In vitro fibrillization of Alzheimer’s amyloid-β peptide (1-42)

AD

Alzheimer’s disease

Aβ

amyloid-β

APP

amyloid precursor protein

HFIP

1,1,1,3,3,3-hexafluoro-2-propanol

TFE

2,2,2-trifluoroethanol

ThT

Thioflavin T

NP

nucleated polymerization

The assembly of peptides and proteins into fibrillar aggregates plays an essential role in the onset of several pathologies known as amyloid diseases¹ including Alzheimer’s disease (AD), Parkinson’s disease, type II diabetes as well as prion diseases. Identification of the key steps and understanding the mechanism of amyloid fibril formation on the molecular level may reveal important information for the development of methods and drugs for the suppression and reversal of amyloidogenesis.

In this paper we have studied the effects of multiple environmental factors on the aggregation of amyloid-β 1 − 42 (Aβ42) peptide, the main component of amyloid plaques, characteristic to AD. Aβ42 is more prone to aggregation than the most common form, Aβ40, and in many papers it has been considered to be responsible for amyloid plaque formation or at least for the initiation of the process. As a rule, fibrillization of amyloidogenic peptides and proteins in vitro is a self-propagating process characterized by a sigmoidal growth curve. The process starts with a relatively long lag-phase that is followed by the fast propagation of fibrillar material. According to the most popular model for Aβ fibrillization, the nucleated polymerization (NP) model,^2,3 the lag-phase is the time required for the formation of oligomeric nucleus large enough to be stable and grow due to the addition of monomers. However, recent studies have shown the importance of secondary nucleation events in the fibril growth process. For example the fragmentation of fibrils, can dominate in the propagation of amyloid growth.⁴ Moreover, several recent studies have shown that amyloid formation in vivo is also initiated by the formation of a limited number of seeds that spread from a single nucleation site and start the formation of plaques e.g. the process involves a secondary nucleation step.^5,6

Fibrillization is known to be enhanced by increasing temperature and affected by changes in pH, addition of organic solvents or other solutes as well as agitation of the reaction mixture. Recently, it has been shown that agitation enhances the fibrillization of Aβ42 peptide only in the initial exponential phase and this has been suggested to be related with the significant role of secondary nucleation.⁷ 

In order to elucidate the nature of the physico-chemical interactions involved in the fibrillization of Aβ peptides we have experimentally studied the effects of multiple environmental factors (temperature, pH, peptide concentration, ionic strength, and denaturing agents) on the kinetics of the Aβ42 fibril formation. The results suggest that the growth rate of Aβ42 fibrils is not under control of the diffusion of peptide molecules to the fibril end. The fibril growth is most likely limited by a monomolecular event that may be related to an “opening” of the conformation and exposition of hydrophobic surfaces. Destabilization of hydrophobic interactions by small amounts of solvents increased the fibrillization rate showing that hydrophobic interactions may stabilize the monomers in the solution. The decrease in the fibrillization rate at lower pH values is caused by the concurrent protonation of H13 and H14 residues that stabilizes the soluble monomer.

Lyophilized Aβ peptides (ultra-pure, recombinant) HFIP forms were purchased from rPeptide (Athens, USA). HEPES, Ultrapure, MB Grade was from USB Corporation (Cleveland, USA). 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP), 2,2,2-trifluoroethanol (TFE), and Thioflavin T (ThT) were from Sigma Aldrich (St. Louis, USA). NaCl (extra pure), ammonium acetate, and ammonium were from Scharlau (Barcelona, Spain). All solutions were prepared in fresh MilliQ water.

Stock solution of Aβ peptides was prepared as follows: 1 mg of the peptide was dissolved in HFIP at a concentration 500 μM to disassemble preformed aggregates.⁸ The solution was divided into aliquots, HFIP was evaporated in vacuum and the tubes with the peptide film were kept at −80° C until used. A 50 μg aliquot was assessed for amino acids in order to check the quality and quantity of the peptide. Before using the Aβ HFIP film was dissolved in water containing 0.02% NH₃ at a concentration of 10-20 μM. After 5 minutes incubation the Aβ stock solution was dissolved with buffer and used for experiments.

Fluorescence spectra were collected on a Perkin-Elmer LS-45 fluorescence spectrophotometer equipped with a magnetic stirrer. Fibrillization was monitored using ThT fluorescence. If not otherwise stated, fresh Aβ stock solution was diluted in 20 mM HEPES and 100 mM NaCl, pH 7.4 containing 3.3 μM of ThT to a final concentration of 5 μM. 400 μl of each sample was incubated at 40 °C if not otherwise stated. ThT fluorescence was measured at 480 nm using excitation at 445 nm. The pH dependence was determined using 20 mM ammonium acetate as a buffer.

The kinetics of Aβ fibrillization could be described as sigmoid curves and the aggregation parameters were determined by fitting the plot of fluorescence intensity versus time to Boltzmann curve

where A₁ is the initial fluorescence level, A₂ – corresponds to the fluorescence at maximal fibrillization level, t₀– is the time when fluorescence is reached half maximum and k – is the rate constant of the fibril elongation. The lag time is calculated as

An aliquot of 5 μl of sample was loaded on a Formvar-coated, carbon-stabilized copper grid (300 mesh from Ted Pella Inc., Redding CA). After 1 min, the excess solution was drained off using a Whatman filter paper. The grid was briefly washed and negatively stained with 5 μl of 2% uranyl acetate. The grid was air-dried and then viewed on a Tecnai G2 BioTwin transmission electron microscope (FEI, Japan) operating with an accelerator voltage of 80 kV. Typical magnifications ranged from 20,000 to 60,000×.

The fibrillization of Aβ was followed by an increase in ThT fluorescence due to the binding of the dye to the fibrils. The representative fibrillization curve on Fig. 1 demonstrates a good fit between the Boltzmann equation used for fitting and experimental data that is confirmed by plotting the residuals versus time in the inset. Parallel measurements at 40 °C with 5 μM Aβ42 showed good reproducibility of the maximal ThT fluorescence 322 ± 17 and the fibril growth rate (k = (1.35 ± 0.25) min⁻¹ (n=7)), however, the variation in the lag-period of the reaction was considerably larger (3.3 ± 2.0 min). The lower reproducibility of the lag period is most probably related with the stochastic nature of the formation of initial fibrils. The fibrillization of Aβ peptides is highly dependent on the agitation rate and conditions.⁹ We have used conditions where further increase in the agitation rate does not speed up the process any more.^7,10

A known drawback of the ThT method is the interference of added compounds or reaction conditions with ThT binding or its fluorescence,^11,12 thus, under certain circumstances the final ThT emission values under different conditions does not allow to compare the amount of fibrils formed. We have confirmed¹¹ that when the maximal ThT fluorescence intensity is achieved the solution does not contain monomeric Aβ42 peptide in detectable amounts. An interesting question is the mathematical model used for data fitting. Boltzmann equation describing a symmetric growth curve has been successfully used to describe the time curves of peptide fibrillization in several papers.^13–16 This equation also gave good descriptions for Aβ42 fibrillization and the parameters determined can be used in the analysis of various effects on the fibril formation and growth rates.

The kinetics of Aβ42 fibrillization was studied at peptide concentrations between 0.5 and 20 µM. The maximal level of ThT fluorescence showed nearly linear dependence from peptide concentration (Fig. 2), and the solutions did not contain Aβ monomers at the end of the process in detectable amounts.

The rate constant for fibril formation increased with an increase in the peptide concentration up to 3 µM, however, at peptide concentration above 3 µM its value (k = 0.52 ± 0.20 min⁻¹ (Fig. 2)) remained constant. In the fibril elongation phase the rate-constant corresponds to the disappearance of monomers¹⁷ and its independence of the peptide concentration means that this is a first-order (or most likely a pseudo-first order) process. It has been suggested that the rate constant can be independent of the peptide concentration when the equilibrium between the peptide fibrillization-competent and incompetent conformations in the solutions is shifted towards the latter¹⁸ e. g. the rate of fibril elongation is limited by obtaining an addition-competent structure in solutions (or any other monomolecular process) and the concentration of nucleation centers or growing fibrils is sufficiently high to bind all the peptide molecules in favorable conformation. The fibrillization rate of an Aβ42 derivative with a methionine residue in the N-terminus was also found to be concentration dependent at concentrations below 5.8 µM.¹⁵ However, in that study different agitation conditions were used and the overall process was also slower. The fibrillization models assuming prevalent primary nucleation predict an extremely sharp shortening of the lag phase with increasing peptide concentrations. Unfortunately, the lag phases showed relatively poor reproducibility in our experiments, however, the similarity of the curves obtained at low and high peptide concentrations shows that lag phase duration does not decrease with the peptide concentration in n^th power (where n is the minimal number of peptide molecules in a fibrillization-competent oligomeric nucleus). The secondary nucleation dominance with stochastic formation of early aggregates on the surfaces surrounding the solution is in agreement with the relative independence of the lag phase on the peptide concentration. Alternatively, the rate of fibril growth may become almost independent of peptide concentration at “supercritical concentration” when oligomer population may become significant with respect to the total peptide concentration, and when aggregates can only grow by monomer addition.¹⁹ However, this does not happen in our conditions since SEC analysis showed the presence of only monomeric peptide in the initial solutions up to peptide concentrations of 20 μM.

Aβ42 fibrillization was studied in the temperature range from 10 to 45 °C. The final level of ThT fluorescence was similar at all temperatures studied indicating that there is no drastic differences between the fibrils formed. The fibrillization rate constant increased and the lag time decreased with increasing temperature. The log k values showed a linear relationship in Arrhenius plot (Fig. 3) and the slope of the relationship corresponded to the activation energy 12 kcal/mole.

In general fibril formation is accelerated at higher temperatures.^20–22 Incubation of Aβ peptide at low temperature (4 °C) and physiological pH is known to lead to oligomerization, whereas fibrillization occurs at higher temperatures.²³ Our data confirm that higher temperatures favor fibrillization, moreover, the effect of temperature is characterized by a linear Arrhenius plot without temperature sensitive “switches” in the reaction mechanism. The ΔH^‡ value is in the same range with ΔH^‡ values determined for Aβ40 aggregation at acidic pH^17,22 and high peptide concentration. ΔH^‡ = 12 kcal/mole has been calculated from the rate of protofibril fibrillization at high peptide concentration,²⁰ a process that can be equivalent to fibril growth.

The relatively high ΔH^‡ value of the process clearly shows that fibrillization is not controlled by a diffusional event and must involve a conformational rearrangement. Thus, it can be concluded that the rate limiting step of fibril growth is a “lock” not a “dock” within the “dock and lock” model.^17,24 The incubation of Aβ40 at higher temperatures can result in reversible beta-sheet accumulation without aggregation,²⁵ thus, the high ΔH^‡ value can reflect the shift of the conformational equilibrium towards the more aggregation-competent peptide conformation. According to a recent model²⁶ the energy barrier of fibrillization does not correspond to the transition state but describes the recovery from off-pathway kinetic traps.

Effects of two fluorinated alcohols, 2,2,2-trifluoroethanol (TFE) and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP), on Aβ42 fibrillization were studied. Both TFE and HFIP produced a bell-shape rate constant dependence of concentration (Fig. 4), with the maximal rate constant values at 5 % and 0.1 % of the solvent, respectively. In conformity with their relative hydrophobicity the effective concentration of TFE is higher than that of HFIP for the same effect. TFE had no effect on the ThT emission of preformed Aβ fibrils showing, that it does not interfere the ThT fluorescence, however, the fibrils formed in the presence of high TFE concentrations showed lower ThT fluorescence.

Fluorinated alcohols at high concentrations (>40%) have been shown to break β-sheet structure, disrupt hydrophobic forces, and favor α-helical conformation of the peptide.^27–29 In the case of Aβ a considerably sharp transition of the peptide into the α-helix form occurs at TFE concentrations 15 to 25%.³⁰ We observed that under the conditions where α-helical structure becomes prevailing (at 20-25% TFE) fibrillization is almost completely suppressed. Thus, most likely the enhancement of fibrillization rate constant by TFE and HFIP arises from the destabilization of intramolecular hydrophobic interactions in the initial conformation. At higher concentration organic solvent stabilizes the conformation of the helical form, thus suppressing the formation of β-sheets. The effects of HFIP and TFE are in agreement with the effects of peptide concentration and temperature indicating that the fibrillization rate is limited by intramoleculecular reorganization in the peptide molecule, and they do not support the involvement of hypothetical α-helical intermediates in the fibrillization process.

The effect of pH on Aβ fibrillization was studied on the wt Aβ42 and three His to Ala mutants (H6A, H13A and H14A). Fig. 5 shows that the fibrillization rate constant of the wild type peptide was independent of the pH in the range of 7-9, whereas, at lower pH fibrillization rate decreased significantly, which can be caused by the protonation of histidines. The local environment of histidines changes considerably during aggregation as shown by upfield shifts of the histidine 2H NMR signals³¹ The lower fibrillization rate at lower pH where the histidines are protonated shows that it is not likely that they form salt bridges with aspartic acid residues in fibrils as suggested. In acidic solution the positively charged histidines can repel each other in structure of the forming fibril thus inhibiting the fibril formation. In order to find out which of the histidines are essential for Aβ fibrillization, the dependence of fibril elongation rate on the pH of the incubation medium was studied for three His to Ala mutants – H6A, H13A and H14A. The pH dependence of the fibrillization rate of H6A peptide was similar to that of the wild type Aβ suggesting that the protonation of H6 does not affect fibril formation. The fibrillization rates of H13A and H14A mutants were constant in the given pH range suggesting that only simultaneous protonation of both histidines H13 and H14 inhibits Aβ fibril formation.

Aβ40 is the major form of the peptide in cerebral spinal fluid, however, the senile plaques consist mainly of Aβ42,³² whereas in vitro studies show that they form mixed aggregates.³³ This phenomenon is difficult to explain considering that the peptides form interlaced amyloid fibrils in vitro.³⁴ Our results showed that in the agitated solutions both peptides demonstrated similar fibrillization kinetics, the lag period for Aβ42 was only twofold shorter and the fibril growth rate twofold faster than that of Aβ40. The kinetic parameters of the fibrillization process depend on the ratio of Aβ40/42 in the mixture. Adding 10 % Aβ40 to Aβ42 did not affect the lag period, however 50% of Aβ40 elongated the lag period almost to the level observed with pure Aβ40. Importantly, 10 and 50 % Aβ42 did not increase the fibrillization rate that started to increase only when the Aβ42 content was 90%. The fibrils of Aβ40, Aβ42, and their mixtures showed similar structures in transmission electron microscopy (Fig. 6). It has been demonstrated earlier in experiments in quiescent solutions that Aβ40 inhibits Aβ42 fibrillization.^35,36 Considering that Aβ40 but not Aβ42 fibrils grow slower in quiescent solutions⁷ the different behaviour of Aβ40/Aβ42 mixtures observed in different papers can be caused by different agitation conditions that is in agreement with the protective effect of Aβ40 in biological context.³⁷ 

Increasing concentrations of denaturating agents GdnCl (Fig. 7(a), 7(c)) and SDS (data not shown) decreased the fibrillization rate and also the final level of ThT fluorescence. As they also had a noticeable effect on the fluorescence of preformed fibrils, their effect on Aβ fibrillization at higher concentrations cannot be reliably studied by this method. However, both denaturating agents inhibited the fibrillization of Aβ42 peptide already at low concentrations.

In contrast urea had no effect on the final level of ThT fluorescence up to 6 M, thus, even 6 M does not shift the fibrillization equilibrium to higher free peptide concentrations. Below these concentrations urea had an inhibitory effect only on the fibrillization rate constant (Fig. 7). The inhibitory effect of urea on Aβ40 fibrillization has been demonstrated earlier at very high peptide concentrations³⁸ and therefore urea is not an efficient reagent for the prevention of fibrillization of peptide samples or solubilizing the fibrils.

In vitro fibrillization kinetics is studied in order to get valuable information about the characteristics of the process and to find clues for putative strategies to suppress the amyloid formation processes. Fibrillization consist of at least four phases where the dominating processes are (i) formation of initial fibrillization centers where fibrils start to grow (ii) their propagation; (iii) growth of fibrils and (iv) fibril maturation, respectively. The fibril growth corresponds to the in vivo situations where amyloid is already started to form in the brain of an AD patient. It seems that in this case the process can be suppressed by lowering the peptide concentration and stabilizing the peptide in the soluble state by the administration of compounds that form complexes with the peptide molecule. In this phase the in vitro process is a good model for the amyloid growth in vivo. The main factor determining the duration of the first phases is agitation that can multiply the number of growth centers, fibril ends, by rupture of the long fibrils. The initial stages of the reaction are difficult to study quantitatively since the variation in the lag periods is considerably high. Nevertheless, it is clear that the formation of the initial fibril is not completely stochastic. Hellstrand et al. studied the concentration dependence of Aβ fibrillization and concluded that Aβ fibril formation arises from a sequence of events in a highly predictable manner.¹⁵ Recently it was shown that the primary nucleation of α-synuclein fibrillization is enhanced more than three orders of magnitude by lipid bilayers.³⁹ Most likely the non-stochastic formation of primary seeds can occur on interfaces also in the case of in vitro studies of Aβ fibrillization. As a rule the fibrillization in vitro is initiated by preformed fibrillary seeds or by extremely high peptide concentrations where the solution contains nonfibrillar peptide aggregates that can transform to growing fibrils. Whether these mechanisms of formation of primary fibrils model of any processes in the brain is not known, the amyloid formation in the brain can also be triggered by other mechanisms such as for example metal induced Aβ aggregation.⁴⁰ 

Fibrillization of Aβ consist of at least four phases (i) formation of initial fibrillization centers; (ii) their propagation; (iii) growth of fibrils; and (iv) fibril maturation. The effects of environmental factors on the kinetics of Aβ fibrillization in the agitated solutions suggest that:

1. The relatively small effect of peptide concentration on the Aβ fibrillization is not in agreement with the model suggesting dominance of primary nucleation in the bulk solution.

2. The physiological 10% content of Aβ42 in the Aβ solution does not significantly enhance fibrillization. The fibrillization of Aβ42 in agitated solutions is only 2.5 times faster than that of Aβ40.

3. The effects of temperature and solvents show that the fibril growth rate in the stationary phase is limited by conformational rearrangement of the peptide molecule during the binding to the fibril. Due to the relatively high ΔH^‡ value the fibrillization is very slow at low temperatures.

4. The fibrillization rate is decreased at pH below 6.5 due to the simultaneous protonation of His13 and His14.

5. Guanidinium chloride is a strong and urea is a very weak inhibitor of Aβ fibrillization.

This work was supported by the Estonian Ministry of Education and Research (grant IUT 19-8) Estonian Science Foundation grants no. 8811 to PP and 9318 to VT. Authors are thankful to Jelena Beljantseva for determining the pH dependences.