Different soluble aggregates of Aβ42 can give rise to cellular toxicity through different mechanisms

Abstract

Protein aggregation is a complex process resulting in the formation of heterogeneous mixtures of aggregate populations that are closely linked to neurodegenerative conditions, such as Alzheimer's disease. Here, we find that soluble aggregates formed at different stages of the aggregation process of amyloid beta (Aβ42) induce the disruption of lipid bilayers and an inflammatory response to different extents. Further, by using gradient ultracentrifugation assay, we show that the smaller aggregates are those most potent at inducing membrane permeability and most effectively inhibited by antibodies binding to the C-terminal region of Aβ42. By contrast, we find that the larger soluble aggregates are those most effective at causing an inflammatory response in microglia cells and more effectively inhibited by antibodies targeting the N-terminal region of Aβ42. These findings suggest that different toxic mechanisms driven by different soluble aggregated species of Aβ42 may contribute to the onset and progression of Alzheimer's disease.

Fig. 1

Soluble Aβ42 aggregates at different aggregation stages show different toxicity mechanisms. a The aggregation Aβ42 was monitored by ThT fluorescence. Each data point represents the mean of three independent biological repeats and its error bar the corresponding standard deviation. Aliquots containing Aβ42 aggregates were taken from the aggregation reaction at two different time points—at the end of the lag phase (15 min, for ‘early’ soluble aggregates) and at the midpoint of the growth phase (30 min, for ‘late’ soluble aggregates). b The lipid bilayer permeability induced by Aβ42 aggregates was measured by immobilizing on glass cover slides hundreds of nano-sized individual lipid vesicles in the presence of a Ca²⁺-sensitive dye. If the aggregates disrupt the integrity of the lipid bilayer, Ca²⁺ from the external medium enters into individual vesicles to a level that can be quantified using highly sensitive total internal fluorescence microscopy (TIRF) measurements. c Early Aβ42 aggregates cause a higher level of bilayer permeability than later Aβ42 aggregates. The data were averaged over three biological repeats, with each repeat involving the analysis of 6000 individual lipid vesicles, the error bars represent standard deviation, and the statistical significance was calculated using a two-sample t-test. d The inflammatory response in microglia was quantified using an ELISA assay to measure the levels of secreted tumour necrosis factor alpha (TNF-ɑ). e In contrast to the bilayer permeability assay, the late aggregates were more effective than the early aggregates. The data are averaged over three biological repeats, each repeat was averaged over three wells, each of which contained 200,000 to 300,000 BV2 cells, the error bars represent the standard deviation among the repeats, and the statistical significance was calculated using a two-sample t-test (unpaired). Data points for each biological replicate shown with red circles. Source data of a, c and e are provided as Source Data files. Aggregates used for the membrane permeability and the inflammation assays were collected from the same aggregation reaction

Fig. 2

Aβ42 aggregates of diverse sizes exhibit different relative toxicity by distinct mechanisms. a Aβ42 aggregation monitored by ThT fluorescence. Soluble aggregates formed at different time points were mixed together and separated via a discontinuous sucrose gradient. Three different time points during the aggregation of 2 µM Aβ42 in SSPE buffer at 37 °C were chosen: (i) At the end of the lag phase (15 min), (ii) at the middle of growth phase (30 min) and (iii) during the plateau phase (60 min). b Aliquots collected at these time points were loaded in a step gradient of 10% to 50% sucrose and ultracentrifuged and the fractions were collected immediately and stored. The sizes of the aggregates present within different densities of the sucrose solution increase with the sucrose density. c The lipid bilayer permeability assay shows that aggregates present at 20% sucrose are the most potent at membrane permeation. d Aggregates presents at 30% sucrose are the most effective at inducing inflammation. These experiments were carried out for two independent aggregation reactions of Aβ42 (n = 2) and the error bars represent the standard deviation of the mean for each field of view (for 2C) and for each well (2D). Source data of a, c and d are provided as a Source Data file

Fig. 3

AFM characterization of aggregates present in sucrose fractions. a–e Representative AFM images of the Aβ42 aggregates present at different fractions of sucrose gradient (left panels), with a magnified example of the species present in each fraction (central panel) and the corresponding cross-sectional dimensions (right panel). The 30% fraction (c) is the only one that shows protofilaments aggregates

Fig. 4

Statistical analysis of aggregates present in different sucrose fractions. As represented by box charts of the a height and b the deconvoluted diameter of the Aβ42 aggregates are shown. As the concentration of sucrose increase, the height as well as the diameter of individual aggregate increases. Source data of a and b are provided as a Source Data file

Fig. 5

Single-particle characterization of the dye-labelled Aß42 aggregates, a Schematic representation of the separation workflow of the Aβ42 aggregates: solutions of dye-labelled monomeric Aβ42 were allowed to aggregate and aliquots removed at the different times described in Fig. 2 (shown as coloured triangles in the graph). The aggregated mixture collected at different time points was then loaded onto a sucrose step gradient that is divided into five fractions, and was subsequently used for experimental measurements. b Illustration of the confocal FRET experiment, showing a schematic of the microfluidic channel used to deliver the samples for measurement. The confocal volume of the excitation beam was set to the centre of the channel. Both the donor and acceptor channel intensities were recorded for every aggregate flowing through the excitation volume. c Maxima of FRET efficiency for increasing sucrose concentrations shifts towards higher FRET efficiencies indicates that size of the aggregates present in sucrose solution increases. Relative FRET efficiency for each sucrose concentration are fitted with Gaussian function (solid lines) to get the maxima (shown in parenthesis). d Schematic of the TIRF imaging for single-aggregate imaging. e Images of aggregates present in all five fractions are shown to represent the sizes detected through TIRF microscopy and are representative of the indicated fractions. The scale bar is 3 µm. The FRET histograms of the fractions show a clear shift from a very narrow, low-intensity distribution in the smaller fractions to a broader higher intensity distribution in the higher fractions and in the pellet. Each histogram is the sum of aggregates of 27 field of views measured in three repeats of a single gradient ultracentrifugation preparation. Source data of c and e are provided as a Source Data file

Fig. 6

Influence of the different sequence regions of Aβ42 on its toxicity mechanisms. a We used five rationally designed antibodies that target different epitopes of the Aβ42 sequence. The antibodies DesAb_3–9 and DesAb_36–42 bind to the N-terminal and C-terminal regions of Aβ42, respectively. Representative experiments showing the concentration of each antibody (x-axis) added to aggregates present in solutions at b and d 20% and c and e 30% sucrose and the reduction of aggregate-induced toxicity. The error bars represent the standard deviation among the field of views (b, c) and among the well (d, e). Antibodies that target C-terminal regions of Aβ42 are more effective at reducing the membrane permeability induced by Aβ42 aggregates. Antibodies that target N-terminal regions of Aβ42 are more potent in reducing the inflammatory response induced by Aβ42 aggregates (d, e). As a positive control, we used lipopolysaccharides (LPS), which is known to induce TNF-α production in microglial cells. For each case, P values are calculated using two-sample t-test to compare the inhibition by most N-terminally binding antibody (DesAb3–9) and C-terminally antibody (DesAb36–42) at their highest concentration (two biological repeats n = 2, the lines are simply guides to the eye). Source data of b–e are provided as a Source Data file