It met with little enthusiasm when originally proposed, and took more than three years to perfect, but a new strategy for detecting serum antibodies could have important ramifications for Alzheimer’s disease diagnostics. Like many groundbreaking ideas, Thomas Kodadek’s was profoundly simple—use a huge library of synthetic antigens to capture circulating antibodies that are specific to disease states. “Nobody thought this would work,” said Kodadek, who is at The Scripps Research Institute, Jupiter, Florida. But as outlined in the January 7 Cell, Kodadek and colleagues have now validated their methodology using a mouse model of multiple sclerosis, and have gone on to demonstrate that it can identify antibodies found only in people with Alzheimer’s. “This has the potential to be a big advance in the AD biomarker field,” noted David Holtzman, Washington University, St. Louis, Missouri, in an e-mail to ARF (see full comment below). “It may also be very useful for biomarker identification for other neurodegenerative and neurological disorders.”

Fishing for disease-specific antibodies (or antigens) is not a new idea, but as Kodadek and colleagues explain, unmodified native antigens are unlikely to work as bait, because those antigens are also found in non-disease states. It is far more likely that disease-specific antibodies recognize unusually modified natural antigens. But if you don’t know what those antigens are, then how can you find the antibody?

Not such a Catch 22 as it seems. First author Muralidhar Reddy and colleagues used synthetic molecules to make a “shape library,” and then used that to capture antibodies in an unbiased fashion. “There has been a mindset that antibodies are only going to bind to their native antigens,” said Kodadek. But as the researchers demonstrated, that is not the case.

Reddy and colleagues used an array comprising thousands of peptoids—octamers of glycine with side groups added by way of combinatorial chemistry. To ensure that none of these antigens are normally found in nature, they added the side groups to the glycine nitrogen instead of the α carbon, and they ensured most of the side chains did not resemble any of the 20 natural amino acids. In this way, they generated an array of molecules that are unlikely to be exact matches for native antigens. They used the array to capture antibodies, which they detected using a secondary anti-IgG antibody with a fluorescent tag.

Could this array pull out disease-specific antibodies? “We started with the simplest disease model imaginable,” said Kodadek. The researchers tested sera from mice with experimental autoimmune encephalomyelitis (EAE). In this model of multiple sclerosis, mice are immunized with Freund’s complete adjuvant and a peptide derived from myelin oligodendrocyte glycoprotein (Mog). The researchers collected antiserum 38 days later and ran it over an array of more than 4,600 peptoids. Unsurprisingly, serum from the Mog/adjuvant-injected mice contained antibodies that bound to Mog peptide that the scientists had spotted on the array. More importantly, the serum also contained antibodies that bound to three of the peptoids. Those peptoids did not capture antibodies from mice treated only with adjuvant. The same three peptoids also captured antibodies from a second set of EAE mice, giving the test some validity. The peptoids also had some antibody specificity, as they failed to recognize antisera from mice that were immunized with ovalbumin. They gave a hint of how severe disease was, because they captured progressively more antibody up to three weeks after immunization.

But what about disease not induced by an active immunization? This is when the researchers turned to serum from people with Alzheimer’s disease. From an array of more than 15,000 spots, three different peptoids (ADP1 to ADP3) detected higher IgG amounts in serum of six AD patients compared to serum from six age-matched controls. Kristin Martin-Cook at the University of Texas Southwestern Medical Center in Dallas provided the samples. In a subsequent validation test, all three peptoids again detected high IgG levels in sera from 16 separate patients. Sera from 14 of 16 controls did not recognize any of the three peptoids, but sera from two other controls did. These could be false positives, or, as the authors write, these volunteers could have preclinical disease. Because all three peptoids behave similarly, the authors believe that the latter explanation is most parsimonious.

Kodadek told ARF that he is now testing patient samples that were taken longitudinally to see if he can track disease progression by comparing antibody capture to the development of symptoms. Like the Mog peptoids, ADP1 to 3 also seem to be relatively specific, because they did not detect antibodies in sera taken from six Parkinson’s disease patients or six people with systemic lupus erythematosus.

Some news outlets are already touting these findings as a “blood test” for AD, but as Tony Wyss-Coray at Stanford University, California told ARF via e-mail, some of these media reports are over the top. “This is certainly not a blood test at this time,” he told ARF. Kodadek and colleagues agree, and are continuing validation studies. A Florida-based diagnostic company, Opko Health Laboratories, where Reddy and several of the coauthors work, has licensed the technology. On Monday, Opko and Bristol-Myers Squibb signed an agreement to pursue AD diagnostics (see company press release). Kodakek, who holds stock options in Opko, told ARF that in second-generation screening, the scientists discovered new antibodies. “The preliminary results look extremely promising,” he said.

One burning question that remains is, What constitutes the natural antigen for these antibodies? Kodadek said that he has not started looking into that in earnest. But in principle, if he can capture enough of the antibody, then he could use it to pull down antigens from AD brain extracts and identify the antigens by mass spectroscopy. As Wyss-Coray writes (see full comment below), there is no consistent evidence to suggest that antibodies specific to AD (or other neurodegenerative diseases) exist, though “dozens of isolated studies in the literature have described antibodies with distinct antigen specificity to be overrepresented in AD compared with healthy controls.” He did note that the small number of peptoids that bind AD sera in this study could indicate that AD patients possess a highly specific immune response against an unknown antigen. “If true, that would most certainly change the current view of this disease,” he wrote.—Tom Fagan


  1. I think this new unbiased technology of looking for IgGs that bind to unique shapes on synthetic peptoids provides a valuable new methodology for discovery of IgGs that may be specific for certain disease states. The authors provide a good proof-of-concept experiment in experimental autoimmune encephalomyelitis (mouse model of multiple sclerosis), and also show that, in a small set of Alzheimer’s disease cases and controls, this approach can identify unique IgGs that are present in AD. They validated their initial finding from N = 6 AD cases and N = 6 controls and in an additional group of N = 16 AD, N = 16 controls and N = 6 lupus patients with sensitivity and specificity greater than 90 percent. While the numbers are small, the data clearly show the promise of this technique in the potential development of a serum biomarker, but also in identifying the unique antigen that these antibodies bind to. If this can be done on a much larger number of samples in which a lot more biomarker information is known (CSF, amyloid imaging, etc.), this has the potential to be a big advance in the AD biomarker field. It may also be very useful for biomarker identification for other neurodegenerative and neurological disorders.

    View all comments by David Holtzman
  2. This report by Kodadek’s group describes a novel approach in the search for disease-specific molecular biomarkers. The authors are screening patient sera to find specific endogenous immunoglobulins that would be characteristic of a given disease. The rationale for the existence of such antibodies is based on the hypothesis that the patient mounts an immune response against disease-specific molecular species (or pathogens) with subsequent production of specific antibodies. There is no consistent evidence that this is the case for Alzheimer’s disease (AD) or other neurodegenerative diseases, but dozens of isolated studies in the literature have described antibodies with distinct antigen specificity that are overrepresented in AD compared with healthy controls.

    The current study takes an unbiased approach to identify novel species of IgG immunoglobulins using a library of 15,000 peptoids, peptide-like small polymers that represent diverse molecular shapes but do not necessarily model native biological molecules. Using this approach, the team isolated three peptoids that captured IgGs in serum from AD patients with at least threefold higher binding activity than in serum from matched controls. This is promising and provides a potential new way to find subtle changes in the immune repertoire of AD patients. Similar to their proof-of-concept work in a mouse model of autoimmune neuroinflammation, it may be possible to identify the nature of the antigen recognized by the peptoid specific IgGs in AD patients.

    The utility as a biomarker test for AD, however, may be a long way off, as many of the aforementioned studies have shown (or rather failed to show). The number of samples used in the current study was very small, and it remains to be shown how the test would perform if another 1,000 samples were measured. The good news is that large AD plasma sample collections exist now (e.g., the Alzheimer’s Disease Neuroimaging Initiative or the Australian Imaging Biomarker and Lifestyle Flagship Study of Ageing), and the specificity of the test could be validated relatively quickly. It is also unclear how many peptoids were identified to be significantly different between disease and control. Three out of 15,000 would be an extremely small number to identify with convincing statistical power using current algorithms, but maybe many hundreds of additional peptoids with detectable but lower binding capacity were found. On the other hand, the small number of peptoids could indicate that AD patients possess a highly specific immune response against an unknown antigen. If true, that would most certainly change the current view of this disease.

    In summary, the current study describes an exciting new tool to rapidly probe the immune repertoire of humans and other species. This could lead to the discovery of new antigens and pathological processes involved in disease and possibly to the development of new biomarkers.

    View all comments by Tony Wyss-Coray
  3. This is good work. The biomarker is the most important tool for studying AD. The authors report results for a small number of patients for whom diagnosis was confirmed at autopsy. It is also very important to diagnose in living subjects who have early AD or MCI. In addition to AD biomarkers, one should also focus on tau, which is another hallmark of AD. AD biomarkers are valuable in different ways. The important factor is to identify the people, or groups, who have greater risk to develop AD.

    View all comments by Suhail Rasool

Make a Comment

To make a comment you must login or register.


External Citations

  1. Opko Health Laboratories
  2. company press release

Further Reading

Primary Papers

  1. . Identification of candidate IgG biomarkers for Alzheimer's disease via combinatorial library screening. Cell. 2011 Jan 7;144(1):132-42. PubMed.