• adhesion force;
  • atomic force microscopy;
  • fibrous proteins;
  • hydrophobic effect;
  • protein engineering

Protein denaturation and misfolding can lead to the formation of amyloid fibrils in organs and tissues. These deposits, or plaques, are detrimental to the proper functioning of the organism, and can destroy the natural balance of the cells, transport, and cell-membrane integrity, and can lead to protein conformation diseases, such as Alzheimer′s, Parkinson′s, type II diabetes, prion-associated encephalopathies, and others.1 Usually, proteins synthesize as non-folded polypeptides. Then, through a partially folded intermediate process, they form monomers consisting of a natural α-helical fibril or random coil conformation. If these non-folded or partially folded proteins undergo degradation or aggregation, they can form a unique, highly misfolded amyloid fibril that is rich in the β-sheet structure.2

Insulin self-assembles in response to environmental conditions. For example, helical polypeptide insulin rapidly grows into amyloid fibrils with cross-β structure under low-pH and high-temperature conditions.3 However, the emphasis in recent studies has been on the aggregation of insulin monomer into fibers, while rarely exploring fibril structural evolution under changing environmental conditions. Researchers have presumed that continuous structure changes in insulin amyloid fibrils due to its fibers are not maintained in the most stable thermodynamic state;4 however, the end results have not been conclusive.

In the past, researchers explored the relationship between fibril structure and aggregation behavior by varying the amino-acid sequences to generate different fibril forms.5 They observed that the aggregation rate increased for fibrils with more hydrophobic functional groups present at the surface, and proposed that hydrophobic surface interactions played an important role in the formation of amyloid aggregates and plaques.6 However, the amino-acid sequence is not the only factor that determines the protein structure. There are numerous possible folding geometries, and environmental conditions vary, thus, it is very difficult to predict the final structural form from the linear amino acid sequence.

In our study, we have focused on directly measuring changes in the surface properties of fibrils during the process of structural evolution under controlled environmental conditions. We characterized the surface properties of insulin fibrils by performing atomic force microscopy (AFM) measurements, and we correlated with AFM imaging to monitor the surface morphological evolution and further calculated the average growth rate within each image in three different stages of incubation times. We observed a chemical change in the fibrils, from hydrophilic to hydrophobic character, which occurred concurrently with a structural transition from the α-rich to the β-rich form between five and seven hours. This demonstrates an important correlation between fibril chemistry and structural evolution, which could lead to new strategies for prevention or early treatment of protein conformational diseases.7

Previous studies had proposed that hydrophobic groups on the fibril surfaces may lead to amyloid aggregation and plaque formation. Here, we used AFM to directly measure the surface adhesion of insulin incubated for two and ten hours by performing force curves on both types under ambient environmental conditions. A force curve is a point-specific measurement on a sample in which the AFM tip vertically approaches the surface, makes contact, deflects, and then retracts from the surface. As the tip is retracted, adhesion forces between the tip and sample cause the cantilever to deflect downward, this can be observed in the force curve profile and used to determine the adhesion force.

Histograms of the adhesion values are shown in Figure 1 a, and representative force curves for each sample type are shown in Figure 1 b, where the arrows indicate the region of the force curve where adhesion was measured.

thumbnail image

Figure 1. a) Histograms of adhesion-force values from insulin fibril growth after ten hours and two hours incubation time, and on a bare silicon substrate. b) Representative individual AFM force–distance curves. The insets in (b) show cartoon depictions of the surface chemistry on the AFM tip (OH-terminated) and on the insulin (top, middle) and silicon (bottom) surfaces.

Download figure to PowerPoint

The adhesion force values, determined by fitting a Gaussian to the histogram peaks, were 12.7±4.8, 50.6±8.2, and 67.8±9.5 nN for the 10-hr, 2-hr, and clean silicon sample surfaces, respectively. This trend was observed in data collected from five different regions of samples. The increment of adhesion value is accompanied with corresponding increment of surface chemistry due to silicon cantilever provides hydrophilic character and makes a strong interaction force to the hydrophilic probe surfaces (Figure 1 b inset).

The AFM-based adhesion measurements carried out on insulin showed that changes in the surface chemistry of the fibrils occurred in relation to changes in their structural form. In the early stages of incubation, the 2-hr-incubated fibril structures, which exhibit hydrophilic character, are the dominant form. With continued heating, fibril aggregates may begin to form, which signaled the structural transformation from the α-helix to the β-sheet form and the alteration of the chemical character from hydrophilic to hydrophobic. The 10-hr-incubated fibril structure possesses a major hydrophobic character so that it shows a lower force value. Significantly, the adhesion values from AFM force measurements for the 10-hr-incubated fibril structure were four times lower than those for the 2-hr-incubated fibril.

To characterize the effects of incubation time on insulin fibril growth that cause a different surface chemistry, we deposited samples prepared as above on clean silicon surfaces at room temperature, and used AFM to image the fibril structures. The AFM image offers the capability of visualization in three dimensions, as well as qualitative and quantitative parameters including height, volume, surface area, and roughness.8 We examined three fibril properties as a function of the incubation time: 1) length, 2) height, and 3) volume. Prior to heating, the insulin “monomers” were spherical (height 0.6±0.4 nm) and dispersed readily on the silicon substrate, as shown in Figure 2 a. After heating for 0.5 hours, we observed a slight aggregation of the monomers (Figure 2 b), and after two hours, linear fibrils had started to form (Figure 2 c, d). After incubating for 2.5 to 3.5 hours (Figure 2 e, f), the average measured length of the insulin fibrils was greater than 1 μm.

thumbnail image

Figure 2. AFM topographical images (1×1 μm2) of Bovine insulin incubated in solution at pH 1.6 and 80 °C for: a) 0, b) 0.5, c) 1.5, d) 2, e) 2.5, f) 3.5, g) 5, h) 7, and i) 10 hours, and then deposited onto a silicon wafer substrate.

Download figure to PowerPoint

After five hours incubation time (Figure 2 g), we observed significant fibril aggregation with high coverage densities after 7–10 hours (Figure 2 h, i).

The fibril thickness, or “height” as measured by AFM, also increased with the incubation time (Figure S1, Supporting Information). The measured fibril heights from the topographical images shown in Figure 2 a–i were 0.6±0.4m, 0.6±0.3, 2.0±0.1, 3.0±0.4, 3.4±0.2, 4.7±0.1, 3.0 ∼6.0, 3.0 ∼9.0, and 3.0 ∼15 nm, respectively. The height of a single protofilament on the silicon substrate was 3.0 ∼4.0 nm. After incubating for five hours, we observed that the protofilaments began to aggregate, and after seven hours, the fibril thickness was four times that of an individual protofilament.

Such a morphological or structural evolution has been widely investigated in the field of insulin self-assembly to fibrils. However, we applied circular dichroism (CD) spectroscopy and Fourier transform infrared (FTIR) spectroscopy to monitor secondary structural transitions of the insulin as a function of the incubation time (Figure S2) to further verify the results of AFM force measurements. CD spectra up to five hours incubation times, showed a double minima at 208 and 222 nm, which indicated the presence of the α-helical structure (Figure S2a). After seven hours, the CD spectra showed a significant decrease in the ellipticity of the feature at 222 nm, accompanied by a shift in the minimum at 208 nm to 218 nm. These changes indicate a structural transformation from α-helix to β-sheet conformation.

FTIR spectra were acquired and are presented in Figure S2b. The amide I band at 1657 cm−1 was sharp at short incubation times, consistent with the presence of the α-helix structure.9 After a five-hour incubation time, we observed a shift and split of the IR band at 1657 cm−1 to 1664 and 1632 cm−1. After seven hours, the peak shift was more pronounced, with the 1632 cm−1 peak becoming the dominant spectral feature. Dzwolak et al.10 proposed that the amide I peak shift in the FTIR spectra was due to an α-helix to β-sheet structural change. Both CD and FTIR data exhibited significant changes in the insulin fibril spectra after seven-hour incubation time, with complete transformation occurring after ten hours. We attribute these changes to the α-helix to β-sheet transformation.

Adhesion force and fibril thickness increased severalfold with the incubation time, which indicates that the insulin fibrils tended to aggregate and became hydrophobic during the fibrillogenesis process. Additionally, CD and FTIR showed a secondary structural change from α-helix to β-sheet transformation within the period of incubation times. These results demonstrate that the fibrillogenesis process starts from a hydrophilic surface with α-helix-rich contents, which converts into a hydrophobic surface with β-sheet-rich aggregates. Thus, changing the surface chemistry was correlated with fibril-growth dynamics and aggregation rate.

A rough mask surface, from AFM scans of both the desired area and the entire area, with the ratio of both values being divided by time, yields statistic information on the sample growth rate. Here, we analyzed the fibril aggregation rate as a function of the incubation time by calculating the total fibril volume within each 1 μm2 AFM image from Figure 2 and plotting the results graphically in Figure 3.

thumbnail image

Figure 3. Plot of the fibril volume per unit area for each AFM image in Figure 2.

Download figure to PowerPoint

We divided the volume-versus-time data into three sections: 0–3.5, 3.5–7, and 7–10 hours due to their individual linear relationship, and then calculated the average growth rate in each section. The fibrils grew slowly within the first section with an estimated rate of 2.4×10−5 μm3 hr−1. From 3.5–7 hours, the rate increased to 1.3×10−4 μm3 hr−1, and then slowed to 3.4×10−5 μm3 hr−1 during the third section (7–10 hours). According to the van der Waals force of molecular interaction, the additional hydrophobic interactions induce fibril aggregation, resulting in a volume change. Assuming that the system′s pressure remains constant, the work that is generated is stacking free energy, as described by Akiyama et al.11 We observed a fivefold increase in the aggregation rate over the same incubation period as the structural and chemical transformations (5–7 hours), which triggered the formation of larger fibril aggregates and assemblies.

Previous studies suggest that the β-sheet structure is the key determinant of aggregation with hydrophobic residues, including valine, isoleucine, phenilanalanine, and tyrosine.1c Namely, the aggregation rate of β-sheet containing amyloid-like fibrils would be promoted by their relatively high hydrophobicity.9 The emerging feature from our work and the related studies12 is that the aggregation-promoting regions usually are hydrophobic, indicating a strong correlation between aggregation rate and hydrophobicity of amyloidogenesis.

In summary, we have investigated the structural transition of insulin fibrils incubated from zero to ten hours at pH 1.6 and 80 °C using several complimentary analytical techniques. AFM force measurements combined with CD and IR results indicate a transition point of the secondary structure from the α-rich to the β-rich form between five and seven hours, during which the aggregation rate increased fivefold, and the fibril surfaces changed from hydrophilic to hydrophobic character. From these results, we infer that the presence of β-rich-form amyloid fibril leads to dense packing of fibril aggregates due to strong hydrophobic interactions between fibrils. Thus, the β-rich form is implicated in amyloid diseases which elicit from plaque formation within bodily tissues and organs.

Experimental Section

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

Sample Preparation

Insulin fibril formation was carried out according to the method described by Jansen et al.,13 using insulin powder from bovine pancreas (Sigma Aldrich, St. Louis, MO, USA) without further purification. A stock 1 mg mL−1 insulin solution was pH-adjusted to 1.6 using diluted HCl (Scharlau, Australia), and then incubated at 80 °C for periods of 0–10 hours.

AFM Measurements

Each sample was prepared for AFM (Asylum Research MFP-3D, USA) measurements by drop-casting a 20 μL aliquot of incubated fibril solution onto a clean section of silicon wafer for 3 min to adsorb the insulin. Unbound fibrils were removed by subsequent rinsing. Topographic images were acquired in the AC mode using an AFM under ambient conditions. A silicon cantilever (Nanoworld Arrow-FMR, Switzerland) with a scan rate of 0.6∼1.0 Hz, and an image resolution of 512×512 pixels, as well as with a spring constant of 2.3±0.4 N m−1 (calibrated using the thermal method),14 was used for all adhesion measurements.

Surface Structural Characterization

The chemical structures of the samples were characterized at various room-temperature incubation times by using CD, FTIR spectroscopy, and AFM. Sample substrates were prepared from sections of a silicon (100) wafer (Tekstarter Co., Ltd.; P-type/Boron dopant). All of the silicon substrates were plasma-cleaned (Harrick Scientific Products, Inc.) for 2 min using dry air as the reactive gas to increase the OH concentration at the surface. This process creates a uniformly hydroxylated surface.15 CD measurements were performed on a Jasco J-810 spectrometer using a cell of 1 mm optical path length. The solvent spectra were the average of three scans. IR spectra were recorded (4 cm−1 resolution 512 scans, at sample compartment vacuum pressure was 0.12 hPa) using a VERTEX 80/80v FTIR spectrometer provided by Bruker Optik GmbH.


  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

The authors would like to thank the National Science Council of Taiwan (NSC 101–2113 M-110–013-MY3), and the National Sun Yat-sen University Biochip Research Group for financial support of this work. Prof. Hsieh also thanks Dr. David Beck for helpful discussions and proofreading.

Supporting Information

  1. Top of page
  2. Experimental Section
  3. Acknowledgements
  4. Supporting Information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.


Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.