Trodusquemine enhances Aβ42 aggregation but suppresses its toxicity by displacing oligomers from cell membranes

Transient oligomeric species formed during the aggregation process of the 42-residue form of the amyloid-β peptide (Aβ42) are key pathogenic agents in Alzheimer’s disease (AD). To investigate the relationship between Aβ42 aggregation and its cytotoxicity and the influence of a potential drug on both phenomena, we have studied the effects of trodusquemine. This aminosterol enhances the rate of aggregation by promoting monomer-dependent secondary nucleation, but significantly reduces the toxicity of the resulting oligomers to neuroblastoma cells by inhibiting their binding to the cellular membranes. When administered to a C. elegans model of AD, we again observe an increase in aggregate formation alongside the suppression of Aβ42-induced toxicity. In addition to oligomer displacement, the reduced toxicity could also point towards an increased rate of conversion of oligomers to less toxic fibrils. The ability of a small molecule to reduce the toxicity of oligomeric species represents a potential therapeutic strategy against AD.

for raw NMR data and the definition of |∆δ|.       Analysis of the unseeded kinetic curves was carried out to probe the effects of trodusquemine on the product of rate constants involving primary pathways (k+kn) and secondary pathways (k+k2).
This product can be further decoupled by analyzing the seeded kinetic curves where specific individual microscopic processes in the aggregation process can be negated, e.g. primary nucleation at low and high concentrations of seed fibrils, or become dominant, e.g. elongation at high seed concentrations. We found from this approach that the unseeded aggregation traces in the presence of different concentrations of trodusquemine could be closely described by varying the rate constants associated with secondary pathways (Fig. 1b). Indeed, the effects on k+k2 are in excellent agreement with the macroscopic kinetic data, as this product of rate constants was observed to be increased with a clear dose dependence, indicating an enhancement by a factor of approximately 5 in the presence of an equimolar ratio of Ab42-to-trodusquemine (Fig. 1f).
Moreover, aggregation traces in the presence of 5% seed fibrils were well-described by fitting for k2 (Fig. 1c). By contrast, fitting the unseeded and seeded kinetic traces by modifying globally just the rate of primary nucleation resulted in much lower quality fits in comparison to fitting globally for k2 (Supplementary Figure 2). These results confirm the conclusion that the acceleration of aggregation observed in the presence of trodusquemine cannot be attributed to an increase in the rate of primary nucleation.
We then sought to distinguish between the relative contributions of secondary nucleation and elongation to the observed increase in k+k2. Experimental traces in the presence of a high concentration of seed fibrils reveal that trodusquemine causes only a small increase in the rate of elongation, as the half-time of the aggregation of Ab42 was reduced by a factor of 0.7 in the presence of an equimolar concentration of the molecule (Fig. 1d). Indeed, the aggregation traces for Ab42 in the presence of trodusquemine at ratios of 10:1 and 3:1 (Ab42-to-trodusquemine) were not significantly changed in the presence of 25% seed fibrils (Fig. 1d); however, in the presence of 5% seed fibrils, where secondary nucleation is rate-limiting, trodusquemine caused a discernable increase in the rate of aggregation (Fig. 1c). These results indicate that the alteration of the rate of elongation cannot by itself recapitulate the experimental data. As explained in the text, the lengths of the final fibrillar products measured by AFM are highly consistent with the conclusion that trodusquemine increases the rate of monomer-dependent secondary nucleation (see main text, Fig. 2b).

Supplementary Note 2: Consideration of fibril height (AFM) and width (TEM).
A close inspection of the height distribution obtained with Ab42 in the absence of trodusquemine reveals a heterogeneous distribution with at least two populations of fibrillar aggregates having smaller (2-4 nm) and larger (4-7 nm) heights (Fig. 2f). We note that the difference in the average cross-sectional height and width of the fibrils measured with AFM and TEM, respectively, is in part a result of their non-cylindrical cross-sectional symmetry, which has an aspect ratio > 1, as observed when, for instance, an amyloid fibril adopts a twisted ribbon conformation 5 . Moreover, the addition of uranyl acetate increases slightly the measured width of each fibril 6 , whereas the AFM technique causes a minor underestimation of fibril heights due to sample deformation 2,7 .
Despite these differences, both methods demonstrate that the presence of trodusquemine during the process of Ab42 fibril formation results in the production of fibrils with increased crosssectional dimensions.

Supplementary Note 3: TEM measurements at earlier stages of Ab42 aggregation.
To assess the effects of trodusquemine on earlier stages of Ab42 fibril formation, in an additional set of experiments, TEM samples were prepared as previously described and deposited at t=1h, a time when Ab42 exits the lag phase of aggregation in the absence of trodusquemine and reaches its half-time of aggregation when co-incubated with an equimolar concentration of Ab42-totrodusquemine. Fibrils formed in the presence of trodusquemine were characterized by the presence of branching aggregates, which suggests that the molecule potentiated nucleation at the fibril surface, a finding that was not observed for Ab42 after 1 h of aggregation in the absence of the molecule (Supplementary Figure 7). Indeed, such structures were also not observed in the plateau phase of the aggregation reaction for Ab42 in the absence of trodusquemine when measured by AFM and TEM (Fig. 2), indicating that the changes observed after 1 h of aggregation in its presence are not related to differences in the fibril mass fraction. We note, in particular, that this pattern of fibril formation in the presence of trodusquemine at early stages in the aggregation reaction is not consistent with an increase in the rates associated with primary nucleation or elongation.

Supplementary Note 4: Determination of the composition and concentration of oligomers necessary to observe toxicity in human neuroblastoma cells.
We first sought to determine the conditions in which Aβ42 oligomers are cytotoxic to human neuroblastoma cells in a reproducible manner. Oligomers were produced as previously described 3 using monomeric Aβ42 as a recombinant or synthetic (Sigma Aldrich, MO, USA) form and subsequently added to cell culture media at a concentration of 12 µM (monomer equivalents). The viability of neuroblastoma SH-SY5Y cells was found to be reduced by 32.5±3.1% and 37.2±4.2% when incubated with the oligomers generated from synthetic or recombinant Aβ42, respectively, as measured by the ability of the cells to reduce 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) (Supplementary Figure 8), indicating that the toxicity of the ADDLs is not significantly different when produced from these sources. Indeed, it has been shown that ADDLs generated from brain-derived and synthetic peptide have similar masses, isoelectric points, recognition by conformation-sensitive antibodies and binding to cultured primary neurons and toxicities 8 .
Following this observation, we elected to generate the oligomers from synthetic Aβ42. We then further tested the ability of the cells to reduce MTT in the presence of 1 µM Aβ42 oligomers in order to minimize the concentrations of trodusquemine administered to the cells, as the molecule alone can cause a decrease in cellular viability (Fig. 3a). At over an order of magnitude lower concentration of oligomers, a significant toxic effect of the aggregates was observed as the cellular viability was found to be decreased by 23.2±2.4% (Fig. 3a).
In an additional set of measurements, we sought to characterize in greater detail the heterogeneity within our Aβ42 oligomer preparation using dynamic light scattering. Oligomers were produced as previously described 3 and diluted to a concentration of 5 µM in 50 mM Tris buffer, 100 mM NaCl,