Nanoscale Control of Amyloid Self-Assembly Using Protein Phase Transfer by Host-Guest Chemistry

Amyloid fibrils have recently been highlighted for their diverse applications as functional nanomaterials in modern chemistry. However, tight control to obtain a targeted fibril length with low heterogeneity has not been achieved because of the complicated nature of amyloid fibrillation. Herein, we demonstrate that fibril assemblies can be homogeneously manipulated with desired lengths from ~40 nm to ~10 μm by a phase transfer of amyloid proteins based on host-guest chemistry. We suggest that host-guest interactions with cucurbit[6]uril induce a phase transfer of amyloid proteins (human insulin, human islet amyloid polypeptide, hen egg lysozyme, and amyloid-β 1–40 & 1–42) from the soluble state to insoluble state when the amount of cucurbit[6]uril exceeds its solubility limit in solution. The phase transfer of the proteins kinetically delays the nucleation of amyloid proteins, while the nuclei formed in the early stage are homogeneously assembled to fibrils. Consequently, supramolecular assemblies of amyloid proteins with heterogeneous kinetics can be controlled by protein phase transfer based on host-guest interactions.

Amyloid-β 1-42 peptide was purchased from AnyGen (Jangheung, Republic of Korea). A protein stock solution was prepared at a concentration of 100 μM in 0.1% FA, and a 50 mM CB [6] stock solution was prepared in 50% FA. We prepared all solutions under acidic conditions because the net charges of amyloid proteins are positive in low pH, and CB [6] preferentially interacts with positively charged molecules. 1 In addition, INS is highly soluble under acidic conditions and is converted from multimeric states (dimer, tetramer, and hexamer) to the monomeric state. 2 Stock solutions were transferred to 4-mL borosilicate glass vials and diluted to a final volume of 1 mL. Protein concentrations were adjusted to 50 μM for INS and LYZ and 10 μM for amyloid-β and human islet amyloid polypeptide to prevent aggregation prior to incubation in the 5% FA solution. All protein solutions were incubated at 50 °C for four days. Agitation was performed at 200 rpm. When we examined CB [6]-mediated fibrillation at a high temperature (60 °C), CB [6] was not effective for controlling the fibrillation process ( Supplementary Fig. S15). In contrast, at a low temperature (40 °C), the fibrillation process itself was not promoted. The fibrillation process using CB [6] was best controlled at 50 °C. calculated using the mathematical definition of PDI described in the Supplementary Text (vide infra).
Thioflavin T (ThT) assays. ThT assays were performed to obtain information about the fibrillation kinetics. A ThT stock solution (1 mg/mL) was prepared. An aliquot of incubated protein solution (25 μL), ThT stock solution (25 μL), and 5% FA (450 μL) were mixed prior to the fluorescence measurements. The excitation wavelength of ThT was set to 452 nm and the emission was scanned from 460 to 490 nm. The emission at 482 nm was used for monitoring the fibrillation kinetics. Since the fluorescence intensity of ThT increases upon complexation with CB [6], the baseline was corrected using a ThT-CB [6] solution without proteins. All data points were normalized using the height of the stationary phase.
Isothermal titration calorimetry (ITC). ITC experiments were performed using a VP-ITC calorimeter (MicroCal, Worcestershire, UK) to obtain an equilibrium association constant (Ka) for the interaction between INS and CB [6]. INS and CB [6] solutions were freshly prepared and degassed prior to each measurement. The reference cell was filled with 50% FA, and an INS solution in 50% FA (100 µM) was loaded into the ITC cell. A CB [6] solution (1 mM) was injected 30 times through the ITC syringe. The interval of each injection was 3 min. The ITC cell was stirred at 502 rpm and maintained at 25 °C. The heats of dilution by CB [6] were subtracted for analysis. All results were treated using the 'single-set-of-sites' model internally stored in Origin 7.0 (associated with the VP-ITC calorimeter).
Ultraviolet-visible spectroscopy. Protein solutions with and without CB [6] were freshly Infrared spectroscopy. Infrared (IR) spectroscopy was performed using a Nicolet iS10 spectrometer (Thermo Scientific, Waltham, MA, USA). A total of 128 spectra for each sample were obtained with a resolution of 4 cm -1 , which were averaged to generate a single spectrum. Fibril samples were prepared following the procedure described in the 'fibril sample preparation' section above. Unwashed samples were directly cast onto a silicon wafer and fully dried for a day under air. Washed samples were loaded into an Amicon 0.5-mL 100k centrifugal filter (Millipore, Billerica, MA, USA) and then washed three times with 5% FA to remove the excess CB [6] and non-fibrillar species in the samples 6 before being loaded onto silicon wafers. All IR spectra were normalized using the highest peak intensity in the range of 1500-1900 cm -1 .
Solution Small-Angle X-ray Scattering (SAXS). All SAXS measurements were performed at the 4C SAXS II beamline of the Pohang Accelerator Laboratory (PAL) in Pohang University of Science and Technology (POSTECH). The concentration of INS was measured as 0.76 mg/mL in 1% FA. The sample-to-detector distance was set to 2 m, and the temperature was fixed at 20 °C. Each measurement was performed four times using fresh INS samples. The scattering patterns were recorded six times (5 s each with 1-s intervals). SAXS measurements of standard proteins (trypsin, human serum albumin, and concalvalin A) were performed following the same procedures. After the measurements, the concentrations of the standard proteins were also measured. The radius of gyration (Rg) and molecular weight (Mw) calibration using the zero-angle scattering intensity (I(0)) were estimated using a procedure in the literature. 3

Supplementary text
Polydispersity index. The polydispersity index (PDI) 4 is the ratio of the mass average degree of polymerization ( ) to the number average degree of polymerization ( ). If the polymer is ideally monodisperse, PDI = 1. The PDI value is usually larger than 1 because of the heterogeneous distribution of the polymers. In the present study, the length of the amyloid fibril was adopted as the degree of polymerization because its length was linearly proportional to the number of protein monomers within the fibril. Temperature is crucial for the kinetics of amyloid fibrillation, in that the activation energy for protein folding/unfolding is related to the melting temperature of the proteins (the melting temperature of INS is approximately 60 °C). 8 When we examined CB [6]-mediated fibrillation at a high temperature (60 °C), CB [6] was not effective for controlling the fibrillation process. In contrast, at a low temperature (40 °C), the fibrillation process itself was inhibited. This seems to be because CB [6] cannot kinetically modulate the selfassembly if the nucleation rate is too fast; if the nucleation rate is too slow, the fibrillation process itself is inhibited.