Understanding dynamic protein behavior on the electrode interface could enhance the efficiency of developing bioelectronics devices, such as biosensors and biofuel cells. To understand the correlation between molecular dynamic behavior and the electrochemical response of proteins adsorbed on the electrode, numerous studies have used spectro-electrochemical techniques that combine electrochemical measurements and various spectroscopy techniques, such as surface-enhanced resonance Raman (SERRS) [1–3] and surface-enhanced infrared absorption (SEIRA) [4–6]. Electrochemical quartz crystal microbalance (EQCM) and surface plasmon resonance (SPR) measurements also have been used to study dynamic molecular adsorption processes [7–9]. These techniques provide excellent opportunities to study redox protein mechanisms and provide overall information concerning the interface between proteins and an electrode; however, the local structure is not obtained, due to a lack of spatial resolution. Moreover, sometimes spectroscopic approaches suffer from the limitation where only protein chromophores are observed. Thus far, ex- or in-situ AFM imaging with electrochemical measurements have been employed to characterize protein adsorption properties on electrodes [10–12]. Furthermore, integrated AFM and scanning electrochemical microscopy (SECM) were developed to obtain in-situ information on electrochemical process at a liquid-solid interface [13, 14]. However, due to a limitation of temporal resolution of conventional AFM, it was difficult to directly monitor rapid protein adsorption processes and the relevant electrochemical processes. In contrast, high-speed atomic force microscopy (HS-AFM) was recently established, which allowed us to directly visualize several dynamic protein events, such as structural changes, molecular interactions, diffusion processes, and molecule adsorption onto a solid surface [15–20].

In this paper, we report the development of an electrochemical AFM based on HS-AFM, which was constructed in the laboratory and can analyze electrochemical measurements with the advantage of fast imaging. This approach could allow direct visualization of dynamic biological molecule behavior at the electrode with simultaneous electrochemical measurements. We applied electrochemical HS-AFM to observe the cytochrome c (cyt c) electrochemical reaction and adsorption processes on a self-assembled monolayers (SAMs)-modified Au electrode surface. Cyt c plays a crucial role as an electron transfer protein in eukaryotic and many prokaryotic respiratory chains. Cyt c is the most extensively studied protein in bioelectrochemistry [21] because cyt c is often regarded as a model metalloprotein for studying electron transfer within and between proteins. It is well known that cyt c electrostatically immobilizes on SAMs of carboxylic acid-terminated alkanethiols via lysine residues surrounding the cyt c heme crevice and exhibits a reversible voltammetric response [22–27]. Numerous studies on direct electrochemistry of cyt c adsorbed on a SAM-modified [25] electrode have employed various surface analysis techniques [5, 9, 11, 28–33]. Until now, however, cyt c molecules adsorbed on electrodes have not been directly observed through real-time imaging using microscopy techniques. This study demonstrates, to our knowledge, for the first time real-time visualization of cyt c adsorption processes on an 11-mercaptoundecanoic acid (MUA)-modified Au electrode with simultaneous electrochemical measurements.

Cyclic voltammetric measurements using our newly constructed EC-HS-AFM were performed to confirm the redox behavior of the 0.1 mM 2,6-dichlorophenolindophenol (DCPIP) in 50 mM pH 7.0 HEPES buffer with a gold electrode. Because reversible redox peaks from DCPIP were observed in a diffusion system, EC-HS-AFM was established as the potential control for the sample stage.

Fig. 1A shows an AFM image of the MUA-modified gold electrode and a cross-section profile with an arbitrarily configured line in the image. The SAM-modified gold surface was flat with a well-packed monolayer and confirmed small pits. This image was consistent with typical AFM images of 11-mercaptoundecanol or MUA SAMs on a Au (111) substrate [34]. The EC-HS-AFM measurements were collected on the MUA-modified gold electrode as the sample stage after cyt c was injected into the analytical solution. Fig. 1 presents AFM images of cyt c molecules adsorbing on the MUA-modified gold surface collected at 2 frame s⁻¹ (see S1 movie). In S1 movie, the cyt c molecules began to adsorb on the MUA-modified gold surface approximately 340 s after injection. The data generated from the time cyt c was added through the beginning of adsorption might be due to cyt c molecules approaching the nearby surface through diffusion. Randomly adsorbing cyt c molecules on the electrode were gradually observed from 340 s to 430 s in S1 movie. Cyt c molecules also frequently disappear immediately after adsorption during this period, which suggests that cyt c molecules rapidly move onto the surface when a small number of molecules adsorb. Next, rapid and remarkable cyt c molecule adsorption was observed from 430 s to 490 s followed by a full adsorption on the electrode surface after 500 s. After 500 s, no noticeable morphologically changes were observed in the AFM images likely because the electrode was full covered by molecules, and the cyt c molecules did not form multilayers.

Cyclic voltammograms (CVs) were sequentially collected while cyt c adsorption was monitored using EC-HS-AFM and are shown from 337.5 s to 517.5 s in Fig. 2. CVs for saturated cyt c adsorption on the gold electrode with an MUA SAM showed the cathodic and anodic peak potentials −122 and −102 mV (vs. Ag wire), respectively, at 522 s. The cyt c midpoint potential was approximately −112 mV (vs. Ag wire). As depicted in Fig. 2, current redox peaks for cyt c were also observed and increased from 340 s to 480 s simultaneously with the cyt c adsorption processes in the AFM images. The peak currents were steady after 500 s. An EC-HS-AFM experiment of the SAM unmodified gold electrode (bare gold surface) was conducted for comparisons of the MUA-modified gold electrode (see S2 movie and S2 Fig.).

The root mean square (RMS) roughness R_RMS, which is given by the standard deviation of height, for the commonly used AFM images were used as roughness profile ordinates and plotted against time in Fig. 3A. Fig. 3B shows a schematic explanation of the cyt c adsorption processes in Fig. 3A and S1 movie. The R_RMS variations indicate that cyt c adsorbed on the electrode. The R_RMS values exponentially increased beginning at approximately 430 s and peaked at 470 s. At approximately the R_RMS peak time, cyt c adsorption was not saturated, but cyt c covered almost half of the sorbable sites on the electrode surface. After 470 s, the R_RMS values immediately decreased because more cyt c molecules adsorbed more on the electrode surface, which likely filled a gap between the immobilized cyt c molecules. We simulated the evolution time for the R_RMS values from random adsorption on a 50 × 50 site based on a change in the number of adsorbing molecules in proportion to the number of the active adsorption sites. The simulation model results show that R_RMS values increase logarithmically with increased surface coverage and reach a maximum at almost half of the coverage (see S3 movie). In Fig. 3A, it is noteworthy that the R_RMS values an exponentially-increased with time, which indicates that the cyt c adsorption process has a cooperative effect. This result suggests that the cooperative effect includes potential interactions, such as electrostatic or hydrophobic interactions between the immobilized cyt c molecules. Reports show that protein adsorption is a cooperative process [35, 36]. The slight change in R_RMS values from approximately 500 s to 550 s indicates saturated cyt c adsorption. We conclude that this result supports protein monolayer formation on the MUA-modified gold electrode as suggested in the AFM S1 movie. The number of the electrochemically active cyt c molecules adsorbed onto the electrode (Γ) was estimated by integrating the observed CV curve reduction peak using the following equation: Γ = Q/nFA. Q is the charge involved in the reaction, n is the number of electrons involved in the redox process (for cyt c-Fe(II) oxidation, n = 1), F is the Faraday constant, and A is the electrode surface area. As depicted in Fig. 3A, the time dependence of the Γ values was also plotted using a sigmoid and the R_RMS values; these results suggest cooperative cyt c adsorption. The value Γ at the steady state after 500 s was calculated as 11 pmol cm⁻² (the largest value in several times), which is consistent with the experimental and theoretical values for full monolayer coverage at 13 pmol cm⁻² [24, 29]. This value also agrees with that of the integrated the peak currents from CV, 9.1 pmol cm⁻² [27]. In addition, Nakano et al. showed that the QCM analysis and AFM imaging supported the cyt c monolayer formation on the MUA surface [11]. The finding of the cyt c monolayer formation agrees with our results of CV and HS-AFM experiments. The difference between our results and their report on the value of electroactive surface coverage could be explained as the way of cyt c immobilization. The cyt c was covalently immobilized on MUA surface in their report, while it was immobilized by the electrostatic interaction in our case. The QCM analysis in their report showed that the value of apparent cyt c surface concentration was 28 ± 12 pmol cm⁻², which was larger than those calculated from CVs (7.2 ± 4.8 pmol cm⁻²) in their report and our data (11 pmol cm⁻²). In the case of the QCM analysis, Furusawa et al. demonstrated that apparent larger frequency changes by the protein immobilization onto the gold surface in the aqueous solutions (ΔF_water) were observed compared with those in the dry air phases (ΔF_air) owing to the interaction with the surrounding water molecules [37]. Myoglobin (Mw: 16.9 kDa), for example, exhibited apparent mass ratio 2.1 [(−ΔF_water)/(−ΔF_air)]. It could happen in the case of cyt c.

The Γ values compared with the R_RMS in Fig. 3A show that both parameters simultaneously increased until 470 s. At approximately 500 s, the Γ values reached the maximum and the R_RMS was returned to the initial value, respectively. On the other hand, although the Γ values began to increase at approximately ~400 s, the R_RMS increased after ~400 s. This observation is presumably due to unimmobilized cyt c molecules on the electrode that were not imaged using HS-AFM because the movement was too rapid, but these molecules which are located adjacent the electrode could exhibit the electrochemical response.

In addition, the peak potential positively shifted with the increasing Γ values (S3 Fig.). A change in cyt c redox potential is induced by several factors, the heme axial ligands [38], the heme environment polarity [39], and the heme solvent accessibility [40] as well as due to the electrostatic interactions between heme propionates and amino acid residues [41, 42]. Recent research has shown that heme propionate hydration shifts the reduction potential of approximately 50 mV toward more positive values [42]. When there is more hydrophobicity and a localized structural change due to well-packed cyt c adsorption on the electrode with MUA SAM, a positive cyt c redox potential shift should be observed that originates with the environment surrounding the heme. The localized structural change of the environment surrounding the heme could modulate several factors that affect redox potential. Further studies are necessary to discern the heme conditions using spectroscopic techniques.

For cyt c adsorption on the saturated electrode, 20 μL of the electrolyte solution of EC-HS-AFM was extracted, and 20 μL of a 4 M NaCl solution was injected into the analytical solution (NaCl at the final concentration 1.13 M) to generate an analytical solution with a high salt concentration. Cyt c molecules desorbed from the MUA-modified gold electrode surface immediately after adding the NaCl solution as shown Fig. 4A and S4 movie. Most of the cyt c molecules desorbed, and the surface with exposed MUA SAMs was observed after 120 s. Similarly, the R_RMS decreased at approximately 120 s but was steady after 120 s as shown Fig. 4B. Certain ordinates of the points were unusual as shown in Fig. 4B; they are due to spike-like noise in AFM images with an abrupt increase in ionic strength. Fig. 4C shows a schematic representation of the desorbing cyt c molecules on the MUA SAM-modified electrode. The results indicate that cyt c molecules desorb due to weaker electrostatic adsorption on electrode-coated MUA SAMs at higher ionic strengths.

In summary, we altered the HS-AFM sample stage to control the electrical potentials and constructed EC-HS-AFM equipment that simultaneously allows direct visualization of dynamic behavior with nanoscale resolution and electrochemical measurements. Using this EC-HS-AFM approach, this study demonstrates, to our knowledge, for the first time direct observation of adsorption processes and the cyt c electron transfer reaction on a SAM-modified gold electrode in real-time. Cyt c molecules were fully adsorbed at approximately 160 s with increasing redox currents from cyt c. The details of cooperative adsorption have not been clarified; however, the findings herein could lead to new, in-depth discoveries on the cyt c adsorption processes. EC-HS-AFM also enables us to directly visualize the structure and dynamics of molecules in response to an applied potential. This use of EC-HS-AFM significantly facilitates studies aimed at understanding the dynamic behavior of biomolecules on an electrode interface and contributes to the further development of bioelectronics.