16 June 2009. Amyloid-β (Aβ) peptide oligomers have come under intense scrutiny as the prime suspects in the synapse loss and neurotoxicity associated with Alzheimer disease. But despite their sticky nature, the oligomers have proved a slippery foe to researchers wanting to understand their physical makeup. Nearly impossible to isolate, constantly changing in solution, the oligomers have been hard to handle with traditional biochemical techniques.
For a new look, Michael Bowers and colleagues at the University of California, Santa Barbara, used a modified mass spectrometry technique to characterize the oligomerization states of Aβ peptides. Published online June 14 in Nature Chemistry, the work nominates tetramers and dodecamers as key players in the formation of larger amyloid conglomerates, and in the toxicity of Aβ42.
Another paper, in the 10 June Journal of Neuroscience, also reports news on oligomer conformation and its role in toxicity. That work, from Gerd Multhaup and colleagues at the Berlin Free University in Germany, shows that changing a single residue in the Aβ42 peptide can enhance oligomer formation while decreasing toxicity. The results suggest that the toxic Aβ42 oligomer takes on a special shape beyond the fact of its multiplexing, and that that shape could be a target for neuron-sparing therapies.
The senior author on the first report, Bowers is a chemical physicist who developed the method of ion-mobility spectrometry, a technique for analyzing protein structures in a gas-phase system. In the technique, a peptide solution is first sprayed out in fine droplets, just as in mass spec. The solvent evaporates, leaving a collection of ionized protein aggregates floating in a gaseous environment. The mixture is then injected into a helium-filled cell where the aggregates drift under a weak electrical current, with their speed determined by the molecule’s 3D shape. By measuring the time it takes for different species to arrive at the other end of the cell, the researchers can calculate a three-dimensional cross-sectional area for the aggregates. (For a more thorough explanation of the technique, see the accompanying News and Views by David Clemmer and Stephen Valentine, Indiana University, Bloomington.)
In collaboration with Gal Bitan and David Teplow, both of University of California at Los Angeles, Bowers turned this technique to analyzing solutions of Aβ peptides. It took two years for first author Summer Bernstein to get a successful look at Aβ42, because the peptide solution quickly clogged the spray nozzle as it aggregated into larger fibrils. When she finally worked out the conditions, she found a mixture of structures that included dimers, tetramers, hexamers, and dodecamers (Bernstein et al., 2005).
The new work compares the aggregation patterns for a number of Aβ peptides. In contrast to Aβ42, the analysis of Aβ40 was easy, Bowers told ARF. That peptide sprayed nicely and produced drift time peaks corresponding to monomers, dimers, and tetramers, but no higher aggregates. This suggested that the hexamers and dodecamers were on the pathway to fibril formation. This idea was supported when solutions of two non-fibrillogenic forms of Aβ42, one carrying a proline at residue 19 and the other with an oxidized Met35 residue, also revealed tetramers as the highest oligomers.
Cross-section measurements revealed a possible structural basis for the different oligomer profiles. By comparing the experimentally determined area to calculated areas based on the possible arrangements of spherical units, the researchers arrived at likely configurations for tetramers of the various peptides. Their measurements suggest that Aβ40 forms a closed tetramer, with two dimers forming a V of about 30 degrees. In contrast, the Aβ42 tetramer was much more open, with an angle approaching 120 degrees. This accessibility could explain why Aβ42 goes on to incorporate more subunits, while the Aβ40 tetramer is a dead end, the authors suggest.
In the same way, the Aβ42 hexamer measurements best fit a planar ring structure, and the dodecamer, two rings on top of one another. The dodecamer structure seemed to resist the addition of more Aβ units, because the investigators did not see oligomers larger than 12-mers, in spite of the fact that their solutions clearly contained larger aggregates that could clog the spray port. The results are consistent with the dodecamer as a metastable intermediate that would need to undergo some kind of conformational rearrangement, possibly involving the acquisition of β-sheet structure to start fibril formation. These results fit with previous crosslinking studies from co-authors Teplow and Bitan (Bitan et al., 2003) supporting the idea of a six-member “paranucleus” as an important unit of assembly.
The formation of hexamers and dodecamers preferentially by Aβ42 could explain the toxicity of 42 over 40, the authors say. Bowers believes the dodecamer is the memory-impairing structure that Karen Ashe and colleagues isolated from mouse brain, and call Aβ*56 (see ARF related news story on Lesne et al., 2006). The dodecamer “is the terminal species observed in our experiments, and has a mass of ~55.2 kDa, which suggests it is the soluble assembly that Lesne et al. observed,” the authors write.
However, the nomination of hexamers and dodecamers as the toxic species runs counter to other research that implicates dimers isolated from human Alzheimer disease brains as the toxic species (see ARF related news story on Shankar et al., 2008), or to work that implicates tetramers (see comment from Gerd Multhaup below). The data from Bowers and colleagues suggest that tetramers, but also dimers of Aβ42 exist in multiple conformations, and that dimers show different cross-sectional areas compared with dimers of Aβ40 or of the Pro19 or oxidized Met forms of Aβ42. If toxicity is a matter of subtle conformational determinants, as many scientists argue, it is possible that these lower-n oligomers could also have differential toxic effects.
Interestingly, another recent paper from Bowers and Teplow using the same technique shows that in mixtures of Aβ40 and 42, oligomer formation tops out with tetramers. The implication there is that Aβ40, which is more abundant than Aβ42 in brain, could be preventing higher oligomer and fibril formation (Murray et al., 2009; see also Kim et al., 2007).
Bowers stressed the role of the hydrophobic tail in oligomer assembly. “The thing that distinguishes Aβ42 is the very long hydrophobic tail from residues 29 to 42. Aβ40 also has a fairly long hydrophobic tail from 29-40, but apparently, those two residues are enough to swing the balance between 40 and 42. It’s astonishing to me that we observed that Aβ40 stops at the tetramer; if you oxidize Met35 in Aβ42 it stops at the tetramer, and if you change residue 19 to proline it stops at the tetramer.”
The critical role of hydrophobic residues, and their effects on conformation, is borne out by the work of Multhaup and colleagues, who looked at the effects of mutating a glycine residue in that same tail region of Aβ42. In their paper, first author Anja Harmeier substituted alanine or isoleucine for Gly33 in the crucial GxxxG dimerization motif. More hydrophobic substitution resulted in a rapid oligomerization of synthetic peptides to higher-order oligomers (16-20-mers). The oligomers seemed to adopt a more compact conformation based on proteolytic cleavage patterns, and molecular modeling indicated that increased hydrophobicity may have promoted β-sheet conformation. These effects were unique to Gly33, as substitution at Gly29 did not affect aggregation state.
The German researchers then tested the toxicity to neuronal cells of the Aβ variants and their oligomeric fractions. They found that the wild-type peptide was most toxic in low-n oligomer fractions (dimers to tetramers), while none of the Gly33 mutant aggregates were. The Drosophila eye photoreceptor assay yielded the same result in vivo, with the Gly33 mutant peptide showing no toxicity. Moreover, the investigators found that the mutation abolished the ability of Aβ42 tetramers to inhibit long-term potentiation in hippocampal slices. They conclude that the toxicity of Aβ42 oligomers relies on a Gly33-dependent conformation, not just the fact of oligomerization itself. Their identification of toxic versus innocuous oligomers should facilitate exploration of the toxic mechanism on cells, they conclude.
While this study and the Bowers work keep conformational issues front and center, an unresolved question remains: Can in-vitro studies with synthetic peptides truly recapitulate the state of Aβ that is produced, and aggregates, in the aging human brain? Until the question of which oligomers are toxic and which might be protective is better understood with natural oligomers, as well, the quest to alter oligomerization as a therapeutic strategy proceeds at some risk.—Pat McCaffrey.
Bernstein SL, Dupuis NF, Lazo ND, Wyttenbach T, Condron MM, Bitan G, Teplow DB, Shea J, Ruotolo BT, Robinson CV, Bowers MT. Amyloid-b protein oligomerization and the importance of tetramers and dodecamers in the aetiology of Alzheimer's disease. Nature Chemistry. 2009 June 14; advance online publication. Abstract
Clemmer DE, Valentine SJ. Protein oligomers frozen in time. Nature Chemistry. 2009 June 14; advance online publication. Abstract
Harmeier A, Wozny C, Rost BR, Munter LM, Hua H, Georgiev O, Beyermann M, Hildebrand PW, Weise C, Schaffner W, Schmitz D, Multhaup G. Role of amyloid-beta glycine 33 in oligomerization, toxicity, and neuronal plasticity. J Neurosci. 2009 Jun 10;29(23):7582-90. Abstract