BSA aggregation was investigated in both the vial and the microfluidic systems. The obtained results here for BSA aggregation were consistent with previous studies 11–13 (Fig. 1B and Supplementary Fig. 2). Within the microfluidic chip, as BSA moved through the channels over time, we noted the development of fibrillary aggregates along the microchannel sidewalls. Therefore, both systems successfully developed fibrillary structures despite their distinct conditions. Although both systems were studied at 70°C (higher temperature than melting point (Tm)), the bulk system (microtube) was operated under sufficient time without stress (no agitation), contrasted with the microfluidic system experiencing various tensions and short exposure (heating was provided by Supplementary Fig. 1). The significance of applying tension to reduce the time taken for protein aggregation in vitro is validated elsewhere. For example, utilizing agitation that breaks down supersaturation barriers 27 and hydrodynamic force that induces conformational alterations and destabilizes the native structure of protein impose stress, thereby promoting protein aggregation rate 28. Overall, both systems exhibited a similar capacity to generate fibrillary aggregates with a slight difference in ThT emission and morphology. It is accepted that BSA aggregation is not dependent on nucleation, and partial unfolding as well as rapid cluster formation drives the aggregation process 11,12,37. The microfluidic system achieved to produce some quasi-native clusters displaying native-like structures in a few seconds. Hence, initially, both systems behaved similarly. The final product from the chip's terminal exhibited significant ThT emission, showing fibrillary aggregate formation, contrasting with the absence of ThT-emitting aggregates in M1 (Fig. 1A). The existence of BSA clusters in M1 was confirmed through further microscopic and spectroscopic characterizations (Fig. 1E right). These clusters did not exhibit ThT emission and were utterly distinct from monomers.
In contrast, at the output of M2, fully developed fibrillary aggregates capable of emitting ThT comparable to V6 and V24 were observed (Fig. 1A). The chip's output displayed more elongated fibrillary aggregates in fewer numbers compared to vial samples (Fig. 1C-F). Protein fibrillation exposes buried hydrophobic patches, promoting intermolecular interactions and aggregation into fibrillary structures due to exposed hydrophobic surfaces. This process disrupts the protein's native structure, potentially leading to a loss of function 38. Accordingly, structural evaluations were carried out to clarify distinct features of the two mentioned BSA aggregation systems.
Structural analyses confirmed the presence of β-sheets and disordered secondary structures, affirming the amyloid nature of these fibrillary formations. Thus, the transition from globular BSA to fibrillary state involved increased β-sheet content and enhanced random coil content due to secondary structure rearrangements (Table 1). Disruption of the protein native folding is a vital step for aggregation in which ordered secondary structures such as helices and sheets likely turn into disordered secondary structures such as random coils to facilitate the exposure of stabilizing elements buried within the native globular structure 39. The M1 sample, including BSA clusters, closely resembled the native BSA in both secondary and tertiary structures. In contrast, the M2 sample showed significant structural alterations in both secondary (Fig. 3A) and tertiary structures (Fig. 3B).
Similarly, within the vial, there is a notable transformation in secondary and tertiary structures, notably burying tryptophan in a more hydrophobic microenvironment and the emergence of β-sheets. Since protein aggregation is in correlation with surface hydrophobicity (lower hydrophobicity reflects higher protein aggregation), samples were analyzed using ANS emission, revealing that minimal hydrophobicity belonged to the V24 sample, suggesting a highly compact structure of BSA monomers corresponding fibrillar aggregates formation during 24 hours 40–42. In this regard, M2 and V6 samples with higher hydrophobicity indexes than the V24 sample had weaker aggregation profiles. Notably, M1 demonstrated the most akin hydrophobicity index to the native state (Figs. 3C and D). BSA fibrillary aggregates of the two studied systems seemed morphologically different; thus, an MTT test was executed. BSA fibrillary aggregates in two systems did not affect MCF-7 cell viability (Fig. 3E), which is consistent with the lack of evidence on BSA aggregates' cytotoxicity elsewhere 11.
Results suggested that the microfluidic system effectively simulated BSA aggregation in a bulk system while only minimal sample volume is needed and high accuracy is secured in a brief timeframe. The microchannels provide a cross-section with a volume of 0.15 nl, which hosts approximately 3 billion BSA monomers at any moment (see supplementary). Upon heating, particle movement and collision probabilities are elevated, and accordingly, the Brownian force caused by the random movement of the particles increases, resulting in the formation of larger particles 26. Heat transfer modeling revealed that the chip's fluid quickly reaches a temperature of 70°C (Supplementary Fig. 4), providing the driving force to instantly form clusters from quasi-native monomers and initiate the BSA aggregation pathway. COMSOL simulation was performed to study the BSA aggregation process in a microfluidic chip, and three main forces consisting of Stoke’s drag force, particle-particle interaction, and Brownian movement were included (Supplementary Movie 1). Brownian movement and Stokes’ drag force were modeled according to Supplementary equations 8 and 9, respectively, and particle-particle interaction was studied using a Lennard-Jones model in COMSOL 43. Parameters needed for the Lennard-Jones model (collision distance and interaction strength) were extracted from DLVO theory in which repulsive electrostatic and attractive van der Waals forces are adeptly combined in its comprehensive framework to describe colloidal aggregation 33,44. The estimated pair potential between two native BSA monomers (as 7 nm spherical particles) was previously plotted by Ikeda et al. 33. Since SEM images obtained here (Fig. 1E) showed that up to M1 output, the fluid mainly contains 200–500 nm BSA clusters rather than monomers (experienced 70°C for a few seconds). Therefore, DLVO theory was performed to visualize intermolecular interaction between 300 nm BSA clusters (considered to be spherical) using MATLAB (Fig. 4A). A crucial and independent correlation was noted between particle size and van der Waals attraction (Supplementary Table 1). For particles up to 100 nm in diameter, the balance between attraction and repulsion operates in favor of repulsion 45. Conversely, particles larger than 100 nm experience van der Waals attraction, leading to the assembling of formed particles and the generation of larger agglomerates. Hence, with increasing particle size, long-range attractions become influential.
Van der Waals attractions are suggested to correspond to the formation of larger clusters in the second and third loops. Simulation results indicated that hydrodynamic forces do not influence particles smaller than 100 nm. In comparison, particles larger than 100 nm are pushed toward the channel wall, resulting in the deposition of BSA clusters on the sidewall (Supplementary Movie 1) 35. Additionally, increasing the shear rate (concluded from simulation) along the channel sidewall leads to a subsequent reduction in particle velocity (Fig. 4B), aiding the adsorption of BSA clusters on the sidewall 36. The hydrophobic nature of the microchannel surface (PDMS) revealed by the contact angle (Fig. 3F and Supplementary Fig. 3) is suggested to facilitate BSA cluster adsorption because, at the initial steps of BSA aggregation, surface hydrophobicity is increased (Supplementary Fig. 5). While hydrophobic surfaces usually exhibit anti-aggregation properties 46, in the current microfluidic chip, they tend to foster aggregation due to the following likely reasons: accumulation of BSA clusters on the sidewalls based on hydrophobic interactions and elevation of cluster local concentration along the channel sidewalls. This leads to prolonged exposure of clusters to pressure and hydrodynamic forces and amplification of shear rate along the channel sidewalls. The mentioned hydrodynamic driving forces, such as shear stress, act as a significant tension factor that facilitates and accelerates protein aggregation via reducing kinetic energy barriers, which prevent protein aggregation by maintaining the native state 47,48.
ThT emission became noticeable on the channel sidewalls one loop after the emergence of 300 nm clusters, and by passing two loops after M1, fibrillar structures roughly 1.5 µm in length formed along the sidewalls (Fig. 2). Subsequently, after around four loops from M1, longer fibrillary structures, about 5.2 µm, emerged. Notably, the absence of ThT emission in the middle of the microchannel was likely due to the presence of faster-moving clusters. These swiftly moving particles in the channel's center experience a low shear rate, impeding their adsorption by the sidewall, resulting in minimal interaction with other particles and rapid transit through the middle of the channel.
Nucleation is a significant step in polymerization processes such as protein aggregation, and its role is emphasized in many peptide/protein aggregation studies 11. However, in some cases, such as albumin, the nucleation process is not necessary for the aggregation and formation of fibrillary deposits. On the other hand, structural conversion between initial aggregated protein monomers (oligomeric states) at different stages of the aggregation process is another important mechanism that drives aggregation 6. The current study, according to a nucleation-free mechanism of BSA aggregation and observing amyloid-like fibrillary aggregates obtained from microfluidic chips formed from quasi-native clusters, strengthens the probability of a structural conversion between M1 and M2. Prolonged adsorption of BSA clusters on the microchannel sidewalls for extended periods increases the chance of quasi-native clusters' conversion into amyloid-like fibrillar aggregates. Interestingly, rapid clusters lacking ThT emission that are swiftly cleared from the center of the microchannel confirmed this hypothesis because these particles do not have sufficient time to convert and experience minor tension. The driving forces that lead to the aggregation of BSA in the chip are shown schematically in Fig. 5.