Observation of spatial propagation of amyloid assembly from single nuclei

Abstract

The crucial early stages of amyloid growth, in which normally soluble proteins are converted into fibrillar nanostructures, are challenging to study using conventional techniques yet are critical to the protein aggregation phenomena implicated in many common pathologies. As with all nucleation and growth phenomena, it is difficult to track individual nuclei in traditional macroscopic experiments, which probe the overall temporal evolution of the sample, but do not yield detailed information on the primary nucleation step as they mix independent stochastic events into an ensemble measurement. To overcome this limitation, we have developed microdroplet assays enabling us to detect single primary nucleation events and to monitor their subsequent spatial as well as temporal evolution, both of which we find to be determined by secondary nucleation phenomena. By deforming the droplets to high aspect ratio, we visualize in real-time propagating waves of protein assembly emanating from discrete primary nucleation sites. We show that, in contrast to classical gelation phenomena, the primary nucleation step is characterized by a striking dependence on system size, and the filamentous protein self-assembly process involves a highly nonuniform spatial distribution of aggregates. These findings deviate markedly from the current picture of amyloid growth and uncover a general driving force, originating from confinement, which, together with biological quality control mechanisms, helps proteins remain soluble and therefore functional in nature.

Fig. 1.

Schema illustrating the design of the microfluidic device used in this study (A). A droplet maker (B) is coupled to a storage element (C) where elongated drops are incubated and can be monitored individually as a function of time. (D) Schema illustrating the encapsulation of protein molecules (blue spheres) and thioflavin T fluorophores (green spheres) into aqueous droplets stabilized by perfluorinated polyether-polyethyleneglycol (PFPE-PEG) block-copolymer surfactants. Misfolded proteins that assemble into cross-β structures can be detected through ThT fluorescence from the droplet after incubation.

Fig. 2.

Propagation of amyloid growth within microdroplets. Single droplets are observed as a function of time: bright field microscopy (A, Upper) and fluorescence microscopy (A, Lower) images acquired at regular intervals during the incubation of a droplet. The widths of all the droplets analyzed here are 30 μm, and their heights are 25 μm leading to a constant Laplace pressure inside all the droplets independent of their length. (B, Inset) The integrated fluorescence intensity from the droplet as a function of time. In B is shown the velocity of the reaction front measured in two different drop geometries; blue lines denote velocity measurements in elongated droplets, green in corrugated droplets as shown in the insert and red for measurements in elongated droplets acquired for protein solutions purified by size exclusion chromatography (Materials and Methods). The gray lines are the corresponding results from the simulations in Fig. 3. (C) The width of the reaction front measured in the microdroplets (blue lines) and from the simulation in Fig. 3 (gray lines).

Fig. 3.

Simulations of the spatial propagation of amyloid growth. The time evolution is computed starting from a single filament at t = 0 and followed by tracking the position and length of individual filaments as they grow (A) with an elongation rate k₊ = 8.9·10⁴ M^-1 s^-1 (see main text) in a monomer concentration of m_tot = 5.2 mM and multiply by fragmentation with a rate constant k_- = 2.0·10^-8 s^-1. The amyloid conversion is observed to proceed as a chemical wave (B) in a markedly similar patter to that measured in the experiments (Fig. 2).

Fig. 4.

Scaling of lag time with system size. In A, the lag times prior to the observation of fluorescence are reported for an array of 52 droplets of varying size measured simultaneously. A fit to the equation τ = c_nV^-1 + τ_g is shown in blue with c_n = 1.7 ± 0.25·10^-7 L^-1 s and τ_g = 94 ± 8 min. The first data point (gray) shows the lag time in a macroscopic sample volume of 200 μL but otherwise identical conditions to the microfluidic data. The size independence in the limit of large systems originates from the presence of multiple nucleation sites for large volumes, as described in the text and schematically illustrated at the bottom of panel A. Multiple sites can be observed directly in droplets of sufficient size: panel B shows the time evolution of aggregation in such a droplet (Upper, bright field; Lower, fluorescence microscopy).