Canonical changes during aging

Classification of dementias

Symptoms of dementia





1)    Clinical manifestations


2)  Cellular pathology


3)  Cholinergic deficits, relationship between cholinergic loss, pathological lesions and



4) Other neuropathological-neurochemical abnormalities


5) Progression of the disease


6) Pathology of the aging brain in relation to AD


7) Clinical-pathological correlations


8) Etiology and genetics of AD


9) Transgenic mice models


10) Pathogenesis


11) Therapy in AD




Foundation lecture 04/13/05





Despite individual variations in life spans and detail of aging, there is in most species an overall consistency in the characteristics of aging that can be described as a canoniocal pattern of aging.  For example, finite numbers of ovarian oocytes are found in virtually all mammals; an age-related loss of germ cells and hormone-producing follicles is the main cause of sterility at midlife. As a consequence of the depletion of hormone producing follicles, many female mammals experience accelerated osteoporosis, which is best seen in humans and laboratory mice and rats.  However, not all irreplacable cells are lost during aging. For example, the GnRH containing neurons in the hypothalamus, which are the proximal drivers of the ovulatory surges of gonadotropins, show no evidence of loss according to counting. Moreover, the obligatory neuron death during aging in the absence of Alzheimer’s disease (AD) continues to be controversial. Other canonic changes of mammals are the accumulation of lipofuscins or aging pigments in non-dividing cells, the decrease of striatal D2 receptors and the proliferation of smooth muscle cells in blood vessels walls. A major objective of biomedical gerontology is to identify the canonical age changes at molecular, cellular and physiological levels. From this slowly emerging normative or canonical pattern, it will become increasingly possible to construct powerful hypotheses about the interrelationships and causal chains.  Historically, it was thought that the primary contribution to age-related cognitive decline were massive cell loss. More recent studies suggest, that most age-associated behavioral impairements result from-region-specific changes in dendritic morphology, cellular connectivity, Ca2+ dysregulation, gene expression that affect plasticity and ultimately alter network dynamics of neural ensembles that support cognition.


1) Issues of neuron loss versus atrophy and astrocytic hyperactivity in evaluating molecular aging changes


It is well known that most neurons in the mammalian brain are postmitotic and therefore at risk for irreversible demage. Brain atrophy has long been accepted consequence of aging in humans. Originally depicted by gross brain weights and then by conventioanl X-ray and other tomographic techniques, it is amply confirmed that decreases in cranial volume occupied by brain parenchyma are accompanied by complementary increases in the volume occupied by cerebral spinal fluid. Although brain shrinkage was not originally believed to be selective for regions, longitudinal studies of aging subjects by CT revealed that cortical atrophy was selective and restricted to certain areas.


Brain Weight and Volume. The decline in brain weight with age is significant but the onset of reduction is unclear. The volume of the brain in the 8th decade is reduced by 6%-10% versus the third decade. The age related changes are more prominent in the frontal lobe which shows a 10% reduction in volume and 15% reduction in cortical thickness. The corpus callosum is decreased in volume by 12-17%, with accentuation in the anterior 2/5th because of the reduction of fronto-temporal interhemispheral fibers.


Neuronal Counts. The widely held public belief in the inevitability of neuron loss during normal aging is also being extensively revised. The gloomy estimate of 100,000 neurons lost per day does not seem to be as plausible as it once did in view of more recent sophisticated measurements. Neuronal shrinkage or atrophy could account for some of the confusion surrounding neuronal losses with age.


Cerebral Cortex. Total neuronal population is not significantly changed. However, there is a severe loss of neurons with more severe involvement of the frontal cortex. In aged-non human primates, there is a 30% reduction in neuron numbers in all layers in area 8A of the dorsolateral prefrontal cortex (Smith et al., 2004). The salient age-related change is shrinkage of large neurons between 10-35%, with consequently increasing numbers of small neurons. Aging predominantly involves the frontal lobes. The comparatively stable numbers of neocortical neurons in normal aging are in contrast to the extensive neuronal depletion in AD ranging from 40-60% associated with up to a 400% increase in astroglia. Nucleolar shrinkage probably represents reduced ribosomal RNA synthesis and possibly changes in ribosomal gene regulation. Substantial decline in the density of synapses, marked changes in dendritic architecture have been documented in the frontal lobe during normal aging (Peters et al., 1998).


Hippocampus. While older manual cell counts revealed neuronal losses of 20-30% in total H with one up to 6% cell loss per decade, more recent automatized studies showed only mild or no significant effect of age on pyramidal cell density. This is in contrast to a significant cell decrease in AD ranging from 19% in presubiculum to 44% in the CA1 and subiculum (S) associated with severe regression of dendritic extent. In the olfactory bulb (OB) there is a linear decrease of the mitral cells with age with occurence of NFT in 47% of individual over 60 yrs. The shrinkage of neurons in the neocortex and hippocampus is paralleled with a substantial decrease (20-40%) of the density of synaptic contacts during normal aging. Ultrastructural quantification using modern stereological methods documented that aging is associated with significant loss of synapses of entorhinal origin in the dentate gyrus (Geinisman et al., 1992). Because this circuitry conveys much of cortically derived information that the hippocampus uses to support learning and memory, it is reasonable to assume that the disruption of the entorhino-hippocampal connectivity might contribute to the cognitive decline of aging.

Interestingly, there is an increase (15%) of gap junctions between granule cells, and between neurons in the CA1 and CA3. The number of Schaffer collateral-CA1 synapses did not change. There is also a reduction in the postsynaptic density of the so-called perforated synapses in the CA1 area.


Subcortical nuclei. Striatum shows a decrease in total volume of 12% but no significant neuronal loss with age. The striatonigral DA system shows mild damage with neuronal losses particularly in the substantia nigra zona compacta (SNC) dorsal tier. In contrast, in PD mainly the ventral tiers are affected. Locus coeruleus (LC): after the age of 65 a total of 24-54% with predominant damage to rostral parts projecting to neocortex and hippocampus (H). In Alzheimer’s disease (AD) there is 40-80% cell loss. The serotoninergic dorsal raphe shows  little variation during aging. In contrast, in AD the depletion of large neurons ranges from 30-70%. For the cholinergic forebrain controversial data are available. While McGeer et al (1984) estimated a 70% loss of large cholinergic neurons, more recent studies did not show significant age-dependent variations in cortical ChAT activity. In AD, magnocellular NBM cell loss ranges from 15-70%.


As mentioned, a hallmark of aging in many brain regions is a progressive atrophy of neurons, but the relationship to neuron death is unclear. The nucleolar shrinkage in the remaining neurons of the SN in PD is particularly puzzling because lesions of this pathway induced hyperactivity in the remaining neurons in young rats. For example, nigral lesions increased the synthesis and release of DA at the terminals or increased TH. In regard to increased DA metabolism, the opposite changes of the TH-mRNA in its cell body and of DA synthesis and release at its striatal terminals imply a dichotomous regulation. It is possible that the efficiency of TH mRNA translation increased several fold to compensate for reduced mRNA.


The role of reactive glia in the aging brain is receiving much attention. Several reports show 10-30 % increases in glial cell mass or number with age in the rodent brain. Possibly, an increase in neural atrophy, resulting in a decrease in neuronal volume, is accompanied by a compensatory increase in the number or volume of glial cells. It is attractive to suppose that neuronal atrophy and loss might induce astrocyte (perhaps microglial) reactivity and proliferation during aging, yet the consequences of these events remain a matter of conjecture. After brain injury in adults, astrocytes remove debris and provide growth factors for neurite outgrowth.   Astrocytes may have a role in guiding axon growth which is pertinent to synaptic plasticity.  On the other hand, injury-induced reactive gliosis in the adult brain may impair neural function. Prolonged glial hyperactivity might result in physical barriers from glial scars. Both the protein and the mRNA for GFAP are increased in the hippocampus after deafferentation.


2) Neurogenesis in adult animals


Until recently, brain neurons were thought to be irreplaceable. The challenge to the ‘no new neuron’ dogma came first by the study of Altman and Das in 1965 in rats showing that newly born neurons appear in the dentate gyrus. Nearly 20 years later Nottebohm and coworkers discovered the addition of new neurons to the vocal centers in canaries.  Adult neurogenesis continue to be a very controversial topics. For example, is neurogenesis outside the dentate gyrus and olfactory bulb also in the frontal cortex in primates, etc. Apparently stress and depression suppress neurogenesis, whereas exercise, stimulating environment and antidepressant drugs give it a boost (Gage, Gould, etc).  Part of the controversy stems from the lack of unequivocal evidence for neuronal versus glial cell proliferation. Also, it is unclear how long neurogenesis lasts and how neurogenesis keeps in balance with neuronal death, preservation of memory traces, etc


3) Biophysical properties of aged neurons, LTP,  LTD


In all subregions of the hippocampus, most electrical properties remain constant over the lifespan. These include resting membrane potential, membrane time constant, input resistenance, threshold to reach action potential and the width and amplitude of Na action potentials. Numerous studies, however, have shown an increase in Ca2+ conductance in aged neurons. CA1 pyramidal cells in the aged hippocampus have an increased density of L-type Ca-channels that might lead to disruption in Ca-homeostasis, contributing to plasticity deficits that occur during ageing. Neurons in the prefrontal cortex (PFC) of aged monkeys, similar  to CA1-CA3 pyramids have a significantly  larger AHP compared to young neurons that relate in part to increases in Ca2+ conductance. There are subtle deficits in old aged rats in the induction or maintenance of LTP and LTD in the hippocampus (Burke and Barnes, 2006).


4) Age-related changes in gene expression


It is known that the maintenance of LTP requires gene expression and de novo protein synthesis. In behaviorally characterized rats, gene expression alterations in area CA1 were found to correlate with age-related cognitive decline. The behaviorally relevant upregulated genes included several that are associated with inflammation and intracellular Ca2+ release pathways, whereas genes associated with energy metabolism, biosynthesis and activity-regulated synaptogenesis were downregulated. Arc and Narp were two of the genes that were shown to be downregulated. Narp and Arc are effector IEGs. After transcription, Narp mRNA translocate to the synapse, where it is released and may act to cluster AMPA receptors on the postsynaptic membrane. After transcription, Arc mRNA localizes selectively to the region of the dendrite that receive the synaptic input that initiated transcription, and is proposed to be involved in the structural rearrangement of activated dendrites. The results should be interpreted with caution, however, as the data reflect resting levels of gen expression. In aged rats, granule cells of the dentate gyrus, but not the pyramidal cells of the CA3-CA1 area, have significantly smaller proportion of neurons that express Arc following spatial exploration. Also, in the CA1 area of young rats CA1 cells have higher levels of c-fos mRNA compared with old animals, although the same proportion of cells express c-fos (Burke and Barnes, 2006).


5)Dynamics of aged neural ensembles


It is widely agreed that modifiable neuronal ensembles support cognition. Therefore, alterations in these networks could be responsible for the behavioral impairments observed with ageing. In young rats, CA1 place fields expand asymmetrically during repeated route following. The magnitude of this place field expansion, however, significantly decreases in aged rats. It is likely that this age-associated reduction in experience-dependent plasticity is due to LTP deficits, as it does not occur when the NMDA receptor antagonist CPP is administered to young rats. In addition to age-related alterations in experience-dependent place-field expansion, the maintenance of place maps also differs between young and adult animals. For example, rats showing poor performance in a spatial task after returned to the familiar environment is correlated to retrieval of an incorrect map. A probable mechanism for map retrieval failures is defective LTP in aged rats.  In young rats, blockade of NMDA receptors or protein synthesis inhibition has been shown to result in map retrieval error when the rat is returned to the same environment (Burke and Barnes, 2006).


6) Impaired DNA repair in ageing


The possibility of age-related changes in the structure of genomic and mitochondrial DNA and the fidelity of its replication continue to be a very fruitful topics in molecular gerontology. The classic somatic mutation hypothesis of aging, now more than 40 years old, proposed that age-related accumulations of mutations in somatic cells account for the limit of life span (Hasty et al., 2003). Genetic defects in genome maintenance, driven by oxidative damage (see below) is a primary cause of aging. Mitochondrial DNA has a much higher mutation frequency than that of nuclear DNA in humans. Evidence indicates that mitochondrial DNA mutations cause deficiencies in respiration and ATP synthetase complexes.


As an example of somatic mutations, several laboratories have shown the spontaneous curing of a germ-line recessive mutations in the Brattleboro (B) rat strain; here a frame-shift mutation that prevents processing of the VP (vasopressin) prohormone. During aging, there is a progressive increase in cells containing normal protein. At 1 month of age, only 0.1 % of the neurons produce normal VP, but by 20 month of age, approx. 3% of the total VP neurons making the protein. The reverse of this germ-line mutation in B rats occurs through somatic mutations that correct the frameshift and appear to occur through mini-rearrangement of the 3’ region of the G deletion.


7) Accumulation of damaged proteins


The simplistic theoretical hope that aging might be accounted for by randomly accumulated errors in macromolecular synthesis is clearly not supported. However, some evidence suggests that the rates of RNA and protein synthesis are altered during adult life.  An example of germane to the neurobiology of aging is the B-amyloid precursor protein mRNA (APP). Three differentially spliced forms of APP mRNA are identified by the lengths of the polypeptides that each encodes: 695, 751 and 770.  The APP 751 transcript has drawn particular attention because it includes a Kunitz protease inhibitor motif; inhibition of protease activity could be a factor in the accumulation of B-amyloid plaques in the brain. Neurons in the cortex and hippocampus contain the 696 and 751 forms. Amyloid plaques are not unique to AD and occur with considerable variability in the brains of normal aged individuals. Possible increases in the relative prevalence of the 751 transcript in relation to plaques and NFT-s in AD are controversial.


The list of specific molecules that increase or decrease in the brain during aging is growing: As with cell atrophy, they are region and cell specific. For example, the hypothalamic content of POMC mRNA decreased by 30%, GFAP mRNA increases and the Thy1 antigen mRNA decreases in the hippocampus.


The search for other mRNAs that change during aging and AD has involved differential screening of cDNA libraries made from intact and deafferented rat hippocampus polyRNA.  These cloning strategies were designed to isolate mRNAs that are increased or decreased at a specific time after lesion.  After a change in a particular mRNA is confirmed by Northern blot, sequence analysis identified by the clone. By this approach 25 mRNAs were identified as responding to deafferentation. ApoE, alfa1tubulin, and synaptosome-associated proteins, Alfa1tubulin mRNA was also shown to increase in AD. In addition to these structural molecules, inflammatory mediator, transforming growth factorB1 increases in the hippocampus after EC lesion.  The appearance of vimentin and alfa1tubulin mRNAs has led to the hypothesis that proteins normally present only during development may be induced by degeneration.  MAPs are expressed in a specific sequence during development, in brains from AD patients, in the rat hippocampus after EC lesion and in reactive astrocytes. The MAPs are constituents of senile plaques (SP) and neurofibrillary tangles (NFT).   A decrease in the presynaptic growth associated protein GAP-43, a substrate for protein kinase C phosphorylation demonstrated in brains from patients with AD suggest that abnormal synaptogenetic responses are present in AD. 


8) Increased free radical damage during aging


Metabolic by-products called reactive oxygen species continually damage cellular macromolecules, including DNA. More than 100 different types of oxidative DNA lesions have been described. These lesions disrupt vital processes such s transcription and replication, which may cause cell death or growth arrest or may induce mutations that lead to cancer. Incomplete repair of such damage would lead to its accumulation over time and eventually result in age-related deterioration. A number of observations support the free radical theory, including the discovery that dietary restriction delays aging and extends life-span in wide range of rodents and other species, possibly by reducing free radical damage. The notion that genomic and mitochondrial DNA could be a major target of continual free radical attack over time is supported by the recent observation that genetic lesions accumulate with age and that dietary restriction reduces this accumulation in rodents (see review of aging research in Science, 299/5611, February 28, 2003). Among defenses against free radical damage, glutathione, the most abundant antioxidant, is hypothesized to maintain the native state of sulfhydryl groups. The nearly twofold age -related increase in the ratios of reduced-to-oxidized glutathione in skeletal muscles of senescent rats could favor oxydation of cystein residues. A shift in redox status is indicated by parallel twofold age-related increases in the ratios of NAD+NADH and NADp+:NADPH.


9) Slowing of axonal transport


Slower axoplasmic flow in peripheral nerves of aged rats is consistent with reduced rates of synthesis. Slow axoplasmic transport is responsible for the movement of cytoskeletal proteins within the axon. Observations that plasticity is impaired in the aging rodent brain and may be aberrant in AD suggest that slow axonal transport may play a role in these events.



10)  Effects of hormones: pituitary, ovary, adrenal cortex


Adrenal corticosteroids are also implicated in neurodegenerative changes during brain aging, particularly in the rat hippocampus. Among others, the age -related loss of large neurons with receptors for corticosteroids and the hyperactive glia in the hippocampus are retarded by chronic adrenalectomy and are prematurely induced in young rats by sustained exposure to corticosteroids. These findings suggest that under some circumstances, sustained stress has adverse effects on brain neurons.

                Certain mutations (Propdf and Pit1) impede pituitary production of GH, TSH and prolactin; reduce growth rate and adult body size and increase life-span by 40-60%. Without GH, the synthesis of circulating IGF-1 is suppressed (insulin-like growth factors). Reduced insulin/IGF activity is commonly associated with extended longevity of nematodes, flies and mice (Tatar et al., 2003). Estrogen and plasticity. Estrogens promote axonal and dendritic plasticity in the hippocampus of male as well as female brains (McEwen et al., 1997). In female mice, ovariectomy severely impairs the reactive hippocampal synaptogenesis that follows entorhinal damage. This effect is reversed by estrogen replacement. Postmenopausal estrogen deficiency may thus suppress the potential for neuroplasticity.




11) Animal models of cognitive and neurobiological aging              


 Rats in the Morris water maze (MWM) learn the escape location in relation to the configuration of cues surrounding the testing apparatus and this capacity requires the functional integrity of the hippocampal formation. During aging the performance in the MWM is declining, although there is a substantial variability among the aged rats and about half of them learn as well as young adults.


Whereas the hippocampus is crucially involved in spatial and episodic memory, the prefrontal cortex (PFC) is necessary for working memory (WM) and executive functions. WM can be measured using the delayed non-matching to sample task. Aged rats and non-human primates show time-dependent deficits on DNMS. Executive functions are measured with the Wisconsin Card Sort Task. Aged humans are impaired on the WCST and make more perseverative errors.







Definitions. Dementia usually denotes a clinical syndrome composed of failing memory and impairment of at least one other cognitive function due to chronic progressive degenerative disease of the brain. The involvement of multiple capacities distinguishes dementia from other disorders such as amnesia and apahasia, that affect a single functional domain (memory or language). The term includes a number of closely related syndromes that are characterized not only by intellectual deterioration but also by certain behavioral abnormalities and changes in personality. The symptoms are different on the basis of their speed of onset, rate of progression, severity or duration. There are several states of dementia of multiple causation and mechanism and that a diffuse degeneration of neurons, albeit, common, is only one of the many causes. Therefore it is more correct to speak of dementias. Most have an insidious onset and develop slowly over a period of many years. These include Alzheimer’s Disease (AD), Parkinson’s Disease (PD), Huntington’s disease (HD), frontotemporal dementia (FTD). Creutzfeldt-Jacob disease also develops insidiously, but is distinquished by a very rapid rate of progression, often spanning a year or less from the onset of the dementia to death. Multi-infarct dementia follows still another pattern. Initial cognitive symptoms develop acutely, but in this case the clinical course typically proceeds in a stepwise fashion over many years, with periods of relative stability punctuated by abrupt deterioration.


Cortical-Subcortical Dementia. In 1974, Albert introduced the term ‘subcortical dementia” to describe the clinical picture of intellectual impairment found in progressive supranuclear palsy (PSP). The dementia was characterized by forgetfullness, slowness of thought process or bradyphrenia, alterations in mood and personality (particularly apathia and depression), plus a reduced ability to manipulate acquired knowledge.  This clinical picture was contrasted with the frank amnestic, aphasic, apraxic and agnostic disorders of AD and other “cortical dementias” such as Pick disease. Subsequently, the concept of subcortical dementia broadened to include a variety of neurological disorders in which the primary pathology was believed to be subcortical. These included Huntington’s disease (HD), Parkinson’s disease (PD) and Wilson’s disease. While the pathology of PSP is almost exclusively subcortical, in both PD and HD significant cortical changes can be observed, particularly in patients with intellectual deterioration. On the other hand, subcortical pathology is known to be typical of AD. As Whitehouse (1986) points out, “the use of the terms ‘subcortical’ and ‘cortical’ tends to emphasize the independence of these regions in the brain”. In fact, the dense pattern of neuronal interconnections between cortical and subcortical regions suggests that the functional organization of the brain does not respect such conventional anatomical distinctions. Animal studies have shown that in functional systems which link cortical and subcortical regions, lesions at any level can produce seemingly identical behavioral deficits. Albert himself (1978) considered this issue and concluded that a better label for subcortical dementia may be “frontosubcortical dementia”. Indeed, when considering the cognitive impairments in AD, HD, PD and PSP, using simple neuropsychological tests, there is little consistent evidence for separating the four diseases into two broad categories of cortical and subcortical dementia. However, with the increased use of more sophisticated tests, the weight of evidence suggest that each disease has its own characteristic picture of impairment, with a small number of shared functions.


Symptoms of dementia


Memory: The gradual development of forgettfulness is a prominent early symptom. Proper names are no longer remembered, the purpose of an errand is forgotten, appointments are not kept, a recent conversation or social event has been forgotten. The patient may get lost, even along habitual routes. Day-to-day evens are not recalled. At the early stage, memory for the remote past is relatively preserved in AD, impairment is expressed in maintaining memories fro recent past. By comparison, patients with HD display a different profile in which declarative memory is relatively spared against a substantial impairment in procedural or habit-like memory that is thought to rely on cortico-striatal circuitry.


Visuo-spatial skills


Attention: the patient is easily distracted by every passing incident.


Language: Language functions tend to suffer almost from the beginning. Vocabulary becomes restricted and conversation, rambling and repetitious. The patient gropes for proper names and common nouns and no longer formulates ideas with well constructed phrases or sentences. Instead, there is a tendency to resort to cliches, stereotyped phrases which may hide the underlying defect during conversations. Subsequently, more severe degrees of aphasia, dysartria, palialia and echolalia may be added to the clinical picture.


Executive function (frontal lobe syndrome): Loss of  capacity of the individual to act purposefully, to think rationally, and to deal effectively with his/her environment. No longer is it possible for the patient to think about or discuss a problem with customary clarity, and he/she fails to comprehend all aspect of a complex situation.  One feature of a situation that some relatively unimportant event may become a source of unreasonable concern or worry. Judgement becomes impaired. As a rule, these patients have little or no realization of such changes within themselves: they lack insight


Emotional instability: Petulance, agitation, shouting, whining if restrained, loss of social grace, loss of the capacity to tolerate frustration and restrictions. The first abnormality may be in the nature of emotional instability taking the form of outbursts of anger, tears or aggressiveness, loss of the capacity to express feelings and impulses.


Progression of dementia. At certain phases of the illness, suspiciousness or frank paranoia may develop. Visual and auditory hallucinations may be added. As the condition progress, all intellectual faculties are impaired, but memory most of all. Patients may fail to recognize their relatives or to recall their names. Apractognosias may be prominent and these defects may alter the performance of the simplest task. There is a kind of psychic inertia. All movements are slow, sometimes suggesting an oncoming Parkinson’s disease. Sooner or later the gait becomes altered in a characteristic manner. In the later stages, physical deterioration is inexorable. Any febrile illness, drug intoxication, or metabolic upset is poorly tolerated, leading to severe confusion, stupor or coma, an indication of the precarious state of cerebral compensation. Finally, these patients remain in bed and succumb to pneumonia or to other intercurrent infection. Some patients, become virtually decorticate, totally unaware of their environment, unresponsive, mute and incontinent.  Naturally, every case does not follow the exact sequence outlined above. Impaired facility with language, in other impairment of retentive memory with relatively intact reasoning may be the dominant feature. Gait disorder may occur early, particularly in patients in whom the dementia is superimposed on Parkinson disease, cerebellar ataxia or ALS.







Alzheimer's disease (AD) currently afflicts over 4 million Americans and it is the eights leading cause of death. AD is the most common form of dementia, accounting for about 50% of all cases. The prevalence of the disease is tightly coupled to age and increases after the fourth decade of life. The prevalence of AD doubles approximately every 5 years after the age 60. The disease affects about 1% of persons aged 60-64 and up to 40% among those 85 years older.  Definitive diagnosis of AD is available only at autopsy, although the increased use of imaging and genetic methods promises an early diagnosis (DeKosky and Marek, 2003).

A variety of risk factors increase the likelihood of developing AD. Age is the most potent of the known risks. Female gender is also a risk factor; the ratio of affected woman to men is 1.2:1 to 1.5-1. A history of head trauma and a low level of educational attainment are additional risks. Genetic risk factors for the late-onset AD  have been identified. The most important type of these is the ApoE-4 allele. The lifetime risk of AD for an individual without the e4 allele is about 10%, whereas for an individual carrying at least one allelele is 30%. However, determining the ApoE genotype cannot be regarded as a diagnostic test for AD, since some individuals who do not bear the risk allelele develop the illness, and some who  have the allelel are spared the disease. Several causative mutations for AD have been identified. These are transmitted in an autosomal dominant fashion with complete penetrance; those inheriting the mutation will manifest the disease in the course of their lifetime. Mutations in the amyloid precursor protein gene (chromosome 21), presenilin 1 gene (chr. 14), or the presenilin 2 gene (chr. 1), produce familial AD. Inherited AD is rare, accounting for <5% of all cases. 

Numerous investigations have attempted to isolate the one aspect of the neuropathology that is most closely associated with the cognitive deficits. Such claims have been made for the amyloid plaques, neurofibrillary tangles, neuronal loss, synaptic loss and cholinergic depletion. Despite our increased understanding of pathological processes present in AD, the primary cause(s) and the pathophysiology is (are) not clear. Despite numerous attempts, a unitary theory that can account for all the clinical and neuropathological features has failed to emerge.


1) Clinical manifestations


AD occurs in middle or late life and is characterized by a progressive dementia. Typically, cognitive impairments appear insidiously. Some fluctuations may occur in individual patients, but more commonly, impairments in memory, language, praxis, visuo-spatial perception, judgement, and behavior become progressively more disabling (see Alois Alzheimer’s first description of the disease in 1907). Individuals with early onset disease usually have the most severe clinical, histopathological and neurochemical abnormalities.  Dynamic imaging studies of AD have demonstrated reductions in regional cerebral blood flow, oxygen consumption, glucose metablolism, and protein synthesis. Correlations have been demonstrated between results of these imaging studies and characteristics of neuropsychological deficits. For example, patients with language impairments exhibit reduced metabolic activity in the left parietal-temporal region. These focal signs have been attributed to an admixture of pathologies including lesions of subcortical neuronal systems afferent to cortex and disease of intrinsic cortical cells. The global dementia and diffuse reductions in electrical activity and metabolism occuring in later stages of the disease are believed to reflect bilateral distribution of these processes. Using PET and non-toxic tracers, B-amyloid can be visualized also in the living brain. Other studies suggest that with longitudinal MRI studies, shrinking hippocampal volume may be an early diagnostic sign of AD. Serum levels of B-amyloid elevated in the blood of people with APP or presenilin mutations, however the sensitivity and specificity of these tests has been limited.  Plate  summarizes the common types of dementing illnesses and their relative frequency. The onset of AD is not abrupt and patients pass through a phase of mild cognitive impairment (MCI) during which they exhibit cognitive deficits that are distinguishable from normal aging, but not meet the full criteria for AD.



2)      Cellular pathology


The current criteria for the pathologic diagnosis of AD require the presence of both neuritic plaques and neurofibrillary tangles. The pathologic diagnosis is confirmed when there are frequent neuritic plaques using the criteria developed by the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) and neruofibrillary tangles in an abundance grade by the Braak and Braak approach as stage V-VI. AD selectively affects several subcortical neuronal populations, including nerve cells of the basal forebrain cholinergic system, the brainstem locus coeruleus (LC) as well as certain neurons in amygdala, hippocampus and neocortex. These at-risk neuronal populations show several types of pathology, including neurofibrillary tangles  and neuritic or senile plaques.

Neurofibrillary tangles (NFT) (Plate ). These perikaryal (intracellular) inclusions are composed of densely packed fibers. Electron microscopically the tangles consist of  highly insoluble protein polymers appearing as 10 nm paired helical filaments (PHF), which are frequently associated with 15 nm straight and 10 nm neurofilaments. Immunocytochemical studies suggest that PHF are composed of a cytoskeletal protein, the microtubule-associated protein tau (MAPT) that is normally present in axons, but not in cell bodies. Tangle-containing neurons eventually die. Nearly all individuals above age 65 have a few NFT-s in their brains (Braak and Braak, 1996). Whenever a brain contains very few NFT, they are always confined to the hippocampus, the nucleus basalis Meynert, the amygdala and the entorhinal-transentorhinal area. This stage of NFT distribution may be associated with mild cognitive impairment (MCI), senescent forgetfulness and preclinical AD. Only after reaching high densities within limbic, paralimbic and inferotemporal areas, do NFT emerge in prefrontal, parietotemporal association areas involved in the neural control of language, attention and perception. The neuropathological criterion for a definitive diagnosis of AD is reached only when NFT clusters emerge in association neocortex (ADRDA-NIA criteria; Alzheimer’s Disease and Related Disorders Association-National Institute on Aging, 1997).


Normally tau binding to microtubules stabilize the axonal cytoskeleton. Phosphorylated tau cannot stick to microtubules. Tau –when differentially posphorylated –promotes the assembly, disassembly, and reassembly of microtubules, as needed by the cell. Apo-E3 (see below), by sticking to tau, may help keep phosphates away, while Apo-E4 fails to do so. This failure leads to a gradual accumulation of tau proteins with phosphates. These phosphates can cause tau to leave its post at the microtubule prematurely. Then, because tau dislikes the cell's watery environment, it may intertwine with another tau molecule to form paired helical filament. These filaments congregate to create NFT. The hyperphosphorylation of tau is achieved by the deregulation of the cell-cycle kinase Cdk5 (cyclin-dependent kinase 5). The physiological function of Cdk5 is not quite clear, it maybe involved in the regulation of actin cytoskeleteon, modulate signaling cascade (DARP32), control axonal transport, etc. 


Plaques (Plate) have a central core of amyloid protein surrounded by astrocytes, microglia an dystrophic neuritis containing paired helical filaments. These spherical foci, containing enlarged argentophilic axons/synaptic terminals/dendrites associated with extracellular amyloid are thought to represent sites of abnormal synaptic interactions. In AD and in aged primates, neurites in plaques are derived from a variety of transmitter systems, including cholinergic, catecholaminergic, and peptidergic neurons. Extracellular amyloid fibrils, the proteins of which are arranged in a predominantly beta-pleated sheet configuration, are important components of plaques. While most senile plaques are of the 'diffuse' type (not associated with altered neurites and reactive glia), some amyloid deposits in association cortices and limbic structures are much more fibrillar and compacted, with surrounding neuritic alterations together with activated microglia and astrocytes (these are the so called classical or neuritic or senile plaques=NP, or SP).

Amyloid β deposits (A β) are found in brains of both mentally intact aged and AD individuals with the same and relatively uniform patterns of distribution in both groups. The amyloid core of plaques can contain multiple species of amyloid B-proteins, including a form ending at amino acid 42 that is prone to aggregation (A β 42) and a slightly shorter species (A β 40) that is normally produced. Brains from patients with AD post mortem are characterized by an anatomically widespread process of amyloid deposition and senile plaque formation (also called neuritic or amyloid plaques). Neuritic plaque densities are highest in the temporal and occipital lobes and lowest in the frontal and limbic cortex. According to recent studies the plaques and their topographical and temporal time tables are independent of the tangle development (Wisniewski, 1996). A correlation between amyloid accumulation and dementia has been debated. A β can exist in various forms as monomers, dimmers, oligomers and highly compacted admixtures of fibrils and smaller aggregates (amyloid plaques).

In addition to the two classical histopathological features, the brains of patients with AD also feature a reduction in synaptic density, loss of neurons and granulovacuolar degeneration in hippocampal neurons.


3) Cholinergic deficits, relationship between the cholinergic loss, pathological lesions of AD and dementia


Loss of basal forebrain cholinergic neurons. Basal forebrain corticopetal cholinergic neurons (Plates ) develop tangles, and on the basis of studies in aged non-human primates, it has been suggested that cholinergic axons/nerve terminals in cortical target areas form neurites in some plaques. When these nerve cells degenerate, cholinergic markers (including presynaptic M2 receptors) are reduced in the amygdala, hippocampus and neocortex. Loss of these basal forebrain (BF) afferents to the amygdala, hippocampus (HI) and neocortex may result in decreased activation of target neurons and thereby, may contribute to some of the clinical manifestations of dementia.

Most studies show the presence of a very substantial but regionally variable loss of cortical cholinergic innervation in AD (Plates ). The severity of this depletion is greatest in the temporal lobe including its limbic, paralimbic, and association components. Thus, in advanced stages of the disease, temporal association areas (areas 20, 21 and 22) appear almost completely devoid of cholinergic fibers whereas the entorhinal cortex, hippocampus and basolateral amygdala retain a considerable residual density of fibres. By contrast the cholinergic fibers in the premotor cortex, the cingulate gyrus, the sensorimotor cortex, and in some of the frontal association areas seemed to be relatively well preserved. There is a general correspondence between the loss of neurons in specific Ch4 sectors and the depletion of cholinergic innervation in their preferential cortical targets. For example, there is a greater preservation of the cholinergic innervation of the cingulate gyrus, which is derived predominantly from Ch4am, when compared with that of the amygdala, which is derived predominantly from Ch4al. This corresponds to the fact that Ch4am is less affected than Ch4al. Basal forebrain cholinergic loss have been reported ranging in magnitude from 30-95% (Plate ), with a parallel reduction of ChAT activity (30-90%). Many of the remaining Ch4 neurons contain neurofibrillary tangles and shrinkage.     

Although the severity of dementia in AD correlates well with the cholinergic deficit, it has become increasingly accepted that AD is not primarily a disorder of central cholinergic neurotransmission, but that the disease involves deficits in multiple neurotransmitter systems and the impaired cognitive functions result from the disturbed interactions among multiple transmitter systems. Whether these profound deficits in cholinergic markers found in end-stage patients are also found in patients with much earlier disease is controversial. More recent studies, using the so called ‘unbiased’ stereological counting methods in patients with mild AD, suggest that there is only a limited loss of BFC neurons (Gilmor et al., 1999), in the early stages of the disease process (Plate ). Similarly, single-photon emission computed tomography (SPECT) studies using 123I-idobenzovesamicol (IBVM), an analogue of vesamicol that binds to the presynaptic vesicular acetylcholine transporter, suggest that cortical cholinergic terminals undergo only small age-dependent losses in normal aging, are depleted more in mildly demented AD patients when disease onset is earlier than age 65 years rather than later, and are not so devastated in AD as implied by postmortem determination of ChAT activity (50-80%).

Relationship between the density of plaques and cholinergic fibers. Although some cortical plaques contain degenerated cholinergic neurites, many plaques contain components from other neurotransmitter systems, therefore there is no one to one relationship between cortical plaques and degenerating cholinergic fibers in AD

Relationship between cortical cholinergic fibers and density of cortical tangles. There is a modest correlation between tangle density and percentage of cholinergic fiber loss. However, there are also discrepancies. For example, the auditory association cortex and the anterior cingulate gyrus contained an almost identical density of tangles, while the former displayed a much greater degree of cholinergic fiber loss.

Relationship between plaques and tangles and loss of BFC neurons. There is a correlation between Ch4 neuronal loss and plaques in anatomically related as compared with unrelated cortical sites (Arendt et al., 1987). Some presumably related cortical areas, however, showed little correlation with Ch4 loss. There is thus only partial support for the contention that cortical plaques by themselves cause retrograde degeneration of Ch1-Ch4 neurons or that pathologically altered Ch1-Ch4 neurons are responsible for the anterograde deposition of cortical plaques. There is no correlation between cell loss in Ch4 sector and the density of tangles in anatomically corresponding cortical areas.

Relationship between cholinergic pathology and cognitive changes A significant negative correlation has been reported between cortical ChAT activity and the degree of dementia (Palmer et al., 1987), whereas cortical levels of other neurochemicals, such as NE, do not appear to show a significant relationship with the degree of dementia. Despite these correlations, it is very unlikely that the cholinergic denervation constitutes the only or even the major determinant of the behavioral deficits in AD since virtually all sectors of limbic and association cortex also exhibit significant cell loss and contain a high density of plaques and tangles. A well designed study by researchers at the Mount Sinai School of Medicine (Davies, Haroutunian, and their colleagues, 1998-1999) compared the postmortem activity of the cholinergic marker enzymes in the cortex of nursing home residents with Clinical Dementia Rating Scale (CDR) scores of 0-5 with a cohort of elderly subjects without AD but with different CDR score.  According to this study the activity of ChAT in 9 neocortical regions did not differ significantly in subjects with CDR score of 0-2, but was significantly lower in subjects with severe dementia (CDR score 4-5). ChAT levels were significantly correlated with severity of neuropathological lesions of AD, as measured by density of neruritic plaques and neurofibrillary tangles. According to this study, although neocortical cholinergic deficits are characteristic of severely demented AD patients, cholinergic deficits were not apparent in individuals with mild AD and were not present until relatively late in the course of the disease (see summary of relationships between pathology, cholinergic deficits and dementia: Plate   ).



4) Other neuropathological-neurochemical abnormalities


Loss of striatal cholinergic neurons. There is no significant change in the neuronal number or perikaryal size of the cholinergic neurons of the dorsal striatum in AD. In contrast, there is a significant loss (74%) of the cholinergic neurons in the ventral striatum (Lehericy et al., 1989).


A cholinergic denervation within the dorsomedial thalamic nucleus has also been reported (Brandel et al., 1991).


Loss of brainstem neurons. A small number of mesopontine (Ch5 and Ch6 according to the nomenclature of Mesulam) cholinergic neurons of the brain stem in AD have been shown to contain neurofibrillary tangles (German et al., 1987; Mufson et al., 1988). In contrast to the BFC, however, these brainstem cholinergic groups show almost no neuronal loss in AD (Woolf et al., 1989).     In some patients, reductions occur in numbers of neurons in the noradrenergic locus coeruleus (LC) and in noradrenergic cortical markers.  The average loss of neurons of the LC in AD has been found to be either similar or greater than the average loss of Ch4 neurons. On the basis of cortical ChAT (choline acetyltransferase, the enzyme responsible for the acetylcholine syntesis) activity and LC neuronal loss, Wilcock and coworkers (1988) divided their AD patients into two groups. In a small subgroup, cortical ChAT activity was close to normal, while LC displayed a marked cellular loss. In the majority of AD cases a substantial loss was observed in cortical ChAT activity, while LC displayed a smaller degree of loss. Neuronal loss in LC has been found to be of greater magnitude within the dorsal aspects of this nucleus, which projects to the cerebral cortex, than within the ventral LC, which projecs to subcortical regions (Marcyniuk et al., 1986). Serotoninergic neurons of the raphe complex may also be affected.

The majority of investigators, report a greater loss of cortical cholinergic (50-80%) as compared with serotoninergic (40-50%) or noradrenergic (30%) innervation. The Ch4 cholinergic neurons display a larger and more consistent loss of neurons (30-95%) in AD when compared with the reported losses in the dopaminergic neurons of the mesencephalon (0-43%), the serotoninergic neurons of the raphe (0-39%) or the histaminergic neurons of the hypothalamus.

Amygdala. Neurofibrillary tangles, senile plaques, and neuronal loss occur in the amygdala and somatostatinergic neurons are one of the affected cell populations. These lesions of the amygdala may contribute to some of the emotional, motivational, and associative abnormalities occuring in patients with AD.

Hippocampus and associated circuits. Neurofibrillary tangles are common in hippocampus and associated regions, with hippocampal pyramidal neurons particularly showing evidence of several types of structural pathology, e.g. Hirano bodies, granulovacuolar degeneration, changes in dendritic arbors. NFT also occur in neurons of the subiculum and layer II of the entorhinal cortex.  In addition, plaques are common in these regions, and a variety of transmitter systems have been implicated in the formation of neurites. Excitatory amino acids may be important transmitters in some at-risk circuits in these regions. Lesions of these hippocampal systems are thought to be important in the genesis of some of the memory deficits occuring in AD. 

The tables in Plate  summarizes quantitative data showing cell loss in hippocampus in aging and AD. Plate  shows NFT tangles in layer II of the entorhinal cortex, the cells of origin of the major component of the perforant pathway that links the entorhinal cortex with the hippocampus. The lower photomicrograph of this plate shows with Alz50 antibody the degenerating terminals of the perforant pathway in the dentate gyrus.


Neocortex. The neocortex consistently shows tangles, plaques and reductions in numbers of neurons. Intrinsic somatostatinergic neurons develop tangles; their processes participate in the formation of plaques and counts of senile plaques correlate with reductions in levels of somatostatin (SS-like) IR. Reductions occur in SS, glutamate and M2 binding sites in AD cortex. More recent studies suggest that levels of CRF, a peptide present in some cortical neurons decreased and there is a reciprocal increase in CRF receptors. Plate 31 shows that NFTs are especially abundant in sensory association regions (like Brodmann’s area 21-22), but much more sparse in primary sensory regions such as the primary auditory cortex (Brodmann;s area 41-42.)



5) Progression of the disease. The primary target


The stereotypic involvement of specific neuronal populations and relative sparing of nearby or intermingled neurons in a number of system degenerations, including AD, Parkinson’s disease and other dementias suggests that progression of these diseases may involve transneuronal degeneration. In other words, irrespective of the primary causal factor, once a population of neurons has been destroyed, transneuronal degeneration of their input and target neurons is likely to contribute to the final pathological picture. The pattern of cell loss is dictated by connectivity. Studies comparing the location of NFTs and SPs in the neocortex  or in the entorhinal-dentate pathway and the hierarchical progression from the transentorhinal area, involving secondarily the hippocampus, and finally reaching the neocortex indeed suggest such a scenario (Braak and Braak). However, the successive involvement of cortical areas is not due to the simple spreading of the lesion from one area to its neighbor. A spared area may be located close to a severely affected one, that should have been lesioned if pathology had progresses by contiguity.

Virtually all the cell types lost in AD are known to be connected directly to tangle-rich regions of the association and limbic cortices or are themselves located within these regions. Ventrally located neurons in the LC that project to spinal cord are spared at the expense of dorsally and centrally located cortically projecting neurons. Within, cortically projecting cells, those with connections with frontal and temporal regions are more severely damaged than those with occipital connections. Dopaminergic neurons of the substantia nigra, projecting to the corpus striatum, are essentially undamaged in comparison with neighboring cortically projecting VTA neurons. Neurons of the midline thalamus projecting to the frontal cortex are damaged, whereas those of the dorsal and lateral nuclei which project to the tangle-free somatosensory cortex, are preserved. Finally, a greater loss of neurons from the nucleus basalis is seen in those regions projecting to the temporal cortex, where tangle densities are highest. Therefore, the cortex, especially the temporal lobe may be the initial site of the lesion with AD. This is strengthened by "longitudinal" studies of patients with Down syndrome in whom plaque development, tangles and particularly a deposition of B-amyloid protein are the earliest pathological changes found in the amygdala and hippocampus. It is proposed that changes within pathways bounded by the hippocampus and the amygdala later "spread out" by way of corticocortical or subcortical connections  to involve the neocortex and areas such as the nucleus basalis and LC (Mann et al). Longitudinal MRI study (Fox et al., 1996) of asymptomatic individuals at risk of autosomal dominant familial AD cases suggest that hippocampal atrophy developes in these subjects before the appearance of symptoms.

Perry et al. (1981) have provided indirect evidence indicating that the loss of cortical cholinergic innervation in AD occurs earlier than the loss of other neurotransmitter systems. Studying temporal neocortex, these investigators divided their cases into neuropathologically mild, moderate, and advanced groups on the basis of cortical plaque counts. In the mild group, only ChAT activity was significantly reduced but other markers were normal. In the moderate group, in addition to ChAT loss, losses in DBH (dopamine-beta-hydroxylase, the enzyme synthesizing noradrenaline) and GAD (glutamic acid decarboxylase, the synthesing enzyme for GABA) and an increase in substance P (SP) were found. Finally, in the severe group, an additional decrease was observed in cholecystokinin (CCK). Biopsy samples of the frontal lobe obtained within a year of the appearance of clinical symptoms have been shown to display up to a 95% reduction of ChAT activity (Bowen et al., 1982), providing further indication that the cholinergic denervation appears quite early in the course of AD.

Recent studies with transgenic animals tend to support the idea that the initial target in AD may be the synapses. Several electrophysiological studies of young mice transgenic for human APP with AD causing mutations have revealed significant deficits in basal synaptic transmission and/or LTP in the hippocampus well before the development of microscopically detectable BA deposits or plaque formation (for ref. see Hardy and Selkoe, 2002). This ‘synaptosis’ would be consistent with findings over the years that patient’s degree of dementia is more highly correlated with the loss of nerve terminals in their brain than with other pathological features, such as plaques or tangles, degree of neuronal loss or extent of gliosis (Terry et al., 1991). Interestingly, in some APP transgenic mouse lines, the number of synaptophysin-positive presynaptic terminals are 30% less than in non-transgenic controls at age 2-3 months, as there soluble BA levels rise, but before BA plaque formation begins  (Hsia et al., 1999). This animal work fits nicely with growing evidence that memory and cognitive deficits in MCI and AD patients correlate best with the soluble pool of cortical BA (Lu et al., 1999). It is particularly encouraging that single systematic injection of an antibody to BA into 22 month-old APP transgenic mice (Dodart et al., 2002; Kotiinek et al., 2002) reversed the memory-deficits in the absence of any change in the overall levels of BA deposits. Work with cultured brain neurons by Cotman and his colleagues also suggest that terminals may undergo a caspase-dependent degeneration without necessarily killing the entire nerve cells, at least not immediately.  Ultimately, as nerve cell loses more and more of its terminals, it will die, because it needs trophic factors (see later), produced by other neurons with which it connects, to survive. Thus there is a long period of neuronal dysfunction that may contribute to cognitive decline before there is frank neuronal loss. Such slow neuronal degeneration may not be considered classic apoptosis.



6) Pathology of the aging brain in relation to AD


The distribution of NFT and SP are independent from each other. Whereas neocortical NFT and neuritic plaques are usually absent in most mentally intact aged persons, BA deposits increase with age and affect up to 80%. BA and diffuse SP in the absence of neuritic pathology appear to be part of normal aging with no clinical significance. Whereas neuritic lesions in neocortex may be absent in 8-36% of demented individuals over age 75, most of them showing large number of NFT in the hippocampus.

Using both histochemical and biochemical methods, an age-related loss of cholinergic innervation of the hippocampus (50%) and entorhinal cortex (35%) has been reported, which became established after 80 years of age (Perry et al., 1992). Thus, there seems to be an age related depletion of cortical cholinergic innervation, but that it emerges late in life and is quite modest when compared to the magnitude of the loss in AD. There are, however, some similarities to the anatomical pattern seen in AD in that the cholinergic innervation of the cingulate gyrus appear to be spared and that of the temporal cortex affected most severely. Investigations of age-related changes in the cholinergic neurons of the basal forebrain have produced inconsistent results. While some studies found no age-related change in the number of Ch4 neurons, others report a 30-70% decrease in number and shrinkage of nucleolar volume, but not until the ninth decade of life (Lacalle et al., 1991). Individuals older than 60 years had an average 20% decrease in presynaptic terminals in the frontal cortex compared with individuals younger than 60 years. (Masliah, Terry).

It may be difficult to differentiate with any certainty, autopsy cases of early AD from mentally intact aged subjects and to determine whether the density and distribution of BA, SP and NFT in nondemented elderly are related to the normal aging process or to incipient AD. Apparently BA begins around 30 years before manifestation of cognitive deficits. At present, the borders between normative aging and AD are not clear, and the etiology of changes in the aging brain awaits further elucidation. Minimum cognitive impairment (MCI) is characterized by abnormal memory in patients who have otherwise preserved intellectual function and no disturbance of activities of daily living. MCI represent a transitional state between normal aging and AD.



7a) Clinical-pathological correlations. General


Impairment in memory may result, in part from a combination of lesions: deafferentation of the amygdala and hippocampus; pathology intrinsic to hippocampus; and abnormalities of circuits linking hippocampus to neocortex. Focal cortical signs (e.g., language abnormalities or visuospatial impairments) may be attributed to deafferentation of specific neocortical regions by brainstem and basal forebrain lesions admixed with intrinsic cortical pathology. Defects in judgement may be attributable to lesions in frontal cortex, while abnormalities in amygdala may contribute to some of the emotional problems occuring in these patients.           

The association between cerebral plaques and tangles, on one hand, and presenile dementia, on the other hand, was first suggested by Alzheimer in 1907. In 1968, Blessed and colleagues provided the first regression analysis between counts of neocortical plaques and psychometric tests, showing a strong correlation. When examined closely, however, the data used in their analyses reveal that many patients on their charts were either demented without plaques, that is did not have histologically diagnosed AD, or were not demented with relatively few plaques and again were not classifiable as having AD. Without those patients not having AD, their data points do not provide a strong correlation on reexamination (Terry et al., 1991). The modest correlation between tangle density and cognitive loss lends to support to the notion that although tangles are valuable markers of the disease and are probable major contributors to hippocampal-entorhinal malfunction, they are only minor players in neorcortical deficiency. On the other hand, recent studies of the Mount Sinai Medical School (Plates) tend to show a correlation between plaques, tangles and severe dementia. More recently, it has been emphasized the correlation between neocortical synaptic loss (using synaptophysin) and neuropsychological assays (Terry, Masliah). It is proposed that synaptic loss is the most powerful and ubiquitous proximal factor leading to dementia (Selkoe, 2002).

AD dissect human memory into distinct compartments (G. van Hoesen). One of the earliest changes in AD occurs in the hippocampus, particularly its subicular component. Other changes early in the illness affect the anterior parahippocampal cortices such as the entorhinal cortex. These lesions disconnect the hippocampal formation from the association cortices and probably underlie the early signs of AD such as changes in the anterograde compartments of the episodic (contextual) memory. Later changes affect the lateral temporal, posterior parahippocampal, temporopolar and anterior insular cortices further exacerbate the problem and contribute to retrograde memory abnormalities in the contextual realms. The involvement of the multimodal association cortices would seem to mark the point in AD when generic (semantic) memory decline, language, abstract reasoning are lost and where even self-identity is lost.


7b) Neuropsychiatric symptoms/clinical –autopsy findings/ involved neural networks


Apathy, depression, agitation. Apathy is among the most common symptoms of AD. Apathy is characterized by a lack of interest in usual activities, hobbies, loss of interest in social; engagements, etc. Depression: inability to experience pleasure and manifested by sadness, feeling of worthlessness, etc. Agitation: excessive motor activity associated with a feeling of inner tension.

Studies with SPECT and PET revealed diminished cerebral blood flow in the anterior cingulate regions in patients with AD and prominent apathy (Craig et al., 1996; Migneco et al., 2001). Histopathological studies have identified relationships between agitation and NFTs in the frontal lobe (Tekin et al., 2001), cholinergic deficits (Minger et al., 2000) and cell loss in the rostral locus coeruleus (Matthews et al., 2002). Also, neuroimaging reveals disproportionally reduced cerebral blood flow in frontal, temporal and parietal regions in patients with aggression and agitation (Hirono et al., 2000) and depression (Minger et al., 2000; Hirono et al., 1998; Starkstein et al., 1995). Quantitative EEG studies reveal increased slow wave posterior activity in patients with AD and depression compared with patients without depressive mood changes (Pozzi et al., 1993).


Psychosis. Is represented in AD by the occurrence of delusions or hallucinations that have their onset after the appearance of the dementia syndrome and are not attributable to some other psychotic disorder or to a substance-induced disorder.

Most quantitative EEG studies have found slowing in patients with psychosis compared with those without delusions and hallucinations (Edwards et al., 2000). Studies using SPECT have identified differences between AD patients with and without delusions. Mega et al. (2000) found lower perfusion in right and left dorsolateral prefrontal, left anterior cingulate and left ventral striatal regions. There were also reductions in the left pulvinar and the dorsolateral parietal cortex bilatereally. Psychosis has been associated with increases in SP in the medial temporal prosubiculum area and NFTs in the middle frontal cortex and with diminished neuron counts in the parahippocampal region of the medial temporal lobe (Forstl et al., 1994). An increase in M2 muscarinic cholinergic receptors is present in the frontal and temporal cortex of patients with AD and psychosis. The increased abundance of these receptors suggests that they have been upregulated in response to the degenerating presynaptic cholinergic system (Lai et al., 2001). This change may provide a basis for the utility of cholinomimetic compounds in the treatment of behavioral disturbances in AD (Cunningham, 2004).

Neural circuits. The pattern of distribution of neuronal and transmitter changes leads to dysfunction of neuronal systems. Two neural networks have been recognized as particularly important to the mediation of neruopsychiatric disturbances: the limbic system and the fronto-subcortical circuits. The limbic system is responsible for assigning the valence of emotional value to human experiences. The limbic system includes both cortical and subcortical structures integrated into a functionally related circuitry involved in emotional processing of sensory stimuli. Cortical regions included in the limbic system are: hippocampal formation, anterior hippocampal gyrys, cingulate gyrus, subcallosal cortical regions, temporal pole, insula and posterior orbitofrontal cortex. Subcortical structures include the amygdala, septal area and substantia inniominata. Five frontal-subcortical circuits linking frontal lobe regions to subcortical structures and back to the frontal lobe have been described (Alexander et al., 1986). Frontal-subcortical circuits mediate executive function (dorsolateral prefrontal-subcortical circuits), motivation (medial prefrontal-anterior cingulated-subcortical), and control of behavior in social contexts (orbitofrontal-subcortical) as well as motor and eye movement functions.


8) Etiology and genetics of AD


In the last ten years good progress has been made in elucidating the resulting changes of single-gene defects in familial forms of AD. Several highly penetrant genes have been cloned for rare, autosomal-dominant, early onset froms of AD. Plate summarizes the major gene defects, including mutations in the amyloid precursor protein and presenilins that lead to increased amyloid deposit and AD. The strategy has been to search for mutations in identified genes in the rare familial forms of the disease, in the hope that this might provide some insight into the much more common sporadic cases of the disease. Unfortunately, these gene defects affect only about 1% of all AD cases, however, several risk factors (ApoE4 and Alfa2 macroglobulin polymorphism) have also been discovered that results in AD in the non-familial form of the disease. It is assumed that in the late-onset forms of AD there may be multiple genes, each with alleles that are common in the general population, they have relatively weak effects on their own, they may produce more than one effect and they may interact with each other (Kennedy et al., 2003).

Amyloid deposits, composed of a 4kD BA (beta amyloid) protein derived from the amyloid precursor protein (APP). APP is a single transmembrane domain (TM) protein with multiple alternate transcripts, which are expressed ubiquitously and present in dendrites, cell bodies and axons. APP is coded by a gene located on the long arm of human chromosome 21. APP exists in three isoforms, 695, 751 and 770 amino acids. APP is synthesized in RER, glycosylated in the Golgi apparatus and delivered to cell surfaces as an integral membrane protein. Cell culture studies have implicated APP in having roles in cell survival, neurite outgrowth, neuroprotection, cell-substrate interactions, modulating LTP. Two isoforms (751 and 770) contain a domain that can act as a serine protease inhibitor (Plate ).

The normal metabolic processing of APP by three proteases generates both amyloidogenic and non-amyloidogenic products. An alfa-secretase acts at residue 612, within the BA sequence, to produce non-amyloidogenic, intramembranous 10-11kDa C-terminus product (p3=alfa-stub) and a secreted APP fragment (sAPP). The  Beta secretase (BACE) cleaves at residue 597 to release a truncutaed sAPP and a membrane-associated, 8-12 kDa C-terminus fragment (CT or beta-stub), containing the whole sequence BA.  The gamma secretase cleaves at various sites presumed to be integrated in the cell membrane to produce an extracellular BA fragment of 39-43 amino acids. (see for explanation Plates ). The larger BA42 peptides are more prone to self-aggregate and are thought to be more pathogenic.

The gamma secretase is a complex includes the presenilins, (Ps1 and PS2) and three other proteins (nicastrin=NCSTN; APH1A and PEN2).

The beta amyloid is implicated in the genesis of AD because all mutations that have been shown in AD increase BA42 deposit (Plates ). The BA peptide contains 14 amino acids of the transmembrane domain and 28 amino acids of the adjacent extracellular domain of APP. Amyloidogenesis could result from a variety of mechanisms, including gene mutations (716, 717, 692, 670/671), gene dosage, altered levels of APP transcripts/isoforms, and abnormalities of post-translational processing. The mutation of the APP gene is a rare cause of AD (6-7 families in the world!). Transgenic mice that overexpress the mutant human APP 695 gene, develop typical plaques as well as difficulties in learning at older age (Hsiao et al., 1996), suggesting that the increased production of BA may be especially prone to formation of plaques. These data seem to confirm the amyloid-cascade hypothesis of Hardy (see below). BA is toxic to neurons in culture. Possible means of toxicity include: 1) disrupting Ca homeostasis (it could form Ca channels itself or increase membrane permeability); 2) inhibiting axonal sprouting and LTP; 3) reducing N/K ATPase activity; 4) altering redox potential; 5) interacting with p75 neurotrophin receptors; 6) disrupting mAChR coupling to G-proteins; 7) impairing glucose and glutamate transport (see Plate  for a schematic summary for B-APP metabolite-induced cellular toxicity).




The Presenilins. The presenilin genes (PS1 and PS2) are two homologous genes encoding polytopic 8 transmembrane proteins. The gene for PS1 was identified using a positional cloning strategy that led to the identification of missense mutation in a gene on chromosome 14 which encodes a 43-kDa protein (Sherrington et al., 1995). PS2 was identified on chromosome 1 on the basis of its homology with PS1; it encodes a 50-kDa protein. The sequence of both proteins predicts the existence of 8 TM domains that share more than 60% amino acid identity (Plates ). Like APP, they appear to be expressed ubiquitously. Although, interestingly, in human PS-2 is primarily localized in the dentate and hippocampal granule cells and in the deep layers in the frontal and temporal cortices. So far, more than 35 mutations, mainly involving the conserved TM domains or a region adjacent to a large intracytoplasmic loop, have been identified. All but one are missense mutations that appear to have a toxic gain-of-function effect (Hardy, 1997). Presenilin mutations cause an increase in the production of BA42 of about 2-3-fold. Mutations in the presenilin gene are thought to account for about 20-25% of all familial AD cases. Presenilin is cleaved to produce an N-terminal (28kDa) and C terminal (18kDa) fragment thought to be an endoproteolytic cleavage. Only the cleaved protein is functional, uncleaved form is degraded through the proteosome. Presenilin proteins have been localized to the smooth and rough ER and the Golgi.        


The function of the presenilins has been clarified unexpectedly by the finding in the C. elegans of a protein (Sel-12), with considerable homology to the presenilins. Sel-12 has been shown to be involved in the Notch signaling pathway. The Notch pathway is an evolutionary conserved cell-cell signaling mechanism involved in cell fate decisions during different cellular and developmental processes, including neural development. In vitro studies of neurons from mouse cerebral cortex revealed that the formation of neuronal contacts results in activation of Notch receptors, leading to restriction of neuronal growth and a subsequent arrest in maturity (Sestan et al., 1999). A critical step in Notch signaling involves intramembraneous cleavage of the protein to liberate the Notch intracellular domain (ICD) and the translocation of the ICD to the nucleus where it functions as a transcriptional coactivator.

Because mutations in Sel-12 can be rescued by the introduction of normal human PS1 or PS2, it seems likely that the presenilins serve the same signaling function in humans. The finding that the null mutations in the Drosophila melanogaster presenilin gene abolish notch signaling and nuclear translocation (Struhl and Greenwald, 1999) serves to emphasize this point. The deletion of PS 1 in mice greatly reduces gamma secretase activity and increases the accumulation of the alfa-stubs and Beta-stubs, the substrates for gamma secretase. This finding and the existence of two conserved aspartates in the presenilin protein (marked D257 and D385 in TM6 and TM7 domains) that are required for the gamma secretase cleavage of APP and the generation of the Notch intracellular domain, has led Selkoe and his colleagues to suggest that PS1 (AND PS2) are components of the the long-sought-after gamma secretase (Aguzzi and Haass, 2003). These studies show, that presenilins regulate both APP processing and Notch signaling by influencing unusual (intramembraneous) protein cleavage events (Plate). The variants of the genes for either of the presenilins change the function of the gamma secretase in a way that somehow leads to increased accumulation of  BA.

Mutations in PS1 and PS2 alter the way that cells handle APP processing. APP is folded and processed in the endoplasmic reticulum and Golgi for transport to the cell interior. PS1 allows misfolded APP to accumulate. Misfolding, in turn, might be what causes APP to be cut in the wrong place, releasing extra BA42. A selective and highly significant increase of BA42 occurs in the plasma and in media from cultured skin fibroblasts of patients with PS mutations. PS1 knocked out mice showed a dramatic defect in the somites (Sisodia et al). Transgenic mice expressing mutant PS1 also show increased BA42 levels in brain. Finally, direct analysis of the patients bearing PS1 mutations demonstrates a significant increase in the density of BA42-containing plaques compared to that found in patients with sporadic AD. The idea that all known genetic mutations causing familial AD are found either in the substrate (APP), or in the proteinase (PS1-gamma secretase), which are the rate-limiting proteins in the production of the amyloid peptide, is intellectually very appealing.  According to Wolozin et al (1996), PS2 mutation induce apoptosis in cultured nerve cells under conditions in which normal PS2 had no effect. The accelerated rate of neuronal cell death that occurs in AD may therefore result from the additive effects of aging, activated apoptosis and BA toxicity.


Interestingly, not all mutations in PS1 lead to dementia via changes in BA production. In one recently reported PS1 mutation (Gly183Val) was associated with Pick’s disease (frontal lobe dementia with taupathy) in the absence of amyloid plaques (Dermaut et al., 2004)


It is interesting to note that a double transgenic mice that overexpress mutated PS1 and APP genes showed deficits in the number of cholinergic synapses in the frontal cortex, implying diminished function of the cholinergic system. Nonetheless, no significant changes in the basal forebrain cholinergic neurons were noticed in these transgenic animals (Wong et al., 1999). Also, no changes in the cholinergic innervation of the entorhinal cortex was noticed – a cortical region that is prominently affected in AD. Interestingly, Kim found that the p75 neurotrophin receptor, that is specifically expressed in basal forebrain cholinergic (BFC) neurons and has been implicated in the regulation of cell survival and death, is a gamma-secretase substrate. This may be the molecular basis for selective degeneration of BFC neurons in AD.




Apo-E4. The gene encodes apolipoprotein E, a protein that binds to cholesterol, which among other functions, transport lipid (cholesterol) between cells. The gene is localized on chromosome 19. The common isoforms of apo-E are encoded by three Apo-alleles, E2, E3 and E4, which differ from one another as a result of amino acid substitutions at two specific codons. The apo-E4 allele is associated with a threefold increase of the risk for AD, meaning that heterozygotes are three times more likely to develop AD, than people carrying alleles for other isoforms of apo-E (Roses group from Duke University). Apo E is synthesized in astrocytes and apoE receptors are abundantly expressed in neurons. The biochemical effect of the Apo E4 protein seems to stabilizing fibrils of amyloid B protein. This stabilization makes it more likely that the B-amyloid fibril will aggregate, and thus initiate the disease process. Apo E is found in senile plaques and is also associated with NFT. By contrast, the E2 and E3 alleles are protective: apo-E3, but not E4, binds to tau; preventing its aggregation may well be related to this difference. The discovery of an increased frequency of the apo-E4 allele in late-onset cases of AD is of considerable importance because of its occurrence in both familial and sporadic forms of AD. For reasons that are not yet understood, apo-E4 appears to also influence the age of onset of AD in individuals with APP mutations but not in patients with mutations in either of the presenilins. E4 allele lowers the age of the disease onset rather than causes the disease. It is important to note, however, that not every individual with the E4 allele will develop AD, so although apo-E4 screening is a useful adjunct to diagnosis, it cannot be used as a reliable predictor of AD in any given case.



Alfa2macroglobulin binds to the same LRP (low-density lipoprotein receptor-related protein) receptor as apo-E4 and could prevent the formation of the insoluble B amyloid fibrils that are considered most toxic to neurons. Some people carrying apo-E4, even those with two copies, appear not to get AD in the presence of the ‘good’ alfa2M gene.         

Cholesterol homeostasis. Karen Duff and coworkers at Nathan Kline Institute in NY have shown that cholesterol is a major risk factor in amyloid accumulation and that pharmacological and genetic manipulation of cholesterol synthesis impacts the generation, transport or clearance of BA peptides (Refolo et al., 2001). Although in clinical AD cases, serum cholesterol level does not correlate with the disease, recent epidemiological studies showed that treatment with cholesterol-lowering statins lowered the risk of developing AD (Wolozin et al., 2000). In experimental animals, reduced cholesterol levels results in reduced B and gamma-secretase activity, and increased alfa-secretase activity, consequently reducing BA levels. Cholesterol bound to apo-E containg lipoprotein particles are secreted by astrocytes and taken up by neurons where they affect signal transduction through low density lipoprotein-related protein (LRP). Cholesterol binds also to several synaptic proteins and is necessary for the formation of synaptic vesicles and for the clustering of certain postsynaptic receptors. The LRP receptor related protein has been directly implicated in synaptic plasticity in hippocampal slices (Zhuo et al., 2000) and the apoE4 isoform is less able to promote neurite outgrowth than other apo-E isoforms (Nathan et al., 1994). (Plates )  New findings suggest a positive feedback in the sense that increased production of BA exacerbates lipid abnormalities.

Increasing  evidence suggests that LRP mediates the efflux of BA from the brain to the periphery. LRP is a  multifunctional signaling and scavenger receptor that can bind a variety of ligands including apoE, alfa2M and APP. LRP antagonists have been shown to reduce the efflux of  BA from the brain by 90%. Cerebral amyloid load was doubled in LRP knockout mice. BA bound to LRP are internalized to endosomes, and are then either delivered to lysosomes, where they are degraded or undergo transcytosis across the blood brain barrier into the plasma. Several other protease enzymes, including IDE (insulin degrading enzyme), NEP (nephrilysin), PLG (plasminogen) are important in BA degradation. The gene for IDE is located on chromosome 10, the NEP on chromosome 3, PLG on chromosome 6 (see Plate by Tanzi and Bertram, 2005).

The role of inflammatory molecules in the pathological process of AD is not fully understood, but they may be involved in a number of key steps in the pathological cascade. It has been shown that IL-1 (and possibly IL-6) can regulate APP synthesis. The promoter of the gene encoding APP contains both heat-shock and interleukin-responsive elements, and IL-1 expression is rapidly induced following brain injury and other pathogenic insults. This may trigger enhanced APP synthesis and BA42 deposition and set up a vicious cycle whereby amyloid deposits stimulate further cytokine production by microglia, leading to even higher expression of APP. While transient glial activation is beneficial as part of the organism’s homeostasis response to injury, abnormal glial activation results in a detrimental neuroinflammatory cycle. Activated microglial cells produce and release potential toxic products, including cytokines, proteases and free radicals, which could damage neuronal cells by 'bystander lysis'. For this reason, activated microglial cells might be the link between BA4 amyloid deposition and neuronal degeneration. It is possible that treatment with certain anti-inflammatory drugs may reduce the release of toxic factors produced by activated microglial cells but, at the same time, they might also affect the rate of aggregation of the BA4 peptide by shifting the balance between factors that promote and those that prevent aggregation. In fact, studies have shown that cyclooxygenase (COX)-2, an enzyme involved in inflammatory mechanisms as well as neuronal activities, is up-regulated in AD. Transgenic mice expressing both the human APP mutation (APPswe) and the human presenilin (PS1-A246E) mutation, with resultant AD type pathology, were crossed with transgenic mice expressing human (h)COX-2 in neurons. At 24 months of age the triple transgenic mice showed a >2 fold elevation of BA immunopositive plaques in frontal cortex and hippocampus, compared to APPswe/PS1-A246E double transgenics (Pasinetti, 2002).


9) Transgenic mice models

APP and PSAPP mice. Unbiased, stereological neuronal counting has established that APP and PSAPP mouse models, despite extensive amyloid deposition in some lines, do not have widespread neuronal loss as seen in AD. By contrast, evidence for progressive synaptic degeneration and dysfunction has been reported in multiple lines of APP and PSAPP mice, although marked variability is observed between different models. However, recent studies have shown that in older Tg2576 mice (21–25 months) synaptophysin immunoreactivity is reduced only around dense cored amyloid plaques and this is associated with electrophysiological changes indicating abnormalities in synaptic function. In addition, 3D multi-photon microscopy has been used to demonstrate altered neurite trajectory, dendritic spine loss and thinning of dendrites in a zone around the dense cored plaques in Tg2576 and PSAPP mice. One of these studies also reported dendritic breakage and axonal abnormalities around amyloid plaques in PSAPP double transgenic mice (4–7 months). Loss of corresponding pre-synaptic markers also implied that the plaques impacted pre-synaptic boutons and post-synaptic structures. Moreover, Aβ immunotherapy has also been shown to resolve at least some of the abnormalities in neuritic structure around plaques in Tg2576. These data imply that dense cored amyloid plaque formation is likely to contribute to the memory loss observed in AD and in older Tg2576 mice.

It is currently uncertain how the synaptic degeneration observed in the APP and PSAPP relates to the process of neurodegeneration in human AD, and whether this synaptic loss could be a precursor of neuronal cell death given longer time periods than are possible in a mouse. However, it is clear, from the absence of neuronal loss in APP and PSAPP models, that physiological Aβ accumulation and amyloid plaque formation do not cause rapid neuronal cell death in vivo. This in turn suggests that Aβ species are unlikely to cause widespread neuronal loss in the absence of secondary pathological events in human patients. This is consistent with human neuropathological studies that show no correlation between overall levels of amyloid deposition and clinical progression of AD.


MAPT pathology – link to neuronal cell death. To examine the role of MAPT in AD and other tauopathies, mouse models have been developed, through expression of MAPT mutant and wild-type transgenes, that recapitulate most of the features of human neurofibrillary pathology (NFT and neuropil threads).Transgenic mouse lines that develop robust neurofibrillary pathology, with both the structural and biochemical features of human NFT, also show significant neuronal loss in affected brain regions. An important recent example is the inducible rTg4510 model  which expresses MAPT with the FTDP-17 mutation P301L ( McGowan et al., 2006). This mouse develops massive neurodegeneration in AD-relevant cortical and limbic structures to the point where global forebrain atrophy is observed and brain weight is significantly reduced. Furthermore, the progression of neurofibrillary pathology and neuronal loss is also correlated with the progression of deficits in spatial reference memory as assessed by the Morris water maze. This and other MAPT transgenic mouse models have therefore shown that the development of neurofibrillary pathology and neurodegeneration are closely linked. The results in the MAPT mice are also consistent with the long established correlation between NFT pathology and the progression of memory deficits in AD.

The mechanism of neuronal cell death in mouse models that develop NFT pathology is uncertain, but most of the data suggest that apoptosis is not the primary mechanism of neuronal cell death.  Recent studies in mice expressing mutant MAPT transgenes have implicated axonal dysfunction and degeneration as an initiator of MAPT-induced neuronal loss.

Transgenic mouse with amyloid plaques and NF. Lewis and colleagues (2001) crossed mutant APP and MAPT (JNPL3) mice, and demonstrated that the double-transgenic TAPP mice developed enhanced limbic neurofibrillary pathology, when compared with single transgenic littermates. Importantly, the enhanced NFT pathology in the TAPP mice was also associated with evidence of neuronal loss in the entorhinal cortex. Similarly, Gotz et al. (2001) found that the intracranial administration of Aβ into mutant (P301L) MAPT mice resulted in the generation of NFTs within the amygdala. Results of these studies strongly suggested that Aβ accumulation accelerated the development of neurofibrillary lesions, although the exact mechanism of this interaction was unknown. Oddo and colleagues (2003) extended these studies by generating a triple-transgenic model (3×Tg-AD), harboring PSEN1M146V, APPswe and MAPT (P301L) transgenes. This model accumulates intraneuronal Aβ, and subequently forms amyloid plaques and MAPT lesions in an age-dependent fashion. In subsequent studies, Oddo et al. (2004, 2005) found that Aβ  immunotherapy (injection of anti-Aβ antibodies, or antibodies specific for oligomeric forms of Aβ) or a gamma-secretase inhibitor  in  3×Tg-AD mice led to the rapid clearance of accumulated Aβ deposition and early MAPT lesions in the cell body. The removal of both Aβ and MAPT lesions proceeded in a hierarchical, time-dependent manner; clearance of accumulated Aβ occurred before a reduction in the MAPT pathology. Furthermore, after clearance of the injected Aβ antibody, Aβ pathology re-emerged before the appearance of MAPT lesions. However, later stage hyperphosphorylated MAPT (antibody AT8 and AT180 positive) lesions were resistant to clearance by Aβ immunotherapy. These results suggested the existence of reversible and irreversible stages of MAPT pathology. The nature of this shift from reversible to irreversible pathological stages is unclear, but one obvious possibility is that MAPT hyperphosphorylation is associated with aggregation into filaments that renders the lesion resistant to clearance.

A complete mouse model of AD has proven elusive, probably because of the short lifespan of the mouse and the nature of the relationship between Aβ accumulation and other pathological features that contribute to the complex phenotype of AD. In particular, based on results from APP and PSAPP mice, it seems likely that Aβ accumulation does not lead directly to neuronal cell death in AD but usually requires the initiation of a secondary process. This explanation is consistent with human neuropathological studies that show no correlation between total amyloid deposition and memory loss in AD. By contrast, mice that develop robust MAPT neurofibrillary pathology almost inevitably develop widespread neurodegeneration, regardless of whether the neurofibrillary lesions are generated through expression of a mutant, wild-type MAPT transgenes or over-expression of a relevant MAPT kinase . These results from the transgenic mice clearly confirm the close relationship between MAPT and neurodegeneration and are consistent with human studies that have reported a correlation between NFT pathology and the clinical progression of AD. However, results from TAPP (MAPTP301L×APPswe) and the 3×Tg-AD mouse models support the hypothesis that Aβ accumulation can accelerate, if not initiate, the formation of neurofibrillary pathology . Thus, the transgenic mouse data are consistent with a scheme for AD pathogenesis  in which the accumulation of aggregated Aβ oligomers accelerates the parallel process of age-dependent formation of MAPT pathology. However, it is the accumulation of toxic MAPT species (again likely to be early aggregates rather than NFT) or other secondary pathological events (e.g. α-synucleinopathy) that initiates neurodegeneration and, by extension, accounts for most of the clinical syndrome.


10) Pathogenesis


The amyloid cascade hypothesis. Taken together, the available evidence favors a model of the disease in which diverse gene defects lead to enhanced production, increased aggregation, or perhaps decreased clearance of BA peptides. These effects allow accumulation first of the highly self-aggregating BA42 peptide and later the BA40 peptide. The gradual cerebral buildup of BA in first soluble and then particulate forms (the microscopically detectable consequence of which is diffuse plaques) appears to result in local microglial and astrocytic activation, with concomitant release of cytokines and acute-phase proteins (McGeer and Rogers, 1992; Eikelenboom et al., 1994; Wood, 1997). By means of these inflammatory changes or by direct BA neurotoxicity, local neurons and their processes can be injured, causing profound metabolic changes -likely including altered tau phosphorylation and paired helical filament formation in some plaque associated neurites and in tangle-bearing cell bodies.  Filamentous BA may alter glia and neurons by causing changes in calcium homeostasis as well as oxidative injury from free-radical formation. The clinically important consequences of these various events is synaptic loss and multiple transmitter deficits. BA depost may trigger caspase activation and apoptosis (Hardy and Selkoe, 2002).


Critique of the amyloid-cascade hypothesis. Amyloid deposit is the central element in the so called amyloid-cascade hypothesis of Hardy, originally formulated in the late eighties. There are several problems  with this model: 1)  plaques  display substantial interindividual variations and that these variations have no known relationship to the pattern of cognitive deficits; Against this argument the proponent of this hypothesis suggest that the degree of dementia in AD correlates much better with BA assayed biochemically than with histology and the concentration of soluble BA species;  2) although the initial memory loss indicates the presence of early limbic dysfunction, the initial plaques tend to appear in association neocortex  rather than limbic areas. In other words, the NFT distribution patterns that fit the expected anatomical substrate of the neuropsychological deficit, whereas the SP do not; 3) extensive neocortical plaque deposits can be seen in non-demented elderly individuals; 4) the correlation between plaque density and dementia is usually poor; 5) The APP/PS1/Alz17 mutant mouse with 30 months of age has massive amyloid deposits in the cortex, hippocampus and basal forebrain, yet there are no behavioral deficits and no neuronal degeneration or NFT formation (Holcomb et al., 1998); 6) Diseases with tau mutations (on chromosome 17; fronto-temporal dementia=FTD) are associated with dementia and massive cell loss, however, there is no amyloid deposit or tangles. This is according to Peter Davies, (known for discovering the cholinergic deficit of AD in 1976), not an AD mice but an amyloid mice. 7) the neurotoxic species of BA and the nature of its effects on neuronal function have not been defined in vivo. However, several lines of evidence suggest that the soluble oligomers of BA, but not the monomers or insoluble fibrils may be responsible for synaptic dysfunction, including inhibition of LTP and changes in the density of synaptophysin (Klyubin et al., 2005). In fact, it is suggested that the large polymeric aggregates represent inactive reservoirs of species that are in equilibrium with the smaller putatively neurotoxic assemblies.

Although the amyloid cascade hypothesis has several weakness, but an alternative hypothesis explaining the cause and early pathogenesis of AD that has as much experimental support has not emerged. The scheme in Plate is from D. Price (1998) suggesting that genetic mutations and risk factors in vulnerable neurons through several (partially unknown) mechanisms could lead to the clinical syndromes of dementia and cell death. At present, the mechanisms that underlie the selective vulnerability of these subsets of neurons are not well understood. However, evidence suggests that vulnerable cell populations exposed to high plasticity burden (LTP, axonal sprouting, dendritic remodeling, reactive synaptogenesis) in the adult brain and the elevated levels of plasticity related cellular activity may increase the vulnerability to degeneration (Mesulam, 1999). According to this hypothesis, NFT and SP/AP represent independent by-produts of initially compensatory but eventually excessive and maladaptive plasticity-related activity. Old age, genetic mutations, low estrogen, stroke, etc simple accelerate the temporal course of events that lead to plasticity failure and therefore lower the age at which the pathological process begins to gather momentum.  The advanced cognitive and mnemonic activities of the human brain impose a very high plasticity burden.


The identification of factors that specifically influence the viability and functions of these systems should allow the design of biological therapies to treat some of the neurological disorders associated with degeneration of these systems. Plate from Cuello summarizes in a cartoon most of the known pathogenetic routes that lead to AD.  This cartoon also includes data showing that increased acetylcholine release in the cortex via M1/M3 receptors and the inositol-3-phosphate-protein kinase C signaling pathway can lead enhanced APP levels- a link between cholinergic and APP mechanisms. Plate  from a recent review from Isacson et al (2002) also show the possible relationships between ACh regulation, trophic factors and BA production.  The fact that in the ancestry of Cherokee Indians the occurrence of AD is lower, and the finding showing higher rates of AD in Japanese men who emigrated to America as opposed to those who remained in Japan gives credence to theories that ancestry, environment, and other factors play additive roles in predisposing a person to AD.


11) Therapy in AD


The conclusion that flows from recent progress in defining the genotype-to-phenotype relationships in familial AD is a growing consensus that the most effective treatments for AD may be those that interrupt an obligatory early step, occurring before a progressive cascade of cell-damaging events.


Currently, the following broad classes of AD drugs are tested:


1)      Protease inhibitors that partially decrease the activities of the enzymes Beta and gamma-secretase that cleave BA from APP (LY411575; LY450139 dihydrate; BACE= Beta-site APP-cleaving enzyme; neprilysin as possible AB-degrading enzyme);  there is a concern that some compounds against gamma secretase might interfere with Notch signaling and other cell surface receptors

2)      Compounds that bind to extracellular AB and prevent its aggregation into cytotoxic amyloid fibrils. This strategy reasons that chelation of Cu2+ and Zn2+ may prevent BA deposition. BA deposition was prevented in APP transgenic mice treated with the antibiotic clioquinol, a Cu2+, Zn2+ chelator.

3)      Immunotherapeutic Approaches. AB maybe considered a self-antigen. It induces not only local innate immune response but also T-cell activation. Although T-cell activation can mediate autoimmune disease, this endogeneous immune response can be beneficial if properly boosted. Parenteral immunization of APP Tg mice with synthetic AB in complete Freund’s adjuvant can markedly decrease the number of AB deposits (Schenk et al., 1999). Also, a single, parenteral administration of a monoclonal antibody to AB has been shown to produced rapid behavioral improvement. However, passive immunization using monoclonal antibodies to AB resulted in brain hemorrhage. It was recently shown that overexpression of TGF-B in the CNS of APP-Tg mice resulted in a significant reduction of AB-plaque burden by promoting microglial clearance of the peptide. Adult mouse astrocytes also showed profound capabilities of degrading AB in vitro and in situ. Thus immune approaches targeted to induce AB antibodies and immune responses may result in activation of microglia and asytrocytes with a beneficial effect in AD. The finding that active vaccination wit hAB had profound AB lowering effects in animal models led to clinical trials in which an AB1-42 synthetic peptide was administered parenterally with a previously tested adjuvant to patients with mild to moderate AD. Although the Phase I study with a small number of patients failed to reveal significant side effects, a subsequent Phase II trial had to be halted in 2001 (Elan clinical trial) after about 5% of the patients came down with meningoencephalitis, a potentially deadly inflammation of the brain and surrounding membrane. In the postmortem case, there was clear evidence of decreased amount of AB plaques in the neocortex with marked T-cell infiltration in the CNS, the tangles remained, however.  Despite the adverse effects of this early human vaccination studies, efforts are in progress to develop more effective immunization methods.

4)      Nonsteroidal anti-inflammatory (NSAIDs) drugs. AB deposits in AD and mouse models induce immune response including microglial activation, secretion of proinflammatory cytokines (IL-1B, TNFalfa) and NO. If microglial or astrocytic activation fails to clear the toxic forms of AB, the innate response becomes chronic and neurotoxic. In this case, dampening the innate immuno response may be beneficial, as a number of clinical studies suggest that anti-inflammatory drugs used in arthritis may delay or slow the progression of AD (Naproxen, Rofecoxib, Celecoxib, Aspirin, Iboprofen, Indomethacin).   Their use, however, can cause considerable side effects, including gastrointestinal bleeding and perforated ulcers.

5)      Statins (Simvastatin, see above at “Cholesterol homeostasis”).

6)      Neuroprotective compounds such as antioxidants (vitamin C, E, ginkgo biloba), MAO inhibitors (selegiline) neuronal calcium channel blockers, or antiapoptotic agents that interfere with the mechanisms of BA-triggered putative neurotoxicity.

7)      Hormone replacement therapy. Estrogens. Estrogen replacement in postmenopausal woman decreases the risk of developing AD (Kawas et al., 1997). Estrogen replacement therapy became lately controversial, especially in the light of studies showing that hormone replacement therapy may increase the risk for breast cancer, heart attack and stroke (see recent article in New York Times, April 23, 2003).

8)      Cholinergic replacement therapy and cholinesterase inhibitors are symptomatic treatments for AD (tacrine, donepezil)

9)      Trophic factors (Plate). Recently, a transgenic mouse has been described that expresses a neutralizing antibody against nerve growth factor (NGF) (Capsoni et al., 2000). In these mice, brain pathology exhibits remarkable similarities to the pathology seen in AD, including amyloid plaques, hyperphosphorylated tau, NFTs in cortical and hippocampal regions and marked cholinergic neuron degeneration. The anti-NGF transgenic mice show more pathology that found in transgenic mice that express mutant APP. Cholinergic neurons and synapses depend on the trophic action of NGF for their function (ref. in Isacson et al., 2002). The survival effects of neurotrophins in cholinergic neurons are mediated via high-affinity tyrosine kinase receptors (trkA). In addition, neurotrophins binds to low-affinity (p75) receptors that can modify the binding and function of neurotrophins when coexpressed with trk receptors. The trkA and p75 receptors are synthesized in cholinergic neurons and NGF is synthesized by the neurons in the target regions and released in an activity-dependent manner. In individuals with AD, there is typically a marked loss of trkA receptors in both cholinergic target neurons and basal forebrain (BF) neurons, which correlates with loss of cholinergic neurons (Mufson et al., 1997). There are unchanged or increased level NGF levels in the hippocampus and cerebral cortex, while the levels in the BF are decreased with age-matched controls. This, together with cholinergic neuron loss and dysfunction, suggests that NGF is not adequately transported retrogradely to the BF or that binding to the trkA receptors or release of NGF from hippocampal interneurons, or both, is compromised in AD and in animal models. Infusion of NGF can prevent degeneration of axotomized cholinergic septal and BF neurons. APP can augment the effect of NGF and versus vica. Selective lesions of the cholinergic BF neurons of rabbits results in BA deposition in cerebral blood vessels and perivascular neuropil – a pathology seen in a rare familial form of AD. Thus, there is a close relationships between trophic factors, APP and ACh regulation. According to recent studies (Moses Chao at NYU, and Wilma Friedman at Rutgers) competitive signaling between trkA and p75 determines cell survival: trkA signaling leads to cell survival, p75 to cell death, although the different forms of NGF may be also important in determining the outcome.  These and other observations are consistent with the idea that there are homeostatic mechanism regulating hippocampal NGF, APP and ACh-mediated activity. A dysregulation of one or more of these three factors would lead to synaptic damage and neuronal cell loss relevant to memory function. Although initial hope for trophic factors as therapeutic agents did not materialize, future models that focus on complex interactions involved in dementias might be more realistic for designing new drugs acting at specific cellular or molecular levels to influence AD and other age-dependent neurodegenerative diseases.


10)  Environmental enrichment. Exposure of transgenic mice to an  “enriched environment” results in pronounced reductions in cerebral amyloid deposits, compared to animals raised under “standard housing” conditions. The enzymatic activity of an of familial AD (FAD)-linked APP degrading endopeptidase, neprilysin, is elevated in the brains of enriched mice and inversely correlated  with amyloid burden. Moreover, DNA microarray revealed selective upregulation in levels of transcripts encoded by genes associated with learning and memory, vasculogenesis, neurogenesis, cell survival pathways, BA sequestration. These studies provide evidence that environmental enrichment leads to reductions in both the steady-state levels of βA peptides and βA deposition and selective upregulation in levels of specific transcripts in brains of transgenic mice (Sisodia and coworkers, 2005). Similarly, studies with elderly suggest that education and physical exercise can reduce a person’s risk of AD. Although in one study no correlation was found between education and the formation of plaques and tangles, but a battery of 19 tests performed periodically in the years before the donors died revealed that people with high level of education better maintained their cognitive abilities. The highly educated participants did’t develop AD until they had about 5x as many plaques and tangles as the less educated participants (Rush Presbyterian-St. Luke’s Study: News Focus, Science, 309, 864, 2005).