Uncovering the universality of self-replication in protein aggregation and its link to disease

Abstract

Fibrillar protein aggregates are a hallmark of a range of human disorders, from prion diseases to dementias, but are also encountered in several functional contexts. Yet, the fundamental links between protein assembly mechanisms and their functional or pathological roles have remained elusive. Here, we analyze the aggregation kinetics of a large set of proteins that self-assemble by a nucleated-growth mechanism, from those associated with disease, over those whose aggregates fulfill functional roles in biology, to those that aggregate only under artificial conditions. We find that, essentially, all such systems, regardless of their biological role, are capable of self-replication. However, for aggregates that have evolved to fulfill a structural role, the rate of self-replication is too low to be significant on the biologically relevant time scale. By contrast, all disease-related proteins are able to self-replicate quickly compared to the time scale of the associated disease. Our findings establish the ubiquity of self-replication and point to its potential importance across aggregation-related disorders.

Fig. 1.. Effect of self-replication.

Illustration of the kinetic curves of aggregate concentration over time without (left) and with self-replication (right), along with a schematic of the reaction in both cases. When aggregation proceeds via nucleation and growth only, without self-replication, each primary nucleation event gives rise to only one fibril, and the aggregate concentration increases gradually. By contrast, when self-replication is present, here illustrated in the form of secondary nucleation, each primary nucleation event gives rise to many fibrils. The positive feedback loop of self-replication leads to exponential growth of aggregate mass and kinetic curves with a much more sudden increase and steeper transition.

Fig. 2.. Aggregation kinetics with and without self-replication.

(A to C) Illustration of the aggregation mechanism used in producing the fits. (D and G) Aggregation kinetics of CsgA, 5 μM monomer (D), and Aβ42, 2 μM monomer (G) (52), with increasing concentrations of preformed seeds, monitored by thioflavin T fluorescence, a reporter of the mass of aggregates formed. While the CsgA behavior is not significantly affected by the presence of seeds, a large effect can be seen for Aβ42 aggregation, even when the seed concentration is up to three orders of magnitude below that of the monomeric protein. (E, F, H, and I) Aggregation kinetics of CsgA (E and F) (53) and Aβ42 (H and I) (54) at a range of monomer concentrations in the absence of seeds. The solid lines are global fits of the integrated rate laws in the absence (E and H) and presence (F and I) of secondary processes. In (F), a significant contribution of secondary nucleation is enforced to illustrate the misfit. Data are recorded in triplicates at each concentration; all points are shown.

Fig. 3.. Rate diagrams of aggregation mechanisms show that self-replication is ubiquitous.

The rate at which new aggregates are produced by secondary pathways, κ, is plotted against the rate at which primary pathways produce new aggregates, λ. On the dashed line, the rates of the two processes are equal. It separates systems dominated by primary nucleation (bottom right corner) from systems dominated by a secondary process (top left corner). In primary nucleation-dominated systems, when secondary processes are too slow, only an upper bound for the rate of secondary pathways can be obtained, and similarly, when primary nucleation is so slow that seeding is required, only an upper bound for the primary rate can be obtained. These cases are here illustrated by elongated points. Proteins are split into three classes: those forming pathological amyloids (bottom left), those forming functional amyloids (bottom right), and those that do not generally form amyloid under physiological conditions (top right). Labels are shown above or to the right of the corresponding data point. Note: Aβ42 (V18S + A21S) is classed as a nonbiological amyloid because it is an artificial mutant of Aβ42, itself pathological, designed specifically to try to affect secondary nucleation [see Thacker et al. (28)]. Details on all proteins are given in the Supplementary Materials.

Fig. 4.. Time scales of self-replication and relevant in vivo time scale.

Bottom right half: The time scale of self-replication is longer than the relevant in vivo time scale, making self-replication unlikely to be able to contribute to the aggregation kinetics in vivo. Top left half: By contrast, the time scale of self-replication is much shorter than the relevant in vivo time scale, making a contribution of self-replication to the process likely. The time scales of self-replication were computed using the fitted rate constants and the protein concentrations in their respective in vivo environments, the latter being the main source of uncertainty. For two systems, prions in mice and tau in Alzheimer’s disease (AD), the self-replication rates have been determined in the in vivo system directly, requiring no further computation or knowledge of the in vivo concentrations (red). For FapC, CsgA, and actin, the experimental aggregation kinetics are described well with a mechanism that does not include self-replication; therefore, we can only obtain a lower limit on the time scale for those proteins. Dashed lines for PrP and tau denote mild shaking conditions (for details, see Materials and Methods).