The idea that intracellular Aβ injures neurons and causes dementia prior to the formation of extracellular deposits is not obviously reconcilable with other evidence indicating that "full-blown" structural pathology precedes dementia (Crystal H, Dickson D, Fuld P, Masur D, Scott R, Mehler M, Masdeu J, Kawas C, Aronson M, Wolfson L. Clinico-pathologic studies in dementia: nondemented subjects with pathologically confirmed Alzheimer’s disease. Neurology 1988; 38: 1682-1687. Abstract. ; Price JL, Morris JC. Tangles and plaques in nondemented aging and "preclinical" Alzheimer’s disease. Ann Neurol 1999; 45: 358-368. Abstract).

Comment from Bruce Yankner, posted 15 May 2000

I would like to comment on several of the points made by Dr. Gouras in the background to our...  Read more

The idea that intracellular Aβ injures neurons and causes dementia prior to the formation of extracellular deposits is not obviously reconcilable with other evidence indicating that "full-blown" structural pathology precedes dementia (Crystal H, Dickson D, Fuld P, Masur D, Scott R, Mehler M, Masdeu J, Kawas C, Aronson M, Wolfson L. Clinico-pathologic studies in dementia: nondemented subjects with pathologically confirmed Alzheimer’s disease. Neurology 1988; 38: 1682-1687. Abstract. ; Price JL, Morris JC. Tangles and plaques in nondemented aging and "preclinical" Alzheimer’s disease. Ann Neurol 1999; 45: 358-368. Abstract).

Comment from Bruce Yankner, posted 15 May 2000

I would like to comment on several of the points made by Dr. Gouras in the background to our discussion. The prevailing amyloid hypothesis of AD does not require that plaques be the major toxic Aβ component. Rather, extracellular aggregated Aβ is proposed to be the major toxic component, and toxicity can be mediated by extracellular Aβ prior to plaque formation. This hypothesis is consistent with in vitro and in vivo Aβ toxicity studies from many labs which show that extracellular aggregated Aβ, not plaques per se, is neurotoxic. The proposal that intracellular Aβ is also toxic and contributes to the neurodegenerative process should be carefully considered. However, there is still not a single report that shows intracellular Aβ toxicity. Sam Gandy's point is important - the literature suggests that dementia occurs only when extracellular pathology is evident. This does not rule out a contribution of intracellular Aβ, but it suggests that such an effect would be unlikely to precede extracellular Aβ deposition.

Several arguments were advanced to support an intracellular Aβ mechanism.

1. Recent reports show intraneuronal Aβ42 in a cultured neuronal cell line, in AD brain and in AD-vulnerable neurons before plaques and tangles. The key issue is not whether Aβ is present in neurons but whether it does anything when sequestered in an intracellular vesicular compartment. Many groups have found that intracellular Aβ, including Aβ42, is a minor component of total Aβ production. We have detected intracellular Aβ42 in human cortical neuronal cultures from normal and Down's syndrome specimens, but this is a minor component of total Aβ, even in 2 month old cultures. More importantly, in AD brains, intracellular Aβ42 appears to be a minor component of total Aβ42 immunoreactivity.

2. It was noted that extracellular Aβ toxicity requires "supraphysiological levels". The micromolar concentrations of aggregated Aβ required for Aβ toxicity in vitro is in the range of Aβ levels in affected AD cortex (Selkoe et al., 1986). Moreover, Aβ toxicity in vivo requires only a single plaque-equivalent dose (Geula et al., 1998). If Aβ toxicity occurred in the physiological nanomolar range, we might all have AD.

3. The behavioral and physiological changes in APP-transgenic mice prior to the appearance of many plaques argues against a plaque-based mechanism. The behavioral studies in APP-transgenic mice have been difficult to interpret because they are very strain-dependent. I think it is safe to say that something is happening to the transgenic brain which is separable from plaque formation, as suggested by Karen Hsiao and Lennart Mucke. However, what it is and whether it is relevant to AD is unclear. It is certainly too early to conclude anything about its cause. My guess is that it may relate to pre-plaque accumulation of fibrillar Aβ.

4. The recent study by Naslund et al., which correlates brain Aβ levels with disease progression, has been taken as evidence implicating soluble Aβ. However, this study does not provide any information about the physical or cellular state of Aβ. The paper suggests that Aβ levels can be elevated prior to the appearance of plaques, but this Aβ could be extracellular, intracellular, soluble, oligomeric or fibrillar.

My own view is that both extracellular and intracellular Aβ may contribute to the neurodegenerative process. However, to seriously consider a role for intracellular Aβ, we need a study that convincingly demonstrates intracellular Aβ toxicity.

View all comments by Samuel Gandy

From various studies examining the toxicity of Aβ, it is clear that random coil monomeric Aβ is not toxic whereas oligomeric aggregates (protofibrils, ADDLs or fibrils) are. Thus the critical issue with respect to Aβ toxicity is its aggregation state. An ideal anti-aggregation therapy would prevent the production of the first toxic assembly of Aβ.

It has been widely assumed that Aβ aggregation is initiated extracellularly. However, we have demonstrated (data recently presented at the Society for Neuroscience in Miami) that SDS-stable oligomers of Aβ (primarily dimers) are first generated intraneuronally. This finding represents the first direct observation of the aggregation state of Aβ in human neurons (earlier studies on intraneuronal Aβ did not deal with aggregation state) and establishes that oligomerization is initiated within cells. Irrespective of whether intraneuronal oligomers are toxic per se or mediate their effect after export to the extracellular space, their site of origin is now known. Thus, it would be desirable to interfere with Aβ dimerization...  Read more

From various studies examining the toxicity of Aβ, it is clear that random coil monomeric Aβ is not toxic whereas oligomeric aggregates (protofibrils, ADDLs or fibrils) are. Thus the critical issue with respect to Aβ toxicity is its aggregation state. An ideal anti-aggregation therapy would prevent the production of the first toxic assembly of Aβ.

It has been widely assumed that Aβ aggregation is initiated extracellularly. However, we have demonstrated (data recently presented at the Society for Neuroscience in Miami) that SDS-stable oligomers of Aβ (primarily dimers) are first generated intraneuronally. This finding represents the first direct observation of the aggregation state of Aβ in human neurons (earlier studies on intraneuronal Aβ did not deal with aggregation state) and establishes that oligomerization is initiated within cells. Irrespective of whether intraneuronal oligomers are toxic per se or mediate their effect after export to the extracellular space, their site of origin is now known. Thus, it would be desirable to interfere with Aβ dimerization within neurons.

View all comments by Dennis Selkoe
View all comments by Dominic Walsh

Reply to Sam Gandy by Gunnar Gouras
We all agree that "full-blown" structural pathology (plaques and tangles) precede and accompany dementia. What we are especially considering is whether there is any role for intraneuronal Aβ in causing neuronal dysfunction early on in the AD disease process. I don't think that anyone is convinced that intraneuronal Aβ has a clear and seperate neurotoxic effect, but rather some of us are intrigued by the possibility that there may be an early pathological role for intraneuronal Aβ42. Dementia is preceded by probably at least a few years of initially subtle and then mild cognitive impairement (MCI), the pathological basis of which is not established.

I appreciate the instructive comments of Dr. Yankner, and would like to add (Re: his point 1) that a major reason for my interest in intracellular Aβ is that I studied the same brains as Naslund et al. did in their ELISA study with immunohistochemistry, and observed the most prominent Aβ as intraneuronal Aβ42 within AD vulnerable neurons and not in the brain...  Read more

Reply to Sam Gandy by Gunnar Gouras
We all agree that "full-blown" structural pathology (plaques and tangles) precede and accompany dementia. What we are especially considering is whether there is any role for intraneuronal Aβ in causing neuronal dysfunction early on in the AD disease process. I don't think that anyone is convinced that intraneuronal Aβ has a clear and seperate neurotoxic effect, but rather some of us are intrigued by the possibility that there may be an early pathological role for intraneuronal Aβ42. Dementia is preceded by probably at least a few years of initially subtle and then mild cognitive impairement (MCI), the pathological basis of which is not established.

I appreciate the instructive comments of Dr. Yankner, and would like to add (Re: his point 1) that a major reason for my interest in intracellular Aβ is that I studied the same brains as Naslund et al. did in their ELISA study with immunohistochemistry, and observed the most prominent Aβ as intraneuronal Aβ42 within AD vulnerable neurons and not in the brain parenchyma. I do not know whether this is from Aβ42 uptake (see Bahr BA et l., 1998) or from endogenous accumulation. I also wonder if he has tried to extract Aβ42 from neuron cell lysate with formic acid (see Skovronsky et al., 1998) or another more stringent method. Lastly, there was one report published in Nature Medicine last year of neuronal degeneration in thepresence of intraneuronal Aβ42 acumulation but not plaque formation in a FAD PS1 trangenic mouse strain (Chui et al., 1999).

View all comments by Gunnar K. Gouras