Cognitivemap.net

Anatomy 103
halts ongoing behaviours and drives exploration, will have connections (1) to the motor areas generating this pattern and (2) to that part of the hypothalamus driving pituitary output in response to environmental uncertainty.
3.2 The internal structure of the hippocampus
Removal of the posterior and temporal neocortex of an animal such as the rat reveals the large sausage-shaped hippocampus underneath (see Fig. 3).
3.1. Introduction
IN the following two chapters we shall review recent anatomical and physiological
studies of the hippocampus and show that structurally and functionally it is well
suited to act as a cognitive map. As we described earlier, the map system is assumed
to have two basic components: a mapping space and a locative mechanism for
building and changing maps. The mapping space consists of a set of interconnected
places, each place being receptive to a large number of potential input items. The
locative system selects the appropriate place for any particular input and effects
changes in the representation of items in response to environmental change. In
keeping with the arguments of the previous section, this locative system derives its
instructions, at least in part, from the movement-programming system.
Anatomically, the mapping space could be provided by a large matrix of identical neurones structured in such a way that each neurone (or group of neurones) would represent a place in a given environment. At least two inputs to this matrix would be required: (1) an input providing sensory information about the environment; (2) an input providing information that the animal is changing its position in space or changing its receptor surfaces so as to sample inputs from a different part of space. In this chapter we shall discuss the nature of hippocampal anatomy in light of these requirements and shall try to identify (1) the mapping space with the matrix of FIG 3. Drawing of the left rat hippocampus. All other forebrain structures except those at the pyramidal cells in the hippocampus, (2) the input about environmental items with the mid-line have been removed. various pathways from the cortical taxon stores via the entorhinal area, and (3) the locative input with the brain stem-septal-hippocampal pathway.† In this animal the hippocampus occupies a large portion of the forebrain. For The mapping system will need several outputs. Since it subserves place hypotheses descriptive purposes it can be divided into a dorsal portion lying just behind the which can guide an animal's behaviour, it should project to areas capable of driving septum, a posterior portion where it begins to bend ventrally and laterally, and a behaviour. Similarly, the mismatch system, which ventral portion lying in the temporal part of the brain. The part of the hippocampus visible on its dorsal aspect is the hippocampus proper, while the fascia dentata is †We shall concentrate our discussion on the hippocampus as the core of the cognitive map. It is clear, buried inside and on the bottom surface of the sausage. The fimbria is a large fibre however, that this core is supported by several ancillary structures whose functions can be characterized as organizing and channelling the inputs and outputs of the hippocampus. These structures, comprising the medial tract which is visible on the lateral edge of the exposed hippocampus. The dorsal and lateral septum rostral to the hippocampus and the subicular area caudal to it, have important roles in the fornix which runs close to the mid-line beneath the corpus callosum is not shown in functioning of the cognitive map. Though considerable knowledge exists concerning the septal region, detailed Fig. 3. Before we discuss these various features of hippocampal anatomy in greater investigation of the subiculum has only recently begun. Where pertinent to the discussion we shall fill in the details of these ancillary structures. detail, it will be useful to look briefly at the phylogenetic differences in hippocampal structure. 104 Anatomy
Anatomy 105
3.2.1. COMPARATIVE HIPPOCAMPAL ANATOMY Recent developments in comparative anatomy suggest that it is no longer reasonable to expect strict homologies between brain structures in animals occupying different branches of the phylogenetic tree. There is considerable evolution within each phylum, so that one can find sophisticated neural organizations as well as primitive ones in that phylum. For example, in addition to the well-described sharks with simple brains one can find sharks with well-developed and highly differentiated forebrains. Similarly, it is now clear that the comparative anatomy of particular brain areas such as the hippocampus is more complicated than the classical linear evolutionary view suggested. While comparative anatomists have traditionally identified a homologue of the mammalian hippocampus in all vertebrates, including the most primitive, the cyclostome (e.g. hagfish and lampreys: Ariens Kappers, Huber, and Crosby 1936, pp. 1248-1255, Heier 1948, Crosby, Dejong, and Schneider 1966), there are several points which must be kept in mind: (1) Homologies can be sought in terms of either structural or functional similarity, and these might not necessarily be the same. (2) The same neural elements could be used differently to produce a module with an entirely different function. Thus, pyramidal cells used in one way might serve to construct a cognitive map in one species, while used in a slightly different way could serve a different function in another species. The grid-like structure of the hippocampus could be used as a spatial map in an animal such as the rat, or as an olfactory correlation device in the reptile, or as a semantic map in the human. (3) The same function could be performed in different species using quite different structural modules. These questions about homologies can only be adequately tested through the use of behavioural experiments where the same form of behaviour is tested in several species, in ways appropriate to each species, and through single-cell neurophysiological experiments in freely moving animals; neither of these has been provided in sufficient detail as yet. We think it is reasonable to assume that behaviours such as homing, migration, FIG. 4. Cross-sections through one hemisphere of the forebrains of different animals and man to and territoriality are evidence of cognitive mapping and suggest as a working show the size and location of the hippocampus (or hippocampal homologues) in relation to other neural areas: (a) Petromyzon (after Heier 1948, p. 32), (b) Tegu lizard after Lohman and Mentink hypothesis that species demonstrating these behaviours have a homologue to the 1972, p. 327), (c) opossum (after Loo 1931, p. 49), and (d) human. The cross-sections have been mammalian hippocampus, usually situated in the medial wall of the pallium (see, drawn roughly equal in size and are, therefore, not in scale relative to each other. e.g., Fig. 4(a)). In the reptile a strip of densely staining pyramidal-shaped cells lines the medial and dorso-medial surface of the hemisphere (the so-called archipallium or associated with dramatic shifts in the location of the hippocampus and pyriform mediodorsal cortex, see Fig. 4(b)). Between the archipallium on the medial surface and the paleopallium, or pyriform area, on the far lateral surface lies the neopallium The reptilian hippocampus can be divided into a small-celled sector on the or dorsal cortex, the presumed homologue of the neocortex of higher animals. The dorso-medial surface and a large-celled sector more dorsally located. While these massive expansion of this neocortex in mammals is have traditionally been considered to be the homologues of the fascia dentata and the hippocampus proper, respectively, this conclusion must be treated with caution in view of recent work on the connections of 106 Anatomy
Anatomy 107
these two areas with each other and with the rest of the brain. For example, the comment on the possible functional significance of these shifts later. small-celled sector, unlike the mammalian fascia dentata, projects not only to the While increasing in absolute size, the hippocampus decreases in relative size with large-celled sector but also directly to the septum and hypothalamus (Lohman and phylogeny. This is entirely due to the expansion of the neocortex, which is more Mentink 1972, Lohman and Van Woerden-Verkley 1976). We shall discuss than 200 times larger in humans than it is in the basal insectivores. Thus, the mammalian septo-hippocampal connections in greater detail later in this chapter. It is neocortex represents 80 per cent of the forebrain in humans, the hippocampus less worth noting here that the sizes of the hippocampus and the septum are highly than 1 per cent; in the basal insectivores the figures are 13 and 8.5 per cent, correlated among the insectivores and primates (Andy and Stephan 1966). respectively (Stephan 1966). In acallosal mammals such as marsupials the hippocampus is folded over upon itself and protrudes into the lateral ventricles (Fig. 4(c)). The two cell layers of the
hippocampus proper and the fascia dentata are bent into two intersecting U's. This 3.2.2. INTERNAL STRUCTURE OF THE MAMMALIAN HIPPOCAMPUS
infolded cross-section is characteristic of all higher animals. In the opossum the The hippocampus in mammals is the paradigm of simple cortex, consisting
posterior tail of the hippocampus, along with the expanding neocortex, moves into a primarily of one basic cell type and its associated interneurones. These basic
temporal position. With further development of the neocortex the rostral portion of neurones are packed together in one layer of a three-layered structure, in contrast to
the hippocampus is invaded by fibres of the corpus callosum and progressively the six layers of neocortex. The transition from the complex structure of entorhinal
decreases in size until, in humans, only a rudiment remains (Fig. 4(d)).
neocortex to hippocampal archicortex can be seen in a horizontal section through The absolute size of the hippocampus increases steadily with phylogenetic the posterior arch of the hippocampus (Fig. 5; facing p. 108). On the basis of development. Stephan (1966) has shown that the hippocampus in man is twice as cytoarchitectonics, Lorente de No (1934) divided the interposed subicular area into large as that seen in monkeys, where it, in turn, is somewhat more than twice as large parasubiculum, presubiculum, subiculum, and prosubiculum (see Fig. 5). Little is as that seen in basal insectivores (e.g. shrews and hedgehogs). Similar conclusions yet known of the function of these four or five layered cortical areas, partly because can be drawn from a study of fibre numbers in the ventral hippocampal commissure. of the numerous fibres of passage which are invariably interrupted by lesions, Andersen (1960c) reported a steady increase in the number of fibres as one goes from vitiating the use of traditional anatomical and behavioural studies. Fortunately, the rat (365,000), through the rabbit (625,700), to the cat (1,191,000). As we have recent anatomical advances involving the use of techniques such as autoradiography seen, the sizes of hippocampus and septum are correlated, and a study of fibre and enzyme transport combined with histofluorescence have begun to reveal the numbers in the post-commissural fornix, an efferent path of the subiculum, indicates connections of these, and other, areas in previously undreamed of detail. Further a similar increase in this ancillary system (Powell, Guillery, and Cowan 1957). There information about the architectonics of these areas can be obtained from Lorente de are five times as many fibres in this tract in man as there are in monkey, cat, and No (1934) and Blackstad (1956). rabbit; these animals, in turn, have five times as many fibres as does the rat. The hippocampus itself is divided into two major U-shaped interlocking sectors, This vast increase in the size of the hippocampus is not uniform through its the fascia dentata (area dentata, dentate gyrus) and the hippocampus proper (cornu dorsal-ventral extent, nor with respect to its internal layering. As we have noted, that ammonis). The fascia dentata has an internal or buried blade and an external or portion of the hippocampus situated most dorsally is pushed aside by the developing exposed blade (Fig. 6).† The hippocampus proper can be further divided. On the neocortex and corpus callosum, leaving only the hippocampal rudiment located basis of differences in cell morphology and fibre projections Cajal (1911), and more dorsally. This sequence does not mean that the human hippocampus is analogous recently Blackstad (1956), identify a regio superior and a regio inferior. The regio primarily to the ventral portions of, for instance, the rat hippocampus. In fact, the superior abuts on the prosubiculum and contains a double row of medium-sized opposite is more likely the case. The dorsal hippocampus contains a greater proportion of the hippocampal subfield known as regio superior than of the subfield known as regio inferior, while the opposite is the case in the more ventral aspects of † The nomenclature used to refer to the two blades of the fascia dentata is extremely confusing, differing from one author to the next. Relative spatial terms such as inner/outer, internal/external, and endal/ectal are the hippocampus. Stephan (1975, p. 511) has noted that, compared with basal potentially ambiguous since they identify different blades depending on whether the whole brain or the insectivores, there is a six-fold increase in regio superior with only a hippocampus is taken as the reference framework. On the other hand, terms such as medial/lateral and two-and-a-half-fold increase in regio inferior in man. We shall dorsal/ventral are only appropriate over limited parts of the structure. We have adopted a suggestion of P. Andersen (private conversation) and called the blade internal to the hippocampus the buried blade and the blade on the external surface of the hippocampus the exposed blade (see Fig. 6). 108 Anatomy
Anatomy 109
Areas of the hippocampal formation (1908) (1909) (1926) Hilus of the dentate Transition blade Area retrosplenialis e FIG. 6. Schematic diagram of the intra-hippocampal connections. Horizontal section through the right hippocampus as in Fig. 5 except that caudal is to the right. Entorhinal fields Medial entorhinal Entorhinal field A Lateral entorhinal pyramidal cells whose main apical dendrite gives off only small side branches and does not divide for several hundred microns (Fig. 6). The regio inferior is the semi-circle close to the fascia dentata containing the giant pyramidal cells whose apical dendrites bifurcate shortly after leaving the soma (Fig. 6). As can be seen in the figure the part of the dendrite close to the cell body has thorn-like spines, and these, as we shall see, receive contacts from the mossy fibres of the dentate granule From Chronister and White 1975, p. 21 cells. Lorente de No (1934) divided the hippocampus proper into four fields, CA1-4, CA standing for cornu ammonis. He called the regio superior CA1 and subdivided and send their axons into the fimbria. However, unlike the CA3 pyramids, they do the regio inferior into CA2 and CA3;† CA4 designated the scattered cells inside the not receive an input from the basket cells, a group of interneurones we discuss hilus of the fascia dentata. Although these cells are not lined up like the pyramidal cells of CA3, they were included in the hippocampus proper because they have The CA2 field, according to Lorente de No, consisted of CA3-type pyramids which pyramid-like characteristics, receive the mossy fibres, did not receive the mossy fibres of the dentate granule cells. The existence of a separate area, defined as such, has been challenged by Blackstad and his colleagues †CA1 is sometimes called Sommers sector in the pathological literature; it appears to be particularly (e.g. Blackstad 1956, 1963), who can find no difference between CA2 and CA3 cells susceptible to anoxia and/or vascular disturbance and is thus often selectively destroyed in diseases such as in terms of their connections or histo-chemical staining properties. Table 2 idiopathic epilepsy and various types of poisoning (Greenfield et al. 1961, Meldrum and Brierley 1972). reproduces Chronister and White's (1975) comparison between Lorente de No's nomenclature and that of Rose and the now seldom used one of Doinikow. They have also added Blackstad and Brodmann to the table. Since both the nomenclature FIG. 8. Prenatal development of the rabbit hippocampus. Cross-sections through the hippocampus at embryonic stages (a) 15 mm, (b) 20 mm, (c) 29 mm, (d) 41 mm, (e) 60 mm, and (f) 90 mm. The insets show the outline of one hemisphere, and the shaded portion the position of the embryonic hippocampus. (Reconstructed from Stensaas 1967a, b, c, d, e, 1968.) FIG. 5. Photograph of a horizontal section of the right side of the adult mouse brain as seen from above. The section passes through the posterior arch of the hippocampus and entorhinal cortex (R.e.). Caudal is up. Note the gradual transition from the six-layered entorhinal cortex through the parasubiculum (Par), p resubiculum (Pres), subiculum (Sub). prosubiculum, (Pros.) to the three-layered cortex of the hippocampus proper (CA1-CA4) and the fascia dentata (F.D.). Lorente de No further subdivided some of these areas (a, b, c). Fi is the fimbria. Nissl stain. (Modified from Lorente de No 1934.) 110 Anatomy
Anatomy 111
of Lorente de No and that of Cajal/Blackstad are used widely at present we shall refer fascia dentata. Examples of these are seen in Fig. 7. As shown in this figure the to both, with a slight preference for Lorente de No. pyramidal cells have both apical and basal dendrites, while the granule cells have With the exception of CA4, the basic pattern in all sectors of the hippocampus is only apical dendrites. the same: an ordered sheet of large neurones whose cell bodies are all packed An example of the most important type of interneurone in the hippocampus, the together in one layer and whose dendrites all run off in the same direction, extending basket cell of Cajal, is also shown in Figs. 6 and 7. Its cell body is situated either in, for many hundreds of microns. Many of the inputs to each sector traverse the or slightly below, the principal cell layer and its axon ascends through the cell-body dendrites at roughly right angles, making synapses en passage within a restricted layer, travelling a considerable distance. The axon moves orthogonal to the region of the dendritic fields of many neurones in turn. The large neurones are the dendrites of the pyramidal or granule cells, giving off numerous descending giant pyramids of CA3, the smaller pyramids of CA1, and the granule cells of the collaterals which end in basket-like plexuses around the cell bodies. The basket cells differ from the pyramidal and granule cells in a number of ways, and this is seen most clearly in CA3. In contrast to the pyramids, the basket cells do not have the characteristic thorns on their apical dendrites and thus are not in contact with the mossy fibre input. Further, the basket cells do not send their axons out of this area, while the pyramidal cells do. Other interneurones (e.g. the stellate and fusiform cells of Cajal) send their axons into the apical dendritic layer to make contact with the distal dendrites of the pyramids and granules. As we shall see in a subsequent section, there is physiological evidence that at least some of the interneurones exert a widespread inhibitory control over the pyramidal and granule cells. 3.2.3. INTERNAL LAYERING 3.2.3(a). Fascia dentata. The fascia dentata has three layers (see Fig. 7): (1) the granule layer containing the densely packed cell bodies of the granule cells; (2) the molecular layer formed by the intertwining apical dendrites of the granule cells and their afferents; (3) the polymorph layer in the hilus of the fascia dentata which merges with the CA4 field and contains the initial segments of the granule-cell axons as they gather together to form the mossy fibre bundle. It also contains a few types of non-granule cells, the most important of which is the basket cell discussed above. 3.2.3(b). Hippocampus proper. Although it is basically a three-layered structure like the fascia dentata, the hippocampus proper has been divided into as many as seven layers, each defined by a particular feature of the large pyramidal cells or their afferents (see Fig. 7). Starting from the ventricular surface these layers are as follows: (1) The alveus, containing the axons of the pyramidal cells, directed towards the fimbria or the subiculum (some afferents to the hippocampus also travel in the alveus); (2) the stratum oriens, a layer between the alveus and the pyramidal cell bodies which contains the basal dendrites of the pyramidal cells and some of the basket cells, as well as afferents from the septum; (3) the stratum pyramidale, or pyramidal layer, which is dominated by the cell bodies of the pyramids; (4) the stratum radiatum and (5) the stratum lacunosum/moleculare FIG.7(a). Examples of CA1 and CA3 pyramidal cells. (After Cajal 1911, Fig. 475 ).(b) Examples of dentate granule cells and a basket cell of Cajal. (After Lorente de No 1934, Fig. 10.) Anatomy 113
which are, respectively, the proximal and distal segments of the apical dendritic tree immature neurones migrate towards the pial surface where they congregate in the (the stratum radiatum accounts for 70 per cent of the tree in the rabbit and 42 per cent cortical plate. As in all cortical tissue, with the exception of the fascia dentata, the in the guinea pig, according to Hjorth-Simonsen (1973)). There are some scattered formation of the cortical plate takes place in an orderly inside/out fashion; newly cells throughout these layers, making contact with various parts of the pyramidal cell formed neurones of the hippocampus proper push up through the older neurones of dendrite. The dominant feature conveying lamination to these layers, however, the cortical plate and come to rest at the edge facing the pial surface. Stensaas comes from other afferents to the dendrites which typically stream past at right (1967c) has suggested that the migrating neuroblasts encounter a condition at this angles, making numerous synapses en passage as described above. These inputs will level which discourages movement and promotes the growth of dendrites. be discussed in greater detail later. In the CA3 field an additional layer is recognized: As development proceeds, the number of neurones in the cortical plate rapidly the stratum lucidum, interposed between the pyramidal cell bodies and the stratum increases and the hippocampus begins to take on its characteristic enfolded shape radiatum, receiving the mossy-fibre input from the dentate granule cells.† (Fig. 8d, e). During this period the future fascia dentata is represented by a pool of germinal cells gathered above the end of the cortical plate towards the fimbria. The development of the fascia dentata lags well behind that of the rest of the 3.3. Ontogenetic Development of the Hippocampus
hippocampus, and indeed behind the rest of the brain in general, continuing postnatally in some species. 3.3.1. GROSS MORPHOLOGY A brief look at the ontogenetic development of the mammalian hippocampus will 3.3.2. POSTNATAL DEVELOPMENT OF FASCIA DENTATA GRANULE CELLS show how it comes to assume its striking, highly laminated, double-horseshoe shape. The postnatal development of the rabbit brain has not been studied anatomically,† Fig.8 (facing p.109) illustrates six stages in the development of the rabbit but in other altricial‡ animals (e.g. rat and mouse) the fascia dentata does not reach hippocampus, compiled from the papers of Stensaas (1967 a,b,c,d,e, 1968). The its final adult form until some time after birth. Both the differentiation and details of hippocampal structure are all drawn to the same scale. Beginning its life as migration of cells have been studied in great detail in the rat using the techniques of a narrow strip of cortical tissue trapped along the medial wall of the hemisphere, the autoradiographic labelling and low-level X-irradiation. The former technique is hippocampus finds nowhere to go as it expands except into the ventricle. It first based upon the fact that tritiated thymidine will only be incorporated into the DNA pushes laterally and then doubles back underneath itself. It has been suggested that of cells which are formed shortly after injection (i.e. during mitosis). Thus, cells in the U-shape of the adult fascia dentata is due to the pressure of the hippocampus the final mitotic stage at the time of injection will be most heavily labelled, while proper invaginating into the granule-cell layer during its development (Bayer and those dividing again subsequently will contain less labelled DNA. Low-level Altman 1974). X-irradiation selectively kills cells which are dividing but has no effect on mature In the earliest stages (Fig. 8 (a)) the cell bodies of the precursors of both neurones cells. The number and distribution of labelled or pyknotic (killed) cells at different (neuroblasts) and astrocytes (spongioblasts) are concentrated in a layer adjacent to times during development can be used to construct a picture of the generation and the ventricle called the matrix lamina. Here, new neuroblasts are continually formed migration of neurones. Using these techniques Angevine (1965) and Altman and his by cell division. After their final division the colleagues (Altman and Das 1965, Altman 1966, Bayer and Altman 1975, Hine and Das 1974) have shown that the vast majority of the pyramids of the hippocampus proper are formed prenatally in the mouse and rat. On the other hand approximately † In addition to histological evidence for lamination, there is also evidence from histochemical studies of 85 per cent of the dentate granules originate after birth, as do a fair proportion of the putative transmitter systems (e.g. Storm-Mathisen and Blackstad 1964, Storm-Mathisen 1976) and oxidative interneurones in all parts of the hippocampus. Of the 85 per cent of granule cells enzyme systems (Mellgren 1973, Mellgren and Blackstad 1967). For instance, Storm-Mathisen (1972) has formed postnatally 25 per cent originate during the first four days after birth and 20 shown that gamma-aminobutyric acid (GABA), as indicated by the presence of glutamic acid decarboxylase (GAD), is primarily distributed in two layers of the hippocampus, the stratum pyramidale and the stratum lacunosum/moleculare. This putative inhibitory neurotransmitter is thus found in the two layers thought to contain inhibitory interneurones, the basket cells in the former layer and the stellate and fusiform cells in the †A recent electrophysiological study has investigated the postnatal development of the commissural latter (cf. Cajal 1911 Figs 473, 474) Other work indicates that inputs to the hippocampus involving both connections in the rabbit (Thomson 1975). nor-adrenergic fibres (Swanson and Hartman 1975) and serotonergic fibres (Moore and Halaris 1975) also ‡Altricial animals are confined to a nest and nursed for some time postnatally. terminate in these layers and that they have inhibitory effects as well. We shall discuss this at length later. Anatomy 115
per cent more in the next four days (Bayer and Altman 1975). The formation of is that a lesion of the commissural system in neonatal rats does not result in synapses between the perforant path axons from the entorhinal cortex and the perforant path fibres occupying sites closer to the granule cell body dendrites of the granule cells takes place in step with cell formation; the biggest (Hjorth-Simonsen and Jeune 1972, Lynch, Deadwyler, and Cotman 1973); they surge in synaptic formation occurs in the period between postnatal days 4 and 11 remain restricted to the outer portions of the dendrite. Interestingly enough, when the number of synapses in the exposed blade doubles every day and the however, a neonatal lesion of the perforant path does result in an encroachment of synaptic density increases 20 times. The total number of synapses in the exposed the commissural fibres into the upper zone normally reserved for the perforant path blade increases 1000-2000 times from the fourth postnatal day to adulthood, while fibres (Lynch, Stanfield, and Cotman 1973, Zimmer 1973). The second series of the buried blade, which develops earlier, increases 300 times (Crain et al. 1973). studies which casts doubt on the Gottlieb-Cowan principle is that done on the Electron microscopic studies of these synapses in the adult rat show that they have synaptic connections within the hippocampus of the mutant 'reefer' mouse (Bliss, very complex geometrical configurations. This adult level of complexity is not Chung, and Stirling 1974, Bliss and Chung 1974). Because of a defective gene for attained until day 25 after birth (Cotman, Taylor, and Lynch 1973); whether this has the control of migrating neuroblasts, these animals end up with a considerably any functional significance is not clear. disorganized hippocampus.† The granule-cell bodies do not line up in an orderly The development of the fascia dentata is organized according to several spatial fashion but instead are scattered haphazardly throughout the whole region of the gradients. As we have already noted, the buried blade develops before the exposed area dentate. Contact between the incoming perforant path axons and the blade. In addition, the ventral fascia dentata develops before the dorsal part. Finally, granule-cell dendrites is achieved by the axons maintaining their normal positions the accretion of new cells to the cell-body layer follows an outside/in gradient. relative to the pial surface and the aberrant granule cells extending their dendrites up Newly formed neuroblasts migrate to the cell layer and nestle in underneath the to meet them. The synaptic contact, then, is a result of the dendrite seeking the axon. existing granules (Angevine 1965, Altman and Das 1965, Schlessinger, Cowan, and An entirely different principle is revealed by the synaptic contacts formed between Gottlieb 1975), in contrast to other cortical areas where the new immigrants push up the Schaffer collaterals (from CA3 pyramids) and the CA1 dendrites. Here, it is the through the older residents and settle at the top of the cortical plate. It was originally axon which alters its course to meet preferred parts of the dendrite; the two layers of thought that the germinal pool from which the granule cells were formed was located cell bodies (more distinct than in the normal mouse) are mirrored by two layers of in the juxtaventricular matrix lamina near the fimbrial end of the hippocampus Schaffer collateral synapses. Electrophysiological experiments by Bliss and his (Altman and Das 1965). Neuroblasts formed here would then have to migrate colleagues show that in both systems functional synaptic contacts are established. through the alveus and pyramidal layer of CA3 to reach the fascia dentata. More However, Zimmer (1974) has strengthened the case for the 'spatial-proximity' recent studies (Schlessinger et al. 1975, Bayer and Altman 1975) indicate that a notion by demonstrating that the reason for the failure of the perforant path fibres to germinal pool of precursor cells is actually set up in the anlage of the fascia dentata encroach upon the termination sites of commissural fibres after decommissuration itself. resides in the presence of ipsilateral association fibres which occupy the newly available sites. In agreement with this, Hjorth-Simonsen and Zimmer (1975) and Zimmer and Hjoith-Simonsen (1975) have demonstrated the importance of spatial 3.3.3. PRINCIPLES OF AXO-DENDRITIC CONNECTION proximity in the termination of afferents from the entorhinal area to the fascia It has been suggested that these gradients of development might explain the dentata both in rabbit and rat. These more recent studies give new currency to the laminated structure of the adult hippocampus (Gottlieb and Cowan 1972). The site of Gottlieb-Cowan principle. contact between a developing axon and a dendrite would be determined by the sequence in which the afferents present themselves and the availability of appropriate 3.3.4. DEVELOPMENT OF THETA receptor sites on the dendrites. This would explain, for example, the fact that the Another approach to the development of the hippocampus is to study the times of commissural afferents from the opposite hippocampus occupy the portion of the onset of the various patterns of the adult EEG. Vanderwolf et al. (1975) have dendrite closest to the soma on the basis that they arrive before the entorhinal studied the onset of theta activity in the rabbit, rat, and guinea afferents. These latter latecomers would be obliged to synapse on the outer segments of the dendrite. † Other structures such as the cerebellum and neocortex are equally disorganized, but these do not concern There are two pieces of evidence which might be taken to argue against a first- us here. come, first-served principle as the only one in operation. The first Anatomy 117
pig.† At birth the EEG's of altricial animals such as the rat and rabbit do not show any (d) extrinsic afferents from outside the hippocampus. We shall discuss each of these theta. A theta pattern (related to movements) develops in both species around 12-14 in turn. days postnatally. A second type of theta (not associated with movements) develops around 22-24 days postnatally in the rabbit. In the guinea pig, a species which is 3.4.1. INTRINSIC AFFERENTS FROM THE SAME SECTOR behaviourally mature at birth, both types of theta can be recorded at birth. There are two types of interactions between pyramidal cells within the same CA field: a direct excitatory one and an indirect inhibitory one. It is likely that the 3.3.5. FUNCTIONAL IMPLICATIONS OF DELAYED MATURATION OF HIPPOCAMPUS indirect inhibitory effect is mediated by an interneurone, probably the basket cells Before we leave the subject of the ontogeny of the hippocampus, we should enquire described above. As we have seen, the axons of the basket cells make contact with into the possible functional significance of the postnatal development of various numerous adjacent pyramidal cells; the basket cells in turn get part of their input features of the adult structure. There are at least two possible results of this delayed from these same pyramidal cells. This feedback loop could form the basis for a development. Either (1) the hippocampus is functioning before the conclusion of Renshaw-type inhibitory circuit. The evidence that the basket cells play this role postnatal neurogenesis, but in a manner different from that in the adult, or (2) the comes from several physiological studies. hippocampus does not function at all until it has completely developed.‡ Aside from In the first of these Kandel and Spencer and their colleagues (e.g. Kandel, the postnatal development of fascia dentata cells there is also postnatal development Spencer, and Brinley 1961, Spencer and Kandel 1961c) showed that stimulation of of the cholinergic input systems to the hippocampus, as shown by Mellgren (1973). the fornix was almost invariably followed by a prolonged inhibition of the CA3 This may account for the fact that the two separate theta patterns noted above develop pyramidal cells which was associated with an intracellular hyperpolarization. This at different times. As we shall suggest later, the CA1 region seems responsible for the potential could be reversed by artificially hyperpolarizing the cell with current triggering of exploration, and the time of onset of theta activity from CA1 agrees injected through the recording electrode; this indicates that the potential was a true nicely with the onset of exploration in neonates (Douglas, Peterson, and Douglas inhibitory post-synaptic potential (IPSP). In some cases the IPSP was preceded by 1973). This suggests a possible function of postnatal development; the delayed an excitatory post-synaptic potential (EPSP) or an antidromic spike. By using a maturation of the hippocampus could have the negative effect of delaying the onset of preparation in which the fornix had been sectioned several weeks prior to the acute exploration. During the period when the animal is still dependent upon the mother it experiment, and thus contained no afferent fibres to the hippocampus, they showed would be dysfunctional for it to be continually leaving the nest to explore its that the inhibitory effects could be generated by antidromically activating the environment. Spontaneous alternation, which is a reliable measure of exploration, efferent fibres from the hippocampus. They also concluded that the excitatory develops around day 28 in rats (Douglas et al. 1973), but is already present in the effects were due to afferents to the hippocampus, a conclusion which has been guinea pig during the first week of life. modified by later work from the same laboratory, as we note shortly. On the basis The possibility that the hippocampus functions in a different way in neonates can of latency measurements they decided that the IPSP was mediated by a direct axon only be tested adequately by studying the properties of single neurones, and this has collateral of the output fibres or by, at most, one interposed interneurone. not yet been done. Andersen, Eccles, and Løyning (1964a,b) confirmed these findings and argued that the inhibition was mediated by a Renshaw-type interneurone which they 3.4. Afferents to the hippocampus
identified as the basket cell of Cajal. This was supported by three lines of evidence: The cells of the hippocampus receive afferents from several sources: (a) (1) the finding of a rapid oscillation on the ascending limb of the IPSP which they intrinsic afferents from cells of the same sector; (b) intrinsic afferents from attributed to the action of the interneurone; (2) the fact that the maximum for the other sectors; (c) commissural afferents from the opposite hippocampus; positive inhibitory potential evoked by weak stimulation of various input paths was centred around the stratum pyramidale; (3) their ability to record from cells with † Theta (θ) activity is a characteristic rhythmic electrical activity, involving frequencies in the range of 4-12 high-frequency bursts to the stimulus and no IPSP, which they felt were the Hz in most species. We will discuss the functional significance of this activity later (see pp. 160-90). interneurones in question. This last group of cells was located in the stratum oriens ‡ This position has been espoused by Altman, Brunner, and Bayer (1973) who pointed out the behavioural parallels between infant rats and adult rats with hippocampal damage. A critical review of their presentation of as well as in the stratum pyramidale. The authors concluded that the widespread the literature on the effects of hippocampal lesions in adults is found in Nadel, O'Keefe, and Black (1975). IPSP following antidromic stimulation is mediated by an axon collateral from the pyramidal cell onto the basket cell, which in Anatomy 119
turn has a widespread inhibitory action on many pyramidal cells. Subsequent studies areas termed CA3b and CA3c), but the upper one continues to the border of (Andersen et al. 1969) indicate that the inhibitory effect of one interneurone may CA1.†All along their course the mossy fibres make numerous synapses en passage extend over 1 mm. or more. Although there is no direct evidence for the fascia with the dendrites they pass, first in the CA4 field and then in the CA3 field. They dentata it is likely that a similar mechanism operates there. make contact both with large spines or excrescences, which are a prominent feature In addition to this indirect inhibitory interconnection between cells in the same of the CA4 and CA3 pyramidal cell dendrites (Hamlyn 1961, 1962, Blackstad and sector there is evidence for a direct monosynaptic excitatory connection between Kjærheim 1961, Laatsch and Cowan 1966), and with the dendrites themselves. A pyramidal cells of the CA3 field. Lebovitz, Dichter, and Spencer (1971) used the combined electrophysiological−electron-microscopical study has shown these de-afferented fornix preparation described above to demonstrate this. In addition to synapses to be excitatory Gray's type I (Andersen et al. 1966). These synapses are the expected antidromic activation of many pyramidal cells and the subsequent particularly interesting in that they contain a large amount of zinc. Thus, zinc levels widespread inhibition found by Kandel and Spencer, they also recorded some cells in the hippocampus increase postnatally in conjunction with the development of the which were orthodromically activated with short latencies. Since a small percentage mossy-fibre system (Crawford and Connor 1972), while lesions of the mossy-fibre of these cells (12.5 per cent) could also be antidromically activated by stimulation of bundle cause a rapid loss of zinc from the synapse (Haug et al. 1971). The the fornix, and therefore must have been pyramidal cells, the authors concluded that hippocampus overall has a concentration of zinc which is three times higher than some CA3 pyramidal cells must make monosynaptic contact with each other. A the rest of the brain (Fjerdingstad, Danscher, and Fjerdingstad 1974a,b).‡ It was possible anatomical substrate for this contact is the longitudinal association pathway originally reported that the temporary depletion of zinc through the use of H2S described by Lorente de No (1934). According to Lorente de No, this pathway arises rendered the mossy-fibre synapses inoperable (von Euler 1962, Segal 1972). in the CA2 and CA3 fields, either as a collateral of the pathway known as the However, subsequent experiments which have used non-toxic, specific chelating Schaffer collateral system or as a collateral of the pathway destined for the fimbria, agents such as diphenylthiocarbazone (Crawford, Doller, and Connor 1973) or and connects cells within the same sector. Hjorth-Simonsen (1973) has confirmed the DEDTC (antabuse, Danscher, Shipley, and Andersen 1975) have failed to replicate existence of this pathway using the Fink-Heimer method, but believes that its origin this effect and have concluded that it might be due to the nonspecific toxic effects is restricted to the same fields as the Schaffer collaterals. of the gas. The substance of which zinc is a component has not yet been identified (Blackstad et al. 1970), nor has its role in synaptic transmission, if any, been discovered.∗∗∗ Edström and Mattsson (1975) have shown that zinc stimulates the 3.4.2. INTRINSIC AFFERENTS FROM OTHER SECTORS rapid transport of material in axons by its effect on microtubules, and this might The major interconnections between the three sectors, demonstrated anatomically form the basis for its action in the mossy-fibre system. (Lorente de No 1934, Raisman, Cowan and Powell 1965, Hjorth-Simonsen 1973) The axons of both the giant CA3 pyramids (Fig. 6) and the modified pyramids of and physiologically (Andersen, Blackstad, and Lømo 1966, Fujita and Sakata 1962, CA4 divide, with one branch entering the fimbria and going to the septum, while Gloor, Vera, and Sperti 1963), are primarily unidirectional, starting from the fascia the other branch, or branches, remain within the hippocampus. Three projections dentata, coursing through the CA3 field, and ending in CA1 (see Fig. 6). have been described for these collaterals: The thin (rat, 0.2-0.5 µm (Laatsch and Cowan 1966), 0.1-0.2 µm (Blackstad (1) The Schaffer collateral system arises from the CA4, CA3c, and perhaps 1963); rabbit, 0.3µm (Hamlyn 1962)) slowly conducting (cat, 0.75-1.0 m s-1 (Gloor, some of the CA3b pyramids and runs in the stratum radiatum of CA1, Vera and Sperti 1963)), unmyelinated axons of the dentate cells gather together in the making powerful excitatory synapses en passage polymorph layer in the hilus of the fascia dentata and course out of the hilus as the mossy fibres of Cajal. They run in two separate bundles, one above the pyramids of † Lorente de No believed that the mossy fibres only extended to his CA2 field, but see p. 109 for a CA3 in the stratum lucidum (Fig. 6) and one below the pyramids in the stratum discussion of Blackstad's criticisms of this view. ‡ They also showed that the hippocampus has a 10 times higher concentration of lead that the rest of the oriens (Fig. 6).† The lower bundle stops abruptly within the CA3 field (between sub- brain, but without localizing it within the hippocampus. ∗∗∗ Zinc is a component in a number of enzyme systems, including alcohol dehydrogenase. The possibility † The exact course of these bundles may differ from species to species, and within different strains of the that Korsakoff's psychosis, which often develops after chronic alcohol over-consumption, is in some way same species. Barber et al. (1974) have shown such differences in different strains of mice, including the related to the zinc in the mossy fibres, or elsewhere, has never been directly tested. Zinc is also a component existence of an intrapyramidal bundle running within the pyramidal layer in one strain. in glutamic acid dehydrogenase as well as being a strong inhibitor of glutamic acid decarboxylase, enzymes important in the metabolism of both glutamic acid and GABA (cf. Storm-Mathisen (1976) for discussion and references). Thus, zinc might play some role in regulating the activity of these putative neurotransmitters. Anatomy 121
(Fig. 6) (Lorente de No 1934, Hjorth-Simonsen 1973, Andersen, Blackstad and Lømo 1966); these fibres are also of rather small diameter (0.1-0.2 µm in the rat (Blackstad 1963)).† (2) Collaterals of CA4 and CA3c (perhaps from the same cells which give off the Schaffer collaterals) bend around the buried blade of the fascia dentata and run parallel to the long axis of the hippocampus, making contact with the proximal dendrites of the dentate granule cells (Fig. 6) (Zimmer 1971). (3) The longitudinal association pathway of Lorente de No, which has already been described (p. 118). The CA1 pyramidal cells send their axons out of the hippocampus via the alveus (Fig. 6). This projection will be discussed later as part of the hippocampal efferent system. It is generally agreed that there is no projection from CA1 to CA3 or CA4 (Raisman, Cowan, and Powell 1966, Hjorth-Simonsen 1973). The possibility of a projection from CA1 to the fascia dentata is somewhat more controversial. Raisman et al. (1966) found degeneration in the dentate area following posterior CA1 lesions, but Hjorth-Simonsen (1973) has argued that their lesions encroached upon the perforant path; lesions which do not do so yield no degeneration in the fascia dentata. The question arises as to the topography of the projections within this circuit: do the granule cells in a small area send their axons to an equally restricted area of CA3, or is the projection more diffuse? Blackstad et al. (1970) and Andersen, Bliss, and Skrede (1971) agree that the projections are extremely precise and restricted, each small group of granule cells sending an excitatory (Andersen, Holmqvist, and Voorhoeve 1966a,b) projection to a small group of cells in the next link of the circuit.‡ Further, the entorhinal input to each of these cell groups is similarly restricted (Andersen et al. 1971, Lømo 1971a, Hjorth-Simonsen 1972), as we shall see shortly. On the basis of these and other anatomical considerations Andersen et al. FIG. 9. Lamellar organization of the hippocampus: (a) Lateral view of the rabbit brain with the parietal and temporal neocortex removed to expose the hippocampal formation. The lamellar slice (1971) proposed that the principal organization of the hippocampus is a sandwich of indicated has been presented separately in (b) to show the proposed circuitry. alv, alveus; ento, many of these relatively independent lamellae, which are roughly perpendicular to the entorhinal cortex; fim, fimbria; pp, perforant path; Sch, Schaffer collateral. (From Andersen et al. long axis of the hippocampus (see Fig. 9). These lamellae, of course, are not 1971). completely independent; as we have noted there are fibre bundles connecting cell groups in the longitudinal direction as well. example, Zimmer, in his study of the projection from the CA3-CA4 field to the Within this restricted projection from one field to the next there is an interesting dentate granule cells, found that degeneration in the middle of the lamella filled the shift in the pattern of termination as one moves from one end of the lamella to the whole of the projection area on the proximal one-third of the dendrites but only part other (Zimmer 1971, Hjorth-Simonsen 1973). For of this projection area at the extremes of the lamella. As one moves from the rostral end of the hippocampus in a temporal direction there is a shift in the localization of † These collaterals may account for the curious finding (Daitz and Powell 1954) that, after a lesion in the fibre terminations; rostrally they terminate proximal to the cell body while fornix-fimbria system, afferents to the hippocampus degenerate normally while the efferents on the hippocampal side of the lesion survive intact. No chromatolytic changes are seen in the hippocampal cells giving rise to both temporally they terminate distal to the cell body. There is a similar medio-lateral the efferent output and the Schaffer collaterals. movement such that the most rostral projections are confined to the exposed blade, ‡ Lynch et al. (1973) supported this story with a horseradish peroxidase transport study of the mossy-fibre while those at the temporal end go to the buried blade. A similar imbricated pattern system; with 0.1 µl injections terminals were restricted to within 250-350 µm. of termination has been reported by Hjorth-Simonsen (1973) for the longitudinal association pathway and, to a lesser extent, for the Schaffer collateral pathway. This pattern strongly suggests that each cell, or group of cells, in CA3-CA4 Anatomy 123
projects to different points on the dendritic trees of several granule cells, either inputs from CA3 as well. Gottlieb and Cowan (1973) have pointed out that this through multiple collaterals or through a single axon making multiple synapses en pattern of commissural innervation indicates that passage. We shall suggest later (pp. 225-7) that a mechanism exists for blocking the 'each field of the hippocampus which contributes to the commissural projection also gives receptivity of most of the dendritic tree of these cells at any given time. The rise to an ipsilateral association pathway which follows the same intrahippocampal course, combined operation of these two mechanisms could act as a sorting device such that and terminates within the same region' (p. 420). a given input would only activate a small subset of its target neurones. In other words, CA1 and fascia dentata receive inputs from CA3 on both sides. The 3.4.3. COMMISSURAL AFFERENTS significance of this restricted, highly specific, pattern is not clear at present. Extensive connections exist between the two hippocampi crossing the midline in the ventral and dorsal hippocampal commissures or psalteria. The ventral hippocampal 3.4.4. EXTRINSIC AFFERENTS TO THE HIPPOCAMPUS commissure consists of fibres running in the fimbria which turn back at the rostral The hippocampus receives extrinsic afferents from a variety of other structures, end of the hippocampus to run in the fimbria of the opposite side (Fig. 3). These including primarily the entorhinal cortex, the medial septal area, and several fibres are relatively small (0.6, 0.9, and 1.2 µm for cat, rabbit, and rat, respectively) brain-stem sites. Difficulties in interpreting the exact input patterns arise because and slowly conducting (3.3, 5.6, and 7.5 m s-1) according to Andersen (1960c). The some brain-stem afferents course through the septum and the retrosplenial cortex on dorsal hippocampal commissure forms a thin layer of fibres which run beneath the their way to the hippocampus. Thus, traditional anatomical techniques can lead to splenium of the corpus callosum at the posterior arch of the hippocampus where it misinterpretations based on lesions in fibres of passage. The same, of course, begins to turn downward. applies to the study of hippocampal efferents, as we shall see. More recently Following large lesions of the contralateral hippocampus, or section of the ventral enzyme histochemical and retrograde transport methods have facilitated the analysis hippocampal commissure, degeneration is observed in the stratum oriens of all of distinct input (and output) pathways. None the less, some controversy persists, fields, in the stratum radiatum of CA1 and, to a lesser degree, CA3, and in the particularly regarding hippocampal efferents, as we note later. Here, we shall proximal (juxtagranular) one-third of the stratum moleculare of the fascia dentata discuss afferents under four headings: (a) entorhinal efferents; (b) brain-stem (Blackstad 1956, Raisman et al. 1965, Laatsch and Cowan 1967). A small amount of afferents; (c) septal afferents; (d) other afferents. The evidence concerning the degeneration in the stratum lacunosum/moleculare of CA1 (and stratum moleculare mediation of hippocampal theta activity by various of these afferents will be of subiculum) follows section of the dorsal hippocampal commissure. This general discussed in a subsequent chapter. picture has been confirmed in an electrophysiological study by Thomson (1975). The site of origin of these commissural projections has been studied in a variety of 3.4.4(a). Entorhinal afferents. Cajal (1911) described three pathways to the ways. In early work Andersen (1959, 1960a, b, c) and Raisman et al. (1965) used hippocampal region from the entorhinal area: (1) the perforant path; (2) the alvear electrophysiological and classical degeneration techniques and observed a similar path; (3) the crossed temporo-ammonic tract. The crossed temporo-ammonic tract pattern; there was a symmetrical connection between homotopic points on the two distributes to the presubiculum; the alvear path (if it exists∗) remains in the deep sides. That is, CA3 projected to CA3; while CA1 (posterior) projected to CA1 layers of the subiculum and contributes only a small number of fibres to the stratum (posterior); CA1 (anterior) did not seem to send commissural afferents, though it did oriens of CA1. Major interest, therefore, has centred on the other pathway, the receive efferents. Finally, there was some debate concerning fascia dentata perforant path, since it appears to be the major avenue of entorhinal afferents to the commissural inputs; Raisman et al. suggested a diffuse origin while Andersen, hippocampus. Bruland, and Kaada (1961a) presented electrophysiological evidence for a non- According to Lorente de No (1934), Blackstad (1958), and Raisman et al. (1965) homotopic commissural input to fascia dentata from CA3. the perforant path originates in the lateral entorhinal cortex (see More recent studies, using either autoradiography (Gottlieb and Cowan 1972) or the retrograde transport of horseradish peroxidase (Segal and Landis 1974), have clarified the picture. There are clear-cut homotopic projections from CA3 to CA3, ∗ Nafstad (1967), in a combined degeneration-electron-microscope study, found fibres of passage in CA1 but terminal degeneration in less than 1 per cent of the synaptic terminals. Similarly, Hjorth-Simonsen (1972) but non-homotopic inputs to both CA1 and fascia dentata. These two areas both could find no terminal degeneration in CA1 following lateral or medial entorhinal lesions. Segal and Landis seem to receive their commissural (1974) confirmed these findings in a horseradish peroxidase transport study; there was no label transported to the entorhinal cortex if the peroxidase was injected into CA1 alone. Finally, Andersen, Homqvist, and Voorhoeve (1966b), on the basis of an electrophysiological study, concluded that the activation of CA1 pyramidal cells from the entorhinal area was not direct but by way of dentate and CA3 sectors. Anatomy 125
Figs. 5 and 6), crosses the subiculum to reach its upper layers, and then either (Gloor, Vera, and Sperti 1963). The perforant path/dentate synapses are subject to courses within these upper layers to enter the hippocampus proper (Fig. 6) or long-term increases in excitability following a brief tetanus of electric shocks in the perforates through the obliterated hippocampal fissure to the fascia dentata (Fig. 6). anaesthetized rabbit (Andersen, Holmqvist, and Voorhoeve 1966a, Lømo 1966, More recent work (Nafstad 1967, Hjorth-Simonsen and Jeune 1972, Van Hoesen, 1971b, Bliss and Lømo 1970, 1973), unanaesthetized rabbit (Bliss and Pandya, and Butters 1972, Hjorth-Simonsen 1973, Van Hoesen and Pandya 1975b) Gardner-Medwin 1971, 1973), unanaesthetized rat (Douglas and Goddard 1975), confirms this pathway but indicates that the fibres originate from the medial and in in vitro slices of rat hippocampus (Deadwyler, Dudek, Cotman, and Lynch entorhinal cortex as well (Figs. 5 and 6). While the lateral pathway ends on the distal one third of the granule cell dendrites the medial pathway occupies the medial one-third (Hjorth-Simonsen 1972, Van Hoesen, Pandya, and Butters 1972, Fifková Inputs to the entorhinal area. Until recently information about inputs to the 1975). Although most of the projection is ipsilateral, there is evidence for a small entorhinal cortex and other parahippocampal areas has been unsystematic and contralateral projection (Goldowitz et al. 1975). There is some indication from one scattered. What was available suggested weak or uncertain inputs from the of Hjorth-Simonsen's experiments and from Van Hoesen and Pandya (1975b) that a following cortical areas: (1) prefrontal and cingulate cortices via the cingulum lesion on the border of the lateral and medial entorhinal cortices yields degeneration bundle (Adey 1951, Adey and Meyer 1952, White 1959, Cragg 1965, Raisman et intermediate to that described above; thus it is possible that one is dealing with a al. 1965, McLardy 1971, Leichnetz and Astruc 1975)∗∗ although Domesick (1969, continuum rather than with discrete projections here. Given this, a dentate granule 1970) has presented strong evidence that, in the rat at least, most of the fibres in this cell could tell where its input comes from by how high up on its dendrites the input bundle are thalamo-cortical and end in cingulate cortex and presubiculum; Shipley synapses. On the other hand, this may be another example of the imbricated pattern (1974, 1975) has demonstrated a projection from the presubiculum to the dorsal and described above (see pp.120-2) for various intrahippocampal projections. As can be medial entorhinal cortex;∗∗∗ (2) temporal cortex (Cragg 1965); (3) parietal areas, seen from Fig. 6 there is also an input to the stratum lacunosum/moleculare of CA3, either directly (Pandya and Kuypers 1969, Pandya and Vignolo 1969, Petras, 1971) the lateral path terminating on the most distal portion of the dendrites, the medial or indirectly via the cingulate cortex (Cragg 1965) and the prepyriform cortex path on the more proximal portion. While Van Hoesen, Pandya, and Butters (1972) (Cragg 1961); (4) pyriform cortex (Powell, Cowan, and Raisman 1965). Niemer, failed to find a projection to the hippocampus proper from the entorhinal cortex in Goodfellow and Speaker (1963) and McKenzie and Smith (1970) have obtained the monkey, Van Hoesen and Pandya (1975b) reported a projection limited to the electrophysiological evidence of projections from large areas of the neocortex to uncal extremity of CA3 and the area between CA1 and CA3 in the rest of the hippocampus, perhaps via entorhinal areas, but the precise pathways were not The Hjorth-Simonsen and Jeune (1972) study provides anatomical evidence for a In addition, there is evidence suggesting that both olfactory (Cragg 1960, 1961, degree of topographical specificity of the entorhinal input to the dentate granule Heimer 1968, White 1965, Price and Powell 1971, Kerr and Dennis 1972) and cells. Lesions of the dorsal entorhinal cortex caused degeneration restricted to the visual (Casey, Cuenod, and MacLean 1965, Cuenod, Casey and MacLean 1965) dorsal hippocampus; progressively deeper lesions yielded degeneration more sensory systems gain access to the entorhinal area. It has been suggested that the posteriorly and then more temporally within the hippocampus. A similar precision in visual input may not be mediated by the neocortex but may be direct from optic this projection has been demonstrated in the horseradish peroxidase study of Segal fibres bypassing the lateral geniculate nucleus (MacLean and Cresswell 1970), or and Landis (1974) and in the careful electrophysiological studies of Lømo (1971a). via the newly discovered optic-tract input to the anterodorsal nucleus of the It thus appears that the lamellar organization proposed for the hippocampal internal thalamus (Conrad and Stumpf 1975). Consistent with this notion is the report by circuitry (see pp. 120-2) also extends to the entorhinal inputs (Andersen et al. 1971) Sager, Nestianu, and Florea-Ciocoiu (1967) of a long-latency (40 ms) visual evoked potential in the hippocampus which survived destruction of the geniculo-striate The perforant path projection to the dentate granule cells has been shown to be system. Cragg (1960) recorded the evoked potentials powerfully excitatory (Andersen, Holmqvist, and Voorhoeve 1966a, Lømo 1971a), as has the projection to the CA3 pyramidal cells ∗ Schwartzkroin and Wester (1975) have recently shown that the Schaffer collateral pathway from CA3 to CA1 also possesses this property, while the CA3-septum pathway does not (Andersen, personal ∗ A projection from the prorhinal area, adjacent to the entorhinal cortex, to the junction of stratum communication). Thus, only the synapses within the hippocampus have this marked plasticity. moleculare/stratum radiatum in field CA1a was also seen in this study. ∗∗ See p, 131 for a description of a direct prefrontal-hippocampus projection reported in this study. ∗∗∗ This input is a highly organized one, indicating that the topographical specificity already noted for the hippocampus and the perforant path input extends back to the presubiculum, and possibly beyond that to the anterior thalamus (Shipley and Sorenson 1975). Anatomy 127
generated in the hippocampus by olfactory-bulb stimulation in several species and found that this input decreased in importance with phylogenetic development. In rat, responses were easily recorded throughout the extent of the hippocampus, while in monkey only a small part of the anteroventral hippocampus responded. Lastly, the amygdala projects to the entorhinal cortex and to the subiculum (Krettek and Price 1974). This pathway could be responsible for the (multi-synaptic) connections from the amygdala to the hippocampus which have been demonstrated electrophysiologically (Gloor 1960, Gloor, Vera, and Sperti 1963).∗ Van Hoesen, Pandya, and Butters (1972, 1975) and Van Hoesen and Pandya (1975a) have provided a thorough analysis of cortical afferents to the entorhinal area in the monkey. They found three main projection areas: (1) the adjacent prepyriform cortex, parasubiculum, and presubiculum (Brodmann's areas 51, 49, and 27) as well as the perirhinal and prorhinal cortex; (2) the adjacent area TF-TH of Bonin and Bailey, temporal cortex caudal to the entorhinal area; (3) the caudal portion of the orbito-frontal cortex, Bonin and Bailey's FF (Walker's areas 12 and 13). With the exception of the input from the perirhinal area, which projects to contiguous portions of the entire entorhinal cortex, all of these sources of input are highly specific in terms of laminar and sub-areal termination within the entorhinal region. Thus, the prepyriform cortex projects primarily to the lateral entorhinal area, as does the orbito-frontal cortex, while areas 49, 27, TF, and TH project primarily to the medial FIG. 10. Convergence of afferent information onto the entorhinal cortex from primary (SA1, entorhinal area. The extent of interconnections in these areas is truly bewildering, AA1, VA1) and secondary (SA2, AA2, VA2) association areas of the neocortex in the monkey. and is best described in Fig. 10, taken from Van Hoesen, Pandya, and Butters (1972). (From Van Hoesen, Pandya, and Butters 1972; copyright 1972 by the American Association for This figure makes it clear that there is a cascading of inputs from a number of the Advancement of Science.) cortical areas through all adjacent regions leading ultimately to the entorhinal cortex. Halaris 1975) of CA1 and CA3, and in a restricted part of the hilar zone beneath the This pattern of inputs to the entorhinal area strongly suggests that the hippocampus is granule cells in fascia dentata. Some inputs are seen to the stratum radiatum and the concerned not with information about any particular modality, but rather with highly stratum oriens, more densely in CA3 than in CA1. There appears to be a higher analysed, abstracted information from all modalities. 5-HT activity in more posterior portions of the hippocampus (Storm-Mathisen and Guldberg 1974). Iontophoretic application of 5-HT inhibits hippocampal 3.4.4(b) Brain-stem afferents. Afferents to the hippocampus arise in several (pyramidal) cells (e.g. Biscoe and Straughan 1966) as does stimulation in the raphe brain-stem areas. The median raphe nucleus, and particularly the nucleus centralis nucleus itself (Segal 1975). Thus, it appears as though a 5-HT inhibitory pathway superior, projects to the hippocampus, accounting for most of the serotonin from the median raphe nucleus to the hippocampus is well established.* De France, (5-hydroxytryptamine, 5-HT) found in this structure (Ungerstedt 1971, McCrea, and Yoshihara (1975) have shown that serotonin can facilitate units in Storm-Mathisen and Fonnum 1972, Kuhar, Aghajanian, and Roth 1972, CA1 and suggested that these effects are mediated via the Schaffer collateral Storm-Mathisen and Guldberg 1974, Lorens and Guldberg 1974, Segal 1975, Moore system. Similar effects of serotonin were seen on cells in the lateral septum and Halaris 1975). This projection courses primarily through the fimbria, fornix, and receiving inputs from the fimbria and CA3. cingulum, terminating mostly in stratum lacunosum/moleculare (Fuxe and Jonsson The locus coeruleus also projects to the hippocampus, accounting for much of the noradrenaline (NA) in this structure (Blackstad, Fuxe and Hökfeldt 1967, Ungerstedt 1971, Lindvall and Bjorklund 1974, Ross and Reis 1974, Fuxe, Hamberger, and Hökfeldt 1968). A recent immuno- ∗ Recent work in the dog suggests that monosynaptic connections exist between amygdala and ventral hippocampus (Kosmal, personal communication); the functional significance of these is unclear. ∗ Wimer, Norman, and Eleftheriou (1973) have reported large differences in the amount of hippocampal 5-HT between two strains of mice and changes in amount of 5-HT with active avoidance training or etherization. Anatomy 129
fluorescence study using dopamine-β-hydroxylase (DBH) as a marker has greatly ventral hippocampus. There are two possible reasons for this discrepancy: (1) clarified the projection pathway and termination sites of this system (Swanson and lesions might have involved fibres of passage, such as those coursing through the Hartman 1975). Fibres course through the medial forebrain bundle and septum, septum from the brain stem (see above); (2) there might be a topographic projection entering the hippocampus via retrosplenial cortex. There does not appear to be a from the medial septum to the hippocampus such that the dorsal medial area projection via the fimbria or fornix. These NA fibres terminate rather specifically in projects to the dorsal hippocampus, and then as one moves ventrally and laterally the stratum lacunosum/moleculare of CA1 and CA3, the hilus of the fascia dentata, within the medial septum the target in the hippocampus moves first posteriorly and and the stratum lucidum of CA3. Here, they are in proximity to the mossy-fibre then ventrally. Evidence for this exists in several studies, and most convincingly in terminals. Mostly inhibitory effects have been reported after iontophoretic application a report by Segal and Landis (1974). They injected horseradish peroxidase into the of NA (e.g. Biscoe and Straughan 1966), and this is also seen after stimulation in the CA3/CA4 field of either the dorsal or ventral hippocampus and examined the locus coeruleus (Segal and Bloom 1974a,b). septum for retrograde transport of this enzyme. Only cells within the medial septum It is worth noting that these two systems, both inhibitory, take primarily were labelled. Further, dorsal hippocampal injections yielded label in cells in the different paths to the hippocampus, NA entering retrosplenially and 5-HT entering medial part of the medial septum, while ventral hippocampal injections yielded rostrally via the fimbria. Stimulation studies (see pp. 154-5) indicate that the lateral label in cells of the more lateral parts of the medial septum. Mellgren and Srebro raphe pathway is influential in the elicitation of a particular pattern in the (1973), in a combined acetylcholine esterase and Fink-Heimer degeneration study, hippocampal EEG (termed SIA) which is different from that elicited via the more reported results consistent with this notion.∗ Finally, studies of the depletion of medial, locus coeruleus, pathway. Stimulation here elicits the characteristic choline acetylase (ChAc) in hippocampus after various lesions also indicate that the hippocampal theta activity. A full discussion of these effects, which are partly projection involving acetylcholine is restricted solely to the medial septal region mediated via the septum, is given in the next chapter (pp. 154-60). Gray et al. (1975) (Srebro et al. 1973, Oderfeld-Nowak 1974, Storm-Mathisen 1972, Storm-Mathisen have provided pharmacological evidence implicating some noradrenaline system in and Guldberg 1974). theta activity (but see Robinson, Pappas, and Vanderwolf (1975) for a contradictory The path taken by the septo-hippocampal fibres is also in dispute. Some authors (e.g. Andersen, Bruland, and Kaada 1961a,b, Powell 1963) trace the fibres in the In several physiological experiments (Grantyn 1970, Grantyn and Grantyn 1970, dorsal fornix and thence across the top of the hippocampus in the alveus, others 1971, 1973a,b) the existence of an input to the hippocampus from the mesencephalic (e.g. Raisman et al. 1965, McLennan and Miller 1974a) trace them in the fimbria, reticular formation has been demonstrated. Some hippocampal neurones respond to and still others (e.g. Lewis and Shute 1967) trace them in both. Finally, Bland, reticular stimulation with a hyperpolarization, some with a depolarization, and others Andersen, and Ganes (1975) found that a knife cut through the apical dendritic with an oscillation between the two. This input survives a lesion of the septum and is region of CA3 eliminated theta activity in their physiological experiment, probably mediated by the neocortex (Grantyn, Grantyn, and Hang 1971). suggesting this as the pathway for the septo-hippocampal fibres. Similarly, in some experiments (Raisman et al. 1965, Raisman 1966, Genton 1969) degeneration after 3.4.4(c). Septal afferents. A major input to the hippocampus arises in the septal area. septal lesions is restricted to hippocampal fields CA3 and CA4 and the fascia There is some disagreement as to the exact origin of these fibres, their course, and the dentata, while others trace degeneration to all hippocampal fields in addition to the extent and locus of their termination within the hippocampus. Behind these strictly fascia dentata (Siegel and Tassoni 1971b; Ibata, Desiraju, and Pappas 1971, anatomical questions lies an important functional one: which part of this input is Hjorth-Simonsen 1973, Mellgren and Srebro 1973, Mosko, Lynch, and Cotman involved in the generation of hippocampal theta activity (see pp. 154-60)? The 1973). In some of these latter studies the CA1 projection was not as dense as the relevant studies are listed in Appendix Table A1 CA3/fascia dentata projection. Studies employing cholinergic stains (Lewis and Most of these studies localize the origins of the septo-hippocampal projection to Shute 1967, Lewis, Shute, and Silver 1967, Storm-Mathisen 1972, Storm-Mathisen the medial septum and the dorsal (vertical) limb of the diagonal band of Broca and Fonnum 1972, Mellgren and Srebro 1973, Mosko et al. 1973) support the idea (DBB); and deny any projection from the lateral septum and ventral (horizontal) limb of a widespread cholinergic input to all fields in both dorsal and ventral of the DBB. Of the four studies reporting discrepant results, only three used classical hippocampus. In addition to the anatomical techniques. All purported to show that medial septum-dorsal DBB projected to dorsal hippocampus while lateral septum-ventral DBB projected to ∗ De France, Kitai, and Shimono (1973a,b) reported antidromic activation of cells in the ventral lateral septum with fimbrial stimulation, but this could not be replicated by McLennan and Miller (1974a). Anatomy 131
long-term loss of acetylcholine esterase (AchE) and choline acetylase (ChAc) studies (e.g. Bland, Kostopoulos, and Phillis 1974) indicates that these inputs are staining after fimbrial section or septal lesions, it has been reported that immediately excitatory, more so in the fascia dentata than in the hippocampus proper.∗ after fimbrial section there is a 60 per cent increase in AchE and ChAc in the hippocampus (Sethy et al. 1973). Further, stimulation of the surface of the septum 3.4.4(d). Other afferents. Aside from these major afferent systems to the (Smith 1972) or the medial septum itself (Dudar 1975) increases the release of AchE hippocampus two other sources of input have recently been uncovered. Heath and from the surface of the hippocampus. Harper (1974) have demonstrated an input from the fastigial nucleus of the There is somewhat more agreement on the question of the final termination of the cerebellum to CA2, CA3, and CA4 and to the fascia dentata. The terminal sites septal afferents to the hippocampus; most authors find a major projection to the include the stratum oriens, stratum radiatum and stratum lacunosum in the stratum oriens and stratum radiatum of the hippocampus and to the polymorph zone hippocampus proper, and the polymorph layer in the fascia dentata. Finally, of the fascia dentata. In addition, some find a small projection to the molecular layer Leichnetz and Astruc (1975) have demonstrated a direct projection in monkey from of CA4 and fascia dentata. No degeneration is found in the cell-body layers. the medial prefrontal cortex to CA1 and CA3 localized within stratum oriens, Although it is assumed by most authors that the septal projection is to the pyramidal stratum pyramidale, and stratum lacunosum/moleculare. This pathway courses cells and granule cells, the heavy concentration of terminals in the stratum oriens and through cingulate and uncinate approaches to the hippocampus, entering either in the zone just above the cell bodies has suggested to Mosko et al. (1973) that the through the alvear or perforant paths. Some degeneration was seen in the target cells might be interneurones and, more specifically, the basket cells of Cajal. commissure and the alvear path of the contralateral hippocampus, indicating a The septo-hippocampal projection is one of the most controversial in the brain, bilateral projection. There was a suggestion of some specificity in the two modes of perhaps because it has been widely studied and used as a model system by so many projection, the alvear input terminating on the basal dendrites and somata of investigators. Yet the exact origin and terminal distribution of this path has not been pyramidal cells (and perhaps basket cells), the perforant input terminating on apical determined. No single methodological difference, such as species or stain used, dendrites. seems sufficient to account for the reported discrepancies.∗ We have noted several possible explanations. First, the use of lesions involves the possibility of destroying 3.4.4(e). Evidence of hippocampal afferents from single unit studies. On the basis of fibres of passage. Second, and consistent with much of the evidence, the septum the anatomical studies reviewed above, one might expect to find three different might project to the hippocampus in an orderly topographic manner. In addition to types of unit responses in the hippocampus: (a) specific responses to complex the medio-lateral septal difference noted above, evidence has been provided for a sensory stimuli mediated by the entorhinal afferents; (b) diffuse arousal responses dorsal-to-dorsal and ventral-to-ventral projection (Mellgren and Srebro 1973) as well to a wide range of sensory stimuli mediated by the brain-stem afferents arriving via as an anterior-to-anterior and posterior-to-posterior projection in the monkey (De the septal-fornix input and, to a lesser extent, via the entorhinal area; (c) motor Vito and White 1966). Consistent differences in the placement of septal lesions information conveyed by the brain-stem afferents from posterior hypothalamus and would then account for the discrepant reports. reticular formation. It should be noted, however, that the third type of input can In summary, there is almost definitely a group of cholinergic fibres centred in the only be demonstrated adequately in animals which are free to move. Motor-related medial septum projecting sparsely to all ipsilateral fields of the hippocampus and to inputs are either movement-generated proprioceptive or vestibular feedbacks on the the fascia dentata. These fibres travel in the fimbria to CA3-CA4 and perhaps inside one hand, or collaterals from the movement-generating circuits themselves on the the CA3 hilus to the fascia dentata. Another projection may travel in the dorsal fornix other. While the existence of the former type of input could be demonstrated in and alveus to CA1 and also to the fascia dentata. The major termination zones are acute, paralysed animals by the passive manipulation of individual limbs or the found just above and below the main cell bodies, and this could indicate terminations whole animal, the collateral motor input would be more difficult to detect and might on the basket cells as well as on pyramidal cells. Evidence from iontophoretic ∗ There was an important difference between CA1 and fascia dentata cells, however: the excitation elicited ∗ It should be noted that Ibata et al. (1971) mention that they could not find terminals in the hilus of the by acetylcholine was blocked by atropine in CA1 but not in fascia dentata. Thus, the CA1 receptors were fascia using a Fink-Heimer method, although they were apparent with a Nauta-Gygax method. muscarinic while the dentate ones may be nicotinic. This difference may be of some significance to an understanding of the two different types of theta activity, as we note later (see p. 166). Anatomy 133
easily be misinterpreted as a long-latency sensory response.∗ Finally, in the studies from our own and from Ranck's laboratory, in which the Studies on the activity of single hippocampal neurones in response to sensory animal was free to move and its movements carefully monitored, units specifically stimulation or during motor behaviour have been published from nine laboratories related to movement were reported. In addition, unit activity which might in part (Brown and Buchwald 1973, Feder and Ranck 1973, Green and Machne 1955, reflect responsiveness to complex sensory inputs was also described. A fuller Lidsky, Levine, and MacGregor 1974a,b, Molnar and Arutyunov 1969, O'Keefe description of these unit types and possible reasons for the apparent discrepancies 1976, O'Keefe and Dostrovsky 1971, Ranck 1973, Segal 1974, Vinogradova 1970, amongst the different studies is given in the section on single-unit activity in the Vinogradova, Semyonova, and Konovalov 1970, Yokota, Reeves, and MacLean freely moving animal (pp. 190-217). 1970).∗∗ Most of these studies report that sensory stimuli have a non-specific arousing effect on hippocampal neurones, usually of rather long latency ( > 100 ms.) Evidence for more specific sensory responses has been reported in four studies 3.5. Efferents from the hippocampal region
(Brown and Buchwald 1973, Green and Machne 1955, Molnar and Arutyunov 1969, Appendix Table A2 lists those studies which have traced the degeneration resulting
Segal 1974). In the first of these, Green and Machne (1955) found that while many from damage to the hippocampus, the dorsal fornix, the fimbria, or the columns of
hippocampal units in the paralysed, unanaesthetized rabbit were multi-modal, the fornix. Until recently, there was considerable agreement that the hippocampus
activated by all the sensory stimuli that they tried, others could only be excited by projects to (1) the lateral preoptic and lateral hypothalamic areas, (2) the septal
one stimulus such as a touch on a particular part of the body and not by visual or region, (3) the thalamus, (4) the mammillary bodies, (5) the rostral mid-brain, and
auditory stimuli. There was no obvious tendency for units responsive to the same (6) caudally to the subiculum and the entorhinal cortex. Somewhat more
stimuli to cluster into anatomically discrete areas. In lightly anaesthetized cats controversial was the origin of a projection to the medial hypothalamus. Within
Molnar and Arutyunov (1969) asked whether hippocampal neurones responded to these accepted projection areas there was some disagreement as to the exact sites of
one or more sensory inputs. When they used single light flashes, clicks or shocks to termination, and there appeared to be clear species differences as well. Almost all of
the skin, the neurones appeared to be unimodal but if repetitive stimuli were the anatomical studies listed used techniques (e.g. Nauta or Fink-Heimer methods)
delivered virtually all neurones responded to all modalities. In the freely moving rat which stained degenerating fibres after lesions. These, as we have noted, cannot
Segal (1974) also reported finding some units which responded to light flashes but distinguish between fibres originating within the area and those originating
not to tones. Brown and Buchwald (1973) concentrated on auditory stimuli and elsewhere but passing through the lesioned area. In the first autoradiographic study
reported two types of specific response in the paralysed, unanaesthetized cat. Some of the hippocampus Swanson and Cowan (1975, personal communication) have
units responded differentially to tones of different frequencies; some units were found that in the rat the hippocampus projects rostrally only to the septum. The
excited by white noise presented to the ipsilateral ear, but were inhibited by the same source of the fibres to the thalamus, hypothalamus, and mammillary bodies is the
noise in the contralateral ear. Finally, Yokota et al. (1970) recorded short-latency more caudally placed subicular region. It had already been established that a major
EPSP's but no action potentials in hippocampal pyramidal cells in chronic monkeys outflow of the CA1 field in the rat (Hjorth-Simonsen 1973) and the rabbit
following shocks to the olfactory bulb. The short latency might indicate a specific (Andersen, Bland, and Dudar 1973) goes to the subiculum. What was formerly
response, but since no other modalities were tested one cannot be sure. In other considered to be a direct connection from hippocampus to thalamus and
studies (O'Keefe 1976, O'Keefe and Dostrovsky 1971, Ranck 1973, Vinogradova hypothalamus must now be considered indirect. One must keep in mind the
1970, Vinogradova, Semyonova, and Konovalov 1970) only non-specific responses possibility that there are minor species differences in the hippocampal efferents, but
to all sensory stimuli tested were found.
a radical difference between the rat and other species is unlikely. Swanson and Cowan (1975a) found that the efferent fibres of the subiculum and hippocampus ∗ Black and his colleagues (see pp. 174-5) have shown that theta activity in the hippocampal EEG can be clustered into different fibre tracts, the former in the post-commissural fornix, the conditioned to a sensory stimulus in the paralysed, unanaesthetized dog, but that this theta activity is probably a latter in the pre-commissural fornix. Chronister, Sikes, and White (1975) have also concomitant of motor outflow signals and not related to sensory processing of the signal since the dogs invariably moved in response to the sensory stimulus when subsequently tested in the unparalysed state. reported that the post-commissural fornix receives fibres from periallocortex ∗∗ Studies of the changes in hippocampal unit activity during sleep and wakefulness have not been included. adjacent to the subiculum, notably presubiculum and parasubiculum, and that these These are discussed in the section on single unit activity in the freely moving animal (pp. 212-13). two components of the post-commissural fornix have different target sites; we shall consider these in the context of a description of both fornix tracts, and Anatomy 135
their sites of termination. Subsequently, we shall discuss the caudally directed there are about five times as many fibres as seen in monkeys. Comparable figures efferents and, finally, the topography of the efferent projections. were reported by Simpson (1952) and Daitz (1953). The various investigators note that the pre-commissural components of the fornix, which could not be quantified, 3.5.1. THE FORNIX-FIMBRIA SYSTEM were at least as extensive as the postcommissural ones. The rostrally directed output fibres of the hippocampus, subiculum, and adjacent The thalamic destination of the post-commissural fornix fibres appears to vary areas gather together in the fimbria and perhaps the dorsal fornix; these join at the from species to species. Valenstein and Nauta (1959), in a comparative study, found anterior part of the hippocampus to become the columns of the fornix. At this level that the fibres distributed primarily to the antero-ventral and antero-medial thalamus many of the fibres cross the mid-line in the ventral hippocampal commissure and run and only sparsely to the intra-laminar thalamus in the rat and monkey, but that this caudally in the contralateral fimbria to reach the other hippocampus; these relationship is reversed in the cat and guinea pig, with the dominant projection connections have been discussed already (p. 122). The bulk of the fibres in the fornix going to the intra-laminar group. Although they had no pure fornix lesions in the continue rostrally to penetrate the septo-fimbrial nucleus at the rear of the septal area. monkey studies they did find a projection to lateralis dorsalis and medialis dorsalis; Here, the fibres split into the two components noted above, the postcommissural and this prompted them to speculate that the fornix input becomes more involved with pre-commissural fornices. As we just noted, the latter originates primarily in the parietal and frontal lobes in higher species. Since the projection of the anterior hippocampus, the former primarily in adjacent allocortical areas. thalamic group is to lateralis dorsalis and medialis dorsalis as well as to cingulate cortex, this change in termination sites with phylogeny amounts to the substitution 3.5.1(a). Pre-commissural fornix. The pre-commissural fornix distributes to most of of a direct for an indirect path. However, these terminations in the monkey must the nuclei of the septal area, including the lateral septum, the diagonal band of Broca, remain uncertain in the absence of pure fornix lesions, especially since there is and the bed nucleus of the anterior commissure; fibres also terminate in the nucleus evidence that the septum has a large input to medialis dorsalis in the cat (Guillery accumbens septi, now thought to relate more to ventral caudate than to septum (cf. 1959); however, Raisman (1966) could not demonstrate this connection in the rat. Heimer and Wilson 1975). Fibres traverse the septal region and continue into the The other major input to the anterior nuclei of the thalamus is from the mammillary preoptic area as the fascicles of Zuckerkandl, some terminating in the lateral bodies via the mammillo-thalamic tract (the tract of Vicq d'Azyr), which indicates pre-optic region while others turn caudally into the medial forebrain bundle to amongst other things that the anterior thalamic nucleus compares the outflow from distribute to the lateral hypothalamus as far caudal as the optic chiasm. The existence the subiculum, pre-, and parasubiculum, but only after a transformation of the of these lateral pre-optic and hypothalamic connections could not be established in subicular output in the mammillary bodies. Gergen (1967) reported that in many the rabbit (Sprague and Meyer 1950, Cragg and Hamlyn 1960), and the pathway here areas of the thalamus of the monkey, and in particular anterior nuclei and might be an indirect one. Access to the more caudal lateral hypothalamic areas is paracentralis, more than half of the cells are influenced by hippocampal stimulation, gained indirectly from the large contribution to the medial forebrain bundle by septal while Parmeggiani (1967) reported responses to low-frequency stimulation of the and pre-optic areas. hippocampus in the following thalamic nuclei: AV, AM, VA, MD, CL, LD, paracentralis, and habenularis. Mok and Mogenson (1974) have confirmed this last 3.5.1(b) Post-commissural fornix. The post-commissural fornix divides into two projection. approximately equal components (A and B of Guillery 1956): one is destined for the The hypothalamic component of the post-commissural fornix distributes its thalamus, the other for the mammillary bodies and rostral brain stem. Chronister et fibres almost entirely to the mammillary bodies, and primarily to the lateral and al. (1975) have shown that the former, thalamic, projection derives from the pre- and posterior parts of the medial mammillary bodies. As we just noted, these nuclei in parasubiculum, while the latter, mammillary, projection derives from the subiculum. turn project to areas AV and AM in the thalamus via the mammillo-thalamic tract. This latter projection has been confirmed in a horseradish peroxidase study by The other projection of this area is to the mid-brain via the mammillo-tegmental Meibach and Siegel (1975). tract. In addition to this indirect connection with the motor areas of the mid-brain, The number of fibres in the post-commissural fornix before and after the thalamic there are direct connections via fornix fibres which bypass the mammillary bodies component has split off has been assessed in several species by Powell et al. (1957). to project directly to the mid-brain. The magnitude and site of termination of this The numbers for rabbit, cat, and monkey are surprisingly similar (c.200,000 before direct projection varies with the species studied (Valenstein and Nauta 1959). In all and 100,000 after) and are about four to five times that seen in the rat. A huge species there is a termination in the rostral part of the mid-brain central grey. increase is seen in humans, where Anatomy 137
In addition, in the guinea pig the fibres distribute to the subthalamic region and the Siegel and his colleagues (Edinger, Siegel, and Trocano 1973, Siegel et al. 1974, central and caudal mesencephalic tegmentum, in particular the nucleus centralis Siegel and Tassoni 1971a, Meibach and Siegel 1975) suggest instead that the split tegmenti superior of Bechterev. The direct subthalamic projection also appears in the in projections lies between the dorsal and ventral hippocampus, rather than the CA monkey, but it is indirect in the cat via a relay in the mammillary bodies. There is fields. They posit that the dorsal hippocampus projects to the medial septum via the evidence from single-neurone recording studies for predominantly, but not dorsal fornix, while the ventral hippocampus projects via the fimbria to the lateral exclusively, excitatory inputs from the hippocampus to the mammillary bodies septum.∗ The autoradiographic study of Swanson and Cowan (1975a, Swanson, (Poletti, Kinnard, and MacLean 1970), the lateral septum, preoptic region and personal communication) favours Raisman's hypothesis. CA1 was found to project various hypothalamic nuclei (Poletti, Kinnard and MacLean 1973), the thalamus rostrally only to the ipsilateral lateral septum. CA3 projected solely to the lateral (Yokota and MacLean 1968), and to the reticular formation (Grantyn et al. 1972). septum as well, but bilaterally and in a highly organized fashion, such that the more Even before the work of Swanson and Cowan (1975a) there was some question as lateral one went in the hippocampus, the more lateral within the lateral septum the to whether the medial hypothalamic pathway to periventricular areas originated in termination appeared. Similarly, the more ventral one went in the hippocampus, the the hippocampus or in the subiculum. Its origin in the hippocampus had been more ventral within the lateral septum was the termination. The confusion rests, to affirmed in the rat (Guillery 1956, Nauta 1956, Valenstein and Nauta 1959, but see some extent, upon the definition of what constitutes the medial septum.∗∗ Raisman et al. 1966, Chronister et al. 1975, for denials), in the guinea pig (Johnson Additionally, lesions have usually encroached upon fibres of passage, making 1959, Valenstein and Nauta 1959), in the rabbit (Cragg and Hamlyn 1960), in the cat clear-cut results unlikely. (Siegel and Tassoni 1971a, but see Valenstein and Nauta 1959 for a denial), but not Thus, according to Swanson and Cowan, the hippocampus projects solely to the in the monkey (Valenstein and Nauta 1959). Raisman et al. (1966) had put forth a lateral septum in a fashion which maintains the precise spatial ordering present in strong case that the pathway actually originated in the subiculum and ran in the the afferent and intrinsic connections. Though the autoradiographic technique is not fimbria. Lesions of the hippocampus or fimbria almost invariably destroy these perfect, it does provide a cleaner analysis than most degeneration methods, and this fibres, but with lesions which spared both (Nauta 1956, Raisman et al. 1966) there work indicates that Siegel's data are probably contaminated by lesions beyond the was no degeneration in the medial cortico-hypothalamic tract. This tract is the most confines of what was intended. If anything, diffusion of autoradiographic label, likely candidate for the hippocampal influence over the pituitary-adrenocortical which constitutes the major methodological problem with this technique, would system (see pp. 357-62). tend to expand the sites thought to receive hippocampal efferents. Thus, we can conclude that there are direct efferent connections between the hippocampus and 3.5.1(c). Topography of the fornix-fimbria system. The topography of the rostrally the lateral, but not medial, septum. directed efferents is somewhat controversial. There are two basic disagreements: (1) are there any rostrally directed efferents from CA1 (2) does the hippocampus project 3.5.2. CAUDALLY DIRECTED EFFERENTS to the medial septum at all? Hjorth-Simonsen (1973) showed, in a degeneration A direct projection to the entorhinal cortex has been convincingly demonstrated by study, that CA1 projected in a topographically organized way to the subiculum. This Hjorth-Simonsen (1971). Earlier studies (Adey, Sunderland, and Dunlop 1957, organization conformed to the lamellar pattern described earlier for the afferent Votaw 1960a) had hinted at this pathway; according to Hjorth-Simonsen it systems. Andersen et al. (1973) extended this notion in an electrophysiological study originates in the CA3 field of primarily ventral hippocampus and terminates in layer in the rabbit. While they could antidromically activate CA3 pyramids from the 4 of the entorhinal area. Deadwyler, West, Cotman, and Lynch (1975) have fimbria, CA1 pyramids could only be antidromically activated from the caudal alveus confirmed the basic features of this pathway in an electrophysiological study. and subiculum. This suggested that CA1 projected only in a caudal direction, a view Long-latency potentials were evoked in all areas of dorsal entorhinal cortex opposed to the traditional story which suggested that CA1 efferents projected via the following dorsal hippocampal CA3 stimulation. The authors suggest that this effect dorsal fornix and dorsal fimbria to the septofimbrial nucleus, the diagonal band of Broca, part of the ventrolateral septum, and the nucleus accumbens septi (Raisman et ∗ Some support for this notion comes from the observation by McLennan and Miller (1974a) that two al. 1966). CA3 and CA4 were thought to project via the fimbria to the dorsolateral distinct potentials can be evoked by fimbrial stimulation: one in the dorsolateral septum, the other more septum, the diagonal band, and the nucleus accumbens. ventral. The latter potential had a long latency and required more posterior fimbrial stimulation, suggesting that it arose in ventral hippocampus or subiculum. Chronister and White (1975) noted that, in Golgi preparations, the medial and lateral septum appear quite different in terms of dendritic arborization patterns. Anatomy 139
was mediated by the longitudinal association bundle, accounting for the effect of visual evoked potential appears similar to that produced by mid-brain reticular dorsal stimulation upon an output presumably generated in the ventral hippocampus. stimulation (Sierra and Fuster 1968). Stimulation of the hippocampus in cerveau A projection from CA1 to the subiculum has been reported (Andersen et al. 1973, isolé cat (in which the influence of structures caudal to the mesencephalon is Hjorth-Simonsen 1973). The most likely pattern of outputs from the hippocampus is eliminated) enhanced the visual cortical evoked potential, but similar stimulation in schematized in Fig. 11. Note that we accept the Swanson and Cowan results in an encéphale isolé preparation (with rostral-caudal connections intact) resulted in a preference to those of Siegel. decreased evoked potential (Redding 1967). In this study reticular stimulation 3.5.3. PHYSIOLOGICAL ACTION OF HIPPOCAMPAL EFFERENTS enhanced evoked potentials in both preparations. Physiological studies of the hippocampal efferents have been largely concerned with These findings indicate that the hippocampal effect is a mixed one, being the role of the hippocampus in the control of sensory inputs. These include studies on predominantly excitatory via rostral thalamic structures and predominantly inhibitory the effects of hippocampal stimulation on sensory-evoked potentials, as well as via more caudal reticular structures. A subsequent study demonstrated that these studies correlating changes in hippocampal activity with changes in evoked effects were mediated by the fornix (Redding 1969). Lesions of the mid-line thalamic potentials. Appendix Table A3 lists these studies. There is good, though not total, nuclei block the enhancement, while the locus of the effect appears to reside in the agreement that hippocampal upper layers of the cortex (Parmeggiani and Rapisarda 1969), again suggesting an indirect effect through the reticular system. One study looking at the long-term effects of fornix lesions on the auditory evoked potential reported an enhancement (Ungher, Rogozea, and Sirian 1971). In a related type of study Pond and Schwartzbaum (1972), Schwartzbaum and Kreinick (1973), and Schwartzbaum (1975) found that the late components of the visual evoked potential were decreased during lever-pressing associated with hippocampal theta activity, as compared with non-theta related behaviours. We shall discuss the relationship of hippocampal slow waves to behaviour shortly.∗ Finally, one other study of the effects of hippocampal stimulation deserves mention, that of Cazard and Buser (1958, 1963). They reported an effect of extreme importance: long-term changes in visual and auditory evoked ∗ A phenomenon which might be related to these controls over sensory input is the invasive hippocampal rhythm (IHR); this is a theta rhythm recorded from the cortex during periods of large amplitude, well-synchronized hippocampal theta. It has been reported to occur, for example, during the conditioned stimulus prior to the onset of an avoidance response (Pickenhain and Klingberg 1967, Fig. 5) or prior to a lever press (Yoshii, Miyamoto, and Shimokochi 1965, cited in Yamaguchi et al. 1967). Yamaguchi et al. have studied this phenomenon in curarized cats. It occurs primarily in occipital and temporal cortex (i.e. above the hippocampus) during high-amplitude, regular theta and is associated with a regularization and slight increase of the evoked potential. The authors maintain that the IHR is not due to volume conduction from the underlying hippocampus. They note that (1) the important feature for the appearance of cortical IHR is the regularity, not the amplitude, of the hippocampal theta, (2) the cortical distribution is limited by bipolar recording, and (3) FIG. 11. Schematic representation of the efferent connections of the hippocampus: a, Zimmer path; there is sometimes a difference in frequency between the cortical and hippocampal rhythms. These reasons do b, mossy fibres; c, Schaffer collaterals; d, pre-commissural fornix; e, medial forebrain bundle; f, not seem particularly compelling; to rule out volume conduction one must either have maps of the potential post-commissural fornix; g, mammillo-thalamic tract. showing that it changes in a way consistent with a cortical, and not a hippocampal, generator and/or single neurone activity lawfully related to the potential. It is quite common for a penetrating electrode to record a cortical theta rhythm which is totally attributable to a generator in the hippocampus. Cortical units that fire in stimulation enhances one or more components of the evoked potential in many areas. synchrony with the cortical theta rhythm have never been observed, although such 'theta' units are easily Effects of hippocampal stimulation have been found in the olfactory bulb, the detected in the hippocampus (O'Keefe 1976, and unpublished observations). Thus, while not denying the primary and secondary visual and auditory cortex, several specific sensory thalamic validity of changes in cortical activity during hippocampal theta, it seems unlikely that a cortically generated theta rhythm has been demonstrated. These criticisms apply equally to those studies which have demonstrated nuclei, mid-line thalamic nuclei, the hypothalamus, and the cerebellum. Several theta rhythm in brain-stem studies without mapping the depth profile or recording single units (e.g. Le Moal and studies have investigated the mechanisms involved in these effects, and the consensus Cardo 1975). Winson (1974) has shown that theta recorded in the brain-stem is attributable to a generator in the is that one pathway involves the projection through the fornix to anterior and mid-line hippocampus. thalamus and to reticular formation. The enhancement of the late components of the John O'Keefe & Lynn Nadel potentials following multiple pairings of visual and auditory inputs preceded by You may copy or reproduce any part of this file for teaching purposes or personal hippocampal stimulation. A tetanus to the hippocampus was usually followed after use. The original text and figures should not be altered in any way. Permission 45-60s by an increase in cortical evoked potentials, particularly in motor cortex; this should be obtained in writing from one of the authors if all or part of any of the increased to a maximum after several minutes. The phenomenon did not always figures or text is to be used in a publication and the source should be occur, and in any particular acute rabbit could only be elicited two or three times. In acknowledged: chronic animals repeated elicitation of the effect over a period of several days resulted in an increased evoked potential to the sensory stimulus alone for several John O'Keefe & Lynn Nadel (1978) The Hippocampus as a Cognitive Map , weeks. The investigators controlled for most of the obvious possible sources of Oxford University Press. artefact, including hippocampal seizures. It would be interesting to see if the pairing was essential, or if the phenomenon, like the facilitation seen in perforant path and You may redistribute the file electronically providing you do not modify it in any Schaffer collateral synapses (see p. 125), is the result of repeated stimulation alone. A few studies have looked at the effects of hippocampal stimulation upon motor activity, and these and other effects resulting from hippocampal stimulation, will be discussed at greater length in Chapter 12. In summary, the hippocampus has extensive connections, either directly or indirectly through its connection to the subiculum, with other limbic nuclei and to sensory and motor areas in the brain-stem, thalamus, and entorhinal cortex. It can modify sensory inputs and effect motor outputs. In addition, it can probably mobilize the pituitary-adrenocortical steroid system under conditions specified later (see pp. 357-62).∗ ∗ Perhaps the most widely known hypothesis linking the hippocampus with other brain areas, particularly limbic areas, is the Papez circuit (Papez 1937). Papez was impressed by the fact that emotions continued after the eliciting factor was removed and opined that this might depend on neural activity circulating within a limbic loop. Since he believed that the cingulate cortex was the cortical receptive area for the emotions, the circuit he proposed was as follows: cingulate cortex-cingulum-hippocampus-fornix-mammillary bodies-mammillothalamic tract-anterior thalamus-cingulate cortex. Parmeggiani, Azzaroni, and Lenzi (1971) have recently studied the transformation of repetitive activity through various parts of this circuit and suggested that it does, in fact, serve as a feedback circuit for the hippocampus. Some of the recent anatomical work cited in this chapter casts doubt on the existence of the circuit, at least in the form envisaged by Papez. The cingulate cortex does not project to the hippocampus, but the anterior thalamus does project to the presubiculum. The hippocampus does not project to the mammillary bodies directly, but does so by way of the subiculum. One new version of this circuit might be: anterior thalamus-presubiculum-mammillary bodies-anterior thalamus. Of course, there are countless circuits which one could imagine through these structures given the freedom of four or five synapses. At this stage of our understanding of the neural structures involved it may not be a very profitable exercise.

Source: http://www.cognitivemap.net/HCMpdf/Ch3.pdf