Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Dynamics of oligomer populations formed during the aggregation of Alzheimer’s Aβ42 peptide

An Author Correction to this article was published on 17 April 2020

This article has been updated

Abstract

Oligomeric species populated during the aggregation of the Aβ42 peptide have been identified as potent cytotoxins linked to Alzheimer’s disease, but the fundamental molecular pathways that control their dynamics have yet to be elucidated. By developing a general approach that combines theory, experiment and simulation, we reveal, in molecular detail, the mechanisms of Aβ42 oligomer dynamics during amyloid fibril formation. Even though all mature amyloid fibrils must originate as oligomers, we found that most Aβ42 oligomers dissociate into their monomeric precursors without forming new fibrils. Only a minority of oligomers converts into fibrillar structures. Moreover, the heterogeneous ensemble of oligomeric species interconverts on timescales comparable to those of aggregation. Our results identify fundamentally new steps that could be targeted by therapeutic interventions designed to combat protein misfolding diseases.

Main

Disorders of protein misfolding have emerged as some of the leading causes of death in the modern world1. Over 50 medical conditions associated with the misfolding and subsequent aggregation of proteins into amyloid fibrils and plaques have been identified, and include Alzheimer’s disease, Parkinson’s disease, Huntington’s disease and amyotrophic lateral sclerosis2,3,4,5. Specifically, Alzheimer’s disease is linked to the self-assembly of the Aβ42 peptide and other length variants derived from the amyloid precursor protein, which results in misfolded fibrillar aggregates that are observed as deposits in the brains of individuals who suffer from this progressive disorder6,7. Fibrils with high molecular weights were shown to be relatively inert, but aggregates with lower molecular weight, denoted oligomers, have emerged as potent cytotoxins with the ability to trigger neuronal death in cell and animal models8,9,10,11,12,13,14. New methods of kinetic analysis have transformed our understanding of the molecular events involved in the formation of the mature Aβ42 fibrils, and reveal in particular that the proliferation of Aβ42 aggregates occurs at the surfaces of existing fibrils through an autocatalytic process known as secondary nucleation15,16,17,18. Despite their fundamental importance, however, the molecular mechanisms that drive the dynamics of cytotoxic oligomers during amyloid fibril formation remain unknown. In the present study, we addressed this issue by providing direct measurements of the time evolution of oligomeric populations of Aβ42 formed during amyloid aggregation. These measurements, together with theory and simulations, define and quantify the fundamental molecular events of Aβ42 oligomer dynamics, and provides new insights into the secondary nucleation step in Aβ42 aggregation.

Results

We first obtained reproducible and quantitative measurements of Aβ42 oligomer concentrations formed during ongoing aggregation reactions that started from a supersaturated solution of 3H-labelled recombinant monomeric Aβ4219. We then used centrifugation to remove the fibrils, and utilized size-exclusion chromatography (SEC) and liquid scintillation counting to identify the resulting oligomer fraction (Fig. 1 and Supplementary Section 1.6). This isotope-based approach is rapid, highly sensitive and does not rely on any form of chemical labelling, and hence provides a method to study oligomer populations quantitatively without perturbing their aggregation behaviour20,21. 3H is used as a tracer that replaces 0.1% of the protons bound to carbon, whereas all the exchangeable and hydrogen-bonding positions contain 1H. As a validation of this approach, we obtained independent measurements of oligomer concentrations using mass spectrometry (MS) with a 15N isotope standard added after collecting the oligomer fractions (Fig. 1 and Supplementary Section 1.7). These measurements, shown in Supplementary Fig. 1, yielded closely similar kinetics compared to the data measured using 3H labelling. Moreover, MS is more sensitive than liquid scintillation counting, which offers the additional possibility to quantify the eluted fractions individually, and hence allows definition of the size distribution of the oligomer population (Supplementary Fig. 2).

Fig. 1: Experimental procedures for the quantitative measurement of Aβ42 oligomer populations during an ongoing amyloid fibril self-assembly reaction using 3H labelling or MS.
figure1

a, We incubated varying concentrations of Aβ42 or Aβ40 monomers and collected aliquots at the desired time points during the aggregation reaction. For each time point (Δt), we used centrifugation to remove the fibrils. We then isolated the oligomeric fraction, which encompassed species in the range of trimers to ~22-mers, through SEC. We used a Superdex 75 column with a void volume of ~7 ml and the monomer was eluted at 14–16 ml. b,c, After separation through SEC, we used liquid scintillation counting (b) or MS (c) to measure the oligomer concentrations. b, We used liquid scintillation counting to measure the absolute mass concentration of peptides that eluted between 7 and 13 ml in the case of 3H-labelled Aβ42. c, We used MS of natural abundance peptides, in which each fraction (1 ml) was lyophilized, redissolved in 20 μl of H2O, supplemented by a defined amount of 15N-Aβ42 (10 pmol) and AspN enzyme, digested overnight and analysed by matrix-assisted laser desorption/ionization–time of flight MS. The peptide concentration in each fraction was determined as the ratio r of the integrated area of the 14N peak at 1,906 m/z and the 15N peak at 1,928 m/z as c = r × 10 nM. The total oligomer concentration at each time point Δt was calculated as the sum over fractions 7–12. The relative Aβ concentration in each fraction was then calculated by dividing c by the summed concentrations over fractions 7–12. d, Observed concentration of oligomers versus aggregation time Δt. This procedure, which requires 10–16 min for oligomer isolation (Supplementary Section 1), provides a rapid and quantitative readout of the time evolution of oligomeric populations. a.u., arbitrary units.

These oligomer-population measurements were then combined with a quantitative kinetic analysis to develop a detailed mechanistic understanding of Aβ42 oligomer dynamics (Supplementary Sections 36). In general terms, the formation of oligomers requires two or more monomers to come together, either through a primary (that is, fibril-independent) or secondary (that is, fibril-dependent) nucleation mechanism. The population of oligomers can then decrease through (1) the conversion of non-fibrillar to fibrillar oligomers, (2) the elongation of fibrillar oligomers or (3) processes that do not lead to the oligomers being a source of new fibrils, such as dissociation into monomers. The population of each aggregate species can be parameterized mathematically through a master equation approach, similar to that developed for fibril formation15,16,17,18, and we exploited self-consistent approaches to develop integrated rate laws (Supplementary Sections 4 and 5) to be compared directly with the experimental data discussed above (Supplementary Section 6).

Using this framework, we first addressed the fundamental question of whether or not the oligomers formed during the aggregation process are elongation competent, that is, are able to sequester further monomers to grow in a manner similar to that established for mature amyloid fibrils16. We measured the time course of the oligomer populations formed from a solution of monomeric Aβ42 at an initial concentration of 5 μM and simulated a series of mechanistic scenarios to examine the quantitative level of agreement between the two. In the simplest scenario, all the oligomers are short fibrillar aggregates with the same rate of elongation as that of mature fibrils (Fig. 2a). In this scenario, any decrease in the oligomer population can only arise through the direct growth of the oligomers into species that are larger than those in the oligomeric fraction captured in the experiment. The mathematical analysis of the master equation in this limit (Supplementary Section 4.3) shows that the oligomer dynamics are described by the same underlying microscopic parameters as those that define the kinetics of growth of the higher-molecular weight fibrils, which are available from the analysis of fibril mass kinetics recorded at varying monomer concentrations (Fig. 2a, centre)16. The kinetics of oligomer formation can therefore be computed directly, without any further free parameters, from the analysis of the bulk amyloid aggregation data. Comparison of the model predictions with the experimental data on oligomer concentrations reveals, however, that the latter exceed the theoretical predictions by five orders of magnitude (Fig. 2a, right); the highest observed oligomer concentration over time was ~80 nM, whereas the model prediction was 280 fM, which reveals that the model outlined in Fig. 2a is unable to describe, even approximatively, the observed behaviour. We thus conclude that Aβ42 oligomers measured in the experiments are structurally distinct from fibrillar aggregates in that they are unable to recruit monomers to grow in size as effectively as mature fibrils can. Yet, all fibrils must originate from the growth of smaller oligomeric structures, which implies that at least some oligomers must undergo a structural conversion to become faster elongating fibrillar structures. It is likely that such a structural reorganization occurs during a conversion step to produce a fibrillar oligomer with very similar molecular packing to that observed in mature fibrils22,23. The conversion process may occur in solution or in contact with fibril surfaces24,25.

Fig. 2: Kinetic analysis of Aβ42 oligomer populations elucidates the molecular pathways of their dynamics during amyloid aggregation.
figure2

ac, Experimental measurements of the fibril formation at varying initial concentrations of Aβ42 (from Cohen et al.16) (centre column) and the time evolution of the concentration of oligomers recorded starting from 5 μM Aβ42 and best fits (solid lines) to the integrated rate laws that correspond to different mechanistic scenarios for Aβ42 oligomer dynamics (right column) (Supplementary Sections 35). a, One-step nucleation that produces elongation-competent oligomers. b, Two-step nucleation via oligomer conversion to growth-competent fibrils. c, Two-step nucleation via conversion of unstable oligomers. Supplementary Section 6 and Supplementary Table 1 give a detailed description of the fitting procedure and a list of fitting parameters. d,e, Experimental measurements of fibril and oligomer kinetics at 5 μM Aβ42 in the absence (d) and presence (e) of 5 μM Brichos chaperone domain from proSP-C to detect the presence of off-pathway oligomers, that is, oligomers that do not appreciably contribute to the reactive flux towards fibrils on the timescale of the experiment. Fibril mass measurements were fitted to the analytical expression for the aggregation time course (Supplementary Equation (25)) to determine how the overall rate constants for primary and secondary nucleation are affected by Brichos (Supplementary Section 6.4). The rate parameters determined in this way were then used to predict successfully the effect of Brichos on the oligomer concentration, without introducing any additional fitting parameters (Supplementary Equation (28)); this shows that suppressing oligomer formation on fibril surfaces affects equally the reactive fluxes towards oligomers and fibrils, which indicates that the majority of oligomers are on-pathway to fibrils.

We next sought to answer the questions of (1) how fast is the conversion rate of oligomers to fibrils and (2) what fraction of oligomers converts into fibrils or, by contrast, dissociates into monomers without giving rise to new fibrils. To address these questions, we first developed a kinetic model in which oligomeric species undergo a structural conversion before growing into mature fibrils, but cannot dissociate to monomers at a substantial rate over the timescale of the experiment (Fig. 2b and Supplementary Section 5). Although this model is consistent with the fibril mass data (Fig. 2b, centre), it cannot reproduce the observed oligomer population dynamics (Fig. 2b, right). Indeed, in this model the conversion rate controls both the maximum oligomer concentration and the rate of oligomer depletion after this maximum concentration; no value for the conversion rate can simultaneously capture the experimentally observed oligomer peak concentration and the timescale for oligomer depletion. To identify the missing element in our analysis, we introduced to our model an oligomer dissociation process that competes with the conversion into fibrils (Fig. 2c). The inclusion of such a dissociation step in the reaction network does not alter the quality of the fit of the fibril mass data (Fig. 2c, centre) but allows the experimental oligomer data to be fitted successfully in this manner (Fig. 2c, right).

Overall, our oligomer data suggest a two-step mechanism for fibril nucleation that involves oligomers as a necessary intermediate step in the formation of fibrils. This mechanism is analogous to the nucleation of crystals in solution, in which a liquid state serves as a precursor to the crystal phase26,27. In this two-step nucleation process, a conversion step from oligomers to fibrillar aggregates competes with dissociation of oligomers back to monomers (Fig. 2c). Of these steps, oligomer conversion is, on average, the slowest under the conditions probed in our experiments. The explicit estimates (Supplementary Table 1) for the (ensemble-averaged) rates of conversion (ρc 9 × 10−6 s−1) and dissociation (ρd 9 × 10−5 s−1) from this analysis make quantitative predictions, for example, for the average half-time of oligomers of about τ = ln(2)/(ρc + ρd)  120 min at 5 μM Aβ42. For Aβ42, we thus found a near absence of oligomeric species that are long-lived compared to the timescale of aggregation; this finding could be crucial for to determine the extent of spatial propagation of oligomers in living systems from their point of formation on amyloid deposits and plaques28. Equally importantly, the ratio of the conversion and dissociation rates gives the fraction of oligomers that convert successfully into fibril nuclei and eventually transform into mature amyloid fibrils. Strikingly, even though the oligomers are the key source of fibrils, we found that less than 10% of oligomers successfully converted into fibrillar species, whereas the remaining 90% of the oligomers dissociated back to the monomeric form.

The next fundamental question is whether the observed oligomers constitute on-pathway species in the process of assembly into fibrils, or are off-pathway structures unable to contribute directly to this process, for example, because they cannot convert into fibrils on the timescale of aggregation. To distinguish between the two possibilities, we introduced into the aggregation reaction solution 5 μM of the Brichos chaperone domain, which was previously shown to suppress secondary nucleation very selectively by binding to fibril surfaces29, and compared its effects both on the population of oligomers and on the reactive flux towards fibrils (Fig. 2d,e). If most oligomers are on-pathway, both processes will be affected equally, whereas if the majority of oligomers is off-pathway, the oligomer concentration and fibril formation rate may be affected in a different manner. The data in the presence of Brichos show that the reactive fluxes towards fibrils and oligomers are affected equally by Brichos, consistent with the majority of the measured oligomeric populations being on-pathway species. A key finding of this work is, therefore, that the population of oligomers is an ensemble of species that interconvert on timescales comparable to those of the overall aggregation, and hence the fraction of oligomers that converts from non-fibrillar forms to growth-competent fibrillar oligomers is determined by the average relative rates of their conversion and dissociation. We also note that in vivo some off-pathway oligomers may form from on-pathway ones by being stabilized through the interaction with other cellular components, for example, metabolites or other proteins.

We then varied the concentration of monomeric Aβ42 in the initial solution to test the ability of our model to capture the observed behaviour and to estimate the reaction orders of the various microscopic processes involved in determining the oligomer populations. These reaction orders coarse grain the monomer concentration dependence of these multistep processes, and contain key information about the rate-determining features of the free-energy landscape of amyloid nucleation (Supplementary Section 5.1). In particular, we found that the time courses of the oligomer concentrations recorded from solutions with initial monomer concentrations of 2.5 μM, 5 μM and 10 μM Aβ42 can be described by our model using the same choice of the rate parameters for all three datasets, and capture the dependence on the monomer concentration by means of the reaction orders of the different processes (Fig. 3a). We found that the rate of oligomer conversion displayed a marked dependence on monomer concentration with the reaction order nconv = 2.7, whereas the oligomer formation step showed a lower reaction order, noligo,2 = 0.9. These numbers yield an overall reaction order for two-step nucleation of γ = (nconv + noligo,2 + 1)/3  1.5, which is consistent with the value previously obtained from bulk kinetics16 (Fig. 3c, Supplementary Fig. 17, Supplementary Section 5.3 and Supplementary Equations (24) and (35)). Thus, compared to our previous work on secondary nucleation of Aβ4216, with direct measurements and analysis of oligomer populations we are now able to decompose the autocatalytic cycle of fibril self-replication into a series of elementary steps, which include oligomer formation, conversion, dissociation and growth, and to quantify the relative importance of each one of these steps (Supplementary Section 6.5). We found that the rate of oligomer depletion was approximatively independent of monomer concentration, whereas, in contrast, the fraction of successful oligomer conversion events increased with increasing total protein concentration (Fig. 3b). Overall, these findings suggest that oligomer dissociation is ‘spontaneous’, that is, independent of oligomer–monomer or oligomer–oligomer interactions, whereas oligomer conversion involves additional interactions with monomers. This conversion step may occur in solution or in contact with the fibril surface24. Indeed, the latter scenario might explain the high structural specificity of the process30.

Fig. 3: Concentration dependence of the molecular pathways of Aβ42 oligomer dynamics.
figure3

ac, Experimental measurements of the time evolution of oligomeric populations at varying concentrations of Aβ42 reveal the concentration dependence of oligomer conversion. a, Global fit of the experimental oligomer concentration data for 2.5, 5 and 10 μM Aβ42 to the integrated rate law that corresponds to the model shown in Fig. 2c. Shaded areas indicate 68% confidence bands (see Supplementary Section 6 and Supplementary Tables 1 and 2 for a list of the fitting parameters). b, Concentration dependence of the fractional contribution of unconverted oligomers towards the reactive flux to mature fibrils. Error bars indicate standard deviation. c, Extracted reaction orders for oligomer formation, oligomer conversion and the overall two-step secondary nucleation. dh, Computer simulation model of the Aβ42-aggregation-probe concentration dependence of oligomer conversion. d, Possible protein (left) and aggregate (right) states in the computer model. e, Mechanism of secondary nucleation in the computer simulations: monomers adsorb onto the fibril surface and detach as oligomers, which then convert into fibrils in solution at a later time. However, based on the analysis of our experimental data, we cannot exclude the possibility that structural conversion and dissociation of Aβ42 oligomers occur in contact with, or close to, the fibril surface. f, Rate of conversion of detached oligomers at varying monomer concentrations. g, The fraction of converted oligomers in the total oligomer population at three different monomer concentrations. h, Reaction orders for oligomer formation, oligomer conversion and the overall two-step secondary nucleation as measured in the simulations.

To provide a structural interpretation of these results, we performed computer simulations using a coarse-grained model of amyloid formation31, which enables us to calculate experimental observables, such as the reaction orders for oligomer formation and conversion, while retaining molecular-level resolution (Fig. 3d,e, Supplementary Section 2 and Supplementary Fig. 3). In this model, protein monomers are described as single rod-like particles that can interconvert between three states (Fig. 3d): (1) a monomeric state, (2) an oligomer-forming state and (3) a fibril-forming state. The monomeric state represents a disordered monomer in solution, which can adsorb onto the surface of a fibril. The oligomer-forming state represents an intermediate state; oligomers formed of particles in this intermediate state can detach from the parent fibril, but have not yet converted into a new fibril. Finally, the fibril-forming state represents a β-sheet-rich state with the ability to form strong lateral interactions. A protein species in its monomeric and oligomer-forming states interacts with particles of the same kind via its ends, which possess an attractive tip. A particle in the fibril-forming state interacts via its sides, which possess an attractive side-patch. This situation mimics directional interactions, such as hydrogen bonding, and drives the formation of fibrillar aggregates. The interaction between two proteins in the fibril-forming state is by far the strongest interaction in the system (Supplementary Fig. 3) and, once formed, the growth reaction is effectively irreversible. Every conversion from the monomeric to the fibril-forming state is slow and is thermodynamically penalized by a change in the excess chemical potential, which reflects the fact that amyloidogenic proteins and peptides, such as Aβ, are not typically found in a β-sheet conformation in solution32,33. Hence the fibril-forming state is energetically unfavourable. However, as particles in this state interact strongly with other particles of the same kind, the interplay of the two competing energy terms gives rise in the simulations to the nucleation barrier for fibril formation.

We observe in the simulations that oligomers produced via secondary nucleation persist for a substantial amount of time in solution before they convert into fibrils. Hence, both oligomer formation and conversion are slow steps in the reaction (Supplementary Section 2). Moreover, most of the oligomers dissociate back to monomers, and multiple oligomers typically form and dissociate before one successful conversion event into a fibril occurs (Supplementary Video 1). We can therefore conclude that the fibril surface serves as an oligomer breeding ‘factory’24. This results in the free-energy landscape sketched in Fig. 3e, which involves an initial oligomer-formation step followed by a large barrier for oligomer conversion. Note the composite nature of the reaction coordinate, which involves two slow degrees of freedom: aggregate size and structure (β-sheet content (see Supplementary Section 2.1 and Supplementary Fig. 4 for a discussion). The reaction rates and scaling exponents measured in the simulations emerge solely from the molecular ingredients and their interactions (Supplementary Section 2.1.3).

As observed in the experiments, oligomer conversion is markedly accelerated at higher monomer concentrations (Fig. 3f), with a high reaction order for oligomer conversion (nconv = 2.0) and a low reaction order for oligomer formation (noligo,2 = 1.2). These simulations not only reproduce the observed experimental behaviour, but also allow interpretation of the underlying molecular behaviour. For example, we found that larger oligomers have a lower free-energy barrier for conversion than that of smaller ones, which render the rate of conversion highly dependent on monomer concentration. At higher monomer concentrations, oligomers are not only more numerous but also larger (Supplementary Figs. 5 and 6), which gives rise to a faster effective conversion and, hence, to a faster overall fibril self-replication. We also found that oligomers can grow in solution, as the size of the oligomers detached from the fibril surfaces is smaller than the average size of converting oligomers (Supplementary Fig. 6). Figure 3g depicts how the fraction of converted oligomers changes with monomer concentration, and the average size of the converting oligomer is depicted in the Supplementary Information. According to our theory, the overall scaling exponent for two-step secondary nucleation can be predicted from the measured reaction orders as γ = (nconv + noligo,2 + 1)/3  1.4. This value is in excellent agreement with the value measured directly in the simulations, γ = 1.5 (Fig. 3h and Supplementary Equation (24)), as well as in the experiments16, which thus provides strong support for the mechanistic picture defined in this study.

Finally, we sought to understand how the results for the Aβ42 peptide are applicable to other systems, such as oligomer dynamics of the length variant Aβ40. Using MS (Supplementary Section 1.7), we measured the time evolution of the concentration of Aβ40 oligomers starting from a solution with a peptide concentration of 10 μM (Supplementary Fig. 8). We then analysed these data using our theoretical framework to determine the rate of oligomer conversion and dissociation for Aβ40 and compared them with those determined for Aβ42 oligomers (Supplementary Table 2). Interestingly, we found that Aβ40 oligomers have a similar rate of conversion (1 × 10−6 s−1) compared to that of Aβ42 oligomers at the same monomer concentration (6 × 10−5 s−1), but note that Aβ40 oligomers are somewhat larger than Aβ42 oligomers as they elute earlier from the SEC column (Supplementary Fig. 2). Moreover, from the rate of oligomer dissociation (1 × 10−4 s−1) about 1% of the Aβ40 oligomers successfully convert into fibrils, a similar result to that for Aβ42. Thus, for Aβ40 we arrive at the same conclusion as for Aβ42, that the vast majority of oligomers does not form fibrils, but rather dissociates back to monomers.

Discussion

We have described an experimental and theoretical approach to elucidate the fundamental molecular pathways that drive the dynamics of oligomers during an ongoing amyloid aggregation reaction. By applying this general approach to Aβ42, we found that, even though all mature amyloid fibrils must originate from oligomers, the majority of oligomers do not convert into fibrils but predominantly dissociate relatively rapidly into monomeric species before the slower conversion step takes place (Fig. 4). This type of mechanism, which fundamentally couples the accumulation of unconverted oligomers with fibril formation, reveals a non-classical nucleation process for Aβ42 amyloid fibrils. The formation of new fibrils occurs in two steps with oligomers as an obligatory intermediate state. The first step is the generation of oligomers through the interaction of monomers with the surface of existing fibrils. The resulting oligomers are a heterogeneous population of aggregates of different size (Supplementary Figs. 2 and 7), which are structurally distinct from fibrils and undergo structural conversions into fibrillar structures. Under the conditions probed in our experiments, the largest free-energy barrier associated with fibril nucleation is oligomer conversion. As intermediate oligomers are unstable species, slow conversion causes most of the oligomers to dissociate back into monomers without forming new fibrils. The free-energy barrier for oligomer conversion is lower at higher monomer concentrations, which explains the observation that the fraction of oligomers that dissociate without forming new fibrils is lower at higher monomer concentrations. The non-classical nucleation behaviour described here for Aβ42 fibril formation is analogous to the two-step nucleation processes observed in crystallization, biomineralization and sickle-cell haemoglobin26,27,34,35,36,37,38,39. Moreover, we established the absence, in our system, of detectable quantities of persistent off-pathway oligomers that cannot convert into fibrils over the timescale of aggregation, although such species may exist under different experimental conditions or for other, particularly larger, amyloidogenic proteins such as α-synuclein40. More generally, our work could be extended to study oligomer dynamics in peptide mixtures; in the presence of additional inhibitors41, these experiments could inform upon the role of off-pathway oligomers in such systems. The methods and results described in the present work are also likely to provide essential insights into the rational development of precise therapeutic strategies for targeting oligomers formed during pathological aggregation reactions.

Fig. 4: Schematic illustration of the reaction pathways of oligomers during amyloid aggregation and the associated reaction rates determined in this work for Aβ42.
figure4

Amyloid fibril proliferation occurs through a two-step nucleation mechanism that involves oligomer formation followed by oligomer conversion into fibrillar structures. The heterogeneous ensemble of oligomers not only includes converting species but also consists mainly of unstable oligomers that can dissociate back to monomers. Oligomers undergo repeated cycles of formation–dissociation before eventually converting into species that are able to grow into new fibrils. The reaction rates are shown here for Aβ42 at a concentration of 5 μM (rate constants are given in Supplementary Section 6.3) and are to be interpreted as averages over the heterogeneous ensemble of oligomers. The geometric mean of the rates of oligomer formation, oligomer conversion and fibril elongation (which constitute the autocatalytic cycle of fibril self-replication (Supplementary Section 5.3)) yields the characteristic rate of amyloid fibril formation (Supplementary Section 6.5 and Supplementary Fig. 17).

Methods

Details of the experimental materials and methods, mathematical modelling, data fitting and computer simulation model are available in the Supplementary Information.

Data availability

The authors confirm that all data generated and analysed during this study are included in this published article and its Supplementary Information. Data are also available from the corresponding authors upon request.

Code availability

All the simulation and data analysis codes are included in this article and its Supplementary Information. Codes are available from the corresponding authors upon request.

Change history

  • 17 April 2020

    An amendment to this paper has been published and can be accessed via a link at the top of the paper.

References

  1. 1.

    Alzheimer’s Association. 2014 Alzheimer’s disease facts and figures. Alzheimers Dement. 10, e47–e92 (2014).

  2. 2.

    Soto, C. Unfolding the role of protein misfolding in neurodegenerative diseases. Nat. Rev. Neurosci. 4, 49–60 (2003).

    CAS  Article  Google Scholar 

  3. 3.

    Dobson, C. M. Protein folding and misfolding. Nature 426, 884–890 (2003).

    CAS  Article  Google Scholar 

  4. 4.

    Chiti, F. & Dobson, C. M. Protein misfolding, functional amyloid, and human disease: a summary of progress over the last decade. Annu. Rev. Biochem. 86, 27 (2017).

    CAS  Article  Google Scholar 

  5. 5.

    Knowles, T. P. J., Vendruscolo, M. & Dobson, C. M. The amyloid state and its association with protein misfolding diseases. Nat. Rev. Mol. Cell Biol. 15, 384–396 (2014).

    CAS  Article  Google Scholar 

  6. 6.

    Selkoe, D. J. Folding proteins in fatal ways. Nature 426, 900–904 (2003).

    CAS  Article  Google Scholar 

  7. 7.

    Hardy, J. & Selkoe, D. J. The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to therapeutics. Science 297, 353–356 (2002).

    CAS  Article  Google Scholar 

  8. 8.

    Petkova, A. T. Self-propagating, molecular-level polymorphism in Alzheimer’s amyloid fibrils. Science 307, 262–265 (2005).

    CAS  Article  Google Scholar 

  9. 9.

    Campioni, S. et al. A causative link between the structure of aberrant protein oligomers and their toxicity. Nat. Chem. Biol. 6, 140–147 (2010).

    CAS  Article  Google Scholar 

  10. 10.

    Qiang, W., Yau, W.-M., Lu, J.-X., Collinge, J. & Tycko, R. Structural variation in amyloid-β fibrils from Alzheimer’s disease clinical subtypes. Nature 541, 217–221 (2017).

    CAS  Article  Google Scholar 

  11. 11.

    Cremades, N. et al. Direct observation of the interconversion of normal and toxic forms of α-synuclein. Cell 149, 1048–1059 (2012).

    CAS  Article  Google Scholar 

  12. 12.

    Benilova, I., Karran, E. & De Strooper, B. The toxic Aβ oligomer and Alzheimer’s disease: an emperor in need of clothes. Nat. Neurosci. 15, 349–357 (2012).

    CAS  Article  Google Scholar 

  13. 13.

    Catalano, S. M. et al. The role of amyloid-beta derived diffusible ligands (ADDLs) in Alzheimer’s disease. Curr. Top. Med. Chem. 6, 597–608 (2006).

    CAS  Article  Google Scholar 

  14. 14.

    Lashuel, H. A., Hartley, D., Petre, B. M., Walz, T. & Lansbury, P. T. Jr Neurodegenerative disease: amyloid pores from pathogenic mutations. Nature 418, 291 (2002).

    CAS  Article  Google Scholar 

  15. 15.

    Knowles, T. P. J. et al. An analytical solution to the kinetics of breakable filament assembly. Science 326, 1533–1537 (2009).

    CAS  Article  Google Scholar 

  16. 16.

    Cohen, S. I. A. et al. Proliferation of amyloid-β42 aggregates occurs through a secondary nucleation mechanism. Proc. Natl Acad. Sci. USA 110, 9758–9763 (2013).

    CAS  Article  Google Scholar 

  17. 17.

    Törnquist, M. et al. Secondary nucleation in amyloid formation. Chem. Commun. 54, 8667–8684 (2018).

    Article  Google Scholar 

  18. 18.

    Michaels, T. C. T. et al. Chemical kinetics for bridging molecular mechanisms and macroscopic measurements of amyloid fibril formation. Annu. Rev. Phys. Chem. 69, 273–298 (2018).

    CAS  Article  Google Scholar 

  19. 19.

    Walsh, D. M. et al. A facile method for expression and purification of the Alzheimer’s disease-associated amyloid β-peptide. FEBS J. 276, 1266–1281 (2009).

    CAS  Article  Google Scholar 

  20. 20.

    Szczepankiewicz, O. et al. N-terminal extensions retard Aβ42 fibril formation but allow cross-seeding and coaggregation with Aβ42. J. Am. Chem. Soc. 137, 14673–14685 (2015).

    CAS  Article  Google Scholar 

  21. 21.

    Nasir, I., Linse, S. & Cabaleiro-Lago, C. Fluorescent filter-trap assay for amyloid fibril formation kinetics in complex solutions. ACS Chem. Neurosci. 6, 1436–1444 (2015).

    CAS  Article  Google Scholar 

  22. 22.

    Colvin, M. T. et al. Atomic resolution structure of monomorphic Aβ-42 amyloid fibrils. J. Am. Chem. Soc. 138, 9663–9674 (2016).

    CAS  Article  Google Scholar 

  23. 23.

    Wälti, M. A. et al. Atomic-resolution structure of a disease-relevant Aβ(1–42) amyloid fibril. Proc. Natl Acad. Sci. USA 113, E4976–E4984 (2016).

    Article  Google Scholar 

  24. 24.

    Anwar, J., Khan, S. & Lindfors, L. Secondary crystal nucleation: nuclei breeding factory uncovered. Angew. Chem. Int. Ed. 54, 14681–14684 (2015).

    CAS  Article  Google Scholar 

  25. 25.

    Orgel, L. E. Prion replication and secondary nucleation. Chem. Biol. 3, 413–414 (1996).

    CAS  Article  Google Scholar 

  26. 26.

    Vekilov, P. G. The two-step mechanism of nucleation of crystals in solution. Nanoscale 2, 2346–2357 (2010).

    CAS  Article  Google Scholar 

  27. 27.

    Sear, R. P. The non-classical nucleation of crystals: microscopic mechanisms and applications to molecular crystals, ice and calcium carbonate. Int. Mater. Rev. 57, 328–356 (2012).

    CAS  Article  Google Scholar 

  28. 28.

    Koffie, R. et al. Oligomeric amyloid associates with postsynaptic densities and correlates with excitatory synapse loss near senile plaques. Proc. Natl Acad. Sci. USA 106, 4012–4017 (2009).

    CAS  Article  Google Scholar 

  29. 29.

    Cohen, S. I. A. et al. A molecular chaperone breaks the catalytic cycle that generates toxic Aβ oligomers. Nat. Struct. Mol. Biol. 22, 207–213 (2015).

    CAS  Article  Google Scholar 

  30. 30.

    Cukalevki, R. et al. The Aβ40 and Aβ42 peptides self-assemble into separate homomolecular fibrils in binary mixtures but cross-react during primary nucleation. Chem. Sci. 6, 4215–4233 (2015).

    Article  Google Scholar 

  31. 31.

    Šarić, A. et al. Physical determinants of the self-replication of protein fibrils. Nat. Phys. 12, 874–880 (2016).

    Article  Google Scholar 

  32. 32.

    Fändrich, M., Fletcher, M. A. & Dobson, C. M. Amyloid fibrils from muscle myoglobin. Nature 410, 165–166 (2001).

    Article  Google Scholar 

  33. 33.

    Allison, J. R., Varnai, P., C. M. Dobson, C. M. & Vendruscolo, M. Determination of the free energy landscape of alpha-synuclein using spin label nuclear magnetic resonance measurements. J. Am. Chem. Soc. 131, 18314–18326 (2009).

    CAS  Article  Google Scholar 

  34. 34.

    Šarić, A., Chebaro, Y. C., Knowles, T. P. J. & Frenkel, D. Crucial role of nonspecific interactions in amyloid nucleation. Proc. Natl Acad. Sci. USA 111, 17869–17874 (2014).

    Article  Google Scholar 

  35. 35.

    Šarić, A., Michaels, T. C. T., Zaccone, A., Knowles, T. P. J. & Frenkel, D. Kinetics of spontaneous filament nucleation via oligomers: insights from theory and simulation. J. Chem. Phys. 145, 211926 (2016).

    Article  Google Scholar 

  36. 36.

    Michaels, T. C. T. et al. Reaction rate theory for supramolecular kinetics: application to protein aggregation. Mol. Phys. 116, 3055–3065 (2018).

    CAS  Article  Google Scholar 

  37. 37.

    Galkin, O. et al. Two-step mechanism of homogeneous nucleation of sickle cell hemoglobin polymers. Biophys. J. 93, 902–913 (2007).

    CAS  Article  Google Scholar 

  38. 38.

    Auer, S., Dobson, C. M. & Vendruscolo, M. Characterization of the nucleation barriers for protein aggregation and amyloid formation. HFSP J. 1, 137–146 (2007).

    CAS  Article  Google Scholar 

  39. 39.

    Lee, C.-T. & Terentjev, E. M. Mechanisms and rates of nucleation of amyloid fibrils. J. Chem. Phys. 147, 105103 (2017).

    Article  Google Scholar 

  40. 40.

    Chen, S. W. et al. Structural characterization of toxic oligomers that are kinetically trapped during alpha-synuclein fibril formation. Proc. Natl Acad. Sci. USA 112, E1994–E2003 (2015).

    CAS  Article  Google Scholar 

  41. 41.

    Garai, K., Posey, A. E., Li, X., Buxbaum, J. N. & Pappu, R. V. Inhibition of amyloid beta fibril formation by monomeric human transthyretin. Protein Sci. 27, 1252–1261 (2018).

    CAS  Article  Google Scholar 

Download references

Acknowledgements

We acknowledge support from Peterhouse (T.C.T.M.), the Swiss National Science foundation (T.C.T.M.), the Royal Society (A.Š.), the Academy of Medical Sciences (A.Š.), the UCL Institute for the Physics of Living Systems (S.C.), Sidney Sussex College (G.M.), the Wellcome Trust (A.Š., M.V., C.M.D. and T.P.J.K.), the Schiff Foundation (A.J.D.), the Cambridge Centre for Misfolding Diseases (M.V., C.M.D. and T.P.J.K.), the BBSRC (C.M.D. and T.P.J.K.), the Frances and Augustus Newman Foundation (T.P.J.K.), the Swedish Research Council (S.L.) and the ERC grant MAMBA (S.L., agreement no. 340890). The research that led to these results received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013) through the ERC grant PhysProt (agreement no. 337969).

Author information

Affiliations

Authors

Contributions

All the authors were involved in the design of the study; T.C.T.M. developed the theoretical model and performed the kinetic analysis; S.L. and K.B. performed the experiments; A.Š. and S.C. performed computer simulations; all the authors participated in interpreting the results and writing the paper.

Corresponding authors

Correspondence to Michele Vendruscolo or Sara Linse or Tuomas P. J. Knowles.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Experimental methods, details on computer simulations, definition and solution of mathematical models of oligomer dynamics, details on data analysis, Supplementary Figs. 1–19, Tables 1 and 2, and Video 1.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Michaels, T.C.T., Šarić, A., Curk, S. et al. Dynamics of oligomer populations formed during the aggregation of Alzheimer’s Aβ42 peptide. Nat. Chem. 12, 445–451 (2020). https://doi.org/10.1038/s41557-020-0452-1

Download citation

Further reading

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing