High-Speed Atomic Force Microscopy Imaging of DNA Three-Point-Star Motif Self Assembly Using Photothermal Off-Resonance Tapping

Summary

Here, we show the imaging protocol for observing biomolecular interactions with photothermal off-resonance tapping (PORT), where we optimized imaging parameters, identified system limits, and investigated potential improvements in imaging three-point-star DNA motif assembly.

Abstract

High-speed atomic force microscopy (HS-AFM) is a popular molecular imaging technique for visualizing single-molecule biological processes in real-time due to its ability to image under physiological conditions in liquid environments. The photothermal off-resonance tapping (PORT) mode uses a drive laser to oscillate the cantilever in a controlled manner. This direct cantilever actuation is effective in the MHz range. Combined with operating the feedback loop on the time domain force curve rather than the resonant amplitude, PORT enables high-speed imaging at up to ten frames per second with direct control over tip-sample forces. PORT has been shown to enable imaging of delicate assembly dynamics and precise monitoring of patterns formed by biomolecules. Thus far, the technique has been used for a variety of dynamic in vitro studies, including the DNA 3-point-star motif assembly patterns shown in this work. Through a series of experiments, this protocol systematically identifies the optimal imaging parameter settings and ultimate limits of the HS-PORT AFM imaging system and how they affect biomolecular assembly processes. Additionally, it investigates potential undesired thermal effects induced by the drive laser on the sample and surrounding liquid, particularly when the scanning is limited to small areas. These findings provide valuable insights that will drive the advancement of PORT mode's application in studying complex biological systems.

Introduction

High-speed atomic force microscopy (HS-AFM) is a rapidly growing imaging technique^1,2,3,4. It operates at speeds that allow researchers to visualize biomolecular interactions in real time^5,6,7,8,9. Photothermal off-resonance tapping (PORT) is an off-resonance imaging mode similar to peak force tapping^10,11, pulsed force mode^12,13, or jumping mode¹⁴. However, rather than vertically oscillating the scanner, PORT vertically oscillates only the cantilever through an excitation laser focused on the cantilever (usually close to the clamping point). The cantilever deforms due to the bimorph effect: a power-modulated excitation laser periodically heats the coated cantilever, which bends due to the different thermal expansion coefficients of the cantilever and the coating materials¹⁵. Cantilever and sample heating can be minimized by using a drive laser that is periodically switched off and back on during each oscillation cycle, rather than using a full sinusoidal drive⁵.

DNA has been used to form biologically relevant, structurally interesting, and biochemically useful motifs for a number of years^{16,17,18,19,20}. In addition, DNA structures have been proven ideally suited to characterize AFM imaging quality²¹ and to assess the tip-effect of high speed AFM²². Blunt-end DNA three-point-stars (3PS) became practical as a programmable model system for investigating the supramolecular organization of similarly structured molecules in otherwise complex biological systems¹⁹. Previously, the self-assembly of lattices formed by blunt-ended trimeric DNA monomers was tracked via HS-AFM²³. Eventually, these organize into large networks with hexagonal order. Here, the self-assembly of DNA 3-point stars¹⁹ is imaged with the PORT technique at scanning speeds fast enough to track the self-assembly and its correction mechanisms²⁴ while assuring minimal disruption of the process or sample damage. As with any HS-AFM mode, there is a trade-off between achievable imaging quality, imaging speed, and the unwanted disturbance of the sample. By choosing the right compromise, one can better understand the self-organization patterns of supramolecular assemblies. This protocol will, therefore, use a similar setup with DNA 3PS as a model system to optimize the parameters specific to PORT. This will allow operation at fast imaging speeds at large enough scan sizes while minimizing sample damage.

Protocol

1. Sample and buffers

NOTE: The DNA tile used in this study is the 3-point-star motif developed at the Mao laboratory at Purdue University^19,25. All oligonucleotides used in this study were purchased from Integrated DNA Technologies, Inc. Gather the necessary materials and reagents.

1. Mix the single-stranded DNAs (ssDNAs) at a 1:3:3 molar ratio (S1 0.6 µM, 1.8 µM for S2, and 1.8 µM for S3) in the annealing buffer (5 mM TRIS, 1 mM EDTA, and 10 mM MgAc₂). The final concentration of the DNA motif must be 0.6 µM (49 ng/µL with the molecular weight of 3PS being 82020.3 g/mol).
2. Place the DNA solution in a heat-resistant container and heat it to 80 °C. Slowly cool the DNA solution from 80 °C to 20 °C over a period of 4 h. This annealing process helps the ssDNA oligonucleotides to form the desired double-stranded DNA motif.
3. For purification, load the annealed DNA solution onto a 3% agarose gel to remove the excess ssDNAs and any unwanted side product. Run the 3% gel at 60 V for ca. 2.5 h in a running buffer containing 0.5x Tris-Borate-EDTA (TBE) and 10 mM Mg(CH₃COO)₂.
4. Identify and locate the band on the gel that contains the DNA motif. Ensure that the band has migrated to a specific position based on its size.
5. Excise the band containing the DNA motif from the gel carefully. Extract the DNA motif from the excised gel fragment by placing it in a gel extraction spin column and centrifuging at 3,000 x g and 4 °C for 10 min.
6. Replace the buffer in the extracted DNA motif with the annealing buffer using a centrifugal concentrator. Centrifuge at 3000 x g at room temperature (or lower) until the concentrated solution is less than 100 µL. Then, add 300 µL of annealing buffer and repeat this step twice to ensure the buffer is replaced.
7. Dilute the DNA motif to 6 nM for imaging purposes. Use a spectrophotometer or other appropriate methods to accurately determine and adjust the concentration. The DNA motif is now ready for imaging.
   NOTE: All buffers in the protocol are of pH 8.0. The sequence information for the three respective bands is as follows:
   S1: AGGCACCATCGTAGGTTTCTTGCCAGGCACCATCGT
   AGGTTTCTTGCCAGGCACCATCGTAGGTTTCTTGCC
   S2: ACTATGCAACCTGCCTGGCAAGCCTACGATGGACA
   CGGTAACG
   S3: CGTTACCGTGTGGTTGCATAGT

2. Cantilever tip growth

1. Cantilever mounting on the SEM cantilever holder: Ensure that cantilevers are clean and free from any contaminants. Mount the cantilevers onto a suitable holder compatible with the SEM system. The custom-built cantilever holder design can be shared upon request.
2. Gas injection: Heat up the precursor gas (e.g., phenanthrene C₁₄H₁₀ precursor for amorphous carbon tips) to be used on the gas injection system to grow the new tip. As soon as the vacuum is below 10^-5 mbar, purge the gas injection line 10 times for 2 s to be sure no undesired remnant air is in the nozzle line (that must be done with the valve to the gun chamber closed).
3. Tip position adjustment: Use scanning electron microscopy (SEM) to locate the end of the cantilever. Tilt the cantilever holder to an angle (e.g., 11° in this case) equivalent to the one the cantilever will present when placed on the AFM cantilever holder with respect to the surface so the grown tip will be perpendicular to the surface while imaging. Adjust the position and focus of the SEM to obtain a clear view of the cantilever's tip, where a sharp carbon nano-tip will be grown.
4. Focused electron beam-induced deposition (FEBID):
   1. Set the deposition parameters on the selected software (in this case, SmartFIB), such as beam current (I) and acceleration voltage (V), working distance (WD), magnification, exposed shape, dose/deposition time, and dwell time. The following parameters were used to grow amorphous carbon tips, which present good mechanical properties for AFM imaging, leading to lengths around 130 nm and radii in the range of 2-4 nm:
      Select spot/dot as exposed shape
      WD = 5 mm
      I = 78 pA, and V = 5 kV
      Dwell time = 1 µs
      Dose = 0.05 nC, and deposition time = 0.64 s
      Magnification = 20000x
   2. Begin the deposition process to grow the tip by irradiating the electron beam in a spot onto the cantilever tip while simultaneously injecting the precursor gas, closing the gas when the deposition is done. In this case, the SmartFIB software performs this automatically after setting up the recipe mentioned above.
5. Post-growth analysis: Perform post-growth SEM imaging to examine the newly grown tip and ensure its quality and characteristics (tip radius and length). Wait 1-2 min after tip growth to be sure all the precursor gas is pumped out to avoid re-deposition during the SEM imaging. Remove the sample holder from the SEM chamber.
6. Cantilever recycling: In case the SEM system also has a combined focused ion beam (FIB) installed, remove a damaged or dirty previously grown tip by milling it with the FIB system using low FIB currents (e.g., 1 pA, to avoid cantilever damage). Perform the tip removal by milling in a cross-section form, from tip end to base, to avoid tip collapse. This will let re-growing a new one.

3. HS-AFM hardware

1. The imaging setup is composed of the custom-built PORT head^{5, 26} (Figure 1A), high-speed scanner²⁶ with a sample disc with mica on top, compatible controller²⁷, high voltage amplifier with the high bandwidth required for high-speed imaging, AFM base, and PC with the required software to control the before mentioned equipment²⁷.
   NOTE: In this case, these are open-source components for which plans can be obtained from the Laboratory for Bio- and Nano- Instrumentation at EPFL^{27, 28}. It is also possible to attend workshops on how to build the PORT head and controller²⁹, as well as to download and use the LabView based software.
2. Carefully place an ultra-short cantilever (e.g., AC10DS or equivalent) under the spring clip on the cantilever holder using tweezers. Assure that the cantilever chip is fixed and stable.
3. Add 50 µL of liquid using a syringe through the left fluid access port, as demonstrated in Figure 1B. With the excitation laser still off, align the read-out laser on the cantilever using the three dedicated knobs on the AFM head shown in Figure 1A. Do this by observing the shadow of the cantilever on a white paper while maximizing the sum (shadow method). Then, center the laser spot on the photodiode using the two dedicated knobs.
4. Switch on and align the drive laser by checking the Excitation Enable box in Excitation VI, so it actuates and oscillates the cantilever. Show the cantilever excitation signal and the cantilever deflection signal on a connected oscilloscope. If the deflection oscillation is too low (or not existent) it might be that the drive laser is very far off from the cantilever. In that case, use the shadow method and turn off the deflection laser for easier alignment. Maximize the oscillation amplitude using the drive laser adjustment knobs shown in Figure 1A.
5. To adjust the cantilever oscillation amplitude in PORT, input in the corresponding control of the software the peak-to-peak voltage sent to the laser diode control circuit (from now on, peak-to-peak AC input) in Excitation VI. To keep the laser diode in the conduction regime and, therefore, lasing, add a DC voltage to the laser diode control circuit (from now on, DC offset input), which is also done from a configuration box in Excitation VI. Both will also affect the laser power, which can be measured with a laser power meter and is used to tune the cantilever oscillation amplitude.
6. Determine the maximum photothermal excitation frequency that can be applied to the cantilever, which must be below the resonance frequency of the cantilever. Determine the resonance frequency by a thermal tune or a frequency sweep. The cantilevers used in this study (AC10DS), whose resonance frequency in liquids is around 400 kHz (1 MHz in air)³⁰, have a quasistatic bending region below 300 kHz⁵. Thus, PORT frequencies below that limit (around 100-200 kHz) must be used.
7. Adjust the imaging parameters and settings for PORT rate and scan rate to the required values in the Excitation and Scan Vis respectively (the parameters used in this article are stated in the figure descriptions). Once the setup and imaging parameters are configured, perform the required imaging experiments to observe and track the molecular interactions.
   NOTE: Since AC10DS cantilevers are no longer sold, an alternative such as Fastscan D or BioLever³¹ may need to be used until an equivalent is available.

4. Obtaining proper interaction curves

1. Initially, approach the sample surface in contact mode by clicking Start on the Z-controller VI set to Contact Mode.
   1. In this mode, the AFM tip comes into direct contact with the sample. Once the surface is reached, perform a force versus distance curve in Ramp VI to obtain the cantilever deflection sensitivity calibration.
   2. Also, estimate the cantilever spring constant, either from cantilever provider specifications (less accurately), obtained with an interferometer, or through a thermal tune-based calibration after the deflection sensitivity calibration. A precise cantilever calibration is essential for an accurate tip-sample force control.
2. After completely retracting the Z-piezo from the surface where the tip cannot reach the surface by clicking Up in the Z controller VI, switch to the PORT mode in the Z controller VI, and turn on the excitation laser in the Excitation VI checkbox. To begin, set the PORT mode to operate at the desired frequency in Excitation VI, which, in this experimental case, is 100 kHz, using an AC10DS cantilever.
3. Record the cantilever-free oscillation close but not touching the surface. Then, turn the Feedback in the Z controller ON in contact with the surface to record and click Correct to obtain the oscillation of the cantilever when the cantilever is intermittently in contact with the surface, both in nanometers. Subtract the free oscillation curve from the intermittent contact curve to obtain the true interaction curve, converting them to pN using the cantilever spring constant (in this case, 0.1 N/m).

5. HS-AFM imaging

1. Prepare a solution of Mg(CH₃COO)₂ at a concentration of 10 mM. Using a Hamilton syringe, inject 50 µL of the prepared 10 mM Mg(CH₃COO)₂ solution into the fluidic channel of the cantilever holder, creating a drop of liquid that englobes the cantilever.
2. Gradually approach the cantilever to the surface by clicking Start on the Z Controller VI. Once the surface is detected, turn the Feedback ON, and find the peak of the interaction curve in the ORT Force Curves VI. Find the lowest force setpoint that allows proper tracking (below 300 pN to avoid the damaging of fragile bio-samples).
3. Set the Scan Size in the Scan VI to 800 nm by 800 nm and the Line Rate to 100 Hz. Scan the surface by clicking the Frame (down) arrow in Scan VI to check the surface quality. If any contaminations are detected, address the issue by cleaning the surface, cantilever, and/or cantilever holder before proceeding.
4. After scanning, retract the cantilever from the surface by clicking Withdraw in the Z controller VI. Remove the buffer solution from the hole in the cantilever holder with a Hamilton syringe to prevent dead volume issues.
5. Prepare a diluted DNA 3PS solution from part 1 of the protocol. Inject 50 µL of the diluted DNA 3PS solution into the dedicated channel of the cantilever holder.
6. Start the imaging process by repeating steps 5.2-5.3, scanning an 800 nm by 800 nm area at a default line rate of 100 Hz (256 lines × 256 pixels). After the initial scan, adjust the imaging size and speed in Scan VI to the specified values for further data acquisition.
7. Keep the input force setpoint of the tip-sample interaction in the Z Controller VI Setpoint box at the lowest level required for proper tracking (below 300 pN) throughout the imaging process to minimize sample damaging/disturbance unless otherwise specified. Repeat this process for all the required sample areas.

6. Image processing

1. Set up the environment for running the customized Pygwy (Python for Gwyddion) batch processing code, ensuring that Python and all the necessary libraries. More information can be found on Gwyddion's website³² are properly installed on the system. This ensures that the code runs smoothly and can access the required functionalities.
2. Open the customized Pygwy batch processing code (provided on request). This will provide access to the tools and functionalities needed for image processing and analysis.
3. Begin the image processing by performing horizontal median line correction on the images. This step aims to remove any irregularities or artifacts present in the scan lines, ensuring that subsequent processing steps are based on accurate data. After plane background removal, apply the line correction. Use scar removal to further enhance the images by eliminating any unwanted marks or imperfections caused by external factors. This step improves the overall visual appearance of the images.
4. Enhance the visibility and contrast of the images by adjusting the color height map. This ensures that features of interest are clearly visualized and stand out in the images.
5. Select the processed images that need to be combined into a video. Determine the desired frame rate for the video. In this case, set it to 7 frames per second (fps).
6. Calculate the duration of each frame. Use Fiji (ImageJ) to combine the processed images into a video. Ensure that the frame rate and frame duration are correctly set.

Results

In this investigation, the dynamic assembly process of DNA 3-point-star motifs into stable islands was successfully observed utilizing the capabilities of the HS-PORT AFM. This technique allowed us to capture the assembly of these structures in real-time. In Figure 2A,B, we get a clear image scanning at 100 Hz and 200 Hz line rates, respectively, for 100 kHz PORT rate (800 nm by 800 nm scan size). This corresponds to 3.9 and 1.95 oscillation cycles per pixel, respectively. However, imaging at higher speeds without concurrent increases in the PORT rate that would allow a minimum of one oscillation cycle per pixel will drastically reduce the image quality, as can be seen in Figure 2C (0.78 of an oscillation cycle per pixel). The imaging speed is limited by the rate at which the cantilever probes the surface with appropriate tip-sample force control since the rate of topography change becomes too fast to track at the surface sampling rate.

In addition to setting a high enough PORT rate, it is important that the PORT curves contain the information necessary to control the forces. Especially hydrodynamic drag and low detection bandwidth can obscure the applied force. Figure 3 displays the cantilever deflection at different distances from the sample's surface and excitation frequency rates using an average of 4096 curves to reduce the noise. The blue curves in Figure 3A (for 100 kHz PORT rate) and Figure 3D (for 500 Hz PORT rate) represent the cantilever motion when the cantilever is far from the surface and oscillates freely (not touching the surface during one oscillation cycle). The red curves show the cantilever deflection when the cantilever oscillates close to the surface and intermittently contacts the surface, probing the sample. To obtain the true interaction curve and, thus, the tip-sample interaction, we subtract the blue curve from the red curve to obtain Figure 3B,E, respectively. An example of a clear interaction curve required for non-destructive bio-sample imaging is depicted in Figure 3B. This configuration leads to the good imaging quality shown in Figure 3C, where the PORT rate is 100 kHz. As we increase the excitation frequency close to the cantilever resonance, at some point, the feedback control deteriorates, even when the cantilever displays enough amplitude oscillation to detach from the surface after raising the excitation laser power (Figure 3E). If the PORT frequency is too close to the cantilever resonance frequency, the cantilever resonance limits the time response of the cantilever dynamics. The cantilever bending, therefore no longer accurately represents the tip-sample interaction, and the interaction curve we obtain is obscured. This leads to a degradation of the image quality, as displayed in Figure 3F.

A compromised interaction curve at higher PORT frequencies might also happen due to the reduced actuation efficiency at higher frequencies³³, which is currently one of the limiting factors of commercial high-speed AFM cantilevers. This decrease eventually reaches a threshold level in the roll-off, at which proper imaging is hindered: the oscillation amplitude might be insufficient to detach from the surface, and the tip is always in contact, damaging the sample structures. To counteract the decrease in amplitude caused at higher PORT rates, we explored an increase in laser power. We used a PORT head that can reach higher laser powers to better observe this effect. The laser power of the excitation laser, which we measured with a power meter, was divided into the DC and AC components. The total power is controlled by adjusting on the software the peak-to-peak voltage (peak-to-peak AC input) and the DC voltage (DC offset input) sent to the laser diode control circuit. Figure 4A displays the peak-to-peak cantilever oscillation amplitude resulting from an increased peak-to-peak AC input. As expected, the oscillation amplitude correlates well with the AC input value. It is important to note that these peak-to-peak amplitudes are higher than what is normally used in PORT imaging. When imaging fragile samples, the AC amplitude is on the order of 10-30 nm. Figure 4B displays the peak-to-peak cantilever oscillation amplitude for increased DC offset inputs. The peak-to-peak oscillation amplitude remains fairly constant over the DC offset voltage range. We attribute the small variations to nonlinearities in the detection. Figure 4C depicts the increase of the AC and DC laser power components resulting from the increasing peak-to-peak AC. Figure 4D depicts the increase of the AC and DC laser power components resulting from the increasing DC input offset.

The effects of increasing the peak-to-peak AC input and DC offset input were tested separately to determine the respective effects on total power and oscillation amplitude. An increase in laser power causes the sample to be disrupted, likely through temperature increase, whereas an increase in oscillation amplitude will increase the impact force on the sample⁵. Increasing exclusively the peak-to-peak AC input will have less impact on the total laser power increase, as shown in Figure 4C, and will predominantly increase the cantilever oscillation amplitude (and consequentially the impact force), as shown in Figure 4A. Increasing the DC offset input will have a greater effect on the increase of laser power output, as shown in Figure 4D, which heats the sample but will have a negligible effect on the deflection amplitude, as seen in Figure 4B. Imaging results are represented in Figure 4E,F. Figure 4E depicts DNA 3PS for the lowest peak-to peak AC input (left image, circled in blue) and highest peak-to-peak AC input (right image, circled in red), which results in sample damage presumably through the higher forces. Figure 4F exhibits DNA 3PS for the lowest DC offset input (left image, circled in blue) and highest DC offset input (right image, circled in red), where structures are damaged.

Figure 1: The HS-AFM setup. (A) The PORT head on the scanner and base of the AFM, showing the (i) knob for adjusting the read-out laser focus on the cantilever, (ii) knob for adjusting the read-out laser position on the cantilever, (iii) knob for adjusting the horizontal position of the laser on the photodiode, (iv) knob for adjusting the vertical position of the laser on the photodiode, and (v) knobs for adjusting the position of the photothermal drive laser on the cantilever. (B) A close-up of the cantilever in the cross-section view of the head with the sample stage, where (vi) is the spring clip holding the cantilever. The channel for liquid injection is marked in red. Please click here to view a larger version of this figure.

Figure 2: DNA 3PS sample imaged with AC10 at 100 kHz PORT rate at different line rates. (A) 100 Hz (3.9 oscillation cycles per pixel), (B) 200 Hz (1.95 oscillation cycles per pixel), and (C) 500 Hz (0.78 oscillation cycles per pixel). The images were taken at setpoints of 350 pN or less. Scale bar 200 nm. A force curve to obtain deflection sensitivity was taken after imaging. Please click here to view a larger version of this figure.

Figure 3: Cantilever deflection, interaction curve, and image quality at 100 KHz and 500 kHz PORT rates. (A) Cantilever deflection (cantilever free oscillation close but not touching the surface in blue, and the oscillation of the cantilever when the cantilever is intermittently in contact with the surface in red), (B) interaction curve and (C) respective image quality for 100 kHz PORT rate. (D) Cantilever deflection (cantilever free oscillation close but not touching the surface in blue, and the oscillation of the cantilever when the cantilever is intermittently in contact to the surface in red), (E) interaction curve and (F) respective image quality for 500 kHz PORT rate. Images were taken at a line rate of 100 Hz. Scale bar 200 nm. A force curve to obtain deflection sensitivity was taken after imaging. Please click here to view a larger version of this figure.

Figure 4: Effect of peak-to-peak AC input voltage DC offset input voltage. Effect of increasing the (A) peak-to-peak AC input voltage and (B) DC offset input voltage on the peak-to-peak oscillation of the cantilever. Effect of increasing (C) peak-to-peak AC input voltage and (D) DC offset input voltage on AC and DC laser power. Imaging quality at minimum (circled in blue) and maximum (circled in red) (E) peak-to-peak AC input voltage and (F) DC offset input voltage. When increasing the peak-to-peak AC input voltage, the DC offset input voltage was kept at 600 mV. When increasing the DC offset input voltage, the peak-to-peak AC input voltage was kept at 200 mV. Sample integrity is compromised for the high-power configuration. The PORT rate was kept at 100 kHz, and the line rate at 100 Hz. Scale bar 200 nm. A force curve to obtain deflection sensitivity was taken after imaging. Please click here to view a larger version of this figure.

Discussion

When imaging delicate biological samples, off-resonance tapping imaging modes in AFM are particularly useful since they can directly control the tip-sample interaction forces¹⁰. Among them, the PORT mode stands out due to the higher oscillation rates it can reach, which enables higher scan rates. As PORT directly and only actuates the cantilever with a laser, it allows excitation at much higher frequencies than conventional off-resonance tapping modes, particularly when using ultrashort cantilevers with high resonant frequencies⁵. However, when using PORT for HS-AFM, one must be careful with the imaging parameters to achieve good imaging quality.

As depicted in Figure 2, increasing scan rates without simultaneously raising PORT rates will lead to bad-quality imaging. First, PORT rates resulting in less than one oscillation cycle per pixel will drastically reduce the image quality. The imaging speed is inherently limited by the rate at which the cantilever probes the surface with appropriate tip-sample force control. In PORT, the feedback control discretely acts at the end of every oscillation cycle³⁴. This limited sampling rate induces a pure delay in the feedback loop, thereby limiting the achievable feedback bandwidth. If the rate of topography change becomes too fast to track at the PORT sampling rate, the topography can no longer be accurately detected or tracked by the feedback loop. The PORT rate, therefore, needs to be increased to faithfully resolve the sample at higher surface speeds. However, the cantilever's resonance frequency intrinsically limits the maximum PORT rate we can use. If the PORT rate is close to the cantilever's resonance frequency, the deflection measured by the optical lever method no longer faithfully represents the tip-sample forces. To obtain a proper tip-sample interaction curve, both the mechanical bandwidth of the cantilever and the electrical bandwidth of the AFM system must be large enough to detect Fourier components of the deflection curve that are several times that of the PORT frequency. If any of these bandwidths are too low, the interaction curve no longer shows the one-sided cut off (Figure 3A), but rather a sinusoidal behavior (Figure 3D) which obscures the actual tip sample interaction (Figure 3E). This leads to poor force control in the imaging (Figure 3F). Even smaller cantilevers displaying high resonance frequency, which possess at the same time low spring constants are therefore required to increase the PORT rate further. Such high PORT rates and scan speeds present a challenge for AFM electronics, requiring a careful software and electronic design that features the required high bandwidths and optimized digital signal processing strategies.

One of the advantages of PORT compared to tapping mode high-speed AFM is that the surface sampling speed can be tuned. In tapping mode AFM, the surface sampling rate is given by the resonance frequency of the cantilever. When operating in PORT well below the cantilever resonance frequency, we assume that each surface sampling provides a valid data point. As such, the PORT detection bandwidth is given by the Nyquist frequency at half the PORT rate. In the case of tapping mode, the detection bandwidth is given by f₀/2Q³⁵. With the latest developments in PORT, we can gradually increase the PORT rate up to the resonance frequency. In that situation, PORT and tapping mode become almost identical because, at those frequencies, PORT cannot accurately detect the cantilever-to-tip sample interaction, as shown in Figure 3E. When operating at the resonance frequency, the larger oscillation amplitude of PORT compared to the tapping mode becomes a disadvantage. The goal for increasing the PORT imaging rate, therefore remains to increase the cantilever resonance frequency to maintain operation with the PORT rate far enough below the resonance frequency.

As we increase the PORT rate to enable faster scan rates, we will face another problem inherent to photothermal cantilever excitation: at higher frequencies, the oscillation amplitude reduces due to a decrease in photothermal excitation efficiency, eventually to the point that we need to increase the laser power, resulting in sample's temperature rising. Increasing the PORT rate will encumber the balancing of laser power and oscillation amplitude, since a higher excitation power is required to compensate for a reduced actuation efficiency that leads to sample damage. At higher levels of laser power, we could no longer observe the sample, despite using the lowest possible force set point (right panels in Figures 4E,F). We need to be careful about how we will increase the cantilever oscillation amplitude and increase the laser power. There are two parameters we can tune to increase the laser power. First, the peak-to-peak AC input which controls the sinusoidal signal sent to the laser diode to sinusoidally vary the delivered laser power intensity. Second, the DC offset input voltage that keeps the laser diode on the conduction regime and, therefore lasing. An increase of the first one has to be combined with an increase of the second one since an increase in the amplitude of the AC signal sent to the laser requires an increase of the DC voltage to avoid crossing the lasing threshold. However, only the peak-to-peak AC input influences the cantilever oscillation amplitude, as displayed in Figure 4A. DC voltages do not lead to variations in oscillation amplitudes (Figure 4B). Thus, if we want sinusoidal actuation, we need to set the DC voltage to the minimum voltage that lets the diode laser be in the lasing regime. Alternatively, we can minimize the heating by periodically switching the laser diode off and back on during each oscillation cycle rather than using a full sinusoidal drive⁵. The temperature oscillation required to actuate the cantilever in photothermal excitation can be more efficiently reached by switching off the laser during a portion of the cycle, which drops the temperature of the cantilever and, therefore, leads to cantilever bending. This will minimize the delivered energy to the cantilever and sample and thus decrease the total temperature in the surroundings, mitigating sample heat damage.

In this study, we show the limits of imaging parameters in HS-PORT that prevent heat sample damage and proper feedback control. Understanding these limits will help make the best use of the current state of the PORT AFM method, allowing better control over the tip-sample interaction and helping to preserve the native state of the DNA motifs in these measurements. Thanks to the high scan rates reached combined with good feedback control, PORT allowed us to obtain clear and detailed images of the DNA 3PS and their interactions^7,8,36. With the possibility of increased imaging speed and direct tip-sample force control²³, we are able to track the growth and reorganization of biomolecular assemblies more precisely and observe detailed dynamics.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
AC10DS	Olympus	BL-AC10FS-A2	Discontinued
Biometra Compact XS/S	Biometra GmbH	846-025-199	Electrophoresis  unit
Biometra TRIO	Biometra GmbH	207072X	thermocycler for annealing
Custom AFM setup	Laboratory for Bio-Nano Instrumentation, Interfaculty Bioengineering Institute, School of Engineering, Ecole Polytechnique Fédérale Lausanne		Obtainable through Laboratory for Bio-Nano Instrumentation
EDTA	ITW Reagents	A5097	In annealing buffer
Laser Power Meter	Thorlabs	PM100D	Digital Handheld Optical Power and Energy Meter Console
Lively 3AP Power Supply, MP-310	Major Science	MP-310	Electrophoresis Power Supply
MgAc2	ABCR GmbH	AB544692	In annealing buffer
TBE	Thermo Scientific	327330010	Running buffer for electrophoresis
TRIS	Bio-Rad	1610719	In annealing buffer