Neuroanatomy deals with the structure of the nervous system. All nervous systems consist of astonishingly similar elements, the nerve cells or neurons. Despite this fact, nervous systems of different animal classes can be organized in strikingly different ways, and in individual brains different anatomical structures can be made out that are obviously related to different functions. In some of these brain parts, one can easily draw conclusions from their particular structure onto the particular kind of information processing in them.
There are two major subdivisions in the vertebrate nervous system: the central nervous system, consisting of the brain and spinal cord, and the peripheral nervous system which connects the central nervous system via nerves to the sensory receptors and the effectors (muscles, glands). In humans, 31 pairs of spinal nerves emanate from the spinal cord, providing the extremities and the trunk with sensory and motor nerve fibers. The head region is provided by 12 pairs of nerves emanating from the brain. One of them, the vagus nerve, also descends to the trunk and innervates inner organs, together with fibers coming from the spinal cord. The two most anterior ‘nerves’, the olfactory and the optic tract, carry already pre-processed sensory information from the olfactory bulb and the eye, respectively, and are considered not as part of the peripheral nervous system but as a part of the brain (central tracts) itself.
A functional distinction, involving both the central and the peripheral nervous system, can be made between the so-called somatic and the vegetative or visceral nervous system. The first deals with the interaction of the animal with the external world, while the vegetative nervous system is involved in the regulation of the body organs (homeostasis of the internal milieu).
Based on embryological criteria the vertebrate brain has been subdivided into five main regions: telencephalon, diencephalon, mesencephalon, metencephalon and myelencephalon (Fig. 1).
Kinds of nervous systems
The first nervous systems can be found in Coelenterates where they form loose nets over extended regions of the body. In the course of evolution an increasing tendency of nerve cells to conglomerate into lumps of nervous tissue can be found, often clearly associated with special tasks in the neural computation underlying behavior. The resulting differentiation of a central nervous system is strikingly different in different groups of animals (e.g. vertebrates, cephalopods, arthropods). In each group, however, a general scheme can be recognized in every single species, suggesting common principles in the control of behavior.
Loose nerve nets are also found around the inner organs of vertebrates as part of the vegetative nervous system.
Nerve cells (Figs. 2 and 3) consist of a cell body (soma) and two kinds of cell processes: dendrites (usually more than one) and one axon (some axonless cells do, however, exist). Both dendrites and axons may have many ramifications. The dendrites are the recipients of signals which they convey towards the cell body in the form of a change in the electrical polarization of the cell membrane. If this change is a depolarization strong enough to reach a certain electrical threshold at the place where the axon leaves the cell body, a new kind of signal arises, the action potential which is propagated along the axon into all its ramifications.
The signal on the axon, unlike that on the dendrites, is conducted with undiminished strength. The axon can therefore be very long and connect distant parts of the nervous system with each other or reach distant muscles anywhere in the body. Long axons are often surrounded by a myelin sheath which consists of several layers of glial cell membrane wrapped around the axon. This prevents ions from crossing the axonal membrane and, by virtue of a "saltatory" conduction from one so-called node of Ranvier(short interruption of the myelin sheath) to the next, makes signal conduction faster than in a ‘naked’ axon.
There are a number of interesting differences between neurons in the central nervous system and sensory neurons of the peripheral nervous system. One example is the neurons which conduct signals from the skin to the spinal cord. Their cell bodies are located close to the spinal cord in the so-called spinal ganglia, i.e. far away from the skin. In these neurons, the dendritic cell process which leads sensory signals in the direction of the cell body has axonal properties: it is very long, conducts action potentials and can be surrounded by a myelin sheath. The neutral term ‘nerve fiber’ refers to both these sensory cell processes and the axons of motor neurons. The two kinds of processes often run together in the same peripheral nerve.
Different neurons form synapses with each other, structural specializations of the membrane which can be seen under the electron microscope (Fig.4) and at which the signals are transmitted from the axon of one neuron, the ‘presynaptic’ neuron, onto a dendrite or cell body of another one, the ‘postsynaptic’ neuron. (More rarely one finds synapses between axon and axon or dendrite and dendrite). Synaptic transmission usually involves a chemical step. The transmitter substance required is contained in the vesicles visible on the presynaptic side (Fig.4).
The location of synapses on dendrites and axons can sometimes already be recognized in the light microscope. In some types of neurons synapses are located on excrescences of the dendritic membrane (dendritic spines; Fig.2). Also, on axonal ramifications, small swellings (synaptic boutons) may indicate the presence of synapses (Fig.3). In the central nervous system, individual neurons often carry many synapses (sometimes tens of thousands), distributed along their dendrites (input synapses) and axonal ramifications (output synapses). Thus, the ramification patterns of neurons are powerful indicators of connectivity patterns. They show where a neuron can receive signals and to where it can distribute them. No synapses are found where an axon is myelinated.
Nervous tissue also contains various kinds of Glia cells. These also consist of a cell body from which cell processes emanate, but they are usually much smaller than neurons (Fig. 5). Glia cells are loosely interspersed between the nerve cells and perform supportive functions. In the central nervous system, the main types are Astroglia, Oligodendroglia and Microglia. The processes of Astroglia cells fill interstices between nerve cell processes and surround blood vessels. They contribute to the formation of the blood-brain barrier, supply nerve cells with nutritive substances and are involved in keeping up the ionic balance in the tissue. Microglia is involved in the repair of tissue damage and seems to be rare in healthy tissue. In the central nervous system, oligodendroglia forms the myelin sheath around axons. In the peripheral nervous system, the myelin sheath is part of so-called Schwann-cells.
Within the central nervous system of vertebrates, one can even distinguish regions of different shading with the naked eye on unstained cuts through the brain. The so-called grey matter contains the cell bodies, dendrites and axonal ramifications of neurons and the synapses between them. It is the place where neurons interact. The white matter is a cable system, mediating the interactions between various parts of the grey matter. It contains axons, but no other neuronal structures. Many of these axons are myelinated, the fatty myelin being responsible for the light color of the ‘white’ matter, in contrast to the ‘grey’ matter which is much poorer in myelin.
With the aid of histological staining methods a more detailed classification of brain structures can be made.
The cytoarchitectonic parcellation of the brain is based on the staining of cell bodies with the aid of basic dyes (Nissl-stain; Fig.6). These stains show local variations in the density and size of cell bodies, thus enabling the delineation of layers or groups of cells, so-called ‘nuclei’, in the various parts of the brain.
Similar maps, largely coincident with the cytoarchitectonic ones can be delineated on preparations stained for myelin (myeloarchitectonics). They show regions of white matter and fiber tracts (see Figure 3 in the article Brain), well distinguished by the arrangement, density and orientation of myelinated axons. Myelin stains also show interesting patterns of myelinated fibers within the grey matter, such as in the different areas of the cerebral cortex.
Another method which reveals some striking local differences of architectonics is the staining of blood vessels (angioarchitectonics). The main difference is found between grey and white matter, the grey matter being more richly vascularized. A few particularly densely vascularized nuclei can be found in the hypothalamus (Fig. 7).
In addition to these traditional histological methods, a variety of histochemical and immunohistochemical techniques are available today, showing the distribution of particular molecules in the brain (chemoarchitectonics). This enables us to investigate the distribution of subtypes of neurons or of particular transmitter substances or of postsynaptic receptor molecules under the microscope (e.g. Amunts et al. 2002). With these methods, we are therefore coming closer to an understanding of the functional mechanisms in different regions of the brain.
A method which is particularly useful for the delineation of cortical areas in the adult human brain is the so-called pigmentoarchitectonics, based on the staining of lipofuscin granules which accumulate in the soma of nerve cells with age (Braak, 1980).
Visualization of individual neurons
Individual neurons with all their cell processes can be isolated from the tissue by the so-called Golgi-method (Fig. 2). With this method, only a small number of neurons are stained, the selection of which seems to be largely random. In the hands of Santiago Ramón y Cajal (1911) and of others, this method has provided us with most of the basic knowledge on connectivity patterns. Later, it became possible to fill individual neurons by intracellular injection of dyes (Fig. 3). This procedure enables us, at the highest level of detail, to visualize neurons, the physiological properties of which had been investigated beforehand.
The cell processes of the neurons form a dense felt, the so-called neuropil, which fills the space between the cell bodies and the blood vessels. In the mouse, the neuropil occupies about 84% of the cortical grey matter (Schüz and Palm, 1989). In the electron microscope one can distinguish the various components: axons, dendrites, dendritic spines, as well as some glia cell processes. Together, they form a compact mass (Fig. 8) leaving very little extracellular space between the individual elements and allowing for a high density of synapses of about 7 x 108/mm3.
Regions in the brain which have been particularly active during life time can be shown post-mortem with the so-called cytochrome oxidase stain. Regions which have been particularly active during the last hour before the death of the animal can be visualized with the deoxyglucose method (Fig. 9). Similar (though not identical) results can be achieved by staining of the protein c-Fos, an immediate early gene product (e.g. Sadananda and Bischof, 2006; Staiger, 2006).
Tracing distant connections
Various methods exist for tracing distant connections in the brain. In the early years of neurohistology these connections were mainly investigated on autopsy material of brains in which localized lesions (due to a stroke or other brain injuries) led to a degeneration of injured fibers. The degenerated pathways could be followed in histological preparations. Nowadays, so-called tracer substances are used which – after injection into the brain of an anesthetized animal - are taken up by the neurons and transported along the axons. Anterograde tracers are taken up by the cell bodies at the injection site and transported towards the terminal fields of their axons; they answer the question: where do the neurons at the injection site project to (Fig. 10)? Retrograde tracers are taken up by the axon terminals and transported towards the cell bodies. They answer the question: where are the neurons located whose axons reach the injection site (Fig.11)?
With the aid of these methods, many of the connections between the various parts of the brain have been revealed, as well as many of the connections between the various areas of the cerebral cortex (e.g. Young et al., 1995).
For more details on tracing and other neuroanatomical methods, see for example Záborsky et al. (2006) and the previous volumes (Heimer and RoBards, 1981; Heimer and Záborsky, 1989).
Since one of the scopes of neuroanatomy is to provide a reasonable basis for functional models, a quantitative assessment of the tissue elements is of importance. One of the difficulties in measuring and counting structures under the microscope is related to the fact that the histological sections are often thinner than the structures under investigation. For example, electron microscopic sections have a thickness of about 60 nm, while synapses (i.e. their disc-shaped membrane specializations) have diameters of around 350 nm. Light microscopic sections have thicknesses of 100 µm and below, while dendritic trees and terminal axonal arbors can ramify over several hundred micrometer.
Quantification of neuronal elements therefore makes assumptions about the extension of structures in the third dimension necessary and/or sophisticated methods of correlation of neighboring sections. The mathematical-geometrical techniques involved in this (see for example Russ and Dehoff, 2000) go by the name of stereology.
In the mouse cortex, the density of neurons is about 9 x 104/mm3 and the number of synapses per cortical neuron is about 8000. The total length of dendrite in one mm3 of cortex is several hundred meters, while the total length of axon in one mm3 of cortical grey matter is several kilometers (Braitenberg and Schüz, 1998).
Some essential aspects of connectivity
Excitation and inhibition
There are two kinds of neurons: excitatory neurons, the kind where a signal in the presynaptic neuron activates the postsynaptic neurons, and inhibitory neurons, in which a signal in the presynaptic neuron diminishes or suppresses the activity of the postsynaptic neurons. By the combination of electrophysiological, histochemical and neuroanatomical methods, much knowledge has accumulated on the anatomical appearance of excitatory and inhibitory neurons and of the two kinds of synapses (Fig. 4). In many parts of the brain it is therefore now possible to investigate the distribution of these two populations under the microscope.
Individual neurons receive usually both excitatory and inhibitory synapses. These may have a characteristic distribution on the receiving neuron: in the cerebral and the cerebellar cortex, as well as the striatum, excitatory synapses are mainly located on dendritic spines, while inhibitory synapses are mainly located on dendritic shafts and cell bodies.
On the presynaptic side, all the synapses of an individual neuron carry the same transmitter substance (in some neurons, it is a mixture of several substances) and are either excitatory or inhibitory. For an overview of the distribution of GABA (gamma-amino butyric acid), the main inhibitory transmitter in most parts of the brain, see the atlas by Mugnaini and Oertel (1985).
Divergence and convergence
Neurons make usually more than one synapse and many of them make and receive hundreds or thousands of synapses. In some cases, an axon makes all of its synapses onto one or a few postsynaptic neurons (e.g. climbing fibers onto Purkinje cells in the cerebellum). In other cases, an axon may distribute its synapses onto thousands of different neurons (e.g. pyramidal cells in the cerebral cortex). Dendritic trees can differ in the degree of convergence of the signals that they receive: all or many of the synapses on a dendritic tree can come from different neurons, or from only a few. The patterns of ramification may be taken as indicators of the degree of divergence and convergence between neurons (see also article Brain).
Connectivity schemes can differ in the degree to which they are predetermined. Exactly predetermined connectivity between individual neurons can be found in invertebrate nervous systems, such as for example the connectivity between the visual receptors in the compound eye of flies and the first visual ganglion. In large pieces of grey matter, the connectivity is predetermined on a coarser level, such as by a specificity between neuronal types (e.g. chandelier cells in the cerebral cortex), or simply by the statistics of the elements in the neuropil, given by the number and arrangement of neurons of different types and their ramification patterns. Such a coarsely predetermined connectivity may then be later refined by learning.
It is still unclear why some postsynaptic structures are located on dendritic spines while others – on the same neuron – are located directly on the dendrites. There is evidence, however, that the spine synapses in particular are involved in the refinement of connectivity by learning. One indication is the fact that synaptic plasticity has been shown for neurons which are densely studded with spines (pyramidal cells in the cerebral cortex, Purkinje cells in the cerebellum, medium spiny cells in the striatum).
Spatially focused vs. diffuse projections
Neurons can make relatively circumscribed projections to one or a few other points in the brain, or they can make wide axonal ramifications that extend over large regions of the brain. These two patterns go along with an important functional distinction: focused projections are a feature of neurons involved in the information processing proper, while highly diffuse connections are a feature of neural systems which provide the background on which information processing is carried out, related, for instance to emotions or attention. These systems are characterized by particular transmitter substances (e.g. dopamine, noradrenalin, serotonin, acetylcholine), and their cell bodies are mainly located in small nuclei in the brain stem or at the basis of the telencephalon (substantia nigra, locus coeruleus, raphe nuclei, nucleus basalis of Meynert, respectively). For an anatomical survey see for example Nieuwenhuys (1985) and Fallon and Laughlin (1987).
Some examples of connectivity schemes
How parts of the brain can differ in their internal connectivity is sketched in Fig.12 for the four largest parts of the brain. (These sketches show only the quantitatively dominant and/or the most characteristic type of neurons).
The neurons in the cerebral cortex (Fig.12 a) form a rich network among themselves. The number of synapses coming from input fibers are only a small percentage of all synapses in the cortex. The neurons in the striatum (Fig. 12 b) also form a network among themselves. However, one fundamental difference between these two parts of the brain is the fact that most of the neurons in the cerebral cortex, the pyramidal cells, are excitatory, while most of the neurons in the striatum are inhibitory.
In both the thalamus (Fig.12 d) and the cerebellar cortex (Fig.12 c), just as in the cerebral cortex, most of the neurons are excitatory. However, in contrast to the cerebral cortex, the excitatory neurons of the cerebellar cortex do not form a network among themselves, and the same seems to be true for the relay neurons in the thalamus (Steriade et al., 1990). The excitatory neurons of the thalamus relay their activity mainly directly to their target structure, the cerebral cortex, and to a certain extent also to inhibitory neurons within the thalamus, while those in the cerebellar cortex (the granule cells) project exclusively onto the various kinds of inhibitory neurons within the cerebellar cortex. Another outstanding feature of the cerebellar cortex is the particular geometry of its neurons which leads to a one-dimensional spread of excitation in latero-lateral direction only.
For more detailed information on the connectivity of these brain structures and for a functional interpretation see for example Steriade et al. (1990), Braitenberg et al. (1997), Braitenberg and Schüz (1998), Miller (2007).
and further reading
- Amunts K., Schleicher A. and K. Zilles (2002) Architectonic mapping of the human cerebral cortex. In: Cortical Areas: Unity and Diversity, A.Schüz and R. Miller (eds.), Taylor & Francis, London, New York, pp. 29-52
- Angevine J.B. and C.W. Cotman (1981) Principles of Neuroanatomy, Oxford University Press, New York, Oxford
- Braak H. (1980) Architectonics of the Human Telencephalic Cortex, Springer-Verlag, Berlin, Heidelberg
- Braitenberg V., D. Heck and F. Sultan (1997) The detection and generation of sequences as a key to cerebellar function: Experiments and theory. Behavioral and Brain Sciences 20, 229-245
- Braitenberg V. and A. Schüz (1998) Cortex: Statistics and Geometry of Neuronal Connectivity (revised edition of Anatomy of the Cortex – Statistics and Geometry, 1991). Springer, Berlin, Heidelberg, New York
- Fallon J.H. and S.E. Laughlin (1987) Monoamine innervation of cerebral cortex and a theory of the role of monoamines in cerebral cortex. In: Cerebral Cortex, Vol. 6, E.G. Jones and A. Peters (eds), Plenum Press, New York, London, pp.41-127
- Heimer L. and M.J. RoBards, eds. (1981) Neuroanatomical Tract-Tracing Methods. Plenum Press, New York, London
- Heimer L. and L. Záborsky, eds. (1989) Neuroanatomical Tract-Tracing methods 2. Recent Progress. Plenum Press, New York, London
- Miller R. (2007) A Theory of the Basal Ganglia and their Disorders. CRC Press, Boca Raton, Florida.
- Mugnaini E. and W.H. Oertel (1985) An atlas of the distribution of GABAergic neurons and terminals in the rat CNS as revealed by GAD immunohistochemistry. In: Handbook of Chemical Neuroanatomy, Vol. 4, Part I, A. Björklund and T. Höfeld (eds), Elsevier, Amsterdam, New York, Oxford, pp.436-622
- Nieuwenhuys R. (1985) Chemoarchitecture of the Brain, Springer, New York
- Ramón y Cajal S. (1911) Histologie du système nerveux de l’homme & les Vertébrés, french edition (1972), Consejo superior de investigaciones cientificas, Instituto Ramón y Cajal, Madrid
- Russ J.C. and R.T. Dehoff (2000) Practical Stereology. New York, Plenum Press
- Sadananda M. and H.-J. Bischof (2006) C-fos induction in forebrain areas of two different visual pathways during consolidation of sexual imprinting in the zebra finch (Taeniopygia guttata). Behav. Brain Res. 173, 262-267
- Schüz A. and G. Palm (1989) Density of neurons and synapses in the cerebral cortex of the mouse. J. Comp. Neurol. 286, 442-455
- Staiger J. F. (2006) Immediate-early gene expression in the barrel cortex. Somatosens. Mot. Res. 23, 135-146
- Steriade M., Jones E.G., and R.R. Llinás (1990) Thalamic oscillations and signaling. John Wiley and Sons, New York
- Young M.B., Scannell J.W., and G. Burns (1995) The analysis of cortical connectivity. Springer, New York, Berlin, Heidelberg
- Záborsky L., Wouterlood F.G., and J.L. Lanciego (2006) Neuroanatomical Tract-Tracing 3. Molecules, Neurons, and Systems. Springer Science + Business Media, New York
- Valentino Braitenberg (2007) Brain. Scholarpedia, 2(11):2918.
- Julia Berzhanskaya and Giorgio Ascoli (2008) Computational neuroanatomy. Scholarpedia, 3(3):1313.
- Eugene Roberts (2007) Gamma-aminobutyric acid. Scholarpedia, 2(10):3356.
- Robert E. Burke (2008) Spinal cord. Scholarpedia, 3(4):1925.
- S. Murray Sherman (2006) Thalamus. Scholarpedia, 1(9):1583.