Concentration-dependent polymorphism of insulin amyloid fibrils

Insulin sample preparation

Human recombinant insulin powder (Sigma-Aldrich cat. No. 91077C) was dissolved in a 20% acetic acid solution containing 100 mM NaCl (reaction solution) to a final concentration of 2 mM (11.6 mg/ml). Insulin concentration was determined by measuring the sample’s absorbance at 280 nm ε=6,335 M⁻¹cm⁻¹, M = 5808 Da. Samples for unseeded aggregation kinetic measurements were prepared by diluting the 2 mM stock solution using the reaction solution and 10 mM ThT stock solution to a range of concentrations from 0.2 mM to 1.0 mM (which contained 100 µM of ThT). For seeded aggregation, insulin fibrils prepared from the 0.2 mM and 1.0 mM samples were sonicated for 10 min using Sonopuls 3100 (Bandelin) ultrasonic homogenizer equipped with a MS73 tip (40% power, 30 s sonication/ 30 s rest intervals). The homogenized fibrils were then diluted with the reaction solution and mixed with the 2 mM insulin and 10 mM ThT stock solutions to yield 0.2 mM and 1.0 mM concentration samples containing 100 µM ThT and a range of fibril concentrations (from 5% to 10⁻⁶ % of monomer mass).

Aggregation kinetics

Insulin aggregation kinetics were monitored at 60 °C without agitation by measuring ThT fluorescence emission intensity (excitation wavelength—440 nm, emission—480 nm) through the bottom of a 96 well non-binding surface plate using Synergy H4 Hybrid Multi-Mode (Biotek) plate reader (readouts were taken every 10 min to minimize plate agitation). For every condition, four independent measurements were performed. Aggregation half-times (t₅₀) were calculated as the time needed to reach 50% of signal intensity. The full concentration range of aggregated insulin samples were centrifuged at 10 000 g for 30 min and the residual unaggregated insulin in the supernatant was determined to be less than 1% of initial protein concentration.

Atomic Force Microscopy (AFM)

After kinetic measurements, samples were diluted with the reaction solution to a 50 µM concentration and 20 µL of each was deposited on freshly cleaved mica and incubated for 1 min. Subsequently, samples were rinsed with one mL of MilliQ water and dried under gentle airflow. Three-dimensional AFM maps were acquired using a Dimension Icon (Bruker) atomic force microscope operating in tapping mode and equipped with a silicon cantilever Tap300AI-G (40 N m⁻¹; Budget Sensors) with a typical tip radius of curvature of 8 nm. High-resolution (1,024 × 1,024 pixels) images were acquired. The scan rate was 1 Hz. AFM images were flattened and analyzed using SPIP (Image Metrology).

Fourier-Transform Infrared (FTIR) Spectroscopy

Insulin fibrils were separated from solution by centrifugation at 10 000 g for 30 min and subsequently resuspended in one mL of D₂O, the procedure was repeated three times. Then the fibrils were resuspended in 0.2 mL of D₂O and sonicated for 1 min using a MS72 tip (with 20% power and constant sonication). Samples were deposited between two CaF₂ transmission windows separated by 0.05 mm teflon spacers. The FTIR spectra were recorded using Vertex 80v (Bruker) IR spectrometer equipped with a mercury cadmium telluride detector, at room temperature under vacuum (∼2 mBar) conditions. 256 interferograms of two cm⁻¹ resolution were averaged for each spectrum. Spectrum of D₂O was subtracted from the spectrum of each sample. All spectra were normalized to the same area of amide I/I’ band (1,700–1,595 cm⁻¹). All data processing was performed using GRAMS software.

Fibril formation at different concentrations

Aggregating a range of insulin concentrations in 20% acetic acid with 100 mM NaCl at 60 °C without agitation reveals a typical kinetic curve pattern, where an increasing insulin concentration leads to shorter aggregation times (Fig. 1A). However, we observe an uneven ratio distribution between ThT fluorescence emission intensities and final fibril concentrations (Fig. 1B). As the concentration of insulin in the sample increases, this ratio shifts ten-fold, indicating either a higher quantum yield or considerably more bound ThT molecules.

The FTIR spectra of aggregated samples exhibit maxima in the amide I/I’ region at ∼1,628 cm⁻¹ with the shoulder at ∼1 641 cm⁻¹, and a small band outside of the amide I/I’ region at ∼1,729 cm⁻¹ (Fig. 1C), which is very similar to previously reported insulin fibrils formed in phosphate buffer at pH ≤2 (Sneideris et al., 2015). However, minor concentration-dependent differences can be observed (Figs. 1C and 1D). The spectra of fibrils, formed at lower insulin concentrations, have a pronounced shoulder at 1,641 cm⁻¹, while a minor band at 1,620 cm⁻¹ appears in the second derivative spectra (Fig. 1D) of samples aggregated at higher protein concentrations. The 1.0 mM and 0.8 mM fibril spectra are nearly identical, while 0.6 mM and 0.4 mM spectra appear to be intermediates between 0.8 mM and 0.2 mM, suggesting the existence of two distinct conformations.

Fibril morphology

The morphology of insulin fibrils formed at different concentrations was compared using AFM. We can see far more small and separated aggregates in samples formed at lower insulin concentration (Figs. 2A–2E). Analysis of variance (ANOVA) reveals that there is a statistically relevant fibril height difference (p = 0.01, n = 50) between the low and high concentration samples (Fig. 2F). Additional AFM images of these conditions are available as Fig. S1.

Seeded aggregation

In order to determine whether observed different fibril templates can propagate at unfavorable conditions, a set of seeded aggregation reactions were performed (Figs. 3A–3D, Fig. S2). In all four cases we observe a fibril-concentration-dependent seeding propensity (Fig. 3E), however, there is an interesting ThT fluorescence distribution, based on the amount and type of seed added (Fig. 3F). When the low concentration fibrils (0.2 mM, further on referred as LCF) are added to 0.2 mM insulin solutions, there are relatively no major differences in the fluorescence intensity at the end of each reaction. The same can be said in the case when the high concentration fibrils (1.0 mM, further on referred as HCF) are added to 1.0 mM insulin solutions. However, when the LCF are added to 1.0 mM insulin solutions, high seed concentrations yield a low fluorescence intensity, which then rises with decreased amount of seeds, eventually resulting in an intensity comparable to HCF. The opposite is observed when HCF are added to 0.2 mM insulin solutions, where high initial fibril concentrations yield an intensity comparable to the seed conformation (when accounted for fibril concentration at the end of the reaction) and an intensity similar to LCF when the seed concentration becomes minimal.

In order to further confirm the self-replication ability of both conformations, fibrils formed during seeded aggregation were examined by FTIR and their spectra were compared to the unseeded aggregation fibril spectra (Figs. 4A–4D). The results show that when a large concentration of preformed fibrils is added to either 0.2 mM or 1.0 mM insulin solutions, the seed self-replicates and maintains its initial secondary structure. On the other hand, when a low concentration of seed is added, the resulting FTIR spectra are similar to their respective environment conformations, rather than the seed.

Seeded fibril morphology

When large amounts of sonicated aggregates are used, there is minimal difference in the length and distribution of fibrils (Figs. 5A–5D), likely due to the large amount of aggregation centers. When the amount of seed used is low, the fibril length and distribution is similar to unseeded aggregation (Figs. 5E–5H). Fibril height distribution reveals a similarity between almost all conditions, except for when 1.0 mM insulin is seeded with low concentrations of either conformation (Fig. 5I), where the height distribution is comparable to unseeded nucleation. Additional AFM images of these conditions are available as supplementary information (Fig. S3).