Aβ peptides inhabit the human brain in various sizes and shapes, and an enduring question among Alzheimer’s disease researchers has been, Which of them is neurotoxic? Recent studies on Aβ purified from human cerebrospinal fluid (Klyubin et al., 2008) and postmortem brain tissue (Shankar et al., 2008) have steered attention toward dimers. However, “we didn’t know whether the toxic species was the dimer itself, or something the dimer formed,” said Dominic Walsh, University College Dublin, Ireland, who was a coauthor on both reports. Now, new data by Walsh and colleagues points to the latter. Analyzing a synthetic Aβ(1-40) dimer, the authors report in the October 27 Journal of Neuroscience that it had no innate toxicity and did not form typical amyloid fibrils, but quickly aggregated into shorter, thinner protofibrils that block long-term potentiation in mouse hippocampus. The synthetic dimers formed toxic protofibrils remarkably faster than did wild-type Aβ monomers. Taken together, the findings suggest that while dimers are not the be-all and end-all for Aβ toxicity, they are essential building blocks for the species causing neurodegeneration.
Recent studies identified Aβ dimers as a component of human CSF and postmortem brain, and Walsh and colleagues found dimers to be highly specific for AD-type dementia (McDonald et al., 2010 and ARF related news story). Generally speaking, the bar has risen toward trying to study Aβ oligomers from in vivo human sources. However, because human CSF and brain samples contain a slew of other Aβ species as well, they are unsuitable for biophysical analyses of the dimers. The scientists needed a purer preparation of dimer, and much more of it. By replacing serine 26, one of two serines in Aβ40, with a cysteine, first author Brian O’Nuallain and colleagues substituted a hydroxyl group with a sulfhydryl group that allows formation of disulfide cross-linked dimers from synthetic S26C monomers. They purified the AβS26C dimers using size exclusion chromatography, and then analyzed their aggregation propensity, structural features, and synaptotoxicity.
Comparing aggregation rates of wild-type Aβ monomer and AβS26C dimer, the difference “was dramatic,” Walsh said. At relatively high concentrations (e.g., 10 micromolar), the dimer completely aggregated in about 12 hours. “If you had a similar concentration of the monomer, it would take days,” he said. With agitation to speed up the reaction, AβS26C dimers also clumped comparatively faster, forming β-sheet-rich assemblies at starting concentrations less than a fourth of what it took to drive aggregation of wild-type monomers. All told, the dimers “aggregated faster and at lower concentrations than Aβ monomers,” Walsh said.
Moreover, the dimer formed something fundamentally different than did the monomer. Monomers morph into fibrils that crash out of solution when centrifuged, but AβS26C dimers did not. “They formed things we had described more than a decade ago as intermediates called protofibrils,” said Walsh, referring to earlier work (Walsh et al., 1997) he did at Harvard Medical School as a postdoctoral trainee of David Teplow, who is now at the University of California, Los Angeles. Protofibrils are shorter than fibrils, and only about half as wide—“not big enough to fall out of solution like true amyloid fibrils do.”
Freshly prepared AβS26C dimers had no readily identifiable secondary structure, and did not impair synaptic plasticity of mouse hippocampal slices. However, when allowed to aggregate, AβS26C-containing assemblies potently blocked long-term potentiation. The data show that “dimers, as distinct from monomers, can assemble into toxic species that persist for a longer period of time,” Walsh said.
The work strikes a chord with a PNAS paper published last month by Torleif Hard, Swedish University of Agricultural Sciences, Uppsala, and colleagues (Sandberg et al., 2010 and ARF related news story). In that study, the researchers locked Aβ monomer into a hairpin conformation by swapping in two cysteines for the alanines at positions 21 and 30. Also this year, researchers at Kyoto University in Japan used another type of cysteine-linked dimer to stabilize protofibrils (Yamaguchi et al., 2010).
These two studies, along with the present work, “are encouraging in similar ways and for a number of reasons,” Hard noted in an e-mail to ARF. “First, they all constitute new methods to ‘engineer’ protofibrils while avoiding fibril formation. Second, they reveal similar and presumably significant features of the Aβ protofibrils: their overall dimensions and a large fraction of secondary β-sheet structure content. (See full comment below.)
One thing the current study leaves unclear is whether the AβS26C dimer “actually mimics the dimer found in AD patients,” Hard wrote. “The S26C dimer may still be a significant building block, and it would be important to find out what it looks like.” Walsh told ARF his team plans to use nuclear magnetic resonance (NMR) spectroscopy and other high-resolution structural techniques to look more closely at dimer structure. The key to designing therapeutics to block Aβ-induced toxicity may lie in determining the structural basis for why dimers, much more so than wild-type monomers, form stable protofibrils so quickly, Walsh said.
On a broader level, the study highlights the importance of structure in understanding how and why Aβ is neurotoxic. “If you want to establish structure-neurotoxicity relationships, then you need to know the structures of the assemblies with which you’re working,” Teplow told ARF. “Rather than asking whether there’s a particular Aβ-derived neurotoxin that causes AD, we should be asking what the structure of that neurotoxin is.”
Walsh agreed. “People are always looking for a single species. It's not a single neurotoxic species, but rather a pathway,” he said. “There's a mixture of different species with different biological activities.” This message reinforces the key conclusion of a recent study by Teplow and colleagues, who also used crosslinking to stabilize Aβ40 oligomers (Ono et al., 2009). That report concluded “that there is no such thing as one neurotoxin,” Teplow said. “The field needs to look at AD as being caused by a distribution of structures, each with its own ability to kill neurons.”—Esther Landhuis