Amid ongoing attempts to enlist the immune system to fight Alzheimer disease, some scientists have wondered about our innate ability to clear this ornery peptide. In this week’s PNAS Early Edition, a team led by Tony Wyss-Coray at Stanford University, Palo Alto, California, reports that healthy people have natural antibodies against a broad array of toxic Aβ species and other amyloidogenic peptides. The antibodies can protect cultured neurons from Aβ toxicity. Though the physiological relevance is not yet clear, and some antibodies seem to decrease with age and advancing AD, the findings may have ramifications for AD immunotherapy. They raise the possibility that antibodies to certain Aβ configurations might correlate with better protection against disease progression, and that effective, possibly safer, alternatives to current immunotherapy approaches can be found.

Previous studies on Aβ autoantibodies have only muddied the issue of their significance to AD. Using various methods to test blood and cerebrospinal fluid samples, different groups have reported that Aβ antibody titers increase (Nath et al., 2003), decrease (see, e.g., Moir et al., 2005 and ARF related news story), or hardly budge (Hyman et al., 2001) in AD patients relative to healthy people. Wyss-Coray’s team set out to do a more comprehensive analysis looking at sundry forms of Aβ in greater numbers of patients—young and old, healthy or with varying AD severity. “In my opinion, this is the definitive paper because their sample sizes are so substantial, and their methods of measuring the antibodies are as expansive as could be done at this point in time,” said Dave Morgan of the University of South Florida, Tampa, in an interview with ARF.

Using peptide microarrays containing numerous modified and mutated Aβ species, as well as control and other amyloidogenic peptides, first author Markus Britschgi and colleagues analyzed plasma antibodies from 36 cognitively normal people and 75 AD patients. In this screen, and in an independent set of 117 samples (62 normal, 55 AD), Aβ autoantibodies specific to oligomeric, fibrillar and other higher-order Aβ forms were about 12 times as prevalent as antibodies to monomeric Aβ. “This is a very important paper showing that anti-oligomer antibodies naturally occur,” said Michael Agadjanyan of the Institute for Molecular Medicine in Huntington Beach, California.

Many of the study participants had antibodies against pyroglutamate forms of Aβ, which are believed to seed oligomerization (see ARF related news story). Unexpectedly, many individuals had antibodies that recognize mutant forms of Aβ, including Dutch (E22Q), Flemish (A21G), and Arctic (E22G) Aβ1-40 peptides that are only found in familial AD cases. This supports the idea that some antibodies to Aβ, such those raised by Charlie Glabe’s lab at the University of California, Irvine, recognize common three-dimensional structures rather than individual primary peptide sequences (see ARF related news story). Some plasma samples from non-demented individuals also contained antibodies against foreign peptides that are unique to other familial dementias. “People have antibodies against oligomers of the ABri and ADan peptides, for example, but people don’t make these peptide unless they have British or Danish dementia,” Wyss-Coray told ARF. “This is consistent with the concept that cross-reactive oligomer-specific antibodies exist naturally.” As further confirmation of this phenomenon, his team immunized vervet monkeys with full-length Aβ and found that this not only boosted production of anti-Aβ antibodies but also expanded their repertoire of antibodies that cross-react with mutant Aβ and other toxic amyloidogenic peptides.

These findings jibe with more recent work from Glabe’s group, which demonstrated that immune serum from Aβ-immunized rabbits could also recognize islet amyloid in transgenic mouse models of diabetes (Kayed et al., 2007). Glabe’s team went on to show that vaccinating Tg2576 AD mice with different amyloid oligomers improved cognition regardless of the peptide antigen sequence. Similarly, others have reported that mice inoculated with potato virus Y, which has a sequence closely matching the N-terminal region of Aβ, produced antibodies that bind plaques and neurofibrillary tangles in AD brain tissue (Friedland et al., 2008). “The implications of these observations is that it may be possible to develop an effective human vaccine using random peptide sequences that form amyloid oligomers and that this vaccine would avoid the autoinflammatory side effects that doomed earlier vaccines,” Glabe wrote in an email to ARF (see full comment below).

For the time being, Wyss-Coray’s new data leave open a fundamental issue—whether natural Aβ autoantibodies have biological significance. “What would be most satisfying, I guess, is if you found that normal people had really high titers of these antibodies and AD patients were invariably in the lower quartile,” Morgan said. Instead, the study found no difference between plasma antibodies from AD and normal populations. Within the AD group, though, people with moderate to severe AD had lower immunity to oligomeric Aβ1-42 than did patients with mild disease. And in an independent cohort of healthy women, some antibodies from older people (70 or above) had reduced reactivity compared to those from younger (21-44 years) age groups.

To further explore the utility of these natural antibodies, Wyss-Coray and colleagues used their peptide arrays to analyze CSF samples from healthy and mild AD patients. These experiments showed that autoantibodies detected in the periphery were in fact reaching the central nervous system. Antibody reactivity was much lower in CSF than in plasma but had a similar range of specificities to oligomeric and pyroglutamate-modified Aβ forms and to ABri, ADan, and mutant Aβ. In a separate set of analyses, the scientists purified plasma IgGs from three study participants and showed that each prep protected against Aβ toxicity when added to cultured mouse hippocampal neurons. Scientists have seen similar neuroprotection in studies with commercially available intravenous immunoglobulin (IVIg) (Szabo et al., 2008), a mixture of pooled human plasma antibodies that is being tested in a Phase 2 trial of mild to moderate AD (see ARF related conference story).

Based on studies in the current paper, the authors estimate that these natural antibodies exist in human plasma at concentrations around five micrograms per kg body weight. “These concentrations are around 100-1000 times below doses used for passive immunotherapy, but could nevertheless have an impact over a lifetime, or a few years,” Wyss-Coray said, noting a recent review (Lutz et al., 2009) suggesting that low-titer, low-affinity autoantibodies can show functional potency. Ultimately, he said, “it all depends on whether the current trials show any clinical benefit. If the immunotherapy doesn’t work, it’s a moot point to discuss the role of natural autoantibodies.”

To Morgan, the existing data already make a strong case that Aβ titers in unvaccinated human plasma are too low to offer therapeutic benefit. “We know they’re not 100 percent protective because there were people with AD who did have high levels of these antibodies,” he told ARF. He also mentioned several parts of the current study—namely, the primate immunization data, and the antibody reactivities compared with those that recognize common human pathogens—that suggest that natural antibody titers are much lower than the levels normally needed for therapeutic efficacy (see full comment below). Nevertheless, he said “the authors’ conclusion that these could play a role in disease onset and progression is certainly one that merits further analysis and testing.”

Toward this end, Wyss-Corray said his team would love to analyze samples from an active immunization trial to get a closer look at the antibody repertoires of participants who fared better clinically. “Maybe they make antibodies against a specific conformation or post-translational modification of Aβ, and that’s why they had clinical improvement,” he told ARF.—Esther Landhuis


Make a Comment

To make a comment you must login or register.

Comments on News and Primary Papers

  1. This is certainly a fascinating observation that healthy young individuals have relatively high concentrations of these circulating antibodies and that their levels seem to decline with age and in AD. It is also important that immunization against Aβ causes a dramatic increase in the levels of these antibodies that recognize not only Aβ, but oligomeric states of unrelated amyloids like ABri and ADan. Since the immune response to amyloids is largely directed against patterns of amino acid side chains on the surfaces of the β-sheets that result from the simple, parallel, and in-register structure of amyloids, it is not surprising that that the antibodies would react with many amyloids because these patterns would be expected to be the same on any amyloid that has the same amino acid. It may be a little speculative, but the reason that these types of antibodies seem to be favored is that these structures are predominantly if not exclusively pathological and do not occur on globular proteins. The immune response to amyloid aggregates appears to be part of the T cell independent innate immunity where B cells are already primed to respond to these antigens. This is like the innate immune response to things like bacterial cell walls and it seems to indicate that evolution has selected these antibodies because they are commonly encountered and the body needs to respond immediately to these threats.

    The results are consistent with some of our recent findings. Since the antibodies recognize generic, common epitopes, the converse prediction is that it doesn't really matter what peptide sequence you use to form amyloid: you will get the same immune response. We showed this is the case for Aβ and islet amyloid polypeptide (see Kayed et al., 2007). We recently extended this to even random peptide sequences that have no more than a four-amino-acid match in the entire human genome (Rasool and Glabe, 2008) and we showed that vaccination with this random sequence or IAPP gives the same therapeutic benefit in transgenic mice (Tg2576). We have also looked at passive immunization with conformation-dependent, oligomer-specific monoclonals and found that they improve cognition and prevent amyloid pathology in Tg2576 and 3XTg-Ag mice.

    The implications of these observations are that it may be possible to develop an effective human vaccine using random peptide sequences that form amyloid oligomers and that this vaccine would avoid the autoinflammatory side effects that doomed earlier vaccines. Even if there is a cytotoxic T cell response to the linear peptide, there is no protein in the human genome that displays this epitope for the T cells to attack, so vaccination against this antigen should have no more consequences than diphtheria toxin or pertussis, against which we vaccinate infants.

    Disclosure: The author has a financial interest in a company that is trying to develop oligomer-specific, conformation dependent antibodies that recognize generic, sequence independent epitopes on the surfaces of amyloid oligomers. This company is Kinexis Inc. of Carlsbad, CA.

    View all comments by Charles Glabe
  2. This is state-of-the-art technology measuring sizable cohorts of human samples and fairly conclusively answers a question that was controversial in the literature regarding the levels of anti-Aβ antibodies in AD versus control samples.

    My take-home messages from the paper are the following:

    1. Titers against Aβ in unvaccinated human plasma are low. The dilution studies show titers of 1:50 to 1:200. Figure 3A shows that the antibody reactivity to Pneumococcal vaccine and Candida albicans are 100-1,000-fold greater than any of the anti-Aβ reactivities. The studies in nonhuman primates show that efficacious vaccines cause several hundredfold increases over non-immunized reactivities (Fig. 3E). Hock et al., although using a considerably different method, show titers greater that 1:10,000 in some vaccinated AD patients (see Hock et al., 2002). Hence, although these antibody titers may be "abundant," they are lower than the levels normally needed to be therapeutic.

    2. Figure 2D shows that roughly 75 percent of the antibodies against Aβ are masked by bound antigen. The declines seen with age in the plasma samples are diminished after unmasking the antibodies (dissociating bound antigen), implying the aging and AD effects observed may be secondary to increased amounts of antigen binding a greater fraction of the available antibodies in addition to, or instead of, reduced antibody levels.

    3. The neuroprotection data are intriguing, but it should be noted this activity requires purification of IgG and concentration of the IgG relative to that found in plasma. It sees unlikely the titers in plasma are sufficient to provide neuroprotection in vitro without this concentration.

    4. One critical missing piece of information regards the apparent affinity of the anti-Aβ antibodies for the antigens. When we did our work on dissociation of antigen-antibody complexes, we discovered a low-affinity/high-capacity binding artifact associated with antibodies that bound to the highly hydrophobic domains of Aβ with affinities (IC50s, in reality) greater than 10 μM (measured by competition with free Aβ), while the anti-Aβ antibodies generated by vaccination had IC50s of 1 μM for free Aβ, or less. Although the only definitive manner of addressing affinity is specific plasmon resonance, these competition studies can still reveal relative affinity. If these antibody activities in human plasma, which react more with the more hydrophobic variants of Aβ, are not effectively competed by preincubation with10-100 μM of the Aβ preparations used in the array, it seems unlikely these would have much impact on brain in vivo.

    In conclusion, this manuscript applies contemporary methods to a vexing problem regarding endogenous antibody titers against the Aβ peptide in humans. The general conclusions are that these titers are relatively low and considerably below those levels found therapeutic in mice, monkeys, and, apparently, humans. The authors conclude, and I concur, that these anti-Aβ activities are most likely derived by cross-reaction from antibodies directed against other antigens to which humans are commonly exposed. In any event, the question becomes, Does a lifetime of exposure to low levels provide some protection? My general impression is that it seems unlikely, but the authors’ conclusion that these could play a role in disease onset and progression is certainly one that merits further analysis and testing.

    View all comments by Dave Morgan
  3. The Britshgi et al. paper nicely confirms our 2005 study (Moir et al., 2005) in which we measured titers of autoantibodies to "low molecular weight oligomeric cross-linked amyloid protein species (CAPS)," which we have long considered to be the neurotoxic form of Aβ in AD. We analyzed titers in AD and non-demented control plasma. While plasma of both non-demented and AD patients were found to contain autoantibodies specific for soluble CAPS, anti-CAPS antibodies in AD plasma were found to be significantly reduced compared with non-demented controls (p = 0.018). Furthermore, age at onset for AD correlated significantly (p = 0.041) with plasma immunoreactivity to CAPS. Based on these data, we suggested that autoantibodies to CAPS are depleted in AD patients and raised the prospect that immunization with anti-CAPS antibodies might provide therapeutic benefit for AD. The new study elegantly confirms and extends our earlier findings. Collectively, these findings reinforce the hypothesis that naturally occurring protective autoantibodies to toxic forms of Aβ exist in plasma and are depleted in AD.

    View all comments by Rudy Tanzi


News Citations

  1. Molecular Economics of AD—Supply, Demand, and the Aβ Glut
  2. Keystone Drug News: Pyroglu Aβ—Snowball That Touches Off Avalanche?
  3. Amyloid Oligomer Antibody—One Size Fits All?
  4. Chicago: More Phase 2 News—PBT2 and IVIg

Paper Citations

  1. . Autoantibodies to amyloid beta-peptide (Abeta) are increased in Alzheimer's disease patients and Abeta antibodies can enhance Abeta neurotoxicity: implications for disease pathogenesis and vaccine development. Neuromolecular Med. 2003;3(1):29-39. PubMed.
  2. . Autoantibodies to redox-modified oligomeric Abeta are attenuated in the plasma of Alzheimer's disease patients. J Biol Chem. 2005 Apr 29;280(17):17458-63. PubMed.
  3. . Autoantibodies to amyloid-beta and Alzheimer's disease. Ann Neurol. 2001 Jun;49(6):808-10. PubMed.
  4. . Fibril specific, conformation dependent antibodies recognize a generic epitope common to amyloid fibrils and fibrillar oligomers that is absent in prefibrillar oligomers. Mol Neurodegener. 2007;2:18. PubMed.
  5. . Antibodies to potato virus Y bind the amyloid beta peptide: immunohistochemical and NMR studies. J Biol Chem. 2008 Aug 15;283(33):22550-6. PubMed.
  6. . Natural human antibodies to amyloid beta peptide. Autoimmun Rev. 2008 Jun;7(6):415-20. PubMed.
  7. . Naturally occurring auto-antibodies in homeostasis and disease. Trends Immunol. 2009 Jan;30(1):43-51. PubMed.

External Citations

  1. Phase 2 trial

Further Reading


  1. . Autoantibodies to amyloid beta-peptide (Abeta) are increased in Alzheimer's disease patients and Abeta antibodies can enhance Abeta neurotoxicity: implications for disease pathogenesis and vaccine development. Neuromolecular Med. 2003;3(1):29-39. PubMed.
  2. . Autoantibodies to redox-modified oligomeric Abeta are attenuated in the plasma of Alzheimer's disease patients. J Biol Chem. 2005 Apr 29;280(17):17458-63. PubMed.
  3. . Autoantibodies to amyloid-beta and Alzheimer's disease. Ann Neurol. 2001 Jun;49(6):808-10. PubMed.

Primary Papers

  1. . Neuroprotective natural antibodies to assemblies of amyloidogenic peptides decrease with normal aging and advancing Alzheimer's disease. Proc Natl Acad Sci U S A. 2009 Jul 21;106(29):12145-50. PubMed.