Quantitative analysis of co-oligomer formation by amyloid-beta peptide isoforms

Abstract

Multiple isoforms of aggregation-prone proteins are present under physiological conditions and have the propensity to assemble into co-oligomers with different properties from self-oligomers, but this process has not been quantitatively studied to date. We have investigated the amyloid-β (Aβ) peptide, associated with Alzheimer's disease, and the aggregation of its two major isoforms, Aβ40 and Aβ42, using a statistical mechanical modelling approach in combination with in vitro single-molecule fluorescence measurements. We find that at low concentrations of Aβ, corresponding to its physiological abundance, there is little free energy penalty in forming co-oligomers, suggesting that the formation of both self-oligomers and co-oligomers is possible under these conditions. Our model is used to predict the oligomer concentration and size at physiological concentrations of Aβ and suggests the mechanisms by which the ratio of Aβ42 to Aβ40 can affect cell toxicity. An increased ratio of Aβ42 to Aβ40 raises the fraction of oligomers containing Aβ42, which can increase the hydrophobicity of the oligomers and thus promote deleterious binding to the cell membrane and increase neuronal damage. Our results suggest that co-oligomers are a common form of aggregate when Aβ isoforms are present in solution and may potentially play a significant role in Alzheimer's disease.

Figure 1. Schematic of the statistical mechanical model used to estimate Aβ oligomer numbers and relative composition.

For the single-species datasets, the model considers oligomers of any length, whereas for the co-oligomerising datasets it considers monomers and dimers as the single-species analysis predicts a very low number of oligomers larger than dimers (Fig. 3b).

Figure 2. Equilibrium oligomer concentrations as a function of the total initial monomer concentration in the aggregation reaction.

(Error bars SD, N (samples) = 3). The oligomer concentration was modelled and fitted separately for both Aβ40 (a), Aβ42 (b), and the 1:1 mixture (c); allowing extraction of the free energies of oligomerization and estimation of the CAC for Aβ40 and Aβ42 (fitted curves shown overlaid). The shaded bounds on these charts are curves plotted using the maximum and minimum free energies of oligomerization, and of fibril formation (given by the CAC) that still lie within the majority of the error bars. (d) The fitted free energies of oligomerization are also shown in comparison to the free energies of fibril formation obtained by direct measurement of the CAC (“Direct”), and also the free energies of fibril formation obtained from the fitted estimation of the CAC (“Fitted”).

Figure 3. Simulation of Aβ40-Aβ42 co-oligomerization equilibrium behaviour at a total Aβ concentration of 1 nM for a range of Aβ42 proportions, using .

Simulations at 5 nM and 10 nM of total Aβ are shown in Supplementary Fig. S3. (a) Total oligomer concentration and composition as a function of Aβ42 proportion. (b) Estimated concentrations of oligomers of different sizes at 1 nM total protein concentration, calculated by assuming that is unchanged from the single-species value (in which case the ratio of Aβ40:Aβ42 is irrelevant). The true distribution will decline with oligomer size even more rapidly, as visual inspection of the data shows to be less favourable than the single-species values. The error bars correspond to averaged uncertainty in the ΔG measurements. (c) The relative concentration of oligomers estimated to be bound to the surface of a neuronal cell, expressed relative to the concentration of oligomers bound to the surface at 1 nM of Aβ40. This result assumes that the relative affinity of co-oligomers for the cell membrane is 2 times higher than the affinity of Aβ40 oligomers, and that the relative affinity of Aβ42 oligomers is 4 times higher than that of Aβ40 oligomers.