General Introduction

CHAPTER 1

General Introduction

1.1. General introduction 13

1.2. Animal models of aging 14

1.2.1. Rodents 14

1.2.2. Nonhuman primates 15

1.2.3. Tree shrews 16

1.3. The hippocampal formation 16

1.3.1. Anatomy of the hippocampal formation 16

1.3.1.1. Macroscopy of the hippocampal formation 17

1.3.1.2. Microscopy of the hippocampal formation 17

1.3.2. Connections of the hippocampal formation 19

1.3.2.1. The intra-hippocampal pathways 19

1.3.2.2. Afferent pathways 21

1.3.2.3. Efferent pathways 21

1.4. Unbiased methods of neuroanatomical quantification 21

1.4.1. Estimation of total neuron number 23

1.4.2. Estimation of total volume 23

1.5. Aging-associated morphological changes in the hippocampal formation

24

1.5.1. Hippocampal neuron numbers 24

1.5.2. Hippocampal volume 25

1.6. Hippocampus-dependent memory 27

1.6.1. Cognition and the hippocampal formation 27

1.6.2. Hippocampus-dependent memory and aging 28

1.7. Aging-associated changes of neurotransmitter systems in the hippocampal formation

29

1.7.1. Glutamate 30

1.7.2. GABA 30

1.7.3. Acetylcholine 31

1.7.4. Serotonin 32

1.7.5. Noradrenaline 32

1.8. Hippocampal vasculature 33

1.8.1. Hippocampal vasculature in various mammals 33

1.8.2. Anatomy of the blood-brain barrier 35

1.8.3. Innervation of the cerebrovasculature 36

1.8.4. Aging-related changes of the blood-brain barrier 36

1.9. Aims and outline of this thesis 38

1.1. General introduction

Over the past few years, the world’s population has continued on its remarkable transition path from a state of high birth and death rates to one characterized by low birth and death rates. Kofi Annan, UN Secretary-General, said on the International Day of the Older Person at 1 October 1998, when the United Nations launched The International Year of Older Persons 1999: “We are in the midst of a silent revolution, one that extends well beyond demographics, with major economic, social, cultural, and psychological and spiritual implications.” In calling the year 1999 The Year of the Older Person, it is recognized and reaffirmed what demographers and many others have known for decades: our global population is aging at an dramatic rate: one out of every ten persons is now 60 years or above; by 2050, one out of five will be 60 years or older; and by 2150, one out of three persons will be 60 years or older. Over the last half of the 20^thcentury, advancing knowledge of hygiene and biomedicine added 20 years to the average life expectancy. Moreover, the aged population is getting older. The oldest old (80 years or older) is the fastest growing segment of the older population: they currently make up 11% of the 60+ age group and will grow to 19% by 2050. Considering the fact that the majority of health care costs are produced within the last five years of living on average, the effect of progressive aging of the population will have serious consequences for the health care system. In 1999, the Netherlands spent 36 billion Euro for health care, which is 9.6% of the gross national product. In 1998, 44% of the total health care costs were related to persons older than 65 years, and these costs will increase to 61% during the next decades. Nevertheless, the changing demographics contributes with approximately 50% to increased costs for health care, another major factor of increased health care costs is the development and usage of new medical technologies (Wagemakers, 2003).

Aging not only affects a whole population, it also affects every individual. As we age, we are more susceptible for diseases, such as heart disease, osteoporosis, diabetes, arthritis and illness in general. Just as practically all organs, the brain is not spared during the aging process. Cognitive performance is usually declining with advancing age. Aging can be considered as the major risk factor for neurological diseases, such as dementia, Alzheimer’s disease, Parkinson’s disease, and depression.

Whereas the age-associated neurological diseases are being more and more understood, the progress in understanding the fundamental aging processes that underlie these pathologies is relatively slow. This thesis will provide a contribution to the understanding of the process of normal brain aging.

A general definition of the term “aging” is the uninterrupted temporal process of normal development that eventually leads to a progressive decline in physiological function and ultimately to death. According to this definition, we start to age since the moment we were born. However, development or maturation is a normal part of this aging. To avoid confusion, the terms “aging” and “aged” in this thesis refer to senescence rather than to development or maturation. The term “age-related”, which often refers to development, is avoided; “aging-related” is used instead.

1.2. Animal models of aging

Mammals have been widely used in biological research, because of their obvious similarities in both structure and function to man. Rats, mice, guinea pigs, and hamsters are in favor because of their small size, ease of handling, and high reproductive rate.

However, no single animal model can ever duplicate the original human condition (i.e., the model is never the same as the prototype), and models never provide final answers, but only offer approximation.

There are several aspects to be considered with various animal models of aging: appropriateness as an analog, genetic uniformity of organisms, background knowledge of biological properties, cost and availability, degree of generalization of the results, ethical implications, housing availability, size of the animal, numbers needed, and life span.

It should be kept in mind, however, that the validity of extrapolation of animal studies to the human condition is further complicated by the question: to which humans? Humans are, of course, highly variable in genetic, cultural, dietary and environmental influences, whereas such a high variety is encountered to a considerably lesser extent in laboratory animals.

Probably the best way forward for gerontological research is to move away from single-species studies, because comparative studies are likely to provide the best insight into the process of aging.

1.2.1. Rodents

The rat has been the species utilized in the majority of studies on the anatomy, physiology, and behavioral functions of the brain. The maximum life span of a rat in the laboratory is in the range of 4 years, being about¹/25of the human life span. Because of the short life span and the relatively low housing costs, it is quite practical to conduct aging studies with rats. Moreover, a whole series of behavioral tasks has been developed and extensively evaluated. However, it should be noticed that laboratory rat species are night-active animals. In many laboratories, the light-dark phases in the animal facilities correspond to the outdoor situation. Therefore, most behavioral test with rats are done in the light phase, i.e. in the rats’ inactive phase, which probably has confounding effects on outcomes. In addition to this complication, there are some essential differences between the rodent and human brain in general, in particular in the hippocampal formation, as pointed out in section 1.3. Furthermore, the rodent’s brain generally still develops after birth, stronger than in nonhuman primates and humans (Bachevalier and Beauregard, 1993).

There are more than 200 major rat strains, which can be grouped into inbred and outbred strains. Rats from a certain inbred strain are genetically virtually identical (e.g. Fischer 344). The amount of genetic variation present in any given colony of outbred rat strain (e.g. Wistar, Sprague-Dawley, and Long-Evans) depends on its history. This disadvantage has to be considered against the advantage of lower costs for outbred strains than inbred strains.

Laboratory mice have a maximum life span of approximately 3 years, being

1/30of the human life span. There are over 450 different strains of inbred mice, and it may be difficult to select an appropriate strain. Nine mouse strains are so-called senescence- accelerated mouse (SAM) strains, and the common aging characteristic of SAM strains is senescence acceleration after normal development and maturation. These SAM mouse strains are not considered in this thesis, because most, and perhaps all, animal models showing “accelerated aging” are more likely to be useful for pathology than for mechanisms of aging (Harrison, 1997).

1.2.2. Nonhuman primates

Among experimental animals, the rhesus monkey (Macaca mulatta) is probably the species closest to the human in which it is practical to conduct a full range of neurobiological studies (Rosene and Van Hoesen, 1987). Because aged nonhuman primates show many of the aging-related neurobiological alterations that have been reported in aging humans, rhesus monkeys are considered to serve as a useful model for normal human aging (Voytko, 1998). Rhesus monkeys in captivity can reach the age of 35-40 years, and may be considered as old when they are between 20 and 25 years.

Rhesus monkeys over 28 years of age are commonly regarded as the oldest of the old (Moss et al., 1988; Voytko, 1998). Therefore, the life span of rhesus monkeys is thought to be approximately¹/3of that of humans.

Because of their relative longevity, keeping rhesus monkeys into middle or old age is related to high costs for caring. Moreover, aged rhesus monkeys mostly have been subjected to some kind of experimental manipulation. Therefore, it is practically impossible to conduct sophisticated aging studies with naive old rhesus monkeys.

Nevertheless, due to recently established primate brain banks, which constantly grow by the collection of –mostly fixed– brains or brain samples from monkey colonies, possibilities arise to conduct aging-related studies and obtain worthwhile data, although some sensitive neurochemical assessments would fail on e.g. immersion-fixed tissue or brain material that has been stored in fixative over years.

The order Primates is divided into groups of species which are similar to each other. The two major groups are Prosimians and Anthropoids (including all monkeys, the apes and humans). The group Anthropoids, or Simians, is further divided into New World monkeys (e.g. the common marmoset), Old World monkeys (e.g. rhesus monkeys and other macaques), and apes (e.g. gibbon, orangutan, chimpanzee, gorilla, and human).

The term “primate” generally includes Prosimians and Simians (monkeys, apes and humans), but in this thesis “primate” will be used only for the group of Anthropoids, simply because not much neurobiological research has been carried out with Prosimians. “Nonhuman primate” as used in the current thesis excludes humans from the term “primate”.

1.2.3. Tree shrews

Originally regarded as primitive primates (Le Gross-Clark, 1956), today tree shrews are considered as an intermediate between insectivores and primates and are placed in the separate order Scandentia (Martin, 1990). A recent analysis using a relational map of mammalian brain architectures supports the identification of Scandentia as a separate mammalian lineage (Clark et al., 2001). The tree shrews’ natural habitat are forests and plantation areas in South-East Asia. Tree shrews reach maturity after about 3 months of age. The life span of tree shrews under laboratory conditions is in the range of 10 years, being approximately ¹/8 of the human life span. In addition, tree shrews from one breeding colony probably show less inter-individual variability than rhesus monkeys and humans.

1.3. The hippocampal formation

1.3.1. Anatomy of the hippocampal formation

Of all the cytoarchitectonic structures in the mammalian brain, the hippocampal formation is perhaps the most conspicuous, because of its unusual and striking macroscopic and microscopic appearance. The hippocampus obtained its name from its macroscopical appearance, resembling a seahorse (=hippocampus in Latin). The hippocampal formation is a phylogenetically old part of the cerebral cortex, also called archicortex. Much of the available information on the hippocampal formation's anatomy dates from the classical Golgi studies of Ramón y Cajal (1911) and Lorente de Nó (1934).

The hippocampal formation is present in all mammalian species. It has developed from a single cortical plate in amphibia into a rather complicated, three- dimensional convoluted structure in mammals. The volume of the hippocampal formation relative to the volume of the whole brain decreases in a phylogenetic sense.

However, the relative decrease of the hippocampal formation is not caused by a reduction in its size, but by the enormous increase in the size of the neocortex (see Fig.

1.1). With all phylogenetic comparisons, it should be kept in mind that statements are not based upon true phylogenic data, derived from examining a genealogical series of related organisms (Rosene and Van Hoesen, 1987). In order to make solid interspecies comparisons of the size of the hippocampal formation, body size must be taken into account, because small animals have large brain-body ratios whereas large animals have small ratios. This is accomplished by allometry, allowing to compare species which are members of different systematic groups. Allometry studies the change in proportion of various parts of an organism as a consequence of its growth. The hippocampal formation is an allometrical progressive brain region, in that there is an increase in its size relative to body weight during evolution (Stephan, 1983; West and Schwerdtfeger, 1985; Rosene and Van Hoesen, 1987). This relationship is not readily apparent, because the neocortex is even more progressive in the allometric sense.

Any difference in the morphological neuronal organization between the brain of animals and the human brain is supposed to be related to functional differences.

At least in the primate, brain structures of phylogenetically progressive size are apparently also progressive with respect to the development of their fiber connections (Schwerdtfeger, 1984).

1.3.1.1. Macroscopy of the hippocampal formation

In many mammals, if not in all, the hippocampal formation appears as an elongated, C-shaped brain structure (Fig. 1.1). During phylogenetic development from insectivores to primates, the topographical intracerebral position of the hippocampus has undergone remarkable changes, in that the mass of the archicortex is found more and more ventrally in the depth of the temporal lobe. This should be seen as the result of the development of the corpus callosum and the neocortex (Schwerdtfeger, 1984).

In the rat brain, the hippocampal formation extends from the septal nuclei of the basal forebrain rostrodorsally, bends over and behind the diencephalon, to the beginning of the temporal lobe caudoventrally (Amaral and Witter, 1995). The long axis of the hippocampal formation in the rat brain is therefore referred to as septotemporal axis. In the tree shrew brain, the long axis of the hippocampal formation is tilted ventrorostrally when compared to the rat (Schwerdtfeger, 1984). Also the tree shrew’s hippocampal formation bends around the diencephalon, but already great parts of the hippocampal formation are situated in the temporal lobe. The long axis of the tree shrew hippocampal formation can be called the dorsoventral axis, because it is almost vertical.

The folding of the hippocampal formation into the temporal lobe is, compared to the rat and tree shrew, more pronounced in the marmoset monkey, and even more prominent in the rhesus monkey. Finally, in the human brain, much of the hippocampal formation lies in the floor of the temporal horn medial of the lateral ventricle (Amaral and Insausti, 1990).

The temporal and septal parts of the rat hippocampal formation are probably equivalent in function to the anterior and posterior parts of the primate hippocampal formation, respectively (Colombo et al., 1998; Moser and Moser, 1998). Considering the phylogenetic shift in the hippocampal position, it can be suggested that the ventral and dorsal parts of the hippocampal formation of the tree shrew correspond with the anterior and posterior parts of the primate hippocampal formation, respectively. In primates, the anterior end of the hippocampal formation bends medially, giving rise to a bulging in the temporal lobe known as the uncus (Rosene and Van Hoesen, 1987; Insausti, 1993).

1.3.1.2. Microscopy of the hippocampal formation

The combined results from studies in various species of rodents and primates indicate that the cytoarchitectonic and histochemical appearance of the hippocampal formation is basically similar. Transverse sections (except those from the most anterior/ventral and posterior/dorsal parts) of the hippocampal formation, whether it be from the rat, tree shrew or primate, reveal a highly comparable image (Fig. 1.2A).

The hippocampal formation consists of the subiculum, the hippocampus (also called hippocampus proper, Cornu Ammonis, or Ammon’s horn), and the dentate gyrus. These three components have one cell layer encapsulated between two polymorph layers, which is the general organization of the so-called allocortex. The polymorph layers contain dendrites and axons, glial cells, and interneurons.

The general organization of the subiculum corresponds to the three layers of the allocortex. The hippocampus can be subdivided into five layers, from the lateral ventricle to the hippocampal fissure: the alveus (containing axons of subicular and hippocampal neurons), the stratum oriens (mainly containing the basal dendrites of the pyramidal cells), the stratum pyramidale (having the somata of the pyramidal neurons), the stratum radiatum and the stratum lacunosum-moleculare (both contain the pyramidal neurons’ apical dendrites). The dentate gyrus is usually divided into three layers: the molecular layer (having the unidirectional dendrites of the granule cells), the granule cell layer (consisting of the granule cell bodies), and the hilus of the dentate gyrus (containing the granule cells’ axons and many interneurons). The hippocampus can be subdivided into three subfields: the CA1, CA2, and CA3 areas (Amaral and

Figure 1.1

Schematic representation of the position of the hippocampal formation (gray) in the brain of a rat, tree shrew, marmoset monkey, rhesus monkey, and human. In all mammalian brains, the hippocampal formation is C-shaped. In the tree shrew, the bending of the hippocampal formation is orthogonal to the lateral plane.

With the development of the neocortex, the hippocampal formation shifts rostrocaudally towards the floor of the temporal lobe.

For each species, data are given of volume proportions of the hippocampal formation and the neocortex as percentages of the total brain. It should be noted that, in contrast to allometric studies, these data do not take into account the brain-body weight relation. Data of volume proportions are taken from Schwerdtfeger (1984).

Bar (rat, tree shrew, marmoset, and rhesus monkey): 1 cm, bar (human): 5 cm

Insausti, 1999; Amaral and Witter, 1995). The cell layer in the subiculum is continuous with that of the hippocampus. The dentate gyrus is discontinuous with hippocampus.

The principal cell type in the subiculum and hippocampus is the pyramidal neuron, the main cell type in the dentate gyrus is the granule cell. Apart from principal neurons, the hippocampal formation contains glial cells and various types of interneurons, which are situated throughout the hippocampal formation.

In this thesis, the term hippocampal formation comprises the subiculum, the hippocampus, and the dentate gyrus. However, different authors may include various components in this term. Amaral and Insausti (1990) and Amaral and Witter (1995), for example, also included the presubiculum, parasubiculum and the entorhinal cortex, because all these components are linked by unique and largely unidirectional projections. These fields are excluded from the term hippocampal formation in this thesis. A major justification for the exclusion is the controversy whether the pre- and parasubiculum have more than three layers, and the fact that the entorhinal cortex contains six layers. Therefore, in accordance with Rosene and Van Hoesen (1987) and Kloosterman et al. (2003), the present definition of the hippocampal formation comprises only the three-layered regions (a single neuronal layer with fiber or plexiform layers above and below the cell layer).

1.3.2. Connections of the hippocampal formation 1.3.2.1. The intra-hippocampal pathways

The intrinsic hippocampal connections have remained remarkably constant during phylogeny. The main input to the hippocampal formation runs from the entorhinal cortex layer II to the dentate gyrus molecular layer, which contains the dendrites of the granule cells (Fig. 1.2B). This connection is called the perforant pathway, for it

“perforates” the subiculum. From the granule cells, the information is passed to the pyramidal neurons in the CA3 subfield (Fig. 1.2B). The fibers from the dentate gyrus to the CA3 region are called the mossy fibers. The CA3 pyramidal neurons send out their axons to the CA1 pyramidal neurons (called the Schaffer collaterals), which on their turn project to the subicular pyramidal neurons (CA1-subicular pathway) (Fig. 1.2B). The main output of the hippocampal formation is from the principal subicular neurons to the entorhinal cortex layer V (Fig. 1.2B) (Kloosterman et al., 2003). (See Amaral and Insausti (1990) and Amaral and Witter (1995) for extensive descriptions.)

Rat studies have demonstrated commissural connection of two hippocampal pathways, namely of the dentate gyrus and the of Schaffer collaterals. Surprisingly, and in contrast to an expected increase of connections during evolution, the commissural connections of the monkey hippocampal formation are drastically reduced (Rosene and Van Hoesen, 1987). It may be difficult to understand this clear species difference, but it could reflect a growing functional association between the hippocampal formation and the increasingly lateralized cerebral cortex. Indeed, there is some evidence of functional lateralization of the cortex of rhesus monkeys (Hamilton, 1983; Vogels et al., 1994).

Furthermore, the right human hippocampus appears particularly involved in memory for locations within an environment, while the left hippocampus seems more involved in context-dependent episodic or autobiographical memory (Smith and Milner, 1981; Frisk and Milner, 1990; Burgess et al., 2002).

Figure 1.2

A Drawing of a transverse plane through the hippocampal formation, demonstrating the various subfields of the hippocampal formation. B Representation of the main intrinsic hippocampal connections (1-4) and the main hippocampal cell types (a-d). C Diagram of the main cortical input to the hippocampal formation, and a schematic diagram of the main intrinsic hippocampal connections (thick arrows). Dashed lines indicate reciprocal connections. Circular arrows in DG and CA3 indicate the presence of associational projections that link different levels of the fields.

Thin arrows indicate connections other than the major ones. The gray arrows between CA1 and the entorhinal cortex indicate connections that are present in primates, but not in other mammals. D Areas providing subcortical input to the hippocampal formation, and subcortical output areas.

Abbreviations: sub: subiculum; CA1, CA2, CA3: CA1, CA2, and CA3 subfields of the hippocampus; hil: hilus of the dentate gyrus; DG: dentate gyrus; h.f.: hippocampal fissure; L.v.:

lateral ventricle; III.v. third ventricle

1.3.2.2. Afferent pathways

Like the major intrinsic hippocampal connections, the major extrinsic connections have remained similar during evolution (Amaral and Witter, 1995). The major afferent input to the hippocampal formation comes from the entorhinal cortex (layer II) and terminates in the outer two thirds of the molecular layer of the dentate gyrus (Fig. 1.2C) (Nyakas et al., 1988). The entorhinal cortex projects in a lesser extent to the subiculum and hippocampus subfields CA1 and CA3.

The entorhinal cortex gets cortical information from the perirhinal cortex and the parahippocampal cortex (often referred to as postrhinal cortex in rats). These areas receive inputs from the neocortex (Fig. 1.2C). In primates, but not in lower mammals, the temporal neocortex connects directly to the entorhinal cortex, and also the subiculum and CA1 area receive direct input from the temporal neocortex (Schwerdtfeger, 1984) (Fig. 1.2C). Subcortical inputs to the hippocampal formation come from structures such as the amygdala, and the septal and basal nuclei (Fig. 1.2D) (Gaykema et al., 1990).

Cortical inputs are regarded to be the major information input, whereas the subcortical inputs probably are modulatory and associated with arousal and attention aspects of cognition (Amaral and Witter, 1995).

1.3.2.3. Efferent pathways

The entorhinal cortex (layer V) receives the main efferent connections of the hippocampal formation (Fig. 1.2C) (Kloosterman et al., 2003). Regarding the subcortical outputs, the hippocampal formation projects back to those structures that provide input to the hippocampal formation (Fig. 1.2D).

In tree shrews, as in rodents, there seems to be no direct fiber connection from certain areas of the temporal and frontal cortices to the hippocampal formation, whereas such connections were found in simian monkeys (Schwerdtfeger, 1984; Rosene and Van Hoesen, 1987) (Fig. 1.2C). The subiculum and CA1 region may be called phylogenetically progressive structures, not only in the sense of allometry, but also in view of their fiber connections to the temporal neocortex (Schwerdtfeger, 1984). Figure 1.2C also shows that in non-primate mammals, as well as in primates, the subiculum and CA1 area in general have more complex connections than, for example, the dentate gyrus and the CA3 area.

1.4. Unbiased methods of neuroanatomical quantification Many methods of neuron counting that have been used in the past and even today require assumptions about the structure of interest or the neurons to be quantified.

Assumptions will always cause biases that have varying and unpredictable effects on the results. These methods are nowadays thus called assumption-based methods. A study is biased when the mean of repeated measurements would not approach the true value.

Biases arise when only a limited number of sections or sites for quantification were chosen, maybe because such sections or sites have a particular appearance, or because it is a “tradition” to study, for example, the dorsal hippocampal formation of the rat.

Another cause of bias is when assumptions are made about the shape of the particles of interest, for example, the assumption that the particle (neuron) is spherical. Bias is also introduced when a relationship is presumed between the number of profiles of an object seen on a thin section and the number of objects within the structure that is sectioned.

The latter bias is easily understood when considering that larger neurons may appear in two sections, yielding two profiles, whereas smaller neurons only appear in one section.

Since the 1980s, a new generation of quantitative methods has emerged.

These so-called stereological techniques are design-based methods, which no longer need assumptions about the method or tissue. Therefore, any methodological bias in the results is eliminated. Whereas assumption-based techniques are based on sections, i.e. on

Figure 1.3

General procedure for stereological evaluations. A The complete structure of interest, for example the tree shrew hippocampal formation, is serially sectioned. Sections are systematically sampled for evaluation. The first section is chosen with a random number. B The total neuron number of the hippocampal formation can be estimated with the optical disector. The position of the optical disector frames is systematical but random. The sampling fraction is known. C The volume of the hippocampal formation is estimated with the Cavalieri method. A systematic point grid is randomly superimposed onto the systematic set of sections, and the number of point hitting the structure of interest is counted (SP). Each point p is associated with a known area a.

With the measured section thickness T, the volume is calculated from the total area and the section thickness.

planes or on two dimensions (2D), stereological techniques use three dimensions (3D) for quantification. The unique and most important feature of these 3D counting methods is that no knowledge or assumptions about the size, shape, or orientation of the objects being counted is required. Therefore, these techniques avoid the potential biases related to the structural geometry of the objects to be counted.

For stereological techniques the complete structure of interest must be available. After serially sectioning, systematic sections are sampled, of which the first is chosen randomly (Fig. 1.3A). This condition of systematic and random sampling must be met in all stereological procedures. Sections sampled as such are then used to estimate the total neuron number (Fig. 1.3B) or the total volume (Fig. 1.3C).

The methods of neuron counting and volume measurement are briefly described in the following. More detailed report are given by (West and Gundersen, 1990; Cruz-Orive and Weibel, 1990; West et al., 1991; West, 1993a, 1994, 1999; Geinisman et al., 1996; Howard and Reed, 1998; Hof and Schmitz, 2000; Perl et al., 2000).

1.4.1. Estimation of total neuron number

There are basically two methods for estimating the total number (est N) of objects: the optical fractionator technique and the NV × Vref method. The optical fractionator technique involves the determination of the number of objects (SQ^-) in a known fraction (f) of a structure, and multiplying this number by the reciprocal of the sampled fraction (1/f), to obtain an estimate of N using the equation est N = SQ^-× 1/f (Gundersen, 1986).

The second method is a two-step process that involves the estimation of both the volume of the structure (est Vref) and the volume density (est NV) of the neurons (Gundersen et al., 1988). The product of the two is an estimate of total number: est N = est NV× est Vref. The Nv´ Vrefmethod is usually preferred when the brain structure of interest, e.g. a human hippocampal formation, is too large for serial sectioning.

1.4.2. Estimation of total volume

The Cavalieri method to estimate total volume is extremely general and therefore applicable to almost anything. With this method, the volume of any object (est V) may be estimated from parallel sections separated by a known distance (T) by summing up the areas of all across sections of the object (SA) and multiplying this by T: est V = SA ´ T (Gundersen et al., 1988). The estimate is completely independent of the orientation of the set of sections and of the shape of the structure of interest. The area (A) per section can be obtained automatically during computer-assisted quantification, or by point-counting.

Point counting employs a sheet with regularly arranged points. Each point (p) is associated with a known area (a). The number of points (SP) covering the structure of interest is counted for all sampled sections, and the total area of interest is calculated as follows: SA = a/p ´ SP. Point counting of histological sections is usually done with a stereological program on a computer to which a microscope is attached. However, a transparent sheet, printed with the points, can be used to calculate the area of macroscopical tissue, for example of human cortex. The Cavalieri method can also be used on magnetic resonance (MR) images.

1.5. Aging-associated morphological changes in the hippocampal formation

1.5.1. Hippocampal neuron numbers¹

Numerous efforts have been made to answer the question whether neuronal loss in the hippocampal formation and entorhinal cortex can, at least in part, account for aging-related decline in cognitive processes such as learning and memory. In the past, most of the data were reported as neuron density per unit volume or area. This literature on neuronal numbers in e.g. the hippocampal formation shows much controversy.

Geinisman et al. (1995) gave an extensive review on neuronal number and densities in the hippocampus of aging subjects. Table 1.1 summarizes the results of some of these reports on neuronal numbers and densities in the hippocampus of aging humans and rhesus monkeys. Whereas some studies reported no aging-associated loss in human hippocampal fields CA1-3 and the subiculum, other studies demonstrated a loss of neurons in all or some of the hippocampal subfields or the dentate gyrus with advancing age. Most, but not all, of these earlier studies used assumption-based counting techniques. More recent studies on neuronal numbers have one thing in common: they aim at obtaining estimates of the total number of neurons in e.g. the hippocampal formation with design-based methods. Since 1984, when Sterio first described the optical disector (Sterio, 1984), assumption-free, unbiased counting methods were gradually introduced into the study of aging-related changes in neuronal numbers. Nowadays, design-based approaches to assess neuronal number in various fields of neurobiological research are increasingly appreciated over assumption-based studies. Especially in aging studies it is important to use design-based studies. For example, it was shown that brain tissue from younger subjects shrinks substantially more than tissue from older subjects during tissue processing (Haug, 1986), and assumption-based neuron counting would lead to a considerable underestimation in brain tissue from old subjects.

A number of reports used stereological techniques to assess hippocampal pyramidal neuronal estimates in human aging. West (1993b) investigated the hippocampal formation of 32 male human subjects (13–85 years), and reported a region- specific loss in the hilus and subiculum. Simic and colleagues (1997) performed a similar experiment with the hippocampal formation of 18 human subjects (aged 16–99). In agreement with West (1993b), they found an aging-associated decrease in the estimated neuronal number in the subiculum. However, the latter study was not able to confirm the loss of hilar neuronal number, as had been found by West (1993b). In contrast, Simic et al.

(1997) reported a loss of CA1 pyramidal neurons with advancing age, whereas West (1993b) did not (Morrison and Hof, 1997). This is surprising, because both studies used the Nv´Vreftechnique, and the individual estimated neuronal numbers were in the same range. However, this discrepancy may be explained by the fact that Simic et al. (1997)

1 Parts are taken from Keuker JIH, Michaelis T, de Biurrun G, Luiten PGM, Witter MP, Fuchs E: Methodological Considerations when Studying the Aging Process in the Nonhuman Primate Brain. In: Erwin JM, Hof PR (eds): Aging in Nonhuman Primates.

Interdiscipl Top Gerontol. Basel, Karger, 2002, vol 31, pp 76–101.

investigated the hippocampal formation of male and female subjects. Therefore, it may be possible that gender differences play a role in aging effects on hippocampal neuronal numbers. Apparently, in aging studies, it is important to control for gender (Passe et al., 1997), because aging effects in e.g. the temporal lobe have been shown to be more profound in men than in women (Bhatia et al., 1993). Another study of hippocampal neuron numbers with human aging investigated the hippocampal formation of subjects aged 46–85 and found no correlation in any hippocampal subfield with age (Harding et al., 1998). Although the NV ´ Vref method was used, this technique was used on cryosections instead of on plastic sections. Furthermore, the subjects were of both sexes, which might explain the contradiction to the results of West (1993b) and Simic et al.

(1997). However, it might also be that the lack of subjects in younger years (15–40) failed to demonstrate a subfield specific neuron loss with aging.

Aging effects on hippocampal neuronal number in nonhuman primates have been less well studied. An assumption-based study reported an aging-related neuron loss in the CA1 region of rhesus monkeys (Brizzee et al., 1980), whereas another non-stereological study of aging rhesus monkeys reported no significant difference in the CA1 region or the subiculum (Rosene, 1993). However, a stereological report showed preliminary data of no decrease of neuronal numbers in the CA1 region, nor in CA2+3, the hilus, and dentate gyrus, but a decreasing trend in the subiculum of aged rhesus monkeys (West et al., 1993). Another preliminary study did not find any loss of neurons in the hippocampal formation of aged rhesus monkeys (Berman et al., 1997). Whether or not nonhuman primate and human aging, with regard to hippocampal neuronal number, follows similar rules has still to be further investigated, preferentially with stereological methods.

Interestingly, the existing rodent literature shows with tremendous consistency that neuron numbers in the hippocampal formation do not decline with aging (Rapp and Gallagher, 1996; Rasmussen et al., 1996; Cimadevilla et al., 1997;

Calhoun et al., 1998).

Taken together, stereological reports so far were not able to demonstrate a loss of neurons in any subfield of the hippocampal formation of aging nonhuman primates. However, as described above, studies on neuronal number in the hippocampal formation during normal human aging show minor controversies. Since similar methods were applied, it is unlikely that the differences arise from the methodology. However, in human studies it is likely that large inter-individual differences in background and life experience cause deviating results. From this point of view, rhesus monkeys or tree shrews from a single colony may provide a good model to investigate aging-related changes in neuronal number in more detail.

1.5.2. Hippocampal volume

Reports that used assumption-based methods to quantify hippocampal volumes show inconsistencies. This is, at least in measurements in vivo, largely due to determination of hippocampal volumes by magnetic resonance imaging (MRI) on a single image scan.

SpeciesAgesRegionMeasureCountingmethodEffectReference Humann=1519-28and>65 yearsSubiculumNBiased;nocorrection29%loss§Shefer,1972 Humann=1847-89yearsHippocampusproperNVBiased;nocorrection5.4%loss/decade*Ball,1977 Humann=3021-91yearsDG,H1,H2,H3NVBiased,Abercrombiecorrection4.7%loss/decade*MouritzenDam,1979 Humann=649-77yearsCA1,CA2,CA3+4NBiased;nocorrectionNocorrelationwithageBrownandCassell,1980 Humann=1969-95yearsSubiculuma(n)/ABiased;correctionfor postmortemdecayandtissue shrinkage

10%loss/decade§Andersonetal.,1983 Humann=8615-96yearsH1NVBiased;correctionfor hemispherevolume,tissue shrinkageandseculartrends

3.6%loss/decade*Milleretal.,1984 Humann=3048-97yearsH1NVBiased;correctionfor hippocampalvolume5.2%loss/decade*Mannetal.,1985 Humann=234-98yearsCA1,CA2,CA3,CA4NVBiased;nocorrection3.8%loss/decade*Manietal.,1986

Human n=5/group

52-58and77-89 yearsDGNVBiased,Abercrombiecorrection32%loss*Bertoni-Freddarietal.,1990 Humann=126-87yearsSubiculum,CA1,CA2, CA3,CA4NVBiased;nocorrectionNochangewithageDaviesetal.,1992 Humann=3213-85yearsSubiculum,CA1,CA2+3, hilus,DGNUnbiased;NV´Vref7.3%and4.2%loss/decade* insubiculumandhilus, resp.

West,1993b Humann=1816-99yearsSubiculum,CA1,CA2+3, hilus,DGNUnbiased;NV´Vref8.0%and3.3%loss/decade* inCA1andhilus,resp.Simicetal.,1997 Humann=1246-85yearsSubiculum,CA1,CA2+3, hilus,DGUnbiased;NV´VrefNochangeswithageHardingetal.,1998 Rhesusmonkey n=10/group4-7and18-28 yearsCA1,CA2,CA3NABiased;nocorrection44%loss*inCA1Brizzeeetal.,1980 Rhesusmonkey n=4-5/groupYoungadulthood andsenescence (mid-twenties)

Subiculum,CA1,CA2+3, hilus,DGNUnbiased;opticalfractionatorNochangeswithage;trend towardsreductionin subiculum

Westetal.,1993 Rhesusmonkey n=?4-14and26-35 yearsSubiculum,CAfields, DGNUnbiased;NV´VrefNochangeswithageBermanetal.,1997 n-numberofindividualsexamined;DG-dentategyrus;CA1-CA4andH1-H5-hippocampalsubfieldsasdefinedbyLorentedeNó(1934)andRose(1938),respectively; N-totalneuronnumber;NV-numericaldensityperunittissuevolume;NA-numericaldensitypersurfacearea;a(n)/A-proportionoftissueareaoccupiedbyneuronalcell bodies;*statisticallysignificantdifference;§datawerenottreatedstatistically.AdaptedfromGeinismanetal.(1995)

Table1.1 Neuronquantificationinthehippocampalformationduringaginginhumansandrhesusmonkeys

However, MRI studies of the volume of the human hippocampal formation that use the Cavalieri method or voxel-based morphometry (a 3D method) reveal an evident aging-associated reduction (Jack et al., 1997; Schuff et al., 1999; Tisserand et al., 2000;

Bigler et al., 2002). It is important to include the complete hippocampal formation in the evaluation, not only with histological investigations, but also when using MRI, because Pruessner et al. (2001) demonstrated that the most anterior and posterior parts of the human hippocampal formation undergo the most prominent aging-associated volume decline.

Postmortem stereological analyses of hippocampal volumes of rodents, nonhuman primates and humans surprisingly do not reach a consensus. One study in cognitively normal humans between 16 and 99 years of age revealed a negative correlation between hippocampal volume and age (Simic et al., 1997). However, another postmortem study in humans of 45 to 86 years showed no aging-related change in total hippocampal volume (Harding et al., 1998). The number of stereological studies on aging-related changes in hippocampal volume of animals is surprisingly small and report conflicting data, as in human studies. A preliminary stereological study in rhesus monkeys suggests a negative correlation between hippocampal volume and age (Berman et al., 1997). No other stereological report could verify these results for nonhuman primates so far. Whereas rodent studies show a high consistency for preserved neurons with aging, it is not clear from stereological measurements of the hippocampal volume of aged rodents whether the hippocampal volume is decreasing. A general increase of the hippocampal formation was reported in aged Sprague-Dawley rats (Amenta et al., 1998), and a selective increase of the outer two thirds of the molecular layer of the dentate gyrus, the total molecular layer, the hilus, and regio inferior (corresponding to CA3) was demonstrated in old Fischer 334 rats (Coleman et al., 1987).

Old deer mice had a reduced hippocampal volume (Perrot-Sinal et al., 1998), whereas the CA1 and dentate gyrus of aged C57BL/6J (B6) mice were unchanged compared to young animals (Calhoun et al., 1998).

1.6. Hippocampus-dependent memory 1.6.1. Cognition and the hippocampal formation

Cognition describes a number of abilities, such as perception, attention, planning, evaluation, thinking, creativity, speech and reading, learning, memory, and association.

Humans undergo aging-related decline in several domains of cognition function. The focus in this thesis will be learning and memory, for which one of the reasons is the fact that learning and memory can be tested with relative ease in animal models.

Furthermore, the hippocampal formation is strongly involved in processes of learning and memory. The first important clue herefore came in the 1950s from the famous patient H. M., whose medial temporal lobes were removed bilaterally in an attempt to stop his epileptic seizures (Scoville and Milner, 1957). After the operation, H. M. experienced a severe anterograde memory impairment that has persisted to this day. The major

symptom is impairment at virtually any kind of learning task in which there is a delay between presentation and recall, particularly if interfering material is presented in between. H. M.’s impairment is mostly limited to his inability to register new facts in his long-term memory. Some of his spatial abilities are compromised, and some are preserved.

The medial temporal lobes and the hippocampal formation in particular have long been implicated in the acquisition of new memories (Squire and Zola, 1996), with visuo-spatial memory predominantly associated with the right (Smith and Milner, 1981) and verbal or narrative memory with the left (Frisk and Milner, 1990). There is now a consensus that the human hippocampal formation is involved in episodic memory, i.e.

memories that concern our ability to consciously recollect personally experienced events.

Also, the hippocampal formation of humans and animals is known to be involved in spatial or topographical memory. However, the hippocampus in both animals and humans has been ascribed a much broader role and involvement in other types of memory than merely episodic and spatial memory. Such a broader classification is

“declarative” memory, for all forms of conscious or explicit memory including episodic, semantic, and familiarity-based recognition. In contrast, the hippocampal formation is not needed for nondeclarative (procedural) or implicit memory.

Spatial navigation abilities can be tested in humans with, for example, a maze or a virtual reality (Lupien et al., 1998; Maguire et al., 1998; Iaria et al., 2003). With respect to animal studies, the two main tasks used to explore the spatial abilities of rats have been the radial-arm maze and the Morris water maze. A spatial tasks for nonhuman primates does exist (Rapp et al., 1997), but is rarely used. The effects of damage to the hippocampal formation on the ability of humans to navigate through a maze, on the ability of rhesus monkeys to remember baited food locations, and on the ability of rats to adequately perform the Morris water maze are remarkable similar (Colombo and Broadbent, 2000).

The strong link between the hippocampal formation and spatial navigation exists beyond mammalian species. Also in birds, the hippocampal region appears to be involved in spatial memory for the locations of stored food (Colombo and Broadbent, 2000). Moreover, there seems to be an association between the intensity of food hoarding and the volume of the hippocampal formation in birds. Furthermore, food-storing bird species perform better on spatial tasks than do species that do not store food (Lee et al., 1998; Pravosudov and Clayton, 2001).

1.6.2. Hippocampus-dependent memory and aging

Aging is accompanied by spared and impaired memory functions. Many studies in humans concern with the assessment of various forms of memory during aging. For example, hippocampus-dependent episodic memory is probably the mostly tested form of memory in aging human individuals.

Animal models serve a variety of purposes in the study of cognition, for example, they can be used for examining basic electrophysiological and biochemical mechanisms, and they can help identify brain structures involved in a certain type of

memory (Kloosterman et al., 2003). Moreover, animal models can help to increase the understanding of memory impairments with normal aging or neuropathological diseases, and these insights can be used to evaluate the potential of therapeutic approaches. Despite numerous advantages of animal models of cognitive aging, however, it is difficult to identify behavioral aspects in animals that are analogous to the human cognitive behaviors. Specifically, with animal models, episodic memory as tested in humans could never be mimicked, because such memory tests in humans are based upon verbal skills. For the sake of highest possible comparison of hippocampus- dependent memory performance in animal models of aging with that of aging humans, this section will be limited to describing hippocampus-dependent, spatial behavior.

Nonetheless, at least some respects of aging-related deficits of cognitive tasks that have been identified in both rodents and nonhuman primate species resemble the types of impairments observed in normal, aged humans.

Only few studies with aging humans concentrate on the ability to navigate in space. Still, it was shown that some aging humans had a lessened ability to find their way through a maze (Lupien et al., 1998). Similarly to aging-associated memory impairments in humans, behavioral assessments that are sensitive to integrity of the hippocampal formation reveal impairments in aged monkeys compared to young adults (Rapp, 1993), although individual differences in performance among aged rhesus monkeys have been reported (Rapp and Amaral, 1991; West et al., 1993). Cognitive studies in rhesus monkeys mostly aim at assessing memory types other than spatial memory.

Nevertheless, a study that assessed spatial information processing in rhesus monkeys revealed an impairment of spatial memory in aged animals, with no mention of individual differences (Rapp et al., 1997).

In agreement with primate studies, behavioral analyses of aged rats show aging-related impairment of hippocampal function. However, not all aged rats are impaired in spatial, hippocampus-dependent tasks, such as the Morris water maze or the radial arm maze. Part of the aged animals perform as good as young rats (Rapp, 1993;

Kadar et al., 1994; Rapp and Gallagher, 1996; Rasmussen et al., 1996).

1.7. Aging-associated changes of neurotransmitter systems in the hippocampal formation

The interactions of the various neurotransmitter systems in the hippocampal formation are numerous and not yet completely understood. The questions of which processes fail during aging and how do impaired transmitter systems affect other neurotransmitter functions is even harder to resolve. Even when considering a single neurotransmitter, several processes could be impaired during aging. For example, the synthesis of a neurotransmitter could be decreased, the transport of the neurotransmitter molecules to the axon terminals could be impaired, the release of the neurotransmitter from synaptic vesicles into the synaptic cleft could be hindered, the binding capacity or the number of (post-synaptic) receptors could be reduced, the receptor turnover rate can be changed,

receptor desensitization may occur, and the removal of the neurotransmitter from the synaptic cleft could be altered. However, lessened neurotransmitter contents might be compensated for, at least to an extent, by increased receptor binding potential or receptor numbers. Apart from altered chemical processes within neurons, neurotransmitter fibers themselves may undergo severe changes. Axons may degenerate and therefore (partly) deplete the hippocampal formation from certain neurotransmitters. In the following paragraphs, only changes of hippocampal afferents will be discussed. However, it should be kept in mind that the synthesis, transport, and release of neurotransmitters and the features of their receptors may also be altered during aging.

The principal cells of the hippocampal formation use glutamate as a neurotransmitter. The activity of the intrinsic hippocampal circuitry is modulated by other neurotransmitters, such as GABA, acetylcholine, serotonin, and noradrenaline. In the following paragraphs, the above-mentioned neurotransmitters systems in the hippocampal formation are shortly introduced. Swanson et al. (1987) and Kobayashi and Amaral (1999) gave extensive reviews on the distributions of these and other neurotransmitter systems in the rat and rhesus monkey hippocampal formation, respectively. The synaptic and non-synaptic interaction of these neurotransmitters are reviewed in detail by Vizi and Kiss (1998).

1.7.1. Glutamate

Glutamate acts as an excitatory neurotransmitter in the major pathways of the hippocampal formation (Fig. 1.4). Glutamatergic neurotransmission takes place almost exclusively across synapses. Principal neurons of layers II and III in the entorhinal cortex, the granule cells and the pyramidal cells of the hippocampus are all glutamatergic.

Principal cells of the hippocampus make also contact to GABAergic interneurons.

Relatively few studies have been conducted to study the relationship between glutamate and aging of the brain. However, consistent findings are a decrease in glutamate content and the decrease in the density of glutamatergic NMDA receptors with aging (Segovia et al., 2001).

1.7.2. GABA

GABA (g-amino butyric acid) is an inhibitory neurotransmitter. Virtually all hippocampal interneurons are GABAergic and they comprise about 10% of the total neuron population of the hippocampal formation. GABA participates mainly in synaptic interactions, i.e. it hardly acts by diffusion from varicosities. GABAergic interneurons mostly make synaptic contacts with principal cells (Fig. 1.4), however, to a smaller extent they innervate other interneurons. GABAergic interneurons that make synaptic contacts in the hippocampal formation are present in the hippocampal formation (ipsi- and contralateral), and in the medial septum and the diagonal band of Broca.

To date, it is not known with certainty whether GABAergic neurons disappear with aging. However, it is suggested that the GABAergic tone is compromised in aging, resulting in a loss of control of excitability and synchronization of the hippocampal principal cells (Vela et al., 2003).

1.7.3. Acetylcholine

Acetylcholine acts as an excitatory neurotransmitter. Cholinergic fibers in the hippocampal formation mainly arise from the medial septal nucleus and the nuclei of the diagonal band of Broca. A lesser portion of cholinergic innervation originates from the basal nucleus of Meynert. The varicose cholinergic fibers are widely distributed in the hippocampal formation, however, the distribution of fibers in rodents is slightly different from that in nonhuman primates and humans. Studies in rats showed that cholinergic fibers mainly terminate on principal neurons, and hardly on interneurons (Yamano and Luiten, 1989; Freund and Buzsaki, 1996). Acetylcholine is released from synapses and at nonsynaptic varicose sites (Fig. 1.4), however, it is not clear to which extent varicosities make synaptic contacts.

Figure 1.4

Simplified schematic diagram of the hippocampal architecture and its afferent pathways. Principal cells (white pyramidal cell) are glutamatergic; interneurons (white round cells) are GABAergic. A granule cell as the principal cell in the dentate gyrus receives glutamatergic input (perforant pathway) from the entorhinal cortex. A CA3 neuron as principal cell receives information from the mossy fibers. A CA principal neuron receives input from CA3 collaterals. Arrows indicate nonsynaptic release of neurotransmitter, small circles indicate synaptic transmission.

Abbreviations: CA: Cornu Ammonis; DG: dentate gyrus; Glu: glutamate; ACh: acetylcholine;

5-HT: 5-hydroxy-tryptamine = serotonin; NA: noradrenaline; GABA: g-amino butyric acid; M.S.

medial septum; R.N. raphe nucleus; L.C.: locus coeruleus.

Adapted from Vizi and Kiss (1998)

Acetylcholine has been ascribed an important role in processes of (spatial) learning and memory (Hortnagl and Hellweg, 1997). Cholinergic basal forebrain neurons have long been known to be susceptible to atrophy and degeneration with aging. This has been described for mice, rats, dogs, monkeys, and humans (Finch, 1993).

However, normal aging does not reliably produce a degeneration of the cholinergic innervation of the hippocampus. That is, aging does not necessarily cause death of cholinergic neurons, but it seems that the functioning of the cholinergic system is impaired during aging instead (Decker, 1987).

1.7.4. Serotonin

Serotonin (5-hydroxytryptamine, 5-HT) is predominantly an inhibitory neurotransmitter. Serotonergic fibers, originating from the raphe nuclei (Amaral and Cowan, 1980), spread all over the hippocampal formation. Serotonergic axons of the dorsal raphe nucleus are very fine and have small varicosities that do not make synaptic contacts (Fig. 1.4). Approximately 20% of the serotonergic fibers in the hippocampal formation make synaptic contact, mostly to GABAergic interneurons (Fig. 1.4). These fibers arise from the median raphe nucleus, and have large boutons that always make synapses.

The serotonergic system is known to modulate mood, emotion, sleep and appetite and thus is implicated in the control of numerous behavioral and physiological functions. Serotonin on its own seems to have no significant influence on learning and memory, however, the serotonergic system is shown to have strong interactions with the cholinergic system, also in the hippocampal formation (Steckler and Sahgal, 1995). The effects of aging on the serotonergic neurotransmission in the hippocampal formation have not been investigated extensively. Studies on serotonin content in the hippocampal formation of aging rodents either report a decrease (Luine et al., 1990), an increase (Van Luijtelaar et al., 1992) or no change (Miguez et al., 1999; Magnone et al., 2000; Stemmelin et al., 2000).

Studies on fiber densities during aging are rare and not comprehensive enough to make solid conclusions. Rodent studies indicate that serotonergic fiber densities decrease with advancing age in at least some regions of the hippocampal formation (Davidoff and Lolova, 1991; Van Luijtelaar et al., 1992; Nishimura et al., 1995).

1.7.5. Noradrenaline

Noradrenaline may have excitatory and inhibitory properties. Similar to the cholinergic and serotonergic innervation of the hippocampal formation, noradrenergic fibers are distributed throughout the entire hippocampal formation. The origin of the fine and varicose noradrenergic fibers, also probably in the rhesus monkey (Amaral and Cowan, 1980), is the locus coeruleus (Fig. 1.4). Noradrenergic varicosities of small sizes make synaptic contacts on GABAergic interneurons and on pyramidal cells. However, only a small part (15%) of the hippocampal noradrenergic varicosities form synapses, thus noradrenaline mainly acts by non-synaptic interactions.

The central functions of noradrenaline are regulation of alertness and of the wakefulness-sleep cycle, maintenance of attention, memory and learning, cerebral plasticity and neuroprotection. Studies on the content of noradrenaline in the aging hippocampal formation either demonstrate a reduction (Stemmelin et al., 2000;

Birthelmer et al., 2003) or no change (Luine et al., 1990; Miguez et al., 1999; Magnone et al., 2000). Although it was suggested that the aged primate brain is prone to degeneration of the locus coeruleus, stereological studies of the locus coeruleus of humans failed to demonstrate an aging-related neuron loss (Mouton et al., 1994; Ohm et al., 1997).

Nevertheless, as with the cholinergic system, a functional impairment may take place rather than loss of noradrenergic neurons.

1.8. Hippocampal vasculature

The brain is highly metabolically active, yet has no effective way to store oxygen or glucose, thus it depends on a large and stable blood supply. The human brain is approximately 2% of the body weight, but uses approximately 15% of cardiac output and accounts for 25% of oxygen consumption.

The vasculature supplying blood to the brain may be primarily distinguished into two categories, based largely upon structure and function: the arterial and capillary systems. The large arteries, as well as penetrating arterioles with smooth muscle cells, control blood flow and influence global perfusion. The smaller intraparenchymal vessels or microvessels, constituting precapillaries and true capillaries, lack smooth muscle cells and control permeability of water and nutrients. Capillaries and postcapillary venules are associated with pericytes. This association has been suggested to regulate endothelial cell proliferation, survival, migration, differentiation, and vascular branching (Hellström et al., 2001). However, cerebral pericytes have been shown to express contractile proteins (Bandopadhyay et al., 2001), although it is yet to be demonstrated whether pericytes have a prominent role in cerebral blood flow (CBF) regulation.

1.8.1. Hippocampal vasculature in various mammals

The blood supply to the hippocampal formation is rather similar across various mammalian species, concerning the external and internal arteries and the internal arterioles (Nilges, 1944; Goetzen and Sztamska, 1992). The arteries to the hippocampal formation arise from the posterior cerebral artery in a rake-like pattern. However, in the various mammals, the posterior cerebral arteries arise from slightly different maternal trunks. In the primate, vascular arcades have developed before the arteries branch into the internal hippocampal arteries (Nilges, 1944).

The hippocampal vascular system, including the microvasculature, is rather strongly developed in comparison to other cortical structures. Studies in the rat hippocampal formation demonstrated that the strata lacunosum-moleculare and oriens have a more extensive vascularization than the stratum radiatum, where the long axis of capillaries tends to be oriented parallel to the apical dendrites (Coyle, 1978). The capillary