Environmental Conditions Affect the Kinetics of Nucleation of Amyloid Fibrils and Determine Their Morphology

Abstract

To understand and tackle amyloid-related diseases, it is crucial to investigate the factors that modulate amyloid formation of proteins. Our previous studies proved that the N47A mutant of the α-spectrin SH3 (Spc-SH3) domain forms amyloid fibrils quickly under mildly acidic conditions. Here, we analyze how experimental conditions influence the kinetics of assembly and the final morphology of the fibrils. Early formation of curly fibrils occurs after a considerable conformational change of the protein and the concomitant formation of small oligomers. These processes are strongly accelerated by an increase in salt concentration and temperature, and to a lesser extent by a reduction in pH. The rate-limiting step in these events has a high activation enthalpy, which is significantly reduced by an increase in NaCl concentration. At low-to-moderate NaCl concentrations, the curly fibrils convert to straight and twisted amyloid fibrils after long incubation times, but only in the presence of soluble species in the mixture, which suggests that the curly fibrils and the twisted amyloid fibrils are diverging assembly pathways. The results suggest that the influence of environmental variables on protein solvation is crucial in determining the nucleation kinetics, the pathway of assembly, and the final fibril morphology.

Introduction

Amyloid fibril formation by misfolded proteins is considered a hallmark of a range of human and animal diseases (1). Amyloid structures usually consist of extended, well-ordered fibrils typically formed by several filaments twisted in a rope-like manner (2). Although they show considerable similarity in their overall structural and morphological properties, amyloid fibrils can be remarkably diverse. This heterogeneity appears to be related to significant variation in the nanoscale structure of the fibrils, which is influenced by factors related to the environmental conditions that control their formation (3).

The kinetics of amyloid fibril formation generally shows a lag period followed by a rapid extension reaction, which is typical of a nucleation-growth mechanism (4). As in all nucleation-dependent processes, the lag phase can be shortened or even removed by seeding the protein sample with preformed aggregation nuclei or fibrillar species. The length of the lag phase can also be affected by the experimental conditions (pH, salt ions, temperature, etc.) or by mutations in the protein sequence (5), and in some cases certain proteins appear to fibrillate through entirely nonnucleated processes (6).

Of importance, different kinds of oligomeric species and small aggregates, which generally appear early during the lag phases of fibrillation, have been observed for many proteins, and during the last decade a great effort has been made to identify, isolate, and characterize them (7–10). Increasing evidence supports the notion that these oligomeric species are involved in the neurotoxic mechanisms of amyloid-related neurodegenerative diseases (11). Certain prefibrillar species, often called protofibrils, possess a high content of β-sheet structure and interact with Congo red and thioflavin T (ThT) (12), which suggests some degree of structural regularity. Some of these prefibrillar states appear to be on-pathway intermediates in the fibrillation process (13), whereas others can be off-pathway (8,14). Preceding the formation of protofibrils, smaller oligomeric states, generally soluble and with a low degree of structural order, appear during the early fibrillation stages. The assembly and structural reorganization of these species lead to the formation of the protofibrils (6,10,15). These oligomers are capable of nucleating the formation of amyloid fibrils efficiently, reducing the lag phase of aggregation.

The use of well-characterized model proteins and peptides, even unrelated to disease, is a potent approach for investigating amyloid fibrils, and in fact many of the advances that have been made regarding this intricate problem have been achieved with model systems. The folding mechanism and conformational stability of the SH3 domain of α-spectrin (Spc-SH3) have become subjects of intense investigation (16–18), which makes Spc-SH3 an ideal model protein for studying the determinants of amyloid formation. We have reported previously that the single-mutant N47A of Spc-SH3 rapidly forms amyloid fibrils under mildly acidic conditions (19). The amyloid-enhancing effect of the N47A mutation is not related to a thermodynamic destabilization of the native state, but to an increase in the rate of fibril nucleation (20). Here, we extended the study of the formation of amyloid fibrils by the Spc-SH3 N47A mutant under a wide range of salt concentrations, pH, and temperature using a variety of biophysical techniques that report on different aspects of the aggregation process. The results reveal that environmental conditions can dramatically influence the aggregation kinetics by modifying the accumulation of fibril precursors at the initial stages of the fibrillation process. Furthermore, we show that these effects are crucial in determining the final morphology of the fibrillar structures.

Materials and Methods

Protein samples

The N47A Spc-SH3 domain was purified as described elsewhere (17). For aggregation experiments, the lyophilized protein was dissolved (unless stated otherwise) in the appropriate buffer at 4°C, centrifuged for 2 min, and filtered through a 0.2 μm filter. Protein concentration was determined by measurement of absorbance at 280 nm using an extinction coefficient of 15220 M⁻¹·cm⁻¹.

ThT binding assay

ThT binding assays were performed as described elsewhere (21). A stock solution of 250 μM ThT was freshly prepared in 25 mM potassium phosphate (pH 6.0). Protein aliquots (10 μL) were diluted into the phosphate buffer containing 12.5 μM ThT and adjusted to a final volume of 1 mL. The ThT was excited at 440 nm with a 2.5 nm slit width, and the fluorescence emission was recorded at 485 nm with a 5 nm slit width. Measurements were carried out in a Perkin Elmer LS-55 spectrofluorimeter (Perkin Elmer, Shelton, CT) at 25°C using a 10 mm pathlength cuvette.

Transmission electron microscopy

For transmission electron microscopy (TEM), protein samples were diluted five times with buffer. The samples (15 μL) were then placed on a formvar-carbon-coated copper grid and allowed to stand for 4 min. The grid was then washed twice with distilled water and stained with 1% uranyl acetate for 1 min. The dried samples were evaluated in a Zeiss 902 electron microscope (Zeiss, Oberkochen, Germany) operating at an accelerating voltage of 80 kV and observed at a magnification of 50000×.

Circular dichroism

Circular dichroism (CD) experiments were performed on a Jasco J-715 spectropolarimeter (Jasco, Tokyo, Japan) equipped with a thermostatized cell holder. Measurements of the far-UV CD spectra (260–210 nm) were made with a 0.1 mm pathlength cuvette, using a bandwidth of 1 nm, a scan rate of 100 nm min⁻¹, a response time of 1 s, and an average of eight scans. For aggregation experiments, the CD signal at 215 nm was monitored as a function of time every 20 s at constant temperature.

Differential scanning calorimetry

Thermally induced aggregation experiments were performed with the use of a DASM4 differential scanning calorimeter. Differential scanning calorimetry (DSC) scans were conducted between 4°C and 110°C at a scan rate of 2°C·min⁻¹. Instrumental baselines, obtained by filling both calorimeter cells with buffer, were systematically subtracted from the thermograms of the sample, and the time response of the calorimeter was corrected. DSC experiments under equilibrium conditions were performed in a VP-DSC instrument (Microcal, Northampton, MA) at a scan rate of 1.5°C·min⁻¹ and a protein concentration of ≈1 mg mL⁻¹. The molar partial heat capacity curves (C_p) were calculated from the DSC data and analyzed using Origin 6.1 (OriginLab, Northampton, MA) according to the two-state unfolding model.

Dynamic light scattering

Dynamic light scattering (DLS) measurements were performed with a DynaPro MS-X instrument (Wyatt, Santa Barbara, CA) using a thermostatized 30 μL quartz cuvette. The protein solution and the buffer were centrifuged and filtered through a 0.02 μm filters before the measurements were obtained. Dynamics software was used in data collection and processing. DLS data were acquired every 45 s until saturation of the signal was achieved. The laser power was adjusted to avoid early saturation.

Results

Effect of salt concentration on the morphology of the amyloid fibrils of N47A Spc-SH3

We examined in detail the morphologies of the fibrillar aggregates produced at mild acid pH (0.1 M glycine buffer, pH 3.2) under different NaCl concentrations by TEM. Incubating a high-concentration protein sample (20 mg mL⁻¹) for 3 days at 37°C in the absence of NaCl produced fibrils with a curly and tangled appearance (19). TEM images with the time dependence of fibril assembly are shown in Fig. S1 of the Supporting Material. The fibrils had an approximate diameter of 6–7 nm as estimated from the images. Strikingly, after two months of incubation at 37°C, the curly fibrils had practically disappeared and only straight and twisted amyloid fibrils were visible in the samples (Fig. 1 a). These amyloid fibrils had considerable linear persistence and several morphological subtypes characterized by different twists of a number of coiled protofilaments, as has been observed for many other amyloids (2). The diameter of the protofilaments (6.9 ± 0.8 nm) is similar to that of the curly fibrils assembled at short incubation times.

Figure 1.

Influence of fibrillation conditions on the morphology of amyloid fibrils of N47A Spc-SH3. TEM images were taken with samples incubated in 0.1 M glycine buffer, pH 3.2, under the following conditions of NaCl concentration, temperature, sample concentration, and incubation time: (a) 0 M NaCl, 37°C, 20 mg mL⁻¹, 2 months; (b) 0.05 M NaCl, 37°C, 8 mg mL⁻¹, 10 days; (c) 0.1 M NaCl, 37°C, 8 mg mL⁻¹, 10 days; (d) 0.1 M NaCl, 37°C, 8 mg mL⁻¹, 1 month; (e) 0.2 M NaCl, 37°C, 8 mg mL⁻¹, 10 days; and (f) 0.1 M NaCl, 70°C, 8 mg mL⁻¹, 6 h. The black segments correspond to 200 nm.

In the presence of intermediate salt concentrations (0.05 M and 0.1 M NaCl), the aggregates obtained after short periods of incubation at 37°C also consisted of only curved fibrils (Fig. 1, b and c) with a 6–7 nm diameter (Table S1). After ∼10 days of incubation, a few straight and twisted amyloid fibrils appeared sporadically, coexisting with the thin curved filaments (Fig. 1 c). At long incubation times of >1 month, the twisted fibrils prevailed in the samples (Fig. 1 d).

Incubation of an 8 mg mL⁻¹ protein sample at high salt concentration (0.2 M NaCl) resulted in the formation within a few minutes of curved fibrils (Fig. 1 e and Fig. S1). The average diameter of these curly filaments was 4.5 ± 0.6 nm, which is significantly lower than that of the filaments formed at lower salt concentrations and suggests a different internal structure. Of importance, these curly fibrils persisted for >1 month of incubation under these conditions without any apparent change in morphology or diameter (Table S1). A dense mass of curly fibrils was also obtained by incubating an 8 mg mL⁻¹ protein solution at 70°C for 6 h in the presence of 0.1 M NaCl (Fig. 1 f).

These results indicate that the N47A Spc-SH3 domain can form amyloid fibrils of different morphologies depending on the environmental conditions and the length of incubation.

To investigate the ability of each type of curly filaments to transform into the mature amyloid fibrils, we prepared two 8.5 mg mL⁻¹ fresh protein samples, each in a buffer containing either 0.05 M NaCl or 0.2 M NaCl. The samples were incubated for 3 days at 37°C to obtain curly fibrils of 6–7 nm and 4–5 nm diameter, respectively. The fibrils were separated from the soluble material by centrifugation, and the fibrils prepared in the presence of 0.05 M NaCl were resuspended in 0.2 M NaCl and vice versa. In a similar experiment, curly fibrils of 6–7 nm diameter prepared in the presence of 0.05 M NaCl were centrifuged and resuspended in the same buffer. The final volume was adjusted to keep the original protein concentration in each sample. The fibril suspensions were then incubated for two months and analyzed by TEM at several time intervals (Fig. S2). Of interest, only curly fibrils were visible in all samples during 1 month of incubation, and only after 2 months did a few twisted amyloid fibrils appear very sparsely in some samples. These results indicate that in isolation, neither type of curly filaments can assemble directly into straight amyloid fibrils, irrespective of the buffer condition. Therefore, it appears that the soluble protein species are essential for morphological conversion, and that this process takes place by disaggregation of the curly fibrils and renucleation of amyloid fibrils under the appropriate conditions.

Effect of salt concentration on the kinetics of amyloid formation

To relate the morphological effects observed with possible changes in the early fibrillation kinetics, we carried out an extensive kinetic analysis of the fibril formation by the Spc-SH3 N47A mutant at pH 3.2 at different NaCl concentrations between 0 M and 0.3 M. Protein samples at a concentration of 8.3 mg mL⁻¹ were incubated at 37°C and several biophysical techniques were used to follow the aggregation process.

At time zero, the far-UV CD spectrum of N47A Spc-SH3 was identical to that of the native wild-type (WT) Spc-SH3 independently of the NaCl concentration. As aggregation progressed, the CD spectrum changed considerably and gradually developed a negative band at ≈215 nm, which is typical of β-sheet structure in the amyloid fibrils (Fig. S3).

The increase in NaCl concentration dramatically accelerated this conformational process (Fig. 2 a). At 0 M NaCl, fibrillation did not take place during a period of months, although it occurred within 3 days if the protein concentration was increased to 20 mg mL⁻¹ (not shown). At 0.05 M NaCl, there was a considerable lag phase of ∼200 min of incubation before the development of negative ellipticity. At higher salt concentrations of 0.1 M and 0.2 M, the lag phase disappeared and the curves displayed two distinct, well-defined phases. The kinetics could be fitted to a single-exponential decay superimposed on a linear dependence (Fig. 2 b). The fast exponential phase involved considerable changes in secondary structure and took place in ∼100 min at 0.1 M NaCl and in ∼20 min at 0.2 M NaCl. The apparent time constant (38 ± 9 min at 0.1 M NaCl) and the normalized amplitude of the fast phase were practically independent of the protein concentration (Fig. 2 b). On the other hand, the rate of formation of additional β-sheet structure during the slow phase was strongly dependent on the concentration of protein, consistent with a polymerization process. At 0.3 M NaCl, the protein underwent very fast amorphous aggregation as evidenced by TEM (not shown).

Figure 2.

Time dependence of fibrillation of N47A Spc-SH3 (8.3 mg mL⁻¹) at 37°C monitored by several techniques. (a) Far-UV CD signal at 215 nm at different NaCl concentrations. (b) Far-UV CD signal at 215 nm in the presence of 0.1 M NaCl at different sample concentrations. (c) ThT fluorescence at different NaCl concentrations. (d) Scattering intensity at different NaCl concentrations.

The kinetics of aggregation was also followed by ThT fluorescence (Fig. 2 c) and DLS (Fig. 2 d). In the presence of 0.05 M NaCl, the development of ThT fluorescence and scattering intensity showed lag phases of ∼150–200 min consistent with the changes observed by CD. At 0.1 M and 0.2 M NaCl, the lag times in the DLS and ThT fluorescence kinetics practically disappeared, indicating much higher rates of nucleation of fibrils.

A detailed analysis of the DLS data allowed us to follow the distribution of particle sizes during the early stages of the aggregation process (see Fig. 3 and Fig. S4). Before incubation, there were only particles with an apparent hydrodynamic radius (R_h) of ≈1.6 nm corresponding to the monomeric native protein. As aggregation progressed, larger particle radii appeared in the distributions. The series of events depended markedly on the NaCl concentration. Fig. 3 shows the time evolution of the R_h of the two smallest peaks in the distributions. At 0.05 M NaCl, most of the protein mass remained with native R_h for ∼200 min and then there was a slight expansion of the apparent R_h up to ∼2.5 nm. Simultaneously, after just ∼100 min, a few particles with an initial R_h of ∼10 nm appeared in the distributions and grew progressively up to 40–50 nm. These particles correspond to small elongating fibrils (19). Further incubation produced particles of up to several micrometers in the size distributions (Fig. S4 b). Similar events were observed at 0 M NaCl and 20 mg mL⁻¹ of protein concentration (Fig. S4 a and Fig. S5).

Figure 3.

Time evolution of particle sizes determined by DLS during fibrillation of N47A Spc-SH3 at 37°C in the presence of different NaCl concentrations. The average hydrodynamic radii of the two smallest peaks in the distributions are plotted as a function of the time of incubation. The NaCl concentrations are 0.05 M (blue), 0.1 M (red), and 0.2 M (black).

In the presence of 0.2 M NaCl, an expansion in R_h of the native particles up to ≈3 nm occurred within the dead time of the experiment. Small fibrils also appeared from the very beginning of the incubation and grew much faster than at low salt concentration. The observations made at 0.1 M NaCl were intermediate between those obtained at 0.05 M and 0.2 M, with an initial expansion in the R_h of the monomeric protein ending at ∼100 min, in good agreement with the end of the fast phase observed by CD (19).

Taken together, these results show that fibrillation of the N47A Spc-SH3 domain takes place in two stages. In the early stage, a fraction of the native protein undergoes extensive conformational changes involving β-sheet formation and the formation of small oligomers of 2.5–3 nm. These oligomers appear to be crucial for nucleation of the protofibrils, which further elongate in the second stage by incorporating additional protein monomers. The increase in NaCl concentration strongly accelerates the first stage, resulting in a dramatic enhancement of the overall fibrillation process.

Effect of temperature on the fibril formation kinetics

We also analyzed the kinetics of aggregation of 8 mg mL⁻¹ protein samples in the presence of 0.05 M, 0.1 M, and 0.2 M NaCl at different temperatures by CD at 215 nm (Fig. S6). The temperature interval for each salt concentration was chosen according to the velocity of the aggregation process. The lag phase observed in the presence of 0.05 M NaCl disappeared in the temperature interval selected, and the kinetic curves fitted well to a single exponential decay plus a linear phase. The rate constant of the fast phase increased strongly with temperature, indicating a high activation energy for the process. In contrast, the rate of the second phase did not change significantly with temperature, suggesting a downhill polymerization for the slow phase. The Arrhenius plots obtained for the first phase are shown in Fig. 4 a. The activation enthalpies were 181 ± 8 kJ mol⁻¹ at 0.05 M NaCl, 128 ± 14 kJ mol ⁻¹ at 0.1 M NaCl, and 90 ± 8 kJ mol⁻¹ at 0.2 M NaCl. These results indicate that an increase in NaCl concentration considerably lowers the number of interactions involved in the energy barrier of fibrillation.

Figure 4.

(a) Arrhenius plot for the rate of fibril nucleation at different NaCl concentrations. (b) Time evolution of the hydrodynamic radii of the smallest particles present during fibrillation in the presence of 0.1 M NaCl at different temperatures: 37°C (black), 42°C (red), and 55°C (blue).

The distributions of particle sizes followed by DLS during aggregation were also strongly affected by the temperature increase, as shown in Fig. 4 b. Higher temperature accelerated the formation of early oligomers, in good agreement with the increase in the rate of the conformational change observed by CD. Of importance, higher temperature favored the accumulation of larger oligomeric species. For instance, the apparent R_h grew up to 2.5 nm at 37°C, 3.0 nm at 42°C, and up to 7–8 nm at the beginning of the incubation at 55°C.

Effect of pH on the rate of fibrillation

The net charge of the polypeptide chain has been shown to considerably impact the fibrillation rates of polypeptides (22). Salt may act by weakening the repulsive electrostatic interactions via a Debye-Hückel screening effect or by direct ion binding (23), favoring oligomerization of partially unfolded species and accelerating the aggregation cascade. To test this hypothesis, we analyzed the effect of pH between pH 2.0 and pH 3.5 on the fibrillation kinetics at 37°C in the presence of 0.1 M NaCl. An increase in the net positive charge of the protein as the pH decreases would result in higher intermolecular repulsion and slower aggregation.

The rate of fibrillation observed by ThT fluorescence increased moderately with the pH reduction (Fig. S7), in marked contrast to the large effects exerted by salt. In addition, this effect was the opposite of that expected according to the initial hypothesis. At pH 3.5 there is some lag phase period that disappears at lower pH, suggesting a higher rate of nucleation. The development of the far-UV CD signal at 215 nm occurred in two phases as described above. A decrease in pH resulted in an earlier appearance of fibrillar aggregates as observed by TEM (Fig. S8), likely due to a faster formation of oligomeric precursors. At longer incubation times, the fibrils looked very similar at all pH values. The formation of small oligomers observed by DLS was accelerated by the reduction of pH (Fig. S9), but less dramatically than by increasing salt concentration or temperature.

Analysis of the thermodynamic stability of the native state

Our previous studies showed that the thermodynamic stability of the Spc-SH3 domain depends strongly on the pH (17); however, we did not analyze the effect of salt. Therefore, to characterize this effect, we followed the thermal unfolding of the N47A mutant of Spc-SH3 domain by DSC under fibrillation conditions but at low protein concentrations to avoid significant aggregation. We also analyzed the effect of pH at 0.1 M NaCl concentration. Under these conditions, all unfolding curves were highly reversible and followed the two-state unfolding model. Table S2 highlights the thermodynamic parameters for the thermal unfolding of the protein under the different conditions. Although the decrease in pH produced a strong destabilizing effect, as previously reported for the WT Spc-SH3, there was only a moderate destabilization produced by salt. A single linear correlation between ΔH and T_m was observed for all conditions (Fig. S10), which is consistent with a native structure that is practically unaffected by pH or NaCl concentration.

Energetics of thermally induced fibrillation and fibril melting probed by DSC

We used DSC to analyze the thermally induced fibrillation of native N47A Spc-SH3 at several NaCl concentrations and protein concentrations (Fig. 5 a and Fig. S11, a–c). Under conditions of fibrillation, the DSC thermograms presented a complicated shape with essentially three endothermic peaks. The first peak can be associated with a competition between the equilibrium thermal unfolding and the time-dependent fibrillation. The latter process is strongly favored by an increase in protein or NaCl concentration, which is reflected in a significant shift of the peak to lower temperature and in a considerable decrease in its area. At high protein concentration, fibrillation is the dominant process, and the area under this peak is an estimate of the overall enthalpy change of this process. This amounts to ∼50–60 kJ mol⁻¹ at 0.05 M and 0.1 M NaCl, and ∼45 kJ mol⁻¹ at 0.2 M NaCl. These values are much lower than those corresponding to the global unfolding of the protein (Table S2), indicating that the thermally induced protofibrils and fibrils are formed by partially unfolded protein molecules.

Figure 5.

DSC analysis of the fibrillation of N47A Spc-SH3. (a) Effect of sample concentration on the DSC thermograms measured with native protein in the presence of 0.05 M NaCl. Numbers alongside each curve indicate the sample concentration in mg·mL⁻¹. (b) DSC analysis of 8.3 mg.mL⁻¹ samples preincubated at 37°C for different times in 0.2 M NaCl. Incubation times are indicated alongside each curve. The scan rate was 2°C min⁻¹ in all experiments. The curves have been displaced artificially along the ordinate axis for clarity.

A second, smaller transition around 65–70°C at 0.05 M and 0.1 M NaCl, and the shoulder near 80°C at 0.2 M NaCl were attributed in our previous work to the melting of intermediate oligomers or protofibrils (19). The stability of these species is higher at 0.2 M NaCl, in good agreement with the effect of NaCl enhancing oligomerization as observed by DLS.

The third high-temperature peak corresponds to the thermal melting of the amyloid fibrils (24). The normalized area of this peak increased with the total protein concentration and the NaCl concentration, reflecting the formation of higher amounts of fibrils during the DSC scan. The T_m of the fibrils also increased with the NaCl concentration by >10°C, indicating a considerable stabilization of the fibrils by salt.

We also used DSC to analyze protein samples preincubated at 37°C for different time periods in the presence of several NaCl concentrations (Fig. 5 b and Fig. S11, d–f). A progressive reduction in area of the first transition with the time of preincubation reflects a preconversion of the native protein into oligomers or fibrils. This process is enhanced by the increase in NaCl concentration. Concomitantly, the high-temperature transition increases, in agreement with a more complete formation of amyloid fibrils. For samples preincubated for long time periods, in which fibrillation is essentially complete, the enthalpy of fibril melting is ∼100 kJ mol⁻¹ at a temperature of 80–90°C, which is significantly lower than the enthalpy change of unfolding of the native protein (≈270 kJ mol⁻¹) estimated at similar temperature.

Discussion

The partial unfolding of the native state and the formation of oligomers is critical for triggering fibrillation

Here we have shown that fibril nucleation of the N47A Spc-SH3 domain involves partial unfolding of the protein. When partially unfolded species accumulate sufficiently, they show a strong tendency to oligomerize and trigger the fibrillation process. In fact, under all conditions studied here, the end of the lag period was approximately coincident with the formation of small oligomeric species (2.5–8 nm) depending on the conditions. The increase in the rate of fibril growth after this event indicates that oligomer formation is a key factor in the mechanism of fibrillation. Fibrillation via soluble oligomeric species is a common theme in the mechanism of amyloid formation of many proteins. More than a decade ago, small metastable protofibrillar aggregates were described as transient intermediates in Aβ-40 and Aβ-42 fibrillation (7). Further work has shown that Aβ-40 and Aβ-42 form earlier oligomeric assemblies with a distribution of molecular sizes (25). The relative tendency of the two Aβ variants to oligomerize appears to play a crucial role in their different propensities for fibrillation. Transient oligomeric intermediates have also been detected during the lag phase of fibrillation of α-synuclein (26), mouse prion protein (27), and β2-microglobulin (28). The importance of oligomeric species is further supported by the observation that amyloid formation by a tandem repeat of the PI3-SH3 domain is strongly accelerated by the presence of preformed amorphous oligomers (10). These so-called molten oligomers have been proposed to facilitate the structural conversion of the protein into small amyloid-like structures that then act as templates for the fibril growth. Similarly, Serio et al. (13) noted that the prion protein Sup35 forms structurally fluid oligomeric species that nucleate amyloid fibrils. Of interest, these authors reported a remarkable independence of the rate of fibril nucleation with protein concentration, as we also observed for the initial conformational change of the N47A Spc-SH3 domain. This has been interpreted as indicating that a conformational conversion takes place within preformed oligomers or nuclei (13,29).

The effect of environmental factors on the kinetics of fibril formation appears to occur at the nucleation step

The kinetics of fibrillation of N47A Spc-SH3 at low salt concentration shows a prolonged lag phase, consistent with a nucleation and growth mechanism. Because the nucleation step is progressively accelerated by an increase in either the NaCl concentration or the temperature, the lag phase is completely abrogated when nucleation is not rate-limiting. Lag phase removal has also been observed for Aβ fibrillation in the presence of moderate concentrations of trifluoroethanol, which partially facilitates folded intermediates of fibril assembly (30). Barstar fibrillation can also occur from stable and well-populated soluble oligomers, with the kinetics showing no lag phase (31). Lowering the pH also increases the aggregation rate and decreases the lag phase, but this effect is much less intense than that of the salt or temperature. This is in marked contrast to the strong reduction in stability observed for this domain with a decrease in pH, as compared with the moderate destabilizing effect exerted by salt. These results indicate that the enhancement of fibril nucleation by environmental factors must have physicochemical purposes other than destabilization of the native state.

Salt may enhance desolvation of the protein by lowering the enthalpy barrier of formation of amyloidogenic species

A critical balance between electrostatic repulsion and hydrophobic interactions is required to trigger amyloid formation (23). This balance can be altered by such factors as the presence of ions and differences in pH, temperature, cosolvents, and pressure. The effect of salt ions has been most extensively studied. In general, it has been observed that an increase in salt concentration accelerates fibrillation, although the extent of this effect is strongly dependent on the aggregating protein and the type of salt. Salt ions may enhance intermolecular interactions by weakening repulsive electrostatic interactions via a Debye-Hückel screening effect or by direct ion binding (23). They may also perturb the hydration shell of the protein molecule (32). A combination of both direct protein-salt interactions and changes in water structure has also been proposed (33), and saline ions may even interact directly as structural ligands with fibrils, where they can coordinate charges and assist in the formation of new fibrils (34). However, in this study, screening of charge-charge repulsion did not appear to be crucial for triggering fibrillation, as evidenced by the aggregation enhancement that occurred with the increase in the net positive charge of the protein. It is thus likely that the role of salt in enhancing fibrillation is related to its influence on the hydration shell of the protein or to a direct ion interaction with the protein groups.

An important result of this work is that an increase in NaCl concentration considerably lowers the activation enthalpy of the conformational change accompanying nucleation. High activation energies ranging between 75 and 130 kJ mol⁻¹ have been reported for the fibrillation processes of Aβ-40 (35), insulin (36), mouse prion protein (27), and α-synuclein (37). In general, these high activation energies have been attributed to the complex molecular changes that accompany fibrillation. These values are of similar magnitude to those reported here, despite an expected variability due to the different protein systems, experimental conditions, and methods used in each study.

The existence of robust enthalpic activation barriers in protein folding and unfolding has been associated with the cooperative desolvation of the relatively large protein surface area that is needed to establish or break the required intramolecular interactions (38). The typical experimental activation enthalpies of folding-unfolding of small globular proteins fall in the range of 70–160 kJ mol⁻¹, comparable to those of amyloid formation. Since amyloid formation of natively folded proteins is usually accompanied by partial unfolding, it is likely that the activation enthalpies of both types of processes share a common physicochemical origin. We speculate here that the reduction in activation enthalpy produced by the increase in salt concentration may be related to its influence upon protein hydration. Salt ions may act by altering the structure and the cooperativity of the solvation layer of the protein, either by a direct binding to the protein surface that would change the effective protein surface or by interacting with the water of the first hydration layer (39). Support for each of these two mechanisms comes from the observation that the effects of different salt ions upon the amyloid formation rates of some proteins correlate well with electroselectivity series (33), whereas in other cases they correlate with the Hofmeister series (32). An alteration in the cooperativity of the protein hydration shell by salt may have dramatic consequences for the energy landscape of the protein, promoting alternative folding or unfolding pathways accessible to the polypeptide chain that may lead to amyloidogenic species.

The kinetics of nucleation determines the final morphology of the fibrils

It is important to note that salt concentration and incubation temperature can dramatically affect the final morphology of the fibrils formed by the N47A Spc-SH3 domain. Under conditions of slow nucleation, curly fibrils are initially formed, but well-ordered, straight, and twisted amyloid fibrils dominate the mixture at long incubation times. In contrast, high salt concentration and high temperature promote fast nucleation of curly filaments, which remain stable for long incubation times. These filaments appear thinner and shorter than those formed at low and intermediate salt concentrations, suggesting a different arrangement of protein monomers within their internal structure. Despite their very different morphologies, all of these types of fibrils have been shown to have an amyloid nature.

Fibril polymorphism has often been used to support a linear hierarchical mechanism of fibril assembly in which less-ordered fibrillar aggregates would be on-pathway kinetic intermediates of the long and twisted amyloid fibrils that constitute the most thermodynamically stable ones (40). Our observations do not support this mechanism, because this morphological change did not take place under any condition studied here when the fibrils were separated from the soluble species. This indicates that the soluble oligomeric species play a crucial role in the morphological conversion.

On the other hand, the coexistence of different fibril morphologies has also led to the alternative hypothesis that there are alternative competing pathways of assembly, which would be selected by the influence of environmental conditions upon the energy landscape of the protein. This has been established, for instance, for β2-microglobulin (8), in which assembly of different kinds of fibrils can occur by diverging routes depending on the conditions. Our results indicate that this type of mechanism also appears to occur for the N47A Spc-SH3 domain.

Biological implications

Our findings have important biological implications because it is increasingly evident that soluble oligomers or protofibrillar species, and not the final amyloid fibrils, are toxic to cells (11). The devastating nature of amyloid disorders has driven living organisms to protect themselves either by evolving protein sequences to minimize the formation of these highly toxic species or by developing protective mechanisms to clear out aggregating-prone species that may induce toxicity (41). Under some circumstances, however, these mechanisms fail because of ageing, oxidative stress, abnormal concentrations of metal ions, altered metabolic states, or other factors. One of these factors could be deregulation of the optimal protein hydration. Local or transient increases in the concentration of ions may reduce the kinetic barrier to the formation of cytotoxic oligomers within the crowded intra- or extracellular milieu, which could mediate additional metabolic alterations in neighboring cells, feeding back the process and creating a cascade of negative impacts. For instance, impaired water and ion homeostasis in the brain was recently related to spongiform encephalopathies (42) and Alzheimer disease (43). By a related mechanism, the catalytic effect of oligomerization of transition metal ions (e.g., Cu²⁺ and Zn²⁺ ions) has been described as the main factor in several amyloid disorders (44). Ions act by directly binding to the protein surface and strongly perturbing the hydration layer. In fact, binding of Cu²⁺ or Zn²⁺ to Aβ produces the release of a large number of water molecules, constituting a major thermodynamic factor for Aβ aggregation (45). The key role of protein hydration alterations is further supported by the observation that certain osmolytes have a protective effect against Aβ or insulin aggregation, which results when they are preferentially excluded from the protein surface, thermodynamically favoring its hydration and stabilizing the monomeric state (46).

Conclusions

We have shown here that salt concentration can strongly modulate the kinetics of the early conformational events that the protein must undergo to nucleate fibrillation. It is likely that these effects occur via an alteration of the protein solvation layer that dramatically modifies the protein conformational landscape and, as a consequence, alters the accessibility of amyloidogenic states. As a result, early oligomers of different sizes accumulate at diverse rates depending on the conditions, and effectively determine the final morphological and (possibly) structural properties of the protofibrils and amyloid fibrils. It is important to understand the physicochemical factors that govern protein hydration under aggregation-prone conditions because early oligomeric species are involved in the neurotoxic mechanisms of amyloid-related neurodegenerative diseases.