Peter Nelson led this live discussion on 26 March 1998. Readers are invited to submit additional comments by using our Comments form at the bottom of the page.

Paper Under Discussion: Neve RL and Robakis NK. Alzheimer's disease: a re-examination of the amyloid hypothesis. TINS. 21 (1);1998:15-19. Abstract

Live discussion held 26 March 1998.

Participants: Nikolaos Robakis, Rachael Neve, Dennis Selkoe, Peter Nelson, Marc Paradis, Chris Weihl, James Vickers, Dave Small, Wilma Wasco, Ben Wolozin, June Kinoshita.

Note: Transcript has been edited for clarity and accuracy.

Robakis: Hi Dennis this is Nick, how're you doing?

Peter Nelson: Hi Dr. Robakis, Dr. Selkoe. I'm Pete Nelson. I guess I'll be moderator. June suggested that I request messages in “private” from the guests, and then put them to you in an orderly way. If that sounds reasonable to you, that's how I'll proceed, but if you have other thoughts or suggestions, that's fine too. Is there anything that you want covered, or not covered, that I should know about now? June suggested also that we wait until about 5 after the hour to begin, and then I'll give a shpiel about the format.

Robakis: OK.

Robakis: Hi Rachael how are you (Tasos, Nick)

Selkoe: Hi Rachael and Nick. Hope all is well.

Robakis: Everything is OK thanks.

Selkoe: Peter, I remember us sitting together at the scope, looking at tangles in sheep brains. I must commend you on the responses you made in your overview. I concur with virtually all of them!

Peter Nelson: Thanks, Dr. Selkoe, nice of you to remember.

Selkoe: Rachael, I'm looking forward to seeing you at McLean next Friday.

Peter Nelson: I guess that maybe Dr. Neve is having technical difficulties....

Selkoe: Hi Marc. I enjoyed our discussion last night.

Peter Nelson: It would be nice if the text of the ongoing conversation was put up for all who come in. However, I think that only the conversation that occurs while you're logged in is visible to you. And, if anyone does not put input for fifteen minutes (NOTE THAT), that person gets automatically logged off. I'll repeat that maybe, in a half-hour or so.

Peter Nelson: Hi June! Give the word, and I'll get this thing going.

June Kinoshita: Hello everybody. Sorry to be late. Let's roll!

Robakis: OK.

Peter Nelson: Hello, welcome one and all to the discussion. Todays regards the role of β -AP in Alzheimer's disease. Three distinguished panelists (Drs. Robakis, Neve, and Selkoe) will be discussing the issue, (one hopes) contentiously. My name is Pete Nelson, and I'll be moderator. Perhaps we should first have everyone introduce themselves....

Selkoe: I'm Dennis Selkoe in Boston. I favor amyloid's central role in AD.

June Kinoshita: I'm with the Alzheimer Research Forum, which hosts these journal clubs. Welcome!

Marc Paradis: Hello All, I'm Marc Paradis, a graduate student at MIT.

Robakis: I'm Nick Robakis at Mt.Sinai School of Medicine in NY.

James_vickers: Hello, James Vickers from Tasmania (Australia not Africa).

Cweihl: I'm Chris Weihl a graduate student at the University of Chicago.

Peter Nelson: People should send questions to the panelists through me, in private. The private message icon is the dark little man in the upper left of the mini-screen. Remember, give input within fifteen minutes, or you're automatically logged off!

Peter Nelson: My first point is to ask Dr. Robakis if he feels that there is a specific toxin that poisons cells in Alzheimer's disease...and, if it is not, B-AP, then what else might it be?

Robakis: I don't think that there's a toxin with the classical meaning of the word. I think that we have a dysfunction of the protein transport and cytoskeletal system of the cell. In fact changes in AB production in APP and PS FAD mutations may be an indicator of a disturbance in cellular protein transport.

Peter Nelson: Again, to Dr. Robakis: So, there's not a substance within neuritic plaques, for example, that causes the neurites to become dystrophic; and the observation that cells that project to the cerebral cortex die only implies that something specific to those cells is sick?

Robakis: There are more than 1/3 of normal aged people have enough neuritic amyloid depositions to satisfy the AD criteria, yet they are normal.

Selkoe: Nick's last statement is incorrect. Certainly people with many diffuse plaques may be normal but normal aged people have small numbers of neuritic plaques generally confined to medial temporal structures.

Robakis: Dennis hi. That's not my impression from many neuropathologists. In any event you will agree with me that there is no correlation between amyloid deposition and synapse or neuronal loss.

Peter Nelson: This comment by Dr. Robakis prompted further questions. However, I shall turn to Dr. Selkoe now. Dr. Selkoe, your work suggests that presenilin mutations work through increasing β -AP. Some people disbelieve this relationship. Could you describe some of your newer thoughts of the interrelationship of Aβ and presenilins?

Rneve: I'm sorry I'm late. Had to reload Ichat.

Selkoe: The fact that PS mutations increase Aβ 42 in cells mice, human brain and human plasma in all published studies (but not Aβ 40) strongly suggests that mutant PS acts by increasing γ secretase cleavage at Aβ42.

Rneve: Do you make a distinction between intracellular and extracellular Aβ 42? Which is the culprit?

Peter Nelson: Wait! Let's stay for a moment on the Presenilin-b-AP connection!

Rneve: The reason I ask: there is some unpublished work showing that PS1 mutations do not increase extracellular Aβ 42 in neurons. How would you respond to that?

Peter Nelson: My most important class in high school was typing!

Selkoe: Who is the source of the unpublished data? Clearly, Cindy Lemere's published data showing significant increases in Aβ 42 plaques in PS1 mutant brains provides direct in vivo relevance to the many reports of increased Aβ 42.

Rneve: Nick presented it at the Soc for Neuroscience meeting, in a session that you were in. He and I have generated those data.

Selkoe: Do you see any increase in Aβ 40 or 42 and how were the neurons made to express mutant PS1?

Rneve: We used HSV vectors. I don't think we saw any overall increase in Aβ , either 40 or 42; isn't that correct, Nick?

Robakis: Yes it's correct.

Peter Nelson: Dr. Neve pointed in passing to a central point, I think, and that is, the central point of difference between you three panelists (pro b-APPtists all, if I'm not mistaken) is now which PART of the b-APP matters, and where it matters (internal or external to the cell). Internal cell processing of b-APP is certainly a hot topic these days. I'd like the three of you to maybe give some thoughts on how your opinions differ in this regard.

Rneve: I would like to hear how you stand on this point at the present, Dennis.

Rneve: The data that we have generated with Nick suggests strongly that expression of PS1 mutants in neurons does not cause increased release of Aβ . However, we have not measured intracellular levels. Isn't there a trend these days among the amyloidophiles that the important increase in Aβ 42 is _in_ the cell?

Selkoe: I think internal APP processing is critical of course, and that Aβ first appears intracellularly and then is quantitatively secreted. It could be that increased intracellular Aβ 42 in for example, PS and APP mutant patients, causes some toxicity. This is not yet known. But we do know that there is increased Aβ 42 accumulation extracellularly. And in PS APP double transgenic mice, this increased extracellular accumulation is associated with surrounding glial and neuritic change. Probably both are important.

Robakis: What I'm saying is that at this point there's not clear data that amyloid β depositions is the main cause of AD. We need further research to clearly say whether amyloid is or is not the primary cause

Robakis: I think that changes in Aβ production in FAD may actually be an indicator of a general protein transport disturbance in the cell.

Selkoe: I agree there is no direct proof but the combination of the four AD-linked genes, all APP transgenic mouse models, and the clear association of fibrillar but not diffuse Aβ plaques in AD and DS brains with surrounding microglial and neuritic changes provides very strong circumstantial evidence that Aβ is necessary although by itself not sufficient for AD to occur.

Peter Nelson: There is clearly decent evidence that plaques matter, and harm nervous tissue. What is the reaction of Dr. Neve to the intuitive observation that amyloid plaques are surrounded by degenerating (PHF-filled) neurites? It seems like something in there is poisonous. Dr. Vickers points out that this could be a mechanical influence, putting stress on the axons...

Wolozin: I don't quite buy that intracellular Aβ is the key, I think that it is much more likely that rat brains are resistant to extracellular Aβ .

Rneve: Those are good points, Dennis. The reason I bring up the data that we obtained with Nick is that, as you know, we showed that expression of FAD APP mutants in neurons causes increased intracellular C100 (and increased extracellular Aβ ). Now, we don't see increased intracellular C100 when we express PS1 mutants in neurons, nor do we see increased extracellular Aβ . So I see a precursor-product relationship between C100 and Aβ (as does everyone else). Whether C100 is simply a source of toxic Aβ or is the culprit itself is still an unanswered question that needs to be answered. I acknowledge that amyloid plaques might be unhealthy for the neurons around them, but I don't think they are the primary cause of the disease.

Robakis: How do we know that in PS-/- cells C-terminal fragments which are substrates for γ-secretase fail to be transported to the appropriate subcellular compartment for γ-secretase cleavage. Please consider that PS1 is present in transport vesicles and may

Wolozin: If C100 were the problem, would you see mutations that produced just C100 and wouldn't you see a major league accumulation of C100 in AD brains?

Selkoe: If C100 were the toxin, then the more Aβ that is produced (and thus the less C100) the better the patient would be. An increase in C100 in humans with AD should be associated with less Aβ and vice versa. And yet I think that mutant PS patients have significantly more Aβ in their brains than sporadic AD cases and certainly than aged normals. So, I think C100, while able to cause toxicity in vitro and in vivo, is not the key toxic species in humans with AD.

Peter Nelson: A response, Dr. Neve?

Robakis: Hi Ben, how're you doing?

Rneve: Dennis, a flaw in that argument is that increases in C100 should not be associated with less Aβ . They will be associated with more Aβ , because it is a byproduct of C100. I just don't think that the increase in Aβ that you get with increased C100 is the problem.

Wolozin: Nick - Doing fine! Adjusting to the extramural world.

Rneve: The experiments that need to be done -- and Donna is working on them -- are to produce C100 that cannot be broken down to Aβ and show that C100 still has its toxic effects. Ikuo did that with the FAD APP mutants, showing that they still caused apoptosis even when Aβ 42 was not produced from them.

Peter Nelson: And yet, La Ferla et al. showed that just Aβ in a transgenic mouse causes apoptosis like crazy....

Rneve: Yes, but that phenotype was very strange. It caused apoptosis but the pathology was nothing like AD pathology.

Marc Paradis: I think that you have to be careful with overexpression and transgenic experiments. Apoptosis is a general response to many kinds of insults, including such blunt genetic alterations.

Rneve: That's an excellent point.

Peter Nelson: Dr. Wolozin wonders how much that matters; apoptosis is apoptosis, I guess.

Robakis: If soluble Aβ were toxic, how do you explain the fact that it is present in almost equal amounts in the CSF of normal and AD people.

Rneve: Another excellent point.

Peter Nelson: I don't think apoptosis should be such a general thing. Dr. Selkoe, do you think Aβ 's mechanism works through apoptosis?

Marc Paradis: In answer to robakis: its really a matter of concentration, levels in CSF may not be high enough to be toxic.

Selkoe: Rachael, the fact that humans with PS and APP mutations have excess Aβ levels in brain, plasma and fibroblast media indicates that Aβ buildup from C100 is definitely happening excessively in the AD state, therefore a C100 model that does not produce Aβ might still produce toxicity from C100 itself but could not be considered a model of the AD state.

Cweihl: Sensitization could occur through environmental insults in sporadic AD.

Robakis: This conference is not about apoptosis. So we should focus on amyloid and C100.

Rneve: I'd like to get back to Nick's questions. And add to that the fact that inter individual variation in soluble Aβ in CSF is far greater than the statistical difference between control and AD patients in soluble Aβ.

Wolozin: Actually, the issue of apoptosis is overblown. Depending on the cell's cytoprotective phenotype the same response can give an apoptotic or a necrotic response. So that is probably why genetic background is important.

Robakis: I want to ask how Dennis responds to the Braak & Braak data showing that AD pathology starts in the absence of amyloid depositions.

Peter Nelson: CSF Aβ ! We need not stay there too long. Nobody puts too much stock in that, I think. Clearly, AD patients have tons of Aβ in their brains!

Selkoe: Nick, Braak did not use sensitive Aβ ICC or biochemical assays to ensure that there is no Aβ accumulation in those brains. We don't know that all of their cases were definitely destined to develop AD.

Rneve: Dennis, how would you go about determining whether C100 or its catabolic products are the key players?

Selkoe: Rachael, even with the great individual variation, there are statistically significant increases in Aβ 42 in PS and APP, plasmas, and fibroblast media that exceed the mean levels in controls. This is the same as many other biological parameters in humans.

Dsmall: I think we have to be very careful in clearly separating amyloid deposition from plaques. There may be deposition where you cannot see plaques.

Peter Nelson: So you are saying, Dr. Selkoe, that you believe that there was non-noticed Aβ in those brains?

Selkoe: Peter, that is correct.

Peter Nelson: Dr. Selkoe has addressed that point before, but perhaps he ought to again: How come NFTs happen in the absence of plaques?

Selkoe: NFTs occur in numerous diseases of diverse cause and likely represent a response to many causes . In AD, I believe NFTs are caused indirectly or directly by Aβ accumulation.

Rneve: What are the data supporting that hypothesis, Dennis?

Peter Nelson: Drs Neve and Robakis have not responded to the point that neuritic plaques appear to be a physical nexus of EXTRAcellular amyloid and INTRAcellular neurofibrillary pathology. It's hard not to think something in the plaques is poisoning those neuronal processes.

Robakis: Dennis, how do you explain the fact that the amount of Aβ produced by FAD PS and APP mutations does not correlate with the years of onset of the disease

Rneve: Well, I said this earlier, I think that something in the plaques is poisoning those neuronal processes, but I am guessing that that is a relatively minor side effect of amyloid deposition in AD and is not a primary cause of the disease.

Selkoe: Bruce Yankner has very nice data showing that Aβ micro injections (200 pg) induce tau immunoreactive neuronal cell bodies in aged but not young rhesus monkeys. There are many other correlations between Aβ accumulation in mouse or human brain and the appearance of NFT or at least tau alterations.

Peter Nelson: I'll say, to look at a mature AD brain, with all its neuritic plaques, it's hard to think that's minor. Not saying it couldn't be....

Robakis: Rachael how can you explain that there are dystrophic neurites where there's no amyloid deposition and neuronal loss were there's no amyloid deposition?

Rneve: Tau alterations can be produced by many, many insults, if one looks at the literature, so it is hard to know what to make of Bruce's data.

Rneve: I can't explain it; perhaps Dennis can?

Peter Nelson: Marc Paradis points out, Bruce Yankner did controls.

Selkoe: Rachael, just as many things can intoxicate neurons including C100 and Aβ . So this doesn't convince us as much as the genetics which clearly point to a role for increased Aβ prior to tangle formation.

Rneve: Can you outline those genetic data, please?

Peter Nelson: Dr. Wolozin asks whether Dennis thinks the ability of human tau to form PHFs (in contrast to mouse tau) has anything to do with the toxicity and the different response of human vs. mouse brains to Aβ.

Selkoe: Peter which question should we address next?

Peter Nelson: Dr. Selkoe, I'd suggest you address the genetic data and segue into why humans appear able to develop neurofibrillary pathology, but mice, only in the presence of amyloid plaques, do not.

Marc Paradis: Sorry, but I must go. Thank you all for a very interesting discussion.

Peter Nelson tells Wolozin Sorry about's a little afield of the issue of the moment, though. My theory is that it pertains to the native amino acid sequence of the tau protein.

Marc Paradis tells June Kinoshita Nice turnout today! See you next week.

Peter Nelson tells Marc Paradis Thanks a lot, Marc!

Wolozin tells Peter Nelson Ug - the issue is whether Aβ kills human cells because it can stimulate PHF formation but it doesn't kill mouse neurons in vivo because no PHF.

Selkoe: All 4 AD linked genes have been shown in vitro and in humans to increase brain Aβ levels. Because APP mutations are a direct, primary cause of AD (even though very rare) we can say that increased Aβ can be an initiating event in AD. I think that all 4 AD genes increase Aβ by different mechanisms and various neuronal and glial changes occur secondarily. Also, DS argues strongly that excess APP and excess Aβ precede tangles and other neuronal changes.

Peter Nelson: Why doesn't Aβ appear to kill mouse neurons in transgenic animals?

Robakis: But the data is not there

Rneve: I think that we can only say that increased Aβ is an event that correlates with AD, not that initiates it.

Robakis: by definition

Selkoe: Bruce [Yankner] shows that rat brain is far more resistant to micro injected Aβ than monkey, both at old age. I think mouse neurons are probably more resistant to Aβ, and the high levels of APPs in Tg brains may also protect neurons.

Peter Nelson: What do you say about the FAD families with βAPP mutations, Dr. Robakis?

Robakis: in what respect?

Peter Nelson: As perhaps the most solid support for βAP proponents, how would you argue that it does NOT support their argument? ...I mean, the linkage of βAPP mutations to the AD phenotype.

Rneve: I would go back to Nishimoto's data, which show clearly that FAD mutants of APP can cause apoptosis even in the absence of the generation of Aβ from these mutants. I know that begs the questions of whether the FAD APP mutants kill in AD by causing apoptosis, but it's quite compelling data.

Robakis: As I have indicated earlier I think these mutations interfere with the intracellular transport. There are other examples of diseases caused by point mutations in transmembrane proteins that interfere with cellular transport.

Selkoe: Of course many transport defects could cause cellular disease but we have to come back to the phenotype in humans with AD and they invariably have large amounts of Aβ extracellularly.

Peter Nelson: After this answer, perhaps we'll wrap up. People should give additional questions to June, who'll pass them along to the panelists and they will be answered later.

Peter Nelson: A quickie: do the FAD mutations increase C-100 too?

Rneve: Yes, they do. All of the FAD APP mutations do.

Robakis: which may indicate problems with protein transport.

Rneve: Touche.

Peter Nelson: How about a last word from each of the panelists.

Robakis: These are very useful discussions because I feel that they advance and improve our views of the disease.

Rneve: I don't think that the correlation of the FAD mutations with increased release of Aβ is any more compelling than, say, the correlation of the FAD mutations with increased C100 or with apoptosis. We need to consider all of these hypotheses.

Selkoe: I think that Brian Cummings data showing correlation coefficients of greater than 0.9 between total Aβ burden and degree of cognitive impairment is very impressive. It can no longer be said that Aβ amount and also plaque amount does not correlate with presence and degree of dementia.

June Kinoshita: I want to thank everyone for being here and doing a marvelous job!

Peter Nelson: Thank you all very much. I hope most of the questions were answered to people's satisfactions.

June Kinoshita applauds fervently.

Rneve: Peter, you did a superb job on the commentary that you prepared for this!

Selkoe: I agree.

June Kinoshita: I want especially to thank Pete too for his deft moderating!

Peter Nelson: Thank you very much, Dr. Neve. I've long been a big fan of your work.

Rneve: I applaud Peter as well -- it was a tough job, done very well.

Robakis: I would like to thank everybody including Dennis, Rachael, June and Peter

June Kinoshita: Thank you Nick for braving the cybernet!

June Kinoshita applauds Robakis fervently.

Rneve: Yes, Nick, you never cease to surprise me.

June Kinoshita: Thank you Dennis and Rachael for your time and energy!

Peter Nelson: Okey-doke. Best to all, this has been very fun.

Rneve: Are there typing courses on the Internet?

June Kinoshita: Farewell until next time.

Peter Nelson tells Robakis A pleasure chatting with you, Dr. Robakis. I've followed your work w/interest for a while.

Peter Nelson tells Selkoe Nice chatting w/you again, Dr. Selkoe.

Selkoe: Peter: Great job. Let's keep in touch.


Background Text
By Peter Nelson

The debate at hand is whether the deposition of extracellular A-β peptide is an early and integral feature in the biochemical progression of Alzheimer's disease (AD). The purpose of this brief discussion is to concisely outline the review article's points questioning the data in support of a fundamental role of Aβ in AD; to describe some counter-arguments that Dr. Selkoe or others might use to justify a fundamental role of β in AD; and finally, to provide some editorial third- party perspectives on the issue.

Main points of the review article

1. The neurotoxicity of β has not been convincingly established.

i. Many different avenues of β neurotoxicity have been described, some overlapping, but others seemingly contradictory of each other. The article notes fourteen such hypothesized routes of β toxicity. There is confusion, rather than consensus, in this field.

"Pro β" response: Simply because β appears to be toxic in many ways is not a good argument that it is not toxic. Some of the hypotheses are perfectly compatible with each other. For example, were β to increase intracellular calcium, it would also be expected to induce oxidative stress, enhance neurotransmitter release, and cause apoptosis in some cells but necrosis in others.

A first-and-foremost plinth supporting the β Hypothesis is the fact that mere mutations in BAPP can cause AD. No other evidence is as persuasive or direct; BAPP is a direct player in AD. Hence, it is just a matter of figuring out how this molecule plays its role.

Third-party perspective: Whether or not each "story" of β neurotoxicity is valid or not, the practice of evaluating each one critically is essential. There is such a fog of seemingly positive results that one may be led to think that they all are true, which is almost certainly not the case. After all is said and done, the β peptide could be practically inert.

ii. The studies that have revealed neurotoxicity of β have used concentrations of the peptide that are much higher than are probably found in vivo.

"Pro β" response: As stated in the review article, the focal concentration of β (in a plaque) is probably much greater than that of CSF of AD patients, and is more likely to contai polymerized β.

Third-party perspective: Such differences between preparations of β exist that they shouldn't be cited in the same breath.

iii. If the β must be polymerized into fibrils (out of solution), it contradicts studies that suggest that specific receptor(s) are being stimulated by β.

"Pro β" response: Both could be true: certain types of toxicity could be stimulated by β in solution interacting with specific ligands; and polymerized β could cause neurotoxicity also.

Third-party perspective: This debate is the product of a difficult peptide preparation. It is hoped that this problem will resolve itself. Otherwise, it will never be clear what, in β preparations, is apparently causing toxicity.

iv. In vivo, β does not appear neurotoxic (transgenic mice, dogs, β injections).

"Pro β" response: In vivo, β transgenic mice do indeed demonstrate neuritic-like plaques, albeit without a PHF component ("neuritic" may be said to refer to dystrophic neurites, which are indeed surrounding some plaques in β transgenic mice). Moreover, in a paper not cited in the review, a transgenic β mouse was shown to develop widespread apoptotic cell death (La Ferla et al, Nature Genetics 9[1] 1995:21-30. Abstract.). Rhesus monkeys, which develop amyloid plaques, undergo mild loss of cognition in old age. And isn't it true, "you can't teach an old dog new tricks"?

Third-party perspective: A difficulty with demonstrating toxicity in vivo is that the relevance to AD is hard to assess. Even if you had a model with plaques and tangles, could you say that you had a true model of AD? After all, some think the processes are independent of each other. Until the true "key" is found, one can only try to discover that key, or put bandaids onto epiphenomena.


2. The importance of β(1-42) in the AD research world may be overblown.

i. Good circumstantial evidence exists that β(1-42) is important in AD plaques; however, the "smoking gun" of toxicity remains elusive.

"Pro β" response: β(1-42) is enriched in neuritic plaques, appears to seed further amyloid deposition, and seems to be toxic in vitro and in vivo. As yet, this still appears the closest thing to a "smoking gun".

Third-party perspective: It's very difficult to find a smoking gun, without finding an antidote that cures AD. Otherwise, one is never sure...

ii. The correlation of amyloid burden and AD dementia has been consistently shown to be poor.

"Pro β" response: Some recent studies underscore the obvious (albeit tautological): you don't have Alzheimer's without amyloid plaques. Moreover, the relation between amyloid burden and dementia is fairly good in some areas of brain. Finally, there may be some turnover of plaques (more so than tangles), which would render it impossible to gauge plaque burden to antemortem dementia.

Third-party perspective: The correlation of amyloid burden with antemortem dementia is the type of finding that depends critically upon neuroanatomy. Certain areas, as demonstrated by many people, develop plaques faster than others.

iii. Tangles and plaques can exist independently of each other.

"Pro β" response: The evidence of tangles occurring without plaques (as publicized by the Braaks [ref. 51 in the review article]), was argued against by Dr. Selkoe in an Alzheimer Research Forum discussion last year. He made the following three points (which I here paraphrase, with apologies to Dr. Selkoe):

a: Tangles occur in lots of neurological disorders. Perhaps the plaque-free brains with tangles would never develop AD.

b: Perhaps there was some amyloid in the Braak brains in areas they didn't stain.

c: Perhaps there were small β deposits not detectable by microscopy that were present in the Braak brains.

Third-party perspective: Apparently, low levels of tangles can occur without any plaque burden with aging. As amyloid burden increases, tangle formation accelerates. This is not such a bash on the β Hypothesis people. Neurons are vulnerable to tangle formation, which kills cells. Plaques cause more tangles to form, leading to dementia.

iv. Evidence that presenilin (PS) mutations increase the ratio of β(1-42):β(1-40) in cells is rendered less impressive by the data showing that simply increasing the amount of wild-type PS in cells and animals leads to increased ratios of β(1-42):β(1-40).

"Pro β" response: PS mutant transgenic mice develop amyloid plaque-like structures, whereas wt PS transgenic mice do not appear to do so.

Third-party perspective: PS seems to affect APP processing. This may or may not directly increase β(1-42), but the point is valid that PS studies more support than detract from the β Hypothesis.

v. The data that show AD patients have increased β in plasma and CSF (in comparison to controls) can be explained by other grounds than a causal role of β in AD.

"Pro β" response: The increased β in AD patients' plasma and CSF are not data integral to the Amyloid Cascade Hypothesis. However, it is intriguing that FAD patients have discernible rises in β in their plasma prior to appearance of AD symptoms.

Third-party perspective: The presence, or lack, of β in the plasma or CSF of AD patients is so fraught with variations, and possibilities for errant analysis, that it those data are hard to interpret.


3. Animal models don't support the β hypothesis.

i. Transgenic animals with upregulated β don't exhibit neuronal loss.

"Pro β" response: Transgenic animals with upregulated β do indeed have neuritic plaque-like structures and inflammation. Moreover, as stated above, transgenic mice with β do indeed exhibit neuronal loss, apoptosis, plaque-like structures, and gliosis (La Ferla, ibid.)

Third-party perspective: Transgenic animals with upregulated β have plaques that have dystrophic neurites. Now, this begs two questions: first, is it intracellular or extracellular β that is the culprit in AD? This is, perhaps, the understated main point that the present review is hammering home. An increasing stream of data supports the notion that INTRAcellular β (or some other BAPP product) could be the culprit in AD. Even if this were true, I don't think that β Hypothesis people such as Dr. Selkoe would be totally perplexed; he and others have been working on intracellular APP trafficking for years.

It is hard for anyone who has seen neuritic plaques to disbelieve that it is an important nidus of Alzheimer pathology. Not only are they an ugly bruise in the neuropil, but they are intimately associated with PHF-containing dystrophic neurites, and tend to connect anatomically with areas where tangles form. It is difficult to wave away the importance of animal models with β-containing neuritic plaque-like structures; and it is hard to accept an animal model that lacks these pathological hallmarks.

ii. The behavioral deficits seen in transgenic animals with upregulated β may not be specific, nor related exclusively to β production.

"Pro β" response: In the case of β transgenic mice with behavioral deficits, it is intriguing that without any discernible neuropathology (in contrast to animals with, say, massively upregulated neurofilament protein), β upregulation can induce behavioral deficits. Again, this is not a datum on which the β Hypothesis hangs its hat.

Third-party perspective: Evaluating behavioral deficits in mice is a nightmare. Until a pathological explanation for behavioral changes can be found, I think they are basically not very interpretable.

iii. Transgenic APP C100 mice appear to be a more faithful model of AD than transgenic mice with just β.

"Pro β" response: Transgenic APP C100 mice, despite their attractive pathological features, lack neuritic-like plaques, which are considered a fundamental neuropathological feature of AD. This is admittedly a circular argument; however, if it is hypothesized that neuritic plaques are NOT fundamental to AD neuropathology, then the burden of proof must be assumed by those who go against data-supported dogma.

Third-party perspective: See comment under 3.i above.


4. Data supports mechanisms other than those involving β (1-42) in AD neuropathology.

i. Errant trafficking of secretory vesicles through Golgi, endoplasmic reticulum could be responsible for AD pathobiology, as "all known FAD mutations occur in transmembranes that are also found in secretory vesicles".

"Pro β" response: Many proteins go through the secretory system in neurons, but ApoE does not appear to be one of them. Genetics is certainly the area which most supports the β hypothesis. Not only do APP mutations cause AD (this point is so pivotal to the β Hypothesis that it is a pity it was not addressed in the present review), but the other FAD mutations, and also Down syndrome patients, tend to further bolster the link between β production and Alzheimer's disease.

Third-party perspective: It is almost certain that intracellular trafficking plays a role in AD, either within or without the β Hypothesis. Proponents of that hypothesis are not invested in that being not the case.

ii. The normal function of PS and/or APP proteins may be disrupted, causing the pathology of AD.

"Pro β" response: If AD were a "loss of function" disease, the production of β could still be fundamental in proteins (including APP and PS) losing their functions.

Third-party perspective: Although β could be a factor in causing "loss of function" changes in AD, the elucidation of those changes could be the necessary key in finding a cure for the disease. Even if such a change were "downstream" of β formation, there could be a spiraling-downward loop which could be altered.

iii. Apoptotic mechanisms may be responsible that are independent of the β per se (involving either APP-->Go--> apoptosis, PS-->apoptosis, a/o other mechanisms).

"Pro β" response: Most currently-proposed mechanisms of apoptotic cell death currently include a place for β. And again, if APP or PS cause apoptosis in AD brain, surely the disruptive influence of β plaques helps lead to that end.

Third-party perspective: The knowledge about apoptosis in the aged brain is more sketchy than most people seem to admit. "Classical apoptosis" could quite simply not occur there. However, in the context of the current debate, were apoptosis to be relevant to AD, β could play a role at many levels.

iv. A myriad of non-β(1-42) mechanisms have recently been put forward for the majority of AD victims that suffer from late-onset disease; these include "inflammatory processes, oxidative damage, and mitochondrial mutations", in addition to Apo E-related mechanisms and others.

"Pro β" response: It is clear that only a minority of FAD cases are caused by BAPP mutations; hence, the β production will be "downstream" of some changes (e.g. mitochondria mutations, Apo E mechanisms) and "upstream" of others (e.g. oxidative stress, inflammatory processes). The fact that in the chaos of the AD brain many changes are noted does not subtract from there being a fundamental role for β in causing that chaos.

Third-party perspective: AD is obviously a heterogeneous disease, as noted by the review authors. Much remains to be learned. It is notable that the current discussants, Drs. Neve, Robakis, and Selkoe, are among the most preeminent and capable of the scientists searching for new knowledge in the field.


  1. The amyloid cascade hypothesis (ACH) does not fit with the natural and molecular history of Alzheimer's disease. Major points:

    • The first argument against the ACH is that tau and APP pathologies always grow in parallel in the human brain (Delacourte A, et al, Nonoverlapping but synergetic tau and APP pathologies in sporadic Alzheimer's disease.Neurology. 2002 Aug 13;59(3):398-407).
    • However, in sporadic AD, you can find tau pathology (in the entorhinal and hippocampal regions) without any trace of amyloid deposits, while the opposite is not true.
    • Furthermore, the evolution of the clinical impairments is fully explained with the progression of tau pathology in specific brain areas (Braak stages I to VI; Delacourte stages 1 to 10).

    Indeed, ACH derives from and is adapted to familial AD and the mutations on APP and PS1. However, even for these genetic cases, the clinical expression appears (generally around the age of 45 years) with the sudden development of tau pathology in neocortical areas.

    Of course, APP dysfunction is instrumental in the etiology of AD. But our hypothesis is that the defect is related to a loss of function of APP, via its carboxy-terminal fragments (APP-CTFs) that disappear in quality and quantitiy in FAD and sporadic AD (Sergeant N et al. Progressive decrease of amyloid precursor protein carboxy terminal fragments (APP-CTFs), associated with tau pathology stages, in Alzheimer's disease. J Neurochem. 2002 May;81(4):663-72). A participation of amyloid deposits is not excluded, but more likely at the last stages.

    In short, AD is a true tauopathy fueled by APP dysfunction (Delacourte & Sergeant, Brain Aging, 2004, vol 4, N°1, 15-21).


    . Progressive decrease of amyloid precursor protein carboxy terminal fragments (APP-CTFs), associated with tau pathology stages, in Alzheimer's disease. J Neurochem. 2002 May;81(4):663-72. PubMed.

Make a Comment

To make a comment you must login or register.


Webinar Citations

  1. Alzheimer's Disease: A Re-examination of the Amyloid Hypothesis

Paper Citations

  1. . Alzheimer's disease: a re-examination of the amyloid hypothesis. Trends Neurosci. 1998 Jan;21(1):15-9. PubMed.

Further Reading


  1. . Phenotypic change caused by transcriptional bypass of uracil in nondividing cells. Science. 1999 Apr 2;284(5411):159-62. PubMed.
  2. . Dirty transcripts from clean DNA. Science. 1999 Apr 2;284(5411):62-3. PubMed.