α-Synuclein fibrils deposited on mica were scanned both in tapping mode and contact mode, respectively, for determining the height and finding the indentation points for the SPI measurements. We determined an average fibril height of 9.0 ± 0.4 nm (N = 60) from the tapping mode images. This average height value was used to determine the effective contact surface in the indentation measurements according to the model shown in Figure 1. The fibril heights measured in contact mode imaging were considerably lower and were therefore not used in determining the average fibril height. This was attributed to the pressure from the tip on the sample. The force exerted on fibrils with the 0.1 N/m cantilever during scanning was between 0.5 and 1 nN.

We performed nanoindentation experiments on five fibrils, each of which was indented 8 times at different locations along its length. A typical force distance curve resulting from this procedure is shown in Figure 2A. The absence of adhesion during the measurements allowed the use of the Hertz model.

Although every fibril was indented 8 times, not all curves were suitable for analysis. For some curves, the r² values of the linear fit did not exceed 0.95 in the part where the tip was indenting the fibril (part 1 in Figure 2C). From the curves that were analyzed, an average elastic modulus of 1.3 ± 0.4 GPa (N = 31) was found for α-synuclein fibrils.

A sample of α-synuclein fibrils deposited on mica was scanned. Figure 3 shows two typical images recorded, with corresponding height and elasticity profiles. The fibrils show considerably lower modulus of elasticity compared to the background. However, the edges of the fibril show increased modulus of elasticity values, also displayed in the cross-section of the fibril shown in Figure 3F. We attribute this effect is due to the changing contact area compared to the contact area shown in Figure 1 where the tip is indenting the middle of the fibril. This artifact is also visible in the height images derived from the harmonic force mode, shown in Figure 3E, and they are therefore not used in further analysis. For each individual fibril, the values for the elastic modulus measured along the fibril were averaged. The average value was 1.2 ± 0.2 GPa (N = 95).

The surface property mapping technique Peakforce QNM is able to image the sample both in ambient conditions and in buffer solution. Figure 4 shows height images and the corresponding elasticity maps obtained with Peakforce QNM of α-synuclein fibrils, obtained in buffer (Figure 4A, B) and in air (Figure 4C, D). These images were obtained with a high setpoint of around 15 nN and show that for both liquid and ambient conditions the height and elasticity ranges which can be obtained with Peakforce QNM are similar. However, this large setpoint causes the fibrils to break, especially in liquid, see Figure 4A.

To prevent damage to the α-synuclein fibrils, a lower setpoint of 1-2 nN was used. This resulted in intact fibrils with significant lower values of the elastic moduli (Figure 5). The elastic modulus for each fibril is determined from the average value of the DMT modulus obtained along the fibril length. This resulted in a modulus of elasticity of 1.3 ± 0.3 GPa (N = 57) for the fibrils in ambient conditions and 1.0 ± 0.2 GPa (N = 59) for those in liquid.

In order to measure the elastic properties of a material, the choice of the cantilever is key. In nanoindentation the highest sensitivity (and thus accuracy) is achieved if the spring constant of the probe cantilever is identical to the effective spring constant of the sample (also referred to as contact stiffness), see Figure 6. If the spring constant of the cantilever is more than 10 times lower or higher than that of the sample, the sensitivity is about 3 times lower, see Figure 6 making the determination of the elastic modulus less accurate. Practically since one does not know the stiffness of the sample a priori, an estimation is necessary. This is also the case for the surface mapping methods. The nominal elastic modulus ranges accessible by HarmoniX and Peakforce QNM are 10 MPa-10 GPa and 0.7 MPa-70 GPa, respectively 18. However, as noted above, this range depends on the cantilever that is used for the measurements and is in practice significantly smaller.

A second point to consider when choosing the cantilever is the adhesion between the tip and the sample. The spring constant of the cantilever should be sufficiently high to create enough force to come loose from the surface. In the PeakForce QNM experiments reported here on protein fibrils, performed in ambient air, an adhesion of few nanonewtons was observed. For reproducible and proper deflection curves in air we used in this case a cantilever with a spring constant of approximately 27 N/m. In the HarmoniX mode the fibrils are measured in a special tapping mode. In this mode reproducible results were obtained with cantilevers with medium stiffness of 2 N/m in ambient conditions. The cantilever used for the nanoindentation measurements (0.1 N/m) showed an incredibly large artifact in both approach and retract curves at the 1 kHz ramp rate in Peakforce QNM in liquid, which was not seen in the nanoindentation measurements. This artifact could be induced by the impact of the effective mass and damping forces at the working frequency of 1 kHz. These hydrodynamic forces acting on the cantilever are frequency-dependent 2728. Although we do not know how the Peakforce QNM software compensates for this, it is possible these effects in liquid could interfere with the measurements. Scanning with stiffer cantilevers with nominal spring constant 2.8 N/m yielded reproducible results.

Finally, in addition to choosing the optimal cantilever stiffness, it is also important to ensure that the resonance frequency for Peakforce QNM imaging is above 10 kHz, in order not to interfere with the 1 kHz ramping.

The calibration of all three methods is difficult and consists of several steps. For all methods one needs the deflection sensitivity, the spring constant of the cantilever and the tip radius. For the SPI experiments the tip radius can be determined afterwards. Both surface mapping methods need the tip radius as an input parameter before measuring. For HarmoniX, in addition to this tip radius, some additional parameters, such as the torsional frequency, are needed. An alternative way of calibration of the surface mapping methods was done with the reference sample (see "Methods" section). In this study, this reference sample is only used in the HarmoniX measurements.

E46K disease mutant α-synuclein was recombinantly expressed and purified as previously described 4. A 100 μM monomeric E46K solution in 10 mM Tris-HCl, 50 mM NaCl, pH 7.4 was incubated at 70°C in Eppendorf tubes under constant shaking. After 27 h, well-defined protein fibrils were formed in solution, which was verified by a Thioflavin T fluorescence assay specific for cross-beta structures characteristic of amyloid fibrils.

Samples for AFM imaging in liquid were prepared by placing 50 μl of a 5× diluted solution containing fibrils on the mica substrate. This solution was allowed to adsorb for 10 min and then washed gently with 200 μl buffer. For imaging, 80 μl of fresh buffer solution was placed on the sample. We used the same buffer solution (10 mM Tris-HCl, 50 mM NaCl, pH 7.4) for both dilution and imaging. For the measurements performed in ambient air, a 10× diluted protein solution was placed on mica substrates and allowed to adsorb in the same manner as described above. Subsequently, the sample was washed with 200 μl milliQ water and dried with a gentle nitrogen stream.

The tip radius was determined with two different methods. First, from the AFM height images of protein fibrils the tip radius was derived from the fibril height-to-width ratio based on tip-sample convolution 2131. Only fibrils that were perpendicular to the scan axis were used. From the tip sample convolution method an average tip radius of 100 nm was determined. Second, the tip was imaged by scanning electron microscopy (Philips XL30 ESEM-FEG). With the SEM, the average tip radius was found to be approximately 80 nm. For both methods, the tip resulted in a considerably larger number than the nominal tip radius provided by the manufacturer. An average value of 90 nm was used in the analysis with an error of 30%.

The cantilever spring constants were determined with the thermal noise method implemented in the Veeco software and were assumed to have a 5% error 33.

A Bioscope II microscope (Veeco, Santa Barbara, CA, USA) was used for the SPI experiments. In order to measure the fibril heights, AFM tapping mode images were recorded in a physiological buffer (10 mM Tris-HCl, 50 mM NaCl, pH 7.4) in tapping mode with low force settings (reduced to 3 nm, 80-90% of the free amplitude) to minimize interaction with the sample. We use silicon nitride probes (MSCT, tip F, 0.5 N/m, Veeco, Santa Barbara, CA, USA) for these measurements. The average fibril height measured in tapping mode is used to determine the surface contact area for all three indentation methods. The indentation measurements were performed with the "Point and Shoot" application in the NanoScope 7.30 (Build R2Sr1.) software. To locate the indentation locations we first imaged the fibrils in contact mode using another probe (MSCT, tip E, 0.1 N/m, Veeco, Santa Barbara, CA, USA). This probe was selected to, on one hand, minimize the forces during contact mode imaging and, on the other hand, to match the spring constant of the cantilever to the stiffness of the sample for the indentation measurements. Every fibril was indented approximately 8 times at different positions along its length. Prior to fibril indentation, force curves were recorded on the mica substrate close to the fibril to determine deflection sensitivities of the cantilevers.

The raw deflection curves, obtained in the SPI mode, were converted to a force separation curve using the deflection sensitivities and the spring constants of the cantilevers in a custom written Matlab program. To extract the elastic modulus from the force separation curve, the Hertz model was used to analyze the force curve 39. This model, in the case of a spherical indenter on a cylinder shaped object, is given in Figure 1 where F is the load, v the Poisson ratio, δ the separation, and E the modulus of elasticity. The equivalent contact radius R_eq for a spherical indenter with radius R_t, with an infinitely long cylinder with radius R_f is given by the expression in Figure 1. The modulus of elasticity was determined from the slope of the curve where F^2/3 was plotted versus the separation. Small segments along this curve were fitted to a linear equation and the r² value was determined for every fit, yielding an elastic modulus as a function of separation. From the point the force increases, the r² value increases and only fits with an r² above 0.95 were used in the analysis. From the point of contact the modulus of elasticity values for the following 2 nm were averaged (20% indentation 12).

Due to the finite thickness effects, the obtained modulus of elasticity is influenced by the stiff underlying substrate (mica). A correction factor for this effect was applied which was a function of the maximum applied force and the value of the uncorrected modulus of elasticity 1636.

All analysis steps were implemented in a custom Matlab program. The algorithm analyzes both the force curve and the r² curve to accurately determine the point-of-contact, that is, the separation at which the tip starts indenting the fibril. This point is defined as the point where the force distance curve leaves the baseline, and r² adopts a value higher than 0.95.

HarmoniX was performed under ambient conditions (that is, at room temperature without further control of humidity) on a Veeco Multimode microscope with a Nanoscope V controller (Veeco, Santa Barbara, CA, USA). The analysis software uses the DMT model 40. Torsional cantilevers (TL01, MikroMasch, Tallinn, Estonia) with a nominal spring constant of 2 N/m were used. The measured vertical and torsional resonance frequencies were 111 kHz and 1.1 MHz, respectively. The system was calibrated with a reference sample (model PS-LDPE, Veeco, Santa Barbara, CA, USA) 20. Since HarmoniX assumes a spherical tip that indents an infinitely large and thick flat elastic surface, the value for the modulus of elasticity needs to be corrected offline. The first correction factor applied is to account for the different geometry, which in these experiments is a spherical tip indenting an infinitely long cylinder, see Figure 1. The correction factor used here is 2.1. The second correction factor was applied to account for the finite sample thickness of the protein fibril. A correction factor of 1.3, determined by the SPI measurements, was used.

Peakforce measurements were done on a Veeco Bioscope Catalyst microscope with a Nanoscope V controller (Veeco, Santa Barbara, CA, USA). The analysis software uses the DMT model 40. The measurements were done both in ambient conditions (uncontrolled humidity, temperature, and air pressure) and physiological buffer (10 mM Tris-HCl, 50 mM NaCl, pH 7.4). The manufacturer provides a list of optimal cantilevers to measure specific ranges of elastic moduli. For the ambient measurements the stiff RTESP cantilevers (26.9 N/m, Veeco, Santa Barbara, CA, USA) were used, due to the high adhesion forces observed for other, less stiff cantilevers. For the measurements performed in buffer we used a medium stiff cantilever: FMR-10 cantilevers (nominal spring constant 2.8 N/m, Nanoworld, Neuchâtel, Switzerland). Here, the elastic moduli are also corrected offline as described for the HarmoniX data (see Harmonic force microscopy).

Using SPIP software (Image Metrology A/S, Lyngby, Denmark), a trace was drawn on top of the fibril to determine the average height from the height images or modulus of elasticity from the stiffness maps of the individual fibrils, according to the procedure described in 4. A point of potential confusion is that both HarmoniX and Peakforce QNM create so-called 'stiffness' maps, which in the software is expressed in units of Pa. Technically this is not correct, since stiffness is expressed in units of N/m. The parameter in these images is a modulus of elasticity which is expressed in Pa. In this manuscript we therefore refer to these values as moduli of elasticity. All images in this article are linewise corrected. Actual measurements are done on uncorrected images.