plane of section. Nissl granules (G) are seen in the cytoplasm. Surrounding the neuronal somata is the neuropil, consisting of the interwoven processes of these and other neurones and of glial cells.

The nucleus is characteristically large, round and euchromatic and contains at least one prominent nucleolus; these are features typical of all cells engaged in substantial levels of protein synthesis. The cytoplasm contains many mitochondria and moderate numbers of lysosomes. Golgi complexes are usually close to the nucleus, near the bases of the main dendrites and opposite the axon hillock.

The neuronal cytoskeleton is a prominent feature of its cytoplasm and gives shape, strength and support to the dendrites and axons. A number of neurodegenerative diseases are characterized by abnormal aggregates of cytoskeletal proteins (reviewed in Cairns et al 2004). Neurofilaments (the intermediate filaments of neurones) and microtubules are abundant in the soma and along dendrites and axons: the proportions vary with the type of neurone and cell process. Bundles of neurofilaments constitute neurofibrils which can be seen by light microscopy in silver stained or immunolabelled sections. Neurofilaments are heteropolymers of proteins assembled from three polypeptide subunits, NF-L (68 kilodaltons [kDa]), NF-M (160 kDa) and NF-H (200 kDa). NF-M and NF-H have long C-terminal domains which project as side arms from the assembled neurofilament and bind to neighbouring filaments. They can be heavily phosphorylated, particularly in the highly stable neurofilaments of mature axons, and are thought to give axons their tensile strength. Some axons are almost filled by neurofilaments.

Microtubules are important in axonal transport, although dendrites usually have more microtubules than axons. Centrioles persist in mature postmitotic neurones, where they are concerned with the generation of microtubules rather than cell division. Centrioles are associated with cilia on the surfaces of developing neuroblasts. Their significance, other than at some sensory endings (e.g. the olfactory mucosa, p. 553), is not known.

Pigment granules (Fig. 3.5) appear in certain regions, e.g. neurones of the substantia nigra contain neuromelanin, probably a waste product of catecholamine synthesis. In the locus coeruleus a similar pigment, rich in copper, gives a bluish colour to the neurones. Some neurones are unusually rich in certain metals which may form a component of enzyme systems, e.g. zinc in the hippocampus and iron in the oculomotor nucleus. Ageing neurones, especially in spinal ganglia, accumulate granules of lipofuscin (senility pigment) in residual bodies, which are lysosomes packed with partially degraded lipoprotein material (corpora amylaceae).

Fig. 3.5 Neurones in the substantia nigra of the human midbrain, showing cytoplasmic granules of neuromelanin pigment.

DENDRITES

Dendrites are highly branched, usually short processes which project from the soma (Fig. 3.2). The branching patterns of many dendritic arrays are probably established by random adhesive interactions between dendritic growth cones and afferent axons that occur during development. There is an overproduction of dendrites in early development, and this is pruned in response to functional demand as the nervous system matures and information is processed through the dendritic tree. There is evidence that dendritic trees may be plastic structures throughout adult life, expanding and contracting as the traffic of synaptic activity varies through afferent axodendritic contacts (for review see Wong & Ghosh 2002). Groups of neurones with similar functions have a similar stereotypic tree structure (Fig. 3.6), suggesting that the branching patterns of dendrites are important determinants of the integration of the afferent inputs that converge on the tree.

Fig. 3.6 Purkinje neurone from the cerebellum of a rat stained by the Golgi–Cox method, showing the extensive two-dimensional array of dendrites.

(By courtesy of Martin Sadler and M Berry, Division of Anatomy and Cell Biology, GKT School of Medicine, London.)

Dendrites differ from axons in many respects. They represent the afferent rather than the efferent system of the neurone, and receive both excitatory and inhibitory axodendritic contacts. They may also make dendrodendritic and dendrosomatic connections (see Fig. 3.9), some of which are reciprocal. Synapses occur either on small projections called dendritic spines or on the smooth dendritic surface. Dendrites contain ribosomes, smooth endoplasmic reticulum, microtubules, neurofilaments, actin filaments and Golgi complexes. Their neurofilament proteins are poorly phosphorylated and their microtubules express the microtubule-associated protein (MAP)-2 almost exclusively in comparison with axons.

Fig. 3.9 The structural arrangements of different types of synaptic contact.

The shapes of dendritic spines range from simple protrusions to structures with a slender stalk and expanded distal end. Most spines are not more than 2 μm long, and have one or more terminal expansions; they can also be short and stubby, branched or bulbous. Free ribosomes and polyribosomes are concentrated at the base of the spine. Ribosomal accumulations near synaptic sites provide a mechanism for activity- dependent synaptic plasticity through the local regulation of protein synthesis.

AXONS

The axon originates either from the soma or from the proximal segment of a dendrite, at a specialized region free of Nissl granules, the axon hillock (Fig. 3.2). Action potentials are initiated here. The axonal plasma membrane (axolemma) is undercoated at the hillock by a concentration of cytoskeletal molecules, including spectrin and actin fibrils, which are important in anchoring numerous voltage-sensitive channels to the membrane. The axon hillock is unmyelinated and often participates in inhibitory axo-axonal synapses. It is unique because it contains ribosomal aggregates immediately below the postsynaptic membrane.

In the CNS, small, unmyelinated axons lie free in the neuropil, whereas in the PNS they are embedded in Schwann cell cytoplasm. Myelin, which is formed around almost all axons >2 μm diameter by oligodendrocytes in the CNS and Schwann cells in the PNS, begins at the distal end of the axon hillock. Nodes of Ranvier are specialized constricted regions of myelin-free axolemma where action potentials are generated and where an axon may branch. In both CNS and PNS, the territory of a myelinated axon between adjacent nodes is called an internode: the region close to a node, where the myelin sheath terminates, is called the paranode. Myelin thickness and internodal lengths are, in general, positively correlated with axon diameter. The density of sodium channels in the axolemma is highest at nodes of Ranvier, and very low along internodal membranes: sodium channels are spread more evenly within the axolemma of unmyelinated axons. Fast potassium channels are present in the paranodal regions of myelinated axons. Fine processes of glial cytoplasm (astrocytic in the CNS, Schwann cell in the PNS) surround the nodal axolemma.

The terminals of an axon are unmyelinated and most expand into presynaptic boutons which may form connections with axons, dendrites, neuronal somata or, in the periphery, muscle fibres, glands and lymphoid tissue. Exceptions include the free afferent sensory endings in e.g. the epidermis, which are unspecialized structurally, and the peripheral terminals of afferent sensory fibres with encapsulated endings (see Fig. 3.30). Axon terminals may themselves be contacted by other axons, forming axoaxonal presynaptic inhibitory circuits. Further details of neuronal microcircuitry are given in Kandel et al (2000).

Fig. 3.30 Some major types of sensory ending of general afferent fibres (omitting neuromuscular, neurotendinous and hair-related types). The traces below each type of ending indicate (top) their response (firing rate [vertical lines] and adaption with time) to an appropriate stimulus (below) of the duration indicated. The Pacinian corpuscle response to vibration (rapid sequence of on-off stimuli) is also shown.

Axons contain microtubules, neurofilaments, mitochondria, membrane vesicles, cisternae, and lysosomes. Axons do not usually contain ribosomes or Golgi complexes, except at the axon hillock: exceptionally, the neurosecretory fibres of hypothalamo-hypophysial neurones contain the mRNA of neuropeptides. Organelles are differentially distributed along axons, e.g. there is a greater density of mitochondria and membrane vesicles in the axon hillock, at nodes, and in presynaptic endings. Axonal microtubules are interconnected by cross-linking MAPs, of which tau is the most abundant. Microtubules have an intrinsic polarity, and in axons all microtubules are uniformly orientated with their rapidly growing ends directed away from the soma towards the axon terminal. Neurofilament proteins ranging from high to low molecular weights are highly phosphorylated in mature axons, whereas growing and regenerating axons express a calmodulin-binding membrane-associated phosphoprotein, growth-associated protein-43 (GAP-43), as well as poorly phosphorylated neurofilaments.

Neurones respond differently to injury depending on whether the damage occurs in the CNS or the PNS. The glial microenvironment of a damaged central axon does not facilitate axonal regrowth, consequently reconnection with original synaptic targets does not normally occur. In marked contrast, the glial microenvironment in the PNS is capable of facilitating axonal regrowth. However, functional outcome of clinical repair of a large mixed peripheral nerve, especially if the injury occurs some distance from the target organ, or produces a long defect in the damaged nerve, is frequently unsatisfactory.

SYNAPSES

Transmission of impulses across specialized junctions (synapses) between two neurones is largely chemical and depends on the release of neurotransmitters from the presynaptic side. This causes a change in the electrical state of the postsynaptic neuronal membrane, resulting in either its depolarization or hyperpolarization.

The patterns of axonal termination vary considerably. A single axon may synapse with one neurone (e.g. climbing fibres ending on cerebellar Purkinje neurones), or more often with many neurones (e.g. cerebellar parallel fibres, which provide an extreme example of this phenomenon (p. 302). In synaptic glomeruli, e.g. in the olfactory bulb and cerebellum, and synaptic cartridges, groups of synapses between two or many neurones form interactive units encapsulated by neuroglia (Fig. 3.7).

Fig. 3.7 The arrangement of a complex synaptic unit. A cerebellar synaptic glomerulus with excitatory (‘+’) and inhibitory (‘−’) synapses grouped around a central axonal bouton. The directions of transmission are shown by the arrows.

Electrical synapses (direct communication via gap junctions) are rare in the human CNS and are confined largely to groups of neurones with tightly coupled activity, e.g. the inspiratory centre in the medulla. They will not be discussed further here.

Classification of chemical synapses

Chemical synapses have an asymmetric structural organization (Fig. 3.8 and Fig. 3.9) in keeping with the unidirectional nature of their transmission. Typical chemical synapses share a number of important features. They all display an area of presynaptic membrane apposed to a corresponding postsynaptic membrane: the two are separated by a narrow (20–30 nm) gap, the synaptic cleft. Synaptic vesicles containing neurotransmitter lie on the presynaptic side, clustered near a dense plaque on the cytoplasmic aspect of the presynaptic membrane. A corresponding region of submembrane density is present on the postsynaptic side. Together these define the active zone, the area of the synapse where neurotransmission takes place.

Fig. 3.8 Electron micrographs demonstrating various types of synapse. A, Cross-section of a dendrite (D) upon which two synaptic boutons (B) end. The upper bouton contains round vesicles, and the lower bouton contains flattened vesicles of the small type. A number of pre- and postsynaptic (P) thickenings mark the specialized zones of contact. B, A type I synapse (S, postsynaptic site) containing both small, round, clear vesicles and also large dense-cored vesicles of the neurosecretory type. C, A large terminal bouton (B) of an optic nerve afferent fibre, which is making contact with a number of postsynaptic processes, in the dorsal lateral geniculate nucleus of the rat. One of the postsynaptic processes (^*) also receives a synaptic contact from a bouton (Bf) containing flattened vesicles. D, Reciprocal synapses (S) between two neuronal processes in the olfactory bulb.

(By courtesy of Professor AR Lieberman, Department of Anatomy, University College, London.)

Chemical synapses can be classified according to a number of different parameters including the neuronal regions forming the synapse, their ultrastructural characteristics, the chemical nature of their neurotransmitter(s) and their effects on the electrical state of the postsynaptic neurone. The following classification is limited to associations between neurones. Neuromuscular junctions share many (though not all) of these parameters, and are often referred to as peripheral synapses. They are described separately on page 63.

Synapses can occur between almost any surface regions of the participating neurones. The most common type occurs between an axon and either a dendrite or a soma, when the axon is expanded as a small bulb or bouton (Fig. 3.8, Fig. 3.9). This may be a terminal of an axonal branch (terminal bouton) or one of a row of bead-like endings, when the axon makes contact at several points, often with more than one neurone (bouton de passage). Boutons may synapse with dendrites, including dendritic spines or the flat surface of a dendritic shaft, a soma (usually on its flat surface, but occasionally on spines), the axon hillock and the terminal boutons of other axons.

The connection is classified according to the direction of transmission, and the incoming terminal region is named first. Most common are axodendritic synapses, although axosomatic connections are frequent. All other possible combinations are found but are less common, i.e. axoaxonic, dendroaxonic, dendrodendritic, somatodendritic or somatosomatic. Axodendritic and axosomatic synapses occur in all regions of the CNS and in autonomic ganglia, including those of the ENS. The other types appear restricted to regions of complex interaction between larger sensory neurones and microneurones, e.g. in the thalamus.

Ultrastructurally, synaptic vesicles may be internally clear or dense, and of different size (loosely categorized as small or large) and shape (round, flat or pleomorphic, i.e. irregularly shaped). The submembranous densities may be thicker on the postsynaptic than on the presynaptic side (asymmetric synapses), or equivalent in thickness (symmetrical synapses). Synaptic ribbons are found at sites of neurotransmission in the retina and inner ear. They have a distinctive morphology, in that the synaptic vesicles are grouped around a ribbon- or rod-like density orientated perpendicular to the cell membrane (Fig. 3.9).

Synaptic boutons make obvious close contacts with postsynaptic structures, but many other terminals lack specialized contact zones. Areas of transmitter release occur in the varicosities of unmyelinated axons, where effects are sometimes diffuse, e.g. the aminergic pathways of the basal ganglia and in autonomic fibres in the periphery. In some instances, such axons may ramify widely throughout extensive areas of the brain and affect the behaviour of very large populations of neurones, e.g. the diffuse cholinergic innervation of the cerebral cortices. Pathological degeneration of these pathways can therefore cause widespread disturbances in neural function.

Neurones express a variety of neurotransmitters, either as one class of neurotransmitter per cell or more often as several. Good correlations exist between some types of transmitter and specialized structural features of synapses. In general, asymmetric synapses with relatively small spherical vesicles are associated with acetylcholine (ACh), glutamate, serotonin (5-hydroxytryptamine, 5-HT), and some amines; those with dense-core vesicles include many peptidergic synapses and other amines (e.g. noradrenaline (norepinephrine), adrenaline (epinephrine), dopamine). Symmetrical synapses with flattened or pleomorphic vesicles have been shown to contain either γ-aminobutyric acid (GABA) or glycine.

Neurosecretory endings found in various parts of the brain and in neuroendocrine glands and cells of the dispersed neuroendocrine system (see below; p. 31) share many features with presynaptic boutons. They all contain peptides or glycoproteins within dense-core vesicles. The latter are of characteristic size and appearance: they are often ellipsoidal or irregular in shape, and relatively large, e.g. oxytocin and vasopressin vesicles in the neurohypophysis may be up to 200 nm across.

Synapses may cause depolarization or hyperpolarization of the postsynaptic membrane, depending on the neurotransmitter released and the classes of receptor molecule in the postsynaptic membrane. Depolarization of the postsynaptic membrane results in excitation of the postsynaptic neurone, whereas hyperpolarization has the effect of transiently inhibiting electrical activity. Subtle variations in these responses may also occur at synapses where mixtures of neuromediators are present and their effects are integrated.

Type I and II synapses

There are two broad categories of synapse: type I synapses, in which the cytoplasmic density is thicker on the postsynaptic side, and type II synapses, in which the pre- and post-synaptic densities are more symmetrical but thinner. Type I boutons contain a predominance of small spherical vesicles approximately 50 nm in diameter, and type II boutons contain a variety of flat forms. Throughout the CNS, type I synapses are generally excitatory and type II, inhibitory. In a few instances types I and II synapses are found in close proximity, orientated in opposite directions across the synaptic cleft (a reciprocal synapse).

Mechanisms of synaptic activity

Synaptic activation begins with arrival of one or more action potentials at the presynaptic bouton, which causes the opening of voltage-sensitive calcium channels in the presynaptic membrane. The response time in typical fast-acting synapses is then very rapid; classic neurotransmitter (e.g. ACh) is released in less than a millisecond, which is faster than the activation time of a classic second messenger system on the presynaptic side. The influx of calcium activates Ca²⁺-dependent protein kinases. This uncouples synaptic vesicles from a spectrin-actin meshwork within the presynaptic ending, to which they are bound via synapsins I and II. The vesicles dock with the presynaptic membrane, through processes not yet fully understood, and their membranes fuse to open a pore through which neurotransmitter diffuses into the synaptic cleft.

Once the vesicle has discharged its contents, its membrane is incorporated into the presynaptic plasma membrane and is then more slowly recycled back into the bouton by endocytosis around the edges of the active site. The time between endocytosis and re-release may be approximately 30 seconds; newly recycled vesicles compete randomly with previously stored vesicles for the next cycle of neurotransmitter release. The fusion of vesicles with the presynaptic membrane is responsible for the observed quantal behaviour of neurotransmitter release, both during neural activation and spontaneously, in the slightly leaky resting condition.

Postsynaptic events vary greatly, depending on the receptor molecules and their related molecular complexes. Receptors are generally classed as either ionotropic or metabotropic. Ionotropic receptors function as ion channels, so that conformational changes induced in the receptor protein when it binds the neurotransmitter cause the opening of an ion channel within the same protein assembly, thus causing a voltage change within the postsynaptic cell. Examples are the nicotinic ACh receptor and the N-methyl-D-aspartate (NMDA) glutamate receptor. Alternatively, the receptor and ion channel may be separate molecules, coupled by G-proteins, some via a complex cascade of chemical interactions (a second messenger system), e.g. the adenylate cyclase pathway. Postsynaptic effects are generally rapid and short-lived, because the transmitter is quickly inactivated either by an extracellular enzyme (e.g. acetylcholinesterase, AChE), or by uptake by neuroglial cells. Examples of such metabotropic receptors are the muscarinic Ach receptor and 5-HT receptor.

Neurohormones

Neurohormones are included in the class of molecules with neurotransmitter-like activity. They are synthesized in neurones and released into the blood circulation by exocytosis at synaptic terminal-like structures. As with classic endocrine gland hormones, they may act at great distances from their site of secretion. Neurones secrete into the CSF or local interstitial fluid to affect other cells, either diffusely or at a distance. To encompass this wide range of phenomena the general term neuromediation has been used, and the chemicals involved are called neuromediators.

Neuromodulators

Some neuromediators do not appear to affect the postsynaptic membrane directly, but they can affect its responses to other neuromediators, either enhancing their activity (by increasing or prolonging the immediate response), or perhaps limiting or inhibiting their action. These substances are called neuromodulators. A single synaptic terminal may contain one or more neuromodulators in addition to a neurotransmitter, usually (though not always) in separate vesicles. Neuropeptides (see below) are nearly all neuromodulators, at least in some of their actions. They are stored within dense granular synaptic vesicles of various sizes and appearances.

ASTROCYTES

Astrocytes are star-shaped glia (Fig. 3.10). Their processes, which ramify through the entire central neuropil (Fig. 3.11), are coupled functionally at gap junctions, forming an interconnected network which ensheathes all neurones, other than at synapses and along the myelinated segments of axons. Some astrocyte processes terminate as end-feet on the basal lamina of blood vessels, and at the pial surface, where they form the glia limitans (glial limiting membrane).

Fig. 3.10 Confocal micrograph of an astrocyte in an adult rat optic nerve, iontophoretically filled with an immunofluorescent dye by intracellular microinjection. (Prepared by Professor A Butt, Portsmouth, and Kate Colquhoun, formerly of Division of Physiology, GKT School of Medicine, London.)

Fig. 3.11 The different types of non-neuronal cell in the CNS and their structural organization and interrelationships with each other and with neurones.

Astrocytes have been divided into two subtypes, fibrous and protoplasmic, on morphological grounds. Fibrous astrocytes occur predominantly in white matter and protoplasmic astrocytes are found mainly in grey matter. The significance of these subtypes is unclear since there are few known functional differences between fibrous and protoplasmic astrocytes. Astrocytes typically have a pale nucleus with a narrow rim of heterochromatin, although this is variable. Their cytoplasm is pale and contains glycogen, lysosomes, Golgi complexes and bundles of glial intermediate filaments that extend into their processes (these last are found particularly in fibrous astrocytes). Glial intermediate filaments are formed from glial fibrillary acidic protein (GFAP): its presence can be used clinically to identify tumour cells of glial origin.

Astrocytes are now thought to provide a network of communication, providing integration and regulation of function in the brain (reviewed in Fields 2004, Volterra & Meldolesi 2005). One form of communication, with other astrocytes, is via interconnecting low resistance gap junctional complexes. There is recent evidence that astrocytic (and other neural) gap junctions may be formed by a novel family of proteins, the pannexins (see p. 7). They signal to each other using intracellular calcium wave propagation, triggered by synaptically-released glutamate. Signalling presumably co-ordinates astrocyte functions that are essential for efficient neuronal activity, such as ion (particularly potassium) buffering; neurotransmitter uptake and metabolism (e.g. of excess glutamate, which is excitotoxic); membrane transport; the secretion of peptides, amino acids, trophic factors etc. The complexity of astrocytic function and dysfunction in neurological disorders is reviewed in Seifert et al (2006).

Injury to the CNS induces astrogliosis, seen as a local increase in the number and size of cells expressing GFAP, and in the production of an extensive meshwork of processes to form a glial scar. It is thought that the microenvironment of a glial scar, which may also include cells of oligodendrocyte lineage and myelin debris, plays an important role in inhibiting regrowth of damaged CNS axons.

Pituicytes are glial cells found in the neural parts of the pituitary gland, the infundibulum and neurohypophysis. They resemble astrocytes, but their processes end mostly on endothelial cells in the neurohypophysis and tuber cinereum.

Blood–brain barrier

Proteins circulating in the blood enter most tissues of the body except those of the brain, spinal cord or peripheral nerves. This concept of a blood–brain or a blood–nerve barrier applies to many substances – some are actively transported across the blood–brain barrier, others are actively excluded. The blood–brain barrier is located at the capillary endothelium within the brain and is dependent upon the presence of tight junctions between endothelial cells coupled with a relative lack of transcytotic vesicular transport. The tightness of the barrier depends upon the close apposition of astrocytes, which direct the formation of endothelial tight junctions, to blood capillaries (reviewed in Abbott et al 2006) (Fig. 3.12, Fig. 3.13).

Fig. 3.12 Astrocytes (A) in rat brain, immunolabelled to show glial fibrillary acidic protein (brown). Fine processes form end-feet (E) on brain capillaries (C). Note that astrocytes have extremely dense, numerous processes: immunostaining only reveals a proportion of the processes. (Prepared by Mr Marios Hadjipavlou, King’s College, London.)

Fig. 3.13 The relationship between the glia limitans, perivascular cells and blood vessels within the brain, in longitudinal and transverse section. A sheath of astrocytic end-feet wraps around the vessel and, in vessels larger than capillaries, its investment of pial meninges. Vascular endothelial cells are joined by tight junctions and supported by pericytes; perivascular macrophages lie outside the endothelial basal lamina.

The blood–brain barrier develops during embryonic life but may not be fully completed by birth. There are certain areas of the adult brain where the endothelial cells are not linked by tight junctions, which means that a free exchange of molecules occurs between blood and adjacent brain. Most of these areas are situated close to the ventricles and are known as circumventricular organs (p. 240). Elsewhere, unrestricted diffusion through the blood–brain barrier is only possible for substances that can cross biological membranes because of their lipophilic character. Lipophilic molecules may be actively re-exported by the brain endothelium.

Breakdown of the blood–brain barrier occurs when the brain is damaged by ischaemia or infection, and is also associated with primary and metastatic cerebral tumours. Reduced blood flow to a region of the brain alters the permeability and regulatory transport functions of the barrier locally: the increased stress on compromised endothelial cells results in leakage of fluid, ions, serum proteins and intracellular substances into the extracellular space of the brain. The integrity of the barrier can be evaluated clinically using CT and fMRI. Breakdown of the blood–brain barrier may be seen at postmortem in jaundiced patients who have had an infarction. Normally, the brain, spinal cord and peripheral nerves remain unstained by the bile post mortem, although the choroid plexus is often stained a deep yellow. However, areas of recent infarction (1–3 days) will also be stained by bile pigment because of the localized breakdown of the blood–brain barrier.

OLIGODENDROCYTES

Oligodendrocytes myelinate CNS axons and are most commonly seen as intrafascicular cells in myelinated tracts (Fig. 3.14, Fig. 3.15). They usually have round nuclei and their cytoplasm contains numerous mitochondria, microtubules and glycogen. They display a spectrum of morphological variation, from cells with large euchromatic nuclei and pale cytoplasm, to cells with heterochromatic nuclei and dense cytoplasm. Oligodendrocytes may enclose up to 50 axons in separate myelin sheaths, and the largest calibre axons are usually ensheathed on a 1 : 1 basis. Some oligodendrocytes are not associated with axons, and are either precursor cells or perineuronal (satellite) oligodendrocytes whose processes ramify around neuronal somata.

Fig. 3.14 The ensheathment of a number of axons by the processes of an oligodendrocyte. The oligodendrocyte soma is shown in the centre and its myelin sheaths are unfolded to varying degrees to show their extensive surface area.

(Modified from Morell & Norton 1980 by Raine 1984, by permission.)

Fig. 3.15 A, An oligodendrocyte enwrapping several axons with myelin, demonstrated in a whole-mounted rat anterior medullary velum, immunolabelled with antibody to an oligodendrocyte membrane antigen. B, Confocal micrograph of a mature myelin forming oligodendrocyte in an adult rat optic nerve, iontophoretically filled with an immunofluorescent dye by intracellular microinjection.

(A, By courtesy of Fiona Ruge; B, prepared by Professor A Butt, Portsmouth, and Kate Colquhoun, formerly Division of Physiology, GKT School of Medicine, London.)

Within tracts, interfascicular oligodendrocytes are arranged in long rows interspersed with single astrocytes at regular intervals. Since oligodendrocyte processes are radially aligned to the axis of each row, myelinated tracts typically consist of cables of axons myelinated by a row of oligodendrocytes running down the axis of each cable.

Oligodendrocytes originate from the ventricular neurectoderm and the subependymal layer in the fetus (p. 368), and continue to be generated from the subependymal plate postnatally. Stem cells migrate and seed into white and grey matter to form a pool of adult progenitor cells which may later differentiate to replenish lost oligodendrocytes, and possibly remyelinate pathologically demyelinated regions.

Nodes of Ranvier and incisures of Schmidt–Lanterman

The territory ensheathed by an oligodendrocyte (or Schwann cell) process defines an internode, the interval between internodes is called a node of Ranvier (Fig. 3.16) and the territory immediately adjacent to the nodal gap is a paranode, where loops of oligodendrocyte cytoplasm abut the axolemma. Nodal axolemma is contacted by fine filopodia of perinodal cells which have been shown in animal studies to have a presumptive adult oligodendrocyte progenitor phenotype: their function is unknown (reviewed in Butt et al 2005). Schmidt–Lanterman incisures are helical decompactions of internodal myelin where the major dense line of the myelin sheath splits to enclose a spiral of oligodendrocyte cytoplasm. Their structure suggests that they may play a role in the transport of molecules across the myelin sheath, but their function is not known.

Fig. 3.16 A node of Ranvier (N) in the central nervous system. The pale-staining axon (A) is ensheathed by oligodendrocyte myelin (arrow), apart from a short, exposed region at the node. Toluidine blue stained resin section (rat tissue).

(By courtesy of Dr Clare Farmer, King’s College, London.)

MYELIN AND MYELINATION

Myelin is formed by oligodendrocytes (CNS) and Schwann cells (PNS). A single oligodendrocyte may ensheathe up to 50 separate axons, depending upon calibre, whereas myelinating Schwann cells ensheathe axons on a 1 : 1 basis.

In general, myelin is laid down around axons above 2 μm diameter. However, the critical minimal axon diameter for myelination is smaller and more variable in the CNS than in the PNS (approximately 0.2 μm in the CNS compared with 1–2 μm in the PNS). There is considerable overlap between the size of the smallest myelinated and the largest unmyelinated axons, and so axonal calibre is unlikely to be the only factor in determining myelination. Moreover, the first axons to become ensheathed ultimately attain larger diameters than those that are ensheathed at a later date. There is a reasonable linear relationship between axon diameter and internodal length and myelin sheath thickness: as the sheath thickens from a few lamellae to up to 200, the axon may also grow from 1 to 15 μm in diameter. Internodal lengths increase about tenfold during the same time.

It is not known precisely how myelin is formed in either PNS or CNS, but in the CNS it has been shown recently to depend in part on expression of a protein (Wiskott–Aldrich syndrome protein family verprolin homologous; WAVE), which influences the actin cytoskeleton, oligodendrocyte lamellipodia formation and myelination (Kim et al 2006). The ultrastructural appearance of myelin is usually explained in terms of the spiral wrapping of an extensive, flat glial process (lamellipodium) around an axon, and the subsequent extrusion of cytoplasm from the sheath at all points other than incisures and paranodes. In this way, the compacted external surfaces of the plasma membrane of the ensheathing glial cell are thought to produce the minor dense lines, and the compacted inner cytoplasmic surfaces, the major dense lines, of the mature myelin sheath (Fig. 3.17). These lines, first described in early electron microscope studies of the myelin sheath, correspond to the intraperiod and period lines respectively defined in X-ray studies of myelin. The inner and outer zones of occlusion of the spiral process are continuous with the minor dense line and are called the inner and outer mesaxons.

Fig. 3.17 Stages in myelination of a peripheral axon.

There are significant differences between central and peripheral myelin, reflecting the fact that oligodendrocytes and Schwann cells express different proteins during myelinogenesis. The basic dimensions of the myelin membrane are different. CNS myelin has a period repeat thickness of 15.7 nm whereas PNS myelin has a period to period line thickness of 18.5 nm, and the major dense line space is approximately 1.7 nm in CNS myelin, compared with 2.5 nm in PNS myelin. Myelin membrane contains protein, lipid and water, which forms at least 20% of the wet weight. It is a relatively lipid-rich membrane and contains 70–80% lipid. All classes of lipid have been found; perhaps not surprisingly, the precise lipid composition of PNS and CNS myelin is different. The major lipid species are cholesterol (the commonest single molecule), phospholipids and glycosphingolipids. Minor lipid species include galactosylglycerides, phosphoinositides and gangliosides. The major glycolipids are galactocerebroside and its sulphate ester, sulphatide: these lipids are not unique to myelin, but they are present in characteristically high concentrations. CNS and PNS myelin also contain low concentrations of acidic glycolipids, which constitute important antigens in some inflammatory demyelinating states. Gangliosides, which are glycosphingolipids characterized by the presence of sialic acid (N-acetylneuraminic acid), account for less than 1% of the lipid in myelin.

A relatively small number of protein species account for the majority of myelin protein. Some of these proteins are common to both PNS and CNS myelin, but others are different. Proteolipid protein (PLP) and its splice variant DM20 are found only in CNS myelin, whereas myelin basic protein (MBP) and myelin associated glycoprotein (MAG) occur in both. MAG is a member of the immunoglobulin supergene family, and is localized specifically at those regions of the myelin segment where compaction starts, namely, the mesaxons and inner periaxonal membranes, paranodal loops and incisures, in both CNS and PNS sheaths. It is thought to have a functional role in membrane adhesion.

In the developing CNS, axonal outgrowth precedes the migration of oligodendrocyte precursors, and oligodendrocytes associate with and myelinate axons after their phase of elongation: oligodendrocyte myelin gene expression is not dependent on axon-association. In marked contrast, Schwann cells in the developing PNS are associated with axons during the entire phase of axonal growth. Myelination does not occur simultaneously in all parts of the body in late fetal and early postnatal development. White matter tracts and nerves in the periphery have their own specific temporal patterns which relate to their degree of functional maturity.

Mutations of the major myelin structural proteins have now been recognized in a number of inherited human neurological diseases. As would be expected, these mutations produce defects in myelination, and in the stability of nodal and paranodal architecture, which are consistent with the suggested functional roles of the relevant proteins in maintaining the integrity of the myelin sheath. The molecular organization of myelinated axons is described in Scherer & Arroyo (2002).

EPENDYMA

Ependymal cells line the ventricles (Fig. 3.18; see Fig. 3.11) and central canal of the spinal cord. They form a single-layered epithelium which varies from squamous to columnar in form. At the ventricular surface, cells are joined by gap junctions and occasional desmosomes. Their apical surfaces have numerous microvilli and cilia which contribute to the flow of CSF. There is considerable regional variation in the ependymal lining of the ventricles, but four major types have been described. These are: general ependyma which overlies grey matter; general ependyma which overlies white matter; specialized areas of ependyma in the third and fourth ventricles; choroidal epithelium.

Fig. 3.18 Ciliated columnar epithelial lining of the lateral ventricle (V), overlying the subventricular zone (SVZ). C, cilia; E, ependymal cells. Mouse tissue, toluidine blue stained resin section.

The ependymal cells overlying areas of grey matter are cuboidal; each cell bears approximately 20 central apical cilia, surrounded by short microvilli. The cells are joined by gap junctions and desmosomes and do not have a basal lamina. Beneath them there may be a subependymal (or subventricular) zone, from two to three cells deep, consisting of cells which generally resemble ependymal cells. The capillaries beneath them have no fenestrations and few transcytotic vesicles, which is typical of the CNS. Where the ependyma overlies myelinated tracts of white matter, the cells are much flatter and few are ciliated. There are gap junctions and desmosomes between these cells, but their lateral margins interdigitate, unlike their counterparts overlying grey matter. No subependymal zone is present.

Specialized areas of ependymal cells called the circumventricular organs are found in four areas around the margins of the third ventricle, namely the lining of the median eminence of the hypothalamus; the subcommissural organ; the subfornical organ and the vascular organ of the lamina terminalis. The area postrema, at the inferoposterior limit of the fourth ventricle, has a similar structure. In all of these sites the ependymal cells are only rarely ciliated and their ventricular surfaces bear many microvilli and apical blebs. They have numerous mitochondria, well-formed Golgi complexes and a rather flattened basal nucleus. They are joined laterally by tight junctions which form a barrier to the passage of materials across the ependyma, and by desmosomes. Many of the cells are tanycytes (ependymal astrocytes) and have basal processes which project into the perivascular space surrounding the underlying capillaries. Significantly these capillaries are fenestrated and therefore do not form a blood–brain barrier. It is believed that neuropeptides can pass from nervous tissue into the CSF by active transport through the ependymal cells in these specialized areas, and in this way access a wide population of neurones via the permeable ependymal lining of the rest of the ventricle.

The ependyma is highly modified where it lies adjacent to the vascular layer of the choroid plexuses.

Choroid plexus

The choroid plexus forms the CSF and actively regulates the concentration of molecules in the CSF. It consists of highly vascularized masses of pia mater enclosed by pockets of ependymal cells. The ependymal cells resemble those of the circumventricular organs, except that they do not have basal processes, but form a cuboidal epithelium which rests on a basal lamina adjacent to the enclosed fold of meningeal pia mater and its capillaries (Fig. 3.19, Fig. 3.20). The cells have numerous long microvilli with only a few cilia interspersed between them. They also have many mitochondria, large Golgi complexes and basal nuclei, features that are consistent with their secretory activity: they produce most components of the CSF. They are linked by tight junctions that form a transepithelial barrier (a component of the blood–CSF barrier), and by desmosomes. Their lateral margins are highly folded.

Fig. 3.19 Choroid plexus within the lateral ventricle. Frond-like projections of vascular stroma derived from the pial meninges are covered with a low columnar epithelium which secretes cerebrospinal fluid. Mouse tissue, toluidine blue stained resin section.

Fig. 3.20 The arrangement of tissues forming the choroid plexus.

The choroid plexus has a villous structure where the stroma is composed of pial meningeal cells, and contains fine bundles of collagen and blood vessels. Choroidal capillaries are lined by a fenestrated endothelium. During fetal life, erythropoiesis occurs in the stroma, which is occupied by bone marrow-like cells. In adult life, the stroma contains phagocytic cells, which, together with the cells of the choroid plexus epithelium, phagocytose particles and proteins from the ventricular lumen.

Age-related changes occur in the choroid plexus which can be detected by neuroimaging. Calcification of the choroid plexus can be detected by X-ray or CT scan in 0.5% of individuals in the first decade of life and in 86% in the eighth decade. The incidence of calcification rises sharply, from 35% of CT scans in the fifth decade to 75% in the sixth decade. Visible calcification is usually restricted to the glomus region of the choroid plexus, i.e. the vascular bulge in the choroid plexus as it curves to follow the anterior wall of the lateral ventricle into the temporal horn.

MICROGLIA

Microglia are small dendritic cells found throughout the CNS (Fig. 3.21) including the retina (p. 692). They are thought to be derived from fetal monocytes or their precursors; the haematogenous cells cross the walls of neural blood vessels and invade CNS tissue prenatally as amoeboid cells. Later they lose their motility and transform into typical microglia which bear branched processes that ramify in non-overlapping territories within the brain. All microglial domains, defined by their dendritic fields, are equivalent in size, and form a regular mosaic throughout the brain. The expression of microglia-specific antigens changes with age and many are downregulated when microglia attain the mature dendritic form.

Microglia have elongated nuclei with peripheral heterochromatin. Their scant cytoplasm is pale staining and contains granules, scattered cisternae of rough endoplasmic reticulum and Golgi complexes at both poles. Two or three primary processes stem from the opposite poles of each cell body and then branch repeatedly to form short terminal processes. The function of microglia in the normal brain is obscure. Like astrocytes, microglia are activated by both traumatic and ischaemic injury. In many diseases including Parkinson’s disease, Alzheimer’s disease, multiple sclerosis, acquired immunodeficiency syndrome (AIDS), amyotrophic lateral sclerosis (motor neurone disease) and paraneoplastic encephalitis, microglia become phagocytic and are actively involved in synaptic stripping and clearance of neuronal debris. Some transform into amoeboid, motile cells.

ENTRY OF INFLAMMATORY CELLS INTO THE BRAIN

Although the CNS has long been considered to be an immunologically privileged site, lymphocyte surveillance of the brain may be a normal, low-grade activity that is enhanced in disease. Lymphocytes can enter the brain in response to virus infections and as part of the autoimmune response in multiple sclerosis. Activated, but not resting, lymphocytes pass through the endothelium of small venules, a process that requires the expression of recognition and adhesion molecules (induced following cytokine activation) and subsequently migrate into the brain parenchyma. Within the CNS, microglia and astrocytes can be induced by T-cell cytokines to act as efficient antigen-presenting cells. After leaving the CNS, lymphocytes probably drain along lymphatic pathways to regional cervical lymph nodes.

Polymorphonuclear leukocyte entry into the CNS is less common than lymphocyte entry, but occurs in the early stages of infarction and autoimmune disease and, in particular, in pyogenic infections. They probably enter the nervous system by passing through the endothelial layer. Monocytes may follow similar pathways in the later stages of inflammation. Within the subarachnoid space, polymorphonuclear leukocytes and lymphocytes pass through the endothelium of large veins into the CSF during the inflammatory phase of meningitis.

PERIPHERAL NERVE FIBRES

Classification of peripheral nerve fibres is based on various parameters such as conduction velocity, function and fibre diameter. Of two classifications in common use, the first divides fibres into three major classes, designated A, B and C, corresponding to peaks in the distribution of their conduction velocities. In man, group A fibres are subdivided into α, β, δ and γ subgroups: fibre diameter and conduction velocity are proportional in most fibres. Group Aα fibres are the largest and conduct most rapidly, and C fibres are the smallest and slowest.

The largest afferent axons (Aα fibres) innervate encapsulated cutaneous mechanoreceptors, Golgi tendon organs and muscle spindles, and some large alimentary enteroceptors. Aβ fibres form secondary endings on some muscle spindle (intrafusal) fibres and also innervate cutaneous and joint capsule mechanoreceptors. Aδ fibres innervate thermoreceptors, stretch-sensitive free endings, hair receptors and nociceptors, including those in dental pulp, skin and connective tissue. Aγ fibres are exclusively fusimotor to plate and trail endings on intrafusal muscle fibres. B fibres are myelinated autonomic preganglionic efferent fibres. C fibres are unmyelinated and have thermoreceptive, nociceptive and interoceptive functions, including the perception of slow, burning pain and visceral pain. This scheme can be applied to fibres of both spinal and cranial nerves except perhaps those of the olfactory nerve, where the fibres form a uniquely small and slow group. The largest somatic efferent fibres (Aα) are up to 20 μm in diameter. They innervate extrafusal muscle fibres (at motor end plates) exclusively and conduct at a maximum of 120 m/s. Fibres to fast twitch muscles are larger than those to slow twitch muscle. Smaller (Aγ) fibres of gamma motor neurones, and autonomic preganglionic (B) and postganglionic (C) efferent fibres conduct, in order, progressively more slowly (40 m/s – less than 10 m/s).

A different classification divides fibres into groups I–IV on the basis of their calibre; groups I–III are myelinated and group IV is unmyelinated. Group I fibres are large (12–22 μm), and include primary sensory fibres of muscle spindles (Group Ia) and smaller fibres of Golgi tendon organs (Group Ib). Group II fibres are the secondary sensory terminals of muscle spindles, with diameters of 6–12 μm. Group III fibres, 1–6 μm in diameter, have free sensory endings in the connective tissue sheaths around and within muscles and are nociceptive and, in skin, also thermosensitive. Group IV fibres are unmyelinated, with diameters below 1.5 μm: they include free endings in skin and muscle, and are primarily nociceptive.

CONNECTIVE TISSUE SHEATHS

Nerve trunks, whether uni- or multifascicular, are surrounded by an epineurium. Individual fasciculi are enclosed by a multilayered perineurium, which in turn surrounds the endoneurium or intrafascicular connective tissue (Fig. 3.22).

SCHWANN CELLS

Schwann cells are the major glial type in the PNS. In vitro they are fusiform in appearance. Both in vitro and in vivo, Schwann cells ensheathe peripheral axons, and myelinate those greater than 2 μm diameter. In a mature peripheral nerve, they are distributed along the axons in longitudinal chains: the precise geometry of their association depends on whether the axon is myelinated or unmyelinated. In myelinated axons the territory of a Schwann cell defines an internode.

The molecular phenotype of mature myelin-forming Schwann cells is different from that of mature non-myelinating Schwann cells. Adult myelin-forming Schwann cells are characterized by the presence of several myelin proteins, some, but not all, of which are shared with oligodendrocytes and central myelin. In contrast, expression of the low affinity neurotrophin receptor (p75^NTR) and GFAP intermediate filament protein (which differs from the CNS form in its post-translational modification), characterize adult non-myelin forming Schwann cells.

Schwann cells arise from multipotent cells of the very early migrating neural crest that can also give rise to peripheral neurones. Neuronal signals regulate many aspects of Schwann cell behaviour in developing and postnatal nerves. Axon-associated signals appear to control the proliferation of developing Schwann cells and their precursors; the developmentally programmed death of those precursors in order to match numbers of axons and glia within each peripheral nerve bundle; the production of basal laminae by Schwann cells; the induction and maintenance of myelination. Relatively little is known about either Schwann cell–axon or Schwann cell–Schwann cell signalling in normal nerves. An extensive literature supports the view that Schwann cells are key players in the acute injury response in the PNS, helping to provide a microenvironment that facilitates axonal regrowth. Few Schwann cells persist in chronically denervated nerves.

Unmyelinated axons

Unmyelinated axons (Fig. 3.23) are commonly 1.0 μm in diameter, although some may be 1.5 μm or even 2 μm in diameter. Groups of up to 10 small axons (0.15–2.0 μm in diameter) are enclosed within a chain of overlapping Schwann cells that is surrounded by a basal lamina. Within each Schwann cell, individual axons are usually sequestered from their neighbours by delicate processes of cytoplasm. Axons move between Schwann cell chains as they pass proximodistally along a nerve fasciculus. It seems likely, on the basis of quantitative studies in subhuman primates, that axons from adjacent cord segments may share Schwann cell columns: this phenomenon may play a role in the evolution of neuropathic pain after nerve injury. In the absence of a myelin sheath and nodes of Ranvier, conduction along unmyelinated axons is not saltatory but electrotonic, and the passage of impulses is therefore relatively slow (approximately 0.5–4.0 m/s).

Fig. 3.23 An unmyelinated axon (A), enclosed by cytoplasmic processses of a Schwann cell (S) and its basal lamina (arrow), from a biopsy of human sural nerve.

Myelinated axons

Myelinated axons (Fig. 3.24) have a 1 : 1 relationship with their ensheathing Schwann cells. The territory of an individual Schwann cell defines an internode (Fig. 3.25): internodal length varies directly with the diameter of the fibre, from 150 to 1500 μm. The interval between two internodes is a node of Ranvier. In the PNS, the myelin sheaths on either side of a node terminate in asymmetrically swollen paranodal bulbs. Schwann cell cytoplasm only forms a continuous layer in the perinuclear (mid-internodal) and paranodal regions. Between these sites, internodal Schwann cytoplasm forms a delicate network over the inner (abaxonal) surface of the myelin sheath. The outer (adaxonal) layer of Schwann cell cytoplasm is frequently discontinuous, and axons are surrounded by a narrow periaxonal space (15–20 nm) which, although nominally part of the extracellular space, is functionally isolated from it at the paranodes. For further details, see Scherer & Arroyo (2002).

Fig. 3.24 Transverse section of sciatic nerve showing a myelinated axon and several non-myelinated axons (A), ensheathed by Schwann cells (S), from a biopsy of human sural nerve.

(By courtesy of Professor Susan Standring, GKT School of Medicine, London.)

Fig. 3.25 General plan of a myelinated nerve fibre in longitudinal section including one complete internodal segment and two adjacent paranodal bulbs, used as a key for the more detailed microarchitecture of specific subregions. A, Transverse electron microscope section through the centre of a node of Ranvier, with numerous finger-like processes of adjacent Schwann cells converging towards the nodal axolemma. Many microtubules and neurofilaments are visible within the axoplasm. B, The arrangement of the axon, myelin sheath and Schwann cell cytoplasm at the node of Ranvier and in the paranodal bulbs.

(A, By courtesy of Professor Susan Standring, GKT School of Medicine, London. B, Redrawn from a figure provided by PL Williams and DN Landon.)

Nodes of Ranvier

PNS nodes of Ranvier (Fig. 3.25) are typically approximately 0.8–1.1 μm in length. The calibre of the nodal axon is characteristically reduced relative to that of the internodal axon: this is most marked in the largest calibre axons. Node gaps are filled with an amorphous gap substance and processes of Schwann cell cytoplasm, and are surrounded by a continuous basal lamina elaborated by the ensheathing Schwann cells. In large calibre axons the surfaces of the paranodal bulbs and of the underlying axon are fluted as they approach the nodes. Fluting produces grooves in the external surface of the myelin sheath that are filled by Schwann cell cytoplasm characterized by large numbers of mitochondria. In smaller fibres this arrangement is less obvious, although the paranodal cytoplasm usually contains mitochondria.

Schmidt–Lanterman incisures

Schmidt–Lanterman incisures are helical decompactions of internodal myelin that occur in every sheath, irrespective of thickness. At an incisure the major dense line of the myelin sheath splits to enclose a continuous spiral band of granular cytoplasm that passes between the abaxonal and adaxonal layers of Schwann cell cytoplasm. The minor dense line of the incisural myelin sheath separates to create a long channel that potentially connects the periaxonal space with the extracellular fluid in the endoneurium. The function of incisures is not known: their structure suggests that they may participate in transport of molecules across the myelin sheath.

SATELLITE CELLS

Many non-neuronal cells of the nervous system have been called satellite cells, including small round extracapsular cells in peripheral ganglia, ganglionic capsular cells, Schwann cells, any cell that is closely associated with neuronal somata, and precursor cells associated with striated muscle fibres (p. 117). Within the nervous system, the term is most commonly reserved for flat, epithelioid cells (ganglionic glial cells, capsular cells) that surround the neuronal somata of peripheral ganglia (Fig. 3.26). Their cytoplasm resembles that of Schwann cells, and their deep surfaces interdigitate with reciprocal infoldings in the membranes of the enclosed neurones. Capsular cells are succeeded by similar cells that enclose the initial part of the dendroaxonal process of unipolar sensory neurones in dorsal spinal ganglia, and these in turn are continuous with the Schwann cells that surround the peripheral and central processes of the neurones.

Fig. 3.26 Sensory neurones in a dorsal root ganglion (rat). Neurones (N) are typically variable in size but all are encapsulated by satellite cells (S). Myelinated fibres are seen above and below the neuronal somata. Toluidine blue stained resin section.

(By courtesy of Dr Clare Farmer, King’s College, London.)

BLOOD SUPPLY OF PERIPHERAL NERVES

The blood vessels supplying a nerve end in a capillary plexus that pierces the perineurium. The branches of the plexus run parallel with the fibres, connected by short transverse vessels, forming narrow, rectangular meshes similar to those found in muscle. The blood supply of peripheral nerves is unusual. Endoneurial capillaries have atypically large diameters and intercapillary distances are greater than in many other tissues. Peripheral nerves have two separate, functionally independent vascular systems: an extrinsic system (regional nutritive vessels and epineurial vessels) and an intrinsic system (longitudinally running microvessels in the endoneurium). Anastomoses between the two systems produce considerable overlap between the territories of the segmental arteries. This unique pattern of vessels, together with a high basal nerve blood flow relative to metabolic requirements, means that peripheral nerves possess a high degree of resistance to ischaemia.

GENERAL FEATURES OF SENSORY RECEPTORS

There are three major forms of sensory receptor, neuroepithelial, epithelial and neuronal (Fig. 3.30).

A neuroepithelial receptor is a neurone with a soma lying near a sensory surface and an axon that conveys sensory signals into the CNS to synapse on second order neurones. This is an evolutionarily primitive arrangement, and the only example in man is the sensory neurone of the olfactory epithelium.

An epithelial receptor is a cell that is modified from a non-nervous sensory epithelium and innervated by a primary sensory neurone with a soma lying near the CNS. Examples are epidermal Merkel cells, auditory receptors and taste buds. When activated, this type of receptor excites its neurone by neurotransmission across a synaptic gap.

A neuronal receptor is a primary sensory neurone that has a soma in a craniospinal ganglion and a peripheral axon ending in a sensory terminal. All cutaneous sensors (with the exception of Merkel cells) and proprioceptors are of this type: their sensory terminals may be encapsulated or linked to special mesodermal or ectodermal structures to form a part of the sensory apparatus. The extraneural cells are not necessarily excitable, but create an appropriate environment for the excitation of the neuronal process.

The receptor stimulus is transduced into a graded change of electrical potential at the receptor surface (receptor potential), and this initiates an all-or-none action potential that is transmitted to the CNS. This may occur either in the receptor, when this is a neurone, or partly in the receptor and partly in the neurone that innervates it, in the case of epithelial receptors. Transduction varies with the modality of the stimulus, and usually causes depolarization of the receptor membrane (or hyperpolarization, in the retina). In mechanoreceptors, transduction may involve the deformation of membrane structure, which causes strain- or voltage-sensitive transducing protein molecules to open ion channels. In chemoreceptors, receptor action may resemble that for ACh at neuromuscular junctions. Visual receptors share similarities with chemoreceptors: light causes changes in receptor proteins, which activate G proteins, resulting in the release of second messengers and altered membrane permeability.

The quantitative responses of sensory endings to stimuli vary greatly and increase the flexibility of the functional design of sensory systems. Although increased excitation with increasing stimulus level is a common pattern (‘on’ response), some receptors respond to decreased stimulation (‘off’ response). Even unstimulated receptors show varying degrees of spontaneous background activity against which an increase or decrease in activity occurs with changing levels of stimulus. In all receptors studied, when stimulation is maintained at a steady level, there is an initial burst (the dynamic phase) followed by a gradual adaptation to steady level (the static phase). Though all receptors show these two phases, one or other may predominate, providing a distinction between rapidly adapting endings which accurately record the rate of stimulus onset, and slowly adapting endings which signal the constant amplitude of a stimulus, e.g. position sense. Dynamic and static phases are reflected in the amplitude and duration of the receptor potential and also in the frequency of action potentials in the sensory fibres. The stimulus strength necessary to elicit a response in a receptor, i.e. its threshold level, varies greatly between receptors, and provides an extra level of information about stimulus strength.

For further information on sensory receptors, see Nolte (2002).

FUNCTIONAL CLASSIFICATION OF RECEPTORS

Receptors have been classified in several ways. They may be classified by the modalities to which they are sensitive, such as mechanoreceptors (which are responsive to deformation, e.g. touch, pressure, sound waves, etc.), chemoreceptors, photoreceptors and thermoreceptors. Some receptors are polymodal, ie. they respond selectively to more than one modality: they usually have high thresholds and respond to damaging stimuli associated with irritation or pain (nociceptors).

Another widely used classification divides receptors on the basis of their distribution in the body into exteroceptors, proprioceptors and interoceptors. Exteroceptors and proprioceptors are receptors of the somatic afferent components of the nervous system, while interoceptors are receptors of the visceral afferent pathways.

Exteroceptors respond to external stimuli and are found at, or close to, body surfaces. They can be subdivided into the general or cutaneous sense organs and special sensory organs. General sensory receptors include free and encapsulated terminals in skin and near hairs. Special sensory organs are the olfactory, visual, acoustic, vestibular and taste receptors.

Proprioceptors respond to stimuli to deeper tissues, especially of the locomotor system, and are concerned with detecting movement, mechanical stresses and position. They include Golgi tendon organs, neuromuscular spindles, Pacinian corpuscles, other endings in joints, and vestibular receptors. Proprioceptors are stimulated by the contraction of muscles, movements of joints and changes in the position of the body. They are essential for the coordination of muscles, the grading of muscular contraction, and the maintenance of equilibrium.

Interoceptors are found in the walls of the viscera, glands and vessels, where their terminations include free nerve endings, encapsulated terminals and endings associated with specialized epithelial cells. Nerve terminals are found in the layers of visceral walls and the adventitia of blood vessels, but the detailed structure and function of many of these endings are not well-established. Encapsulated (lamellated) endings occur in the heart, adventitia, and mesenteries. Free terminal arborizations occur in the endocardium, the endomysium of all muscles, and connective tissue generally. Visceral nerve terminals are not usually responsive to stimuli which act on exteroceptors, and do not respond to localized mechanical and thermal stimuli. Tension produced by excessive muscular contraction or by visceral distension often causes pain, particularly in pathological states, which is frequently poorly localized and of a deep-seated nature. Visceral pain is often referred to the corresponding dermatome (see Fig. 15.12). Polymodal nociceptors (irritant receptors) respond to a variety of stimuli such as noxious chemicals or damaging mechanical stimuli. They are mainly the free endings of fine, unmyelinated fibres that are widely distributed in the epithelia of the alimentary and respiratory tracts: they may initiate protective reflexes.

Interoceptors include vascular chemoreceptors, e.g. the carotid body, and baroceptors, which are concerned with the regulation of blood flow and pressure and the control of respiration.

FREE NERVE ENDINGS

Sensory endings that branch to form plexuses occur in many sites (Fig. 3.30). They occur in all connective tissues, including those of the dermis, fasciae, capsules of organs, ligaments, tendons, adventitia of blood vessels, meninges, articular capsules, periosteum, perichondrium, Haversian systems in bone, parietal peritoneum, walls of viscera and the endomysium of all types of muscle. They also innervate the epithelium of the skin, cornea, buccal cavity, and the alimentary and respiratory tracts and their associated glands. Within epithelia, free sensory endings lack Schwann cell ensheathment and are enveloped instead by epithelial cells. Afferent fibres from free terminals may be myelinated or unmyelinated but are always of small diameter and low conduction velocity. When afferent axons are myelinated, their terminal arborizations are not. These terminals serve several sensory modalities. In the dermis, they may be responsive to moderate cold or heat (thermoreceptors); light mechanical touch (mechanoreceptors); damaging heat, cold or deformation (unimodal nociceptors); damaging stimuli of several kinds (polymodal nociceptors). Similar fibres in deeper tissues may also signal extreme conditions, which are experienced, as with all nociceptors, as pain. Free endings in the cornea, dentine and periosteum may be exclusively nociceptive.

Special types of free ending are associated with epidermal structures in the skin. They include terminals associated with hair follicles (peritrichial receptors) which branch from myelinated fibres in the deep dermal cutaneous plexus: the number, size and form of the endings are related to the size and type of hair follicle innervated. These endings respond mainly to movement when hair is deformed and belong to the rapidly adapting mechanoreceptor group.

Merkel tactile endings lie either at the base of the epidermis or around the apical ends of some hair follicles, and are innervated by large myelinated axons. Each axon expands into a disc that is applied closely to the base of a Merkel cell in the basal layer of the epidermis. The cells, which are believed to be derived from the neural crest, contain many large (50–100 nm) dense-cored vesicles, presumably containing transmitters, which are concentrated near the junction with the axon. Merkel endings are slow-adapting mechanoreceptors and are responsive to sustained pressure and sensitive to the edges of applied objects.

ENCAPSULATED ENDINGS

Encapsulated endings are a major group of special endings that includes lamellated corpuscles of various kinds (e.g. Meissner’s, Pacinian), Golgi tendon organs, neuromuscular spindles and Ruffini endings (Fig. 3.30). They exhibit considerable variety in their size, shape and distribution, but share a common feature, in that each axon terminal is encapsulated by non-excitable cells.

Meissner’s corpuscles

Meissner’s corpuscles are found in the dermal papillae of all parts of the hand and foot, the anterior aspect of the forearm, the lips, palpebral conjunctiva and mucous membrane of the apical part of the tongue. They are most concentrated in thick hairless skin, especially of the finger pads, where there may be up to 24 corpuscles per cm² in young adults. Mature corpuscles are cylindrical in shape, approximately 80 μm long and 30 μm across, with their long axes perpendicular to the skin surface. Each corpuscle has a connective tissue capsule and central core composed of a stack of flat modified Schwann cells (Fig. 3.31). Meissner’s corpuscles are rapidly adapting mechanoreceptors, sensitive to shape and textural changes in exploratory and discriminatory touch: their acute sensitivity provides the neural basis for reading Braille text.

Fig. 3.31 A tactile Meissner’s corpuscle in a dermal papilla in the skin, demonstrated using the modified Bielschowsky silver stain technique.

(By courtesy of N Cauna, University of Pittsburgh.)

Pacinian corpuscles

Pacinian corpuscles are situated subcutaneously in the palmar and plantar aspects of the hand and foot and their digits; the external genitalia; arm; neck; nipple; periostea; interosseous membranes; near joints, and within the mesenteries. They are oval, spherical or irregularly coiled and measure up to 2 mm in length and 100–500 μm or more across: the larger ones are visible to the naked eye. Each corpuscle has a capsule, an intermediate growth zone and a central core that contains an axon terminal. The capsule is formed by approximately 30 concentrically arranged lamellae of flat cells approximately 0.2 μm thick (Fig. 3.32). Adjacent cells overlap and successive lamellae are separated by an amorphous proteoglycan matrix that contains circularly orientated collagen fibres, closely applied to the surfaces of the lamellar cells. The amount of collagen increases with age. The intermediate zone is cellular and its cells become incorporated into the capsule or core, so that it is not clearly defined in mature corpuscles. The core consists of approximately 60 bilateral, compacted lamellae lying on both sides of a central nerve terminal.

Fig. 3.32 A Pacinian corpuscle in human dermis.

Each corpuscle is supplied by a myelinated axon, which initially loses its myelin sheath and subsequently loses its ensheathing Schwann cell at its junction with the core. The naked axon runs through the central axis of the core and ends in a slightly expanded bulb. It is in contact with the innermost core lamellae, is transversely oval and sends short projections of unknown function into clefts in the lamellae. The axon contains numerous large mitochondria, and minute vesicles, approximately 5 nm in diameter, which aggregate opposite the clefts. The cells of the capsule and core lamellae are thought to be specialized fibroblasts but some may be Schwann cells. Elastic fibrous tissue forms an overall external capsule to the corpuscle. Pacinian corpuscles are supplied by capillaries that accompany the axon as it enters the capsule.

Pacinian corpuscles act as very rapidly adapting mechanoreceptors. They respond only to sudden disturbances and are especially sensitive to vibration. The rapidity may be partly due to the lamellated capsule acting as a high pass frequency filter, damping slow distortions by fluid movement between lamellar cells. Groups of corpuscles respond to pressure changes, e.g. on grasping or releasing an object.

Golgi tendon organs

Golgi tendon organs are found mainly near musculotendinous junctions (Fig. 3.33), where more than 50 may occur at any one site. Each terminal is related closely to a group of muscle fibres (up to 20) as they insert into the tendon. Golgi tendon organs are approximately 500 μm long and 100 μm in diameter, and consist of small bundles of tendon fibres enclosed in a delicate capsule. The collagen bundles (intrafusal fasciculi) are less compact than elsewhere in the tendon, the collagen fibres are smaller and the fibroblasts larger and more numerous. A single, thickly myelinated 1b afferent nerve fibre enters the capsule and divides. Its branches, which may lose their ensheathing Schwann cells, terminate in leaf-like enlargements containing vesicles and mitochondria, which wrap around the tendon. A basal lamina or process of Schwann cell cytoplasm separates the nerve terminals from the collagen bundles that constitute the tendon. Golgi tendon organs are activated by passive stretch of the tendon, but are much more sensitive to active contraction of the muscle. They are important in providing proprioceptive information that complements the information coming from neuromuscular spindles. Their responses are slowly adapting and they signal maintained tension.

Fig. 3.33 The structure and innervation of a Golgi tendon organ. For clarity the perineurium and endoneurium have been omitted to show the distribution of nerve fibres ramifying between the collagen fibre bundles of the tendon.

NEUROMUSCULAR SPINDLES

Neuromuscular spindles are essential for the control of muscle contraction. Each spindle contains a few small, specialized intrafusal muscle fibres, innervated by both sensory and motor nerve fibres (Fig. 3.34, Fig. 3.35). The whole is surrounded equatorially by a fusiform spindle capsule of connective tissue, consisting of an outer perineurial-like sheath of flattened fibroblasts and collagen and an inner sheath that forms delicate tubes around individual intrafusal fibres (Fig. 3.36). A gelatinous fluid rich in glycosaminoglycans fills the space between the two sheaths.

Fig. 3.34 A neuromuscular spindle, showing nuclear bag and nuclear chain fibres within the spindle capsule (green); these are innervated by the sensory anulospiral and ‘flower spray’ afferent fibre endings (blue) and by the γ and β fusimotor (efferent fibre) endings (red). The β efferent fibres are collaterals of fibres innervating extrafusal slow twitch muscle cells (pink, bottom left).

Fig. 3.35 Nuclear bag and nuclear chain fibres in a neuromuscular spindle. Dynamic β- and γ-efferents innervate dynamic bag 1 intrafusal fibres, whereas static β- and γ-efferents innervate static bag 2 and nuclear chain intrafusal fibres.

Fig. 3.36 Neuromuscular spindle in transverse section in human extraocular muscle. The spindle capsule (C) encloses intrafusal fibres (IF) of varying diameters. Typical muscle fibres (M) in transverse section are shown above the spindle. Toluidine blue stained resin section.

There are usually 5–14 intrafusal fibres (the number varies between muscles) and two major types of fibre, nuclear bag and nuclear chain fibres, which are distinguished by the arrangement of nuclei in their sarcoplasm. In nuclear bag fibres, an equatorial cluster of nuclei makes the fibre bulge slightly, whereas the nuclei in nuclear chain fibres form a single axial row. Nuclear bag fibres are greater in diameter than chain fibres and extend beyond the surrounding capsule to the endomysium of nearby extrafusal muscle fibres. Nuclear chain fibres are attached at their poles to the capsule or to the sheaths of nuclear bag fibres.

The intrafusal fibres resemble typical skeletal muscle fibres, except that the zone of myofibrils is thin around the nuclei. One subtype of nuclear bag fibre (dynamic bag 1) generally lacks M lines, possesses little sarcoplasmic reticulum and has an abundance of mitochondria and oxidative enzymes, but little glycogen. A second subtype of bag fibre (static bag 2) has distinct M lines and abundant glycogen. Nuclear chain fibres have marked M lines, sarcoplasmic reticulum and T-tubules, and abundant glycogen, but few mitochondria. These variations reflect the contractile properties of different intrafusal fibres (Boyd 1985).

Muscle spindles receive two types of sensory innervation via the unmyelinated terminations of large myelinated axons. Primary (anulospiral) endings are equatorially placed and form spirals around the nucleated parts of intrafusal fibres. They are the endings of large sensory fibres (Group Ia afferents), each of which sends branches to a number of intrafusal muscle fibres. Each terminal lies in a deep sarcolemmal groove in the spindle plasma membrane beneath its basal lamina. Secondary (flowerspray) endings, which may be spray-shaped or anular, are largely confined to nuclear chain fibres, and are the branched terminals of somewhat thinner myelinated (Group II) afferents. They are varicose and spread in a narrow band on both sides of the primary endings. They lie close to the sarcolemma, though not in grooves. In essence primary endings are rapidly adapting, while secondary endings have a regular, slowly adapting, response to static stretch.

There are three types of motor endings in muscle spindles. Two are from fine, myelinated, fusimotor (γ) efferents and one is from myelinated (β) efferent collaterals of axons that supply extrafusal slow twitch muscle fibres. The fusimotor efferents terminate nearer the equatorial region where their terminals either resemble the motor end-plates of extrafusal fibres (plate endings) or are more diffuse (trail endings). Stimulation of the fusimotor and β efferents causes contraction of the intrafusal fibres and consequently, activation of their sensory endings.

Muscle spindles signal the length of extrafusal muscle both at rest and throughout contraction and relaxation, the velocity of their contraction and changes in velocity. These modalities may be related to the different behaviours of the three major types of intrafusal fibre and their sensory terminals. The sensory endings of one type of nuclear bag fibre (dynamic bag 1) are particularly concerned with signalling rapid changes in length that occur during movement, whilst those of the second bag fibre type (static bag 2), are less responsive to movement. The afferents from chain fibres have relatively slowly adapting responses at all times. These elements can therefore detect complex changes in the state of the extrafusal muscle surrounding spindles and can signal fluctuations in length, tension, velocity of length change and acceleration. Moreover, they are under complex central control: efferent (fusimotor) nerve fibres, by regulating the strength of contraction, can adjust the length of the intrafusal fibres and thereby the responsiveness of spindle sensory endings. In summary, the organization of spindles allows them to monitor muscle conditions actively in order to compare intended and actual movements, and to provide a detailed input to spinal, cerebellar, extrapyramidal and cortical centres about the state of the locomotor apparatus.

JOINT RECEPTORS

The arrays of receptors situated in and near articular capsules provide information on the position, movements and stresses acting on joints. Structural and functional studies have demonstrated at least four types of joint receptor: their proportions and distribution vary with site. Three are encapsulated endings, the fourth a free terminal arborization.

Type I endings are encapsulated corpuscles of the slowly adapting mechanoreceptor type and resemble Ruffini endings. They lie in the superficial layers of the fibrous capsules of joints in small clusters and are innervated by myelinated afferent axons. Being slowly adapting, they provide awareness of joint position and movement, and respond to patterns of stress in articular capsules. They are particularly common in joints where static positional sense is necessary for the control of posture (e.g. hip, knee).

Type II endings are lamellated receptors, and resemble small versions of the large Pacinian corpuscles found in general connective tissue. They occur in small groups throughout joint capsules, particularly in the deeper layers and other articular structures (e.g. the fat pad of the temporomandibular joint). They are rapidly adapting, low-threshold mechanoreceptors, sensitive to movement and pressure changes, and they respond to joint movement and transient stresses in the joint capsule. They are supplied by myelinated afferent axons, but are probably not involved in the conscious awareness of joint sensation.

Type III endings are identical to Golgi tendon organs in structure and function: they occur in articular ligaments, but not in joint capsules. They are high-threshold, slowly adapting receptors and may serve, at least in part, to prevent excessive stresses at joints by reflex inhibition of the adjacent muscles. They are innervated by large myelinated afferent axons.

Type IV endings are free terminals of myelinated and unmyelinated axons that ramify in articular capsules, the adjacent fat pads and around the blood vessels of the synovial layer. They are high-threshold, slowly adapting receptors and are thought to respond to excessive movements, providing a basis for articular pain.

SKELETAL MUSCLE

The most intensively studied effector endings are those that innervate muscle, particularly skeletal muscle. All neuromuscular (myoneural) junctions are axon terminals of somatic motor neurones. They are specialized for the release of neurotransmitter on to the sarcolemma of skeletal muscle fibres, causing a change in their electrical state which leads to contraction. Each axon branches near its terminal to innervate from several to hundreds of muscle fibres, the number depending on the precision of motor control required.

The detailed structure of a motor terminal varies with the type of muscle innervated. Two major types of ending are recognized, innervating either extrafusal muscle fibres, or the intrafusal fibres of neuromuscular spindles. In the former type, each axonal terminal usually ends midway along a muscle fibre in a discoidal motor end-plate (Fig. 3.37), and usually initiates action potentials that are rapidly conducted to all parts of the muscle fibre. In the latter type, the axon gives off numerous branches that form a cluster of small expansions extending along the muscle fibre; in the absence of propagated muscle excitation, these excite the fibre at several points. Both types of ending are associated with a specialized receptive region of the muscle fibre, the sole plate, where a number of muscle cell nuclei are grouped within the granular sarcoplasm.

Fig. 3.37 The neuromuscular junction. A, Whole-mount preparation of teased skeletal muscle fibres (pale, faintly striated, diagonally orientated structures). The terminal part of the axon (silver-stained, brown) branches to form motor end-plates on adjacent muscle fibres. The sole plate recesses in the sarcolemma, into which the motor end-plates fit, are demonstrated by the presence of acetylcholinesterase (shown by enzyme histochemistry, blue). B, Diagrammatic representation of the axonal motor end-plate and the deeply infolded sarcolemma. C, Electron micrograph showing the expanded motor end-plate of the axon filled with vesicles containing synaptic transmitter (ACh) (above); the deep infoldings of the sarcolemmal sole plate (below) form subsynaptic gutters.

(A By courtesy of Dr Norman Gregson, Division of Neurology, GKT School of Medicine, London; C By courtesy of Professor DN Landon, Institute of Neurology, University College London.)

The sole plate contains numerous mitochondria, endoplasmic reticulum and Golgi complexes. The terminal branches of the axon are plugged into shallow grooves in the surface of the sole plate (primary clefts), from where numerous pleats extend for a short distance into the underlying sarcoplasm (secondary clefts) (Fig. 3.37B,C). The axon terminal contains mitochondria and many clear, 60 nm spherical vesicles similar to those in presynaptic boutons, which are clustered over the zone of membrane apposition. It is ensheathed by Schwann cells whose cytoplasmic projections extend into the synaptic cleft. The plasma membranes of the axon terminal and the muscle cell are separated by a 30–50 nm gap and an interposed basal lamina which follows the surface folding of the sole plate membrane into the secondary clefts. The basal lamina contains specialized components including specific isoforms of type IV collagen and laminin and the heparan sulphate proteoglycan, agrin. Endings of fast and slow twitch muscle fibres (p. 111) differ in detail: the sarcolemmal grooves are deeper, and the presynaptic vesicles more numerous, in the fast fibres.

Junctions with skeletal muscle are cholinergic; the release of ACh changes the ionic permeability of the muscle fibre. Clustering of ACh receptors at the neuromuscular junction depends in part on the presence in the muscle basal lamina of agrin, which is secreted by the motor neurone, and is important in establishing the post-junctional molecular machinery. When the depolarization of the sarcolemma reaches a particular threshold, it initiates an all-or-none action potential in the sarcolemma, which is then propagated rapidly over the whole cell surface and also deep within the fibre via the invaginations (T-tubules) of the sarcolemma (p. 108), causing contraction. The amount of ACh released by the arrival of a single nerve impulse is sufficient to trigger an action potential. However, because ACh is very rapidly hydrolysed by the enzyme AchE, present at the sarcolemmal surface of the sole plate, a single nerve impulse only gives rise to one muscle action potential, i.e. there is a one-to-one relationship between neural and muscle action potentials. Thus the contraction of a muscle fibre is controlled by the firing frequency of its motor neurone.

Neuromuscular junctions are partially blocked by high concentrations of lactic acid, as in some types of muscle fatigue.

AUTONOMIC MOTOR TERMINATIONS

Autonomic neuromuscular junctions differ in several important ways from the skeletal neuromuscular junction and from synapses in the CNS and PNS. There is no fixed junction with well defined pre- and post-junctional specializations. Unmyelinated, highly branched, postganglionic autonomic axons become beaded or varicose as they reach the effector smooth muscle. These varicosities are not static but move along axons. They are packed with mitochondria and vesicles containing neurotransmitters which are released from the varicosities during conduction of an impulse along the axon. The distance (cleft) between the varicosity and smooth muscle membrane varies considerably depending on the tissue, from 20 nm in densely innervated structures such as the vas deferens to 1–2 μm in large elastic arteries. Unlike skeletal muscle, the effector tissue is a muscle bundle rather than a single cell. Gap junctions between individual smooth muscle cells are low resistance pathways which allow electronic coupling and the spread of activity within the effector bundle: they vary in size from punctate junctions to junctional areas of more than 1 μm in diameter.

Adrenergic sympathetic postganglionic terminals contain dense-cored vesicles. Cholinergic terminals, which are typical of all parasympathetic and some sympathetic endings, contain clear spherical vesicles like those in the motor end-plates of skeletal muscle. A third category of autonomic neurones has non-adrenergic, non-cholinergic endings which contain a wide variety of chemicals with transmitter properties. Conjugated purine (ATP, a nucleoside), is probably the neurotransmitter at these terminals, which are thus classed as purinergic. For further reading on purinergic signalling, see Burnstock (2006). Typically, their axons contain large, 80–200 nm, dense opaque vesicles, congregated in varicosities at intervals along axons. Such terminals are formed in many sites, including the external muscle layers and sphincters of the alimentary tract, lungs, walls of blood vessels, urogenital tract and in the CNS (see Chapter 15). In the intestinal wall, neuronal somata lie in the myenteric plexus, and their axons spread caudally for a few millimetres, mainly to innervate circular muscle. Purinergic neurones are under cholinergic control from preganglionic sympathetic neurones, via postganglionic sympathetic fibres. Their endings mainly hyperpolarize smooth muscle cells, causing relaxation, e.g. preceding peristaltic waves, opening sphincters and, probably, causing reflex distension in gastric filling.

Autonomic efferents also innervate glands, myoepithelial cells, adipose and lymphoid tissues.

ACTION POTENTIAL

The action potential is a brief, self-propagating reversal of membrane polarity. It depends on an initial influx of sodium ions which causes a reversal of polarity to about +40 mV, followed by a rapid return towards the resting potential as potassium ions flow out (the detailed mechanism differs somewhat between CNS and PNS). The rapid reversal process is completed in approximately 0.5 msec, followed by a slower recovery phase of up to 5 msec, when the resting potential is fully restored. Once the axon hillock reaches threshold, propagation of the action potential is independent of the initiating stimulus, thus the size and duration of action potentials are always the same (described as all-or-none) for a particular neurone, no matter how much a stimulus may exceed the threshold value.

Once initiated, an action potential spreads spontaneously and at a relatively constant velocity, within the range of 4–120 m/s. Conduction velocity depends on a number of factors related to the way in which the current spreads, e.g. axonal cross-sectional area, the numbers and positioning of ion channels, and membrane capacitance (influenced particularly by the presence of myelin). In axons lacking myelin, action potential conduction is analogous to a flame moving along a fuse. Just as each segment of fuse is ignited by its upstream neighbour, each segment of axon membrane is driven to threshold by the depolarization of neighbouring membrane. Sodium channels within the newly depolarized segment open and positively charged sodium ions enter, driving the local potential inside the axon toward positive values. This inward current in turn depolarizes the neighbouring, downstream, nondepolarized membrane, and the cyclic propagation of the action potential is completed.

Myelinated fibres are electrically insulated by their myelin sheaths along most of their lengths, except at nodes of Ranvier. The distance between nodes, referred to as the internodal distance, is directly related to axon diameter and varies between 0.2 and 2.0 mm. Voltage-gated sodium channels are clustered at nodes, and the nodal membrane is the only place where high densities of inward sodium current can be generated across the axon membrane. Like conduction in unmyelinated axons, conduction in myelinated axons is self-propagating, but instead of physically adjacent regions of membrane acting to excite one another, it is the depolarization occurring in the neighbouring upstream node that excites a node to threshold. Reaching threshold causes the sodium channels at the node to open and generate inward sodium current, but instead of this acting on the adjacent membrane, the high resistance and low capacitance of the myelin sheath directs the current toward the next downstream node, exciting it to threshold and completing the cycle. The action potential thus jumps from node to node, a process known as saltatory conduction, which greatly increases the conduction velocity.

A number of disorders of the CNS and PNS include demyelination as a characteristic feature. Perhaps most common amongst these is multiple sclerosis, which is characterized by primary demyelination at scattered sites within the white matter of the CNS (it is now recognized that axonal loss also contributes to the progression of multiple sclerosis). Primary demyelination is the loss of the myelin sheath with axonal preservation, and is usually segmental, ie. it rarely extends along the entire length of an affected axon. The phenomenon is associated with conduction block because the newly exposed, previously internodal, axolemma contains relatively few voltage-sensitive Na⁺ channels. There is experimental evidence that conduction can be restored in some demyelinated axons, and experimental and clinical evidence that remyelinated axons can conduct at near normal speeds, because even though their sheaths are thinner than the original myelin sheaths, the safety factor (i.e. the factor by which the outward current at a quiescent node next to an excited node exceeds the minimum current required to evoke a response) is greater than 1. The myelin loss that occurs in the early stages of Wallerian degeneration in both CNS and PNS, usually distal to a site of trauma but also in response to a prolonged period of ischaemia or exposure to a neuronotoxic substance, is accompanied by axonal degeneration (the term secondary demyelination is sometimes used to describe this form of myelin loss).

In both myelinated and unmyelinated fibres there is an irreducible interval at the end of an action potential, termed the refractory period, during which another action potential cannot be triggered. This determines the maximum frequency at which action potentials can be conducted along a nerve fibre: its value differs in different neurones and affects the amount of information which can be carried by an individual fibre.

Axonal conduction is naturally unidirectional, from dendrites and soma to axon terminals. When an action potential reaches the axonal terminals, it causes depolarization of the presynaptic membrane and as a result, quanta of neurotransmitter (which correspond to the content of individual vesicles) are released to change the degree of excitation of the next neurone, muscle fibre or glandular cell. For further information, see Kandel et al (2000).