The first clinical trial may be dead, but AD vaccination research is alive and kicking. Yesterday, researchers led by Steven Paul at Lilly Research Laboratories in Indianapolis, reported in the online Nature Neuroscience that a single shot of an Aβ-binding antibody reversed memory deficits in transgenic mice even while these mice's brains contained copious amyloid plaques.

At first glance, the study is surprising. For one, it appears paradoxical that memory would improve overnight in a model of a disease thought to be due to a decade-long pathologic process that eventually kills populations of neurons and thus robs the brain of its abilities. For another, the study runs counter to the prevailing notion that amyloid deposits are the chief cause of neural toxicity, instead lending support to a growing view that a changeable, soluble pool of non-fibrillar Aβ may do some damage, as well. Finally, it suggests functional improvement might be gained from limited applications of passive immunization, a type of therapy not often used for chronic diseases.

The scientists, working with colleagues at Washington University School of Medicine in St. Louis, Missouri, and elsewhere, extended their previous studies of the m266 monoclonal antibody in PDAPP transgenic mice, which had shown that repeated immunization over five months could lower brain Aβ levels by acting as a peripheral "sink" (see ARF news item). In today's paper, first author Jean-Cosme Dodart et al. treated 2 year-old PDAPP mice with m266 for six weeks and found that aged old mice performed almost as well in an object recognition memory task as did young wild-type control mice, even though they had as many Aβ deposits in their hippocampus and cerebral cortex as did untreated PDAPP mice, which perform poorly in this test (Dodart et al, 1999).

Then the authors injected a single dose of m266 into middle-aged PDAPP mice and tested them the next day. Again, the treated PDAPP mice performed almost as well as did age-matched wild-type controls. Roughly the same was true with the holeboard task, a learning test. The effect was dose-dependent, as was the concentration of Aβ peptides detected in plasma. The scientists detected complexes of Aβ bound to antibody in all plasma samples but only in cerebrospinal fluid of mice treated with the highest doses of m266.

The Dodart et al. paper appears to contradict previous research on active Aβ immunization studies, which tended to report that behavioral improvement occurred along with a reduced amyloid burden. The study does not settle the question whether these antibodies work from the periphery or by entering the brain.

This study does, however, bolster a line of evidence arguing that soluble Aβ may be as neurotoxic as plaques, though in a different way. For example, one study found that soluble Aβ levels, measured in postmortem brains, could distinguish between people who had plaque pathology but no dementia and those who also had clinical dementia (Lue et al. 1999). Another study found that soluble Aβ levels correlate better with disease severity than do plaque levels (McLean et al, 1999), and a recent paper implicated soluble Aβ, but not plaques, in spatial learning deficits in another AβPP-transgenic mouse strain (Koistinaho et al. 2001.) A paper in the current issue of Nature reports that soluble Aβ oligomers interfere with long-term potentiation in rats (see ARF news item) and another one recently reported similar data in rat hippocampal slices (Wang et al. 2002). Finally, Bill Klein and others at Northwestern University Medical School reported Aβ oligomers, or ADDLs, cause neurological dysfunction in the hippocampus long before neurons degenerate (Lambert et al. 1998)

It is fascinating to wonder how, mechanistically, such marked improvement in memory can occur overnight. Moreover, it remains to be seen if any of this effect-seen in mice whose abnormal production of human AβPP is not generally accompanied by massive neuronal death-can be replicated in human patients, who do lose significant populations of hippocampal and entorhinal cortex neurons before they notice the first symptoms. As famed angiogenesis researcher Judah Folkman likes to quip: "If you are a mouse, we can take good care of you."—Gabrielle Strobel


  • Q & A with Steven Paul-Posted 9 April 2002

    Q: Memory improvement after one day at first blush seems paradoxical when one thinks of AD as a slow, degenerative process that eventually kills off populations of neurons. How does it work?
    A: This transgenic mouse model does not develop a lot of neurodegeneration. The difference between the mouse model and human AD is that while the mouse develops abundant neuritic plaques, we do not see frank loss of neurons, certainly not the massive neurodegeneration that occurs in late-stage human AD. The soluble forms of Aβ are toxic to some forms of memory; we only tested two learning/memory tasks. The antibody can somehow acutely block those toxic effects on whatever biological processes underlie those forms of memory.

    Last week, Walsh et al. showed that small oligomers of Aβ were particularly effective in blocking LTP, suggesting again, much like our data, that these small oligomers may themselves disrupt a molecular substrate for memory formation (see related ARF news items). This antibody also prevents the deposition of amyloid plaques (see related ARF news items).

    Q: So there must be a separate mechanism from what is generally thought, namely that Aβ ultimately induces cell death by forming fibrils and aggregating?
    A: It would be different, yes. That highlights the complexity of how this little peptide might contribute to the pathogenesis of AD. I suspect that if this paper were to have any relevance to human AD, it would be very early in the disease-perhaps when people are having mild cognitive impairment-that the peptide is disrupting memory, quite apart from its effect of depositing aggregate-forming fibrils and killing neurons.

    Q: How long does the one-shot improvement last? Did you test again after a week, or a month?
    A: We have not tested that yet. I expect it may last as long as the antibody is interacting with the peptide, and that probably relates to the half-life of the antibody in the plasma or brain, which is a few days.

    Q: Dominic Walsh and Bill Klein have antibodies for non-fibrillar species of Aβ. How about testing if their levels correlate with cognitive function in these PDAPP mice?
    A: We have not done so, that would be a worthwhile experiment.

    Q: What exactly does m266 bind? Monomers, oligomers, protofibrils?
    A: We know it binds to monomeric Aβ but don't know what other forms it may recognize. It could bind to other low-n oligomers, as well. We do not seem to be able to decorate plaques with this antibody in vivo, so it does not bind well to more complex, aggregated forms of Aβ.

    Q: Some of the results of other vaccination studies were contradictory because they varied with the genetic background of the strains used. Is yours a PDAPP-specific effect, and have you tried it on other strains?
    A: We have not tried it on other strains yet. How well this can be extrapolated to other mouse strains is not clear at this point. How it will translate to humans is even less clear.

    Q: With regard to treating humans, eventually, how much of a risk with passive immunization is serum sickness, and does this infrequent vaccination scheme encourage you because it raises the specter that AD could be treated more like an acute, not a chronic, disease.
    A: I do not think serum sickness is not going to be a huge problem at the doses given in this disease. You would not be giving huge doses like we do against virus infections. We have been giving humanized antibodies safely now for a variety of conditions. In light of the Elan data I think the question becomes, can an antibody to the peptide be safe without producing an "inflammatory" response? We do not know the answer to that, but with a well-defined antibody recognizing a well-defined epitope, you improve your chance of avoiding cellular immunity, which is what happens when you actively immunize with peptide.

    Q: A T cell response is said to have occurred in the Elan trial.
    A: Right. Whether that response caused the problems I do not know.

    Q: Finally, papers are always a step behind the real research front. Where is this in terms of future clinical trials?
    A: We are continuing to explore it as a possible avenue. But there are still so many complexities, including making sure we have the right antibody that would be safe and effective in the animal model. All the preclinical toxicology work still needs to be done, so we are a ways off from being able to test this in humans. Whether human trials will occur will depend on the results of a whole series of tests that still need to be done. But I think the general strategy of immunotherapy is still interesting.

    Q: Will this paper give a boost to those who believe soluble Aβ deserves more attention?
    A: Let me emphasize that we are in no way saying that deposited forms of Aβ that are fibrillized into amyloid plaques are harmless. We are quite certain that that will disrupt memory, too. We are just saying that the soluble form can be damaging, as well. We like the fact that this antibody, when administered chronically, also reduces deposited Aβ as well.-Steven Paul, Lilly Research Laboratories, Indianapolis.


  1. "The findings by Dodart and colleagues are very interesting. However, as Steven Paul points out in the Q and A session on this website, it remains to be seen if a similar effect will be observed in other transgenic AβPP mouse models and eventually in AD patients. Reversal of behavioral deficits was not associated with reduction in amyloid plaque burden or alterations in levels of total brain Aβ, but was significant at doses that allowed detection of Aβ-antibody complexes in the CSF. However, levels of soluble brain Aβ or the presence of antibodies bound to plaques were not measured.

    Increase or Decrease of Soluble Aβ?
    The authors speculate that the behavioral improvements may be caused by efflux of soluble Aβ out of the brain. This may be true, but the reversal of memory deficits may as well be caused by a rapid increase in soluble Aβ within the CNS, derived from plaque Aβ. However, this acute increase may not be sufficient to significantly reduce plaque burden. This alternative explanation should come as no surprise as numerous laboratories have shown low levels of Aβ to have neurotrophic effects in cell culture, which may translate into a beneficial neuromodulatory effect in vivo. Follow-up studies measuring soluble Aβ within the brain should clarify this issue.

    It is certainly difficult to compare behavioral studies in various immunized transgenic AβPP models of different background strains. And yet, a treatment induced-increase in brain soluble Aβ may actually explain similarities within the reported behavioral studies, whereas Dodart et al. mention that their data seem to contrast with previous findings (Janus et al., 2000; Morgan et al., 2000). Janus et al. observed behavioral improvement associated with reduction in plaques but no change in total brain Aβ. Although they suggested that cognitive improvement might be due to reductions in potentially toxic protofibrils, it may also have been caused by an increase in soluble brain Aβ levels. Morgan et al. observed a partial reversal of cognitive deficits in AβPP/PS1 mice, though cerebral amyloid burden as measured by immunohistochemistry was not significantly reduced. The authors suggested that a decrease in soluble Aβ might explain the cognitive improvement in the immunized mice, but this potential connection was not measured in their study.

    Our results following 7 months of treatment suggest that the reduction in soluble brain Aβ is less than that of plaque Aβ (see related ARF news items), but we did not analyze the behavior of the mice. All the behavioral studies may fit nicely together if the cognitive improvements were caused by an increase in soluble Aβ within the CNS. This view may seem to contradict findings suggesting toxicity of soluble Aβ species (see news story), but we emphasize that the in-vivo ratio of Aβ monomers/oligomers and protofibrils/fibrils in transgenic mice and AD brain has not been thoroughly established, and any future biochemical findings will likely vary depending on the methods used. Furthermore, Aβ oligomers found in these mice may be less stable than those detected in AD brains (Kalback et al., 2002), and their toxicity has not been demonstrated. It is likely that most of the soluble Aβ in mouse brain are monomers, which may be trophic instead of toxic.

    Previously, these authors showed that peripheral injection of anti-Aβ antibody decreased brain amyloid burden without binding to amyloid plaques, but the presence of Aβ-antibody complexes was not measured in the CSF (see related ARF news items) While both these studies show an extensive increase in plasma Ab levels, they cannot be easily compared because of differences in parameters measured. The present results need to be replicated in other AβPP transgenic models. They also need to be correlated with levels of soluble brain Aβ, as well as amounts of various AβPP fragments that may affect behavior. In the PDAPP model, Aβ is predominantly generated within the CNS, whereas in the Tg2576 model Aβ is formed in various peripheral organs as well. It is, therefore, likely that any sequestration of Aβ from the CNS to the periphery will be greater in the PDAPP model, and a higher dose may be needed to achieve a similar effect in Tg2576 mice.

    The predictive value of the transgenic AβPP mouse models are questionable also because the plaques in the AβPP23 and the Tg2576 mouse models are much more soluble than those in AD, though vascular amyloid is as insoluble in these mice as in AD (Kuo et al., 2001; Kalback et al., 2002). This finding suggests that the mouse plaques may be more easily removed than those in AD. It may be explained by transgenic mice having fewer posttranslational modifications within the Aβ peptides, as well as by differences in the composition of amyloid-associated proteins. However, a contradicting finding in the Tg2576 model has been reported (Kawarabayashi et al., 2001), in which the portion of Aβ requiring formic acid for extraction showed an age-related increase and eventually reached similar levels as seen in AD brain. This controversy regarding the ratio of soluble versus insoluble Aβ in Tg mice needs to be resolved, and a similar study should be performed in the PDAPP model. Future therapy studies should also attempt to determine the ratio of various Aβ species.

    Overall, therapeutic findings in these mouse models must be interpreted cautiously, as they may not apply to the human condition. It should be noted, however, that although amyloid burden in AD patients and AβPP transgenic mice has been reported to be similar, human plasma Aβ levels are several-fold lower than those observed in the Tg2576 mice. Therefore, a smaller amount of antibodies per weight may be needed in humans than in transgenic mice to achieve similar therapeutic results, although we should not expect that antibodies will lead to clearance of existing plaques in AD brains, which postmortem require formic acid for solubilization."


    . A beta peptide immunization reduces behavioural impairment and plaques in a model of Alzheimer's disease. Nature. 2000 Dec 21-28;408(6815):979-82. PubMed.

    . A beta peptide vaccination prevents memory loss in an animal model of Alzheimer's disease. Nature. 2000 Dec 21-28;408(6815):982-5. PubMed.

    . APP transgenic mice Tg2576 accumulate Abeta peptides that are distinct from the chemically modified and insoluble peptides deposited in Alzheimer's disease senile plaques. Biochemistry. 2002 Jan 22;41(3):922-8. PubMed.

    . Comparative analysis of amyloid-beta chemical structure and amyloid plaque morphology of transgenic mouse and Alzheimer's disease brains. J Biol Chem. 2001 Apr 20;276(16):12991-8. PubMed.

    . Age-dependent changes in brain, CSF, and plasma amyloid (beta) protein in the Tg2576 transgenic mouse model of Alzheimer's disease. J Neurosci. 2001 Jan 15;21(2):372-81. PubMed.

  2. The remarkable finding described by Dodart et al. in Nature Neuroscience adds an important page to the evolving story of Aβ toxicity in Alzheimer's disease. It builds on two related discoveries. First, thanks to the pioneering work of Dale Schenk and colleagues (Schenk et al, 1999), we have known for three years that active and passive vaccination can have a major impact on brain chemistry, a terrifically surprising and important discovery. Schenk's original findings showed that vaccination with fibril-enriched preparations of Aβ could significantly lower amyloid plaques in transgenic mice models for AD. Second, since the work of Lambert et al., 1998, we've also known that small oligomers of Aβ, soluble and globular in structure, have potent CNS effects. The disruptive activity of oligomers (aka "ADDLs") is likely to account for the imperfect correlation between dementia and plaque burden in Alzheimer's disease.

    Particularly relevant to the study by Dodart et al, ADDLs rapidly inhibit LTP, a major paradigm for synaptic memory formation (Lambert, ibid, see also more recent works by Wang et al., 2002, and Walsh et al, 2002, (see related ARF news items). The fast and selective nature of LTP inhibition indicates that it is not a consequence of neuronal degeneration. Because these synaptic effects appear dysfunctional rather than degenerative, we have proposed that memory loss in AD (and in mild cognitive impairment) could actually be reversed, not just slowed down (see, e.g., Klein et al., 2001). As seen in Dodart et al., the rapid cognitive reversal in antibody-treated transgenic mice provides strong support for this hypothesis.

    Dodart's findings underscore the potential value in developing vaccines that target soluble toxins. The usefulness of such vaccines was suggested by earlier work from Morgan et al., 2000, who found that vaccination improved memory performance in transgenic mice whether or not plaques were eliminated. We might expect that ideal therapeutic antibodies would be able to discriminate between toxic oligomers and physiological monomers. Toxicity-neutralizing antibodies with such specificity recently have been generated by vaccination of rabbits with ADDLs (Lambert et al., 2001). Even greater specificity ultimately may be crucial. For example, vaccines that target soluble toxins but avoid insoluble amyloid deposits may circumvent the CNS inflammation recently uncovered in human clinical trials (see vaccination live chat).

    In a wider venue, recent evidence by Bucciantini et al. indicates that other protein misfolding diseases, previously associated with amyloid deposition, also may have pathogenic sub-fibrillar species (see related ARF news items). It is possible, therefore, that approaches developed for targeting such species in AD ultimately may have broad application.


    . Immunization with amyloid-beta attenuates Alzheimer-disease-like pathology in the PDAPP mouse. Nature. 1999 Jul 8;400(6740):173-7. PubMed.

    . Diffusible, nonfibrillar ligands derived from Abeta1-42 are potent central nervous system neurotoxins. Proc Natl Acad Sci U S A. 1998 May 26;95(11):6448-53. PubMed.

    . Soluble oligomers of beta amyloid (1-42) inhibit long-term potentiation but not long-term depression in rat dentate gyrus. Brain Res. 2002 Jan 11;924(2):133-40. PubMed.

    . Targeting small Abeta oligomers: the solution to an Alzheimer's disease conundrum?. Trends Neurosci. 2001 Apr;24(4):219-24. PubMed.

    . A beta peptide vaccination prevents memory loss in an animal model of Alzheimer's disease. Nature. 2000 Dec 21-28;408(6815):982-5. PubMed.

    . Vaccination with soluble Abeta oligomers generates toxicity-neutralizing antibodies. J Neurochem. 2001 Nov;79(3):595-605. PubMed.

  3. This paper makes important advances in the immunization field on two fronts. First, it follows up on previous papers, such as that out of Dave Morgan's work, suggesting that the vaccine can reduce the cognitive deficit associated with the accumulation of Aβ, without removing the neuritic plaques. The reason is presumably that soluble Aβ oligomers and proto-fibrils are being reduced. Examples of such protofibrils are given in work on ADDLs by the work or Bill Klein and Grant Krafft. The current work confirms the observation that cognitive deficits can be improved without eliminating neuritic plaques, and focuses attention on the toxicity associated with soluble Aβ.

    A second reason that this work might be particularly important is because it furthers prior work by Steve Paul's group showing that administration of peripheral anti-Aβ can reduce plaque load, and now the cognitive deficit associated with plaque load. Given potential difficulties with the active immunization model, this passive immunization model takes on increased importance (although there is a possibility that the problems with the human trials of the Aβ vaccine were not directly due to the Abeta antibodies). One can imagine that humanized anti-Aβ antibodies might be useful clinically.

    View all comments by Benjamin Wolozin

Make a Comment

To make a comment you must login or register.


News Citations

  1. Early Diagnosis of Alzheimer's—Making Use of the Blood-Brain Barrier
  2. Earliest Amyloid Aggregates Fingered As Culprits, Disrupt Synapse Function in Rats

External Citations

  1. Dodart et al, 1999
  2. Lue et al. 1999
  3. McLean et al, 1999
  4. Koistinaho et al. 2001
  5. Wang et al. 2002
  6. Lambert et al. 1998

Further Reading

Primary Papers

  1. . Immunization reverses memory deficits without reducing brain Abeta burden in Alzheimer's disease model. Nat Neurosci. 2002 May;5(5):452-7. PubMed.