By Jennifer Altman, based on her original report in Alzheimer Actualités, a newsletter published by the Ipsen Foundation. This report is part of the Alzforum Discussion Contemplating the Centennial. We thank the foundation for their permission to post Altman's work.
25 January 2007. One hundred years ago, in January of 1907, Alois Alzheimer published a paper entitled “On a unique disease of the cerebral cortex” (Über eine eigenartige Erkrankung der Hirnrinde) in the Allgemeine Zeitschrift fur Psychiatrie und Psychisch-gerichtliche Medizin (Alzheimer, 1907). This was the print version of an oral presentation Alzheimer had given to an audience of fellow psychiatrists on 3 November 1906 at the Psychiatric Clinic in the German university city of Tuebingen. By 1911, Alzheimer’s name was associated with the disease he had described, which we now know as pre-senile or early-onset dementia, that is, dementia starting before 60 years old, with distinctive pathological changes in the cerebral cortex. A century after Alzheimer's lecture, the world’s leading researchers into Alzheimer’s and related neurodegenerative diseases assembled on 3-4 November 2006 in the same institute to review the achievements of the past 100 years. Part scientific meeting, part celebration, in 2 days the major developments were summarized, the current state of research outlined, and speculations made for therapeutic developments in the coming 20 years. The meeting was organized by a committee of German and Swiss scientists in conjunction with the IPSEN Foundation, and an associated volume of articles has been published .
This occasion was unlike any other scientific meeting I have attended, more like a family reunion, with reminiscences, compliments, jokes, and one or two ancient rivalries revisited (no violence, though!). Several members of the community were mourned, others wished well in recovering from illness. However, some considered we should not be celebrating because, although the biology and pathology of these devastating afflictions are now fairly well understood, cure and prevention are still a long way off.
The enormous challenge posed by Alzheimer disease (AD) was becoming recognized by 1976, but it was the advances of the 1980s and early 1990s that set the stage for the explosion of research in the past 15 years. This has not been confined to AD; similar pathological processes are being identified in other neurodegenerative conditions, including Parkinson and motor neuron diseases. Many of these advances have been discussed over the years and the presentations at this meeting gave only a flavor of the huge effort put in by numerous teams worldwide. Here I will briefly review the development of the field, then discuss recent advances and future prospects, including some that are covered in the publication  by authors who were unable to speak at the meeting. This is an opportunity to get an overview of our understanding of the pathophysiological processes that result in neurodegenerative disease and the therapeutic possibilities that are being explored.
Defining the Disease
In his 1906 presentation and 1907 paper , Alzheimer correlated precise observations of a single patient’s behavior with a careful description of the pathology. He described the characteristic neurofibrillary tangles (NFTs) and what he termed “miliary foci,” which are now known as senile or amyloid plaques.
A big debate, which continued right up to the 1990s and still causes some confusion, is what conditions can be considered as Alzheimer disease (AD) (Yves Christen, Paris, France ). The case of Auguste D. presented by Alzheimer in 1906 was clearly pre-senile dementia, as onset was at age 54. Initially Alzheimer’s name was attached only to this early form of dementia, a relatively rare condition that was distinguished from the dementia of old age, commonly termed senile dementia. Molecular and genetic advances of the past 20 years have clearly established the pathological similarities among many cases of senile dementia and pre-senile dementia. The name of Alzheimer now applies to cognitive decline leading to dementia that is characterized by plaques and tangles in the cerebral cortex. Pre-senile (or early-onset) and senile (late-onset) dementia are considered part of the same disease continuum. Other forms of senile dementia associated with different patterns of neurodegeneration have been identified, including frontotemporal dementia, as well as multi-infarct dementia, caused by repeated small strokes.
In the 60 years following Alzheimer’s presentation, few significant advances were made in understanding what was then considered a rare condition. Interest started to stir in the 1960s with the advent of the electron microscope and its application to neuropathology. The history of research into Alzheimer disease (AD) since then has recently been summarized . Attention first turned to the NFTs in AD brains, which were well suited for such studies (Robert Terry, La Jolla, USA). The fibrils that composed them were shown to be paired with a helical conformation, hence the name paired helical filaments (PHF); changes in synapses close to plaques were also described. By the end of the decade, Blessed and colleagues had published the first quantitative assessment of dementia associated with pathological changes in the cortex.
In the mid 1970s, the loss of cholinergic neurons early in the disease and its correlation with memory decline was established (Peter Davies, New York, USA). Although the cholinergic deficit is not the primary cause of AD, this remains one of the most significant findings linking pathology with behavior. It led to the development of the cholinesterase inhibitor drugs that presently constitute three of the four approved drugs for treating AD (the fourth drug acts on NMDA receptors).
Early in 1990s, it became clear that the AD pathology had several origins. The early-onset form is predominantly inherited, and a variety of mutations has now been identified. No clear pattern of inheritance or cause has been associated with the late-onset form, although the E4 variant of the ApoE gene is an important risk factor (Alan Roses, Research Triangle Park, USA), the only undisputed one identified so far.
The Amyloid Cascade Hypothesis
Another big technological advance, the development of molecular biology and genetics in the early 1980s, shifted attention to the chemical composition of the plaques and tangles. The main component of plaques had been identified as an amyloid protein by Paul Divry in 1927 by the way it reacted to certain histological dyes (Terry). In 1984, an amyloid protein was purified and sequenced from deposits in cerebral blood vessels by George Glenner and Cai’ne Wong (Cai’ne Wong, Scarborough, Maine, USA) and the following year, the same protein was isolated from plaques (Colin Masters, Melbourne, Australia). This established a causal link between AD and cerebral amyloid angiopathy, a condition marked by bleeding in the brain (Blas Frangione, New York, USA; Mathias Jucker, Tuebingen, Germany ). This protein, named β amyloid (now known as amyloid-β and usually abbreviated as Aβ), was soon shown to derive from a parent protein, the amyloid precursor protein (APP), the gene for which was cloned in 1987 (Konrad Beyreuther, Heidelberg, Germany; Rudolph Tanzi, Boston, USA; Dmitry Goldgaber, New York, USA).
Early in the 1990s, mutations were identified in the APP gene, located on chromosome 21, which were linked to families with several members suffering from early-onset AD (Alison Goate, Saint Louis, USA; Peter St George-Hyslop, Toronto, Canada) and cerebral amyloid angiopathy (Frangione). This strongly implicated plaques and particularly Aβ as the primary cause of neurodegeneration. Support came with the discovery that many families with early-onset AD have mutations in two genes coding for presenilins (Gerard Schellenberg, Seattle, USA), proteins involved in an enzyme that releases Aβ from APP. Moreover, Down syndrome sufferers almost invariably develop AD pathology as a result of the duplication of the part of chromosome 21 that carries the APP gene (Edward Koo, La Jolla, USA ). About the same time, Aβ was claimed to promote both the growth and death of neurons in culture (Bruce Yanker, Boston, USA).
These discoveries led to the hypothesis that Aβ triggers a cascade of events in the brain that leads to the dysfunction and death of neurons. The amyloid cascade hypothesis was reinforced by subsequent findings and today still dominates thinking about the progress of the disease and its pharmacological treatment.
NFTs and the Tauopathies
The focus on Aβ has tended to play down the significance of the NFTs, even though their distribution in the brain closely mirrors the course of cognitive decline (Heiko Braak, Frankfurt, Germany). NFTs appear in the hippocampus before plaques start to form in the cortex, whereas plaque distribution is uniform and does not correlate with the advance of symptoms (Charles Duyckaerts and Jean-Jaques Hauw, Paris, France ).
The main constituent of the PHFs that form the NFTs was isolated in the late 1970s and identified as tau, a protein associated with the microtubules responsible for the transport of molecules along axons. The tau in the PHFs, however, has an abnormal number of phosphate groups attached compared with that in normal axons, a state known as hyperphosphorylation (Khalid Iqbal, New York, USA). Another component of PHFs was identified in 1987 as ubiquitin, a protein that labels malformed proteins for destruction (Yasuo Ihara, Tokyo, Japan). This confirmed the indestructible nature of the PHFs, already indicated by the way NFTs survive as “ghost” tangles or “tombstones” once neurons have died and degenerated.
Using the recently introduced technique of immunostaining, researchers demonstrated tau in NFTs in the brain in 1985 (Jean-Pierre Brion, Brussels, Belgium). By 1988 its gene had been cloned and sequenced (Michel Goedert, Cambridge, UK). It was another 10 years before mutations in the tau gene associated with neurodegenerative diseases were described, in some families with a form of frontotemporal dementia in which patients also have parkinsonian symptoms (FTDP) (in another form of FTDP, the inclusions contain ubiquitin but not tau) (Mike Hutton, Jacksonville, USA).
These patients have inclusions in neurons that contain tau but do not resemble NFTs; tau-containing pathological inclusions are also found in a range of other neurodegenerative diseases, including Pick disease, some cases of Parkinson disease, and in myotonic dystrophy (Christine van Broeckhoven, Antwerp, Belgium; André Delacourte, Lille, France ). All diseases with tau pathology are now classed as tauopathies, emphasizing the importance of pathological forms of the tau protein in neurodegeneration.
The Biology of APP, Aβ, and Tau
Understanding the normal biological role of the proteins associated with AD pathology is both fundamental to understanding the process of neurodegeneration and essential for designing drugs that will safely halt or prevent the process. Surprisingly little is known even now about the normal functions of APP and Aβ —a hole that urgently needs addressing. APP, a large protein that threads through intracellular membranes, is found in all types of cells in most species, so must have a basic biological function. In neurons, it seems to be associated with the recycling of synaptic vesicles (John Cirrito and David Holtzman, St Louis, USA ), and possibly with axonal transport (Eva-Maria Mandelkow et al., Hamburg, Germany ). It is normally cut by the enzyme α-secretase into two portions, one of which, known as APPs, may promote neurite growth and protect against damage (Falk Fahrenholtz, Mainz, Germany ).
The Aβ peptide, a sequence of 40–42 amino acids, spans the middle of the APP molecule, lying partly within the transmembrane domain. The α-secretase cuts in the middle of this sequence, so produces no Aβ or amyloid formation. Under other conditions, which remain to be defined, two other enzymes, the β- and γ-secretases, come into play, releasing the Aβ peptide. β-secretase cuts at the start of the Aβ sequence and γ-secretase in the intramembrane domain (Christian Haass, Munich, Germany). Aβ release by β- and γ-secretase action is now known to be not solely pathological but part of normal cell function; it is just beginning to emerge that Aβ may be involved in synaptic receptor physiology. There is also an indication that it may be involved in sealing damaged blood vessels (John Hardy, Bethesda, USA).
The function of normal tau protein is much better understood. Molecules of tau attach to the surface of microtubules, stabilizing them and promoting transport along axons from cell body to terminals. This involves phosphate residues attaching to and detaching from the tau molecules, a process requiring the enzyme MARK (Eckhardt Mandelkow, Hamburg, Germany). When, for an unknown reason, the tau molecules collect too many phosphate residues, that is, become hyperphosphorylated, they become insoluble and aggregate to form PHFs, with the result that the axon and later the cell body becomes choked with PHFs, axonal transport is blocked, and slowly the cell dies. However, indications are that this process can be reversed. Adding MARK to cultures of neurons containing tau aggregates reverses the blockage and restores transport. Treating mice engineered to produce hyperphosphorylated tau with compounds that stabilize microtubules also prevents neurons from dying and protects the mice from neurodegeneration (John Trojanowski, Philadelphia, USA).
The Various Faces of Amyloid
Although in AD the word “amyloid” is usually applied to deposits of the Aβ protein, chemically it refers to any protein deposit made up of fibrils that have a protein backbone folded into a conformation known as a β-sheet, with the sheets lying perpendicular to the long axis of the fibril. The β-sheet conformation is a pathological form of proteins that are normally found in the body in the α-helical conformation. What causes the transformation is not well understood, but the result is a protein that forms stable, relatively insoluble fibrils that stick together to form aggregates; these cause problems for the tissues in which they are deposited as well as destroy the biological function of the proteins. The amyloid found in AD plaques and tau in PHFs are both amyloids (Eckhard Mandelkow), as are the prion protein found in scrapie and Creuzfeldt-Jakob disease, and the intraneuronal inclusions found in Lewy body dementia, Parkinson disease, and motor neuron disease, among others (Virginia Lee, Philadelphia, USA).
Both from the perspective of the pathological protein deposits and of genetics, it now seems helpful to view all the neurodegenerative diseases as a continuum. At one end are the cerebral amyloid angiopathies, where amyloid deposits form in cerebral blood vessels but not in the neural tissue; at the other end are amyotrophic lateral sclerosis (motor neuron disease) and the type of FTDP in which the inclusions are formed largely of ubiquitin, and Parkinson disease with α-synuclein inclusions. Mutations in relevant genes have now been identified for all these types of inclusions, and many patients with these conditions also develop dementia (van Broeckhoven; Lee).
The overlap between conditions along the continuum accounts for much of the heterogeneity in the relationship between genetic mutation and clinical symptoms. For instance, patients with one mutation in presenilin-1 have symptoms of AD together with spastic paraparesis and a distinctive type of “cotton-wool” plaques; another presenilin mutation has been found in patients with Pick disease—many other examples have been reported (van Broeckhoven).
Producing Aβ—the Secretases
The plaque amyloid is mainly composed of the 42-amino-acid form of Aβ, which more easily forms fibrils than the 40-amino-acid form. The form that is produced depends on where the γ-secretase cuts (Haass), and a shift to producing Aβ42 is a prelude to plaque formation (Sangram Sisodia, Chicago, USA). Several mutations in the APP gene and the presenilin genes promote this shift (Steven Younkin, Jacksonville, USA). The Aβ42 form is also thought to be more neurotoxic than the Aβ40 form (Yanker).
When APP was first sequenced, a big puzzle was how Aβ could be released from the membrane domain, as enzymes were considered unable to operate within the fatty, water-repelling environment of the membrane. The discovery of presenilin led to its identification as part of the γ-secretase, which is a complex of four proteins (Bart de Strooper, Leuven, Belgium). The APP molecule attaches between two of the other subunits and is pulled into an aqueous microenvironment in the center of the complex, where presenilin acts as the catalyst (Michael Wolfe, Boston, USA; Takeshi Iwatsubo, Tokyo, Japan). One nasty surprise for those looking to inhibit γ-secretase action was that the enzyme has other substrates as well as APP, including a protein called Notch, which is essential during embryonic development and in the regulation of cell division in adult skin and intestine. Presenilin also seems to determine the switch between cutting APP and Notch. Some way will have to be found to ensure that drugs aimed to prevent the γ-secretase cutting at amino acid 42 are selective for APP.
Another way to prevent Aβ from accumulating would be to inhibit the β-secretase, which was identified in 1999 as a membrane-bound enzyme known as BACE1 (Martin Citron, Thousand Oaks, USA). Given that Aβ seems to have a function in normal cells, even if this has not yet been fully identified, it is rather surprising to find that mice lacking the gene for BACE1 have no obvious physical abnormalities. However, they do have subtle cognitive and emotional problems, raising the possibility that BACE1 is somehow involved in cognition (Donald Price, Baltimore, USA). There are also questions about whether BACE1 inhibitors may damage myelin.
Compromising Neuronal Function
Just how plaques and NFTs cause neurodegeneration is still being debated. The current view is that plaques, and particularly the Aβ fibrils that are their main constituent, do not kill neurons. Rather, they cause neuron dysfunction through destroying synapses and dendrites, resulting in widespread deterioration of cortico-cortical connections. In contrast, NFTs do kill neurons, because they block axonal transport, and so cause specific disconnection of one area from another (Bradley Hyman, Boston, USA). One early consequence of this disconnection is the loss of the cholinergic input from the basal forebrain to the cortex, important for memory processing (Davies; John Growdon, Boston, USA ).
In the past two years, a new candidate culprit has emerged: the damage may be caused not by the insoluble Aβ in the plaques but by shorter, soluble forms of Aβ that have been named variously amyloid-β-derived diffusible ligands (ADDLs; William Klein, Northwestern University, Chicago, USA ), Aβ*56 (Karen Ashe, Minneapolis, USA) and soluble oligomers (Charles Glabe, Irvine, USA ; Dominic Walsh, Dublin, Ireland ). These oligomers take on a globular rather than fibrillar form and seem to accumulate around cortical neurons, attaching themselves to the post-synaptic neuron at or close to the NMDA neurotransmitter receptors that are essential for memory storage. In genetically engineered mouse models of AD, they seem to be the cause of characteristic synapse loss, increased production of hyperphosphorylated tau, and deteriorating memory.
One reason for the accumulation of Aβ, either as oligomers or as fibrils, may be a loss in the balance between the biological production and breakdown of excess Aβ. One of the key enzymes for breaking down Aβ is neprilysin, and inactivating the gene producing neprilysin in a mouse model of AD resulted in an increase in Aβ oligomers, with the consequences detailed above (Takaomi Saido, Saitama, Japan). Other Aβ-degrading enzymes have since been found, and a deficit of this sort could be a cause of the late-onset, non-hereditary form of AD.
The time has perhaps come to take a longer look at the amyloid cascade hypothesis and to consider the biological basis of AD in more totality (Hardy). It is possible that the focus on amyloid has blinded the field to a more fundamental process that triggers the formation of both plaques and NFTs. Such a reassessment is particularly urgent as many of the therapeutics under development aim to reduce Aβ production and plaque formation. It would be wise to establish the roles of APP, APPs, and Aβ before interfering with their natural production. Likewise, efforts to understand the full breadth of function of the β- and γ-secretases should accompany efforts at interfering with them therapeutically.
Disease Models, Diagnostics, and Imaging
Another technological advance that has had a great impact on research into AD has already been alluded to—the creation of mouse models by genetic manipulation. It is a delusion to think that this technology can reproduce AD in a mouse, and still there is no good animal model for the full course of the disease—mice do not get AD (Ashe). Twelve mouse strains have had the human APP gene with a given AD mutation successfully introduced, and all produce amyloid deposits. Some strains form plaques, but none have produced tangles. More complete models have become available recently, but these have their own limitations (Frank LaFerla , Irvine, USA). However, the genetically manipulated mice do help to gain an understanding of various processes and to check on unforeseen consequences of inhibiting the actions of molecules such as the secretase enzymes (Price).
Great strides have also been made in a different area of technology: imaging the disease process in humans. Starting with computerized tomography scans in the late 1970s, the first quantification of changes in the brain were obtained by drawing around individual images, cutting out the images and weighing them (Mony de Leon, New York, USA). Now high-resolution magnetic resonance scans and automated analysis are allowing almost routine serial examinations of individual patients. Change over time can be plotted and used both for diagnosis and monitoring in drug trials (Nick Fox, London, UK). A radioactive tracer known as PIB that stains plaques in vivo has been developed and can be detected by positron emission tomography (William Klunk, Pittsburgh, USA). Although at present it is used mainly for research purposes, a variant suitable for clinical diagnosis is being developed. Employing these tracers in a mouse carrying a mutated human APP gene that produces plaques has shown that once plaques form, they remain a constant size, while neurites surrounding the plaque can be seen to retract and realign themselves (Hyman).
In the clinical assessment of dementia, the category of “mild cognitive impairment” was introduced a few years ago to indicate people in the earliest stage of decline into dementia. Always controversial, its usefulness is being challenged by both careful behavioral monitoring of individuals, including assessments from caregivers (John Morris, St Louis, USA), and repeated imaging of plaque load in individuals using PIB. In about 30 percent of patients diagnosed with mild cognitive impairment, plaques are few and these patients never develop AD; in contrast, more than half have a plaque load similar to that of early AD (Klunk). Some researchers argue that the concept of dividing patients into categories of mild cognitive impairment and AD is outworn and needs replacing (Monique Breteler, Rotterdam, Netherlands). A development crucially needed for early diagnosis and to provide simple outcome measurements in clinical drug trials—a rapid test that could be performed on a blood sample in a local clinic—remains elusive. Several candidates have been promoted over the years, but none has stayed the course. One current contender examines platelets for the effect of acetylcholinesterase inhibitors on APP processing and the ratio of α-secretase to β-secretase (Elena Marcello et al., Milan, Italy ). Another promising direction is an assay for elevated Aβ42 in plasma (Younkin).
Therapeutics in Development
Treatment for AD falls into three categories: symptomatic relief, disease modulation, and prevention. The only approved drugs, cholinergic drugs and a NMDA receptor modulator, largely treat symptoms and will continue to have a role even when drugs that can alter the course of the disease are approved (Leon Thal, La Jolla, USA). Most drugs in development are of this modulatory type and in the future, variants of such drugs may be suitable to prevent the disease from developing.
In-vivo imaging of plaques in mice shows that deformed neurites in the vicinity of plaques recover when the plaque is dissolved by experimental therapeutic treatment (Hyman), giving hope that if safe compounds are found to remove Aβ deposits, function will recover. At present, most therapeutics in development aim to slow the rate of Aβ deposition or reduce the plaque burden. Much effort is going into development of selective inhibitors or modulators for the β- and γ-secretase enzymes (Price), although enhancing the action of α-secretase could help to tip the balance away from overproduction of Aβ (Farenholtz ). Targets such as neprilysin are also being tested (Saido).
Much attention has been given to therapeutic immunization against Aβ. Although immunization in mouse models results in a decrease in the size and number of plaques (Dale Shenk, South San Francisco, USA; Hyman), clinical trials have run into trouble because some patients have developed meningoencephalitis (Roger Nitsch, Zurich, Switzerland). This may be due to some unexpected action of the antigen used. Alternative possibilities are that the antibodies made in response to the injected antigen block a biologically useful function of Aβ (Hardy) or that the inflammation is a reaction to the dissolving plaque material. A concern with all the anti-Aβ therapeutics is that removing plaques but leaving NFTs will turn AD into a pure tauopathy (Trojanowski); some candidate drugs for treating the NFTs by stabilizing microtubules are being investigated (see above).
It is unlikely that a single “magic bullet” drug will be found; rather, a cocktail of more or less specific drugs will need to be tailored to individual requirements (Todd Golde, Jacksonville, USA). A major stumbling block in the development of drug therapies is the enormous cost of trials and the difficulty in defining clear outcomes, highlighting the need for simple tests to measure markers of the disease process (Thal). One method being developed measures the rate of production and clearance of Aβ; it is showing some promise (Holtzman). Another problem is the format of clinical trials required by the Food and Drug Administration in the USA. It is poorly suited to diseases like AD, where patients are unlikely to recover and the best that can be expected is to halt the patient’s decline; changes to the FDA requirements are urgently needed (Golde).
The dream of a cure for AD is probably just that. By the time symptoms appear, about 60 percent of the neurons in the entorhinal cortex, the first region to be affected in the disease, have been lost (Hyman). Many researchers feel that once clinical diagnosis is made, the only possibilities are drugs that slow down the course of deterioration and give symptomatic relief. Even with today’s resources, however, considerable gains can be made by caring for patients individually, with compassionate consideration for their shrinking personality, rather than by viewing them collectively as a group beyond help. Highly skilled nursing, tailored drug regimens, and a stimulating environment are essential (Bengt Winblad, Huddinge, Sweden). The decision of the English National Institute for Clinical Excellence to withdraw anticholinesterase drugs from certain classes of AD patients was heavily criticized—clinical experience shows that these drugs can significantly improve the quality of life for many patients and their caregivers, and there is no justification for denying their use.
With the continuing increase in the number of elderly people, the concomitant rise in AD cases will place an enormous burden on both public and private resources. Today the cost of caring for patients with dementia in the USA is estimated at around $248 billion, three times that for diabetes, and this is likely to increase fourfold by 2050 (Winblad). Consensus is growing that neurodegeneration starts decades before the disease becomes apparent. Besides illustrating the awesome plasticity of the human nervous system, in practical terms this means that the highest priority is to identify those at risk and to find ways of preventing their deterioration (Breteler). A proactive program of genetic testing, blood tests, and brain scans for those over 50 was foreseen (Dennis Selkoe, Boston, USA), although only those in the richer countries are likely to be able to afford such screening.
The final speakers at the meeting were asked to speculate about future developments. Most were upbeat, estimating that markers for determining outcomes and predicting risk will be available within 10 years and that in 20 years’ time there will be therapeutics for delaying disease onset (Fox, Holtzman, Selkoe). One problem with prevention programs is public resistance to changes in life style; education is essential but perhaps a new marketing concept, such as “brain preservation,” would have more impact than “protect yourself from AD” (Breteler). From a research point of view, two central concerns are to increase funding, which will require private as well as public money (Zaven Khachaturian, Potomac, USA), and to attract high-caliber young scientists to the field (Beyreuther).
The meeting ended on a general tone of optimism. Although the challenge ahead was likened to “putting men on the moon” (Khachaturian), the prospects for managing neurodegenerative diseases within the next 20 years are positive. Maybe it is time for our societies to start to consider how they are going to accommodate large numbers of healthy, active 90 year olds. There’s a thought!
1. Jucker, M. et al. Alzheimer: 100 Years and Beyond. Heidelberg, Springer Verlag (2006)
2. Alzheimer A. Über eine eigenartige Erkrankung der Hirnrinde Allgemeine Zeitschrift fur Psychiatrie und Psychisch-gerichtliche Medizin. 1907 Jan ;64():146-8. Reproduced in , with English translation by Clifford Saper and Horst Herbert. Abstract
3. Perry G, Avila J, Kinoshita J, Smith MA. Alzheimer’s Disease. A Century of Scientific and Clinical Research. IOS Press, 2006.
4. Klein, W.L. Molecules that disrupt memory circuits in Alzheimer's disease: the attack on synapses by Abeta oligomers (ADDLs). In: eds Bontempi, B., Silva, A. J. and Christen, Y. Memories: Molecules and Circuits. Heidelberg, Springer Verlag, in press.
5. Kayed R, Head E, Thompson JL, McIntire TM, Milton SC, Cotman CW, Glabe CG. Common structure of soluble amyloid oligomers implies common mechanism of pathogenesis. Science. 2003 Apr 18;300(5618):486-9. Abstract
6. Walsh DM, Klyubin I, Fadeeva JV, Cullen WK, Anwyl R, Wolfe MS, Rowan MJ, Selkoe DJ. Naturally secreted oligomers of amyloid beta protein potently inhibit hippocampal long-term potentiation in vivo. Nature. 2002 Apr 4;416(6880):535-9. Abstract
7. Oddo S, Caccamo A, Shepherd JD, Murphy MP, Golde TE, Kayed R, Metherate R, Mattson MP, Akbari Y, LaFerla FM. Triple-transgenic model of Alzheimer's disease with plaques and tangles: intracellular Abeta and synaptic dysfunction. Neuron. 2003 Jul 31;39(3):409-21. Abstract