16 November 2007. The first AD vaccine to be tested in humans, Elan’s AN-1792, stalled in phase 2b when some of the recipients developed meningoencephalitis. The inflammation was blamed on self-reactive T cells that infiltrated the brain (see ARF related news story) in response to immunization with full-length Aβ42 peptide. After that unfortunate turn, Elan shifted its attention to testing passive vaccination with a humanized monoclonal antibody (see ARF related news story) and to a new, different active vaccine. For their part, academic researchers focused their efforts on different immunogens. The current crop of active vaccine candidates features smaller fragments of Aβ that induce B cells to produce antibody, but are too small on their own to activate T cells (e.g., Lemere et al., 2007). The Catch-22 is that maximally stimulating B cells for high-level antibody production requires help from activated T cells. But how to get that help without eliciting unfriendly autoimmunity?
In the November 14 Journal of Neuroscience, David Cribbs and Michael Agadjanyan from the University of California, Irvine, describe a vaccine that takes a divide-and-conquer approach to the B/T cell problem. Their immunogen consists of an Aβ1-11 peptide to activate antibody production, and a separate, synthetic T cell-activating peptide (the pan-DR epitope, or PADRE; see ARF meeting report) to provide help without the risk of generating Aβ-reactive T cells (see ARF meeting report for more background on the vaccine). Giving the vaccine to adult AD mice resulted in the clearance of plaques and insoluble Aβ with an efficiency proportional to the level of Aβ antibody raised. On the safety side, the vaccination did not trigger T cell infiltration or microhemorrhage in the animals’ brains.
Despite these hopeful signs, the vaccine did not decrease soluble Aβ in adult animals with established plaque pathology. This result supports the idea that Aβ vaccines will be useful as preventive measures to forestall amyloid accumulation, but may not work in a therapeutic mode after plaque pathology is established, the authors write.
To test the vaccine, first author Irina Petrushina and colleagues used Tg2576 mice, and found that immunization resulted in significant reductions in Aβ40 and Aβ42 in brain. Mice were immunized beginning at 9 months of age, and analyzed on average 10 months later. The vaccine resulted in the production of anti-Aβ antibodies, and activated T cells that were specific for the non-self PADRE peptide. In contrast, mice vaccinated with full-length Aβ42 produced Aβ-reactive T cells.
Like people in the AN-1792 trial, the mice showed a variable response to vaccination. Importantly, the researchers found that antibody titers correlated with the removal of Aβ42; that is, a higher immune response equated to more efficient clearance of amyloid deposits. Only the mice with the highest titers showed a statistically significant reduction in insoluble Aβ40 and Aβ42.
Despite the disappearance of plaque, the levels of soluble Aβ or oligomers detected in the brains of immunized animals did not change. The antibodies generated by this vaccine have previously been shown to recognize soluble, fibrillar, and oligomeric forms of Aβ (Mamikonyan et al., 2007). In vitro, the antibodies delayed fibril formation, but did not disrupt preformed oligomers. However, the in vitro binding did not translate into reduction of soluble complexes in the vaccinated mice.
All this argues for using vaccination early in the course of AD, Agadjanyan told Alzforum. “The whole idea of vaccines is to be preventative, not therapeutic,” he said. “While it is obviously preferable to start immunizing before widespread neuronal loss, it may also be necessary to begin before plaque is established in the brain or blood vessels.”
This idea draws support from other recent data that show the clearance of soluble Aβ and improvements in behavior in animal models are possible if immunization is started early (see ARF related news story and Chen et al., 2007). Agadjanyan said his group has seen similar clearance of soluble Aβ using a DNA vaccine if they start treatment in young animals. On the flip side, late treatment could have the opposite effect: in human studies, researchers have reported an elevation of soluble, potentially toxic forms of Aβ in the brain of patients from the AN-1792 trial (Patton et al., 2006). And indeed, the fate of Aβ solubilized from plaques—Can it be cleared? Could it be toxic? Will it redeposit?—is a focus of concern and close observation in ongoing vaccination studies.
This vaccine has several potential advantages over the full Aβ42 immunogen, Agadjanyan said. The first, which the current paper shows, is that separating the T cell epitope from the Aβ peptide results in high antibody production without generating Aβ-reactive T cells. The second, which remains to be proven, is that using a promiscuous T cell activator like the PADRE peptide should increase the response to the vaccine. In the AN-1792 trial, only 20 percent of immunized patients developed Aβ antibodies after two or three injections. The large majority of non-responders might reflect variability in the HLA histocompatibility antigens in the population, so that only people with a particular HLA-DR subtype respond well to Aβ. Because PADRE activates T cells bearing any of the 15 major forms of the HLA-DR antigen, it may elicit a better response in the general population, Agadjanyan speculated.
His team’s vaccine is not ready for human testing, because it cannot be made in sufficient quantities, Agadjanyan said. However, the group is working on an alternative approach to scale up synthesis of the Aβ-PADRE immunogen for such studies.—Pat McCaffrey.
Petrushina I, Ghochikyan A, Mktrichyan M, Mamikonyan G, Movsesyan N, Davtyan H, Patel A, Head E, Cribbs DH, Agadjanyan M. Alzheimer's disease peptide epitope vaccine reduces insoluble but not soluble/oligomeric Abeta species in amyloid precursor protein transgenic mice. J Neurosci. 2007 Nov 14; 27:12721-12731. Abstract