In order to study the microscopic processes involved in the aggregation of IAPP, and to resolve quantitatively the underlying rate constants of the microscopic processes underpinning the aggregation reaction, it is essential to generate highly reproducible experimental measurements of aggregation over a range of peptide concentrations. Through optimising the experimental conditions, including the sample purity, the initial conditions, the inertness of all surfaces involved, the area of all surfaces and the linearity of the reporter used (Hellstrand et al., 2010; Meisl et al., 2016), the recorded reaction profile from such experiments is an accurate reflection of the molecular processes leading from monomers to fibrils, and the data can be analysed to reveal information about these processes. The ability to generate such data has previously been impeded, in particular, by the difficulty in purifying IAPP and isolating the monomeric species from other aggregated species. To overcome these obstacles, we have developed a protocol using successive rounds of size-exclusion chromatography (SEC). Successive rounds of SEC guarantee the quality of the measurements of the aggregation, and allow for subsequent theoretical analysis of the aggregation mechanism, as previously employed for other proteins and peptides, such as Aβ (Cohen et al., 2013).

The protocol was developed by exploring a variety of conditions in terms of buffer composition, pH, temperature and size-exclusion resin in order to produce IAPP monomers without excessive losses of the peptide. We found that a rigid allyl dextran/bisacrylamide matrix (as in the Tricorn 10/300 GL Sephacryl S-100 HR column) has minimal interaction with IAPP, which allows for good separation of the monomer from aggregates and enables isolation of homogeneous samples (Figure 1). An initial HiPrep 16/60 Sephacryl S-100 HR SEC shows the presence of higher-ordered species together with the monomeric IAPP (Figures 1A,B). Pooling the fractions containing monomeric IAPP and subjecting them to another round of size exclusion (using a Tricorn 10/300 GL Sephacryl S-100), an extremely pure monomeric IAPP sample was obtained. With this protocol the initial conditions are well-controlled, and we can study IAPP aggregation kinetics and derive the molecular mechanism of aggregation (Figures 1C,D).

Starting with pure monomeric IAPP samples, with concentrations ranging from 2 to 10 µM IAPP, the aggregation process was studied by monitoring the fluorescence of ThT as a function of time. In the present work, the aggregation process was studied at mildly acidic pH (5.3) at an ionic strength of 183 mM to reflect the physiological environmental conditions inside the β-cells where IAPP is proposed to aggregate initially (Hutton, 1982). All experiments were initiated with a temperature jump from 0 to 37 C.

The aggregation curves of recombinant IAPP presented here are sigmoidal-like (Figure 2) and qualitatively similar to the data presented in other publications studying the aggregation of IAPP, or shorter fragments of IAPP, using tyrosine anisotropy (Padrick and Miranker, 2002), light scattering (Ruschak and Miranker, 2007) or ANS fluorescence (Kayed et al., 1999). The curves are also qualitatively similar to those observed for non-amidated IAPP (Lundqvist et al., 2021). The ThT fluorescence intensity over time was found to be highly reproducible and dependent on the initial concentration of IAPP monomers in solution (Figure 2). Moreover, both the final fluorescence intensity, as well as the overall rate of aggregation, were found to increase with the concentration of IAPP in a systematic manner (Figure 2).

The half-time of the aggregation process, t_1/2 (time to formation of half of the final aggregate mass), versus initial monomer concentration [m] ₀ was fitted to a power law, t_1/2 ∝ [m] ₀ ^γ, in order to quantify the concentration dependence of the aggregation. The scaling exponent γ is related to the reaction order of the dominant nucleation mechanism generating new aggregates (Cohen et al., 2012). The scaling exponent for IAPP is found to be γ ∼ −1.26 ± 0.05 (Figure 3A). The value of this scaling exponent is reproducible between discrete batches of the peptide; however, deviation of the absolute t _1/2 values can be observed, which may originate from errors in the estimates of the concentration of IAPP (Supplementary Figure S1). A scaling exponent of approximately −1.2 to −1.3 excludes aggregate proliferation by fragmentation as a dominant mechanism, in which case γ ≈ −0.5 is expected, i.e., a weaker monomer dependence as new aggregates are generated by fibrils alone (Cohen et al., 2012). Moreover, the scaling exponent for IAPP is in the range of that obtained for Aβ42 (γ = -1.3), which has previously been shown to aggregate via a surface-catalysed secondary nucleation mechanism (Cohen et al., 2013).

The kinetic traces of IAPP aggregation were analysed using the AmyloFit platform (Meisl et al., 2016) to connect the macroscopic measurements of protein aggregation to the microscopic mechanisms using chemical kinetics. The aim of the analysis is to describe the entire set of kinetic data at all different IAPP monomer concentrations using a single rate law (Figures 3B–D). The fitting results for IAPP show that the data are collectively best described by a model in which the main source of new aggregates is a process involving fibril surfaces catalysing the nucleation of IAPP monomers (Figure 3D). Indeed, the results show excellent agreement between this model and the experimental kinetic data (Figure 3D). Conversely, when the data were fitted to other aggregation mechanisms, i.e. nucleation-elongation (Figure 3B) or nucleation-elongation and fragmentation (Figure 3C), they were not well-described by the kinetics models.

The aggregation mechanism of IAPP was further investigated by aggregation experiments of IAPP in the presence of preformed seeds (Figure 4). For an aggregation process in which secondary nucleation is the dominant process, addition of small amounts (on the order of 1% or less) of preformed fibril seeds will result in a significant reduction of t _1/2 . This phenomenon is not observed in aggregation processes which include only primary nucleation and elongation, since the elongation of the small quantity of seeds would be within the noise level of the experiment and there would not be a shift in t _1/2 (Cohen et al., 2012). As shown in Figure 4, a significant acceleration in the aggregation reaction is observed in the presence of preformed seeds. The acceleration of aggregation increases as a function of the seed concentration (Figure 4). In fact, t _1/2 shows a linear dependence on the logarithm of the seed concentration, which is expected of self-seeding behaviour in the presence of secondary pathways (Arosio et al., 2014) (Figure 4B). The results also reveal that even low concentrations of seeds, i.e., 0.5%, are sufficient to accelerate the reaction, which confirms that secondary processes are involved in the aggregation mechanism of IAPP. The elongation rate constant, k ₊ , was estimated by using the initial gradient of the highly seeded data (33%) and the average fibril length determined from cryo-TEM measurements, to be approximately 5^.10⁵ M⁻¹s⁻¹ (see materials and methods) (Meisl et al., 2014). Taken together, the experimental and theoretical results reported here reveal unambiguously that IAPP aggregates through a mechanism governed by surface catalysed secondary nucleation.

The pH (5.5) selected for the present study mimics the intragranular pH (5–6) of the insulin-secretory granule, where IAPP is located (Hutton et al., 1982). While some of the IAPP will be secreted from these granules, the data from experiments performed at pH 7.5 (Supplementary Figures SI4, I5) are better fitted if secondary nucleation is included, implying that the mechanism we describe at pH 5.5 extends to neutral pH.

The results for IAPP aggregation presented here can be compared to other peptides that have been shown to aggregate via a mechanism dominated by secondary nucleation (Aβ40, Aβ42 and α-synuclein) (Cohen et al., 2013; Buell et al., 2014; Meisl et al., 2014). The scaling exponent γ of IAPP under native conditions, pH 5.3 and an ionic strength of 183 mM, where its net charge is close to +2 is in the same range as previously detected for Aβ42 (determined under conditions where its net charge is close to -3) (Figure 5). Future quantitative comparison at identical solution conditions and as a function of pH and salt to modulate the influence of electrostatic interactions will provide additional insights into the commonalities and specifics of the aggregation mechanisms among different proteins and peptides.

Three different cell biology assays were performed in order to study the biological activity of species generated during the IAPP aggregation reaction. Insulinoma cells were exposed to IAPP samples taken at different time points of the aggregation reaction (Figure 6). These samples included: 1) IAPP at the start of the aggregation process, where it is mostly monomeric (t ₀ ); 2) IAPP at the t _1/2 of the aggregation process, where there is a significant population of oligomers is expected to coexist with monomers and fibrils; 3) IAPP at the end of the aggregation process, 2.4 t _1/2 , where there are predominantly fibrillar species (but a smaller fraction of both oligomers and monomers remains) (Michaels et al., 2020). The ability of these samples, from the three different time points in the aggregation time course, to cause cellular dysfunction either by early or late cell toxicity readouts were assessed.

The level by which the samples disrupted cellular membranes and induced Ca²⁺ influx, which is widely regarded as a phenomenon associated with the toxicity of aggregates (Flagmeier et al., 2017), were measured after 30 min of treatment as an early readout. The Fluo-4 calcium indicator was used to measure intracellular calcium levels. A significant increase in the amount of calcium influx for cells exposed to IAPP samples at t _1/2 and 2.4 t _1/2 were observed (Figure 6C). These results suggest that IAPP aggregation intermediates are able to trigger Ca²⁺ influx within insulinoma cells.

The apoptotic marker Hoechst 33,342 was also used to investigate the 30 min treated cell samples. Interestingly, a significant increase in the signal were observed for the cells treated with the t _1/2 sample compared to the t ₀ sample (Figure 6D). Also the 2.4 t _1/2 sample showed an increased signal i.e., the samples containing species other than monomers shows an increase in cytotoxicity after 30 min treatment.

The CellTiter-Glo luminescent assay was used to estimate the viability of the cells by measuring the ATP levels produced by metabolically active cells after a 24 h treatment by the three different samples. A decrease in viability occurred in the cells treated with the t _1/2 sample (Figure 6E). The 24 h assay also reveals that the t ₀ sample also decreased the cell viability, which could be attributed to monomeric IAPP sample aggregating, i.e., producing oligomers, during the course of the 24 h assay.

Taken together, these results show that only the t _1/2 samples, where oligomeric species are present at their maximum concentration for a secondary nucleation driven reaction, are able to both trigger significant calcium influx and cause a significant decrease of cell viability.

The recombinant IAPP peptide (KCNTATCATQRLANFLVHSSNNFGAILSSTNVGSNTY) was produced as described previously (Rodriguez Camargo et al., 2015) with the following modifications. All expression media were supplemented with ampicillin (Duchefa Biochemie A0104.0010) to a final concentration of 50 μg/ml. Protein expression was induced by adding IPTG to a final concentration of 0.4 mM. After induction, the cultures were incubated at 20 C whilst shaking at 200 rpm overnight in baffled flasks. Cultures were harvested by centrifugation at 5,000 g. The resulting pellet from 1 L culture was resuspended in 25 ml of 20 mM HEPES, 0.1 mM EDTA, 2 M urea, 50 mM NaCl, pH 8. The eluate, after the chitin step, containing the fusion protein leader-IAPP was concentrated from 100 to 15 ml using 3 kDa cutoff centrifugal filters (Amicon) where after the buffer were changed by using six disposable PD 10 Desalting Columns (GE Healthcare, 17-0851-01) to the standard buffer (20 mM HEPES, 0.1 mM EDTA, 2 M urea, 50 mM NaCl, pH 8.0) in parallel. The lyophilised fraction of IAPP from the HPLC step was dissolved in 35 mM sodium acetate buffer pH 5.3 keeping the concentration of peptide lower than 100 µM and oxidised by the addition hydrogen peroxide to a final concentration of 3 mM and incubated for 6 h at 4 C after which the peptide was again lyophilized.

The lyophilised powder was dissolved in 5 ml, 8 M urea, 35 mM acetate and pH 5.3 to give an approximate protein concentration of 70 μM, as determined by absorbance spectroscopy using ε₂₈₀ = 1615 L mol⁻¹ cm⁻¹. The solution was loaded onto a HiPrep 16/60 Sephacryl S-100 HR (17,116,501 GE Healthcare) size exclusion column and proteins were separated using a flow rate of 1.0 ml/min. Fractions containing only monomeric IAPP were combined, yielding 18.6 µM IAPP in 14 ml, aliquoted, lyophilized and stored at -80 C for use in the kinetics experiments.

Just prior to each kinetics experiment, the lyophilised peptide was solubilised in 1 ml, 35 mM sodium acetate, pH 5.3, and loaded onto a Tricorn 10/300™ GL Sephacryl S-100 HR size exclusion column (17,061,201 GE Healthcare) equilibrated in the same buffer. Fractions containing monomeric IAPP were collected, combined, and kept on ice until use in the kinetics experiments.

Discarded options for size-exclusion chromatography. Several buffer components (urea, GuHCl), pH values (4.0, 5-0, 5.3, 5.5 and 6.0). NaAc concentrations (from 35 to 1.7 mM), temperatures (above 4 C) and column types [Superdex 30 increase 10/300 GL (GE Healthcare), Superdex 75 increase 10/300 GL (GE Healthcare)] were discarded during initial optimisation of monomer isolation because they lead to loss of more than 80% of the peptide, or the purity of the monomeric sample was compromised.

Novex™ 10–20% Tricine Protein Gels, 1.0 mm (Invitrogen™ EC66255BOX) were used for the SDS-PAGE analyses. The gels were run for 45 min at 150 V in Tris/tricine running buffer (0.1 M Tris, 0.1 M tricine, 0.1% SDS, pH 8.25) and stained with InstantBlue® Protein Stain (SKU: ISB1L Expedeon).

Monomer of oxidized IAPP was isolated using SEC in 30 mM acetic acid pH 5.3, dried under vacuum and dissolved in 4 μl 0.1% Trifluoroacetic acid (TFA), 2% acetonitrile (ACN). Matrix solution, 0.5 µl consisting of 5 mg/ml α-cyano-4-hydroxy cinnamic acid, 80% ACN, 0.1% TFA, were mixed with 1 µl sample and spotted on a MALDI stainless steel plate. MS spectra were acquired using a Autoflex Speed MALDI TOF/TOF mass spectrometer (Bruker Daltonics, Bremen, Germany) in positive reflector or positive linear mode. The observed mono-isotopic mass of 3,900.9 Da corresponds to full-length oxidized and amidated IAPP.

All samples were prepared in low-binding tubes (Axygen) on ice. Monomeric IAPP in 35 mM acetate buffer pH 5.3 was mixed with a 2.5 M stock solution of KCl and a 2 mM stock solution of ThT (CalBiochem, purchased from Sigma-Aldrich, Product code 596,200, dissolved in H₂O and filtered through a 200 nm filter and concentration determined using absorbance) to a final concentration of 33 mM acetate, 150 mM KCl, pH 5.3, 20 µM ThT. Solutions were pipetted into wells of a 96-well Half Area Black/Clear Flat Bottom PEGylated Polystyrene plate (Corning® 3,881), 80 µl per well in triplicates. The ThT fluorescence was monitored through the bottom of the plate over time as a reporter of the amount of aggregates formed using a plate reader (FLUOstar Omega, FLUOstar Galaxy or FLUOstar, BMG Labtech) equipped with 440 nm excitation filter and 480 nm emission filter.

Fibril seeds were prepared by adding a 5 µM IAPP solution in 33 mM acetate, 150 mM KCl, 20 μM ThT, pH 5.3, in the 96-well plate and incubating the solution in the plate reader as described above. The ThT fluorescence was monitored to ensure that the aggregation reaction was complete before collecting the seeds. The collected fibril seeds were added to monomer IAPP solutions to final seed concentrations ranging from 0 to 33% of the monomer concentrations (in monomer equivalents). The solutions were pipetted in triplicate to the 96-well plates and the aggregation process was followed by monitoring the ThT fluorescence in the plate reader at 37 C under quiescent conditions, as described above.

All analyses involving the determination of the midpoint of the aggregation reaction (t _1/2 ) and the global analysis of the kinetic data were performed using the online Amylofit platform (Meisl et al., 2016). For the global analysis of the kinetic data the following rate law was used, which describes the IAPP species distribution over time and allows for the inclusion of secondary nucleation: [ M ] t [ M ] ∞ = 1 − ( 1 − [ M ] 0 [ M ] ∞ ) e − k ∞ t ⋅ ( B − + C + e ϰ t B + + C + e ϰ t ⋅ B + + C + B − + C + ) k ∞ 2 k ¯ ∞ ϰ where the following definitions are used: κ = 2 [ m ] 0 k + [ m ] 0 n 2 k 2 λ = 2 k + k n [ m ] 0 n c C ± = k + [ P ] 0 κ ± k + [ M ] 0 2 [ m ] 0 k + ± λ 2 2 κ 2 k ∞ = 2 k + [ P ] ∞ k ¯ ∞ = k ∞ 2 − 2 C + C − κ 2 B ± = k ∞ ± k ¯ ∞ 2 κ and where [ m ] 0 is the initial concentration of soluble monomers, [ M ] 0 and [ M ] ∞ are the mass concentration of fibrils at the start and the end of the reaction respectively, and [ P ] 0 and [ P ] ∞ are the number concentration of aggregates at the start and end of the reaction respectively; n c and n 2 are the reaction orders relative to the monomer of the primary and secondary nucleation pathways respectively; k + , k 2 , and k n are the rate constants for elongation, secondary nucleation, and primary nucleation, respectively. The equations derived for models with fragmentation, have been described previously (Meisl et al., 2016).

Fibrils were collected after reaching the plateau phase in the aggregation process and analysed by cryogenic transmission electron microscopy (TEM). The samples were loaded as a liquid film on a lacey carbon filmed cooper grid (01881F Lacey F/C, 200 mech Cu; PELCO No.160). A layer of sample less than 300 nm thick was produced on the grid by blotting the extra liquid away at the back of the grid using a filter paper, followed by flash freezing the grid in liquid ethane and whereafter it was stored in liquid nitrogen. The grid preparation was carried out in a controlled environment vitrification system to ensure the stable temperature and humidity in order to maintain the original state of the sample. Images were recorded using a 120 kV electron microscope (Philips CM120 BioTWIN Cryo) with a CCD camera. The size of the fibrils, for statistical purposes, were analysed using ImageJ (Schneider et al., 2012).

The estimation of the elongation rate constant was performed as described previously (Meisl et al., 2014, 2017). Firstly, the fibril sizes were estimated from length and thickness measurements based on cryo-EM images (Supplementary Figure S6). The average length was determined to be 900 ± 400 nm, and the cross area 80 ± 20 nm². Assuming a protein density of 1.3 g/ml and the molecular mass of 3,906 Da for the IAPP monomer, each fibril was estimated to consist of approximately 14,400 monomers. Subsequently, using the results of the strongly seeded aggregation (33% seeds), the initial gradient, d M / d t / t = 0 = 2 k + m ( 0 ) P ( 0 ) was derived. To estimate the number of seed fibrils, P (0), the mass concentration of seed fibrils (in monomer equivalents), M (0), which is known, was divided by the number of monomers per seed fibrils. The elongation rate constant was then determined to be approximately 5⋅10⁵ M⁻¹s⁻¹, within a factor of 3, based on the heterogeneous length and thickness distribution in the fibril samples.

INS-1 832/13 rat insulinoma cells (Merck KGaA, Darmstadt, Germany) were cultured in RPMI-1640 (Thermofisher, United Kingdom) supplemented with 2 mM l-glutamine, 1 mM sodium pyruvate, 10 mM HEPES, 0.05 mM β-mercaptoethanol and 10% heat-inactivated fetal bovine serum. The cell cultures were maintained at 37 C in a 5.0% CO₂ humidified atmosphere and grown until 80% confluence for a maximum of 20 passages.

Cell viability was measured using the CellTiter-Glo Luminescent Cell Viability Assay (Promega) according to the manufacturer’s instructions. Briefly, the cells were plated into a white opaque 96-well plate and treated for 24 h with samples containing 5 µM IAPP from an aggregation time course (6 replicates per condition). The final concentration of IAPP was 1.25 µM (monomer equivalent). Luminescence values were measured using a plate reader (ClarioStar Omega BMG Labtech, Aylesbury, United Kingdom), and cell viability was expressed as a percentage vs untreated cells (taken as 100%).

The cytosolic calcium ion (Ca²⁺) levels were measured by exposing the INS-1 832/13 cells loaded with 2.0 μM Fluo4-AM to samples containing 5 µM IAPP taken at different time points of an aggregation reaction (3 replicates per time-point). The final concentration of IAPP was 2.5 µM (monomer equivalent). The emitted fluorescence was recorded after excitation at 488 nm using the fluorescence microscope Cytation5 Cell Imaging Reader and quantified by means of the Gen5 Data Analysis software (BioTek Instruments, Winooski, VT).

INS-1 832/13 cells were treated with samples containing 5 µM IAPP from an aggregation reaction (3 replicates per time-point). The final concentration of IAPP was 2.5 µM IAPP. Cells were stained with the apoptotic marker Hoechst 33,342. The emitted fluorescence was recorded after excitation at 350 nm using the fluorescence microscope Cytation5 Cell Imaging Reader and quantified by means of the Gen5 Data Analysis software (BioTek Instruments, Winooski, VT).

For cellular assays, comparisons between groups were performed using one-way ANOVA followed by Bonferroni’s multiple comparison test. A p-value lower than 0.05 was considered statistically significant.