Modulating amyloids’ formation path with sound energy

Significance

Beyond their functional roles, amyloidogenic assemblies are also associated with a large number of diseases, including neurodegenerative. Over the past decades, essential knowledge about amyloidogenic proteins has been gathered, but we are still limited in understanding external fields’ effect on amyloids. This study demonstrates that sound waves can induce chemical modifications and structural perturbations in proteins. We further show that sound energy causes not only the expected nucleation and fragmentation of amyloids but also structural distortion, an observation that may lay the basis for development of therapies. The resultant transformation is determined by the density of the delivered acoustic energy. Our results uncover basic principles underlying sound energy’s effect on proteins, which may be exploited in various protein chemistry settings.

Abstract

Protein folding is crucial for biological activity. Proteins’ failure to fold correctly underlies various pathological processes, including amyloidosis, the aggregation of insoluble proteins (e.g., lysozymes) in organs. The exact conditions that trigger the structural transition of amyloids into β-sheet-rich aggregates are poorly understood, as is the case for the amyloidogenic self-assembly pathway. Ultrasound is routinely used to destabilize a protein’s structure and enhance amyloid growth. Here, we report on an unexpected ultrasound effect on lysozyme amyloid species at different stages of aggregation: ultrasound-induced structural perturbation gives rise to nonamyloidogenic folds. Our infrared and X-ray analyses of the chemical, mechanical, and thermal effects of sound on lysozyme’s structure found, in addition to the expected ultrasound-induced damage, evidence of irreversible disruption of the β-sheet fold of fibrillar lysozyme resulting in their structural transformation into monomers with no β-sheets. This structural transition is reflected in changes in the kinetics of protein self-assembly, namely, either prolonged nucleation or accelerated fibril growth. Using solution X-ray scattering, we determined the structure, the mass fraction of lysozyme monomer, and the morphology of its filamentous assemblies formed under different sound parameters. A nanomechanical analysis of ultrasound-modified protein assemblies revealed a correlation between the β-sheet content and elastic modulus of the protein material. Suppressing one of the ultrasound-derived effects allowed us to control the structural transformations of lysozyme. Overall, our comprehensive investigation establishes the boundary conditions under which ultrasound damages protein structure and fold. This knowledge can be utilized to impose medically desirable structural modifications on amyloid β-sheet-rich proteins.

Amyloid fibrils are a group of linear protein assemblies that share a common structure, fibrillation pathway, and physical characteristics (1, 2). These assemblies are associated with more than 60 diseases, including the neurodegenerative Alzheimer’s and Parkinson’s diseases (3–6) as well as lysozyme-based amyloidosis (7), which is relevant for this study. Generally, amyloids are formed from proteins via misfolding or unfolding mechanisms (8) when subjected to certain destabilizing conditions. For example, changes in pH, ionic strength, or temperature can disrupt the protein’s native fold and prompt a conformational transformation into hydrogen (H)-bonded β-sheet-rich structure (9, 10). This transformation is accompanied by a protein phase transition from a soluble monomeric state into solid nanoscale fibrils. The structure, self-assembly pathway, and biophysical characteristics of different amyloids have been extensively studied using various standard and nonstandard analytical techniques (11, 12). Yet, when it comes to the effects of external fields on amyloids, our understanding remains limited.

This limitation stems from the intricacy of both the underlying physicochemical and biological phenomena. Fibrillar protein self-assembly generates heterogeneous complexes that are dynamic and transient in their nature (13). This aspect has attracted much attention due to the high toxicity of misfolded/unfolded protein molecules and oligomeric complexes (14) and their role as aggregation accelerators (15). External fields, and specifically mechanical ones such as ultrasound (US), generate multiple physical and chemical changes in aqueous media. The chemical effect, the so-called sonolytic effect, triggers the formation of cavity bubbles, which, in turn, produce shock waves, hotspots, and free hydroxyl radicals as well as molecular hydrogen, oxygen, and hydrogen peroxide (16). Dissolved proteins may react with H^• and OH^• radicals (16), causing further chemical changes. For example, free radicals can initiate oxidation-reduction chemical reactions that result in the formation of new covalent bond disulfide (S–S) bridges in cysteine (Cys)-containing proteins (17). The physical effects include mechanical-acoustic streaming and shear stresses, turbulence, and thermal gradients (18–21). Under such conditions, the native fold of proteins may be easily disrupted, leading to unanticipated new protein–protein interactions and often to the further acceleration of aggregation processes (22).

Surprisingly, US-induced damage to the protein structure has been found to be favorable for various industrial and research applications, an aspect extensively explored by our group (23). Food (24), and other industries, for example, routinely use the fact that mechanical agitation of proteins dispersed in biphasic water–oil systems can prompt folded proteins to “open” and sequester protein molecules to the water–oil interface. Other applications include the use of US-made protein capsules, termed microspheres, for drug and gene delivery (25–28) and sonochemistry-mediated functionalization of surfaces with antibacterial protein-based agents (29). We have previously shown that US can be used to modify the structure of proteins in order to tune their intrinsic characteristics, e.g., modifying the fluorescence of green fluorescent protein (29) or chemically functionalizing proteins with polynucleic acids and/or synthetic polymers (30, 31).

In studies of protein conformational disorders (32, 33), the damaging effects of US were shown to amplify small quantities of protein fibrillar aggregates through cycles of sonication at 20-kHz frequency (34). In this process, mature fibrils break into smaller species due to US exposure, and the small fibrillar fragments seed new aggregate growth when incubated in the vicinity of natively folded protein monomers. Traditionally, the “imaging” regime of US (at >10-MHz frequency) is considered not to damage protein folding, even though this assumption has not been experimentally proven. Of particular interest is high-intensity US (20-kHz frequency), as it can simultaneously exert several effects, including chemical modification, thermal fluctuation and mechanical perturbation. Recent studies have demonstrated that the sonication at 20 kHz of monomeric structurally diverse amyloid-forming proteins results in the formation of aggregates with tinctorial, structural, and seeding properties similar to those of amyloids (34). Interestingly, the response of amyloidogenic protein species to 20-kHz US delivered under variable conditions (at different power and energy values) is yet to be understood.

Even though the general impact of US on amyloidogenic protein structure has been established as “damageable,” the exact contribution of each sound-derived effect (mechanical, thermal, or chemical) is largely unknown. The transition point between damageable and nondamageable US regimes is also awaiting discovery. We demonstrate here that, in addition to the protein damage it causes, US can also, under specific conditions, impose structural changes on fibrillating protein species that differ from those of amyloids. These two opposing effects of US on protein folding highlight the need for a fundamental understanding of the effect of mechanical energy on amyloidogenic proteins.

With the aim of establishing the boundary conditions for the damageable and nondamageable regimes of US and to unveil the structural modifications imposed on proteins by US, in this study we performed a systematic investigation of the chemical (free radicals), physical (shear stress), and thermal (gradient in temperature and local heating with hotspots) effects of US and probed their separate and combined impact on fibrillating amyloidogenic protein species. We chose lysozyme as a model amyloid protein as its structure and the amyloid fibrillation pathway are both known (35–37). To define the effect of US regimes on a natively folded amyloid-forming protein, we exposed lysozyme monomers to variable US energies and resolved the US-imposed structural and behavioral changes using an ensemble of analytical methods, including Fourier transform infrared spectroscopy (FTIR), small-angle X-ray scattering (SAXS) analysis, electron and atomic-force microscopy, and fluorescence assays (ThT). We observed that the exposure of protein monomers to low-energy sound waves indeed leads to the rapid formation of nucleated β-sheet-rich species, which accelerate further the amyloidogenic growth process, as postulated in previous literature reports (38). However, short exposure of these monomers to the high-energy US leads to a slight increase in the β-sheet content, without any evidence of protein denaturation, and to a decrease in aggregation rates—changes indicative of nonamyloidogenic structural characteristics. Thus, the exposure to low-energy US induces an amyloidogenic aggregation process, while to high energy US suppressing fibrillation.

Further analysis of the chemical (free-radical formation), physical (mechanical agitation), and thermal effects on the native protein’s fold and aggregation behavior showed that structural changes in monomeric lysozyme are modulated by the thermal effect and by the rate of the delivered thermal energy. When mature protein fibrils were subjected to US exposure, we found that the amount of free radicals formed by the sonication at the energy level we used did not induce distinguishable changes in the protein aggregation, which is in agreement with previously published data (39). The structural analysis showed an irreversible decrease in the β-sheet content (from 47 to 19%) in the lysozyme fibrils treated with US (7 W 1,000 J), as well as altered fibril nanomechanics and aggregation behavior. Such changes are imposed by the combined mechanical and thermal energies. In summary, we were able to establish the damageable and nondamageable regimes of US and to delineate which US-derived effects more greatly contribute to the damage imposed on the native protein’s fold that can lead to structural modifications.

Results and Discussion

Intense ultrasonic waves induce fluctuations in pressure, specifically compression (positive pressure) and rarefaction (negative pressure) phases, as shown in SI Appendix, Fig. S1 A, Top. When an aqueous solution is irradiated with such waves, cavity bubbles form owing to the presence of dissolved gases and/or local water evaporation (21). The bubbles tend to expand at low-pressure levels and shrink at high-pressure levels until they reach a critical size, whereupon they collapse and release their internal high pressure (>1,000 atm) and high temperature (>5,000 K) (SI Appendix, Fig. S1 A, Bottom). The rapid heating/cooling rates (10¹⁰ K/s) inside the cavity bubbles turn them into micro-/nano reactors. The collapse of the bubble is an almost adiabatic process, which releases not only high temperature and pressure but also powerful cavitation-generated shock waves and microjets. Thus, these sonolytic reactions occur under the action of shock waves. When protein monomers disperse in an aqueous solution due to US irradiation, the molecules tend to accumulate dominantly at the gas–liquid interface of the cavity bubbles (40) (SI Appendix, Fig. S1B) and come into close proximity with one another upon bubble collapse. In the presence of protein nanofibrils, the bubbles collapse near the surface of the fibril (SI Appendix, Fig. S1C). Such conditions may lead to three possible outcomes: protein denaturation, structural changes, or no changes to the protein fold when the delivered sonic energy is insufficient (SI Appendix, Fig. S1 D and E).

Effects of Sound Energy on the Morphology of Natively Folded Lysozyme Monomers and Aggregated Lysozyme Nanofibrils.

To study the impact of various US parameters (SI Appendix, Fig. S2) on the self-assembling amyloid-forming protein lysozyme, we focused on the sonolytic effects on natively folded protein monomers and self-assembled protein nanofibrils. Our atomic force microscopy (AFM) (SI Appendix, Fig. S3 A and B) and TEM (Fig. 1 A and B) analyses show that exposure to US waves prompts the conversion of soluble protein into particle-like supramolecular assemblies.

Fig. 1.

Sonolytic effects on lysozyme protein monomers and lysozyme protein nanofibrils. (A) TEM image of lysozyme protein monomers. (B) TEM image of lysozyme aggregates formed under US irradiation (7 W 1,000 J). (C) DLS analysis of the average size distribution (nm) of protein aggregates formed from protein monomers under irradiation of US with variable parameters. The applied power (W) and energy (J) values are shown in the chart (see detailed evaluation of the US parameters in SI Appendix, Fig. S2). Inset is a graphical representation of the mechanism of aggregate formation under the action of US. (D) TEM image of lysozyme protein nanofibrils. (E) TEM image of lysozyme fragments formed from lysozyme nanofibrils under US irradiation (7 W 1,000 J). (F) Average fibril length distribution analysis of protein fibrillar fragments formed under US with variable parameters. The applied power (W) and energy (J) values are shown in the chart (see detailed evaluation of the US parameters in SI Appendix, Fig. S2). Inset is a graphical representation of the mechanism of US-induced fibril fragmentation.

The size of the particle aggregates grew as the delivered US energy increased, from 4 nm for untreated protein to 1,100 nm (SI Appendix, Fig. S3 B, iv) when the protein was exposed to 1,000 J—reflecting a 300× change. We systematically tuned the US energy from 100 J to 1,000 J (between 2 W and 7 W) and tracked the changes in the size of the formed particles using dynamic light scattering (DLS). The DLS analysis (Fig. 1C) showed that exposure of a monomeric protein solution to US energy of 400 J at a low power of 2 W generates protein particles with a multimodal size distribution of 4 nm (indicative of a monomeric species) and also larger aggregated particles that are 17 and 133 nm in size (SI Appendix, Fig. S3 B, ii). Exposure to 720 J at 5 W increased the level of heterogeneity of the protein assemblies and produced three populations with different sizes: 4 nm, 64 nm, and 525 nm (SI Appendix, Fig. S3 B, iii). The higher energy input (1,000 J, 7 W) leads to the formation of micron-sized particles (1,100 nm) in addition to the monomeric species (SI Appendix, Fig. S3 B, iv).

When aggregated lysozyme nanofibrils (see AFM image in SI Appendix, Fig. S3C and TEM image in Fig. 1D) were exposed to sound energy, we observed the formation of smaller fragments that were contaminated with a small fraction of amorphous aggregates (SI Appendix, Fig. S3D and Fig. 1E). This observation is consistent with previous literature reports. The length of the fragments decreased linearly with the increase in the delivered energy (Fig. 1F). Interestingly, this linear relationship between fibril length shortening and increase in sound energy ceases at the critical length value of 30 to 60 nm (SI Appendix, Fig. S3C), below which the fibrils cannot be fragmented anymore.

Characterization of the Structural Changes Imposed by Sound Energy on the Protein Monomers and Aggregated Protein Nanofibrils.

To establish the structural organization of particles and fragments formed following exposure of lysozyme monomers or nanofibrils to US, we performed a FTIR analysis of the changes in the vibration bands corresponding to amide I. This allowed us to resolve the differences between the native protein fold and the β-sheet-rich aggregated amyloid. The vibration spectra of the aggregated amyloid structures are characterized by the presence of intermolecular β-sheet content (1,610 to 1,625 cm⁻¹) and antiparallel amyloid β-sheets (1,690 to 1,705 cm⁻¹) (41). These spectra differ from the characteristic vibrations of the native protein fold are 1,625 to 1,635 cm⁻¹ for β-sheet content, 1,635 to 1,665 cm⁻¹ for α-helix and random coils, and 1,665 to 1,690 cm⁻¹ for β-turns.

We found that exposure to US changes the secondary structure of monomeric lysozyme to change profoundly (summarized in Fig. 2 A and B; see full spectra in SI Appendix, Figs. S4 and S5).

Fig. 2.

FTIR analysis of lysozyme monomers and fibrils exposed to US irradiation. (A and B) Amide I FTIR spectra of untreated lysozyme monomers (A) and monomers exposed to US at 7 W 1,000 J (B). (C) Secondary structural content of random coils, α-helixes, and β-sheets for untreated and US-treated monomeric lysozyme. (D and E) Amide I FTIR spectra of mature lysozyme fibrils (D) and fibrils exposed to US at 7 W 1,000 J (E). A comparative analysis of the secondary structure in % of fibrils treated with US with various parameters (F). (G and H) SDS gels of untreated and US-treated monomers (G) and fibrils (H).

Specifically, at US power of 2 W, when the delivered energy was increased from 0 J (untreated) to 400 J, the random coil and α-helix content of untreated monomers decreased from ~70 to ~54%, while the β-turn and antiparallel β-sheet content, typical of aggregated proteins, increased, as expected. At US power of 7 W, when the delivered energy was increased from 0 J (untreated) to 1,000 J, the aggregated β-sheet content in the protein monomers increased to 29%, while the observed 35% increase in β-turns was associated with a more disordered state (compared with amyloidogenic β-sheets).

The maximum is reached at US energies of 1,000 J and higher, where thermal denaturation/hydrolysis of the protein monomers is observed (see Fig. 2G). An analysis of the changes in the secondary structure of the US-formed lysozyme fibrillar fragments revealed a trend of a decrease in aggregative amyloidogenic β-sheets and an increase in less-ordered content, namely, β-turns, α-helixes and random coils (Fig. 2 D–F).

To elucidate whether lysozyme monomers polymerize (via the formation of dimers, trimers, tetramers, or x-mers) and fibrils depolymerize (i.e., undergo partial or full denaturation), we performed a gel electrophoresis analysis (Fig. 2 G and H). We observed that the exposure of monomeric lysozyme to sound energy of up to 500 J does not significantly change the natural tendency of lysozyme to form complexes. The gel electrophoresis analysis showed monomeric and dimeric bands at 14 kDa and 28 kDa without any sign of protein degradation (Fig. 2G). It is worth mentioning that lysozyme has a natural tendency to form dimers spontaneously under ambient conditions. At US energies of 500 J and higher, however, large aggregates, as well as protein proteolytic fragments, formed. This observation indicates that the protein damage is due to US-derived effects in an aqueous environment. Interestingly, whereas the gel electrophoresis analysis of mature lysozyme fibrils exposed to low US energies showed the presence of fibrillar content, monomers and products of proteolysis, but no dimeric complexes, at US energies of 1,000 J or higher, dimeric content appeared alongside fibrils, monomers, and proteolytic products.

To unravel whether the changes in the secondary content and in the formation of protein complexes originate from the mechanical perturbation or from the temperature gradient created by US, we compared the contribution of each effect separately in the absence of US agitation (see Materials and Methods section). The FTIR spectroscopy analysis found no structural change in the protein monomers under mechanical mixing. However, the exposure to heat (SI Appendix, Fig. S6) led to the thermal denaturation of protein monomers. When the fibrillar species were exposed to mechanical mixing and heat, either separately or simultaneously, no pronounced structural changes were detected (SI Appendix, Fig. S6). These results indicate the sensitivity of soluble monomeric protein molecules to temperature, which cause the protein to adopt a fibrillar, β-sheet-rich structure via the self-assembly mechanism.

Furthermore, we probed the chemical effect of ultrasonic treatment on monomeric and fibrillar protein species. To this end, we first evaluated the concentration of radicals formed in aqueous media during the sonication by using chemical dosimetry (42). Next, the plasma discharge source was used to generate free radicals in aqueous media without causing any thermal and mechanical (mixing and perturbations) effects. The plasma discharge method is known to produce reactive oxygen species from the air to the solution phase (43). The level of H^• and OH^• radicals during the US was detected by the chemical dosimetry method (42). The results show that the produced radical did not impose any structural and behavioral changes on protein monomers and protein fibrils. Therefore, it cannot be seen as a driving force for protein aggregation. We further reconfirmed this observation by performing two types of analyses. The first involved using high-resolution electrospray ionization mass spectrometry to track the changes in monomeric protein species upon exposure to US. The second method was an amino acid analysis of the possible sensitivity of the specific amino acid from the lysozyme protein sequence to sound energy. The results are summarized in SI Appendix, Figs. S7 and S8 and the SI Appendix. Interestingly, changes in lysozyme’s molecular weight (~14,304 Da, SI Appendix, Fig. S7) or the pronounced difference between the untreated and US-treated amino acid fraction (in pmole) (SI Appendix, Fig. S8) was detected. This observation reinforces the negligible impact of the chemical effect of US on self-assembling lysozyme protein species.

SAXS Analysis.

To further gain insights into the structural changes imposed by US irradiation on monomeric and aggregated fibrillar protein constructs, we performed a SAXS analysis at the European Synchrotron Radiation Facility (ESRF) (see Materials and Methods section). The data were analyzed using D+ software (44). The atomic model of hydrated lysozyme monomer (PDB ID 1aki) was used in our SAXS analysis. To fit the SAXS data (Fig. 3 A–D), we modeled several protein complexes formed under the action of US by docking the atomic model of lysozyme monomers at various locations and orientations (see SAXS Models in the Materials and Methods section).

Fig. 3.

Small-angle X-ray scattering (SAXS) analysis of lysozyme protein assemblies. (A–D) Analysis of azimuthally integrated background-subtracted scattering intensity $I$, as a function of $q$, the magnitude of the momentum transfer vector $\overrightarrow{q}$. (A) soluble monomers, (B) protein assemblies formed from lysozyme monomers under US irradiation, (C) aggregated lysozyme fibrils, (D) protein species formed from lysozyme fibrils under US irradiation. The Insets show a ribbon diagram of monomeric lysozyme (A), and flat and spherical lysozyme assemblies formed under US (B). (E) Model of a mature lysozyme fibril resolved based on the SAXS analysis. (F) Model of lysozyme fragments formed under US resolved based on the SAXS analysis. SI Appendix, Fig. S9 shows the computed SAXS curves of lysozyme monomers (SI Appendix, Fig. S9A), spherical aggregates (SI Appendix, Fig. S9B), flat aggregates (SI Appendix, Fig. S9C), fibril (SI Appendix, Fig. S9D), and fibril fragments (SI Appendix, Fig. S9 E and F). The mass fractions of the structures that best fitted the SAXS data in B, C, and D are shown in SI Appendix, Fig. S9 G–I. (G and H) Analysis of the chemical kinetics of lysozyme aggregation for (G) untreated and US-treated monomers and (H) untreated and US-treated fibrils.

We conducted a comparative SAXS analysis of untreated monomers and monomeric protein samples treated with US (Fig. 3 A and B, respectively). In the case of the untreated monomers, the model (Fig. 3A, blue curve) was obtained by multiplying the form-factor of the hydrated monomer atomic model (SI Appendix, Fig. S9A) by a Lorentzian structure-factor with an amplitude A of 0.784 ± 0.001, full width at half maximum, $\mathrm{\Gamma }=1.005\pm 0.001 {\mathrm{nm}}^{-1}$, centered at $q_c=1.075\pm 0.001 {\mathrm{nm}}^{-1}$. The structure factor represents monomer-monomer interactions. The structure factor approximates the mean distance, d_m, between interacting monomers in the solution. Assuming a hexagonal-like arrangement (45), $d_m={\displaystyle \raisebox{1ex}{$4\pi $}\!\left/ \!\raisebox{-1ex}{$\sqrt{3}{q}_c$}\right.}=6.75\pm 0.01 \mathrm{nm}$.

We observed that exposure of natively folded monomers to US triggered the formation of aggregates that coexisted with soluble monomers (Fig. 3B). The monomer form-factor (SI Appendix, Fig. S9A) was multiplied by a Lorentzian with $A=1.326\pm 0.001$, $\mathrm{\Gamma }=0.715\pm 0.001 {\mathrm{nm}}^{-1}$, and $q_c= 1.285 \pm 0.001 {\mathrm{nm}}^{-1}$, corresponding to a mean spacing of 5.64±0.01 nm between soluble monomers. The shorter spacing between the monomers indicate that the attracting interactions between the monomers became stronger following US. To the monomer model we added a model of a lysozyme fibril, created by docking copies of the atomic model of a hydrated lysozyme monomer at positions ${\overrightarrow{r}}_{\mathit{ijk}}=(x_i,y_j,z_k)$, where $x_i=i{D}_x$, $y_j=j{D}_y$, $z_k=k{D}_z$, and $i,j$ and $k$ vary between 0 and $⌈{\displaystyle \raisebox{1ex}{$W$}\!\left/ \!\raisebox{-1ex}{$D_x$}\right.}⌉-1$, $⌈{\displaystyle \raisebox{1ex}{$d$}\!\left/ \!\raisebox{-1ex}{$D_y$}\right.}⌉-1$, and $⌈{\displaystyle \raisebox{1ex}{$H$}\!\left/ \!\raisebox{-1ex}{$D_z$}\right.}⌉-1$, respectively. The height, $H,$ width, $W$, and depth, $d$, of the fibril were 400 nm, 300 nm, and 6.0 ± 0.2 nm, respectively. We note, however, that our largest dimensions’ resolution was limited by our lowest scattering angles. Hence, $H$ and $W$ could be larger. The spacing between the centers of adjacent monomers in the $x, y$ and $z$ directions where $D_x=D_y=D_z=3.0\pm 0.1 \mathrm{nm}$. We took into account the contribution of thermal fluctuations, assuming a harmonic potential between the nearest-neighbor monomers, as explained in ref. 46, using a lattice elastic constant $\kappa =2\pm 1 \times {10}^{-19} \mathrm{J} {\mathrm{nm}}^{-2}$. From the equipartition theorem and the value of $k$, we estimated the root mean squared displacement (RMSD) of each monomer within the fibril to be: $〈{\left|\overrightarrow{u}\right|}^2〉=\sqrt{\frac{k_{B}T}{\kappa }}\approx 0.14\pm 0.03 \text{nm}$. This RMSD value is below the Lindemann melting threshold value (roughly ∼0.1 of the interparticle distance (~0.3 nm)), suggesting the fibril was rather stiff.

After introducing thermal fluctuations, here and in subsequent fibril models, we deleted half of the last layer by removing monomers with indexes $⌈{\displaystyle \raisebox{1ex}{$W$}\!\left/ \!\raisebox{-1ex}{$2D_x$}\right.}⌉<i<⌈{\displaystyle \raisebox{1ex}{$W$}\!\left/ \!\raisebox{-1ex}{$D_x$}\right.}⌉$, $j=⌈{\displaystyle \raisebox{1ex}{$d$}\!\left/ \!\raisebox{-1ex}{$D_y$}\right.}⌉-1$, and $0\le k<⌈{\displaystyle \raisebox{1ex}{$H$}\!\left/ \!\raisebox{-1ex}{$D_z$}\right.}⌉.$ The computed SAXS curve of this flat aggregate model is shown in SI Appendix, Fig. S9B and illustrated in the Inset of Fig. 3B.

In addition, the contribution of large spherical aggregates was taken into account by the Porod approximation (SI Appendix, Fig. S9C), given by ${\displaystyle \raisebox{1ex}{$\alpha $}\!\left/ \!\raisebox{-1ex}{$q^4$}\right.}$ (when $\alpha =2.90\pm 0.01\times {10}^{-7}$ is a constant calibrated based on the atomic model of the spherical object, illustrated in the inset of Fig. 3B). When the aggregates are large enough, our lowest q-range meets the Porod analysis requirements (47). Thus, our analysis reveals that US waves triggered the formation of spherical (Fig. 3 B, Upper Inset) and soft flat (Fig.3 B, Bottom Inset) aggregates. The mass fraction of lysozyme monomers in each of the structures is shown in SI Appendix, Fig. S9G.

Next, we analyzed by SAXS the US-triggered structural changes in the fibril (Fig. 3 C and D). For untreated lysozyme fibrils (Fig. 3C), the model was calculated using the hydrated monomer form-factor and a fibril model, calculated with the following parameters: $D_x=D_y=D_z=3.0\pm 0.1 \mathrm{nm}$,

$H=400 \mathrm{nm}$, $W=12.0\pm 0.4 \mathrm{nm}$, and $d=6.0\pm 0.2 \mathrm{nm}$. We took into account the contribution of thermal fluctuations, assuming a harmonic potential between the nearest-neighbor monomers in the fibril, with a lattice elastic constant (46) $\kappa =1.3\pm 0.6\times {10}^{-19} \mathrm{J} {\mathrm{nm}}^{-2}$ and RMSD of ~0.18 ± 0.03 nm, indicating the fibril structure was still rather stiff (this model is illustrated in Fig. 3E). The computed SAXS curve of this fibril model is shown in SI Appendix, Fig. S9D. The mass fraction of the monomers and fibrils is shown in SI Appendix, Fig. S9H.

For fibrils treated with US (Fig. 3D), the model was calculated using the monomer form-factor and two fibril fragments (illustrated if Fig. 3F) with the following parameters: fibril #1 $H=150\pm 20 \mathrm{nm}$, $w=12.0\pm 0.4 \mathrm{nm}$, $d=6.0\pm 0.2 \mathrm{nm}$ (corresponding to 50 × 4 × 2 monomers, SI Appendix, Fig. S9E); fibril #2 $H=33\pm 9 \mathrm{nm}$, $w=12.0\pm 0.4 \mathrm{nm}$, $d=6.0\pm 0.2 \mathrm{nm}$ (corresponding to 11 × 4 × 2 monomers, SI Appendix, Fig. S9F). In both fibril models, we deleted half of the last fibril layer. We took into account the contribution of thermal fluctuations, assuming a harmonic potential between the nearest-neighbor monomers with a lattice elastic constant (46) $\kappa =1.7\pm 1\times {10}^{-19} \mathrm{J} {\mathrm{nm}}^{-2}$ (RMSD ~ 0.15 ± 0.03 nm). The mass fraction of fibrillar species in each of the structures is shown in SI Appendix, Fig. S9I. Thus, our SAXS analysis showed that protein fibrils exposed to US tend to fragment, producing fibrils that are tighter and shorter in length. The SAXS models for untreated full-length fibrils and fibril fragments following US exposure are presented in Fig. 3 E and F, respectively.

Analysis of US-Imposed Changes in the Chemical Kinetics of Fibrillar Protein Self-Assembly.

To unravel whether US-generated assemblies from lysozyme monomers and fragments from lysozyme nanofibrils share amyloidogenic behavioral characteristics, we used a standard Thioflavin T (ThT)-binding assay. ThT dye undergoes a red-shifted change in its emission spectra (measured maxima at 490 nm) upon binding to amyloid structures. Monitoring the changes in ThT emission intensity enables one to characterize specific molecular events in the protein aggregation process, including primary nucleation, elongation, and secondary nucleation. The chemical kinetics analysis is expected to shed light on whether β-sheet-rich structures formed from monomeric lysozyme under US irradiation are identical to amyloidogenic nuclei that seed amyloid growth and whether fragmented species formed from mature fibrils by the US are similar to those of natively formed amyloid fragments that accelerate amyloid growth. Thus, to investigate the changes in chemical kinetics in response to US exposure, we mixed natively folded lysozyme monomers (i.e., untreated) with lysozyme monomers that were exposed to US with different parameters at a ratio of 1:1 by volume (Fig. 3G). We observed pronounced differences in the lag phase of the protein fibrillation process, which is associated with the transformation of natively folded monomers into oligomers and then into nuclei. Low-amplitude sonication (20%) (2 W power, 100 to 400 J sound energy) causes a very small acceleration in protein aggregation, whereas more aggressive conditions (7 W power, 750 to 1,000 J or higher) lead to a reduction in the fibrillation rate.

As irradiation of aqueous media with US can produce three types of effects, chemical (free radicals), mechanical, and thermal (21), we next compared the impact of each effect on natively folded monomers. This comparison was achieved via either a mixing protein solution (see Materials and Methods section), where we aimed to mimic acoustic streaming and turbulence, or heating. The kinetic analysis (SI Appendix, Fig. S10 A and B) revealed that mechanical forces do not cause any changes in the aggregative behavior of lysozyme monomers, while increase in temperature at sound energy level (see Materials and Methods) strongly effect the lag phase of protein aggregation. This observation is explained by the high stability of the lysozyme secondary structure, which is indeed sensitive to changes in pH and temperature, but has low sensitivity to shear forces (48). Overall, the structural characteristics and amyloidogenic aggregation behavior of US-treated monomers confirm that the exposure of natively folded monomers to US indeed triggers the formation of relatively large assemblies, but these assemblies do not exhibit classical amyloid characteristics.

We next investigated the effect of sound energy on behavioral characteristics of mature protein fibrils (Fig. 3H). In general, the growth of an amyloid fibril progresses via the addition of monomers to each end of the fibrillar assembly. When a mature fibril is fragmented into smaller species, the number of ends increases (SI Appendix, Fig. S3 C and D). For example, when a mature fibril with two ends is broken into two fragments, the total number of ends increases to four. We examined the changes in the chemical kinetics of amyloid growth owing to US exposure using the above-described ThT assay, and further analyzed the aggregation rates due to US exposure by assessing the number of fibril ends. As expected, the results show that the exposure of lysozyme nanofibrils to US accelerates the growth phase of amyloid fibrils, with no lag phase observed (Fig. 3H). The growth phase is linearly dependent on the number of free ends (SI Appendix, Fig. S11). The correlation between the structural analysis of the US-formed lysozyme fibrillar fragments (which tend to exhibit a decrease in aggregative amyloidogenic β-sheets and an increase in less-ordered content) and kinetics studies (Fig. 2F and SI Appendix, Fig. S5) indicate that sound energy perturbs the aggregative fold of amyloid fibrils. The phase diagram summarizing the correlation between the applied sound energy and the effect of US on monomeric proteins and on fibrillar lysozymes is shown in SI Appendix, Fig. S12). Our analysis of thermal and mechanical effects on fibrils’ growth kinetics (SI Appendix, Fig. S10B) reveals that mechanically fragmented fibrils seed new aggregation growth, as expected, whereas thermally treated fibrils do not exhibit any change in the fibril elongation process.

Physical Characteristics of Supramolecular Fibrillar Assemblies Formed with Sound Energy.

One of the notable characteristics of amyloid protein fibrils is their material properties, specifically, their nanomechanics. Thus, we performed modulus mapping of the lysozyme assemblies generated from monomers (Fig. 4B) and fibrils after exposure to US (Fig. 4D).

Fig. 4.

Nanomechanical analysis of lysozyme monomers and fibrils. Comparison of the height, elastic modulus, and deformation of (A) natively folded lysozyme monomers and (B) aggregates formed as a result of US treatment of monomers at 7 W 1,000 J. Comparison of the height, elastic modulus, and deformation of (C) mature fibrils and (D) fragments after sonication at 7 W 1,000 J.

The identification of elastic modulus for untreated monomers is questionable due to the fact that monomeric protein species are small in their size and cover the surface with a very thin layer (~2 nm); therefore the determination of the modulus is affected by the mica substrate. The measured average elastic modulus for assemblies formed from monomers under US irradiation depicted inhomogeneity with measured values from 8.37 ± 2.82 GPa to 15.02 ± 6.03 GPa. The formed aggregates are not homogeneous in nanomechanical characteristics with stiffer and softer regions and are different from nanomechanics of fibrillar assemblies, which are fairly uniform. Our results show that exposure of lysozyme monomers to US triggers the formation of larger particles, as shown in Fig. 4B. This observation is in good agreement with our results from structural analysis, indicating the appearance of the small fraction of β-sheet conformation upon exposure of the monomeric protein to US. The nanoindentation analysis of fibrillar species exposed to US shows that average elastic modulus of sonicated fragments is 12.03 ± 2.02 GPa, which is similar to mature fibrils (elastic modulus 11.95 ± 2.51 GPa) (Fig. 4D).

Conclusions

Our comprehensive study shows that exposing lysozyme protein in its monomeric soluble state to US leads to the formation of two types of aggregates: nonamyloidogenic supramolecular assemblies and complexes with a small fraction of β-sheets that deviate from the classical amyloid state. Such newly-adopted structural characteristics are reflected in changes in the proteins’ aggregative behavior, resolved by a chemical kinetics analysis. The exposure of fibrillar species to US leads to the irreversible disruption of the β-sheet fold and to a structural transformation into protein monomer-like species with decreased β-sheet fraction. Interestingly, the analysis of the separate contribution of the chemical, mechanical, and thermal effects of sound on lysozyme’s structure showed that even though the high-intensity US at 20 kHz gives rise to the formation of the reactive radical species, the contribution of the chemical effect, namely, the formation of cavity bubble interfaces, is negligible. Rather, the structure-modifying effect originates from the shock waves created upon cavity bubble collapse and the release of ether thermal energy that preferentially affects the structure of the protein monomers, or of mechanical energy that results in the fragmentation of the fibrillar assemblies. Overall, our investigation establishes the effect of US on the structure, behavior, and properties of self-assembling protein species. This knowledge can be utilized as a tool for imposing structural modifications on amyloid β-sheet-rich proteins in highly controlled settings.

Materials and Methods

Preparation of Lysozyme Fibril/Monomer Solution.

Monomer solution: Lyophilized egg white lysozyme (Sigma-Aldrich) was dissolved in 20 mM NaCl + 0.01 M HCl at 20 mg/mL (1.4 mM). For in vitro fibril formation, the solution was incubated for 72 h at 65 °C. After incubation, a gel was obtained. Fibrils remained stable in solution and were stored at 4 °C.

US Treatment of Lysozyme Fibrils/Monomers.

Lysozyme fibril/monomer solutions were subjected to sonication using an ultrasonic processor (QSONICA) operating at 20 kHz. The sonication was carried out at various amplitudes (20 to 70%, power 2 to 7 W), and varying time intervals (1 to 4 min).

Mechanical Mixing of Fibrils/Monomers.

Mechanical mixing was performed via magnetic stirring at 1,000 rpm for 4 min. 

Thermal treatment. This was performed using a thermal mixing block, heated to 90 °C (the maximum temperature reached during the sonication).

Kinetic Measurements Based on the ThT Assay.

To follow the kinetics of protein fibrillation, we used ThT dye, which binds to the β-sheet structure of protein aggregates. Upon binding, the quantum yield of the dye increases by multiple orders of magnitude, permitting the use of ThT as a highly sensitive reporter for amyloid species. For the kinetics assay, lysozyme monomer solutions/amyloid suspensions were mixed with 20 µM ThT and placed in a 96-well plate (Greiner Bio-One GmbH, Germany) at 65 °C in a CLARIOstar (BMG LABTECH) plate reader. ThT fluorescence was measured at excitation and emission wavelengths of 440 nm and 490 nm, respectively. The reaction was stopped once the saturation phase of kinetics was reached and detected.

Morphological Analysis by AFM.

Sample preparation: A small droplet (50 µL) of a fibril/monomer suspension was deposited onto freshly cleaved mica, allowed to stand for 2 min, washed several times with distilled water, and then dried. AFM tapping-mode measurements were carried out using AC240 cantilevers on a NanoWizard 4 instrument in an acoustic enclosure (JPK Instruments). JPK Data Processing Software was used for image analysis, and ImageJ software for statistical data collection.

Protein Secondary Structure Elucidation Using FTIR Spectroscopy.

Structural analysis of the lysozyme fibrils and monomers was performed using an FTIR spectrometer Nicolet iS50 with an attenuated total reflection (ATR) accessory (Thermo Scientific). The samples were used without further pretreatment and were loaded into the ATR holder and analyzed by subtracting a water reference. The atmospheric compensation spectrum was subtracted from the original FTIR spectra and a second derivative was applied for further analysis. To resolve the spectra, OMNIC spectroscopy software (Thermo Scientific) and OriginLab software were used.

DLS for Monitoring Protein Aggregation.

DLS experiments were carried out with a Zetasizer (Malvern Panalytical) DLS instrument at 25 °C with a protein concentration of 2 mg/mL using a laser wavelength of 780 nm and a scattering angle of 90°. DLS data were analyzed using Zetasizer’s operating software.

Transmission Electron Microscopy (TEM).

Fibrils and monomers were dried onto freshly discharged 300 mesh formvar carbon-coated TEM grids and negatively stained with 2% aqueous uranyl acetate. The samples were observed with a FEI Tecnai T12 120 kV transmission electron microscope.

SDS–PAGE.

Lysozyme fibrils/monomers were loaded on sodium dodecyl sulfate (SDS)-polyacrylamide gels (15% acrylamide). The gels were stained with silver according to the standard silver-stain protocol. The gels were fixated with a fixation solution of 50% methanol, 12% acetic acid, and 0.05% formaldehyde (37%) for 1 h. Next, the gels were rehydrated for 1 h with 50% ethanol, followed by a reduction for 1.5 min in a solution of 0.02% Na₂S₂O_3. The gels were rinsed thoroughly with double-distilled water, then stained with a silver solution containing 0.2% AgNO₃ and 0.075% formaldehyde (37%). Finally, the gels were rinsed again thoroughly with double-distilled water and developed in a solution containing 6% Na₂CO₃, 0.05% formaldehyde (37%), and 0.0004% Na₂S₂O₃.

AFM Nanomechanical Measurements.

Imaging and nanomechanical properties measurements were conducted using PeakForce QNM on a Multimode AFM (Bruker). AC160 probes (Olympus) with a spring constant of 40 N/m were used. The effective tip radius was calibrated on a sample of HOPG with a modulus of 18 GPa, deformation depths were kept to about 1 nm. The Bruker software (Nanoscope 9.2) was used to compute the elastic modulus maps using the DMT model for fitting the force vs. deformation curves. Gwyddion open-source software was used for the image analysis. All averaged modulus values were calculated from the center of each fibril or aggregate.

SAXS Measurements.

All solution SAXS measurements were measured at ID02 beamline at the ESRF. The analysis was performed using a beam size of $32.4\times 145 {\mathrm{\mu m}}^2$ (vertical and horizontal, respectively), the photon energy of 12.23 keV, Eiger2 4M (Dectris AG) detector, a sample-to-detector distance of 3.114 m, and exposure time of 0.1 s (49).

SAXS Models.

All the models were computed after translating their center of mass to the origin. We assumed an instrument Gaussian resolution function with a SD of 0.001, as explained in ref. 45. Other D+ computational parameters were: Integration Method: Monte Carlo (Mersenne Twister), Integration iterations: 10⁶, Monomer Grid Size: 20, Number of Generated Points: 300, Convergence: 0.001, Update interval: 500 ms. The hydration layer was computed assuming an outer solvent electron density of $340 \mathrm{e} {\mathrm{mm}}^{-3}$, using the dummy atoms method, solvation thickness of 0.28 nm, solvation probe radius of 0.14 nm, solvent voxel size of 0.2 nm, $C_1=1$, and solvent electron density of $333 \mathrm{e} {\mathrm{mm}}^{-3}$. We used the Hybrid method with a reciprocal grid for the monomer. These parameters are explained in our earlier publication (44).

The atomic model of the fibrils was created by docking copies of the atomic model of the hydrated lysozyme monomer into the fibril assembly symmetry (describing the rotations and translations of each monomer subunit). In a fibril of height $H$, depth $d$ and width $W$, the location of a specific monomer, ${\overrightarrow{r}}_{\mathit{ijk}}=(x_i,y_j,z_k)$, is $x_i=i{D}_x$, $y_j=j{D}_y$, $z_k=k{D}_z$, where $i,j$ and $k$ vary between 0 and $⌈{\displaystyle \raisebox{1ex}{$W$}\!\left/ \!\raisebox{-1ex}{$D_x$}\right.}⌉-1$, $⌈{\displaystyle \raisebox{1ex}{$d$}\!\left/ \!\raisebox{-1ex}{$D_y$}\right.}⌉-1$, and $⌈{\displaystyle \raisebox{1ex}{$H$}\!\left/ \!\raisebox{-1ex}{$D_z$}\right.}⌉-1$, respectively. $H, W$, and $d$ vary between different fibril models. The spacing between the centers of adjacent monomers in the $x, y$ and $z$ directions, where $D_x=D_y=D_z \mathrm{i}\mathrm{s} 3.0\pm 0.1 \mathrm{nm}$ in all the models. The orientation of the monomer was the same in all the samples, given by ${\alpha }_{\mathit{ijk}}={\beta }_{\mathit{ijk}}={\gamma }_{\mathit{ijk}}=0$, using the Tait-Brain notation. We took into account the contribution of thermal fluctuations, assuming a harmonic potential between the nearest-neighbor monomers, as explained in our earlier publication (46).

After introducing thermal fluctuations, in all the fibril models, we deleted half of the last layer by removing monomers with indexes $⌈{\displaystyle \raisebox{1ex}{$W$}\!\left/ \!\raisebox{-1ex}{$2D_x$}\right.}⌉<i<⌈{\displaystyle \raisebox{1ex}{$W$}\!\left/ \!\raisebox{-1ex}{$D_x$}\right.}⌉$, $j=⌈{\displaystyle \raisebox{1ex}{$d$}\!\left/ \!\raisebox{-1ex}{$D_y$}\right.}⌉-1$, and $0\le k<⌈{\displaystyle \raisebox{1ex}{$H$}\!\left/ \!\raisebox{-1ex}{$D_z$}\right.}⌉.$ To all the models, a very small constant background factor was added to optimize the fit.

Mass-Spectrometry Analysis of Lysozyme Monomers.

To analyze whether lysozyme monomers undergo chemical changes, the control solution of untreated monomers and US-treated monomers (1,000 J, 7 W) were subjected to a mass spectrometry (MS) analysis. Salts were removed by dialysis against water. Samples were diluted to the appropriate concentration prior to the MS analysis. The analyses were carried out on a Waters Xevo G2-S QTof mass spectrometer with an electrospray ionization source operating in the positive mode. The samples were directly infused at a flow rate of 10 μL/min. All the spectra were acquired in the mass range of 50 to 2,000 m/z. The positive matches were limited, with a mass error of no more than 5.0 ppm. The following settings were applied: capillary voltage of 3.00 kV, cone gas flow of 20 L/h, source temperature of 120 °C, and cone voltage of 30 V. The desolvation temperature was 250 °C, and the desolvation gas (N₂) flow rate was 400 L/h. All measurements were done using Leucine–Enkephalin (200 μg/μL, acetonitrile: H₂O containing 0.1% formic acid (1:1, v/v)) as a lock-spray reference at a flow rate of 10 μL/min, to ensure mass accuracy and follow resolution mode. Data acquisition and recording were done by Waters MassLynx v4.2 software.

Amino Acid Analysis.

The amino acid analysis was performed to probe whether US causes changes to the amino acid composition of lysozyme. The Waters AccQ Tag amino acid analysis method, which is based on a derivatizing reagent developed specifically for amino acid analysis, was applied. Waters AccQ Fluor Reagent (AccQ Fluor™ Reagent Kit) converts amino acids to stable, fluorescent derivatives.

The chromatographic separation of the labeled amino acids was performed on a Waters Alliance 2695 equipped with an autosampler and a column heater coupled to a Waters 474 Scanning Fluorescence Detector with a 5 μL flow cell set to λ_ex 250 nm and λ_em 395 nm and a Waters 996 photodiode array detector with an absorbance detection set to 254 nm (for reagent peak monitoring). Data acquisition and integration were carried out with Empower software.

The separation was achieved using an AccQ.Tag for a hydrolysate amino acid analysis column (3.9 × 150 mm, Waters) at 37 °C with a flow rate of 1 mL/min. Solvent A: AccQ.Tag Eluent A (Acetate-Phosphate buffer); Solvent B: H₂O:Acetonitrile (40:60, v/v). Two gradient elutions were used to determine the amino acids’ composition in this study. The first separation method, AccQ.Tag amino acid analysis, was used to determine the 20 natural amino acids. The second method was designed to get separation between AMQ (a byproduct of the AccQ.Fluor reagent) and Hydroxy proline.

The first gradient elution was set as follows: 0 to 0.5 min, 100 to 98% A; 0.5 to 1.5 min, 98 to 93% A; 15 to 19 min, 93 to 90% A; 19 to 32 min, 90 to 67% A; 32 to 33 min, 67% A; 33 to 34 min, 67 to 0% A; 34 to 37 min, 0% A; 37 to 38 min, 0 to 100% A; 38 to 50 min, 100% A for column equilibration. The second gradient elution was set as follows: 0 to 0.5 min, 100 to 96.5% A; 0.5 to 15 min, 96.5 to 93% A; 15 to 19 min, 93 to 90% A; 19 to 32 min, 90 to 67% A; 32 to 33 min, 67% A; 33 to 34 min, 67 to 0% A; 34 to 37 min, 0% A; 37 to 38 min, 0 to 100% A; 38 to 50 min, 100% A for column equilibration. The injection volume was 7 μL. Prior to the injection of the tested samples, a calibration standard mixture solution (Sigma-Aldrich, A2908) was injected three times, to ensure system repeatability and for determination and quantification purposes. In addition, two additional injections were done at the end of the sequence.

Sample preparation: Each sample was inserted into a disposable culture tube (6 x 50 mm ASTM Type 1, borosilicate glass) supplemented with HCl 6M (for amino acids analysis, Sigma) containing 0.1% phenol, flushed with nitrogen to reduce the oxygen concentration and sealed under vacuum (<20 mbar) and hydrolyzed for 2 h at 110 °C, in order to digest all the peptide bonds. The culture tubes were heated to 400 °C before their use, to avoid amino acid contamination. After the hydrolysis step, the samples were dried at ambient temperature under vacuum (<20 mbar) for 3 h. Then, 60 μL AccQ.Fluor Borate buffer (Waters) was added to each reconstituted sample and vortexed. Then, 20 μL labeling reagent (Waters) was added and each sample was vortexed and transferred to a HPLC vial, sealed and heated at 55 °C for 10 min.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.