Ultrasonication-dependent production and breakdown lead to minimum-sized amyloid fibrils

Abstract

Because of the insolubility and polymeric properties of amyloid fibrils, techniques used conventionally to analyze protein structure and dynamics have often been hampered. Ultrasonication can induce the monomeric solution of amyloidogenic proteins to form amyloid fibrils. However, ultrasonication can break down preformed fibrils into shorter fibrils. Here, combining these 2 opposing effects on β₂-microglobulin (β2-m), a protein responsible for dialysis-related amyloidosis, we present that ultrasonication pulses are useful for preparing monodispersed amyloid fibrils of minimal size with an average molecular weight of ≈1,660,000 (140-mer). The production of minimal and monodispersed fibrils is achieved by the free energy minimum under competition between fibril production and breakdown. The small homogeneous fibrils will be of use for characterizing the structure and dynamics of amyloid fibrils, advancing molecular understanding of amyloidosis.

Amyloid fibrils are supramolecular assemblies exhibiting a long unbranched fibrillar morphology ≈10 nanometers in diameter, the deposition of which is associated with >30 degenerative diseases including Alzheimer's disease, prion disease, and dialysis-related amyloidosis (1–3). The past decade has seen progress in our biophysical understanding of amyloid fibrils using various approaches including solution and solid state NMR (4–6) and X-ray crystallography (7, 8). However, the polymeric properties of amyloid fibrils, where huge size and heterogeneous nature result in insoluble and noncrystalline assemblies, are an obstacle to techniques such as X-ray crystallography, solution NMR spectroscopy and other conventional spectroscopic measurements. To overcome the analytical problems confronting studies of amyloid fibrils, it is worth establishing a strategy for producing monodispersed samples of amyloid fibrils. If amyloid fibrils of a well-defined molecular weight and improved solubility are formed reproducibly, a more general application of various types of fibril samples to a series of preexisting spectroscopic measurements will be accomplished.

As a strategy to produce amyloid fibrils of uniform and minimal size, ultrasonication has several potential applications. Although ultrasonication was originally used to prepare seeds from preformed fibrils (9), which were further applied to the amplification of infectious prion proteins (10, 11), ultrasonication-dependent fragmentation is becoming an important approach to analyzing the properties of fibrils (12, 13). However, its strong effect of agitation has recently been found to accelerate the fibril nucleation of several proteins and peptides (14–18). Interestingly, amyloid fibrils produced by ultrasonication-induced fibrillation were very short with apparently similar lengths as determined by AFM (15, 19), suggesting that short fibrils of homogeneous molecular size are formed efficiently under ultrasonication in combination with the opposing effects of fragmentation and of nucleation (Fig. 1).

Fig. 1.

Schematic illustration representing the small-size and monodispersed size distribution of amyloid fibrils under ultrasonication pulses. The schematic free energy diagrams in the absence (black) and presence of ultrasonication (red) are shown, representing that the ultrasonication pulses modulate the free energy landscape to form a minimum, producing minimum-sized and monodispersed fibrils.

In this study, by applying analytical ultracentrifugation, we have investigated the size distribution of amyloid fibrils of β₂-microglobulin (β2-m) formed spontaneously under pulses of ultrasonication. β2-m, a light chain of the type I major histocompatibility antigen, is the main component of the amyloid fibrils deposited in the synovia of the carpal tunnel of patients suffering from dialysis-related amyloidosis (9, 20). We will first demonstrate the uniformity of amyloid fibrils generated by ultrasonication based on the distribution of sedimentation coefficients determined by a sedimentation velocity analysis. We also measured the size distribution of the amyloid fibrils obtained by the ultrasonication of long fibrils to evaluate the effects of ultrasonication-induced fragmentation. Furthermore, weight-average molecular weight was determined by sedimentation equilibrium taking advantage of the small and uniform size of ultrasonicated fragments.

Results

Ultrasonication-Induced Spontaneous Fibrillation.

Previously (15), we found that repetitive ultrasonication pulses of 1 min with a quiescent incubation period of 9 min resulted in an overshoot of fibrillation after a lag time of several hours, followed by the breakdown of the preformed fibrils. The overshoot phenomenon suggested that by reducing the period of quiescent incubation, one can obtain fibrils of minimal size by escaping the overshoot, the most efficient approach for forming them. Here, the ultrasonication-induced fibrillation reaction was performed by applying repetitive pulses of a fixed period of 1 min with various quiescent incubation periods from 6 sec to 9 min. As for the NaCl concentration of the protein solutions, 38 mM, lower than previously, was selected to prevent fibrils from associating during the sedimentation analysis (21) (see Figs. S1 and S2).

In all pulse cycles used, an abrupt increase in ThT fluorescence was observed after a lag time (Fig. 2 A–D), consistent with ultrasonication-induced spontaneous fibrillation (15). When the fibrils sampled after the 1-day incubation period (arrows 2 in Fig. 2 A–D) were subjected to a sedimentation velocity analysis, all samples exhibited considerably simple boundary profiles with stable plateaus, indicating a sharp distribution of s_20W values at ≈36, 29, 27, and 17 S for quiescent periods of 9 min, 4 min, 2 min, and 6 sec, respectively (Fig. 2 E–H, closed circles in Fig. 2 I–L).

Fig. 2.

Size distribution of β2-m amyloid fibrils formed by ultrasonication-induced spontaneous fibrillation. Monomeric β2-m was subjected to the periodic 1-min ultrasonication followed by a period of silence without sonication lasting 9 min (A, E, and I), 4 min (B, F, and J), 2 min (C, G, and K), or 6 sec (D, H, and L). (A–D) Kinetics of ultrasonication-induced fibrillation monitored using ThT fluorescence. The results of 3 independent experiments (filled circles, triangles, and gray squares) are shown. The products of the triangle and filled circle plots were subjected to a sedimentation velocity analysis at 2 different time points, labeled 1 and 2, respectively. (E–H) Sedimentation boundary profiles of fibrils sampled after the continued ultrasonication (filled circles in A–D). The data were recorded at 20,000 rpm (30,900 × g) and 5 °C by monitoring the absorbance at 280 nm, in which several traces at intervals of 10 min (E–G) or 8 min (H) were presented. (I–L) Integral distribution plots (s_20W) of ultrasonication-generated fibril products sampled immediately after the abrupt increase of ThT fluorescence (triangles) and after the continued ultrasonication (filled circles).

As expected, the overshoot became less pronounced as the pulse interval became shorter (Fig. 2 I–L), and with a 6-sec quiescent period, the size of amyloid fibrils reached equilibrium almost directly accompanied by only a slight decrease in average size during the consecutive ultrasonication after the explosive phase, in accordance with negligible changes in ThT fluorescence intensity (closed circles in Fig. 2 D and L). Furthermore, smaller fibrils were produced as the period became shorter, which suggests that the converged size of amyloid fibrils is regulated by the frequency of ultrasonication pulses applied during the fibrillation process. The fibril size obtained when the quiescent period was 6 sec long was the smallest possessing 17 S and additionally, a significant unsedimenting fraction assigned to monomeric β2-m with 1.6 S at 50,000 rpm (192,900 × g) coexisted (Fig. 2H). The results suggest that the decrease in the size of fibrils accompanies the shift of equilibrium in the direction of depolymerization, i.e., the critical monomer concentration (CMC) increases with the decrease in fibril size. The presence of the overshoot phase was also suggested by the change in ThT fluorescence intensity, although the exact dependency of ThT fluorescence on fibril size is unknown (Fig. 2 A–D).

Characterization of the Ultrasonication-Induced Amyloid Fibrils.

To gain more information about the shape and molecular weight of monodispersed amyloid fibrils obtained by ultrasonication-induced spontaneous fibrillation, AFM, EM, and a sedimentation equilibrium analysis were performed. Although the amyloid fibrils formed with 6-sec intervals were the smallest, it was difficult to reproduce them due to an occasional unexpected overshoot (gray squares in Fig. 2D). Therefore, for the sake of reproducibility, we selected the amyloid fibrils formed at 2-min intervals.

AFM images scanned immediately after the explosive fibrillation phase and after the 1-day incubation period showed plenty of small fibrils without any large aggregates, consistent with observations made in our study in ref. 15 (Fig. 3 A and B). With regard to the uniformity of fibril size, the AFM images indicated uniform fibril lengths, which were further confirmed with EM images (Fig. 3C). A time-dependent decrease in fibril length was also observed on comparison of the AFM images between the 2 time points, supporting an overshoot of fibril size as observed in the sedimentation velocity analysis (Fig. 3 A and B). No serious change in conformation or chemical properties was detected by far-UV CD, HPLC, and MS analyses (SI Text and Fig. S3).

Fig. 3.

Characterization of the β2-m fibrils formed by ultrasonication-induced fibrillation. The amyloid fibrils formed by periodic 1-min ultrasonication followed by a 2-min silence were subjected to AFM, EM, and sedimentation equilibrium analyses. (A and B) AFM images of fibrils sampled immediately after the abrupt increase in ThT fluorescence (A) and those sampled after applying 1-day ultrasonication (B). (C and D) EM image (C) and sedimentation equilibrium analysis (D) of fibrils sampled after the 1-day ultrasonication treatment. The sedimentation equilibrium measurement was performed by measuring absorbance at 280 nm at a rotor speed of 3,000 rpm (700 × g), and 5 °C. The solid line represents a theoretical fitted curve assuming a single species model with an offset. The corresponding residuals are shown in Upper. (Scale bars: B, 1 μm; C, average length of fibril fragments estimated from the result of sedimentation equilibrium and velocity experiments based on the rod-like model, details of which are given in SI Text.)

Furthermore, the small size of the ultrasonicated fibrils allowed applying the sedimentation equilibrium method in which sedimentation and diffusion to be successfully balanced at 3,000 rpm (700 × g) and 5 °C (Fig. 3D). By virtue of the uniformity of the sample fragments, the experimental data were fitted by an equation assuming a single size component and no interfibrillar interactions (see SI Text). By using the solvent density at 5 °C calculated from the database in Ultrascan 8.0 and the partial specific volume (v̄) at 5 °C obtained in our previous experiment (12), we obtained a molecular weight (M_w) value of 1,660,000 ± 20,000 (140 mer), where error is the fitting error. The A₀ value was 0.2, slightly higher than that observed for the sedimentation velocity (0.07), which would be a result of the difference in rotor speed between sedimentation velocity and equilibrium: A small number of fragments sedimented at 20,000 rpm (31,000 × g) remained as an offset at 3,000 rpm (700 × g) during the equilibrium analysis (see SI Text).

It is also important to investigate the distribution width of molecular weight and the M_w value of amyloid fibrils formed by the ultrasonication pulses. As seen in small residual errors of the fitting even with the equation assuming a single size component (Fig. 3D), very sharp distribution of molecular weight was suggested by sedimentation equilibrium measurements. In contrast, when the s values in the abscissa of the sedimentation velocity profile were converted into molecular weights by using the M_w value obtained from the sedimentation equilibrium, a wide range of molecular weight distribution with the standard deviation of 1,660,000 ± 1,290,000 was estimated (refer to Fig. 2K, plot 2). At the present stage, the difference in distribution width between these 2 types of ultracentrifugation methods remains unclear, suggesting that more sophisticated processing of sedimentation profiles is required for more precise analysis of molecular weight distribution.

Effect of Ultrasonication on the Preformed Fibrils.

β2-m amyloid fibrils with lengths of several micrometers prepared by a seed-dependent extension reaction were also subjected to repeated 1-min pulses each of which was followed by a 9-min quiescent incubation at 7 °C and 700 watts (see SI Text). When cycles of ultrasonication pulses were applied, the sedimentation patterns showed a marked decrease in fibril size and distribution width, demonstrating the production of isomeric fibrils (Fig. 4). AFM images also showed the effective fragmentation of amyloid fibrils by the pulses (Fig. 5 A–D). When 2 β2-m amyloid fragments having sedimentation coefficients of 44 S and 32 S obtained after 18 and 72 pulses of ultrasonications, respectively, the difference confirmed with EM images (Fig. 5 E and F), were subjected to a sedimentation equilibrium analysis assuming a single size component, they gave M_w of 3,310,000 ± 70,000 (279 mer) and 2,040,000 ± 30,000 (172 mer), respectively (Fig. 5 G and H), where errors are the fitting errors. Far-UV CD, HPLC, and MS analyses of conformation and chemical properties revealed no serious changes in either of these fibrils (Fig. S3).

Fig. 4.

Dependency of the size distribution of fragmented amyloid fibrils on the number of ultrasonication pulses applied. The frequency, power of output, and pulse length for the ultrasonicators are described in Materials and methods. (A–D) Sedimentation boundary profiles of fragmented fibrils (0.3 mg/mL) without sonication (A), or with 1 pulse (B), 18 pulses (C), and 72 pulses (D) of 1-min ultrasonication. Each sedimentation pattern was recorded at 10,000 rpm (7,700 × g) (A) or 17,000 rpm (22,300 × g) (B–D) and 5 °C by monitoring the absorbance at 280 nm, and several traces at intervals of 10 min (D), 11 min (A), or 12 min (B and C) are presented. (E) The integral distribution plots of β2-m amyloid fibrils treated by 1(black), 3 (red), 9 (blue), 18 (yellow), 33 (green), 54 (magenta), and 72 pulses (cyan) of ultrasonication. The average sedimentation coefficient reached 44 S after 3 pulses, and, although no remarkable change in fibril size was detected between 3 and 33 pulses, the sedimentation coefficient decreased again and eventually reached 32 S after 72 pulses. (Inset) Result for amyloid fibrils before the ultrasonication treatment (white). A distorted integral distribution plot for the fibril samples without ultrasonication represents too large a range of sedimentation coefficients with no stable plateau, failing to extrapolate correctly in the van Holde–Weischet analysis.

Fig. 5.

Characterization of the β2-m fibrils fragmented by ultrasonication. (A–D) AFM images of the fibrils before the ultrasonication treatment (A) and those treated with 1 (B), 18 (C), and 72 pulses (D) of ultrasonication. (E and F) EM images of the fibrils treated by 18-pulse (E) and 72-pulse (F) ultrasonication, with sedimentation coefficients of 44 S and 32 S, respectively (Fig. 4). (G and H) Sedimentation equilibrium measurements of ultrasonicated fibrils of β2-m. The data for fibrils treated by 18-pulse and 72-pulse ultrasonication are plotted under the same condition as used in Fig. 3D. (Scale bars: D, 1 μm; E and F, average length of fibril fragments estimated from the result of sedimentation equilibrium and velocity experiments based on the rod-like model, details of which are given in SI Text.)

By further applying repetitive ultrasonication, insoluble aggregates with white turbidity were formed as a nonideal by-product, and were prominent after 360 pulses (Fig. S4). Simultaneously, the dissociation of a significant amount of monomer and/or small oligomers was also observed with a marked offset from the baseline at 10,000 rpm (7,700 × g), which was assigned to monomeric and/or oligomeric β2-m upon slow sedimentation at 53,000 rpm (216,700 × g). This result indicates that the fragmenting effect of ultrasonication leads to a dissociation into monomers without any convergence of fibril size at 7 °C, where fibrillation does not occur.

To evaluate the effectiveness of the ultrasonication technique in the measurements of NMR spectroscopy, the amyloid fibrils treated by 18 pulses of ultrasonication were further subjected to one dimensional ¹H-NMR measurements. The result clearly represents that the ultrasonicated amyloid fibrils gave many sharp peaks of amyloid fibrils whereas the usual long amyloid fibrils without ultrasonication did not give any peaks, indicating that the ultrasonication-generated amyloid fibrils will be useful for the direct measurement of NMR spectra of amyloid fibrils (Fig. S5 and SI Text).

Discussion

The mechanism underlying the ultrasonication-induced production of small monodispersed fibril fragments is thought to be a combination of 2 opposing effects of ultrasonication, the formation and breaking down of fibrils (Fig. 1). Although the energy barrier of the nucleation phase is generally high, preventing the spontaneous formation of amyloid fibrils under quiescent conditions, ultrasonication can promote nucleation by lowering the energy barrier resulting from the assistance of agitating effects (14, 15). Once the nucleus is formed, the growth phase proceeds via the incorporation of the monomers into the ends of seed fibrils in a template-dependent manner (9). Under normal extension conditions without the fragmenting effect of ultrasonication, the energy landscape is broad, and might be modulated by additional mechanisms including breaking down and rejoining reactions especially for long fibrils (22–25), resulting in a wide range of fibril lengths (black line in Fig. 1). The increase in the monomeric fraction with the decrease in fibril length (Fig. 2 and Fig. 4) suggests that CMC decreases with an increase in fibril length, implying a more complicated mechanism than that approximated by a unique constant of fibril polymerization. In contrast, the combination of fragmentation with the dynamic nature of the equilibrium between the monomeric dissociation and association occurring at the end of each fibril, which has been demonstrated by H/D exchange of amyloid fibrils in the SH3 domain (13) (red line in Fig. 1), will serve to produce small isometric amyloid fibrils. The position of the energy minimum and the steepness of the landscape of fragmentation would be determined by a balance between the elongation kinetics and fragmentation efficiency, the latter of which is presumably determined by the frequency of ultrasonication pulses. Additionally, the intrinsic physical properties such as fragility of individual fibrils (23, 25–27) and intensity of ultrasonication would affect the equilibrated fibril size.

Another accomplishment of the current study is the direct determination of molecular weight by applying the sedimentation equilibrium method. The sedimentation velocity technique has recently been applied to the characterization of the size distribution of prefibrillar oligomers or larger aggregates involving fibrillation (28). With regard to the application to amyloid fibrils themselves, worm-like amyloid fibrils of apolipoprotein C-II with an advantageous soluble nature represent the first instance of successful sedimentation velocity measurements, by which interfibrillar tangling (29) and fibril breaking and rejoining (24) during the process of fibril formation were successfully observed. However, there has been no report for the sedimentation equilibrium of amyloid fibrils.

The accomplishment of sedimentation equilibrium for amyloid fibrils made possible the estimation of their hydrodynamic shapes. By using M_w obtained from the sedimentation equilibrium measurement with the sedimentation coefficient obtained from the sedimentation velocity measurement, we estimated the values of diameter and length for each fragment as 5.0 nm × 99 nm for the 27-S amyloid fibrils formed at 2-min intervals and 6.9 nm × 104 nm and 5.7 nm × 93 nm for the 44-S and 32-S amyloid fibrils formed by the fragmentation, respectively, assuming that the shape of ultrasonicated β2-m fibril fragments is approximately cylindrical (SI Text) (30). All of the estimated sizes of amyloid fibrils agreed with the EM images (Figs. 3C and 5 E and F; the scale bar in each panel represents the length calculated from the frictional coefficient).

From a series of analytical ultracentrifugation investigations performed for fibrils of β2-m, we conclude that ultrasonication is a powerful technique for producing monodispersed fibrils with a well-defined molecular size. The use of ultrasonication-generated amyloid fibrils combined with various conventional techniques is expected to bring about clarification of the wide spectrum of physicochemical properties of amyloid fibrils. In particular, the faster rotational motion with the decrease in size of ultrasonicated amyloid fibrils, which has brought about properties closer to those of a soluble and homogeneous protein solution, might extend to direct applications in solution NMR studies, or the formation of amyloid crystals using isometric fragments as a building block for X-ray crystallography (7, 8). Indeed, the ultrasonicated amyloid fibrils gave many sharp NMR peaks specific to the amyloid fibrils by the faster rotational motion accompanied by the decrease in size by ultrasonication, whereas the usual long amyloid fibrils obtained by seeding did not give any peaks because of an extensive broadening of signals (Fig. S5). This result demonstrates that the ultrasonication technique will become a promising procedure to achieve the direct measurements of NMR spectra of amyloid fibrils to reveal more detailed structural and dynamic properties. Additionally, it might also be expected that the ultrasonication diminishes the polymorphic feature of amyloid fibrils at the molecular level (i.e., various patterns of twisting between protofilaments or various microscopic structures inside the fibrils), resulting in an improvement of the quality of solid-state NMR spectra. Other beneficial applications include the detailed evaluation of kinetic and thermodynamic feature of the fibril formation and the quantitative analysis of the cytophysiological effects such as toxicity and infectiveness of amyloid fibrils, both of which had been difficult to be accomplished because of poor dispersibility and unclarified concentrations of seed ends in the conventional method for fibril preparation. Overall, the development of analytical techniques in combination with the ultrasonication-induced fragmentation method should help to clarify the properties of fibril formation and structure.

Materials and Methods

Ultrasonication Treatments.

Recombinant human β2-m expressed as reported in ref. 31 was dissolved at a concentration of 0.3 mg/mL in a 3.2 mM HCl solution (pH 2.5) containing 38 mM NaCl and was placed on a water bath-type ultrasonicator (ELESTEIN SP070-PG-M, Elekon, Tokyo). Ultrasonication pulses were applied to the samples for 1 min followed by a quiescent period varying from 6 second to 9 min, a process that was repeated during the incubation. The samples were ultrasonicated from 3 directions, i.e., bottom and 2 side walls of the incubating bath, the ultrasonication pulses from which cross one another at the sample position (15). The frequency and output of the sonication were set to 17–20 kHz and 350 watts, respectively, and temperature was maintained at 37 °C. To monitor the formation of amyloid fibrils, a 5-μL aliquot of sample was mixed with 1 mL of 5 μM ThT in a 50 mM glycine-NaOH buffer (pH 8.5) at 25 °C, and the fluorescence intensity at 485 nm of this solution was measured with an excitation wavelength of 445 nm at different points of time (9). The fragmentation of long amyloid fibrils several micrometers in length was also carried out by using the ELESTEIN SP070-PG-M. Long β2-m amyloid fibrils several micrometers in length were formed by the seed-dependent fibril extension method established by Naiki et al. (9). β2-m was dissolved at a concentration of 0.3 mg/mL in a 3.2 mM HCl solution (pH 2.5) containing 38 mM NaCl and the extension reaction was carried out by adding 5 μg/mL of ultrasonicated fibrils as seeds at 37 °C under quiescent conditions. Then, 500-μL aliquots of the formed fibrils were sealed into tubes and ultrasonicated for 1 min, before being incubated for 9 min without sonication, a process that was repeated. The frequency and the power of output were set to 17–20 kHz and 700 watts, respectively, and the temperature was maintained at 7 °C throughout the treatment.

Analytical Ultracentrifugation.

The size distribution of the amyloid fibrils was monitored by sedimentation velocity and sedimentation equilibrium measurements that were performed using a Beckman-Coulter Optima XL-1 analytical ultracentrifuge. For the sedimentation velocity experiments, the samples were first centrifuged at 3,000 rpm (700 × g) for 5 min to stabilize the temperature, and after precentrifugation, the rotor speed was increased to 10,000–20,000 rpm (7,700–30,900 × g) and absorbance data at 280 nm were collected at intervals of 10–20 min. All measurements were carried out at a constant temperature of 5 °C with a radial increment of 0.003 cm in the continuous scanning mode. The sedimentation coefficients that were corrected to s_20W (standard solvent conditions: The density and velocity of pure water at 20 °C) were obtained from the data by the van Holde–Weischet method with the software UltraScan 8.0 (www.ultrascan.uthscsa.edu) (32), using the partial specific volume of amyloid fibrils determined in our study in ref. 12.

For the sedimentation equilibrium experiments, samples were centrifuged at 3,000 rpm (700 × g) at 5 °C and equilibrium concentration profiles were recorded by monitoring absorbance at 280 nm across the centrifugation cell with a radial increment of 0.001 cm in the continuous scanning mode. Detailed methods for the data analysis of sedimentation equilibrium results are described in SI Text.

AFM Measurements.

Fibril samples were diluted 5-fold with water and 50 μL was spotted onto a freshly cleaved mica plate. After 1 min, the residual solution was blown off with compressed air. AFM images were obtained using a Nano Scope IIIa (Digital Instruments). The scanning tip used was a phosphorus (n)-doped Si (Veeco, spring constant = 20–80 N/m, resonance frequency = 245–289 kHz), and the scan rate was 0.5 Hz.

Transmission Electron Microscopy.

Amyloid fibrils (0.3 mg/mL) were diluted 10-fold with water and immediately placed on a 400-mesh carbon-coated copper grid (Nissin EM). The excess solution was removed with filter paper after the sample had stood for 1 min and the fibrils adsorbed on the grid were negatively stained with a 2% (wt/vol) uranyl acetate solution. Electron micrographs were acquired using a transmission microscope (JEM100CX and JEM1200EX; JEOL) at 80 kV with magnification at ×29,000 or ×30,000.