The paper by Lesne et al. is interesting. It would be more convincing if it had included additional controls/information relating to the Aβ oligomer.

For example, do the authors have evidence for this oligomer from in-vitro preparations of Aβ42? If not, why not? If they do, is it ThT-reactive?

Could the authors present TEM evidence of the oligomer, either generated via the transgene or from in-vitro preparations?

If, as I have assumed, the oligomer is only formed in vivo, perhaps only in transgenes, and has not been identified in in-vitro preparations, then some speculation as to why this should be so would be pertinent. It is apparently quite stable, as the authors were able to isolate it for subsequent injection into rats.

In relation to the final experiments in which the isolated oligomer was injected into rat brains, a control consisting of "the vehicle" is surely not sufficient to demonstrate activity of this particular oligomer. We are all aware that injections of Aβ cause behavioral changes in the rat. The authors could have used a positive control, for...  Read more

The paper by Lesne et al. is interesting. It would be more convincing if it had included additional controls/information relating to the Aβ oligomer.

For example, do the authors have evidence for this oligomer from in-vitro preparations of Aβ42? If not, why not? If they do, is it ThT-reactive?

Could the authors present TEM evidence of the oligomer, either generated via the transgene or from in-vitro preparations?

If, as I have assumed, the oligomer is only formed in vivo, perhaps only in transgenes, and has not been identified in in-vitro preparations, then some speculation as to why this should be so would be pertinent. It is apparently quite stable, as the authors were able to isolate it for subsequent injection into rats.

In relation to the final experiments in which the isolated oligomer was injected into rat brains, a control consisting of "the vehicle" is surely not sufficient to demonstrate activity of this particular oligomer. We are all aware that injections of Aβ cause behavioral changes in the rat. The authors could have used a positive control, for example, aggregated Aβ, to try to demonstrate that it was not simply the injection of Aβ, in any form, that produced the behavioral differences. In addition, the authors might have tried to demonstrate that the oligomer was actually present in the rat brain. Upon injection, it might have immediately aggregated or dissolved; we have no way of knowing.

The authors may be correct in their assertion that this oligomer causes the behavioral changes seen in both transgenic mice and rats, though the research as presented does not appear to do more than suggest a relationship. Given the weight afforded to research published in Nature, it is surprising that the lack of suitable controls was not commented upon in the accompanying News and Views.

View all comments by Chris Exley

Does this paper provide a new model of memory loss? No, but it advances our understanding of the basis of memory loss in a well-known transgenic mouse model of Alzheimer disease. Above all, the paper offers us a concrete biochemical entity to study and compare against other Aβ oligomer species that various groups have themselves found in recent years.

The paper fits nicely with prior studies that address the major question of what brain changes account for the deficits in memory and cognition in AD. Here is some historical context of this work: In the early 1990s, DeKosky and Scheff, 1990, as well as Robert Terry and Robert Katzman (Terry et al., 1991), showed that loss of synapses was the best correlate of the declines of memory and cognition in AD. Plaques did not correlate with memory and cognition, and tangles correlated slightly. But in these studies of the early 1990s, loss of synapses only accounted for about half the losses of memory and cognition in AD. Where might the...  Read more

Does this paper provide a new model of memory loss? No, but it advances our understanding of the basis of memory loss in a well-known transgenic mouse model of Alzheimer disease. Above all, the paper offers us a concrete biochemical entity to study and compare against other Aβ oligomer species that various groups have themselves found in recent years.

The paper fits nicely with prior studies that address the major question of what brain changes account for the deficits in memory and cognition in AD. Here is some historical context of this work: In the early 1990s, DeKosky and Scheff, 1990, as well as Robert Terry and Robert Katzman (Terry et al., 1991), showed that loss of synapses was the best correlate of the declines of memory and cognition in AD. Plaques did not correlate with memory and cognition, and tangles correlated slightly. But in these studies of the early 1990s, loss of synapses only accounted for about half the losses of memory and cognition in AD. Where might the missing 50 percent be?

In 2003, our group showed that the brains of AD patients were deficient in a protein, called dynamin 1, that is crucial to the functioning of synapses and, hence, for memory formation and information processing in the brain (Yao et al., 2003; Coleman and Yao, 2003). More specifically, dynamin 1 is a key protein in the trafficking of presynaptic vesicles that contain neurotransmitters. In 2005, Brent Kelly, Robert Vassar, and Adriana Ferreira showed that Aβ peptide caused depletion of dynamin 1, and they confirmed our major finding by showing depletion of dynamin 1 in a mouse model of AD (Kelly et al., 2005).

The current paper by Lesne et al. specifies the form of Aβ that probably was responsible for the loss of dynamin 1 described by Kelly et al. in the Tg2576 mouse model of AD, and by Yao et al. in human AD cases.

These papers all fit together when one posits that a major part of the missing 50 percent in DeKosky’s, Scheff’s, and Terry’s earlier observations lies in defective functioning of synapses that remain structurally present but are unable to function optimally due to deficient expression of dynamin 1 (and other molecules related to synaptic function), and, further, that this deficient expression of dynamin 1 is caused by a specific form of Aβ, which Lesne et al. have now identified. At present, more attention is being paid to Aβ effects on specific transmitters than on vesicle recycling. I believe the latter deserves focused exploration, as well. For one, it would be interesting to know whether Aβ effects on synaptic vesicle trafficking are selective.

One of the major questions unanswered by Lesne et al. lies in the fact that the mouse model they used contains a mutated form of the human APP molecule that is found in only a small percent of AD patients. The work of Kelly et al. apparently used the wild-type form of Aβ. Would Lesne et al. have obtained similar results with the wild-type form of Aβ?

View all comments by Paul Coleman

This is an impressive and important contribution. It links the appearance of a particular multimeric species of the amyloid-β peptide—Aβ*56—to a specific behavioral perturbation, and induces the same perturbation in naïve rats by reintroducing the Aβ*56 species purified from Tg2576 brains. It begins to address the conundrum that Aβ levels, soluble or insoluble, do not correlate with the onset and severity of behavioral changes in these animals.

Expectedly, this report stimulates a raft of questions, not with the work itself, but in teasing out more of the details and in stimulating new approaches. It will also energize corroboration of their findings in other Tg mouse models, as well as a search for correlates in Alzheimer disease brain. The findings from those studies will either further validate the animal model or set limits on its interpretation, both of which will be valuable. To begin to understand this complex paper, you must also study the supplementary information...  Read more

This is an impressive and important contribution. It links the appearance of a particular multimeric species of the amyloid-β peptide—Aβ*56—to a specific behavioral perturbation, and induces the same perturbation in naïve rats by reintroducing the Aβ*56 species purified from Tg2576 brains. It begins to address the conundrum that Aβ levels, soluble or insoluble, do not correlate with the onset and severity of behavioral changes in these animals.

Expectedly, this report stimulates a raft of questions, not with the work itself, but in teasing out more of the details and in stimulating new approaches. It will also energize corroboration of their findings in other Tg mouse models, as well as a search for correlates in Alzheimer disease brain. The findings from those studies will either further validate the animal model or set limits on its interpretation, both of which will be valuable. To begin to understand this complex paper, you must also study the supplementary information provided online.

Hypothesizing that a particular species of Aβ was responsible for deficits in memory retention that progress in a specific pattern with age in Tg 2576 mice, the authors developed a sequential extraction method to distinguish pools of Aβ peptide in the brains of these mice. They were able to empirically quasi-separate “extracellular-enriched” and “intracellular-enriched” fractions, although it is not clear how detergent-containing extractions can avoid solubilizing membranes. Perhaps a specific detergent/protein ratio limits the effect of the detergent.

Combined with primary neuronal and astrocyte culture work, this fractionation allowed them to conclude that oligomers of Aβ larger than trimers assemble extracellularly. Interestingly, cultured cells contained only trimers, while the extracellular species were trimers and tetramers, but not hexamers, nonamers, or Aβ*56 (dodecamers), which are observed in the whole aged animals. This could be due to the fact that the cultured cells are embryonic, not mature neurons, and they are not in a tissue environment. What is apparent is that oligomer formation in cells yields a different size spectrum of products from those reported in the literature for synthetic peptides and from those observed in aged (>6 months) animals (this article). This may be due to clearance mechanisms and/or modulating intra- and extracellular processes.

The importance of these observations is that they direct attention to extracellular events as being important in the maturation of trimers and tetramers into Aβ*56. This has not previously been appreciated. It also suggests that the change occurring at 6 months of age in Tg 2576 mice may be in the cellular environment. This will undoubtedly launch new lines of research focusing on that aspect. An extracellular target is also an easier one to reach with therapeutics.

In the studies reported here, Aβ*56 is purified by immunoprecipitation. Along with the specificity of that method for Aβ peptides comes the risk of missing cryptic epitopes. As an example, in Lesne et al., hexamers do not seem to IP well, yet they can be readily seen in non-IPed material. Antigen recovery by boiling of the blots before immunodetection affects the relative intensity of the species observed, as noted by the authors.

The authors are careful not to claim that Aβ*56 is the only species capable of inducing behavioral effects; indeed, they do not report testing the other SEC-separated species they isolate.

The story, of course, is far from complete with the identification of an Aβ*56 species that reproduces particular behavioral deficits in rats that are seen in the transgenic APP mice. Transgenic mice are at best a partial model of the uniquely human Alzheimer disease. There is minimal neuronal cell death in mice and in most cases (including Tg 2576), only moderate synapse loss is apparent. While more detailed study may reveal subtle changes in synaptic architecture/function, the devastation in the Alzheimer brain is not recapitulated. Thus, mice may represent a model of the earliest stages of a process that becomes Alzheimer disease in humans. That is a particularly important issue, because it is precisely at that stage at which intervention is likely to be most effective and detection of pathology first possible.

View all comments by Harry LeVine III

This study is impressive both for the breadth and detail of the experiments undertaken. Using the well-characterized Tg2576 APP transgenic mouse line, the authors searched for the appearance of an Aβ species that coincided with the first observed changes in spatial memory. Starting at 6 months, the time when cognitive changes are first apparent, the authors detected Aβ species that migrated on SDS-PAGE as nonamers and dodecamers. Aβ monomer, trimer, and hexamer were seen at earlier time points and were therefore not considered to have a deleterious effect on cognition. Indeed, comparison of spatial memory and the levels of Aβ monomer, trimer, hexamer, nonamer, and dodecamer revealed that only nonamer and dodecamer levels correlated with memory impairment.

The authenticity of these various Aβ species as discrete assemblies was confirmed using a gel filtration paradigm previously employed to fractionate cell culture-derived low-n oligomers (Walsh et al., 2005), and was combined with immunoaffinity chromatography to achieve...  Read more

This study is impressive both for the breadth and detail of the experiments undertaken. Using the well-characterized Tg2576 APP transgenic mouse line, the authors searched for the appearance of an Aβ species that coincided with the first observed changes in spatial memory. Starting at 6 months, the time when cognitive changes are first apparent, the authors detected Aβ species that migrated on SDS-PAGE as nonamers and dodecamers. Aβ monomer, trimer, and hexamer were seen at earlier time points and were therefore not considered to have a deleterious effect on cognition. Indeed, comparison of spatial memory and the levels of Aβ monomer, trimer, hexamer, nonamer, and dodecamer revealed that only nonamer and dodecamer levels correlated with memory impairment.

The authenticity of these various Aβ species as discrete assemblies was confirmed using a gel filtration paradigm previously employed to fractionate cell culture-derived low-n oligomers (Walsh et al., 2005), and was combined with immunoaffinity chromatography to achieve purification of the dodecamer.

The authors then conducted the most important and compelling experiment of their study: They injected purified dodecamer into the ventricle of normal pre-trained rats and tested if the injected dodecamer could alter spatial memory. Rats given dodecamer showed a dramatic fall-off in performance; thus, the dodecamer shown to correlate with decreased cognition in Tg2576 mice was also capable of directly mediating impairment of memory in normal rats.

These studies demonstrate for the first time that a soluble, brain-derived form of Aβ can directly mediate brain dysfunction in the absence of neurodegeneration. They open up new avenues of investigation, and yet, as with all scientific advances, the Lesne study raises more questions than it answers. Going forward it will be vitally important to validate the human relevance of the Tg2576 dodecamer—is it present in human brain or CSF? Can it be detected in other animal models of AD? While there is no doubt that Aβ dodecamer present in Tg2576 brain is capable of impairing memory, it is not clear if this species also exists in human brain. Based on the novel homogenization protocol used by Lesne et al., one would predict that Aβ dodecamer should be present in the interstitial fluid and by extension should be readily detectable in CSF. To my knowledge, no such Aβ assembly has been detected in human CSF to date. Indeed, in prior studies, high-molecular-weight Aβ oligomers were not detected in human CSF, whereas Aβ monomer, dimers and trimers were consistently detected (Ida et al., 1996; Walsh et al., 2000).

Of course, the Tg2576 line is a model for AD, and, like all models, it may differ from the human condition. For instance, the authors demonstrate a time-dependent increase in "extracellular-enriched" Aβ monomer, yet in humans it is well documented that CSF Aβ falls with increasing disease severity (Nitsch et al., 1995; Andreasen et al., 1999; Lewczuk et al., 2003). Moreover, recent studies indicate that the fall in Aβ monomer (it has been previously demonstrated that the Takeda ELISA does not readily detect Aβ oligomers; see Morishima and Ihara, 1998) is due to sequestration into senile plaques (Fagan et al., 2005). Together, these results suggest that the overall economy of Aβ in the Tg2576 may be different from that of human brain, and raise the possibility that Aβ assemblies other than or in addition to Aβ dodecamer underlie the memory loss that characterizes AD.

View all comments by Dominic Walsh

Amyloid-β protein dodecamer in the brain impairs memory in the Tg2576 mouse
The experience from genetic findings in the early 1990s strongly point to Aβ as the culprit in Alzheimer disease. However, we still do not understand how Aβ confers cognitive dysfunction and neuronal atrophy. Recent years have witnessed an increased interest in soluble Aβ oligomers as being the important pathogenic form of Aβ. This article is a significant contribution to the field. Most impressive is perhaps the author’s ability to isolate a soluble Aβ species from the brain and prove that it affects cognition. The research team, headed by Karen Ashe, has for a long time sought the elusive Aβ species responsible for cognitive decline in their transgenic mouse model Tg2576, which harbors the Swedish APP mutation.

Tg2576 lack neuropathology and are cognitively unimpaired until 6 months of age, when spatial memory declines but then remains stable for another 7-8 months. Animals aged more than 14 months develop neuropathology including neuritic plaques containing amyloid-β peptides and...  Read more

Amyloid-β protein dodecamer in the brain impairs memory in the Tg2576 mouse
The experience from genetic findings in the early 1990s strongly point to Aβ as the culprit in Alzheimer disease. However, we still do not understand how Aβ confers cognitive dysfunction and neuronal atrophy. Recent years have witnessed an increased interest in soluble Aβ oligomers as being the important pathogenic form of Aβ. This article is a significant contribution to the field. Most impressive is perhaps the author’s ability to isolate a soluble Aβ species from the brain and prove that it affects cognition. The research team, headed by Karen Ashe, has for a long time sought the elusive Aβ species responsible for cognitive decline in their transgenic mouse model Tg2576, which harbors the Swedish APP mutation.

Tg2576 lack neuropathology and are cognitively unimpaired until 6 months of age, when spatial memory declines but then remains stable for another 7-8 months. Animals aged more than 14 months develop neuropathology including neuritic plaques containing amyloid-β peptides and further cognitive deficits. The authors posited the existence of an Aβ oligomeric form, designated Aβ*, responsible for early cognitive decline in Tg2576. Two criteria were used: Aβ* should appear at 6 months of age and remain stable between 6-14 months of age. The best correlation was found between Aβ 12-mers and spatial memory.

Curiously, non-transgenic mice also show a trend toward impaired spatial memory at 6 months of age (Figure 1a). It would be interesting to investigate whether increase in Aβ* is coincident with cognitive deficits also in other APP mouse models, since cognitive dysfunction is known to be highly dependent upon strain background.

Most surprisingly, levels of Aβ*56 in brain do not increase upon onset of senile plaque deposition when total Aβ levels increase 100-fold (Kawarabayashi et al. 2001). This would tend to suggest a dichotomous model of Aβ amyloidosis in the brain, where Aβ* formation is unrelated to senile plaque formation. It would be interesting to determine turnover of endogenous Aβ* in the brain, especially since Aβ* confers a transient effect on memory retention. What would happen with levels of Aβ* in Tg2576 following acute or chronic treatment with a potent γ-secretase inhibitor or in a TET-off APP transgenic model?

Most important would be, of course, to investigate if Aβ* exists in the brain or CSF of Alzheimer disease patients, and if Aβ* levels are linked to mild cognitive impairment (MCI) and further cognitive decline in the human disease.

View all comments by Lars Lannfelt
View all comments by Lars Nilsson

To their credit, the authors have attempted to look for early changes in the TG 2576 mouse model, which are more likely to deal with pathogenesis than pathogenic consequences. Lesne et al. have identified an unusual, high molecular-weight component in the brains of these mice that contains Abeta determinants and is only present before amyloid deposits accumulate. The claim that this material is necessarily all derived from extracellular spaces is questionable, since it was isolated from detergent-solubilized brain tissue. It is also not clear how much of the 56K band is made up of Abeta peptides. The authors describe an Abeta-derived peptide as representing the "core" of the material, but careful mass spec analysis should have revealed how much and what else was present in the sample. Until this is done, it is premature to declare this a special form of Abeta. I also agree that the biological activity of this material has not yet been studied adequately.

View all comments by Vincent Marchesi

Star-struck by Amyloid
Lesne and colleagues show that Aβ*56 is found in cognitively impaired Tg2576 animals without Aβ plaques, but not in unimpaired animals, and that it correlates to early declines in memory but not later ones. Notably, when isolated and injected into rats, Aβ*56 leads to reversible cognitive deficits. This is an interesting study and will definitely appeal to supporters of the amyloid hypothesis. However, before we get ahead of ourselves, a few salient aspects bear remembrance.

First, different groups have reported that knockout of PS1 (i.e., no Aβ and probably no Aβ*56, either), while attenuating Aβ pathology in APP mutant transgenic mice, does not cure cognitive deficits (Dewachter et al., 2002; Saura et al., 2005). Therefore, cognitive deficits do not relate to Aβ (in any guise, even *). Second, mitochondrial, apoptotic, and oxidative events all precede frank Aβ deposition and are linked to cognitive decline in APP transgenic mice (Pratico et al., 2001; Reddy et al., 2004). Since oxidative stress leads to increases in Aβ (Yan et al., 1995;...  Read more

Star-struck by Amyloid
Lesne and colleagues show that Aβ*56 is found in cognitively impaired Tg2576 animals without Aβ plaques, but not in unimpaired animals, and that it correlates to early declines in memory but not later ones. Notably, when isolated and injected into rats, Aβ*56 leads to reversible cognitive deficits. This is an interesting study and will definitely appeal to supporters of the amyloid hypothesis. However, before we get ahead of ourselves, a few salient aspects bear remembrance.

First, different groups have reported that knockout of PS1 (i.e., no Aβ and probably no Aβ*56, either), while attenuating Aβ pathology in APP mutant transgenic mice, does not cure cognitive deficits (Dewachter et al., 2002; Saura et al., 2005). Therefore, cognitive deficits do not relate to Aβ (in any guise, even *). Second, mitochondrial, apoptotic, and oxidative events all precede frank Aβ deposition and are linked to cognitive decline in APP transgenic mice (Pratico et al., 2001; Reddy et al., 2004). Since oxidative stress leads to increases in Aβ (Yan et al., 1995; Li et al., 2004), we suspect this is the true star. Third, related to these issues, mutations in APP cause increases in oxidative stress (Yamatsuji et al., 1996; Hashimoto et al., 2000).

In sum, Aβ*56 may be the brightest star, but, as any amateur astronomer can attest, “stars that burn brightest burn fastest and thus have the shortest lifetimes.”

References:
Dewachter I, Reverse D, Caluwaerts N, Ris L, Kuiperi C, Van den Haute C, Spittaels K, Umans L, Serneels L, Thiry E, Moechars D, Mercken M, Godaux E, Van Leuven F. Neuronal deficiency of presenilin 1 inhibits amyloid plaque formation and corrects hippocampal long-term potentiation but not a cognitive defect of amyloid precursor protein [V717I] transgenic mice. J Neurosci. 2002 May 1;22(9):3445-53. Abstract

Hashimoto Y, Niikura T, Ito Y, Nishimoto I. Multiple mechanisms underlie neurotoxicity by different types of Alzheimer's disease mutations of amyloid precursor protein. J Biol Chem. 2000 Nov 3;275(44):34541-51. Abstract

Li F, Calingasan NY, Yu F, Mauck WM, Toidze M, Almeida CG, Takahashi RH, Carlson GA, Flint Beal M, Lin MT, Gouras GK. Increased plaque burden in brains of APP mutant MnSOD heterozygous knockout mice. J Neurochem. 2004 Jun;89(5):1308-12. Abstract

Pratico D, Uryu K, Leight S, Trojanoswki JQ, Lee VM. Increased lipid peroxidation precedes amyloid plaque formation in an animal model of Alzheimer amyloidosis. J Neurosci. 2001 Jun 15;21(12):4183-7. Abstract

Reddy PH, McWeeney S, Park BS, Manczak M, Gutala RV, Partovi D, Jung Y, Yau V, Searles R, Mori M, Quinn J. Gene expression profiles of transcripts in amyloid precursor protein transgenic mice: up-regulation of mitochondrial metabolism and apoptotic genes is an early cellular change in Alzheimer's disease. Hum Mol Genet. 2004 Jun 15;13(12):1225-40. Epub 2004 Apr 28. Abstract

Saura CA, Chen G, Malkani S, Choi SY, Takahashi RH, Zhang D, Gouras GK, Kirkwood A, Morris RG, Shen J. Conditional inactivation of presenilin 1 prevents amyloid accumulation and temporarily rescues contextual and spatial working memory impairments in amyloid precursor protein transgenic mice. J Neurosci. 2005 Jul 20;25(29):6755-64. Abstract

Yamatsuji T, Matsui T, Okamoto T, Komatsuzaki K, Takeda S, Fukumoto H, Iwatsubo T, Suzuki N, Asami-Odaka A, Ireland S, Kinane TB, Giambarella U, Nishimoto I. G protein-mediated neuronal DNA fragmentation induced by familial Alzheimer's disease-associated mutants of APP. Science. 1996 May 31;272(5266):1349-52. Abstract

Yan SD, Yan SF, Chen X, Fu J, Chen M, Kuppusamy P, Smith MA, Perry G, Godman GC, Nawroth P, et al. Non-enzymatically glycated tau in Alzheimer's disease induces neuronal oxidant stress resulting in cytokine gene expression and release of amyloid beta-peptide. Nat Med. 1995 Jul;1(7):693-9. Abstract

View all comments by Hyoung-gon Lee
View all comments by Akihiko Nunomura
View all comments by George Perry
View all comments by Mark A. Smith
View all comments by Xiongwei Zhu

This exciting paper set outs to define the site and conformation of Aβ in the brain that may be critical for cognitive dysfunction in Tg2576 mice. The co-occurrence of Aβ*56 with behavioral alterations is quite interesting, yet aspects of the study are surprising. Aβ* does not progressively increase, while Alzheimer disease and Tg2576 mice are characterized by progressive synaptic pathology. Aβ* appears at the onset of what seems to be a progressive decline in behavior in Tg2576 mice, were it not for transient improvement at 13 months, which surprisingly also occurs in wild-type mice.

The data used to support that Aβ* accumulates extracellularly in Tg2576 mice are challenging. As suggested in previous comments (LeVine; Marchesi), it would seem difficult to be certain that one is mainly looking at extracellular peptides after detergent treatment (0.01 percent NP-40; 0.1 percent SDS) and homogenization of the intricate mass of neurons and processes of brain by 10 passages through a 20-gauge needle. The authors did provide some data on other intracellular proteins not leaking...  Read more

This exciting paper set outs to define the site and conformation of Aβ in the brain that may be critical for cognitive dysfunction in Tg2576 mice. The co-occurrence of Aβ*56 with behavioral alterations is quite interesting, yet aspects of the study are surprising. Aβ* does not progressively increase, while Alzheimer disease and Tg2576 mice are characterized by progressive synaptic pathology. Aβ* appears at the onset of what seems to be a progressive decline in behavior in Tg2576 mice, were it not for transient improvement at 13 months, which surprisingly also occurs in wild-type mice.

The data used to support that Aβ* accumulates extracellularly in Tg2576 mice are challenging. As suggested in previous comments (LeVine; Marchesi), it would seem difficult to be certain that one is mainly looking at extracellular peptides after detergent treatment (0.01 percent NP-40; 0.1 percent SDS) and homogenization of the intricate mass of neurons and processes of brain by 10 passages through a 20-gauge needle. The authors did provide some data on other intracellular proteins not leaking out in the process, although one might expect cytoskeleton proteins, such as tau and MAP2, to be more readily retained in cells in the presence of detergent compared to a hydrophobic/lipid-associated peptide such as Aβ. A readily releasable marker such as LDH could have been helpful.

In the supplement, the authors compare previous work addressing Aβ increases in Tg2576 mice, including work from our lab on intraneuronal Aβ42 increases. Their interpretation appears inconsistent with our immunoelectron microscopy studies demonstrating pathological intraneuronal Aβ42 increases and oligomerization in Tg2576 mouse brains (Takahashi et al., 2002; 2004). Remarkably, a fascinating report relating Pin1 and amyloid by Lu and colleagues (Pastorino et al., 2006) in the following issue of Nature provided further confirmation of intraneuronal Aβ accumulation in MVBs of Tg2576 mice.

Can the findings of Lesne et al. be reconciled with evidence for intraneuronal Aβ? It is possible that Aβ* is the specific oligomer responsible for the invariable association of intracellular pathology with Aβ oligomers (Takahashi et al., 2004), which could then be released from neurites following destruction of the plasma membrane from within. This is not inconsistent with a subsequent important role for extracellular Aβ* as well. A potentially similar scenario of intracellular amyloid formation is described in a recent article on diabetes, another common amyloid associated age-related disease (Paulsson et al., 2006).

View all comments by Gunnar K. Gouras

I would just like to comment on the questions/remarks that followed our article. First and foremost, I would like to point out that we did not write in the article that Aβ*56 is an assembly composed of 12 units of Aβ. We did not include any hard data that would directly demonstrate this statement. What we did mention, however, is the possibility that Aβ*56 could represent a 12-mer because of the following observations: 1) Aβ trimers are formed intracellularly and are secreted by neurons in vivo and in vitro; 2) Aβ-immunoreactive species of high molecular weights (above 20 kDa) migrate at molecular weights that match theoretical migrations for 6-mer, 9-mer, and 12-mers of Aβ1-42. It remains to be determined whether these proteins/assemblies are only composed of Aβ, but we postulated so due to the fact that trimers are predominant in vitro and in vivo and only multiples of three monomers appear to form these Aβ-immunoreactive larger structures in vivo. Further analyses are underway to confirm our hypothesis.

View all comments by Sylvain Lesne

Normally, soluble Aβ molecule (39-43 amino acids) undergoes conformational changes in disease and is deposited in the brain as insoluble fibrils, oligomers and protofibrills. Previously it was demonstrated that Aβ neurotoxicity required insoluble fibril formation (mainly Aβ42 and to lesser degree Aβ40) (Lorenzo, 1994) and the fibrils served as inducers of neuronal apoptosis (Loo, 1993). Recently, emphasis has shifted to smaller soluble Aβ. Aβ42 dimers and trimers naturally secreted from a 7PA2 cell line were suggested to be responsible for the disruption of cognitive functions (Cleary, 2005). Importantly, intraventricular injection of such Aβ42 small oligomers inhibited long-term potentiation (LTP) in rat hippocampus and an anti-Aβ monoclonal antibody (6E10) that binds to N-terminal region of Aβ42 prevented this inhibition (Klyubin, 2005). It has also been demonstrated that passive immunization with...  Read more

Normally, soluble Aβ molecule (39-43 amino acids) undergoes conformational changes in disease and is deposited in the brain as insoluble fibrils, oligomers and protofibrills. Previously it was demonstrated that Aβ neurotoxicity required insoluble fibril formation (mainly Aβ42 and to lesser degree Aβ40) (Lorenzo, 1994) and the fibrils served as inducers of neuronal apoptosis (Loo, 1993). Recently, emphasis has shifted to smaller soluble Aβ. Aβ42 dimers and trimers naturally secreted from a 7PA2 cell line were suggested to be responsible for the disruption of cognitive functions (Cleary, 2005). Importantly, intraventricular injection of such Aβ42 small oligomers inhibited long-term potentiation (LTP) in rat hippocampus and an anti-Aβ monoclonal antibody (6E10) that binds to N-terminal region of Aβ42 prevented this inhibition (Klyubin, 2005). It has also been demonstrated that passive immunization with monoclonal antibodies (NAB61), that specifically recognizes a pathologic conformation present in Aβ dimers, soluble oligomers and higher order species of Aβ, resulted in rapid improvement in spatial learning and memory (Lee, 2006). Other authors showed that 12-mer oligomers of Aβ42, also known as Aβ-derived diffusible ligands (ADDLs) increased about 70-fold in AD patient’s brains over controls (Gong, 2003; Klein, 2006). Collectively, these data suggest that the Aβ oligomers of various sizes are the most pathologic substrate responsible for disrupting neuronal functions and cognitive decline in AD.

The current paper showed that although different forms of Aβ42 are deposited in the brains of aged APP/Tg2576 mice, the memory deficits are induced in 6-month-old or older mice by that accumulated 12-mer oligomer, termed Aβ*56. This is an interesting paper that indicates that levels of soluble or insoluble Aβ do not fully correlate with behavioral changes in this mouse model of AD. Instead only levels of 9-mer (p = 0.0169; r2 = 0.4505) and 12-mer (p = 0.0014; r2 = 0.6556) oligomers showed an inverse correlation with spatial memory. These data indicates that AD therapy should target particular species of Aβ that are responsible for AD like pathology and memory deficits. Thus, if Aβ*56 is a major player in AD that is implicated in memory deficits in middle-aged Tg2576 mice, then learning about misfolding of human Aβ peptide is an important task. As we gain more knowledge of the mechanisms of assembly of Aβ peptides more potent AD therapies will be developed.

One important conclusion from this paper is that such AD therapy should be based on the prevention of accumulation of oligomers, rather than on clearing already formed toxic forms of Aβ. In other words AD therapy that can block oligomerization of Aβ should be started earlier, before the accumulation of monomers in the brains. This can be done by the prototype AD vaccine that will be able to generate antibodies that can bind all forms of Aβ and block oligomerization of the monomeric peptide. Of course such a vaccine should be used in middle-aged healthy people to prevent generation of AD-like pathology, rather than for vaccination of elderly AD patients with immunosenescence (therapeutic vaccine strategy). Such a vaccine would have to be safe and should be able to generate high titers of antibodies that can block oligomerization of β amyloid peptide, although they can be specific to any form of Aβ (monomers, oligomers, and fibrils).

Thus, the important aim of AD-immunotherapy research must be the identification of the most safe and immunogenic form of the vaccine that can generate therapeutically potent antibodies that can block oligomerization of the peptide. For example, using an epitope vaccine strategy we recently generated polyclonal antibodies specific to the N-terminal region of Aβ that can not only bind all forms of Aβ, but also delay oligomerization of Aβ42 in vitro (paper submitted). In the nearest future it will be important to test the ability of this and other prototype AD vaccines to block generation of Aβ*56 in the brains of immunized APP/Tg 2576 or other APP/Tg mice.

View all comments by Michael G. Agadjanyan