Fiber-dependent amyloid formation as catalysis of an existing reaction pathway

Abstract

A central component of a number of degenerative diseases is the deposition of protein as amyloid fibers. Self-assembly of amyloid occurs by a nucleation-dependent mechanism that gives rise to a characteristic sigmoidal reaction profile. The abruptness of this transition is a variable characteristic of different proteins with implications to both chemical mechanism and the aggressiveness of disease. Because nucleation is defined as the rate-limiting step, we have sought to determine the nature of this step for a model system derived from islet amyloid polypeptide. We show that nucleation occurs by two pathways: a fiber-independent (primary) pathway and a fiber-dependent (secondary) pathway. We first show that the balance between primary and secondary contributions can be manipulated by an external interface. Specifically, in the presence of this interface, the primary mechanism dominates, whereas in its absence, the secondary mechanism dominates. Intriguingly, we determine that both the reaction order and the enthalpy of activation of the two nucleation processes are identical. We interrogate this coincidence by global analysis using a simplified model generally applicable to protein polymerization. A physically reasonable set of parameters can be found to satisfy the coincidence. We conclude that primary and secondary nucleation need not represent different processes for amyloid formation. Rather, they are alternative manifestations of the same, surface-catalyzed nucleation event.

Fig. 1.

Kinetics of iapp_20-29 fiber formation. (A) Representative kinetics of 700 μM de novo reactions conducted in the presence (circles, red) and absence of a CH₂Cl₂:aqueous interface. Reactions without CH₂Cl₂ were prepared by diluting an organic stock solution of iapp_20-29 into an aqueous buffer solution at 25°C. The latter is prepared by diluting a filtered, 10× buffer solution with water either obtained directly from a Milli-Q purification system (triangles, green) or additionally passed through a 0.2-μm syringe filter (squares, blue). The t₅₀ of these reactions are shown in Inset. (B) Kinetic profiles of reactions shown in A, with time axis renormalized to the t₅₀ of the representative reactions. For triangles and squares, data are overlaid with fits to a sigmoid. A simulated reaction profile for actin (magenta) is shown as representative of wholly polymer-independent nucleation. The first 250 sec of a CH₂Cl₂:aqueous interface-mediated reaction profile fits to the function at², where a is a constant (Inset, black line). (C) Measurement of reaction profiles at alternative protein concentrations. Shown are 2,000 μM (×) and 300 μM (+) reactions in the presence of a CH₂Cl₂:aqueous interface (Lower). Kinetics are also shown for 1 mM (squares) and 500 μM (circles) reactions without CH₂Cl₂ (Upper). After renormalizing time to the reaction t₅₀, the respective profiles overlay (Right).

Fig. 2.

Comparison of fibers formed in the presence and absence of a CH₂Cl₂:aqueous interface. (A and B) Negatively stained transmission electron micrographs of fibers formed at 1 mM in the absence (A) and presence (B) of a CH₂Cl₂:aqueous interface. (C) Representative x-ray diffraction patterns of unaligned fibers prepared in the presence (right) and absence (left) of the CH₂Cl₂:aqueous interface. Two reflections characteristic of β-sheet structure are evident: ≈4.7 Å, corresponding to the distance between peptides within the same β-sheet, and ≈8.5 Å, corresponding to the intersheet distance. (D) Fibers formed in the presence and absence of a CH₂Cl₂:aqueous interface seed fiber formation of IAPP_20-29 monomer with the same efficiency. Representative reactions showing 800 μM IAPP_20-29 de novo (squares) or seeded with 100 μM fibers formed in the presence (circles) and absence (triangles) of the interface. Lines shown are fits to sigmoidal and exponential equations, respectively.

Fig. 3.

Kinetic profiles measured by light scatter and soluble monomer concentration are closely similar. A representative 700 μMiapp_20-29 fiber formation was split and monitored simultaneously by NMR (triangles) and by 90° light scatter at 400 nm (squares). Successive 1D ¹H spectra were taken approximately every 10 min. The absolute monomer concentration was determined by comparing the area of the phenyl peak to the area of TMSP, which was added at a fixed concentration.

Fig. 4.

Fiber formation occurs during the lag phase. A representative 1 mM reaction (circles, orange) was diluted to 800 μM at 150 sec (inverted triangles, magenta) and 300 sec (triangles, blue) into the lag phase with kinetics subsequently monitored. The ratio of t₅₀ of diluted reactions to t₅₀ of reactions initiated at 800 μM (×) were determined (Inset). The gray fraction shows the contribution of controls in which reactions initiated at 800 μM were agitated at 150 sec and 300 sec, respectively.

Fig. 5.

Determination of the reaction order for nucleation and elongation processes. (A) Representative data showing the concentration dependence of de novo reactions conducted in the presence (circles, red) and absence (triangles, green) of the CH₂Cl₂:aqueous interface. Reaction orders were obtained by simultaneous fitting of 8 and 13 independent data sets, respectively, to a power law with a common exponent. (B) The quality of the global analyses are shown by plotting measured vs. fitted t₅₀ for all data sets. A line with a slope of one is shown in black. The reaction orders (Inset) are shown with errors corresponding to a 95% confidence interval. (C) Concentration dependence of fiber elongation. Reactions were conducted with concentrations of monomer ranging from 1 mM to 100 μM in the presence of 50 μM preformed fibers. Representative reaction profiles for two seeded kinetics containing 900 μM (triangles) and 400 μM (×) monomer are shown, respectively (Inset).

Fig. 6.

Temperature dependence of reactions conducted in the presence (circles, red) and absence (triangles, green) of the CH₂Cl₂:aqueous interface. Protein concentrations were 350 μM and 1 mM, respectively. Fits to determine activation enthalpies are shown (see text). Representative kinetics are shown (Inset).

Fig. 7.

Global analysis of fiber formation. (A) Eqs. 1 and 2 were used to conduct a simultaneous fit of the concentration dependence of reaction profiles in the presence (circles, red) and absence (triangles, green) of the CH₂Cl₂:aqueous interface. For clarity, all data and the best fit (m = 4, n = 4, black) are shown with time renormalized to the t₅₀ of an arbitrarily chosen data set. (B) χ² for the fits in A are shown for fixed values of m and n. (C) Global analysis was also performed on synthetic data created with m = 4, n = 2 by using an equivalent number of data points as in B. Contours for B and C are shown on a log₁₀ scale from lowest (blue) to highest (red) in increments of 0.33. (D and E) The rates of nucleation were calculated for a 700 μM profile by using the constants determined in A for aqueous (D) and CH₂Cl₂:aqueous (E) interface-mediated fiber formation, respectively. Loss of monomer A₁(t) (green, −CH₂Cl₂, or red, +CH₂Cl₂), the rate of fiber-independent nucleation (dE₁/dt, magenta), and the rate of fiber-dependent (dE₂/dt, black) nucleation are shown. The point in time where the rates of nucleation by these mechanisms are equal is indicated with an arrow.