Structural and morphological dynamics of “on‐path” and “off‐path” oligomers of human islet amyloid polypeptide

TL;DR

This study explores how human islet amyloid polypeptide aggregates form and how their structures relate to toxicity in Type 2 Diabetes.

Contribution

The study identifies two distinct early oligomer types with different structural and kinetic roles in IAPP aggregation.

Findings

Donut-like oligomers (DO) are 'on-path' and form parallel β-sheets, leading to fibril formation.

Round oligomers (RO) are 'off-path', disordered, and persist during aggregation, contributing to toxicity.

Both structural evolution and persistent 'off-path' oligomers increase cytotoxicity in pancreatic β cells.

Abstract

The deposition of cytotoxic human islet amyloid polypeptide (IAPP) aggregates is a hallmark feature of Type 2 Diabetes. However, the structural evolution and cytotoxicity of IAPP aggregate species remain poorly understood. This study combines kinetics, biophysical and cell assays to resolve the morphological dynamics of IAPP aggregation. Using atomic force microscopy (AFM) and atomic force microscopy Infrared (AFM‐IR) spectroscopy, we observed two distinctly different types of oligomers, donut‐like (DO) and round oligomers (RO), formed at the early stages of protein aggregation. DO were dominated by parallel β‐sheet secondary structure. Their evanescence is linked to the formation of IAPP fibrils, which also had parallel β‐sheet secondary structure. In contrast, RO had primarily disordered secondary structure and persisted throughout the course of fibril formation. This structural and kinetic analyses showed that RO were “off‐path”, while DO were “on‐path” protein aggregates. Cell toxicity assays indicated that structural evolution of IAPP amyloids as well as persistent “off‐path” oligomeric species both contribute to high cytotoxicity in pancreatic β cells. These results revealed a complex mechanism of IAPP aggregation which is highly important in the context of the prevention of pathological protein aggregation.

Figures

ThT kinetics of IAPP aggregation. (a) ThT kinetic curves of IAPP aggregation at a concentration of 40 μM. (b) Calculated t lag and t half values of IAPP aggregation. (c)–(e) AmyloFit results for kinetic data points fitted to (c) secondary nucleation dominated, (d) nucleation elongation, and (e) fragmentation dominated models with corresponding mean residual error (MRE) values. (f) Schematic of secondary nucleation dominated aggregation model. Kinetic curves represent the average of n = 3 replicates ± SD, and bars represent mean ± SD of n = 3 calculated values. AmyloFit curves show the fitted line plotted over individual data points from each of n = 3 individual replicates.

A11 dot‐blot of IAPP aggregates formed at 1 h, 4 h, 8 h, 24 h, and 48 h. (a) Image of representative A11 anti‐oligomer dot‐blot from serially diluted 40 μM IAPP aggregate samples formed at 1, 4, 8, 24, and 48 h. (b) Quantification of the integrated density of each dot.

AFM images of IAPP aggregates and respective height data. (a) AFM images of 40 μM IAPP aggregates observed at different time points during protein aggregation. Scale bars are 200 nm. DO, donut oligomer; FB, fibril; RO, round oligome. Red circles highlight RO species and green circles highlight DO species. (b) Violin plots of heights of 40 μM IAPP aggregates; n = 30 per species.

IR spectra of IAPP aggregate species and calculated secondary structure makeup. (a) Collected IR spectra of 40 μM IAPP aggregates observed at different time points during protein aggregation. DO, donut oligomer; FB, fibril; RO, round oligomer. Individual spectra were median normalized to ensure comparative consistency. (b) Relative secondary structure percentages calculated from Gaussian peak fitting of collected IR spectra. Spectra represent the average of n = 30 spectra per aggregate species. Bar graphs represent mean ± SD of n = 3 calculated values obtained from the coaveraging of 10 spectra from the n = 30 spectra collected per species. Significance was calculated using two‐way ANOVA with Tukey's post hoc test. *p < 0.05; p < 0.01; p < 0.001; p < 0.0001. NS, nonsignificant differences.

ROS levels of pancreatic β cells exposed to IAPP aggregates formed at 1, 4, 8, 24, and 48 h. ROS values obtained from the CellROX Deep red assay using n = 3 independent BRIN‐BD11 cell cultures exposed to IAPP aggregate populations formed at different stages of aggregation. IAPP samples were prepared at a concentration of 40 μM and diluted to a final concentration of 12 μM in each well. Bars represent mean ± SD of n = 3 replicates. Statistics were performed using ANOVA with Tukey's post hoc test. *p <0.05, *p <0.001.

Tables

TABLE 1. AmyloFit kinetic parameters of IAPP aggregation.

Constant	Value
k n	1.58e‐8 conc−nc+1 time−1
k 2	6.94e+6 conc−n 2 time−1
k +	3.10e+6 conc−1 time−1
n c	0.783
n 2	1.85

Full text

INTRODUCTION

1

Type 2 diabetes (T2D) is a severe pathology that affects over 462 million people globally and is projected to impact 783 million by 2045 (Cao et al., 2024; Verma et al., 2024; Yu et al., 2025). In the United States, currently 1 out of 10 people are diagnosed with T2D, while an incidence rate of this pathology is ~1.2 million new cases per year. This results in the estimated economic burden of T2D over $327 billion per year in the US alone (Asiamah‐Asare et al., 2025; Ma et al., 2025; Perrone et al., 2025). Over 90% of T2D patients have a progressive aggregation of islet amyloid polypeptide (IAPP) into toxic oligomers and fibrils. (Mukherjee et al., 2017; Westermark & Grimelius, 1973) These protein aggregates accumulate in the pancreatic islets causing β cell death, resulting in insulin insufficiency and chronic hyperglycemia (Asthana et al., 2018; Giri et al., 2018; Kanatsuka et al., 2018; Laakso, 1999). IAPP oligomers and fibrils can also trigger the aggregation of amyloid β (Aβ) and α‐synuclein (α‐syn), amyloidogenic proteins associated with Alzheimer's and Parkinson's diseases, respectively (Moreno‐Gonzalez et al., 2017; Mucibabic et al., 2020).

In vitro studies show that under physiological conditions, IAPP rapidly aggregates forming morphologically heterogeneous oligomers and β‐sheet‐rich fibrils (Abedini & Schmidt, 2013; Cao, Marek, et al., 2013; Roder et al., 2020). Although the secondary structure of IAPP fibrils is well understood (Roder et al., 2020) structural organization of IAPP oligomers remains unclear (Abedini et al., 2016; Jeong & An, 2015). This is primarily because the morphological heterogeneity and transient nature of IAPP oligomers limits solid‐state NMR‐ and cryo‐EM‐based analyses of their secondary structure. Nevertheless, Hsu and co‐workers demonstrated that aggregating, IAPP formed toxic oligomers that had α‐sheet structure (Hsu et al., 2024). This structure was distinct from standard α‐helices or β‐sheets and was responsible for high cytotoxicity exerted by these oligomers. It should be noted that there is also no clear understanding of the cytotoxicity of IAPP oligomers and fibrils (Huang et al., 2007; Kim et al., 2021; Lin et al., 2007; Rivera et al., 2014). Some studies reported that fibrils, unlike IAPP oligomers, exerted no cytotoxic effects, while other studies found both IAPP oligomers and fibrils toxic to β cells (Abedini et al., 2016; Cao, Abedini, et al., 2013; Lin et al., 2007; Meier et al., 2006; Sitton et al., 2025). For instance, Abedini and co‐workers showed that low‐order oligomers formed by IAPP lack extensive β‐sheet structure and induce pancreatic β‐cell death by upregulating pro‐inflammatory markers and reactive oxygen species (ROS) (Abedini et al., 2016).

In this study, we used atomic force microscopy (AFM) and atomic force microscopy Infrared spectroscopy (AFM‐IR) coupled to kinetic and cell toxicity assays to investigate molecular mechanisms of IAPP aggregation. Previously, AFM and AFM‐IR were used by our group to resolve structural dynamics of Aβ and α‐syn oligomers (Zhaliazka & Kurouski, 2022; Zhou & Kurouski, 2020). Zhaliazka and co‐workers found that Aβ first formed oligomers with parallel β‐sheet secondary structure (Zhaliazka & Kurouski, 2022). These oligomers propagated into fibrils with substantially slower rates compared to the oligomers with anti‐parallel β‐sheet that were formed at later stages of Aβ aggregation. However, once formed, oligomers with anti‐parallel β‐sheet rapidly propagated into fibrils. As a result, at 72 h after the initiation of Aβ aggregation, a steady‐state equilibrium between the aggregates with parallel and antiparallel β‐sheet secondary structures was reached. After that, Aβ aggregates with antiparallel β‐sheet remained as a subpopulation, while the oligomers and fibrils with parallel β‐sheets dominated. Using AFM and AFM‐IR, Zhou and Kurouski investigated the secondary structure of α‐syn oligomers formed at different time points of protein aggregation (Zhou & Kurouski, 2020) It was found that at early time points, a variety of structurally distinct oligomers were formed. Some of them were dominated by parallel β‐sheets, whereas others had a mixture of α‐helix and disordered protein secondary structures. The researchers also found that early‐stage oligomers possessed a high amount of antiparallel β‐sheet that progressively decreased as the oligomers developed into fibrillar species. Based on these findings, the researchers concluded that α‐syn fibril formation was driven by the conversion of anti‐parallel to parallel β‐sheet secondary structure. Recently, Rodriguez and co‐workers utilized AFM and AFM‐IR to characterize the evolution of the secondary structure of transthyretin (TTR) that was taking place during pathological protein aggregation into oligomers and fibrils (Rodriguez et al., 2023) The researchers found that early‐stage oligomers were dominated by α‐helix and disordered secondary structure. These oligomers did not yield fibrils and, therefore, should be called “off‐path” oligomers. The second type of oligomers dominated by parallel β‐sheet appeared at 9 h after the initiation of TTR aggregation. These oligomers rapidly propagated into fibrils (“on‐path”).

Expanding upon these findings, we aimed to determine whether IAPP aggregation also results in “on‐path” and “off‐path” oligomers. To achieve this, we analyzed the aggregation kinetics of IAPP at 37°C using thioflavin T (ThT) assay. Next, we took solution aliquots at 1, 4, 8, 24, and 48 h. Solution aliquots were deposited onto gold‐coated silicon substrates, dried, and analyzed using AFM and AFM‐IR. Finally, we performed cytotoxicity assays using pancreatic β cells to understand the cytotoxic effects of IAPP aggregates formed at different time points of protein aggregation.

RESULTS

2

IAPP aggregation kinetics reveal a short lag‐time that quickly transitions into fibril elongation dominated by secondary nucleation

2.1

ThT assay revealed that under physiological conditions, IAPP aggregated with a short lag‐time (t lag) of 3.11 ± 0.098 h (Figure 1a,b). The lag‐time was followed by a rapid increase in the ThT intensity which indicated the formation of fibrils. We used a half‐maximal fluorescence intensity to determine half‐time (t half), which represents the rate of fibril formation. For IAPP, t half was 3.94 ± 0.098 h (Figure 1b).

ThT kinetics of IAPP aggregation. (a) ThT kinetic curves of IAPP aggregation at a concentration of 40 μM. (b) Calculated t lag and t half values of IAPP aggregation. (c)–(e) AmyloFit results for kinetic data points fitted to (c) secondary nucleation dominated, (d) nucleation elongation, and (e) fragmentation dominated models with corresponding mean residual error (MRE) values. (f) Schematic of secondary nucleation dominated aggregation model. Kinetic curves represent the average of n = 3 replicates ± SD, and bars represent mean ± SD of n = 3 calculated values. AmyloFit curves show the fitted line plotted over individual data points from each of n = 3 individual replicates.

This short time between t lag and t half indicates that the rate at which IAPP monomers form oligomers is much slower compared to the rate of secondary nucleation, a process when monomers aggregate on the surfaces of formed fibrils. To better understand these processes, we utilized AmyloFit software and fitted aggregation kinetics data to the previously established amyloid aggregation models (Meisl et al., 2016). We found that our data demonstrated the best fit best to a secondary nucleation dominated model (MRE = 0.000663) (Figure 1c and Table 1), as compared to nucleation elongation (MRE = 0.00848) (Figure 1d) and fragmentation dominated (MRE = 0.000723) (Figure 1e) models. Together, these results indicate that secondary nucleation catalyzed by fibril formation is the primary mechanism of IAPP aggregation. These results also suggested that both fibrils and smaller oligomeric aggregates could exist within the same population (Figure 1f).

To determine if oligomeric populations persist within IAPP aggregate populations over time, we utilized an A11 antibody dot‐blot assay to quantify relative number of oligomers present over the time (Figure 2). We observed that, indeed, there are persistent populations of oligomers present at 1, 4, 8, 24, and 48 h time points of IAPP aggregation. These results are in good agreement with what was observed for our kinetic modeling and indicate that there exists persistent oligomer populations that do not lead to fibril formation during the first 48 h of IAPP aggregation. It is also important to note that previous studies have shown that amyloid oligomers are in equilibrium with available monomer pools, which undergo dynamic aggregation and disaggregation based upon the available monomer pool (Figure 1f) (Fawzi et al., 2010). It is possible that these persistent off‐path oligomers exist in conjunction with dynamic populations of oligomers which aggregate and disaggregate in equilibrium with monomeric protein. Thus, persistent off‐path oligomers which form in early stages of aggregation as well as those that continuously form over time collectively contribute to similar levels of oligomers present during the first 48 h of IAPP aggregation as they do not efficiently evolve into fibrils.

A11 dot‐blot of IAPP aggregates formed at 1 h, 4 h, 8 h, 24 h, and 48 h. (a) Image of representative A11 anti‐oligomer dot‐blot from serially diluted 40 μM IAPP aggregate samples formed at 1, 4, 8, 24, and 48 h. (b) Quantification of the integrated density of each dot.

The aggregation of IAPP results in the formation of heterogeneous mixtures of multiple transient oligomer species which structurally evolve over time

2.2

We utilized AFM to examine morphological properties of IAPP aggregates formed at different stages of protein aggregation (Figure 3). At 1 h, two morphologically different oligomers with round and donut‐like shapes were observed. Donut‐like oligomers (DO) had a mean height of 17.48 ± 2.78 nm and had a pronounced central gap. Round oligomers (RO) exhibited a uniform spherical shape. These aggregates were slightly smaller compared to DO with a mean height of 14.26 ± 2.42 nm. Although both DO and RO were also observed at 4 h, their heights slightly changed. DO observed at 4 h were slightly smaller (13.78 ± 2.47 nm) than 1 h DO (~17 nm). At the same time, RO observed at 4 h exhibited a greater variability in height (8.9–23.3 nm) compared to 1 h RO (9.8–18.7 nm). These morphological changes point to the transient nature of both DO and RO aggregates and evolution of their structures. AFM analysis of IAPP aggregates formed at 8, 24, and 48 h revealed no DO oligomers. However, similar in shape and height, RO were co‐observed together with fibrils at 8 h and 24 h samples. It should be noted that a very small amount of fibrillar species was observed already at 1 h, which indicates fast propagation of IAPP oligomers into fibrillar species. We also found a gradual decrease in the size of RO observed at 4, 8 and 24 h. Specifically, the average height of RO observed at 4 h was 14.8 ± 3.72 nm, while the average heights of RO observed at 8 h and 24 h were 12.55 ± 2.83 nm and 9.95 ± 2.48 nm, respectively. Finally, fewer oligomers, if any, were observed at 48 h by AFM.

AFM images of IAPP aggregates and respective height data. (a) AFM images of 40 μM IAPP aggregates observed at different time points during protein aggregation. Scale bars are 200 nm. DO, donut oligomer; FB, fibril; RO, round oligome. Red circles highlight RO species and green circles highlight DO species. (b) Violin plots of heights of 40 μM IAPP aggregates; n = 30 per species.

Next, we used AFM‐IR to resolve the secondary structure of individual IAPP aggregates observed at different stages of protein aggregation (Figure 4). For this, a metallized scanning probe was placed on the surface of individual oligomers and fibrils. After that, the sample surface was illuminated by pulsed tunable IR light that induced thermal expansions in the aggregates. These thermal expansions were recorded by the metallized scanning probe and converted into IR spectra. The acquired IR spectra exhibited amide I and II bands centered around 1660 and 1525 cm^−1^, respectively. We performed Gaussian peak fitting of amide I to quantify the relative amounts of parallel β‐sheet (1615–1640 cm^−1^), disordered (1641–1670 cm^−1^), and anti‐parallel β‐sheet (1671–1699 cm^−1^) in IAPP oligomers and fibrils.

*IR spectra of IAPP aggregate species and calculated secondary structure makeup. (a) Collected IR spectra of 40 μM IAPP aggregates observed at different time points during protein aggregation. DO, donut oligomer; FB, fibril; RO, round oligomer. Individual spectra were median normalized to ensure comparative consistency. (b) Relative secondary structure percentages calculated from Gaussian peak fitting of collected IR spectra. Spectra represent the average of n = 30 spectra per aggregate species. Bar graphs represent mean ± SD of n = 3 calculated values obtained from the coaveraging of 10 spectra from the n = 30 spectra collected per species. Significance was calculated using two‐way ANOVA with Tukey's post hoc test. *p < 0.05; **p < 0.01; ****p < 0.001; ***p < 0.0001. NS, nonsignificant differences.

We found that DO and RO species observed at 1 h had significantly different anti‐parallel β‐sheet and disordered secondary structure makeup (p <0.05). DO had ~40% of parallel and anti‐parallel β‐sheet with ~20% of disordered secondary structures. At the same time, RO were dominated by anti‐parallel β‐sheet (75%) with a small amount of anti‐parallel β‐sheet (~20%), Figure 5 and Table S1. AFM‐IR also revealed structural evolution of DO. Specifically, DO observed at 4 h had the same amount of anti‐parallel β‐sheet (~40%) as DO observed at 1 h. However, we observed an increase in the amount of parallel β‐sheet (60%) and a significant decrease in disordered protein in 4 h DO compared to 1 h DO. We also found that IAPP fibrils had a very similar secondary structure to DO. Specifically, these aggregates were dominated by parallel β‐sheet (60%–80%) with a small amount of anti‐parallel β‐sheet (20%–40%). This structural similarity suggests that IAPP fibrils evolved from DO. This conclusion is further supported by structural analysis of RO observed at different stages of protein aggregation. We found that RO observed at 4 h had predominantly disordered secondary structure (70%) with a small amount of anti‐parallel β‐sheet (20%). The same structural organization was observed for RO present at 24 h. RO found in 8 h sample had nearly equal amounts of parallel β‐sheet, disordered, and anti‐parallel β‐sheet in their secondary structure. Importantly, none of these aggregates had a secondary structure similar to IAPP fibrils. Therefore, we conclude that RO are “off‐path” aggregates that although appear simultaneously with “on‐path” DO, do not evolve into fibrils.

*ROS levels of pancreatic β cells exposed to IAPP aggregates formed at 1, 4, 8, 24, and 48 h. ROS values obtained from the CellROX Deep red assay using n = 3 independent BRIN‐BD11 cell cultures exposed to IAPP aggregate populations formed at different stages of aggregation. IAPP samples were prepared at a concentration of 40 μM and diluted to a final concentration of 12 μM in each well. Bars represent mean ± SD of n = 3 replicates. Statistics were performed using ANOVA with Tukey's post hoc test. *p <0.05, **p <0.001.

The structural evolution of IAPP amyloids over time result in unique cytotoxicity profiles in β cells

2.3

To determine if the structural evolution of IAPP amyloids influences their cytotoxicity, we treated BRIN‐BD11 rat pancreatic β cells with IAPP amyloid populations formed after 1, 4, 8, 24, and 48 h of aggregation (Figure 5). We utilized the flow cytometry CellROX Deep Red assay to quantify the amount of ROS produced by β cells exposed to these aggregate populations. We found that after 1 h of IAPP aggregation, cytotoxic IAPP aggregates were present and significantly elevated ROS levels (13.77% ± 1.82%) when compared to the negative control (7.27% ± 0.93%). Aggregates formed after 4 h (11.63% ± 0.76%) and 8 h (11.37% ± 0.38%) of IAPP aggregation elicited similar toxicity to β cells as those formed after 1 h. This suggests that the appearance of highly cytotoxic species populates early in the aggregation of IAPP and persist over time. On the other hand, after 24 h of aggregation, aggregates formed elicited much higher cytotoxicity (20.47% ± 2.11%) compared to earlier timepoints. This heightened cytotoxic effect persists after 48 h of aggregation (17.9% ± 0.66%). Together, this demonstrates that the time of IAPP aggregation strongly influences the toxicity of the formed aggregates in cultured β cells. Specifically, early‐stage IAPP aggregates elicit cytotoxic effects to β cells, and these effects are exacerbated with longer time of aggregation.

DISCUSSION

3

In this study, we demonstrated that structurally and morphologically distinct populations of IAPP aggregates co‐existed simultaneously during different stages of IAPP aggregation. ThT kinetics revealed that IAPP aggregates and forms fibrils through the secondary nucleation dominated mechanism. This is consistent with findings by Camargo and colleagues, who found that secondary nucleation is the dominant mechanism of toxic IAPP aggregate formation (Rodriguez Camargo et al., 2021). These results indicate that both oligomers and fibrils can exist within the aggregate population simultaneously, with β‐sheet rich fibrils acting as the main nucleus for further amyloid formation. Further, this secondary nucleation mechanism suggests the continuous formation of amyloid oligomers from monomers over time, and the persistence of off‐pathway, stable oligomers. Similar findings were observed by Chen and coworkers, who found that α‐syn will form persistent, stable oligomeric species (Chen et al., 2015). This is also consistent with previous findings by our group for TTR, α‐syn, and Aβ (Rodriguez et al., 2023; Zhaliazka & Kurouski, 2022; Zhou & Kurouski, 2020). Additionally, our results in the A‐11 dot blot assay demonstrated similar levels of oligomers present throughout all tested stages of IAPP aggregation. Together, these results suggest that IAPP aggregation leads to the formation of transient on‐pathway and stable off‐pathway oligomers which exist simultaneously with fibrils over different stages of aggregation, and that this phenomenon is conserved across several amyloidogenic proteins.

At early stages of IAPP aggregation (1 and 4 h), we observed heterogeneous populations of ROs, DOs, and very few fibrils via AFM. However, only DOs and fibrils remained after 8, 24, and 48 h of aggregation, with an increase in the total contribution of fibrils. These results indicate that DOs evolve into other amyloid species, while ROs persist throughout aggregation and fibril formation. Our AFM‐IR analysis further supports this conclusion. We observed that IAPP DOs adopt primarily β‐sheet conformations at 1 h of aggregation, and these conformations evolve to become more β‐sheet rich after 4 h of aggregation. Specifically, an increase in parallel β‐sheet content in DOs is observed over time. In the same light, we observed that IAPP fibrils adopt a primarily parallel β‐sheet conformation after 1 h of aggregation, which becomes more prevalent after 48 h. ROs, on the other hand, display a more variable ensemble of conformations over time, indicating that the ROs formed by IAPP are highly dynamic and do not adopt a conserved β‐sheet conformation. Together, this data suggests that DOs adopt β‐sheet rich conformations early in IAPP aggregation and evolve into IAPP fibrils over time, while ROs are highly dynamic, persistent oligomeric species that exist simultaneously with IAPP fibrils. It is important to note that annular protofibrils (APFs) composed of lower order round oligomers has been previously reported and do not contribute to fibril formation (Kayed et al., 2009; Lasagna‐Reeves et al., 2011). However, these APFs require the presence of a membrane to form, as fibrils are insufficient to induce APF formation from oligomers, and are morphologically distinct from the DOs observed in this study. Thus, we conclude that DOs are a unique on‐pathway oligomer that contributes directly to fibril formation.

Previous reports have suggested that lower‐order IAPP oligomers are the most toxic species of IAPP aggregate (Janson et al., 1999; Jeong & An, 2015; Kayed et al., 2009), however, the mechanisms of toxicity elicited by the diverse species of IAPP amyloids still remain incompletely understood. The accumulation of ROS within affected cells has been characterized as a hallmark of IAPP‐mediated toxicity, among other amyloidogenic proteins (Freeman et al., 2013; Pilkington et al., 2016; Zraika et al., 2009). Consistent with these previous findings, we observed increased levels of ROS present within β cells exposed to aggregates formed by IAPP in as little as 1 h of aggregation. This observed level of toxicity was maintained for aggregates formed after 4 and 8 h as well, indicating that early‐stage amyloids oligomers are toxic to β cells. However, after 24 and 48 h of aggregation, formed IAPP aggregates elicited significantly greater ROS as compared to 1, 4, and 8 h aggregate populations. This indicates that the evolution of IAPP amyloids overtime, in addition to the presence of persistent oligomers, contributes to their toxic effects. This is consistent with findings with Pilkington and colleagues, who observed that mature IAPP fibrils elicited the greatest accumulation of ROS as compared to earlier stage aggregates (Pilkington et al., 2016). Together, these findings demonstrate that both the presence of persistent oligomers formed during the early stages of IAPP aggregation and the structural evolution of IAPP amyloids overtime synergistically contribute to cytotoxicity in β cells.

LIMITATIONS OF THE STUDY AND FUTURE DIRECTIONS

4

This study is limited by the use of dried and substrate‐adsorbed samples, which may not fully capture the behavior of IAPP aggregate species in physiological solutions. Further, toxicity assays were only performed using one marker of toxicity in an immortalized cell line. Future studies which confirm these results in primary cells and model organisms would greatly strengthen the presented evidence. Additionally, the molecular mechanisms by which IAPP aggregate populations elicit cytotoxicity will be of great value, and these mechanisms will be focused on in future studies to better understand the link between amyloid structural evolution and cytotoxicity.

EXPERIMENTAL SECTION

5

Protein preparation

5.1

Synthetic human amylin (1–37) was purchased from AnaSpec as a lyophilized powder. The powder was dissolved in 100% 1,1,1,6,6,6,‐hexafluoroisopropanol at a concentration of 1 mg/mL and incubated for 24 h at 4°C. The protein solution was then filtered using a 0.22 μm syringe filter and dried using pressurized nitrogen gas until the solution dried into a film. The formed film was freeze‐dried at −80°C for 48 h then stored at −20°C for no more than 1 month.

Protein aggregation

5.2

Protein film samples were allowed to thaw at room temperature before being placed on ice and suspended in chilled sterile Milli‐Q water to reach a protein concentration of 100 μM. Protein solutions were added to sterile microcentrifuge tubes with sterile 1X PBS to make a final concentration of 40 μM and allowed to aggregate at 37°C with no agitation. Aggregation time points were collected from the aggregation tubes and immediately deposited for AFM imaging.

Thioflavin T (ThT) kinetics

5.3

ThT kinetics experiments were performed in 96‐well nonbinding plates. Each well with a single technical replicate contained 105 μL of solution with a final concentration of 40 μM IAPP and 50 μM ThT in 1X PBS. PBS and ThT were added first to each well, and all empty wells were filled with Milli‐Q water. The plate was prechilled to 4°C before adding chilled IAPP solution to a concentration of 40 μM. This prechilling step is essential to prevent aggregation from a sudden temperature change. The plate was then incubated quiescently at 37°C in a Tecan Spark plate reader, and fluorescent measurements were taken every 10 min for at least 24 h. All experiments were performed in triplicates and repeated twice to ensure reproducibility. Kinetic curves are representative of the average of one of these sets of triplicates. Lag time and half‐time calculations were performed on each of the three triplicates, and bar graphs represent the mean ± SD.

AmyloFit kinetic modeling

5.4

Kinetic curves were uploaded into AmyloFit software using the supplied manual and as described previously (Meisl et al., 2016) Models were fit to each model using global fits for all rate constants and nucleus sizes, and by adjusting the protein concentration to 40 × 10^−6^ M. Bins were set to 50 and each model was allowed to complete all 50 bins. The quality of the fit was judged based upon the mean residual error (MRE).

AFM and AFM‐IR

5.5

For both AFM and AFM‐IR, samples were prepared as described for ThT kinetic assays without the addition of ThT. Sample aliquots were deposited on gold‐coated silica substrate that was washed with DI water. For each sample, 3.5 μL of protein solution was placed on the substrate and allowed to air‐dry for 15 min. After that, a micropipette was used to remove the excess of the protein solution, and the surface was dried at room temperature. Dried samples were washed gently with DI water then gently dried under compressed air. AFM images were collected using a Bruker Nano‐IR 3 system. Budget Sensors ContGB‐G gold‐coated AFM scanning probes were used to collect both AFM images and IR spectra. Scanning probes were optimized using a polymethyl acrylate (PMMA) standard. Due to instrumental chip transition falling in the amide I window, the probe was also optimized on prepared IAPP fibrils at 1630 cm^−1^. Samples were scanned at a rate of 0.7 Hz with a resolution of 256 pts. × 256 pts. A total of 30 spectra were collected per time point per species. The spectra from 1648 to 1652 cm^−1^ were removed from analysis to remove interference from instrumental chip transitions.

AFM height data analysis

5.6

Height data was collected for each individual species analyzed per time point sample, and fibrillar species were measured in several locations along their lengths to ensure accurate reflection. Fibrils that appeared overlapped were avoided. Donut‐shaped oligomers were measured from the outside of the donut shape to the inner depression, taking a cross section of the highest point of the species. Each oligomeric species had only one respective height point collected.

IR spectral fitting

5.7

The collected IR spectra were normalized and smoothed in MATLAB using the Eigenvector PLS toolbox. The IR spectra were imported into GRAMS™ spectroscopy software to peak fit with a Gaussian function. To do this, the amide I region was baselined, peak fitting was performed, and the area for generated curves was calculated. We assigned the following peak windows to their respective secondary structures: parallel β‐sheet (1615–1640 cm^−1^), disordered (1641–1670 cm^−1^), anti‐parallel β‐sheet (1671–1699 cm^−1^).

A11 dot‐blot analysis

5.8

IAPP samples were prepared as described for ThT kinetics without the addition of ThT. IAPP samples were prepared at a concentration of 40 μM and aggregated for their respective time points at 37°C quiescently before being serial diluted in TBS. Diluted samples were applied to a TBS‐prewet 0.22 μm nitrocellulose membrane (Bio‐Rad) with a Bio‐Rad Bio‐Dot Apparatus (Bio‐Rad) using the manufacturer's instructions. The membrane with protein bound was then blocked with 1% BSA in TBS before being washed with TBS‐Tween20 (TBST) three times. The primary A11 anti‐oligomer antibody was diluted to a concentration of 1 μg/mL in TBS, and this solution was applied to the membrane in a volume of 100 μL. The membrane was washed again three times with TBST. 100 μL Goat anti‐Rabbit IgG (H + L) Secondary Antibody, HRP (Invitrogen) diluted in 1x TBS to a final concentration of 0.18 μg/mL was then added. After three washes with TBST and a final two rinses with TBS, the membrane was developed with SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Scientific) using the manufacturer's instructions before exposing. The membrane was exposed and imaged using GE Amersham Imager 600 Luminescence Image Analyzer (General Electric). The integrated density method was utilized, and each dot was measured in Fiji/ImageJ after subtracting background. Integrated density values were plotted in JASP software.

Cell culture and toxicity assays

5.9

BRIN‐BD11 Rat pancreatic β cells were purchased from Millipore and maintained at 37°C in a 5% CO_2_ incubator using RPMI 1640 media supplemented with 10% FBS and Normocin. Cells were trypsinized from their flask and seeded at a concentration of 30,000 cells/well in a 96‐well treated plate. Cells were allowed to reach ~80% confluence by incubating overnight before treating with 30 μL of sterile 40 μM IAPP samples to yield a final concentration of 12 μM IAPP aggregates per well. IAPP samples were prepared identically to samples for AFM, AFM‐IR, and A11, maintaining both concentration and time of aggregation for direct comparison between assays. Exposure to IAPP aggregates continued for 24 h before staining cells with CellROX Deep Red (Invitrogen) per the manufacturer's instructions. Fluorescence intensity was quantified using flow cytometry using at least 3 independent cultures per condition and repeating each experiment twice.

AUTHOR CONTRIBUTIONS

Daniel Warren: Conceptualization; investigation; writing – original draft; methodology; validation; visualization; writing – review and editing. Jadon Sitton: Conceptualization; investigation; writing – original draft; writing – review and editing; visualization; methodology; validation. Dmitry Kurouski: Conceptualization; writing – review and editing; project administration; supervision; resources.

CONFLICT OF INTEREST STATEMENT

The authors declare no competing financial interests.

Supporting information

Table S1. Results of Tukey test for amount of secondary structure in different IAPP aggregates.