Atomic force microscopy

Atomic force microscopy (AFM) is a raster scanning technique allowing to visualize the topography of a surface with subnanometer resolution. The principle and application were introduced by Binning, Quate and Gerber in the mid eighties of the last century.⁹⁴ It is based on the direct interaction between the scanning tip and the atoms of the scanned surface. AFM allows to map the topography of the samples in 3D and can, in contrast to other high resolution scanning techniques like electron microscopy, also be applied in water and is not limited to electrically conductive samples. Compared to recently developed high resolution fluorescence techniques like STED or localization microscopy it has a superior resolution in both axial and lateral dimension of up to 0.1 and 1 nm,^{95, 96} but requires direct interaction of the scanning tip to the surface of interest. These properties make it an excellent tool to study biological samples under native conditions.

Setup and working principle

The working principle of an atomic force microscope is shown in Figure 3.5.

A1

A2

B1

B2

IRlaser adjustable

mirror

cantilever photodiode

x,y,z piezo actuators controller

computer feedback loop

0 nm 3 nm

ld lo

STxB

Figure 3.5: Setup of an AFM experiment. The cantilever is moved on the surface and reflects a light beam onto the photodiode. The position of the signal on the diode controls the movement of the piezo actuators by a feedback loop. Object sizes and the angles of the light beams are chosen arbitrary for clarity.

The scanning probe is a cantilever bearing a sharp tip on the bottom side. Using a z-piezo actuator the cantilever can be moved towards the surface until the tip interacts with the sample resulting in a bending of the lever. This bending can be detected using a light beam, usually emitted by a laser or super radiant diode, which is reflected from the top side of the cantilever. The beam gets directed onto a detector built of four photodiodes by an adjustable mirror. The long path of the light transfers small displacements of the cantilever into a larger measurable changes on the diodes. Bending of the cantilever upon contact to the surface changes to position of the reflected beam on the diode array. Using this principle, both the vertical and lateral deflection of the scanning tip can be calculated. The resulting signal is coupled into a feedback loop controlling the movement of the z-piezo actuator. It is amplified by proportional-integral-differential controller.

This element compares a given value for the cantilever deflection (setpoint) with the actual measured value and creates a signal to compensate for a possible difference.

The proportional gain creates a fast feedback proportional to the measured difference which can result in an overcompensation. The integral gain measures the temporal changes resulting in a precise but slow amplifier signal. The differential gain creates a spurious signal if a change in the input is detected.

These feedback loops allow to control the z-position and deflection of the cantilever.

To obtain a map of the surface, additionally x-y piezo actuators are either integrated into the sample holder or the cantilever stage. They allow to position the scanning tip laterally with nanometer precision. Scanning the surface in x-y direction, while controlling the z-position of the cantilever, creates a 3D image of the sample. Images are usually scanned line by line. One dimension is scanned fast (fast scan axis) while the other is scanned subsequently (slow scan axis).

Modes and scanning

The AFM setup can be applied to investigate a variety of properties of surfaces.

Using force spectroscopy, mechanical properties of the sample can be studied. A commonly used mode is the AC or tapping-mode. The cantilever is excited to oscillate in z-direction while scanning allows to minimize the interaction of the tip with the sample. In this thesis the contact mode was used. The cantilever is brought into contact with the sample, applying a defined force onto the surface.

Two different modes are available using the contact mode. The cantilever can be moved at a constant height. The resulting image shows the forces acting on the lever as a function of the lateral coordinates. This mode allows fast scanning but has several disadvantages. A high force is applied which might deform a soft substrate and steep, high ascends might damage the scanning tip. To avoid these obstacles, the slower, more regulated constant force mode can be used. The applied force is set to the setpoint value and the feedback loop controls the z position of the cantilever to keep a constant force, changing the height of the probe. The sample shown in Figure 3.5 is scanned in constant force contact mode. If the applied force is low compared to the samples hardness the topography of the sample can be accurately scanned. However, if the feedback loop is too slow the image can still be distorted by the cantilever moving along the surface.

Imaging phase-separated lipid membranes allows to identify the l_o phase by its elevated height of ∆h ≈ 0.5-1.0 nm.⁹⁷ It is crucial to apply a low force to avoid

mechanical deformation of the l_o phase. Figure 3.6 shows a schematic trajectory of a cantilever scanning a phase-separated membrane with bound STxB.

cantilever

Sphingomyelin Cholesterol DOPC liquid-ordered (lo)

low STxB density

liquid-ordered (lo) high STxB density

STxB

Gb3

Figure 3.6:Schematic drawing of the cantilever substrate interaction in AFM. The lateral resolution of the AFM does not resolve individual proteins but measures an averaged height over the .

The black line shows the measured topography compared to the real sample. The x,y resolution of the AFM used for the measurements used in this thesis was set to 59×59 nm. At the border of the ld and lo phase no sharp increase in height is found but a smeared transition between the two phases due to the cantilever shape and scanning speed. This results in a broadening of the height distributions used for analysis (vide infra) but does not impair the analysis of the height differences because the imaged phases are in the size range of several micrometers. However, the pixel size is large compared to the crystallographic dimensions of STxB.⁶⁰ A single pentamer cannot be resolved. The height measured corresponds to the product of the protein height of approx. 2 nm and its surface density. Densely packed STxB results in a height close to 2 nm while areas with less protein have a lower height.

Experimental procedure

AFM topography maps were recorded using a JPK NanoWizard I. Membranes prepared on mica in custom made PTFE chambers were imaged in PBS at 20 °C using cantilevers with a nominal spring constant of k= 0.03 N m⁻¹ (Ultrasharp CSC38/no al). The cantilever was mounted and immersed into the solution to equilibrate. Images were recorded at a scan speed of 30 µm s⁻¹ (fast scan axis) with a resolution of 512 x 512 pixels in contact mode using constant force. The setpoint was chosen manually and adjusted during measurements. One pixel corresponds to an

area of 59×59 nm. Using higher lateral resolution was found to deform the samples, resulting in lower apparent heights of the lipid phases and the bound protein.

Images were analyzed using Gwyddion 2.26 or 2.34.⁹⁸Binned height histograms were created from manually chosen regions in the images. The histogram was fitted in matlab using multiple Gaussian distributions. The height of the lowest phase (l_d) was set to 0. Values for individual histograms are given as (h₂−h₁)±(σ₂+σ₁) (note that the matlab2014 fitting toolbox parameter c for Gaussian distributions isσ·√

2).

Pooled values for the height differences are given as mean±standard deviation.

Images are visualized using the ’gold’ color coding included in Gwyddion.