Distinguishing crystal-like amyloid fibrils and glass-like amorphous ...

17 downloads 0 Views 1MB Size Report
Sep 4, 2012 - glass-like amorphous aggregates formed rapidly without a lag phase. ..... we propose that ultrasonication is a powerful approach to breaking.
Distinguishing crystal-like amyloid fibrils and glass-like amorphous aggregates from their kinetics of formation Yuichi Yoshimuraa, Yuxi Lina, Hisashi Yagia, Young-Ho Leea, Hiroki Kitayamaa, Kazumasa Sakuraia, Masatomo Soa, Hirotsugu Ogib, Hironobu Naikic, and Yuji Gotoa,1 a Institute for Protein Research, Osaka University, 3-2 Yamadaoka, Suita, Osaka 565-0871, Japan; bGraduate School of Engineering Science, Osaka University, 1-3 Machikaneyama, Toyonaka, Osaka 560-8531, Japan; and cFaculty of Medical Sciences, University of Fukui, Fukui 910-1193, Japan

Edited by José N. Onuchic, Rice University, Houston, TX, and approved July 31, 2012 (received for review May 16, 2012)

Amyloid fibrils and amorphous aggregates are two types of aberrant aggregates associated with protein misfolding diseases. Although they differ in morphology, the two forms are often treated indiscriminately. β2 -microglobulin (β2m), a protein responsible for dialysis-related amyloidosis, forms amyloid fibrils or amorphous aggregates depending on the NaCl concentration at pH 2.5. We compared the kinetics of their formation, which was monitored by measuring thioflavin T fluorescence, light scattering, and 8-anilino1-naphthalenesulfonate fluorescence. Thioflavin T fluorescence specifically monitors amyloid fibrillation, whereas light scattering and 8-anilino-1-naphthalenesulfonate fluorescence monitor both amyloid fibrillation and amorphous aggregation. The amyloid fibrils formed via a nucleation-dependent mechanism in a supersaturated solution, analogous to crystallization. The lag phase of fibrillation was reduced upon agitation with stirring or ultrasonic irradiation, and disappeared by seeding with preformed fibrils. In contrast, the glass-like amorphous aggregates formed rapidly without a lag phase. Neither agitation nor seeding accelerated the amorphous aggregation. Thus, by monitoring the kinetics, we can distinguish between crystal-like amyloid fibrils and glass-like amorphous aggregates. Solubility and supersaturation will be key factors for further understanding the aberrant aggregation of proteins. protein aggregation ∣ metastability ∣ glass transition ∣ ultrasonication

M

isfolding of proteins leads to two major forms of aberrant aggregates: amyloid fibrils and amorphous aggregates. Amyloid fibrils, unique linear self-assemblies with an ordered cross-β structure, are associated with several diseases such as Alzheimer’s disease and type II diabetes (1, 2). Conversely, amorphous aggregates, including various types of assemblies without ordered intermolecular interactions, are common to most proteins. Amorphous aggregates form even for amyloidogenic proteins when they fail to form amyloid fibrils. Aggregates of α-crystallin in patients with cataracts are considered to be amorphous (3), indicating that amorphous aggregates are also associated with serious diseases (4). Various proteins which are not directly responsible for diseases can also form these two types of aggregates in vitro (5). Moreover, amorphous aggregates are frequently observed during the process of protein expression and purification. It is intriguing to address whether inclusion bodies are amorphous aggregates or amyloid fibrils, or a mixture of the two (6, 7). Recently, it has been suggested that the propensity for aggregation is correlated with the structural classification of proteins (8) and that the thermodynamic stability of the native states is closely linked with the kinetic barriers of aggregation (9). However, it is often difficult to distinguish between amyloid fibrils and amorphous aggregates even though the two are usually distinct in their morphologies (10), so they tend to be treated together in both experimental and theoretical studies. The mechanism of amyloid fibrillation has become increasingly clearer. It is coupled with the denaturation of proteins (11–14), and it is a nucleation-dependent reaction because addi14446–14451 ∣ PNAS ∣ September 4, 2012 ∣ vol. 109 ∣ no. 36

tion of seeds skips a lag phase of the kinetics. The nucleation is accelerated upon agitation of the solution by various methods including stirring and shaking (15, 16). Recent studies have focused on the accelerating effects of ultrasonic irradiation on amyloid fibrillation (17–21). These characteristics of amyloid fibrillation are similar to those of crystals (22, 23). The equilibrium and kinetic mechanisms of amyloid fibrillation have been explained on the basis of classical and atomistic nucleation theories elaborated for crystals (24). In other words, amyloid fibrillation, like crystallization, is a process whereby a supersaturated amyloidogenic protein or peptide adopts a water-excluded ordered assembly. Here, supersaturation refers to a metastable state in which the solution is kinetically kept stable although the concentration of dissolved solutes is larger than the thermodynamic solubility. In the metastable state, proteins hold soluble for an extended period because of a high free energy barrier. In contrast to amyloid fibrillation, we know little about the kinetics of amorphous aggregation. It is likely that amorphous aggregation also occurs coupled with the denaturation of proteins because unfolded conformations with exposed hydrophobic surfaces exhibit a greater propensity to aggregate. However, the role of nucleation in amorphous aggregation is unclear. In general, solutes exhibit a glass transition when the forces of exclusion from water are too strong to establish metastability. It might be useful to assume that amorphous aggregation is similar to the glass transition. In fact, theoretical studies of protein folding have advanced significantly by taking the glass transition into account, establishing a funnel view of protein folding with minimal frustration (25–27). In the case of protein misfolding leading to amyloid fibrils or amorphous aggregates, an analogy might be useful because both amyloid fibrils and amorphous aggregates form by intermolecular interactions. This paper aims to distinguish between amyloid fibrils and amorphous aggregates in their kinetics of formation. We used β2 -microglobulin (β2m), a protein responsible for dialysis-related amyloidosis (28). β2m forms both amyloid fibrils and amorphous aggregates at pH 2.5 depending on the NaCl concentration (29). The β2m solutions subjected to ultrasonication at a low concentration (≤500 mM) of NaCl efficiently produced amyloid fibrils after a lag period. In contrast, the solutions at a high concentration (>500 mM) of NaCl spontaneously formed amorphous aggregates without a lag phase, and no accelerating effect was produced by seeding or ultrasonication. The results suggest that Author contributions: Y.Y., H.Y., Y.-H.L., H.N., and Y.G. designed research; Y.Y., Y.L., H.Y., and H.K. performed research; Y.Y., Y.L., K.S., M.S. and H.O. analyzed data; and Y.Y., Y.-H.L. and Y.G. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1

To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/ doi:10.1073/pnas.1208228109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1208228109

a high free energy barrier of nucleation is determined by the ordered structures of amyloid fibrils and that amorphous aggregation occurs promiscuously without a high free energy barrier. Based on an analogy to the crystallization and glass transition of substances, we propose that solubility and supersaturation are key factors for further understanding the aberrant aggregation of proteins. Results NaCl-Dependent Formation of Amyloid Fibrils and Amorphous aggregates. We first investigated the amyloid fibrillation and amor-

Fig. 1. Formation of β2m amyloid fibrils and amorphous aggregates at 0 (gray), 100 (red), and 1000 (blue) mM NaCl monitored by measuring light scattering at 350 nm (A) and ThT fluorescence at 480 nm (B). The solution was subjected to agitation with a stirring magnet. The inset in panel A is a close-up view of the early time course at 1000 mM NaCl.

Yoshimura et al.

Fig. 2. Effects of the NaCl concentration on the kinetics and morphology of β2m aggregation. The β2m solutions were subjected to agitation with a stirring magnet at 600 rpm. (A, B) Dependencies on the NaCl concentration of the maximum ThT fluorescence at 480 nm (A) and the lag time of the increase in ThT fluorescence at 480 nm (B). The solid lines are drawn as an eye-guide. In panel B, the lag time at 0, 900, and 1000 mM NaCl was not able to be quantified because ThT fluorescence did not show a significant enhancement (see Fig. S2). (C–F) AFM images of the β2m aggregates at 100 (C), 300 (D), 500 (E), and 1000 (F) mM NaCl. The white bars represent 1 μm. (G–J) TEM images of the β2m aggregates at 100 (G), 300 (H), 500 (I), and 1000 (J) mM NaCl. The black bars represent 250 nm.

We then monitored amyloid fibrillation and amorphous aggregation using 8-anilino-1-naphthalenesulfonate (ANS) (Fig. S3). At 100 mM NaCl, ANS fluorescence showed a time course with a lag phase similar to that monitored using light scattering or ThT fluorescence under stirring. However, the ANS fluorescence increased markedly at 0 and 50 mM NaCl producing visible aggregates at the end of the reaction (Fig. S3A), which probably formed because of the strong electrostatic attraction between the positively charged β2m and the negatively charged ANS at pH 2.5 under the low ionic conditions (32). At higher NaCl concentrations (300, 500, and 1000 mM), the ANS fluorescence showed a saturating kinetics without a lag phase (Fig. S3B). The aggregated β2m at 100 mM NaCl showed an emission maximum of ANS fluorescence at 484 nm, while those at the higher NaCl concentrations showed blue-shifted emission maxima (from 474 to 478 nm) (Fig. S3E). Because amorphous aggregates have more hydrophobic surface exposed to solvent than amyloid fibrils, it is likely that the increase in ANS fluorescence at 100 mM NaCl with a lag phase predominantly represents the formation of fibrils and that the increase at higher NaCl concentrations without a lag phase represents amorphous aggregation. Moreover, it is possible that ANS promotes nonspecific hydrophobic interactions resulting in additional amorphous aggregation. Thus, ANS fluorescence might be useful to monitor amyloid fibrillation and amorphous aggregation although additional factors contributing to the fluorescence increase should be taken into account. We examined the products formed under various salt conditions with atomic force microscopy (AFM) (Fig. 2 C–F) and transmission electron microscopy (TEM) (Fig. 2 G–J). The AFM and TEM images showed that the β2m solutions at 100 and 300 mM NaCl formed rigid amyloid fibrils, whereas the solution PNAS ∣ September 4, 2012 ∣

vol. 109 ∣

no. 36 ∣

14447

BIOPHYSICS AND COMPUTATIONAL BIOLOGY

phous aggregation of β2m at 0, 100, and 1000 mM NaCl by monitoring light scattering at 350 nm and thioflavin T (ThT) fluorescence at 480 nm (Fig. S1). Both amyloid fibrils and amorphous aggregates were expected to show an increase in light scattering, but only amyloid fibrils an increase in ThT fluorescence (30, 31). The monomeric β2m solutions at a concentration of 0.1 mg∕ mL at pH 2.5 were subjected to agitation with a stirring magnet at 600 rpm. In the absence of NaCl, neither light scattering nor ThT fluorescence exhibited an increase in intensity, indicating that the acid-unfolded β2m remains monomeric because of the strong intramolecular and intermolecular electrostatic repulsions. Light scattering at 100 mM NaCl increased in intensity after a lag time of 1.5 h (Fig. 1A). ThT fluorescence at 100 mM NaCl also increased after a lag time of 1.3 h (Fig. 1B), consistent with the kinetics monitored by light scattering. It should be noted that, without stirring, the β2m remains monomeric at least for several hours at 100 mM NaCl (see below). In the presence of 1000 mM NaCl, light scattering significantly increased in a few minutes without a lag time, whereas ThT fluorescence did not show an enhancement, representing amorphous aggregation. This reaction was independent of agitation (see below). To evaluate the dependency of β2m aggregation on the NaCl concentration, we examined the reactions at various NaCl concentrations ranging from 0 to 1000 mM monitored using ThT fluorescence (Fig. S2). The intensity of ThT fluorescence showed a maximum at 300 mM NaCl, but decreased with any further increase in the NaCl concentration, resulting in a bell-shaped profile (Fig. 2A). This NaCl-dependent formation of β2m fibrils monitored with ThT was consistent with that reported previously (29), in which the seed-dependent formation of β2m fibrils was measured at various salt concentrations. Amyloid fibrillation at various NaCl concentrations showed a minimum of the lag time at around 300 mM, while an increase of the lag time was detected at higher NaCl concentrations (Fig. 2B). It is likely that the rapid formation of amorphous aggregates followed by slow reconfiguration to more stable amyloid fibrils occurs above 300 mM NaCl, contributing to the increase in the lag time above 300 mM.

at 500 mM NaCl produced a mixture of amyloid fibrils and amorphous aggregates with the former dominant. In the presence of 1000 mM NaCl, amorphous aggregates prevailed without amyloid fibrils. Thus, β2m formed amyloid fibrils and amorphous aggregates depending on the NaCl concentration. Dominant products were amyloid fibrils and amorphous aggregates at the NaCl concentrations below and above 500 mM, respectively. Effects of Different Types of Agitation. Amyloid fibrillation, a nucleation-dependent reaction, is accelerated by agitation of the solution. Stirring as employed in Fig. 1 is one conventional method of agitation. Ultrasonication is another, and more powerful for promoting fibrillation (17–21). We compared the effects of stirring and ultrasonication at 0, 100, and 1000 mM NaCl on the aggregation of β2m by monitoring light scattering at 350 nm (Fig. S1). The solution in the absence of NaCl did not show an increase in light scattering with stirring or ultrasonic pulses (Fig. S4), indicating it was undersaturated with monomers thermodynamically stable. At 100 mM NaCl, both stirring and ultrasonication produced amyloid fibrils, whereas no amyloid fibril formed under the quiescent conditions (Fig. 3A). The results indicate that the supersaturated solution is metastable in the absence of agitation, that is, kinetically stable, and that stirring and ultrasonication break the metastability. Ultrasonication accelerated the amyloid fibrillation more efficiently than stirring with a lag time of 0.5 h, versus 1.5 h for stirring. In contrast, in the presence of 1000 mM NaCl, amorphous aggregation monitored by light scattering occurred even without stirring of the solution (Fig. 3B, Fig. S4). The saturating kinetics was independent of stirring or ultrasonication, indicating that no high free energy barrier of nucleation exists in amorphous aggregation (SI Kinetic Barriers of Nucleation, Fig. S5). The effects of agitation (i.e., stirring or ultrasonication) were also monitored at various concentrations of NaCl by ThTand ANS fluorescence (Fig. 4). When monitored using ThT fluorescence, significant acceleration by ultrasonication in comparison with stirring was evident at 100, 300, and 500 mM NaCl (Fig. 4 A–C). Ultrasonication-dependent acceleration showed a more cooperative kinetics than stirring probably because secondary nucleation (i.e., ultrasonication-dependent fragmentation of the fibrils) is a frequent occurrence with ultrasonic irradiation. The stronger ThT fluorescence exhibited by the amyloid fibrils produced by ultrasonication than by stirring suggests that ultrasonication produces more ordered fibrils with stronger ThT binding. Alternatively, because the fibrils formed by ultrasonication are short and homogeneous, this may provide stronger ThT binding. At 1000 mM

Fig. 3. Effects of various forms of agitation on the growth of β2m amyloid fibrils at 100 (A) and 1000 (B) mM NaCl monitored by measuring light scattering at 350 nm. The solutions of β2m were agitated with a stirring magnet or irradiation with ultrasonication pulses. The effects of seeding were also examined. Light scattering intensity at 350 nm was monitored as a function of time. For comparison, the kinetics under quiescent conditions are also shown. When the effects of ultrasonic pulses or seeding were examined, the solutions were stirred. 14448 ∣

www.pnas.org/cgi/doi/10.1073/pnas.1208228109

NaCl, neither ultrasonication nor stirring increased ThT fluorescence (Fig. 4D), confirming the absence of amyloid fibrils. When monitored by ANS fluorescence, notable accelerating effects of ultrasonication were observed at 100 and 300 mM NaCl (Fig. 4 E and F). As described above, ANS itself is likely to promote additional amorphous aggregation, as revealed by the kinetics without a lag phase even at 300 mM NaCl. However, the marked and cooperative kinetics at 100 and 300 mM NaCl suggest that ultrasonication accelerated predominantly amyloid fibrillation. At 500 and 1000 mM NaCl, ANS fluorescence increased rapidly within a dead time of the measurements, independent of agitation, and subsequent minor change was observed (Fig. 4 G and H), revealing that the dominant products are amorphous aggregates. Seed-Dependent Growth of β2m Aggregates. Seeding is an important procedure for evaluating nucleation-dependent mechanisms (22). We investigated the ability of aggregates preformed under ultrasonication to act as seeds. First, the seeding experiments were performed at 100 and 1000 mM NaCl and monitored by measuring light scattering at 350 nm (Fig. 3). With the seeds added, the β2m solution at 100 mM NaCl showed an increase in light scattering without a lag phase (Fig. 3A), consistent with previous reports (33). Although the overall reaction was slower than that induced by ultrasonication, this is explained by the absence of active secondary nucleation promoted by ultrasonication. In contrast, seeding did not accelerate the amorphous aggregation at 1000 mM NaCl monitored using light scattering, as was the case of ultrasonication (Fig. 3B). The effect of seeding was measured at several salt concentrations with ThT and ANS fluorescence (Fig. 4). Seeding induced an increase in ThT fluorescence and a removal of the lag times at 100, 300, and 500 mM NaCl (Fig. 4 A–C). The cooperativity of the reaction was less than that of ultrasonication-induced fibrillation because of the absence of active secondary nucleation. When monitored using ANS fluorescence, seeding effects were evident at 100 and 300 mM NaCl, where amyloid fibrillation was accelerated (Fig. 4 E and F). In the presence of 500 and 1000 mM NaCl, where amorphous aggregates are the dominant product, seeding did not change the kinetics, again confirming that the free energy barriers of amorphous aggregation are low (Fig. 4 G and H).

Discussion Crystal-Like Amyloid Fibrils and Glass-Like Amorphous Aggregates.

We compared the amyloid fibrils and amorphous aggregates of β2m at pH 2.5 in their kinetics of formation. The fibrils formed at moderate concentrations of NaCl by a nucleation and growth mechanism in which a high free energy barrier associated with the nucleation rate-limits the overall reaction. In contrast, the amorphous aggregates formed rapidly at high concentrations of NaCl without a high free energy barrier, so that the reaction was independent of seeding or agitation of the solution. It might be possible that, even if the high energy barrier persists for amorphous aggregation, the overall aggregation occurs rapidly without discernible lag phase under the conditions where the growth rate is very fast. However, the simulation with Finke–Watzky mechanism, a two-parameter model for describing protein aggregation kinetics (34), argues that this is unlikely to occur (SI Kinetic Barriers of Nucleation, Fig. S5). It has been pointed out that amyloid fibrillation is similar to crystallization (22, 23). Both amyloid fibrils and crystals form by a nucleation and growth mechanism. Amyloid fibrils and crystals are also similar in that they grow immediately after the addition of seeds (i.e., preformed crystals or fibrils). The seeds work as a template for crystallization (35) and fibrillation (22), bypassing the high free energy barrier of nucleation. As established for crystallization, amyloid fibrillation continues until the solute Yoshimura et al.

concentration reaches its thermodynamic solubility (36). This thermodynamic solubility is sometimes called the critical concentration analogous with the critical micelle concentration at which water-excluded amphiphathic solutes form micelles. The amorphous aggregation observed here is similar to the glass transition. Lattice models suggest that amorphous aggregates of amyloidogenic proteins show a glassy behavior (37) in which the heterogeneous conformations are fixed by strong attractive forces producing various sites of interaction. In the case of native proteins, glass-like states mean amorphous aggregates as observed by the intense salting-out procedure with a high concentration of salt, such as ammonium sulfate and NaCl. Here, it has been suggested that domain swapping leads to amorphous aggregates followed by reconfiguration to more stable species (27). The suggested mechanism explains the rapid formation of amorphous aggregates followed by slow reconfiguration to more stable amyloid fibrils observed for various amyloidogenic fibrils. In fact, this reconfiguration is likely to occur for β2m at pH 2.5 under the transition regions where amyloid fibrils and amorphous aggregates coexist. Thus, distinguishing between amyloid fibrils and amorphous aggregates in their kinetics of formation is similar to distinguishing between the kinetics of crystallization and glass transition of substances. Because thermodynamic solubility and kinetic supersaturation are key factors in protein crystallization, they are also keys to understanding amyloid fibrillation and amorphous aggregation. Now, we would like to address the roles of solubility and supersaturation in amyloid fibrillation and amorphous aggregation. Crystals and glasses of native proteins. First, we illustrate the saltdependent phase transition of a folded protein (e.g. hen egg white lysozyme at 20 mg∕mL with a pH around 4.7) from a soluble state to a crystalline state and finally to a glass state (Fig. 5A). Here, we assume a salting-out experiment starting in the absence of salt, where the protein solution is undersaturated with a thermodynamic solubility higher than 100 mg∕mL (Region 1). The solubility decreases with an increase in the concentration of NaCl and becomes equal to the concentration of lysozyme at 0.36 M NaCl Yoshimura et al.

(38). However, even at a slightly higher NaCl concentration, the supersaturated solution is apparently soluble, i.e. metastable (Region 2), where spontaneous nucleation does not occur. In the presence of NaCl higher than 0.63 M, a labile region appears where spontaneous nucleation occurs and crystals form by a nucleation-growth mechanism (Region 3) (38). A further increase in the NaCl concentration produces a glass region where too many nuclei lead to amorphous aggregation (Region 4). The schematic representation of phase transitions shown in Fig. 5A is generally used to explain the phase transition of such substances as silica and polymers, where the abscissa is usually temperature. The competition between crystallization and glass transition has been elaborated for understanding the mechanism of protein folding (25–27), in which the folding temperature corresponds to the temperature of crystallization. To achieve cooperative and rapid protein folding, the temperature of glass transition should be much lower than the folding temperature so as to construct a smooth folding funnel with minimal frustration. Here, an analogy seems applicable to proteins self-assembling into amyloid fibrils and amorphous aggregates. Amyloid fibrils and amorphous aggregates of β2m. The distinct formation of amyloid fibrils and amorphous aggregates of β2m at pH 2.5 can be illustrated with a phase diagram dependent on the NaCl and protein concentrations (Fig. 5B). This type of diagram is often used for representing crystallization and amorphous precipitation depending on the concentrations of protein and precipitant (35). The phase diagram consists of a soluble region (Region 1), metastable region (Region 2), labile region (Region 3), and glass region (Region 4). At pH 2.5 in the absence of salt, β2m is largely unfolded and electrostatic repulsion among positive charges keeps the solubility relatively high. Thus, β2m in the acid-denatured state at 0.1 mg∕mL is undersaturated with monomers thermodynamically stable. Addition of NaCl at 100 mM shields the charge repulsion, resulting in a significant reduction of the solubility probably to less than 0.1 mg∕mL. This is accompanied by an anion-induced conformational transition often producing a compact molten-globule state (39). However, supersaturation keeps PNAS ∣ September 4, 2012 ∣

vol. 109 ∣

no. 36 ∣

14449

BIOPHYSICS AND COMPUTATIONAL BIOLOGY

Fig. 4. Effects of agitation on the growth of β2m amyloid fibrils monitored by measuring ThT fluorescence at 480 nm (A–D) and ANS fluorescence at 480 nm (E–H). The solutions of β2m were agitated with a stirring magnet or irradiation with ultrasonic pulses. The effects of seeding were also examined. The NaCl concentrations were 100 (A, E), 300 (B, F), 500 (C, G), and 1000 (D, H) mM NaCl. When the effects of ultrasonic pulses or seeding were examined, the solutions were stirred.

Fig. 5. Schematic diagrams for the phase transitions producing distinct aggregates. (A) NaCl-dependent conformational transition of a folded protein (e.g. hen egg lysozyme at pH 4.7). As a representative parameter, specific volume is plotted against NaCl concentration. (B) Phase diagrams of β2m at pH 2.5 depending on the NaCl and β2m concentrations. Agitation causes a downward shift of the metastability curve, which is indicated by an arrow. Filled circles indicate the experimental conditions employed in this study, illustrating the agitation-triggered fibrillation at 100 mM NaCl. In panels A and B, Regions 1, 2, 3, and 4 represent thermodynamically stable region with soluble monomers, metastable region without spontaneous nucleation, labile region with spontaneous nucleation, and glass region produced with too many nuclei, respectively. (C) General phase diagram of the conformational states of peptides and proteins dependent on conformational uniqueness and concentration. The representative conformational states are monomers, crystals, amyloid fibrils, and amorphous aggregates. In this phase diagram, amorphous aggregates of unfolded proteins and those of folded proteins are not distinguished. As for crystallization and amyloid fibrillation, supersaturation critically determines the phase transition as shown in panels A and B.

the unfolded β2m at 0.1 mg∕mL apparently soluble in the metastable region. In other words, the metastable state exists because of a high energy barrier of nucleation (40, 41). When seeds are introduced, fibrils form, reducing the monomer concentration to the solubility limit. Upon a further increase in the concentration of NaCl, the labile region, where nucleation occurs spontaneously producing fibrils, starts. The boundary between the metastable and labile regions is shifted downward upon agitation. In the case of crystallization, it has been suggested that ultrasonication decreases the energy barrier of nucleation by reducing the metastable region of the phase diagram (38). In the glass region, where the driving forces of spontaneous nucleation are too strong, too many nuclei result in glassy amorphous aggregates. In other words, interactions between unfolded β2m occur promiscuously and noncooperatively. As summarized above, the distinct formation of amyloid fibrils and amorphous aggregates can be defined by their locations in the phase diagram (Regions 1–4). The diagram indicates that supersaturation is a critical factor determining conformations under the conditions chosen. While solubility is a thermodynamic property, amyloid fibrillation and amorphous aggregation are kinetically controlled (42). Effects of Agitations. Agitation, including stirring or ultrasonication, is a kinetic factor modifying the apparent phase diagram (Fig. 5B). It is likely that the boundary between the metastable and labile regions is shifted downward upon agitation. Note that the protein concentration of 0.1 mg∕mL at 100 mM NaCl is located in the labile region in the presence of agitation, whereas it is located in the metastable region in the absence of agitation (Fig. 5B). A downward shift of the glass region may also occur, introducing transient and local glass regions. Although the real physical events responsible for amyloid nucleation are unclear, it is possible that the local and transient glass conformation provides effective nuclei for fibrillation. In the case of crystallization, it has been suggested that ultrasonication decreases the energy barrier of nucleation by reducing the metastable region of the phase diagram (38). Practically, one possible mechanism of ultrasonication-dependent nucleation is the formation of amorphous aggregates at the air-water interface. Ultrasonication produces cavitation microbubbles accompanied by local high pressure and high temperature 14450 ∣

www.pnas.org/cgi/doi/10.1073/pnas.1208228109

(43). It has been suggested that proteins condense and aggregate at the surface of the microbubbles, producing glassy amorphous aggregates. Amorphous aggregates are likely to contain seed-competent conformations. Thus, the role of ultrasonication might be to induce a transient and local glass state (Region 4) in the labile (Region3) or metastable (Region 2) region. These effects may also occur with other forms of agitation including stirring and shaking. General Phase Diagram for Aggregation of Proteins. Previously, we proposed a phase diagram for conformational and aggregational states of proteins determined by size and uniqueness of conformation (i.e., folded or unfolded) (44). When peptides are very short, they can form crystals above their solubility limit. For example, some short amyloidogenic peptides form microcrystals (45, 46). As peptides become longer, although still relatively short, they form amyloid fibrils above their solubility limit. When very long and unfolded, they form amorphous aggregates above their solubility limit. However, even extremely long polypeptides can form crystals when they fold into a unique conformation. Here, we consider that the size of peptides or proteins should be replaced by conformational uniqueness to accommodate the various conformational states of peptides and proteins (Fig. 5C). Native proteins with a unique conformation can form crystals independent of their size. If the conformation is highly flexible and various intermolecular interactions are possible, a glass state may form. Relatively short peptides and proteins in unfolded states can form amyloid fibrils above their solubility limit. However, one of the most important factors in determining amyloidogenicity is the role of supersaturation. A large number of previous reports as well as the present results argue that amyloid fibrillation starts under supersaturation. On the other hand, the kinetic barriers of glass transition are low compared to those of crystallization. The present results of rapid and saturating kinetics, as well as no effects of seeding or agitation on amorphous aggregation, suggest that the free energy barrier of nucleation is not high in comparison with that of amyloid fibrillation.

Conclusions We have summarized the key characteristics of aggregates formed by denatured proteins. First, amyloid fibrillation and amorphous aggregation are determined by the thermodynamic solubility of the respective peptides and proteins in water. This solubility Yoshimura et al.

1. Chiti F, Dobson CM (2006) Protein misfolding, functional amyloid, and human disease. Annu Rev Biochem 75:333–366. 2. Andersson A, et al. (2008) Amyloid deposition in transplanted human pancreatic islets: A conceivable cause of their long-term failure. Exp Diabetes Res 2008:562985. 3. Sugiyama M, et al. (2010) SAXS and SANS observations of abnormal aggregation of human α-crystallin. Chem Biodiversity 7:1380–1388. 4. Truscott RJ (2005) Age-related nuclear cataract-oxidation is the key. Exp Eye Res 80:709–725. 5. Dobson CM (2003) Protein folding and misfolding. Nature 426:884–890. 6. Wang L, Maji SK, Sawaya MR, Eisenberg D, Riek R (2008) Bacterial inclusion bodies contain amyloid-like structure. PLoS Biol 6:1791–1801. 7. de Groot NS, Sabate R, Ventura S (2009) Amyloids in bacterial inclusion bodies. Trends Biochem Sci 34:408–416. 8. Niwa T, et al. (2009) Bimodal protein solubility distribution revealed by an aggregation analysis of the entire ensemble of Escherichia coli proteins. Proc Natl Acad Sci USA 106:4201–4206. 9. Agostini F, Vendruscolo M, Tartaglia GG (2012) Sequence-based prediction of protein solubility. J Mol Biol 421:237–241. 10. Stranks SD, et al. (2009) Model for amorphous aggregation processes. Phys Rev E 80:051907. 11. Kelly JW (1996) Alternative conformations of amyloidogenic proteins govern their behavior. Curr Opin Struct Biol 6:11–17. 12. McParland VJ, et al. (2000) Partially unfolded states of β2 -microglobulin and amyloid formation in vitro. Biochemistry 39:8735–8746. 13. Khurana R, et al. (2001) Partially folded intermediates as critical precursors of light chain amyloid fibrils and amorphous aggregates. Biochemistry 40:3525–3535. 14. Uversky VN, Li J, Fink AL (2001) Evidence for a partially folded intermediate in α-synuclein fibril formation. J Biol Chem 276:10737–10744. 15. Sluzky V, Tamada JA, Klibanov AM, Langer R (1991) Kinetics of insulin aggregation in aqueous solutions upon agitation in the presence of hydrophobic surfaces. Proc Natl Acad Sci USA 88:9377–9381. 16. Giehm L, Otzen DE (2010) Strategies to increase the reproducibility of protein fibrillization in plate reader assays. Anal Biochem 400:270–281. 17. Stathopulos PB, et al. (2004) Sonication of proteins causes formation of aggregates that resemble amyloid. Protein Sci 13:3017–3027. 18. Ohhashi Y, Kihara M, Naiki H, Goto Y (2005) Ultrasonication-induced amyloid fibril formation of β2 -microglobulin. J Biol Chem 280:32843–32848. 19. Yamaguchi K, Matsumoto T, Kuwata K (2008) Critical region for amyloid fibril formation of mouse prion protein: Unusual amyloidogenic properties of the helix 2 peptide. Biochemistry 47:13242–13251. 20. Chatani E, et al. (2009) Ultrasonication-dependent production and breakdown lead to minimum-sized amyloid fibrils. Proc Natl Acad Sci USA 106:11119–11124. 21. So M, et al. (2011) Ultrasonication-dependent acceleration of amyloid fibril formation. J Mol Biol 412:568–577. 22. Jarrett JT, Lansbury PT, Jr (1993) Seeding “one-dimensional crystallization”of amyloid: A pathogenic mechanism in Alzheimer’s disease and scrapie? Cell 73:1055–1058. 23. Come JH, Fraser PE, Lansbury PT, Jr (1993) A kinetic model for amyloid formation in the prion diseases: Importance of seeding. Proc Natl Acad Sci USA 90:5959–5963. 24. Cabriolu R, Kashchiev D, Auer S (2010) Atomistic theory of amyloid fibril nucleation. J Chem Phys 133:225101.

Yoshimura et al.

Materials and Methods Self-Assemblies of β2m. Expression and purification of human β2m was described in SI Materials and Methods. Lyophilized β2m was dissolved in 3.2 mM HCl (pH 2.5) at a concentration of 1.0 mg∕mL, and then diluted 10-fold in a 1 cm cuvette with 3.2 mM HCl containing a series of NaCl concentrations. The volume of the solution in the cuvette was 2 mL. The solution was incubated at 25 °C. Methods of agitation with a stirring magnet and ultrasonic irradiation and seeding experiments are provided in SI Materials and Methods. Light Scattering, ThT Fluorescence, and ANS Fluorescence. Light scattering, ThT fluorescence, and ANS fluorescence were measured using a Hitachi fluorescence spectrophotometer F4500 with the excitation wavelengths at 350 nm, 445 nm, and 350 nm, respectively. The lag time for aggregation was defined as the time at which ThT fluorescence or light scattering reach 1/10th of the maximum. Fluorescence of ThT and ANS was monitored by adding ThT at 5 μM and ANS at 50 μM to the solution, respectively. Microscopic Images. AFM and TEM images were obtained using a Digital Instruments Nanoscope IIIa scanning microscope (Veeco) and a HITACHI H-7650 transmission microscope (Hitachi) respectively, as reported previously (20). ACKNOWLEDGMENTS. We thank Ms. Kyoko Kigawa for the expression and purification of β2m. This work was supported by the Japanese Ministry of Education, Culture, Sports, Science and Technology. 25. Bryngelson JD, Onuchic JN, Socci ND, Wolynes PG (1995) Funnels, pathways, and the energy landscape of protein folding: A synthesis. Proteins 21:167–195. 26. Nymeyer H, Garcia AE, Onuchic JN (1998) Folding funnels and frustration in off-lattice minimalist protein landscapes. Proc Natl Acad Sci USA 95:5921–5928. 27. Wolynes PG (2005) Energy landscapes and solved protein-folding problems. Philos Trans R Soc A 363:453–467. 28. Yamamoto S, Gejyo F (2005) Historical background and clinical treatment of dialysisrelated amyloidosis. Biochim Biophys Acta 1753:4–10. 29. Raman B, et al. (2005) Critical balance of electrostatic and hydrophobic interactions is required for β2 -microglobulin amyloid fibril growth and stability. Biochemistry 44:1288–1299. 30. Naiki H, Higuchi K, Hosokawa M, Takeda T (1989) Fluorometric determination of amyloid fibrils in vitro using the fluorescent dye, thioflavin T. Anal Biochem 177:244–249. 31. LeVine H, 3rd (1999) Quantification of β-sheet amyloid fibril structures with thioflavin T. Methods Enzymol 309:274–284. 32. Nielsen L, et al. (2001) Effect of environmental factors on the kinetics of insulin fibril formation: Elucidation of the molecular mechanism. Biochemistry 40:6036–6046. 33. Naiki H, Gejyo F (1999) Kinetic analysis of amyloid fibril formation. Methods Enzymol 309:305–318. 34. Morris AM, Watzky MA, Agar JN, Finke RG (2008) Fitting neurological protein aggregation kinetic data via a 2-step, minimal/“Ockham’s razor” model: The Finke–Watzky mechanism of nucleation followed by autocatalytic surface growth. Biochemistry 47:2413–2427. 35. Bergfors T (2003) Seeds to crystals. J Struct Biol 142:66–76. 36. Jarrett JT, Lansbury PT, Jr (1992) Amyloid fibril formation requires a chemically discriminating nucleation event: Studies of an amyloidogenic sequence from the bacterial protein OsmB. Biochemistry 31:12345–12352. 37. Dima RI, Thirumalai D (2002) Exploring protein aggregation and self-propagation using lattice models: Phase diagram and kinetics. Protein Sci 11:1036–1049. 38. Crespo R, Martins PM, Gales L, Rocha F, Damas AM (2010) Potential use of ultrasound to promote protein crystallization. J Appl Crystallogr 43:1419–1425. 39. Goto Y, Calciano LJ, Fink AL (1990) Acid-induced folding of proteins. Proc Natl Acad Sci USA 87:573–577. 40. Auer S, et al. (2007) Importance of metastable states in the free energy landscapes of polypeptide chains. Phys Rev Lett 99:178104. 41. De Simone A, et al. (2011) Experimental free energy surfaces reveal the mechanisms of maintenance of protein solubility. Proc Natl Acad Sci USA 108:21057–21062. 42. Jarrett JT, Berger EP, Lansbury PT, Jr (1993) The carboxy terminus of the β amyloid protein is critical for the seeding of amyloid formation: Implications for the pathogenesis of Alzheimer’s disease. Biochemistry 32:4693–4697. 43. Ogi H, Tomiyama Y, Shoji Y, Mizugaki T, Hirao M (2006) Effects of dissolved and ambient gases on sonochemical degradation of methylene blue in high-amplitude resonant mode. Jpn J Appl Phys 45:4678–4683. 44. Yanagi K, et al. (2011) Hexafluoroisopropanol induces amyloid fibrils of islet amyloid polypeptide by enhancing both hydrophobic and electrostatic interactions. J Biol Chem 286:23959–23966. 45. Nelson R, et al. (2005) Structure of the cross-β spine of amyloid-like fibrils. Nature 435:773–778. 46. Sawaya MR, et al. (2007) Atomic structures of amyloid cross-β spines reveal varied steric zippers. Nature 447:453–457.

PNAS ∣ September 4, 2012 ∣

vol. 109 ∣

no. 36 ∣

14451

BIOPHYSICS AND COMPUTATIONAL BIOLOGY

depends on the structure and length of the peptides and proteins. What seems missing in many previous studies is a recognition that amyloidogenicity is a property determined by the concentration of peptides or proteins relative to solubility; that is, critical concentration. Below the solubility limit, even a highly amyloidogenic peptide will remain dissolved. Above the solubility limit, peptides and polypeptides form crystals, amyloid fibrils, or amorphous aggregates depending on the stability and kinetics of the respective aggregated forms. Considering that both thermodynamic and kinetic factors are involved, distinct forms can coexist or apparent conformations can change with time, e.g. rapid formation of amorphous aggregates followed by the formation of amyloid fibrils. Another important characteristic is the impact of supersaturation, the exact mechanism of which is still unknown. In the metastable region, even strongly amyloidogenic peptides or unfolded proteins will remain soluble. However, the introduction of seeds or various types of agitation modify the phase diagram dramatically, releasing the kinetic trap and establishing the equilibrium determined by the intrinsic solubility of peptides or proteins. Finally, we propose that ultrasonication is a powerful approach to breaking metastability, possibly by introducing transient and local glass or labile states into the metastable region.