Using gentle ultrasound to break down tiny amyloid clots in a lab-on-chip model

Porcine plasma was obtained from Innovative Research, Inc. (United States; No. IGPCPLANAC500ML) and stored at −20°C until use. Plasma aliquots (1 mL) were used for microclot formation. rtPA was purchased from Sigma-Aldrich (MilliporeSigma, United States; No. T0831) and reconstituted according to the manufacturer's protocol to achieve a final working concentration of 3 μg/mL. Thioflavin T (ThT) was purchased from Sigma-Aldrich (St. Louis, MO, United States; No. T3516). Vevo MicroMarker^® contrast agents (FUJIFILM VisualSonics Corp., Japan; No. 11678) were used as exogenous microbubbles and were diluted in PBS buffer per manufacturer protocol.

Amyloid fibrin(ogen) microclots were generated by subjecting porcine plasma to repeated freeze-thaw cycles, followed by incubation at 37°C. Plasma aliquots (1 mL) underwent 12 freeze-thaw cycles, with each cycle consisting of freezing at −20°C and thawing at 37°C for 2 hours. The aliquots were diluted 1.5 x before the ultrasound treatment.

The LOC device was designed to replicate a 6 mm diameter popliteal vein, providing a controlled environment for targeted ultrasound exposure of amyloid microclots. The popliteal vein was selected due to its accessibility for transcutaneous ultrasound exposure rendering it a relevant site for potential clinical applications of ultrasound. The device molds were designed using Onshape (version 1.188, PTC Inc., United States) CAD software and fabricated via digital light processing (DLP) technology using a 285 D Printer (CADWorks3D, Canada). One side of the channel was made slightly flat (5.8 mm instead of a full 6 mm circle) to create a smooth surface for bonding the silicone layer. This made it easier to peel off the PDMS and allowed better imaging, while still keeping the overall shape close to that of a popliteal vein. After UV treatment, Momentive RTV615 silicone elastomer (RS Hughes, United States) was cast into the molds and cured following manufacturer-recommended conditions to form the microfluidic channels. The cured elastomer was then bonded onto a 20 µm silicone sheet (Gteek, United Kingdom) using plasma surface treatment.

The schematic representation of the LOC platform, transducer arrangement, and acoustic bath setup is shown in Figure 1. The experimental setup consisted of a custom-designed and 3D printed acoustic water bath, which housed the LOC device, a piezoelectric transducer, and supporting components. The transducers were mounted at the bottom of the bath. The LOC chip was mounted on top of the water bath using a custom-designed holder, securing its position at the pre-measured focal plane of the ultrasound transducers to maintain consistent acoustic exposure across all experimental conditions.

A function generator (DG4162, Rigol Technologies Co., Ltd., China) controlled the ultrasound frequency generated continuous stepped pulse waveform, while an amplifier (Model 1040L, Electronics & Innovation, Ltd., United States) regulated the voltage to the transducers. The bath was filled with degassed water as a coupling medium for ultrasound wave transmission while minimizing acoustic attenuation.

Experiments were conducted under static flow conditions in the LOC platform. Ultrasound was applied at four different frequencies: 150, 300, 500 kHz, 1 MHz. The mechanical index (MI) is defined as MI = PNP √f , where PNP is the measured peak negative pressure (MPa), and f is the ultrasound frequency (MHz). The MI was maintained at 0.3 for all the experiments by adjusting the ultrasound pressure accordingly for all frequencies which corresponds to PNP of 1 MHz = 0.30 MPa, 500 kHz = 0.21 MPa, 300 kHz = 0.16 MPa, and 150 kHz = 0.11 MPa. The thin silicone membrane at the base of the chip allows acoustic waves to enter the microfluidic chamber while minimizing attenuation. Inside the LOC channel, ultrasound waves interact with the plasma and amyloid microclots, generating acoustic streaming, cavitation, and shear forces. These interactions induce microstreaming and localized mechanical forces, which contribute to the mechanical fragmentation and potential enzymatic enhancement of clot lysis.

Thioflavin T was used as a fluorescent dye to stain amyloid fibrils. A stock solution of 30 mM Thioflavin T was prepared in phosphate-buffered saline (PBS), pH 7.4. The solution was filtered using a 0.2 µm syringe filter to ensure clarity. Samples were incubated with Thioflavin T at final concentration of 30 μM room temperature for 30 min prior to imaging. Amyloid microclot lysis was evaluated based on clot count reduction and size reduction. The microclots morphology and count were monitored before and after ultrasound exposure, and changes in clot size were analyzed using fluorescence microscopy. This concentration was selected based on previously published studies demonstrating that 30 μM offers an optimal balance between signal intensity and minimal fluorescence quenching (Mai et al., 2024; Sakharov et al., 2000; Xue et al., 2017).

Imaging was conducted using the Countstar BioTech Automated Cell Counter (Model IC1000), which uses epifluorescence microscopy with a fixed-focus optical system and a 5-megapixel CMOS camera. A ×10 objective lens was used to capture the fluorescent microclots, and a GFP filter was used for ThT-labeled samples. Image analysis was performed using segmentation and particle detection plugins in ImageJ software, which measured clot area and count. Equivalent circular diameter (D_eq) was calculated from the measured area for each microclot and analyzed in a custom Python script.

All statistical analyses were conducted using GraphPad Prism version 10.4.1. Group comparisons for clot count and average clot diameter were performed using one-way ANOVA. Sidak's post hoc multiple comparisons test was used to identify pairwise differences between groups. A p-value of less than 0.05 was considered statistically significant.

Each experimental group included three biologically independent samples (n = 3), except for Figure 2, where two replicates (n = 2) were analyzed. For each replicate, three non-overlapping microscopic fields of view were imaged. Every image contained thousands of individual microclots. Clot count and diameter were quantified using ImageJ version 2.16.0 with particle segmentation and custom Python scripts. Diameter was determined using effective diameter to estimate the equivalent size for irregular shapes. Data are reported as mean ± standard deviation (SD).

Development of a reproducible amyloid fibrin microclot model is essential for investigating their role in clinical disorders including Long COVID, Alzheimer's, type 2 diabetes (Kell et al., 2025; Turner et al., 2023a). Various in vitro methods have been reported to produce amyloid microclots including drying, heating, exposure to acidic pH, hydrophobic surfaces, and organic solvents (Mai et al., 2024).

In this study, we used the freeze-thaw method and incubation at biological temperature to induce amyloidogenesis. Freeze-thawing results in ice-water interfaces in plasma and introduces structural stress into proteins which could induce protein misfolding and aggregation (Bader et al., 2014). Moreover, the incubation at room temperature can promote amyloidogenesis through oxidative stress and spontaneous fibrinogen misfolding that ultimately enhances the formation of amyloid fibrin microclots. Freeze-thaw-based amyloid formation is a cell and enzyme-free approach. In this method, protein structural modification results from intrinsic protein dynamics rather than external biochemical interactions, offering reproducibility and control over aggregation conditions (Graether et al., 2003; Filippov et al., 2008).

To verify amyloid formation by freeze-thawing, we employed the Thioflavin T (ThT) dye, which selectively binds to β-sheet-rich amyloid fibrils and enhances fluorescence upon interaction. Fluorescent ThT staining is widely used in amyloid research as a reliable marker for detecting the amyloid transition of fibrinogen, owing to its high specificity for cross-β structures (Turner et al., 2023a; Kell et al., 2022; Laubscher et al., 2021; Pretorius et al., 2016).

Figure 2A shows the fluorescence microscopy images of freeze-thawed plasma samples stained with ThT. In the earlier cycles, there are no visible ThT-labeled amyloid structures. However, green, fluorescent ThT-labeled microclots emerge starting from cycles 7, indicating a gradual amyloid microclot formation. By cycle 8, a marked increase in ThT-tagged microclots can be observed which shows significant amyloid microclot formation and with each cycle more amyloid microclots appear in the sample. Figure 2B shows the quantitative clot counts as a function of cycles, confirming a significant increase in amyloid microclots after cycle 7. This cycle-dependent increase in amyloid microclot counts confirms amyloidogenesis by repeated freeze-thaw cycles and incubation. The formation of microclots with repeated freeze–thaw cycles and incubation are previously reported, showing that freeze-thaw processes generate ice-water interfaces that impose structural stress on plasma proteins and lead to misfolding and aggregation. Fibrinogen, in particular, has been shown to adopt abnormal conformations under these conditions and covert into amyloid-like forms (Graether et al., 2003; Filippov et al., 2008; Cao et al., 2003). This experiment was repeated twice to ensure reproducibility. However, since plasma contains natural proteins, batch-to-batch variations are inherent. As a result, the data presented here is primarily aimed to show the trend of increasing clot formation with successive cycles rather than serving as an exact reference for clot numbers. Although we use a particle detection algorithm to identify discrete microclots, accumulation of submicron amyloid clots can lead to diffuse green fluorescence across the field. These clots are likely too small or diffuse to be detected as distinct particles but still bind Thioflavin T (ThT) and result in a uniform background signal. This background signal may reflect early-stage amyloid formation occurring before the assembly of detectable microclots.

After amyloid microclot formation, we explore how different ultrasound frequencies impact both microclot morphology and count. The different mechanical interactions of ultrasound with microclots are frequency-dependent, as acoustic forces such as acoustic radiation force and acoustic streaming exhibit varying levels of dominance based on their inherent frequency ranges (Rasouli et al., 2023). The balance between these forces dictates whether ultrasound leads to microclot fragmentation, which is our aim, or instead results in displacement or even aggregation of the microclots. Hence, identifying the optimal frequency and experimental conditions for effective clot disruption is crucial for the potential therapeutic applications of our ultrasound platform. To this end, we tested the effect of different ultrasound frequencies on the lysis of microclots by subjecting them to 150, 300, 500 kHz, and 1 MHz.

This experiment was also repeated in the presence of rtPA, a serine protease used as a thrombolytic agent to dissolve blood clots. rtPA works by binding to fibrin in microclots and then converting plasminogen into plasmin, which subsequently dissolves the clot through a process known as fibrinolysis. We sought to determine whether ultrasound increases the rate of clot dissolution when rtPA is used and whether a possible synergistic effect of the acoustic forces and pharmacological thrombolysis can be observed.

Figures 3A, B show fluorescence microscopy images of ThT-tagged microclots treated with different ultrasound (US) frequencies for US-only and US + rtPA groups. The control samples show a high density of large (over 30 microns) amyloid microclots, which are uniformly distributed across the image. Upon exposure to 150 kHz waves, the microclots exhibit a marked reduction in size. At higher frequencies, however, less fragmentation can be observed, and the clots remain intact.

To quantify microclot lysis under US treatment, image analysis was performed, and the average clot diameter and count of clots above 30 μm at each testing condition were calculated for three samples to ensure statistical robustness. Figures 3C, D show the average clot diameter and Figures 3E, F show the number of clots over 30 μm as a function of ultrasound frequency for both US and US + rtPA samples. In agreement with the microscopy images, a significant reduction in clot size is observed at 150 kHz, while exposure to higher frequencies leads to less pronounced fragmentation. US + rtPA shows a similar trend of reduction in average diameter and large clot count at 150 kHz. For US + rTPA the mean clot diameter decreases from ∼20 µm in the control to ∼7 μm at 150 kHz, confirming significant fragmentation at this frequency. At 300 kHz and higher frequencies there is no statistically significant change in clot size for both US and US + rTPA.

The findings indicate that 150 kHz ultrasound is the most effective frequency for reducing both large clot count and size in both US-only and US + rtPA groups, whereas higher frequencies (300 kHz–1 MHz) result in less pronounced disruption. One potential explanation for clot lysis at 150 kHz is the formation of endogenous microbubbles, which can result in boundary-layer-driven acoustic microstreams (Rasouli et al., 2023). During the 150 kHz experiment, bubbles became visibly present, whereas no such bubble formation was observed at higher frequencies. Plasma and most biological fluids naturally contain gas nuclei that act as precursors for bubble formation under ultrasound exposure. Upon formation, these bubbles oscillate in response to ultrasound actuation and create localized streaming which in turn can enhance clot lysis.

To investigate whether the mechanism of fragmentation at 150 kHz was directly linked to cavitation and acoustic streaming, we compared the effect of ultrasound exposure on degassed and non-degassed plasma samples. Cavitation requires dissolved gas nuclei to form microbubbles and thus degassing the plasma minimizes endogenous cavitation and bubble formation.

Figure 4 shows the fluorescence microscopy results of microclots in degassed and non-degassed plasma. Clot lysis and reduced micro clot size was observed in the gassed plasma sample after treatment with 150 kHz acoustic waves as shown by the images in Figure 4A. The same ultrasound waves failed to lyse microclots in degassed plasma samples and they remained largely intact. The absence of visible bubble formation in the degassed sample during the test as compared to the non-degassed sample suggests that clot lysis at 150 kHz can be attributed to the formation of endogenous bubbles and their attendant acoustic streaming.

Quantitative data on microclot size after ultrasound exposure is depicted in Figure 4B for both degassed and non-degassed samples. The left graph investigates microclot lysis in degassed vs. non degassed plasma where cavitation effects are inhibited. The y-axis is the mean clot diameter, and the x-axis is frequency. Although the degassed samples showed a minor decrease in clot size it is considered statistically nonsignificant with diameters ranging from ∼17 to ∼14 µm compared to ∼7 µm in non-degassed sample. Moreover, the number of large clots over 30 µm decreased for non-degassed sample to approximately a third of the control while in the degassed sample the variation was not statistically significant. This supports the hypothesis that clot lysis observed at 150 kHz is due to the cavitation, likely through cavitation-induced acoustic streams.

Previous results showed that cavitation is crucial for clot fragmentation at 150 kHz. In this section, we introduce exogenous microbubbles (MBs) into the plasma to promote stable cavitation. We use gas-filled microbubbles ∼1–2 µm in diameter which are coated with a phospholipid shell to render them more stable. In response to acoustic waves, these microbubbles undergo rapid compression and expansion cycles and generate localized fluid motion known as acoustic streams. Unlike endogenous cavitation, which relies on the presence of dissolved gas nuclei that varies between samples, contrast agents provide more consistent and controlled bubble dynamics and acoustic streaming. We hypothesize that the use of exogenous bubble can further increase lysis. In this section we also investigate the impact of MBs in combination with rtPA on clot lysis to assess whether this approach would further enhance enzymatic clot degradation.

Figure 5 shows the fluorescence microscopy images of ThT-stained plasma in the MB and MB + rtPA groups at the tested frequencies. In both groups, clot fragmentation is more pronounced at 150 kHz and 300 kHz, showing a significant reduction in clot size. Fragmentation is also observable with less impact at 500 kHz, while at 1 MHz, the effect diminishes significantly. Clot size analysis shown in Figure 5 further supports these observations and confirms a substantial decrease in clot size. The average clot diameter significantly decreases from ∼18 µm in the control group to ∼7 μm at 150 kHz, ∼8 μm at 300 kHz, and ∼10 μm at 500 kHz, while at 1 MHz, the clot size remained about 10 µm in the MB group. Additionally, the large clot count decreased from ∼450 in the control to ∼70 at 150 kHz, 90 at 300 kHz, 140 at 500 kHz, and 350 at 1 MHz. The results confirm the capability of the ultrasound system in combination with MBs for fragmentation and elimination of large microclots.

A similar trend, with a slightly stronger effect, can be seen in the MB + rtPA conditions, where the average clot diameter decreases from 19 µm in the control to 7 μm at 150 kHz, and the large clot count declines from ∼500 to 50 at 150 kHz. The observed reduction in clot size and count suggests a potential synergistic interaction between ultrasound, microbubbles, and enzymatic lysis, warranting further investigation to optimize this effect. This is particularly relevant given that acoustic streaming is reported to be one of the strongest mechanisms to induce normal advection and enhancing mixing (Rasouli and Tabrizian, 2019). Strong and rapid mixing can accelerate enzymatic clot degradation by facilitating the binding of rtPA to fibrin within the clot, thereby promoting the conversion of plasminogen to plasmin.

These results indicate that microbubble-assisted ultrasound stimulation is more effective for clot fragmentation at lower frequencies (150–500 kHz). The addition of exogenous contrast agents enhances cavitation and acoustic streaming, resulting in more effective clot disruption. Also, in MB added groups the clot fragmentation was observed across multiple frequencies, while in the US-only group, lysis was observed only at 150 kHz. The enhanced efficacy observed at 150–300 kHz aligns with prior studies showing that microbubble volumetric oscillation is markedly stronger in this range. Volumetric expansion ratios can exceed 30-fold at these frequencies, while at 1 MHz, expansion is typically below 2-fold under comparable pressure conditions (Ilovitsh et al., 2018).

The clot lysis in the presence of microbubbles can be attributed to boundary-layer-driven acoustic microstreaming generated by oscillating stable or semi-stable microbubbles which creates strong localized shear forces capable of breaking down the brittle amyloid fibrin(ogen) microclots. This also provides additional targeting specificity to the microclots, as they are more susceptible to fragmentation than other biological particles. Moreover, the streaming-induced mixing may improve enzymatic access by exposing binding sites and facilitating the interaction of rtPA with the microclots (Školoudík et al., 2025). The absence of lysis in degassed conditions supports the role of cavitation-driven streaming in this process. These results emphasize the potential of MB-assisted ultrasound therapy for targeting persistent amyloid microclots in conditions such as Long COVID.