General structural features of neurons are the perikarya, dendrites and axons

The stereotypical image of a neuron is that of a stellate cell body, the perikaryon or soma, with broad dendrites emerging from one pole and a fine axon emerging from the opposite pole. This impression stems from the older work of Purkinje, who first described the nerve cell in 1839, and of Deiters, Ramón y Cajal and Golgi (see [3]) at the end of the nineteenth century and the early twentieth century. However, this picture does not hold true for many neurons. The neuron is the most polymorphic cell in the body and defies formal classification on the basis of shape, location, function, fine structure or transmitter substance. Although early workers described the neuron as a globular mass suspended between nerve fibers, the teased preparations of Deiters and his contemporaries soon proved this not to be the case. Later work, using impregnation staining and culture techniques, elaborated on Deiters' findings. Before the work of Deiters and Ramón y Cajal, both neurons and neuroglia were believed to form syncytia, with no intervening membranes. Today, of course, we are familiar with the specialized membranes and the enormous variety of nerve cell shapes and sizes. They range from the small, globular cerebellar granule cells, with a perikaryal diameter of approximately 6 to 8 μm, to the pear-shaped Purkinje cells and star-shaped anterior horn cells, both of which may reach diameters of 60 to 80 μm in humans. Perikaryal size is generally a poor index of total cell volume, however, and it is a general rule in neuroanatomy that neurites occupy a greater percentage of the cell surface area than does the soma. For example, the pyramidal cell of the somatosensory cortex has a cell body that accounts for only 4% of the total surface area, whereas from its dendritic tree the dendritic spines alone claim 43% (Mungai, quoted by Peters et al. [3]). Hyden [2] quotes Scholl (1956), who calculated that the perikaryon of a “cortical cell” represents 10% of the neuronal surface area. In the feline reticular formation, some giant cells possess ratios between soma and dendrites of about 1:5. A single axon is the usual rule, but some cells, like the Golgi cells of the cerebellum, are endowed with several axons, some of which may show branching.

The extent of the branching displayed by the dendrites is a useful index of their functional importance. Dendritic trees represent the expression of the receptive fields, and large fields can receive inputs from multiple origins. A cell with a less developed dendritic ramification, such as the cerebellar granule cell, has synapses with a more homogeneous population of afferent sources.

The axon emerges from a neuron as a slender thread and frequently does not branch until it nears its target. In contrast to the dendrite and the soma, the axon is myelinated frequently, thus increasing its efficiency as a conducting unit. Myelin, a spirally wrapped membrane (see Chap. 4), is laid down in segments, or internodes, by oligodendrocytes in the CNS and by Schwann cells in the PNS. The naked regions of axon between adjacent myelin internodes are known as nodes of Ranvier (see below).

Neurons contain the same intracellular components as do other cells

No unique cytoplasmic inclusions of the neuron distinguish it from any other cell. Neurons have all the morphological counterparts of other cell types, the structures are similarly distributed and some of the most common, the Golgi apparatus and mitochondria, for example, were described first in neurons (Fig. 1-4).

The nucleus is large and usually spherical, containing a prominent nucleolus. The nucleochromatin is invariably pale, with little dense heterochromatin. In some neurons, such as the cerebellar granule cells, the nucleoplasm may show more differentiation and dense heterochromatin. The nucleolus is vesiculated and clearly delineated from the rest of the nucleoplasm. It usually contains two textures: the pars fibrosa, which are fine bundles of filaments, and the pars granulosa, in which dense granules predominate. An additional juxtaposed structure, found in neurons of the female of some species, is the nucleolar satellite, or sex chromatin, which consists of dense but loosely packed, coiled filaments. The nucleus is enclosed by the nuclear envelope, made up on the cytoplasmic side by the perikaryon inner membrane, which sometimes is seen in continuity with the endoplasmic reticulum (Fig. 1-5), and a more regular membrane on the inner, or nuclear, aspect of the envelope. Between the two is a clear channel of between 20 and 40 nm. Periodically, the inner and outer membranes of the envelope come together to form a single diaphragm, a nuclear pore (Fig. 1-5). In tangential section, nuclear pores are seen as empty vesicular structures, approximately 70 nm in diameter. In some neurons, as in Purkinje cells, that segment of the nuclear envelope which faces the dendritic pole is deeply invaginated.

The perikaryon, or body of the neuron, is rich in organelles (Fig. 1-4). It often stands out poorly from a homogeneous background neuropil, most of which is composed of nonmyelinated axons and dendrites, synaptic complexes and glial cell processes. Closer inspection shows that, like all cells, the neuron is delineated by a typical triple-layered unit membrane approximately 7.5 nm wide. Among the most prominent features of the perikaryal cytoplasm is a system of membranous cisternae, divisible into rough or granular endoplasmic reticulum (ER), which forms part of the Nissl substance; smooth or agranular ER; subsurface cisternae; and the Golgi apparatus. Although these various components are interconnected structurally, each possesses distinct enzymological properties. Also present within the cytoplasm are abundant lysosomes; lipofuscin granules, which also are termed aging pigment; mitochondria; multivesicular bodies; neurotubules; neurofilaments; and ribosomes.

Nissl substance consists of the intracytoplasmic basophilic masses that ramify loosely throughout the cytoplasm and is typical of most neurons (Figs. 1-4 and 1-5). The distribution of Nissl substance in certain neurons is characteristic and can be used as a criterion for identification. By electron microscopy (EM), this substance is seen to comprise regular arrays or scattered portions of flattened cisternae of the rough ER surrounded by clouds of free polyribosomes. The membranes of the rough ER are studded with rows of ribosomes, which produce the granular appearance of the rough ER. A space of 20 to 40 nm is maintained within cisternae. Sometimes, cisternal walls meet at fenestrations. Unlike the rough ER of glandular cells or other protein-secreting cells, such as plasma cells, the rough ER of neurons probably produces most of its proteins for use within that neuron, a feature imposed by the extraordinary functional demands placed on the cell. Nissl substance does not penetrate axons but does extend along dendrites.

Smooth endoplasmic reticulum is present in most neurons, although it is sometimes difficult to differentiate it from the rough ER owing to the disorderly arrangement of ribosomes. Ribosomes are not associated with these membranes, and the cisternae usually assume a meandering, branching course throughout the cytoplasm. In some neurons, the smooth ER is quite prominent, for example, in Purkinje cells. Individual cisternae of the smooth ER extend along axons and dendrites (see Chaps. 8 and 9).

Subsurface cisternae are a system of smooth, membrane-bound, flattened cisternae that can be found in many neurons. These structures, referred to as hypolemmal cisternae by Palay and Chan-Palay [1], abut the plasmalemma of the neuron and constitute a secondary membranous boundary within the cell. The distance between these cisternae and the plasmalemma is usually 10 to 12 nm, and in some neurons, such as the Purkinje cells, a mitochondrion may be found in close association with the innermost leaflet. Similar cisternae have been described beneath synaptic complexes, but their functional significance is not known. Some authors have suggested that such a system may play a role in the uptake of metabolites. Membrane structures are described in Chapter 2.

The Golgi apparatus is a highly specialized form of agranular reticulum and is visualized best using the metal impregnation techniques of Golgi. Ultrastructurally, the Golgi apparatus consists of aggregates of smooth-walled cisternae and a variety of vesicles. It is surrounded by a heterogeneous assemblage of organelles, including mitochondria, lysosomes and multivesicular bodies. In most neurons, the Golgi apparatus encompasses the nucleus and extends into dendrites but is absent from axons. A three-dimensional analysis of the system reveals that the stacks of cisternae are pierced periodically by fenestrations. Tangential sections of these fenestrations show them to be circular profiles. A multitude of vesicles is associated with each segment of the Golgi apparatus, particularly “coated” vesicles, which proliferate from the lateral margins of flattened cisternae (Fig. 1-6) (see Chap. 9). Such structures have been variously named, but the term alveolate vesicle seems to be generally accepted. Histochemical staining reveals that these bodies are rich in acid hydrolases, and they are believed to represent primary lysosomes [7]. Acid phosphatase also is found elsewhere in the cisternae but in lesser amounts than in alveolate vesicles.

The lysosome is the principal organelle responsible for the degradation of cellular waste. It is a common constituent of all cell types of the nervous system and is particularly prominent in neurons, where it can be seen at various stages of development (Fig. 1-4). It ranges in size from 0.1 to 2 μm in diameter. The primary lysosome is elaborated from Golgi saccules as a small, vesicular structure (Fig. 1-6). Its function is to fuse with the membrane of waste-containing vacuoles, termed phagosomes, into which it releases hydrolytic enzymes (see Chap. 41). The sequestered material is then degraded within the vacuole, and the organelle becomes a secondary lysosome; it is usually electron-dense and large. The matrix of this organelle will give a positive reaction when tested histochemically for acid phosphatase. Residual bodies containing nondegradable material are considered to be tertiary lysosomes, and in the neuron some are represented by lipofuscin granules (Fig. 1-7). These granules contain brown pigment and lamellar stacks of membrane material and are more common in the aged brain [7].

Multivesicular bodies usually are found in association with the Golgi apparatus and are visualized by EM as small, single membrane-bound sacs approximately 0.5 μm in diameter. They contain several minute, spherical profiles, sometimes arranged about the periphery. They are believed to belong to the lysosome series prior to secondary lysosomes because they contain acid hydrolases and apparently are derived from primary lysosomes.

Neurotubules have been the subject of intense research [8]. They usually are arranged haphazardly throughout the perikaryon of neurons but are aligned longitudinally in axons and dendrites. Each neurotubule consists of a dense-walled structure enclosing a clear lumen, in the middle of which may be found an electron-dense dot. Axonal neurotubules display 5-nm filamentous interconnecting side-arms known to be involved in axoplasmic transport in association with the proteins dynein and kinesin (see Chap. 28). The diameter of neurotubules varies between 22 and 24 nm. High-resolution studies indicate that each neurotubule wall consists of 13 filamentous subunits arranged helically around a lumen (see also Chaps. 8 and 28).

Neurofilaments belong to the family of intermediate filaments and usually are found in association with neurotubules. The function of these two organelles has been debated for some time [8,9], and current views of their roles in the maintenance of form and in axoplasmic transport are discussed in Chapters 8 and 28. Neurofilaments have a diameter of approximately 10 nm, are of indeterminate length and frequently occur in bundles. They are constant components of axons but are rarer in dendrites. In the axon, individual filaments possess a minute lumen and interconnect by proteinaceous side-arms, thereby forming a meshwork. Because of these cross-bridges, they do not form tightly packed bundles in the normal axon, in contrast to filaments within astrocytic processes (see Fig. 1-14), which lack cross-bridges. Neurofilaments within neuronal somata usually do not display cross-bridges and can be found in tight bundles. A form of filamentous structure finer than neurofilaments is seen at the tips of growing neurites, particularly in the growth cones of developing axons. These structures, known as microfilaments, are 5 nm in size and are composed of actin. They facilitate movement and growth since it has been shown that axonal extension can be arrested pharmacologically by treatment with compounds that depolymerize these structures. The biochemistry of neurotubules and neurofilaments is dealt with in more detail in Chapter 8 and in Soifer [8] and Wang et al. [9].

Mitochondria are the centers for oxidative phosphorylation. These organelles occur ubiquitously in the neuron and its processes (Figs. 1-4 and 1-6). Their overall shape may change from one type of neuron to another, but their basic morphology is identical to that in other cell types. Mitochondria consist morphologically of double-membraned sacs surrounded by protuberances, or cristae, extending from the inner membrane into the matrix space [7].

The axon becomes physiologically and structurally divisible into the following distinct regions as it egresses: the axon hillock, the initial segment, the axon proper and the axonal termination [3]. The segments differ ultrastructurally in membrane morphology and the content of the rough and smooth ER. The axon hillock may contain fragments of Nissl substance, including abundant ribosomes, which diminish as the hillock continues into the initial segment. Here, the various axoplasmic components begin to align longitudinally. A few ribosomes and the smooth ER persist, and some axoaxonic synapses occur. More interesting, however, is the axolemma of the initial segment, the region for the generation of the action potential, which is underlaid by a dense granular layer similar to that seen at the nodes of Ranvier. Also present in this region are neurotubules, neurofilaments and mitochondria. The arrangement of the neurotubules in the initial segment, unlike their scattered pattern in the distal axon, is in fascicles; they are interconnected by side-arms [3,9]. Beyond the initial segment, the axon maintains a relatively uniform morphology. It contains the axolemma without any structural modification, except at nodes and the termination, where submembranous densities are seen; microtubules, sometimes cross-linked; neurofilaments, connected by side-arms; mitochondria; and tubulovesicular profiles, probably derived from the smooth ER. Myelinated axons show granular densifications beneath the axolemma at the nodes of Ranvier [6,10], and synaptic complexes may occur in the same regions. In myelinated fibers, there is a concentration of sodium channels at the nodal axon, a feature underlying the rapid, saltatory conduction of such fibers [11] (see Chaps. 6 and 10). The terminal portion of the axon arborizes and enlarges at its synaptic regions, where it might contain synaptic vesicles beneath the specialized presynaptic junction.

The dendrites are the afferent components of neurons and frequently are arranged around the neuronal soma in stellate fashion. In some neurons, they may arise from a single trunk, from which they branch into a dendritic tree. Unlike axons, they generally lack neurofilaments, although they may contain fragments of Nissl substance; however, large branches of dendrites in close proximity to neurons may contain small bundles of neurofilaments. Some difficulty may be encountered in distinguishing small unmyelinated axons or terminal segments of axons from small dendrites. In the absence of synaptic data, they often can be assessed by the content of neurofilaments. The synaptic regions of dendrites occur either along the main stems (Fig. 1-8) or at small protuberances known as dendritic spines or thorns. Axon terminals abut these structures.

The synapse is a specialized junctional complex by which axons and dendrites emerging from different neurons intercommunicate [12]. This was proposed first by Sherrington in 1897, who also proposed the term synapse. The existence of synapses was immediately demonstrable by EM and can be recognized today in a dynamic fashion by Nomarski and confocal optics, light microscopy and scanning EM. With the development of neurochemical approaches to neurobiology, an understanding of synaptic form and function becomes of fundamental importance. As was noted in the first ultrastructural study on synapses (Palade and Palay in 1954, quoted in [4]), synapses display interface specialization and frequently are polarized or asymmetrical. The asymmetry is due to the unequal distribution of electron-dense material, or thickening, applied to the apposing membranes of the junctional complex and the heavier accumulation of organelles within the presynaptic component. The closely applied membranes constituting the synaptic site are overlaid on the presynaptic and postsynaptic aspects by an electron-dense material similar to that seen in desmosomes and separated by a gap or cleft of 15 to 20 nm. The presynaptic component usually contains a collection of clear, 40- to 50-nm synaptic vesicles. These synaptic vesicles are important in packaging, transport and release of neurotransmitters and after their discharge into the synaptic cleft, they are recycled with the axon terminal [6,13]. Also present are small mitochondria approximately 0.2 to 0.5 μm in diameter (Figs. 1-8–1-10). Occasionally, 24-nm microtubules, coated vesicles and cisternae of the smooth ER are found in this region. On the postsynaptic side is a density referred to as the subsynaptic web, but apart from an infrequent, closely applied packet of smooth ER or subsurface cisternae belonging to the hypolemmal system, there are no aggregations of organelles in the dendrite. At the neuromuscular junction, the morphological organization is somewhat different. Here, the axon terminal is greatly enlarged and ensheathed by Schwann cells; the postsynaptic or sarcolemmal membrane displays less density and is infolded extensively.

Before elaborating further on synaptic diversity, it might be helpful to outline briefly other ways in which synapses have been classified in the past. Using the light microscope, Ramón y Cajal (see [14]) was able to identify 11 distinct groups of synapses. Today, most neuroanatomists apply a more fundamental classification schema to synapses, depending on the profiles between which the synapse is formed, such as axodendritic, axosomatic, axoaxonic, dendrodendritic, somatosomatic and somatodendritic synapses. Unfortunately, such a list disregards whether the transmission is chemical or electrical, and in the case of chemical synapses, this classification does not address the neurotransmitter involved.

In terms of physiological typing, three groups of synapses are recognized: excitatory, inhibitory and modulatory. Some neuroanatomical studies [14] have claimed that excitatory synapses possess spherical synaptic vesicles, whereas inhibitory synapses contain a predominance of flattened vesicles (Fig. 1-8). Other studies [15] have correlated this synaptic vesicular diversity with physiological data. In his study on the cerebellum, Gray [15] showed that neurons, with a known predominance of excitatory input on dendrites and an inhibitory input on the cell body, possessed two corresponding types of synapses; however, although this interpretation fits well in some loci of the CNS, it does not hold true for all regions. Furthermore, some workers consider that the differences between flat and spherical vesicles may reflect an artifact of aldehyde fixation or a difference in physiological state at the time of sampling. In light of these criticisms, it is clear that confirmation of the correlation between flattened vesicles and inhibitory synapses is required.

Another criterion for the classification of synapses by EM was introduced in 1959 by Gray [15]. Briefly, certain synapses in the cerebral cortex can be grouped into two types, depending on the length of the contact area between synaptic membranes and the amount of postsynaptic thickening. Relationships have been found between type 1 synapses, which have closely apposed membranes over long distances and a large amount of associated postsynaptic thickening, and excitatory axodendritic synapses. Type 2 synapses, which show less close apposition and thickening at the junction, are mainly axosomatic and believed to be inhibitory. This broad grouping has been confirmed in the cerebral cortex by a number of workers, but it does not hold true for all regions of the CNS.

Most of the data from studies on synapses in situ or on synaptosomes have been on cholinergic transmission. There is a vast family of chemical synapses that utilize biogenic amines (see Chap. 12) as neurotransmitter substances. Morphologically, catecholaminergic synapses are similar but possess, in addition to clear vesicles, slightly larger dense-core or granular vesicles of variable dimension (Figs. 1-8 and 1-10). These vesicles were identified first as synaptic vesicles by Grillo and Palay (see Bloom [16]), who segregated classes of granular vesicles based on vesicle and core size, but no relationship was made between granular vesicles and transmitter substances. About the same time, EM autoradiographic techniques were being employed and, using tritiated norepinephrine, Wolfe and co-workers [17] labeled granular vesicles within axonal terminals. Catecholaminergic vesicles generally are classified on a size basis, and not all have dense cores. Another, still unclassified, category of synapses may be the so-called silent synapses observed in CNS tissue both in vitro and in vivo. These synapses are morphologically identical to functional synapses but are physiologically dormant.

Finally, with regard to synaptic type, there is the well-characterized electrical synapse [18], where current can pass from cell to cell across regions of membrane apposition that essentially lack the associated collections of organelles present at the chemical synapse. In the electrical synapse (Fig. 1-11), the unit membranes are closely apposed, and indeed, the outer leaflets sometimes fuse to form a pentalaminar structure; however, in most places, a gap of approximately 20 nm exists, producing a so-called gap junction. Not infrequently, such gap junctions are separated by desmosome-like regions [3]. Sometimes, electrical synapses exist at terminals that also display typical chemical synapses; in such cases, the structure is referred to as a mixed synapse. The comparative morphology of electrical and chemical synapses has been reviewed by Pappas and Waxman [18].

Molecular markers can be used to identify neurons

Characterization of the vast array of neuron-specific cytoskeletal elements, such as intermediate filaments, microtubules and their associated proteins [19,20], and the neurotransmitters and their receptors [6,21] has led to the development of correspondingly large numbers of molecular and immunological probes, which now are applied routinely in neuroanatomical analyses. The neuron is incapable of participating in T-cell interactions via the expression of major histocompatibility complex (MHC) antigens or the production of soluble mediators, such as cytokines. However, neurons do possess unique proteins, some of which are antigenic, that normally are sequestered by the BBB from the circulating immune system. This theoretically renders the CNS vulnerable to immune-mediated damage should the BBB be breached.