Extracellular Vesicles Slow Down Aβ(1–42) Aggregation by Interfering with the Amyloid Fibril Elongation Step

Formation of amyloid-β (Aβ) fibrils is a central pathogenic feature of Alzheimer’s disease. Cell-secreted extracellular vesicles (EVs) have been suggested as disease modulators, although their exact roles and relations to Aβ pathology remain unclear. We combined kinetics assays and biophysical analyses to explore how small (<220 nm) EVs from neuronal and non-neuronal human cell lines affected the aggregation of the disease-associated Aβ variant Aβ(1–42) into amyloid fibrils. Using thioflavin-T monitored kinetics and seeding assays, we found that EVs reduced Aβ(1–42) aggregation by inhibiting fibril elongation. Morphological analyses revealed this to result in the formation of short fibril fragments with increased thicknesses and less apparent twists. We suggest that EVs may have protective roles by reducing Aβ(1–42) amyloid loads, but also note that the formation of small amyloid fragments could be problematic from a neurotoxicity perspective. EVs may therefore have double-edged roles in the regulation of Aβ pathology in Alzheimer’s disease.


SUPPLEMENTARY DATA (TABLES AND FIGURES)
Table S1.Mean and mode size (diameter) of EV particles collected from SH-SY5Y and HEK293-T cells as determined by NTA, corresponding to the data shown in Fig. 1a of the main text.

Extracellular vesicles slow down Aβ(1-42) aggregation by interfering with the amyloid fibril elongation step
Figure S1.Quantification of protein amounts in the EVs.The total protein amount in SH-SY5Y and HEK293-T EV samples was quantified using the Pierce BCA Protein Assay Kit as described in the main text.The protein amount per EV was obtained by dividing the protein mass with the EV particle concentration in the sample as determined by NTA.The experiment was performed in biological duplicate (N=2).

SUPPLEMENTARY TEXT: Estimation of fibril and vesicle concentrations 2.1 Number of lipids in a 100 nm synthetic lipid vesicle
In the manuscript text we convert lipid concentrations into lipid vesicle (particle) concentrations for previously published data on synthetic DOPC liposomes in order to compare the amounts of synthetic lipid vesicles and EVs in the aggregating Aβ(1-42) samples.
The number of lipids in a 100 nm lipid vesicle was calculated by taking into account two leaflets, a bilayer thickness of ~5 nm 1 and an average head-group area of 70 Å 2 per DOPC lipid 2 using the following equation: This results in ~ 8.1e04 lipids per vesicle.

Number of monomers in an Aβ(1-42) fibril
In the manuscript text we estimate the number of fibrils (per mL) formed in the EV-containing samples based on the following assumptions: a fibril length of 300 nm, an interstrand distance (rise) of 4.7 Å (ref), a packing of two Aβ(1-42) monomers per layer in the fibril 3,4 and full conversion of monomers into fibrils 5 .

Figure S2 .
Figure S2.Residual monomer content at the aggregation end-point.SDS-PAGE was used to analyse the residual monomeric Aβ(1-42) content in the samples at the aggregation endpoint.a-b Coomassie-stained gel showing the appearance of faint monomeric Aβ(1-42) bands at ~4.5 kDa in samples aggregated in absence and presence of increasing concentrations of a SH-SY5Y and b HEK293-T EVs.c Sample information and quantification of the bands as determined by relative density comparison with Image Lab (Biorad).

Figure S3 .
Figure S3.Aggregation of Aβ(1-42) in Opti-MEM medium control.Normalized data of 2 µM Aβ(1-42) aggregation with increasing volumes of Opti-MEM.Opti-MEM was added in volumes corresponding to an assumed EV concentration of 1.8E9 -7.2E9 particles/mL to reflect the EV concentrations used in the study.

Figure S4 .
Figure S4.Kinetic modelling of experimental data using the AmyloFit web-based tool.Normalized kinetic profiles of the aggregation of 2µM Aβ(1-42) in presence of indicated concentrations of EVs, corresponding to the data shown in Fig. 3a, d in the main text.The solid lines represent fits to the data based on a saturated secondary nucleation aggregation model with different parameters set as global or fixed.a, d, k+kn set as free parameter (b, e) k+k2, set as free parameter, (c, f) k+kn and k+k2 set as free parameters.g-h Change in k+kn and k+k2 corresponding to the fits in (e-f), illustrating the random change in k+kn.

Figure S5 .
Figure S5.Modelling of experimental data using the AmyloFit web-based tool.Normalized kinetic profiles of aggregation of 2µM Aβ(1-42) in presence of SH-SY5Y EVs ac or HEK293-T EVs d-f and increasing concentrations of pre-formed fibril seeds.Solid lines represent predictions based on kinetic modelling of reduction of the rate constant for primary nucleation, kn.The rate constants for secondary nucleation (k2) and elongation k+ were kept constant.

Figure S6 .
Figure S6.Modelling of experimental data using the AmyloFit web-based tool.Normalized kinetic profiles of aggregation of 2µM Aβ(1-42) in presence of SH-SY5Y EVs ac or HEK293-T EVs d-f and increasing concentrations of pre-formed fibril seeds.Solid lines represent predictions based on kinetic modelling of reduction of the rate constant for secondary

Table S3 . Fitted kinetic parameters for Aβ(1-42) aggregation in presence of HEK293-T EVs
(FigureS3d-f).k+kn is the product of the rate constants for elongation (k+) and primary nucleation (kn).k+k2 is the product of the rate constants for elongation and secondary nucleation (k2).KM is the Michaelis constant.A multi-step secondary nucleation dominated model was used in the web-based tool AmyloFit.KM was determined for Aβ(1-42) aggregation with no additives and was kept as a global constant for all further analyses.

Table S4 . Fitted kinetic parameters for SH-SY5Y EVs
. k+ is the rate constant for elongation, k2 is the rate constant for secondary nucleation, and kn is the rate constant for primary nucleation.