Ultrasonic frequency and intensity dependence in ${\beta }_2$-microglobulin amyloid seed detection: Balancing cavitation-induced nucleation, fragmentation, and radical formation

Abstract

Amyloid fibrils are aberrant proteinaceous aggregates involved in several intractable diseases called amyloidosis. Detecting fragments of amyloid fibrils, so-called seeds, in human biofluids is a promising approach for their early-stage diagnosis. Ultrasonication of amyloidogenic protein solutions has the potential to enhance the detection sensitivity of amyloid seeds in an accelerated manner through the effects of ultrasonic cavitation. However, the effects of acoustic frequency and intensity on ultrasonic seed detection have not been investigated. In this study, we explore the optimized acoustic conditions for rapid and sensitive amyloid-seed detection and sonochemical mechanisms behind it. Our results show that maximum detection performance is achieved at moderate acoustic intensities with frequencies below 109 kHz, where the balance between ultrasonic enhancement of nucleation and fragmentation pathways of amyloid formation is suitable for the detection of amyloid seeds. Under these conditions, the detection time for amyloid seeds is reduced fivefold while achieving detection sensitivity of 1 pM. We further find that excessively high-intensity ultrasonication is likely to cause fragmentation of protein monomers into smaller peptides at lower frequencies, while that at frequencies above 200 kHz introduces reactive radical species, hindering the ultrasonic seed detection assay. These results highlight the critical role of optimizing acoustic parameters for the application of ultrasonication in the amyloid-seed detection, offering a pathway for rapid diagnostic assays for amyloidosis.

1. Introduction

Amyloid fibrils are aberrant proteinaceous aggregates characterized by an ordered structure and needle-like morphology [1], [2]. Once formed in vivo, these fibrils deposit on biological tissues and/or in the central nervous system, eventually leading to the onset of severe diseases known as amyloidosis, such as Alzheimer’s disease and Parkinson’s disease [3], [4]. Since amyloid fibrils cause irreversible damage to biological tissues after the onset of clinical manifestation, early diagnosis of amyloidosis is crucial for effective intervention in these diseases. However, reliable methods for the early diagnosis of amyloidosis have yet to be established.

From a physicochemical standpoint, amyloid fibrils are crystal-like aggregates formed from a supersaturated monomer solution [5], [6], [7], [8], [9], [10]. Macroscopically, their formation follows a nucleation-growth mechanism [11]. Primary nucleation is the rate-limiting step in which soluble monomers associate to form nuclei after a certain lag time. Once primary nucleation occurs, fibril growth proceeds rapidly as monomers are added to the ends of the fibrils. The seeding reaction further features the crystal-like nature of amyloid fibrils [12]; when preformed fibrils, commonly termed seeds, are introduced into the supersaturated monomer solution, amyloid formation is immediately initiated, as the seeds act as a template of fibrils to bypass the rate-limiting nucleation.

The formation of amyloid fibrils in vivo is a hallmark of the onset of amyloidosis; therefore, amyloid seeds are considered promising biomarkers for these diseases [13]. The seeding reaction has proven to be effective in detecting amyloid seeds in biological samples [13], [14]: Biological samples are added to the supersaturated monomer solutions, and the kinetics of amyloid formation is investigated. If seeds are present in the biological sample, they accelerate amyloid formation compared to samples without seeds due to the seeding reaction. The difference in the kinetics of amyloid formation allows us to detect the seeds in the biological samples. For example, Shahnawaz and co-workers successfully differentiated types of amyloidosis by analyzing the kinetics of amyloid formation caused by structural variations in cerebrospinal fluid seeds, using the protein misfolding cyclic amplification method [15], [16]. This technique amplifies seeds by shaking samples, allowing the detection of trace amyloid seeds in the samples of patients with Parkinson’s disease. However, it requires more than $150$ h for detection [16], which limits its clinical applicability. This limitation has driven researchers to develop methodologies aimed at reducing the detection time [17], [18], underscoring the importance of rapid and sensitive seed detection for clinical use.

Previous studies have demonstrated that ultrasonication affects amyloidogenic protein solutions by promoting both amyloid nucleation and fibril fragmentation [19], [20]. Specifically, ultrasonic irradiation of monomeric protein solutions accelerates amyloid formation by enhancing primary nucleation [21], [22], [23], [24], while ultrasonic irradiation of preformed fibrils induces their fragmentation into shorter fibrils [25], [26]. Moreover, in seeding reactions, ultrasonication improves detection sensitivity and shortens detection time compared with conventional shaking methods [20], [27], [28]. These effects are attributed to ultrasonic cavitation: cavitation bubbles generate air–water interfaces that locally concentrate protein monomers and facilitate nucleation [29], [30], [31], [32], [33], while their collapse imposes mechanical stress that fragments fibrils [25], [34]. Previous study indicated that the physical characteristics of cavitation bubbles affect the resultant seed-detection ability under acoustic fields: Several types of surfactants were used to modulate the surface tension of cavitation bubbles, and it was found that adding surfactants at an optimal concentration for seed detection improved the detection sensitivity by a factor of 100–1000. This finding provides important evidence that cavitation bubbles play a critical role in seed-detection assays under acoustic fields. It is well-known that the dynamics of acoustic cavitation is governed by frequency and intensity of the acoustic field [35], [36]. These facts lead to a presumptive hypothesis that change in the acoustic frequency and intensity for acoustic seed detection assay can influence the performance of the assay. However, the effects of acoustic frequency and intensity on seed-dependent amyloid formation have not yet been studied systematically.

In this study, we investigate the seed-dependent amyloid formation of ${\beta }_2$-microglobulin (${\beta }_2$m), a causative protein of dialysis-related amyloidosis [37], [38], under various acoustic fields to identify the optimum acoustic conditions for ${\beta }_2$m seed detection. For this purpose, we design and construct a dedicated sonoreactor equipping a Langevin-type piezoelectric transducer and fluorometer, enabling us to change acoustic frequency and intensity by replacing the transducer and changing the applied voltage. We evaluate the performance of the seed detection under various acoustic conditions in the frequency range of 26 to 450 kHz where the dominant effects of ultrasonic cavitation transition from mechanical to chemical [39]. Then, we explore the optimum conditions for the seed detection and discuss the sonochemical mechanism behind it using several analytical techniques.

2. Materials and methods

2.1. Preparation of the sample solution

The wild-type $\beta 2$m monomer was expressed in Escherichia coli (BL21) and purified as previously described [40]. Note that the ${\beta }_2$m monomer expressed in Escherichia coli has a methionine residue at the N-terminus in addition to the wild-type sequence. The purified $\beta 2$m was lyophilized and stored at $-20{}^{\circ }$C until just before the experiment. The lyophilized $\beta 2$m monomers were then dissolved in deionized water, after which the solution was filtered by a membrane filter with a pore diameter of $220\mathrm{nm}$ (Millipore, SLGVR04NL) to remove the preformed aggregates, if any. The concentration of the $\beta 2$m monomer was determined by measuring the absorbance at $280\mathrm{nm}$ using a molar extinction coefficient of $1.93\times 10^4$ M⁻¹cm⁻¹. In the amyloid-formation experiment, the sample solution includes 8.5 $\mathrm{\mu }$M (0.1 mg/mL) $\beta 2$m monomer, $20$ mM HCl, $80$ mM NaCl, $5\mu$M thioflavin-T (ThT), and different concentrations of $\beta 2$m seeds between $1$ pM and $10$ nM. Note that the seed concentration corresponds to the equivalent concentration of monomers constructing the seeds.

For the seed detection assay, the preparation of the seed solution should be carefully performed because of their reversible nature, especially under ultrasonication. When seeds at a concentration below their solubility limit are exposed to ultrasonic irradiation, they dissolve into monomers [41]. Therefore, seed solutions with a dilute concentration of less than 1 nM were added to the supersaturated monomer solutions within 30 minutes after dilution to avoid seed dissolution.

2.2. Construction of sonoreactor with fluorospectrometer

We developed the sonoreactor to irradiate multiple protein solutions with ultrasound (Fig. 1a). 135-$\mathrm{\mu }$L sample solutions were dispensed into 18 wells in a custom-made plastic plate, and then, the plastic plate was sealed by a plastic film with a thickness of $0.1\mathrm{mm}$ (WATSON, 547-KTS-HC). The plastic plate was immersed in a cylindrical water bath with the sealed side facing down. The water inside the bath was degassed by a deaerator (KAIJO, WRS-40006A), and its temperature was kept at $37$ °C by circulating the water through a temperature control system (TAITEC, Personal-11). The bottom plate of the water bath, which is made of stainless steel with a thickness of $2\mathrm{mm}$, is equipped with an ultrasonic transducer (Fig. 1b). The transducer was integrated with the stainless plate by spot-welded bolts and bonded with epoxy resin glue. We prepared three types of plate-integrated PZT transducers with fundamental frequencies of 26.0, 109, and $241$ kHz (KAIJO). By replacing these transducers, we performed the fibril-formation experiments with five different frequencies of 26.0, 47.6, 109, 241, and $450$ kHz to investigate the frequency effects over wide frequency range. We used higher modes of the transducers with fundamental frequencies of 26 and $241$ kHz for experiments with 47.6 and $450$ kHz, respectively, which were confirmed by acquiring the admittance spectra (Fig. 1c,d) using a network analyzer (NF Corporation, FRA51615). The voltage waveform generated by a synthesizer (NF Corporation, WF-1973) was amplified by a bipolar amplifier (NF Corporation, HSA4051) and applied to the transducer.

Fig. 1.

(a) Schematic illustration of the experimental system constructed for ultrasonic amyloid formation assays. (b) Photograph of the sonoreactor with a Langevin-type transducer with a fundamental frequency of 26.0 kHz and plastic plate containing sample solutions. The unit of the scale in the photograph is a centimeter. (c,d) Admittance spectra of ultrasonic transducers with fundamental resonant frequencies, $f_r$, of (c) 26.0 and (d) 241 kHz, respectively, which show higher mode resonance of the transducer. (e) Temperature change of sample solutions included in the plastic plate measured by thermocouples under various acoustic frequencies and applied voltages. The error bars indicate the standard deviation among 18 independent sample solutions.

We employed a ThT fluorescence assay for evaluating the time course of amyloid formation. Because the ThT molecule specifically binds to the $\beta$-sheet structure of amyloid fibrils and emits high-intensity fluorescence [42], an increase in the ThT fluorescence intensity corresponds to the increase in the amount of formed fibrils. The maximum excitation and emission wavelengths of the ThT fluorescence are 450 and 492 nm, respectively. For the fluorescence measurement, the sample solution was irradiated with an LED light with a center wavelength of $453\mathrm{nm}$ (Kobo Sawaki, FOLS-01). This LED light passed through a short-pass filter with the cut-off wavelength of $475\mathrm{nm}$ (Edmund, 84-691) before reaching the sample solution. The ThT fluorescence intensity of each sample solution was detected through a long-pass filter with the cut-off wavelength of $475\mathrm{nm}$ (Edmund, 84-737) and an optical fiber connected to a photomultiplier tube (Hamamatsu Photonics, H10722-210). The fluorescence signal received by the photomultiplier tube was collected using a digitizer (National Instruments, NI USB-5133) and transferred to a PC. We used Matlab (Version R2020b, Mathworks) to control the ultrasonic irradiation, rotation stage, and fluorescence acquisition. The fluorescence intensity of each sample solution was recorded every $3$ minutes during the amyloid-formation experiment. Here, we perform the amyloid formation under the acidic condition, which leads to decrease in the ThT fluorescence intensity of sample solutions compared to a neutral condition [43]. However, our system can adjust detection sensitivity of fluorescent signal from sample solutions by changing the bias voltage applied to the photomultiplier, so that we can perform the ThT fluorescence assays for a variety of samples. The assay conditions used in this study are summarized in Table S1.

2.3. Calibration of acoustic intensity through temperature measurement

The net input energy into the sample solution was determined based on the temperature increase of the solutions during ultrasonic irradiation. The sample plate was irradiated with ultrasound under various conditions for $50$ s, with 18 wells filled with 135 $\mathrm{\mu }$L of deionized water and equipped with thermocouples (Fig. S1a). The temperature change of the solution was analyzed by calculating the slope of the temperature curve during the initial 10 s of ultrasonication (Fig. S1b). Temperature changes under different acoustic intensities were measured by varying the voltage applied to the transducer at different frequencies (Fig. 1e). Using these data, we converted the applied voltages used in the experiments into the corresponding input power levels delivered to the sample solution.

2.4. Atomic force microscopy

For the atomic force microscopy (AFM) visualization of ${\beta }_2$m samples, the sample solution was diluted 10-fold with deionized water, and a droplet of the solution was placed on a mica plate. After drying, the surface of the mica was rinsed with deionized water, and then, the AFM image was acquired in tapping mode (HITACHI, AFM5000II).

2.5. Reversed-phase liquid chromatography

The ${\beta }_2$m monomer solution at a concentration of 0.1 mg/mL was subjected to ultrasonication for $10$ min with a duty ratio of 67% at 47.6 and $450$ kHz with various input power levels in a glass vial. The effects of the ultrasonication on the ${\beta }_2$m monomer were analyzed by means of reversed-phase chromatography. For the reversed-phase chromatography, we used a high-performance liquid chromatography system (GILSON) with a C4 300-Å column (5C4-AR-300, COSMOSIL). The eluted protein was detected by absorbance at 280 nm. The chromatographic data were collected with Clarity (Version 7, DataApex).

2.6. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis

The ${\beta }_2$m monomer solution at a concentration of 0.1 mg/mL was subjected to ultrasonication for $10$ min with a duty ratio of 67% at 47.6 and $450$ kHz with various input power levels in a glass vial. 16 $\mathrm{\mu }$L of each sample was mixed with 4 $\mathrm{\mu }$L of 5-fold concentrated sample buffer containing 2-mercaptoethanol applied to the 5%–15% gradient gel (Nacalai tesque, Kyoto, Japan) after heating at 95 °C for 10 min. Then, the electrophoresis was performed with the constant current of 20 mA.

2.7. Potassium iodide oxidation assay

Potassium iodide (KI) was dissolved in deionized water at a concentration of 100 mM. The KI solution was poured into a 2-mL glass vial and subjected to ultrasonication at a duty ratio of 67% with various input power levels at 47.6 kHz and 450 kHz. Then, the absorption spectra in the range of 260–480 nm of the KI solutions after ultrasonication were acquired by a spectrophotometer (U-3000, Hitachi) to evaluate the amount of generated radical species.

2.8. Liquid chromatography-mass spectroscopy

The ${\beta }_2$m monomer solution at a concentration of 0.1 mg/mL was subjected to ultrasonication for $10$ min with a duty ratio of 67% at 47.6 and $450$ kHz with various input power levels in a glass vial. The samples were analyzed using an LCQ Fleet mass spectrometer (ThermoFisher, Bremen, Germany) equipped with an HPLC. Salts and buffer components were removed using an AdvanceBio Desalting-RP column (Agilent Technologies, Santa Clara, CA) with a 10%–50% acetonitrile gradient containing 0.1% formic acid. Mass spectra were generated by integrating ion signals across the entire elution range.

3. Results and discussion

3.1. Ultrasonic amyloid formation assay from ${\beta }_2$m monomer solution

We first conducted an amyloid formation assay using the developed sonoreactor at 47.6 kHz (Fig. 1a,b). The ${\beta }_2$m monomer solution without preformed seeds was prepared, and all 18 wells of the sample plate were filled with the same monomer solutions to evaluate the uniformity of amyloid formation among different wells (Fig. 2a). The samples were exposed to ultrasonication at a 10% duty cycle, and ThT fluorescence intensity was monitored over 8 h (Fig. 2b,c).

Fig. 2.

(a) General scheme of ultrasonically accelerated amyloid formation assay using the developed sonoreactor coupled to a fluorospectrometer. Here, all the 18 wells are filled with ${\beta }_2$m monomer solutions: [${\beta }_2$m] $=$ 8.5 $\mathrm{\mu }$M, [HCl] $=$ 20 mM, [NaCl] $=$ 80 mM, and [ThT] $=$ 5 $\mathrm{\mu }$M. (b) The ThT time-course curves of ${\beta }_2$m monomer solutions under ultrasonication with the frequency and amplitude of 47.6 kHz and 29 mW, respectively. (c) Distribution of the lag time at each well position. The average and standard deviation of the lag time are 1.6 ± 0.2 h. (d) AFM image of formed amyloid fibrils. The scale bar denotes $1\mathrm{\mu }\mathrm{m}$.

The resulting ThT fluorescence curves exhibit a characteristic sigmoidal behavior: After an initial lag phase of $\sim 1.5$ h, the fluorescence intensity sharply increases, followed by a gradual decline. Such sigmoidal curves are typical indicators of amyloid fibril formation [44]. An AFM image taken after 8 h of ultrasonication confirms the presence of fibrillar structures in the ${\beta }_2$m samples (Fig. 2d), demonstrating that the developed sonoreactor successfully induces amyloid formation from ${\beta }_2$m monomers.

The lag time for amyloid formation in each well was evaluated to assess the uniformity of the reaction induced by ultrasonication among different wells. Here, the lag time is defined as the time at which the ThT fluorescence intensity reaches 20% of the maximum value. Fig. 2c shows the distribution of lag times across all wells, indicating that no positional dependence of lag times is observed. This result demonstrates that the axial symmetric design of the sonoreactor and the rotation movement minimizes positional non-uniformity in ultrasonic intensity. The lag time is 1.6 $\pm$ 0.2 h (coefficient of variation; CV $=$ 15.2%). The ${\beta }_2$m monomer solution used here does not form amyloid fibrils under quiescent conditions within 50 h (Fig. S2), indicating that the developed sonoreactor efficiently induces amyloid formation from ${\beta }_2$m monomers.

3.2. Ultrasonic seed detection at 47.6 kHz with different input power levels

Ultrasonic assays for ${\beta }_2$m seed detection at 47.6 kHz with different input power levels were performed to investigate the dependence of seed detection performance on acoustic intensity. Preformed ${\beta }_2$m amyloid fibrils were diluted and added to ${\beta }_2$m monomer solutions. The seed concentrations were 10 nM, 100 pM, and 1 pM, and the monomer solution without seeds was also prepared as a negative control.

Here, we determined this seed concentration range based on several previous studies, which demonstrated clinical success in detecting amyloid seeds from biological samples [16], [45]. For example, Okuzumi and co-workers succeeded in distinguishing patients with PD from healthy individuals by detecting amyloid seeds of $\alpha$-synuclein based on the assay with the seed detection sensitivity of approximately 69 pM [45]. In practice, the seed detection is carried out by analyzing the difference in the ThT kinetic curves [15]. Specifically, seeds in the sample solution are considered present if the mean lag time plus its standard deviation in seeded samples does not exceed the mean lag time minus its standard deviation in seed-free samples (Fig. S3).

We prepared sample solutions with seed concentrations across the clinically applicable range (1 pM–10 nM). The samples were dispensed into the sample plate, and the time course of ThT fluorescence intensity was monitored under various ultrasonic conditions (Fig. 3a). The resulting ThT fluorescence curves were analyzed across the different seed concentrations, and the detection sensitivity under each ultrasonic condition was evaluated as described below.

Fig. 3.

(a) General scheme of ultrasonic seed detection assay. The sample solution with three different seed concentrations and without seed is prepared and dispensed in the sample plate with 4 or 6 replicates. For all the sample solutions, the ThT time-course curves are acquired, and the lag times are calculated to investigate the acoustic-field dependency of the ultrasonic seed detection assay. (b-e) The ThT fluorescence curves of ${\beta }_2$m samples with various seed concentrations under 47.6-kHz ultrasonication with the input power level of (b) 0 mW (without ultrasonication), (c) 1 mW, (d) 10 mW, (e) 29 mW, and (f) 56 mW, respectively. For clear representation, these figures are shown without error bars. All curves and curves with error bars are shown in the supplementary information (Figs. S4 and S5). (g) Results of ultrasonic seed detection assay at 47.6 kHz with various input powers. The error bars denote standard deviation among independent samples ($n$$>$ 3).

We applied four different input power levels at 47.6 kHz and no ultrasonic irradiation as a negative control (Fig. 3b–f) and analyzed the lag times for all conditions (Fig. 3g). Here, it should be noted that, although the ThT time-course curves shown in Fig. 3b–f do not have error bars for visual clarity, the curves with error bars and individual curves are appeared in the supplementary information to show the data variability (Figs. S4 and S5). To compare seed detection performance across ultrasonic conditions, we defined three indices: (i) the detection limit, (ii) detection time, and (iii) discrimination sensitivity. (i) The detection limit is the lowest seed concentration detectable under each condition (see Fig. S3 for a graphical explanation). (ii) The detection time is the lag time of the seed sample with the lowest seed concentrations detectable ($La{g}_{low.}$). We also use the acceleration ratio as a relative metric of the detection time, which is the ratio of the lag time between the assay without ultrasonic irradiation and one with ultrasonication under a certain condition. (iii) The discrimination sensitivity is the ratio of the lag time of the sample without seeds ($La{g}_{mon.}$) to that of the 1 pM seed sample ($La{g}_{mon.}/La{g}_{low.}$). A higher discrimination sensitivity indicates clearer differentiation between seeded and unseeded samples, and thus greater detection performance. An optimized assay condition is characterized by a lower detection limit, a shorter detection time, and a higher discrimination sensitivity. These values are summarized in Table 1.

Table 1.

Results of the ultrasonic seed detection assay at various frequencies and input power levels. Note that the discrimination sensitivity could not be calculated for the condition without ultrasonication (0 mW) because the ThT fluorescence intensity did not increase for the sample solution without seeds.

Frequency	Power level	Detection limit	Detection time	Discrimination sens.	Acceleration ratio
w/o US	0 mW	1 pM	13.86 ± 2.20 h	N.A.	N.A.
26.0 kHz	3 mW	100 pM	2.16 ± 0.16 h	1.54	6.42
26.0 kHz	5 mW	1 pM	2.49 ± 0.11 h	3.32	5.57
26.0 kHz	14 mW	1 pM	2.73 ± 0.85 h	1.80	5.08
47.6 kHz	1 mW	1 pM	16.58 ± 3.56 h	1.67	0.84
47.6 kHz	10 mW	1 pM	2.07 ± 0.52 h	3.53	6.70
47.6 kHz	29 mW	100 pM	1.08 ± 0.83 h	1.03	12.8
47.6 kHz	56 mW	100 pM	0.92 ± 0.18 h	1.00	15.1
109 kHz	9 mW	1 pM	8.58 ± 1.59 h	1.51	1.62
109 kHz	17 mW	1 pM	5.37 ± 1.46 h	2.44	2.58
109 kHz	22 mW	1 pM	3.36 ± 0.95 h	2.08	4.13
109 kHz	38 mW	1 pM	3.27 ± 0.16 h	1.70	4.24

Without ultrasonication (Fig. 3b), the sample without seeds does not exhibit an increase in the ThT fluorescence intensity. As the seed concentration increases, an increase in ThT fluorescence intensity is observed, and the lag time for amyloid formation is shortened in a dose-dependent manner. For the sample with 1 pM seeds, the lag time is approximately 15 h. Ultrasonication with an input power of 1 mW (Fig. 3c) has no significant effects on amyloid formation of samples with and without seeds.

Ultrasonication at 10 mW significantly accelerates amyloid formation in both seeded and unseeded samples (Fig. 3d). Importantly, the ThT fluorescence curves of the seeded samples are clearly separated from those of the unseeded samples, indicating successful detection of 1 pM seeds with accelerated kinetics. The detection time is reduced by approximately sevenfold compared to that without ultrasonication, and the discrimination sensitivity is the highest among all tested conditions at 47.6 kHz (Table 1). Further increases in ultrasonic intensity (Fig. 3e,f) mainly accelerate amyloid formation in the unseeded samples rather than in the seeded ones, likely because seeded samples form fibrils almost immediately after the reaction starts, leaving little opportunity for additional acceleration.

Under 56 mW ultrasonication, the ThT fluorescence curves exhibit an abnormal decrease after reaching maximum intensity (Fig. 3f). Such an abrupt decrease in ThT fluorescence has been previously reported [46]. It was indicated that the decrease in the ThT fluorescence intensity after reaching a maximum intensity is likely due to formation of ThT-negative intermediate aggregates under excessive ultrasonic irradiation. The stronger the input power of ultrasonic irradiation, the higher the tendency of the denaturation, implying that this denaturation is presumably caused by mechanical effects of ultrasonic cavitation [46]. We discuss this phenomenon later and conclude that it is presumably due to fragmentation of ${\beta }_2$m monomers into shorter peptides by the mechanical effects of ultrasonic cavitation.

The performance of the ultrasonic seed detection assay at 47.6 kHz is optimized at a moderate input power level. At low intensity, the effects of ultrasonication on amyloid formation are negligible, providing no benefit, for example, a shortened detection time. At moderate intensity, ultrasonic irradiation reduces the detection time by approximately sevenfold compared to without ultrasonication, while maintaining high detection sensitivity down to 1 pM with clear discrimination. Under this condition, the balance between ultrasonic enhancement of nucleation and fragmentation is optimized for the seed detection, as discussed later). In contrast, at high intensity, ultrasonication accelerates amyloid formation even in the absence of seeds, resulting in amyloid formation independent of seeds. Clinically, formation of seed-independent amyloid fibrils can cause a false positive, which means that samples without seeds are incorrectly judged as seed-containing ones. This can occur because ultrasonic irradiation may induce the formation of seed-independent amyloid fibrils through nucleation enhancement, even in seed-free samples.

3.3. Ultrasonic seed detection at frequencies other than 47.6 kHz with different input power levels

We also performed the ultrasonic seed detection assay at ultrasonic frequencies of 26.0, 109, 241, and 450 kHz with various input power levels. Under 26.0 kHz (Fig. 4a-d: The curves with error bars are again shown in the supplementary information, Figs. S6 and S7.), the 5-mW ultrasonication shows clearer separation of samples with 1-pM seeds from samples without seeds than 3- and 14-mW ultrasonication. In the case of 5 mW, the detection limit of 1 pM is achieved with a detection time of $\sim$2.5 h, and the discrimination sensitivity is 3.32, which is similar to the best result obtained at 47.6 kHz. The tendency for seed detection to be optimized at moderate intensity is consistent with the results at 47.6 kHz.

Fig. 4.

(a–c) ThT time-course curves of ${\beta }_2$m samples with various seed concentrations under 26.0-kHz ultrasonication at input power levels of (a) 3 mW, (b) 5 mW, and (c) 14 mW. (d) Results of the ultrasonic seed detection assay at 26.0 kHz with various input power levels. Error bars denote standard deviation among independent samples ($n>3$). (e–h) ThT fluorescence curves of ${\beta }_2$m samples with various seed concentrations under 109-kHz ultrasonication at input power levels of (e) 9 mW, (f) 17 mW, (g) 22 mW, and (h) 38 mW. (i) Results of the ultrasonic seed detection assay at 109 kHz with various input power levels. Error bars denote standard deviation among independent samples ($n>3$). For clarity, the ThT time-course curves are shown without error bars. All curves, including those with error bars, are shown in the supplementary information (Figs. S6–S9).

Under 109 kHz (Fig. 4e-i: The curves with error bars are again shown in the supplementary information, Figs. S8 and S9.), we tested four input power levels. All four conditions achieve the detection limit of 1 pM. Increasing the input power level shortens the detection time of 1-pM seeds. The discrimination sensitivity reaches the maximum value of 2.44 when the input power is 17 mW, moderate intensity among tested conditions. These results are consistent with those at 26.0 and 47.6 kHz.

Under 241 kHz and 450 kHz (Figs. S10-13), the performance of the assay deteriorates; the ThT fluorescence curves exhibit large intensity deviations. This is due to the subtle increase in ThT fluorescence intensity. Furthermore, the fluorescence increase is not observed in wells where a fluorescence increase is expected. The mechanism by which the assay performance declines at frequencies above 200 kHz is associated with the ultrasonic generation of radical species [47], [48]. Several previous studies have demonstrated that the generation of radical species is more efficient at ultrasonic frequencies above approximately 200 kHz compared with lower frequencies ($<$ 100 kHz) [49], [50], [51]. According to the Minnaert resonance relation, $f_r=\frac{1}{2\pi R_0}\sqrt{\frac{3\gamma P_0}{\rho }}$ [52], the resonance radius of cavitation bubbles decreases with increasing frequency. Therefore, at low frequencies, larger cavitation bubbles can grow and collapse more violently, producing intense mechanical stress. In contrast, at higher frequencies, smaller bubbles are formed in greater numbers, which mainly contribute to chemical effects through more uniform and frequent collapses. In our experimental setup, the threshold at which radical generation becomes significant is approximately above 200 kHz, which will be shown in detail through several experimental results later.

In summary, the ultrasonic seed detection assay at five frequencies with different input power levels shows that ultrasonication at moderate intensity and frequencies below 109 kHz effectively shortens the time required to detect amyloid seeds while maintaining clear separation between samples containing a small amount of seeds and those without seeds. On the other hand, ultrasonication at frequencies above 200 kHz disrupts the assay. We further investigate the sonochemical mechanisms underlying these results.

3.4. Optimum condition for ultrasonic seed detection and its underlying mechanism

Previous studies have shown that ultrasonication of solutions containing amyloidogenic proteins primarily accelerates two pathways, namely primary nucleation and fragmentation (Fig. 5a) [20], [25]. For sensitive seed detection, the desirable effects of ultrasonication are extensive fragmentation of pre-existing seeds without enhancement of primary nucleation. Fragmentation of mature fibrils produces numerous short fibril fragments [25], thereby increasing the number of fibril termini available for monomer addition and subsequent elongation. Thus, enhancement of the fragmentation pathway facilitates rapid and sensitive seed detection through amplification of seeds. In contrast, enhancement of the primary nucleation pathway deteriorates detection sensitivity [20], as it promotes the formation of seed-independent fibrils from monomers, potentially leading to false-positive results.

Fig. 5.

(a) Reaction scheme of amyloid formation under ultrasonication. Red arrows indicate the reaction pathways enhanced by ultrasonication. (b–e) AFM images of ${\beta }_2$m fibrils (b) before ultrasonication and after ultrasonication at 47.6 kHz with input power levels of (c) 19 mW, (d) 42 mW, and (e) 56 mW. Scale bar denotes $1\mathrm{\mu }\mathrm{m}$. (f) Length of fibrils after ultrasonication with various input power levels. Each plot shows the mean $\pm$ standard deviation. Fibril lengths were analyzed over 100 fibrils. (g) Lag time for amyloid formation from monomer solutions under various input power levels. Each plot shows the mean $\pm$ standard deviation. (h) Schematic representation of the optimum input power for seed detection, where the balance between enhancement of nucleation and fragmentation is optimized for sensitive and rapid detection.

We examined the dependence of ultrasonic enhancement of aforementioned two pathways on input power levels, considered separately with respect to the fragmentation and nucleation pathways. Preformed mature fibrils were subjected to ultrasonication at 47.6 kHz at various input power levels, and the resulting fibril lengths were analyzed by AFM imaging (Fig. 5b–f).

The average length of ${\beta }_2$m fibrils prior to ultrasonication is approximately $1\mathrm{\mu }\mathrm{m}$ (Fig. 5b,f). Under ultrasonication at 19 mW, the fibril length remains essentially unchanged (Fig. 5c,f), suggesting that fragmentation is negligible under this condition. By contrast, fibrils exposed to 42-mW and 56-mW ultrasonication exhibit short fragment morphologies with an average length of approximately 150 nm (Fig. 5d,e). These observations indicate that ultrasonic enhancement of the fragmentation pathway follows a threshold-dependent behavior: When the input power level is below a certain threshold, no significant fragmentation of mature fibrils occurs, whereas once the threshold is exceeded, ultrasonication leads to fragmentation into short, particle-like structures. Notably, further increases in input power do not reduce fibril length beyond this point.

In contrast to fragmentation, the degree of ultrasonic enhancement of primary nucleation monotonically increases as a function of the input power level. This is indicated by the result that the lag time of amyloid formation from the monomer solution is monotonically shortened as the input power level increases (Fig. 5g). By taking into account these two different input-power-level dependencies, the optimum condition for accelerated detection of amyloid seeds, while maintaining clear discrimination from samples without seeds, should appear at a moderate input power (Fig. 5h).

At input power levels below the fragmentation-enhancement threshold (Fig. 5h(i)), the effect of ultrasonication is negligible. When the input power level is slightly above the fragmentation threshold (Fig. 5h(ii)), ultrasonication generates short fibril fragments during the seed-detection assay, leading to rapid amplification of pre-existing seeds. In this power range, ultrasonication also enhances primary nucleation, but samples with seeds remain clearly distinguishable from those without seeds, as fragmentation enhancement is the dominant effect. Therefore, this power range is optimum for rapid and sensitive seed detection under ultrasonication. At higher input powers (Fig. 5h(iii)), no further enhancement of fragmentation occurs, and only nucleation pathway is further promoted, deteriorating seed detection sensitivity by producing seed-independent fibrils from monomers.

In summary, the optimal input power of ultrasonication for seed detection is determined by a balance between ultrasonic enhancement of fragmentation and primary nucleation, which are favorable and unfavorable for sensitive detection, respectively. Our results indicate that this trend persists below 109 kHz, suggesting no significant frequency dependence of seed-detection performance within this range. In contrast, at frequencies above 241 kHz, the ultrasonic seed-detection assay did not yield reproducible results due to the generation of radical species, as discussed below. Although no clear optimum frequency for seed detection was found below 109 kHz, the upper limit of 109 kHz simply reflects the range of acoustic frequencies examined in this study rather than a fundamental threshold. These results suggest that the seed detection assay should be conducted under acoustic conditions where extensive radical generation does not occur, and that, in the absence of radical species, the assay performance is primarily determined by acoustic intensity. The underlying mechanism is further examined in the following section.

3.5. Frequency-dependent inactivation mechanisms of ${\beta }_2$m monomers by excessive ultrasonication

We investigate the mechanism by which the ultrasonic seed detection assay fails to detect amyloid seeds at high frequencies above 241 kHz. Furthermore, we examine the behavior of protein molecules under ultrasonic conditions, where the ThT fluorescence curves abruptly decrease after reaching their maximum at low frequencies below 100 kHz (Fig. 3f). Ultrasonic frequencies of 47.6 kHz and 450 kHz were employed as representative low- and high-frequency conditions, respectively. The input power levels were set to 56 mW and 134 mW at 47.6 kHz and 450 kHz, respectively, which represent the maximum input power levels of our experimental setup.

A 2-mL solution of ${\beta }_2$m monomers (8.5 $\mathrm{\mu }$M) was poured into a glass vial and sonicated for 10 minu with a duty cycle of 1.0-s ultrasonication followed by a 0.5-s pause (Fig. 6a). During the 10-min intermittent ultrasonication, samples were fractionated at different time points and mixed with ThT dye. The mixed samples were then incubated with shaking for 60 h, and their ThT fluorescence time courses were monitored to investigate the effects of excessive ultrasonication on amyloid formation by ${\beta }_2$m monomers. Notably, ultrasonication was applied only to alter the monomer state prior to amyloid formation, and amyloid formation itself was monitored under shaking without ultrasonication. This approach allows us to isolate the effects of excessive ultrasonication on the ability of ${\beta }_2$m monomers to form amyloid fibrils.

Fig. 6.

(a) Schematic illustration of the assay performed to investigate the inactivation behavior of ${\beta }_2$m monomers upon excessive ultrasonication. The monomer solution is subjected to ultrasonication and then fractionated at various time points. Each fraction is mixed with ThT dye, and amyloid formation kinetics are monitored via ThT fluorescence. (b,c) Time-course curves of ThT fluorescence intensity for ${\beta }_2$m monomers exposed to excessive ultrasonication at (b) 47.6 kHz and (c) 450 kHz. Durations in the legend indicate the exposure times to ultrasonication.

Without ultrasonication, ${\beta }_2$m monomers form amyloid fibrils under shaking after a lag time of approximately 20 h (Fig. 6b,c). In contrast, excessive ultrasonication of the monomers prolongs the lag time and reduces the final ThT fluorescence of amyloid fibrils at both low and high frequencies (Fig. 6b,c). When the irradiation time exceeds 5 min, amyloid formation from the monomers is completely suppressed. These observations indicate that excessive ultrasonication renders the ${\beta }_2$m monomers incapable of amyloid formation.

We next analyze the ${\beta }_2$m monomers subjected to 10-min ultrasonication at various input power levels using reversed-phase chromatography (Fig. 7a,b), SDS–PAGE (Fig. 7c), and liquid chromatography–mass spectrometry (LC–MS) (Table 2 and Fig. S14). At the low frequency, ultrasonication at input power levels below 19 mW does not alter the elution profile of intact ${\beta }_2$m monomers in the chromatogram (Fig. 7a) and the corresponding SDS–PAGE band pattern (Fig. 7c). Under these conditions, the molecular weight determined by LC–MS matches that of intact ${\beta }_2$m (11,860 Da) [53] (Table 2). These findings demonstrate that ultrasonication under such mild conditions does not modify the chemical structure of ${\beta }_2$m monomers, consistent with the reproducible induction of amyloid formation observed in seed detection experiments.

Fig. 7.

(a,b) Reversed-phase chromatograms of ${\beta }_2$m monomers after 10 minutes of ultrasonication at different input power levels: (a) 47.6 kHz and (b) 450 kHz. The control (Ctrl.) shows the chromatogram of the intact monomer without ultrasonication. (c) SDS-PAGE analysis of ${\beta }_2$m monomers after ultrasonication at different input power levels at 47.6 kHz and 450 kHz. Numbers below the gel indicate the input power used for each sample. The control (Ctrl.) denotes the intact monomer without ultrasonication. (d,e) Absorption spectra of potassium iodide (KI) after 10 minutes of ultrasonication at various input power levels: (d) 47.6 kHz and (e) 450 kHz. The inset in panel (d) shows a magnified view of the spectra. The control (Ctrl.) denotes the spectrum of KI solution without ultrasonication. (f) Schematic illustration of the inactivation mechanism through which ${\beta }_2$m monomers lose the ability to form amyloid fibrils upon excessive ultrasonication at 47.6 kHz and 450 kHz.

Table 2.

LC-MS analysis of ${\beta }_2$m monomer molecular weights after 10 min ultrasonication at various input power levels (47.6 kHz and 450 kHz). The intact ${\beta }_2$m monomer has a molecular weight of 11,860 Da [53]. Values marked with asterisks (*) are unreliable because of low peak intensities, suggesting that ${\beta }_2$m monomers may have been fragmented into shorter peptides of random lengths.

Frequency (kHz)	Power level(mW)	Molecular weight (Da)
47.6	1	11,863
47.6	10	11,861
47.6	19	11,863
47.6	42	11,173/11,908${}^{\ast }$
47.6	56	11,967${}^{\ast }$
450	5	11,864
450	20	11,909
450	48	11,911
450	97	11,909
450	134	11,895

At the low frequency, high-intensity ultrasonication (42 and 56 mW), corresponding to the conditions which caused abrupt decrease in ThT fluorescence intensity (Fig. 3f), shifts the reversed-phase chromatogram peaks from those of intact ${\beta }_2$m monomers (Fig. 7a) and produces smeared bands in SDS–PAGE (Fig. 7c). These observations indicate that ultrasonication modifies the chemical structure of the monomers. Consistently, LC–MS spectra reveal the disappearance of the intact monomer peak (Fig. S14), implying decomposition of ${\beta }_2$m into shorter peptides. KI oxidation assay [54] suggests that radical generation under these conditions is negligible (Fig. 7d), indicating that mechanical effects of cavitation collapse, such as shock waves, is the primary cause of fragmentation [55], [56]. Importantly, fragmentation of ${\beta }_2$m monomers differs fundamentally from fibril fragmentation described in Fig. 5a. Monomer fragmentation involves cleavage of intramolecular peptide bonds within a single monomer, whereas fibril fragmentation shortens fibrils without altering monomer chemistry by disrupting the intermolecular hydrogen bonds stabilizing the fibrillar assembly.

At the high frequency (450 kHz), where the ultrasonic seed detection assay fails to yield reproducible results, ultrasonication induces a shift in the intact chromatographic peak under all conditions except at the lowest input power level (5 mW) (Fig. 7b). Strikingly, above 20 mW, the shifted peaks exhibit nearly identical profiles regardless of input power levels, in contrast to 47.6 kHz, where higher input power levels cause progressive peak broadening. Consistently, smeared bands appear in SDS–PAGE under conditions where the chromatographic peaks are shifted (Fig. 7c). Unlike at the low frequency, the KI oxidation assay confirms the generation of radical species during ultrasonication at the high frequency (Fig. 7e). Absorption spectra of KI solutions exposed to powers above 20 mW are nearly identical, indicating that radical production reaches a plateau at this threshold. This behavior is consistent with previous reports showing that, above a certain acoustic power, radical production hits a plateau likely due to bubble coalescence, gas depletion, and acoustic attenuation effects [50], [57]. The plateau correlates with the uniform chromatographic profiles, suggesting that radical-mediated chemical modification of monomers underlies the observed spectral changes.

LC–MS analysis supports this interpretation (Table 2 and Fig. S13): Although the main molecular-weight peak persists even at powers above 20 mW, it shifts slightly to a higher mass (＋50 Da). Such a shift can be attributed to covalent addition of radical species (e.g., O$\cdot$ or $\cdot$OH) to the side chains of ${\beta }_2$m monomers. Amyloid fibrils exhibit a tightly packed atomic configuration within the fibril core, involving various side chain, as revealed by the cryo-electron microscopy [58]. The covalent addition of radical species to these side chains presumably sterically prevent ${\beta }_2$m monomers from adopting the amyloid fibril conformation. Another possibility is the reduction of the intramolecular disulfide bridge of denatured ${\beta }_2$m monomers. The native ${\beta }_2$m monomer possesses a disulfide bridge between Cys25 and Cys80, which remains even under acidic denaturing conditions. This disulfide bond is essential for amyloid formation of denatured ${\beta }_2$m monomers [59]. Reduction of this bond by radical species [60] would increase configurational entropy of the peptide chain and thus thermodynamically favor the unfolded state over the amyloid state, leading to fibril formation inhibition. Taken together, these findings strongly suggest that the loss of reproducibility in ultrasonic seed detection assays above 200 kHz arises from radical generation and subsequent covalent modification of ${\beta }_2$m monomers.

In summary, excessive ultrasonication of ${\beta }_2$m monomer solutions leads to monomer inactivation, thereby compromising the reproducibility of the ultrasonic seed detection assay (Fig. 7f). Notably, the underlying mechanisms differ between frequencies. At the low frequency, inactivation arises from fragmentation of protein molecules at the amino-acid level, likely driven by mechanical stress associated with the violent collapse of transient cavitation bubbles [61]. In contrast, at the high frequency, inactivation occurs once the acoustic-field intensity exceeds the threshold for radical generation. The resulting covalent modifications render ${\beta }_2$m monomers incompetent for amyloid assembly, eventually impairing assay reproducibility.

3.6. Challenges in transferring acoustic seed-detection methods to clinical diagnostics

To transfer acoustic seed-detection methods to clinical diagnostics, several challenges must be addressed in further sonochemical research. In clinical settings, the throughput of the assay (i.e., the number of samples that can be processed in parallel) is critical for both reducing assay time and ensuring statistical reliability. In this study, we developed a sonoreactor dedicated to investigating the acoustic field dependency of seed-dependent amyloid formation, which can analyze up to 18 samples simultaneously. To enhance assay throughput, we have also developed several automated sonoreactors compatible with commercially available 96-well plates [20], [62], which have been partially applied in clinical studies [23]. In addition, controllability of amyloid formation according to study objectives is essential for reliable seed-detection assays using clinical samples. Clinical specimens such as cerebrospinal fluid, serum, and plasma exhibit diverse environmental conditions. Therefore, studies using acoustic seed-detection assays under molecularly crowded conditions, mimicking real biological environments, are important for optimizing acoustic parameters for clinical samples. Although research on acoustic seed detection in crowded environments remains limited, current results suggest that ultrasonic methods possess superior seed-detection capability even in solutions containing high concentrations of serum albumin, compared with conventional shaking-based methods [28]. In summary, further development of sonoreactors to enable high-throughput processing of biological samples, together with expanded sonochemical studies on seed detection under physiological conditions, will be essential for translating this technology to clinical diagnostics.

4. Conclusion

In this study, we established the optimal conditions for detecting ${\beta }_2$m amyloid seeds under controlled ultrasonication using a custom-built sonoreactor. At 26.0, 47.6, and 109 kHz, moderate acoustic intensities provided an optimized balance between ultrasonic enhancement of nucleation and fragmentation, enabling reliable detection of clinically relevant seed concentrations ($\sim$1 pM) with a fivefold reduction in assay time. In contrast, excessive ultrasonication at these frequencies was likely to fragment ${\beta }_2$m monomers into short peptides, leading to loss of seeding competence due to fragmentation. At higher frequencies (241 and 450 kHz), the generation of radical species chemically modified ${\beta }_2$m monomers, rendering them incompetent for amyloid formation and thereby compromising assay reproducibility. Together, these findings highlight the frequency- and intensity-dependent effects of ultrasonication on amyloidogenic proteins and provide insight into defining effective operational acoustic conditions for amyloid seed detection assays in clinical applications.

CRediT authorship contribution statement

Kichitaro Nakajima: Writing – review & editing, Writing – original draft, Validation, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Kakeru Hanada: Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Data curation. Masatomo So: Formal analysis, Data curation. Keiichi Yamaguchi: Writing – original draft, Formal analysis, Data curation. Yuji Goto: Writing – review & editing, Validation, Supervision, Conceptualization. Hirotsugu Ogi: Writing – review & editing, Supervision, Project administration, Methodology, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary material related to this article.

MMC S1

The supplementary material includes supplimentary figures and tables, Fig. S1-S14 and Table S1, S2.