The Aβ42 peptide comes in a variety of aggregation states, from monomers to small oligomers to fibrils to frank amyloid. In his talk, Koichi Iijima presented data suggesting that changing the intraneuronal aggregation state of Aβ could lead to different patterns of neurotoxicity in his fruit fly model of Aβ accumulation.

When he was with Yi Zhong at Cold Spring Harbor Laboratory in New York, Iijima developed a Drosophila model of AD by overexpressing either Aβ40 or Aβ42 in neurons (see ARF related news story). The flies with Aβ42 had learning deficits, reduced lifespan, amyloid buildup, and neurodegeneration. The flies also showed both plaques and intraneuronal accumulation of Aβ42, mainly in the protein secretory pathway and lysosomes.

Now starting his own lab at Thomas Jefferson University, Philadelphia, Pennsylvania, Iijima is exploring how the aggregation state of Aβ might affect its toxicity. To do this, Iijima and colleagues took the tack of expressing Aβ42 mutants in their Drosophila AD model. They chose the Arctic mutation to increase aggregation, and made a leucine to proline substitution at position 17 to produce a peptide less likely to aggregate.

Their hypothesis was that the different conformations might lead to different phenotypes, and they were right. They found that the Arctic form of Aβ accumulates in neuronal cell bodies and causes cell body degeneration. In contrast, the aggregation-defective mutant appeared mainly in neurites and induced neurite degeneration. Previously, they had shown that wild-type Aβ42 accumulated in both locations.

In addition to neuron loss, the flies developed age-related short-term memory deficits and motor problems, and had a shorter lifespan. All these effects were more pronounced with the Arctic mutation, consistent with its enhanced pathogenicity in people. But the aggregation-deficient Aβ had its own sticking point—it caused greater memory deficits than wild-type Aβ.

Their results show that Aβ42 with different aggregation properties causes distinct behavioral and pathological phenotypes. The coincidence of localization and pathology strengthens the idea that intraneuronal Aβ is the toxic species in this model. Iijima hypothesized that wild-type Aβ may assume different conformations in cells, leading to its own demonstrated multifaceted toxicity.

Generic amyloid as antigen…
Many proteins unrelated to Alzheimer disease show a propensity to aggregate into amyloid structures. Because of this, and from Charlie Glabe’s work with oligomer-reactive antiserum, it has been proposed that peptides of varying primary structure adopt a common toxic oligomeric conformation. To test this idea, Marcelo Vieira, a graduate student working with Sergio Ferreira and Fernanda De Felice at the Federal University of Rio de Janeiro, Brazil, investigated the neurotoxicity of soluble oligomers from one of these unrelated proteins, hen egg white lysozyme (HEWL).

Vieira chose HEWL because the human lysozyme protein is well-known to be amyloidogenic, and variants of it cause hereditary systemic amyloidoses. He showed that under partially denaturing conditions, HEWL formed β-sheet-rich soluble oligomers, which then went on to form fibrils. The oligomers (at 5 uM) were toxic to rat cortical neurons in culture. At higher concentrations (20 uM), the oligomers induced tau phosphorylation at sites associated with AD pathology (Thr212/Ser214 and Ser404). Finally, Vieira showed that injecting the oligomers into the rat cerebral cortex caused neurodegeneration not only in the cortex, but also in the hippocampus, thalamus, and amygdala, as measured by Fluoro-Jade C staining. This effect was not seen with fibrillar preparations of lysozyme or Aβ. An unexpected result was that the HEWL oligomers did not react with a commercially available preparation of the A11 anti-oligomer antiserum, in contrast to Glabe’s original report, which showed that the antiserum reacted with oligomers of human lysozyme (see ARF related news story). Nonetheless, Vieira’s results support the toxic conformation theory.

One corollary of that theory holds that antibodies that can neutralize harmful conformations should be neuroprotective. Using that idea as the basis for a novel vaccine strategy, Becky Stodola, working with Pritam Das and Todd Golde at the Mayo Clinic in Jacksonville, Florida, is testing the use of heterologous amyloids as immunogens to induce an anti-plaque response. The advantage of this approach lies in the use of non-human protein sequences that lower the risk of generating self-reactive antibodies. The structure of amyloid, with its large molecular weight, repetitive structure, and poor clearance and complement activation properties, also means it is more likely to induce a T cell-independent response, which could further reduce unwanted inflammatory reactions. Stodola showed that immunization of mice of various strains with amyloid formed from peptides derived from several adenoviral or bacterial proteins, the yeast prion sup35, or even a butoxycarbonyl dipeptide-induced antiserum that recognized fibrillar Aβ. The antisera have not yet been tested against monomeric or oligomeric Aβ, Stodola said. The group is now working on evaluating the heterologous amyloid vaccines to see if they reduce Aβ deposition in mice.

…Or Oligomer
Taking the passive vaccination approach, Matthew Seager from Merck, West Point, Pennsylvania, presented a poster on Merck’s anti-oligomer antibody development. Working with Grant Krafft at Acumen Pharmaceuticals, South San Francisco, California, they have produced multiple high-affinity, humanized monoclonal antibodies that preferentially recognize soluble Aβ oligomers. The antibodies block oligomer binding to neurons, and their new data showed that weekly administration lowered brain Aβ in mice. Antibodies with therapeutic potential are now in testing for their ability to improve memory and behavior in mice.

Another Merck researcher, Gene Kinney, had a separate presentation on selecting peptides for Aβ immunization. By keeping peptides short (eight or fewer amino acids), Kinney and colleagues hope to avoid triggering T cell responses, which require the presentation of longer peptides. By systematically scanning the entire Aβ sequence with overlapping peptides, they discovered scattered immunogenic sites that were not predicted based on their position in the Aβ sequence. Immunization of Rhesus monkeys with a dual antigen featuring two different peptides conjugated onto one carrier protein produced high levels of amyloid-reactive antibodies in Rhesus monkeys, with no indication of amyloid-specific T cell activation.—Pat McCaffrey.