Characterizing Individual Protein Aggregates by Infrared Nanospectroscopy and Atomic Force Microscopy

Francesco Simone Ruggeri; Tomas &#352;neideris; Sean Chia; Michele Vendruscolo; Tuomas  P. J. Knowles

doi:10.3791/60108

JoVE Journal > Biochemistry

生物化学

Characterizing Individual Protein Aggregates by Infrared Nanospectroscopy and Atomic Force Microscopy

Published: September 12, 2019

doi:

10.3791/60108

Francesco Simone Ruggeri¹, Tomas Šneideris^1,2, Sean Chia¹, Michele Vendruscolo¹, Tuomas P. J. Knowles^1,3

¹Centre for Misfolding Diseases, Department of Chemistry,University of Cambridge, ²Institute of Biotechnology, Life Sciences Center,Vilnius University, ³Cavendish Laboratory, Department of Physics,University of Cambridge

Summary

We describe the application of infrared nanospectroscopy and high-resolution atomic force microscopy to visualize the process of protein self-assembly into oligomeric aggregates and amyloid fibrils, which is closely associated with the onset and development of a wide range of human neurodegenerative disorders.

Abstract

The phenomenon of protein misfolding and aggregation results in the formation of highly heterogeneous protein aggregates, which are associated with neurodegenerative conditions such as Alzheimer’s and Parkinson’s diseases. In particular low molecular weight aggregates, amyloid oligomers, have been shown to possess generic cytotoxic properties and are implicated as neurotoxins in many forms of dementia. We illustrate the use of methods based on atomic force microscopy (AFM) to address the challenging task of characterizing the morphological, structural and chemical properties of these aggregates, which are difficult to study using conventional structural methods or bulk biophysical methods because of their heterogeneity and transient nature. Scanning probe microscopy approaches are now capable of investigating the morphology of amyloid aggregates with sub-nanometer resolution. We show here that infrared (IR) nanospectroscopy (AFM-IR), which simultaneously exploits the high resolution of AFM and the chemical recognition power of IR spectroscopy, can go further and enable the characterization of the structural properties of individual protein aggregates, and thus offer insights into the aggregation mechanisms. Since the approach that we describe can be applied also to the investigations of the interactions of protein assemblies with small molecules and antibodies, it can deliver fundamental information to develop new therapeutic compounds to diagnose or treat neurodegenerative disorders.

Introduction

Over 40 million people worldwide are currently affected by neurodegenerative disorders, such as Alzheimer’s (AD)¹ and Parkinson’s (PD)² diseases. More generally, more than fifty pathologies are associated at the molecular level with protein misfolding and aggregation, a process that leads to the proliferation of insoluble fibrillar protein aggregates, known as amyloid deposits³^,⁴. The molecular origins of neurodegeneration and its links with protein conformational changes of proteins leading to amyloid formation, however, remain unclear, in large part because of the high level of heterogeneity, transient nature and nanoscale dimensions of the pathological aggregates⁴^,⁵.

Highly successful investigations of protein structures in the last several decades have been based widely on the use of bulk methods, including X-ray crystallography, cryo-electron microscopy and nuclear magnetic resonance spectroscopy⁵^,⁶^,⁷^,⁸^,⁹. Within this class of techniques, infrared (IR) spectroscopy has emerged as a sensitive analytical tool to unravel the chemical properties of biological systems such as proteins⁸. IR methods allow the quantification of protein secondary and quaternary structural changes during their misfolding and aggregation. In addition, in order to further decipher at the microscopic level the mechanistic details involved in the complex free energy landscapes of protein during their aggregation, a major advance has been the development of chemical kinetics tools to extend to complex self-assembly pathways including amyloid fibrils formation⁵^,⁶^,⁷^,¹⁰^,¹¹^,¹². However, bulk spectroscopic methods provide only average information on the heterogeneous ensemble of species present in solution or involved in specific microscopic steps, thus rendering the investigation of the biophysical properties of individual aggregated species challenging at the nanoscale level¹³^,¹⁴.

Several microscopy techniques with the capability of operating on scales smaller than the diffraction limit of light have emerged in the last decades. This class of methods includes electron microscopy (EM) and atomic force microscopy (AFM). While scanning electron microscopy (SEM) and transmission electron microscopy (TEM) provide two-dimensional (2D) images of a specimen, AFM has emerged in the last decades as a powerful and versatile technique to study three-dimensional (3D) morphologies, as well as the nanomechanical properties of a sample with sub-nanometer resolution¹³^,¹⁴^,¹⁵^,¹⁶^,¹⁷^,¹⁸^,¹⁹^,²⁰^,²¹^,²²^,²³^,²⁴^,²⁵^,²⁶^,²⁷. The rationale behind studying protein aggregation via AFM is that this approach enables the investigation of the morphology of individual species present in solution¹³^,¹⁴^,¹⁶^,¹⁷^,¹⁹^,²⁰^,²¹^,²⁵^,²⁷^,²⁸^,²⁹^,³⁰^,³¹^,³²^,³³^,³⁴^,³⁵^,³⁶^,³⁷. In particular, by monitoring the sample as a function of time, AFM allows the investigation of the evolution of the morphology of the species within the sample, which makes it possible to follow and visualize the pathways of amyloid formation²³^,²⁵^,³⁸^,³⁹^,⁴⁰^,⁴¹^,⁴². Furthermore, AFM enables the quantification of structural parameters such as cross-sectional heights and lengths of the individual species present in solution¹³^,¹⁹^,³⁰^,³¹^,³²^,³³^,³⁴^,³⁵^,³⁶^,³⁷^,⁴⁰^,⁴³^,⁴⁴^,⁴⁵^,⁴⁶^,⁴⁷^,⁴⁸. However, the study of a single biophysical property, such as morphology, is often not sufficient when studying heterogeneous and complex biological systems. AFM, SEM or TEM imaging methods alone do not readily reveal the chemical properties of heterogeneous species of amyloid aggregates at the nanoscale.

A major advance for the analysis of heterogeneous biological samples at this scale has been made recently with the development and application to the field of protein aggregation of infrared nanospectroscopy (AFM-IR)²⁴^,²⁶^,³⁸^,⁴²^,⁴⁹^,⁵⁰^,⁵¹^,⁵². This innovative method exploits the combination of the spatial resolution of AFM (~1−10 nm) with the chemical analysis power of IR. The AFM-IR technique is based on the measurement of the photothermal induced resonance effect driven by an IR laser, and on the measurement of the thermal expansion of the sample under investigation by the AFM tip. The sample can be illuminated by the IR laser directly from the top or from the bottom in total internal reflection, similarly as in conventional infrared spectroscopy²⁴^,⁴²^,⁵²^,⁵³. The IR laser can be pulsed with typical frequencies in the order of hundreds of kilohertz (1−1000 kHz) and tuned over a wide spectral range, typically between 1000−3300 cm^-1. Although the laser source covers an area of ~30 µm diameter, the spatial resolution of the AFM-IR technique is determined nominally by the AFM tip diameter, which detects the local thermal expansion of the system. AFM-IR is well suited to study biological samples because the IR signal is proportional to their thickness up to 1−1.5 µm, and the resulting IR spectra are generally in agreement with the corresponding FTIR transmission spectra¹³^,⁵⁴^,⁵⁵. For this reason, established methods of analysis in spectroscopy can be readily applied, such as the study of chemical shifts, band shape change and de-convolution by second derivatives analysis⁵². Overall, combining the spatial resolution of AFM with the chemical recognition power of IR spectroscopy, AFM-IR enables the simultaneous acquisition of a wide range of morphological, mechanical and chemical properties of a sample at the nanoscale.

Here, we illustrate a protocol for the characterization of the process of protein aggregation that exploits the combination of in vitro fluorescence assays, high-resolution AFM imaging and nanoscale AFM-IR. This combined approach has already excelled in providing detailed results in studying the chemical and structural properties of individual micro-droplets formed by protein aggregates, in the study of liquid-liquid protein phase separation, and in investigating the heterogeneity and biophysical properties of individual aggregated species at the nanoscale²³^,²⁶^,³⁸^,⁴⁵^,⁵⁰^,⁵³^,⁵⁶^,⁵⁷.

Protocol

1. Aggregation assays on fluorescence plate readers

NOTE: The protocol described here is an example of how to study the aggregation of any protein or peptide by chemical kinetics. In particular, it describes an optimized protocol to study the aggregation of the Aβ42 peptide, which is involved in the onset and progression of Alzheimer’s disease⁵⁸^,⁵⁹. A similar protocol can be adjusted and adopted towards studying the aggregation of any protein or peptide.

Obtain a highly pure monomeric solution of Aβ42 through ion-exchange and size-exclusion chromatography techniques to distinguish protein fractions containing Aβ42, and to isolate the monomeric fraction from other aggregated forms of Aβ42 obtained⁵⁸^,⁵⁹.
Dilute the Aβ42 peptide in 20 mM sodium phosphate buffer, 200 µM EDTA at pH 8.0⁵⁸ to a desired final concentration ranging between 1−5 µM, in 1.5 mL low-bind tubes. Samples collected for AFM measurements should not contain the fluorescent dye used in the aggregation assay (thioflavin T, ThT), as it might introduce artefacts during sample deposition for AFM analysis. Thus, prepare triplicates of two identical solutions: i) the first containing monomeric Aβ42 to follow the process of aggregation by AFM; ii) the second containing the Aβ42 monomeric with addition of 20 µM ThT as a tracer to monitor the kinetics of aggregation.
NOTE: It is important to perform steps 1.1 and 1.2 on ice. This procedure is to ensure that the monomeric Aβ42 solution does not aggregate until initiated in the plate reader. When handling samples, careful pipetting is essential to avoid the introduction of any air bubbles that may affect the protein sample.
Pipette 80 µL of each sample in each well of a 96-well, half-area plate of black polystyrene with clear bottom and nonbinding surfaces.
Seal the plate with a foil to minimize the evaporation of the sample over the course of the aggregation.
Equilibrate the temperature of the plate reader to 37 °C.
Set up the aggregation protocol to read fluorescence measurements at fixed time-point intervals, at an excitation wavelength of 440 nm and an emission wavelength of 480 nm. The aggregation should be initiated in quiescent conditions.
Insert the plate into the plate reader.
NOTE: It is important to be careful while handling the plate, to ensure that the aggregation is not triggered prior to starting the plate reader. Use the gain adjustment settings to ensure the optimal readout of the fluorescence of each sample.
Start the measurement.
NOTE: The aggregation experiment is concluded when the ThT fluorescence sigmoidal curve reaches a plateau⁵⁸.

2. Sample preparation for AFM and nano-IR measurements

Glue highest quality grade V1 mica disc to high quality magnetic stainless-steel disc using adhesive tabs or double-sided tape.
Etch mica by placing a piece of adhesive tape onto the mica surface and pulling off the top layer of the mica. This procedure will produce clean, atomically flat surface, suitable for sample deposition.
NOTE: Etched surface must be uniform and smooth.
CAUTION: Do not touch or breathe directly above freshly etched mica, as this will introduce artefacts.
Pause/stop the plate reader measurement to collect the time point of interest during the aggregation process of the protein.
Remove the sealing foil and collect 10 µL aliquot of the samples without ThT from the well into a 1.5 mL low-bind tubes.
Deposit the 10 µL of the sample on the freshly etched mica. For AFM-IR measurements samples must be deposited on freshly stripped gold substrate.
Incubate the solution on the mica for 1 min to allow physisorbtion.
NOTE: Longer incubation time would allow better absorption on the surface but may induce artificial self-organization and self-assembly²⁵. To avoid these effects, spray deposition may be exploited²⁵.
Rinse three times with 1 mL of ultrapure water.
Dry under a gentle flow of nitrogen to measure the sample in air environment.

3. AFM imaging of the morphology of protein aggregates

NOTE: Morphology measurements can be performed both in contact and dynamic mode, in the following steps the latter is described since it reduces lateral forces to measure the 3D morphology of the sample with high resolution. AFM-IR measurements will be performed in contact mode to enhance AFM-IR signal-to-noise ratio.

Turn on the AFM system at least 30 min before the measurements in order to enable the system to reach thermal stability. Click on Setup, then on Probe. In Probe Change window click on Next to prepare Z, Focus Stage.
Turn off the AFM beam switch (if it is on). Unlock the dovetail locks, disconnect the head from the system and in Probe Change window click on Next.
Mount the AFM cantilever on the probe holder. Connect the head to the system and lock the dovetail locks.
NOTE: Optimal cantilevers to study biological samples in dynamic mode AFM have spring constant ranging between 2−40 N m^-1 and an apex radii ranging between 2−8 nm¹⁴.
Turn on the AFM laser beam switch and in Probe Change window click on Next.
In the pop-up window click on No if the cantilever type did not change. Adjust optical alignment knobs to find the cantilever. Adjust focus onto cantilever and click on Next.
In the pop-up window click on Yes if the focus is on the cantilever.
Position the laser beam at the end of the cantilever using the knobs controlling position of the detection laser. Maximize the total signal measured by the four-quadrant photodiode to at least >1 V using the knobs controlling the position of the deflected laser beam on the position sensitive photodiode (PSPD). In Probe Change window click on Next and then on Close.
Wait 15 min for the cantilever to reach thermal stability. Readjust the position of the deflected laser beam on the PSPD if necessary.
Click on NCM SWEEP, choose desired amplitude of oscillation, click on Use Phase, click on auto and tune the cantilever close to the maximum of its first free resonance of oscillation, which is of approximately 300 kHz for a cantilever with a spring constant of 40 N m^-1.
NOTE: Tuning the cantilever out of the maximum of its free resonance assures higher stability of the measurements¹⁴.
Place the sample on the sample holder. Choose suitable imaging parameters. Typical scanning rate is 0.3−1.0 Hz for a scan area of 1 x 1 µm² to 5 x 5 µm². Typical resolution needed is between 256 x 256 and 1024 x 1024 pixels. Click on Scan Area to choose the scan size and the pixel number.
Focus the optical view on the sample. Click on Approach to approach the sample surface. Once approach is completed click on Lift 100 µm to rise the AFM tip 100 µm above the surface of the sample. Click Expand on the optical image of the sample. Click on the Focus Stage bar to focus the view on the surface of the sample.
Use the arrows to move in the region of the sample of interest. Engage the surface by pressing Approach button. Click on Line Scan button, and check if the tip is following the surface well, if necessary, adjust the Set Point.
Start imaging the sample surface by pressing Scan button. During imaging, to avoid large imaging force and keep consistency of within independent samples, maintain a constant regime of phase change not exceeding Δ20°⁴².
Enter file name and choose base directory where the acquired image will be saved.

4. Infrared nanospectroscopy measurements of protein aggregates

Turn on the AFM-IR system and the infrared (IR) laser⁶⁰ 30−60 min before measurements for thermal stabilization. Typical lasers for AFM-IR instruments are optical parameter oscillators (OPO) and quantum cascade lasers (QCL).
Open the built-in software to control the instrument. Click on file and then on new to open a new nanoIR file. Press the button initialise to start the AFM-IR system.
Open the instrument cover and mount on the AFM-IR system a silicon gold coated probe with a nominal radius of 30 nm and a spring constant of 0.2 N/m to measure the sample in contact mode.
Click on Load in the section AFM probe and then next. In focus on probe section click on the arrows to focus the camera on the cantilever. In the section sample XY movement, use the arrows to place the cross-air at the end of the cantilever.
Rotate the knobs controlling the position of the detection laser to position laser at the end of the cantilever. Rotate the laser knobs to detect and maximize the sum measured by the four-quadrant photodiode to a value >3 V. Rotate the deflection knob to adjust the cantilever deflection to -1 V and then click next. Close the cover of the instrument.
In the section focus on sample use the arrows to focus the camera on the sample. Then, in the section sample XY movement use the arrows to move in the region of interest of the sample and click on approach and engage.
In the microscope window select as inputs the channels of interest to map the biophysical properties of the sample. In particular, choose height for morphology, Amplitude 2 for IR absorption and PLL frequency to map tip-sample contact resonance.
In the AFM scan section of the controls window, use similar parameters as in section 3 (i.e., scan rate 0.1−1.0 Hz, between 256 x 256 and 1024 x 1024 pixels). Choose as value of the gains according to sample roughness: integral I = 1−10 and proportional P = 10−30. Then, click on scan to acquire a morphology map.
NOTE: The morphology can be similarly measured in the dynamic tapping mode, but AFM-IR measurements are performed here in the contact mode to have higher signal to noise ratios.
After the mapping of morphology is finished, in the microscope window, click on the height map to position the probe on the top of one aggregate. Then in the nanoIR section of the controls window, click on start IR to illuminate the sample with the IR laser.
To focus the infrared laser on the cantilever, click on optimisation. In this window, write a wavenumber where the sample is going to have high absorbance and click add. For protein, a typical value is 1655 cm^-1. Click on scan to find the IR laser position and click on update to align its position with the cantilever. Close the window.
In the section general of the nanoIR setting, write a wavenumber where high absorbance in the relative field is expected. Then, deactivate the Band Pass Filter option and look at the meter reading and at the fast Fourier transform (FFT) of the cantilever response. In the FFT window move the green cursor to read the resonance frequency of the cantilever. A typical value of the FFT of the resonance of the cantilevers is around 300 kHz. Write this resonance frequency value in the general section in the freq. centre field and use a freq. window of 50 kHz.
NOTE: Select a laser power that is low enough to not saturate the meter reading and to have distortion in the cantilever response and to not overheat the sample.
Click on laser pulse tune window to use the resonance enhanced mode. Choose a frequency centre of 300 kHz, a tune range of 50 kHz and a duty cycle of the laser of 5%. Click on acquire to sweep the pulse rate of the laser. By using the cursor in the graph, tune the laser pulse to the frequency of the mechanical response of the thermal expansion of the sample absorbing the IR light. Then, select the option PLL to monitor the contact resonance between the sample and the tip. Press the zero button in the PLL window and tick enable to track the sample-tip contact resonance. Choose an integral gain I = 0.5 and proportional gain P = 10. Close the window.
Open again the optimisation window and find the position of the IR laser for at least 3 wavenumbers corresponding to major absorbance bands of the sample (amide band I-II-III) and for at least one wavenumber for each chip of the laser.
Click on Tools | IR Background Calibration | New. In the window select the spectroscopic region of interest (1800−1200 cm^-1 to study protein samples); choose the same pulse rate as determined in step 4.12 and a duty cycle of 5%; click on fast acquisition and select the laser speed, typical range for a quantum cascade laser is between 20−500 cm^-1. Click on acquire to measure the IR laser background. This background spectrum will be used for normalization of measured nanoscale localized spectra. Close the window.
NOTE: If a fast laser and resonance enhanced mode is not available, skip step 4.12 and select stepped spectra instead of fast in both the background and IR spectra acquisition windows. However, single aggregate sensitivity will not be reached.
In the IR spectra settings, choose an IR spectrum resolution between 1−4 cm^-1 and a number of co-averages of at least 64x. Click on acquire to measure a nanoscale localized IR spectrum in the protein range (1800−1200 cm^-1).
To acquire a nanoscale resolved chemical map, select IR imaging option, choose a wavenumber of interest (e.g., 1655 cm^-1 for amide band I) and click on scan in the AFM scan window.
Once the mapping is completed, go to file and save the measurements. To analyze the acquired maps of morphology, contact-resonance and chemistry, as well as the nanoscale localized spectra, use the built-in AFM image processing software. Spectra and Images can be saved as .csv or .axz files for further detailed analysis with commercial software.

5. Image processing and analysis of cross-sectional dimensions

Flatten raw images¹⁴^,¹⁷^,¹⁹^,⁴¹^,⁴²^,⁵⁹ using built-in AFM image processing software or commercial software.
NOTE: The aggregates should be masked from the calculation to avoid artefacts of analysis and underestimation of their measured height.
Flatten the image by 0 order plane fit subtraction.
Flatten the image by plane and then line by line at a 1^st regression order, the second step is repeated until the flat baseline in line profile of the image is reached.
If the sample is very crowded, contains exceptionally high aggregates or scanner bow artefact is present, flatten image using 2^nd regression order fit.
Measure aggregate height and width from a line profile taken perpendicular to the fibril axis¹⁴^,¹⁷^,¹⁹^,⁴¹^,⁴²^,⁵⁹.
Measure fibril length parallel to the fibril axis¹⁴^,¹⁷^,¹⁹^,⁴¹^,⁴²^,⁵⁹.

Representative Results

A representative time course of Aβ42 aggregation, as measured by the ThT fluorescence assay, is shown in Figure 1. The aggregation process is commonly characterized by a sigmoidal curve, where a lag phase is initially observed, and is followed by a steep growth phase, before the curve reaches a plateau when an equilibrium steady state is reached⁶^,⁷^,⁵⁸. It is essential to ensure that an optimized aggregation protocol is used to generate high-quality data to study the molecular details pertaining to aggregation processes⁵⁸.

High resolution of AFM enables to investigate the morphology and heterogeneity of the aggregated species at different time points of the process. During the aggregation, monitored by the ThT assay, aliquots of the sample in the plates are prepared for single aggregate investigation by AFM and nano-IR (Figure 1). The typical process flow of manual sample preparation is shown in Figure 2. At the completion of the measurement of the 3D morphology of the sample by high resolution and phase-controlled AFM, the maps are flattened to remove non-linearity of the piezoelectric scanner and reduce sources of error in post-processing analysis of sample morphology (Figure 3). Subsequently, an accurate and sensitive single molecule statistical analysis can be performed, as shown in Figure 4. From a 3D morphology map, it is possible to extract aggregate cross-sectional height, width and length (or diameter in case of spheroidal particles), which allows to distinguish and characterize distinct species of aggregates present during the aggregation time course²³^,³⁸. Typical time points of interest to investigate the process of aggregation are the lag phase, the growth phase and the plateau phase (Figure 1). During the lag phase, monomeric and oligomeric species of Aβ42 are primarily present. When visualized by AFM, monomeric and oligomeric species of Aβ42 typically appear as spheroidal particles 1−15 nm in diameter and 0.3−2 nm in height (Figure 1, bottom left)³⁸^,³⁹. Formation of elongated protofilaments, protofibrils and fibrils is visible during the growth phase of Aβ42 aggregation time course (Figure 1, bottom middle)³⁸^,³⁹. Typically, protofilaments appear as elongated features hundreds of nanometers in length and 0.5−2 nm in height, while protofibrils appear as elongated linear or curvilinear aggregates 1−5 nm in height and hundreds of nanometers in length³⁸^,³⁹. During the plateau phase, fibrils are the dominant species of Aβ42 aggregates. Aβ42 fibrils typically appear as unbranched, thread-like structures, with a cross-sectional diameter of 6−10 nm and length in the order if micrometers (Figure 3, bottom right)³⁸^,³⁹. Remarkably, this schematic representation of the morphological properties of the aggregates is a general feature of most aggregating proteins and peptides¹³^,¹⁴.

After the investigation of the sample morphology, nano-IR can be used to investigate the chemical properties of the individual protein aggregate species present during the process of aggregation, by acquiring nanoscale-resolved IR maps and spectra both in air and native liquid environment²⁴^,²⁶^,³⁸^,⁴²⁴⁹^,⁵⁰^,⁵¹^,⁵²^,. Figure 5 shows a schematic illustration of the AFM-IR setup. An IR source is used to illuminate the sample from the bottom in total internal reflection or directly from the top as in Figure 5a. If the IR light is absorbed by the sample, it will excite the corresponding molecular vibrational energy transition levels of the species present. The vibrational energy is dissipated inside the sample in the form of thermal heating, which causes the thermal expansion of the sample. This expansion is measured at the nanoscale by the AFM tip in contact with the sample with a resolution in the order of 10 nm. At each pulse, the cantilever detects the thermal expansion. In particular, the fast expansion of the sample kicks out the cantilever from contact with the sample, and after the kick out the cantilever rings down at its natural frequencies. In order to enhance the sensitivity of AFM-IR, it is possible to tune the laser pulse frequency at the same frequency of the oscillation of the cantilever⁵⁴. In order to work in this resonance-enhanced mode, it is necessary to have IR sources that can be pulsed in a wide range of frequencies, such as quantum cascade lasers that operate typically in a range between 1−1000 kHz. The peak-to-peak amplitude and the fast Fourier transform of the ringdown signal, termed IR amplitude, of the raw cantilever deflection are proportional to the IR light absorbed. These signals are detected in real time by measuring the deflection of a red laser focused on the top of the cantilever. In order to acquire chemical maps, the laser wavenumber is fixed at a certain wavenumber and the IR amplitude signal is collected at each point of the map, while, to acquire IR spectra, the position of the cantilever is fixed in a position of interest and the laser wavelength is swept along the spectroscopic range of interest. The ultimate resolution of AFM-IR enables the measurement of the chemical properties of protein aggregates with a cross-sectional height of approximately 5 nm, as represented in Figure 6.

Figure 1: Monitoring of the aggregation time course in vitro via ThT fluorescence and AFM.
Samples taken during the lag phase, growth phase, and plateau phase of the aggregation process as detected by ThT fluorescence were imaged via high-resolution AFM. This figure has been adapted from Ruggeri et al.³⁸. Please click here to view a larger version of this figure.

Figure 2: Schematic representation of the substrate preparation and sample deposition for AFM measurements.
(A, B) Mica etching using adhesive tape. (C) Sample deposition. (D) Sample incubation. (E) Sample rinsing with ultrapure water. (F) Sample drying under gentle flow of nitrogen. (G) Sample imaging using microfabricated AFM cantilever with sharp tip on its end. (H) Processed image of amyloid fibrils. Please click here to view a larger version of this figure.

Figure 3: AFM image processing procedure⁴².
The top of each panel shows a profile line of the sample surface (red line) illustrated by the corresponding AFM image, while the lower part shows a histogram of the height of all pixels in the image. (a) Raw AFM image before image flattening procedure. (b) AFM image after the processing procedure using the whole plane flattening. Fibrillar structures (pink color) were masked from the flattening procedure. (c) Image processed using line-by-line flattening procedure. (d) Final image after the image processing procedure. This figure has been adapted from Ruggeri et al.⁴².

Figure 4: Single aggregate statistical analysis of AFM images.
(a) Example of the tracing of the heights and lengths of fibrillar structures, indicated with 1 and 2. (b) Graph with sections of the traced fibrils and their average height. (c) Example of a histogram showing the average height of fibrillar structures. (d) Graph with normalized profile of the traced fibrils 1 and 2. (e) Histogram distribution of the normalized profile height points. This figure has been adapted from Ruggeri et al.⁴². Please click here to view a larger version of this figure.

Figure 5: Principle of function of the AFM-IR method.
(a) Absorbed IR light causes the thermal expansion of the sample, exciting the mechanical resonances of the AFM cantilever in contact with the sample. The amplitude of the cantilever oscillations is proportional to the IR absorption. (b) IR absorption maps are obtained scanning the cantilever on the sample while fixing the laser wavelength. (c) Tip and sample in contact behave as a system of coupled springs whose resonant frequency increases monotonically with the intrinsic stiffness of the sample⁵⁰. (d) Localized spectra are obtained by sweeping the laser wavelength while fixing the position of the AFM cantilever. (e) IR spectrum of protein. (f) In summary, AFM-IR enables the simultaneous study of morphological, mechanical and chemical properties at the nanoscale²⁶. Please click here to view a larger version of this figure.

Figure 6: Infrared nanospectroscopy (nano-IR) of a single amyloid aggregate.
(a) AFM morphology map. (b) IR absorption map at the laser resonance peak at 1658 cm^-1. (c) Cross sectional dimensions of the fibril height. (d) IR spectra on different positions of the fibrillar structure (marked in panel b with blue circles) and the substrate (marked in panel b with green circles). The average net signal deriving from the aggregate structure (solid black line) was obtained by subtracting the averaged background signal (solid green line) from the averaged fibril signal (solid blue line). This figure has been adapted from Ruggeri et al.⁴².

Discussion

The first critical step in this protocol is the preparation of monomeric proteins, such as in the case of Aβ42 solution described in steps 1.1 and 1.2. It is essential to initiate the aggregation process from a highly pure, monomeric solution, as the presence of oligomeric or aggregated species may result in poor reproducibility of the aggregation kinetics⁵⁸, and induce artefacts in the AFM measurements (e.g., fibrillar species will be evident at the initial stages of the aggregation), which may lead to the misinterpretation of the data. Highly reproducible kinetics data of amyloid formation based on ThT fluorescence assay, in association with the master equation formalism of chemical kinetics⁵^,⁶^,⁷, have allowed to define the Aβ42 aggregation mechanism in terms of its underlying molecular events. Chemical kinetics connects the microscopic steps underlying amyloid formation with their macroscopic manifestations by considering the different ways in which new aggregates can form and grow, which are for instance elongation at the aggregate ends or secondary nucleation on the aggregate surface. However, chemical kinetics by itself does not directly enable the visualization of the possible nucleation phenomena at the nanoscale requiring their combination with single molecule methods.

The second critical step in this protocol is the substrate preparation and sample deposition procedures described in steps 2.2 and 2.6−2.8. To avoid artefacts, the sample must be deposited on a clean, atomically flat surface. Proper etching of the mica is essential to achieve artefact-free, high-resolution in AFM measurements. The sample deposition time is also extremely important, as longer incubation time allows better absorption on the substrate surface. However, it might also induce artificial self-organization and self-assembly²⁵, which may induce artefacts (e.g., surface-induced aggregate species) that may lead to data misinterpretation. In addition, the mica surface is negatively charged, which means that only positively charged molecules easily absorb on it. If the net charge of the sample is negative, the surface of the mica can be positively functionalized using APTES for a better phisisorption²³. Microfluidic spray deposition²⁵ can be exploited to avoid these effects and artefacts and deposit the sample in a single step and artefact-free manner.

The third critical step is the proper setup and the choice of imaging parameters for sample imaging via AFM and AFM-IR described in section 3. The AFM tips used for the sample imaging should be sharp enough (apex radii of 2−8 nm) to achieve high-resolution and minimize convolution effects (broadening of the sample features by the tip)¹⁴, which may induce uncertainties in the image of the sample. The choice of imaging mode, contact or dynamic, is also important. For sample imaging via conventional AFM, the dynamic mode is preferred over the contact mode as the latter mode induces large lateral tip-sample frictional forces that may cause sample damage and introduce artefacts in the measurements (e.g., reduction in the sample height due to nanoindentation)¹⁴^,⁶¹^,⁶². Conversely, the contact mode is preferred for the measurements via nano-IR to enhance AFM-IR signal. For measurements in dynamic mode, the cantilever should be tuned just slightly below (tapping mode) or above (non-contact mode) the maximum of its first free resonance of oscillation to assure higher stability of the measurements¹⁴. The imaging resolution, which depends on the pixel number and scan area, should be high enough to capture the smallest degree of detail present in a specimen (e.g., 1024 x 1024 pixels for a 4 x 4 µm² area)¹⁴^,⁶³. Low imaging resolution can induce distortions and uncertainties in the image of the sample due to the loss of the vertical and lateral information upon digitisation of the signal¹⁴. The scan rate, used for imaging, should be low enough for the tip to be able to follow surface features properly as well as to have enough time to acquire chemical information¹⁴. One of the most important imaging parameters is the imaging force. In contact mode, it is fundamental to use a low interaction force to preserve structure of the sample. In dynamic mode, the energy dissipation on the samples should be kept constant in order to consistently compare morphology of distinct samples; consistent imaging of independent samples can be obtained by maintaining a constant regime of a phase change not exceeding Δ20°¹⁴^,⁴². Large imaging forces should be avoided as they may induce distortions and uncertainties in the sample images.

Thermal drift caused by the expansion and contraction of AFM parts due to thermal fluctuations can induce distortions and artefacts in the image of the sample⁶⁴^,⁶⁵. Drifts in the vertical direction may cause the cantilever to lose track of the surface as well as to crash into the surface, while drifts in the lateral direction usually result in elongation of the surface features and image distortion, which makes it difficult to achieve precise measurements of the sample features. The effect of these thermal drifts can be minimized by accurate temperature control of the laboratory as well as giving enough time (around 30 min) for the system to become stable or by performing fast scans⁶⁶^,⁶⁷^,⁶⁸^,⁶⁹^,⁷⁰.

The quality of the nano-IR imaging and spectra collection by AFM-IR can be affected by several factors, of which the most important are: (i) a wrong division of the IR signal on the sample by the IR background, (ii) a large variation of nanomechanical contact between tip and the sample, and (iii) an excessive heating of the sample causing its softening. To correctly divide the IR spectrum of the sample by the collected background, it is crucial that they are collected at the same laser power. Indeed, the spectral line shape of the IR laser background depends on the power of the laser. Then, in order to avoid the influence of the mechanical properties of the sample into the measured chemical information, it is crucial to monitor and track the contact resonance between the sample and the tip during spectra and image acquisition. For spectra acquisition, ideally, it is sufficient to pulse the laser at a fixed contact resonance. However, if the spectrum is acquired on a large spectroscopic range, high intensity peaks could cause strong heating of the sample and its softening, thus changing the tip-sample contact resonance and artificially reducing the IR peak amplitude. For this reason, it is important to track contact resonance during spectra acquisition in order to verify that the spectrum is not affected by excessive heating of the sample.

In conclusion, conventional AFM and nano-IR are capable of investigating with high resolution the morphological, structural and chemical properties of the individual species forming during protein aggregation²⁴^,³⁸. However, they lack the capability of chemical kinetics to follow their rapid kinetics of formation in native bulk conditions. In order to unravel the conformational changes that protein undergo during their aggregation and misfolding, it is necessary to develop and apply novel biophysical methods capable of bringing together the capabilities of bulk biophysical methods with the investigation of the heterogeneity and ultrastructural properties of protein aggregation at the nanoscale. This approach represents a fruitful avenue to address the challenge of understanding the problem of protein self-assembly and its role in health and disease. Indeed, single aggregate approaches are capable to unravel and elucidate the molecular mechanisms of protein aggregation polymorphism and formation. This information is central to address the challenge of understanding protein aggregation and its role in the onset and progression of human diseases, as well as understanding their biophysical properties for biotechnology applications.

Disclosures

The authors have nothing to disclose.

Acknowledgements

The authors thank Swiss National Foundation for Science (SNF) for the financial support (grant number P2ELP2_162116 and P300P2_171219), the Darwin College, Erasmus+ program for the financial support (grant number 2018-1-LT01-KA103-046719-15400-P3) and the research leading to these results has received funding from the European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013) through the ERC grant PhysProt (agreement number 337969), the Newman Foundation (T.P.J.K.) and The Cambridge Centre for Misfolding Diseases (C.G., M.V., and T.P.J.K.).

Materials

AFM-IR system	Anasys Instruments	nanoIR 2 or 3	Systems to measure thermal expansion in contact and resonance mode
Corning 96-well Half Area Black/Clear Bottom Polystyrene NBS Microplate	Corning	3881
Corning Microplate Aluminium Sealing Tape	Corning	6570
Double Sided Adhesive Discs	AGAR Scientific	AGG3347N
FLUOstar Omega	BMG Labtech	415-101	Platereader
Mica Disc 10mm V1	AGAR Scientific	AGF7013
Park NX10 AFM system	Park Systems	N/A	Atomic Force Microscope
Platypus Ultra-Flat Gold Chips	Platypus Technologies	AU.1000.SWTSG
PPP-NCHR-10 cantilevers	Park Systems	PPP-NCHR-10
Protein LowBind Tubes, 2.0mL	Eppendorf	30108132
Silicon gold coated cantilevers	Anasys Instruments	PR-EX-nIR2
SPM Specimen Discs 12mm	AGAR Scientific	AGF7001

References

Selkoe, D. J., Hardy, J. The amyloid hypothesis of Alzheimer’s disease at 25 years. EMBO Molecular Medicine. 8, 595-608 (2016).
Poewe, W., et al. Parkinson disease. Nature Reviews Disease Primers. 3, 1-21 (2017).
Chiti, F., Dobson, C. M. Protein misfolding, functional amyloid, and human disease. Annual Review of Biochemistry. 75, 333-366 (2006).
Chiti, F., Dobson, C. M. Protein misfolding, amyloid formation, and human disease: A summary of progress over the last decade. Annual Review of Biochemistry. 86, 27-68 (2017).
Knowles, T. P. J., Vendruscolo, M., Dobson, C. M. The amyloid state and its association with protein misfolding diseases. Nature Reviews Molecular Cell Biology. 15, 384-396 (2014).
Meisl, G., et al. Molecular mechanisms of protein aggregation from global fitting of kinetic models. Nature Protocols. 11, 252-272 (2016).
Knowles, T. P. J., et al. An analytical solution to the kinetics of breakable filament assembly. Science. 326, 1533-1537 (2009).
Barth, A. Infrared spectroscopy of proteins. Biochimica et Biophysica Acta (BBA) – Bioenergetics. 1767, 1073-1101 (2007).
Fitzpatrick, A. W. P., et al. Atomic structure and hierarchical assembly of a cross- amyloid fibril. Proceedings of the National Academy of Sciences of the United States of America. 110, 5468-5473 (2013).
Cohen, S. I. A., et al. Proliferation of amyloid-β42 aggregates occurs through a secondary nucleation mechanism. Proceedings of the National Academy of Sciences of the United States of America. 110, 9758-9763 (2013).
Michaels, T. C. T., Lazell, H. W., Arosio, P., Knowles, T. P. J. Dynamics of protein aggregation and oligomer formation governed by secondary nucleation. Journal of Chemical Physics. 143, (2015).
Šarić, A., Michaels, T. C. T., Zaccone, A., Knowles, T. P. J., Frenkel, D. Kinetics of spontaneous filament nucleation via oligomers: Insights from theory and simulation. The Journal of Chemical Physics. 145, 211926 (2016).
Ruggeri, F. S., Habchi, J., Cerreta, A., Dietler, G. AFM-based single molecule techniques: Unraveling the amyloid pathogenic species. Current Pharmaceutical Design. 22, 3950-3970 (2016).
Ruggeri, F. S., Šneideris, T., Vendruscolo, M., Knowles, T. P. J. Atomic force microscopy for single molecule characterisation of protein aggregation. Archives of Biochemistry and Biophysics. 664, 134-138 (2019).
Iljina, M., et al. Nanobodies raised against monomeric α-synuclein inhibit fibril formation and destabilize toxic oligomeric species. BMC Biology. 15, 1-14 (2017).
Schilling, C., et al. Sequence-Optimized Peptide Nanofibers as Growth Stimulators for Regeneration of Peripheral Neurons. Advanced Functional Materials. 1809112, 1-15 (2019).
Chiki, A., et al. Mutant exon1 huntingtin aggregation is regulated by T3 phosphorylation-induced structural changes and crosstalk between T3 phosphorylation and acetylation at K6. Angewandte Chemie – International Edition. 56, 5202-5207 (2017).
Sieste, S., et al. Water-dispersible polydopamine-coated nanofibers for stimulation of neuronal growth and adhesion. Advanced Healthcare Materials. 7, 1-11 (2018).
De, S., et al. Different soluble aggregates of Aβ42 can give rise to cellular toxicity through different mechanisms. Nature Communications. 10, 1541 (2019).
Chang, K. C., Chiang, Y. W., Yang, C. H., Liou, J. W. Atomic force microscopy in biology and biomedicine. Tzu Chi Medical Journal. 24, 162-169 (2012).
Variola, F. Atomic force microscopy in biomaterials surface science. Physical Chemistry Chemical Physics. 17, 2950-2959 (2015).
Drolle, E., Hane, F., Lee, B., Leonenko, Z. Atomic force microscopy to study molecular mechanisms of amyloid fibril formation and toxicity in Alzheimer’s disease. Drug Metabolism Reviews. 46, 207-223 (2014).
Ruggeri, F. S., et al. Identification and nanomechanical characterization of the fundamental single-strand protofilaments of amyloid α-synuclein fibrils. Proceedings of the National Academy of Sciences of the United States of America. 115 (28), 7230-7235 (2018).
Ruggeri, F. S., et al. Infrared nanospectroscopy characterization of oligomeric and fibrillar aggregates during amyloid formation. Nature Communications. 6, 1-9 (2015).
Ruggeri, F. S., et al. Microfluidic deposition for resolving single-molecule protein architecture and heterogeneity. Nature Communications. 9, (2018).
Qamar, S., et al. FUS phase peparation is modulated by a molecular chaperone and methylation of arginine cation-π interactions. Cell. 173, 720-734 (2018).
Habchi, J., et al. Cholesterol catalyses Aβ42 aggregation through a heterogeneous nucleation pathway in the presence of lipid membranes. Nature Chemistry. 10, 673-683 (2018).
Sweers, K. K. M., Stöckl, M., Bennink, M. L., Subramaniam, V., Uversky, V. N., Lyubchenko, Y. L. Characterizing nanoscale morphologic and mechanical properties of α-Synuclein amyloid fibrils with atomic force microscopy. Bio-nanoimaging: Protein Misfolding & Aggregation. , 309-322 (2014).
Goldsbury, C., et al. Amyloid structure and assembly Insights from scanning transmission electron microscopy. Journal of Structural Biology. 173, 1-13 (2011).
Knowles, T. P. J., Smith, J. F., Devlin, G. L., Dobson, C. M., Welland, M. E. Analysis of structural order in amyloid fibrils. Nanotechnology. 18, (2007).
Knowles, T. P. J., Mezzenga, R. Amyloid fibrils as building blocks for natural and artificial functional materials. Advanced Materials. , 6546-6561 (2016).
Knowles, T. P. J., Oppenheim, T. W., Buell, A. K., Chirgadze, D. Y., Welland, M. E. Nanostructured films from hierarchical self-assembly of amyloidogenic proteins. Nature Nanotechnology. 5, 204-207 (2010).
Knowles, T. P. J., Smith, J. F., Craig, A., Dobson, C. M., Welland, M. E. Spatial persistence of angular correlations in amyloid fibrils. Physical Review Letters. 96, 1-4 (2006).
Knowles, T. P. J., et al. Twisting transition between crystalline and fibrillar phases of aggregated peptides. Physical Review Letters. 109, 158101 (2012).
Knowles, T. P., et al. Role of intermolecular forces in defining material properties of protein nanofibrils. Science. 318, 1900-1903 (2007).
Smith, J. F., Knowles, T. P. J., Dobson, C. M., MacPhee, C. E., Welland, M. E. Characterization of the nanoscale properties of individual amyloid fibrils. Proceedings of the National Academy of Sciences of the United States of America. 103, 15806-15811 (2006).
Sokolov, D. V. Atomic force microscopy for protein nanotechnology. Methods in Molecular Biology (Clifton, N.J.). 300, 323-367 (2013).
Ruggeri, F. S., et al. Influence of the β-sheet content on the mechanical properties of aggregates during amyloid fibrillization. Angewandte Chemie – International Edition. 54, 2462-2466 (2015).
Jeong, J. S., Ansaloni, A., Mezzenga, R., Lashuel, H. A., Dietler, G. Novel mechanistic insight into the molecular basis of amyloid polymorphism and secondary nucleation during amyloid formation. Journal of Molecular Biology. 425, 1765-1781 (2013).
Adamcik, J., Mezzenga, R. Study of amyloid fibrils via atomic force microscopy. Current Opinion in Colloid and Interface Science. 17, 369-376 (2012).
Deguire, S. M., et al. N-terminal huntingtin (Htt) phosphorylation is a molecular switch regulating Htt aggregation, helical conformation, internalization, and nuclear targeting. Journal of Biological Chemistry. , (2018).
Ruggeri, F. S., et al. Nanoscale studies link amyloid maturity with polyglutamine diseases onset. Scientific Reports. 6, 1-11 (2016).
Adamcik, J., et al. Understanding amyloid aggregation by statistical analysis of atomic force microscopy images. Nature Nanotechnology. 5, 423-428 (2010).
Lin, Y. C., Komatsu, H., Ma, J., Axelsen, P. H., Fakhraai, Z. Quantitative analysis of amyloid polymorphism using height histograms to correct for tip convolution effects in atomic force microscopy imaging. RSC Advances. 6, 114286-114295 (2016).
Mannini, B., et al. Stabilization and characterization of cytotoxic Aβ40 oligomers isolated from an aggregation reaction in the presence of zinc ions. ACS Chemical Neuroscience. , (2018).
Medalsy, I., Hensen, U., Muller, D. J. Imaging and quantifying chemical and physical properties of native proteins at molecular resolution by force-volume AFM. Angewandte Chemie – International Edition. 50, 12103-12108 (2011).
Dufrêne, Y. F., Martínez-Martín, D., Medalsy, I., Alsteens, D., Müller, D. J. Multiparametric imaging of biological systems by force-distance curve–based AFM. Nature Methods. 10, 847-854 (2013).
Takai, E., et al. Scanning electron microscope imaging of amyloid fibrils. American Journal of Biochemistry and Biotechnology. 10, 31-39 (2014).
Dazzi, A., et al. AFM-IR: Combining atomic force microscopy and infrared spectroscopy for nanoscale chemical characterization. Applied Spectroscopy. 66, 1365-1384 (2012).
Volpatti, L. R., et al. Micro- and nanoscale hierarchical structure of core-shell protein microgels. Journal of Materials Chemistry B. 4, 7989-7999 (2016).
Galante, D., et al. A critical concentration of N-terminal pyroglutamylated amyloid beta drives the misfolding of Ab1-42 into more toxic aggregates. The International Journal of Biochemistry & Cell Biology. 79, 261-270 (2016).
Ruggeri, F. S., et al. Identification of oxidative stress in red blood cells with nanoscale chemical resolution by infrared nanospectroscopy. International Journal of Molecular Sciences. 19, 1-14 (2018).
Ramer, G., Ruggeri, F. S., Levin, A., Knowles, T. P. J., Centrone, A. Determination of polypeptide conformation with nanoscale resolution in water. ACS Nano. 12 (7), 6612-6619 (2018).
Dazzi, A., Prater, C. B. AFM-IR: Technology and applications in nanoscale infrared spectroscopy and chemical imaging. Chemical Reviews. 117, 5146-5173 (2017).
Lahiri, B., Holland, G., Centrone, A. Chemical imaging beyond the diffraction limit: experimental validation of the PTIR technique. Small. 9 (3), 439-445 (2013).
Ansaloni, A., et al. One-pot semisynthesis of exon 1 of the huntingtin protein: New tools for elucidating the role of posttranslational modifications in the pathogenesis of Huntington’s disease. Angewandte Chemie – International Edition. 53 (7), 1928-1933 (2014).
Khalaf, O., et al. The H50Q mutation enhances α-synuclein aggregation, secretion, and toxicity. Journal of Biological Chemistry. 289, 21856-21876 (2014).
Hellstrand, E., Boland, B., Walsh, D. M., Linse, S. Amyloid β-protein aggregation produces highly reproducible kinetic data and occurs by a two-phase process. ACS Chemical Neuroscience. 1, 13-18 (2010).
Limbocker, R., et al. Trodusquemine enhances Aβ42 aggregation but suppresses its toxicity by displacing oligomers from cell membranes. Nature Communications. 10 (1), 225 (2019).
Lu, F., Belkin, M. A. Infrared absorption nano-spectroscopy using sample photoexpansion induced by tunable quantum cascade lasers. Optics Express. 19, 1902-1904 (2011).
Jiao, Y., Schäffer, T. E. Accurate height and volume measurements on soft samples with the atomic force microscope. Langmuir. 20, 10038-10045 (2004).
Müller, D. J., Engel, A. The height of biomolecules measured with the atomic force microscope depends on electrostatic interactions. Biophysical Journal. 73, 1633-1644 (1997).
Heymann, J. B., Moller, C., Muller, D. J. Sampling effects influence heights measured with atomic force microscopy. Journal Of Microscopy-Oxford. 207, 43-51 (2002).
Marinello, F., Balcon, M., Schiavuta, P., Carmignato, S., Savio, E. Thermal drift study on different commercial scanning probe microscopes during the initial warming-up phase. Measurement Science and Technology. 22, 9 (2011).
Marinello, F., Bariani, P., De Chiffre, L., Savio, E. Fast technique for AFM vertical drift compensation. Measurement Science and Technology. 18, 689-696 (2007).
Ricci, D., Braga, P. C. Recognizing and avoiding artifacts in AFM imaging (Clifton, N.J.). Methods in Molecular Biology. 242, 25-27 (2004).
Canale, C., Torre, B., Ricci, D., Braga, P. C., Braga, P. C., Ricci, D. Recognizing and avoiding artifacts in atomic force microscopy imaging. Atomic Force Microscopy in Biomedical Research. 736, 31-43 (2011).
Marinello, F., Carmignato, S., Voltan, A., Savio, E., De Chiffre, L. Error Sources in Atomic Force Microscopy for Dimensional Measurements: Taxonomy and Modeling. Journal of Manufacturing Science and Engineering. 132, 030903 (2010).
Ukraintsev, E., Kromka, A., Kozak, H., Reme, Z., Rezek, B., Frewin, C. L. Artifacts in Atomic Force Microscopy of Biological Samples. Atomic Force Microscopy Investigations into Biology – From Cell to Protein. , (2012).
Tsukruk, V. V., Singamaneni, S. . Scanning Probe Microscopy of Soft Matter: Fundamentals and Practices. , (2011).

Play Video

PDF

DOI

DOWNLOAD MATERIALS LIST

Cite This Article

Ruggeri, F. S., Šneideris, T., Chia, S., Vendruscolo, M., Knowles, T. P. J. Characterizing Individual Protein Aggregates by Infrared Nanospectroscopy and Atomic Force Microscopy. J. Vis. Exp. (151), e60108, doi:10.3791/60108 (2019).