In Situ Spectroelectrochemical Investigations of Electrode-Confined Electron-Transferring Proteins and Redox Enzymes.

This perspective analyzes recent advances in the spectroelectrochemical investigation of redox proteins and enzymes immobilized on biocompatible or biomimetic electrode surfaces. Specifically, the article highlights new insights obtained by surface-enhanced resonance Raman (SERR), surface-enhanced infrared absorption (SEIRA), protein film infrared electrochemistry (PFIRE), polarization modulation infrared reflection-absorption spectroscopy (PMIRRAS), Förster resonance energy transfer (FRET), X-ray absorption spectroscopy (XAS), electron paramagnetic resonance (EPR), and differential electrochemical mass spectrometry (DMES)-based spectroelectrochemical methods on the structure, orientation, dynamics, and reaction mechanisms for a variety of immobilized species. This includes small heme and copper electron shuttling proteins, large respiratory complexes, hydrogenases, multicopper oxidases, alcohol dehydrogenases, endonucleases, NO-reductases, and dye decolorizing peroxidases, among other enzymes. Finally, I discuss the challenges and foreseeable future developments toward a better understanding of the functioning of these complex macromolecules and their exploitation in technological devices.

Introduction

Electron-transferring proteins and redox enzymes are key players in a large variety of metabolic processes, including photosynthetic and respiratory energy transductions. Their redox cofactors are quite diverse, including metallic centers and organic molecules that span a potential range of about 1.5 V.[1] The use of solution electrochemical and spectroelectrochemical methods, mainly with UV–vis and IR detection and with the aid of redox mediators, is a powerful approach for assessing thermodynamic redox parameters of these complex biomolecules, as well as for studying redox-linked structural changes.[2−4] Alternatively, redox proteins can be immobilized on the surface of electrodes following a variety of biocompatible or biomimetic approaches[5−9] to perform either mediated or direct electrochemistry. The latter approach circumvents protein diffusion and allows for some control on the protein–electrode distance and relative orientation, thus facilitating kinetic direct electron transfer (ET) studies by conventional electrochemical methods. This methodology and particularly the so-called protein film voltammetry technique have provided a great insight into the mechanistic aspects of several electron-transferring proteins and redox enzymes.[10] Moreover, it provides the basis for exploiting bioelectrocatalysis in enzymatic fuel cells, enzyme electrosynthesis, bio-photovoltaic devices, and biosensors.[6,11−14] However, conventional electrochemical techniques, such as voltammetry and chronoamperometry, provide limited information on the immobilized proteins. Coupling protein film electrochemistry with various types of spectroscopic detection allows us to obtain more detailed information on the structure, orientation, conformational dynamics, electronic properties, and chemical transformations associated with the ET and catalytic cycles of the immobilized redox proteins and enzymes.[3,9,15−18] In this perspective, I discuss selected examples of spectroelectrochemical studies of surface-confined redox proteins and enzymes, with focus on the contributions over the last 5 to 10 years. The spectroelectrochemical techniques employed in these selected examples include surface-enhanced resonance Raman (SERR) spectroscopy,[9,16] surface-enhanced infrared absorption (SEIRA),[7] protein film infrared electrochemistry (PFIRE),[19] polarization modulation infrared reflection–absorption spectroscopy (PMIRRAS),[20] electrochemical Förster resonance energy transfer (FRET),[21] in situ X-ray absorption spectroscopy (XAS),[22] in situ electron paramagnetic resonance (EPR),[23] and differential electrochemical mass spectrometry (DMES).[24] In all of the cases, proteins are either physisorbed or chemisorbed onto the working electrode (WE) surface, which is usually functionalized with different organic films that range from self-assembled monolayers (SAMs) of mercaptans on Au or Ag to more complex constructs that may include supported lipid bilayers for biomimetic immobilization of membrane proteins.[5−9,16]

In the first part of the article, I introduce a recently developed theoretical framework for rationalizing the heterogeneous protein ET step, based on kinetic data collected using electrochemical and spectroelectrochemical methods. The second part provides an overview of lessons learned from spectroelectrochemical studies of 13 different types of proteins. The material is organized according to the proteins rather than by techniques as the aim is to exemplify the kind of information that can be obtained rather than introducing the technical aspects of each technique, which, however, can be found in the specific references provided in each case. This information crucially contributes to a better understanding of the functioning of these biomolecules and paves the way for their utilization in devices that rely upon protein immobilization.

On the Heterogeneous ET Step

Proteins are rather flexible macromolecules capable of exploring highly multidimensional and hierarchical free energy landscapes at physiological temperatures. Accordingly, kinetic and thermodynamic ET parameters of redox proteins are not only determined by crucial structural elements, such as the coordination sphere of a redox-active metal but also by protein flexibility and dynamics. For example, as discussed in subsequent sections, flexibility at different levels of the metalloprotein structure has been found to be a determinant for electron shuttling and alternative functions of multifunctional proteins such as cytochrome c (Cyt-c).[25] These dynamical features, particularly the low-frequency motions, are likely to be affected upon protein immobilization onto either electrodes or onto biological interfaces, as well as by the high viscosities typical of crowded media such as cells and organelles.[26] Therefore, while the Marcus semiclassical expression for long-range (nonadiabatic) ET reactions[27] is useful for most in vitro studies in diluted aqueous solutions, some of its underlying assumptions may not be fulfilled for proteins immobilized on electrodes or even under physiological conditions.[28] For example, water–protein nuclear fluctuations are assumed to be much faster than electron tunneling at the crossing point of reactant and product parabolas. This assumption may break down due to either slowdown of nuclear modes imposed by the medium or enhanced electron tunneling probability. So far, most theoretical and experimental efforts to address dynamical effects in ET have focused on ultrafast reactions.[29,30] For a slower ET process in the sub-millisecond timescale, direct electrochemistry and spectroelectrochemistry have proven useful tools for extracting thermodynamic and kinetic parameters, specifically for electron shuttling metalloproteins that contain the metal center partially exposed or very close to the surface. In recent years, different groups have investigated the heterogeneous ET reactions of a variety of native and non-native redox metalloproteins, including heme proteins such as cytochromes c, c6, and b562, as well as binuclear (CuA) and mononuclear (T1) copper proteins either physisorbed or chemisorbed on metal electrodes coated with SAMs of mercaptans with different compositions.[26,31−46] A common feature experimentally observed for all of these systems is a transition in the distance dependence of the zero driving force ET rate constants (kET0) from an exponential decay at long protein–electrode distances to a plateau at shorter distances. Some examples are shown in Figure A,B. This type of behavior has been rationalized using a variety of arguments, depending on the experimental information available in each case. This includes a two-state conformational or orientational gating from a redox-inactive to a redox-active orientation of the adsorbed protein, as well as an electric-field controlled dynamical gating from populations with low electronic coupling to populations with higher electronic coupling.[40−43,47,48] In these models, the rate-limiting step at the thinner SAMs is reorientation rather than ET. On the other hand, it has been shown that in some cases the measured ET quantities reflect a distribution of orientations,[39] and that this distribution may be the origin of the unusual distance dependencies exemplified in Figure , even if protein orientations do not change during the experiment.[49]

Figure 1

kET0 of heme (A) and copper (B) proteins immobilized on SAM-coated electrodes as a function of the chain lengths, as determined electrochemically or spectroelectrochemically. The lines are fittings to Matyushov′s equations. Panel (A) includes Cyt-c wired to pyridinyl-terminated SAMs (black) adsorbed on COOH/OH (red) and COOH (orange) SAMs and cross-linked to COOH/OH-SAMs (light green), as well as Cyt-c6 (light blue) and Cyt-D (green) on CH3/OH-SAMs. Panel (B) includes azurin (Azu; red) and CuA (green) native proteins, as well as five chimeric proteins. Light blue, blue, and wine symbols correspond to the CuA scaffold that has been engineered to host the T1 copper sites of azurin (Azu-CuA), amicyanin (Ami-CuA), and the CBP protein (CBP-CuA), respectively. Pink and orange symbols are CuA chimeras containing the Thermus thermophilus scaffold and loops from other species (Tt-3L-Hs and Tt-1L-E, respectively). (C) ET activation energies of proteins adsorbed on thick (C15-SAMs; blue) and thin (C5-SAMs; red) coatings. The data correspond to WT Cyt-c, its point mutants Y67F and E66Q, Cyt-c nitrated at Tyr74 (NO2-Cyt-c), WT Azu, and the chimeras Azu-CuA, Tt-3L-Hs, and Tt-3L-At. Figures are created using data taken from refs (26, 31, 32, 35, 36, 38, 39).

Interestingly, ET activation energies of some adsorbed proteins, determined from either the temperature or overpotential dependencies of kET, were found to increase upon shortening the tunneling distances (Figure C),[26,31,32,34,37,50,51] in parallel with a change of sign of the activation volume.[37]

Waldeck and co-workers interpreted these findings, which are not consistent with a gating model, in terms of a change of the ET regime from nonadiabatic at the thicker SAMs to friction-controlled ET in the plateau region.[32,45,46] Albeit with some variability depending on the protein and immobilization method, the transition is generally observed at SAM thicknesses of about 10 methylene groups, which correspond to through-bond tunneling distances of ca. 19 Å (Figure A,B). For longer distances, i.e., in the nonadiabatic regime, the rate constant can be treated in terms of the Marcus semiclassical equation integrated to account for the density of states of the metal electrode, which can be approximated aswhere Δ = Δ0 exp(−β(R + R0)) is the protein–electrode coupling strength, which decays with the distance R at a rate β ≈ 1 Å–1. kB is the Boltzmann constant, ℏ is the reduced Planck constant, φ is the electrode overpotential with respect to the protein reduction potential (E°′), erfc is the complementary error function, and λ is the effective ET reorganization energy, which differs from the Marcus definition for nonergodic systems.

For shorter electrode–protein distances, eq is empirically corrected by a crossover parameter g to account for dynamical effects (friction control)[52]The limiting cases of nonadiabatic and friction–controlled ET reactions are obtained for g ≪ 1 and g ≫ 1, respectively. The latter has been ascribed to solvent dynamics affecting the crossing at the top of the activation barrier when the medium relaxation time exceeds the time of electron tunneling at the activated state. In this case, the crossover parameter for the electrochemical reaction of the surface-confined species is defined as[52]where the average Stokes-shift relaxation time τ represents the decay of the time-correlation function of the reaction coordinate. Note that the combination of eqs –3 leads to the experimentally observed distance dependence of kET0 (Figure A,B), including the plateau and exponential regions. This agreement, however, is only apparent as the relaxation times required to enter the frictional control regime, i.e., g ≫ 1, are about 200 ns, which is 6 orders of magnitude slower than the longitudinal relaxation time in bulk water and 3 orders slower than molecular dynamics (MD) simulation estimates of τ for protein/SAM/electrode systems.[53] This strong discrepancy reveals that low-frequency dynamical features need to be explicitly considered, in addition to fast nuclear medium modes. Waldeck et al.[32,45,46] addressed this issue using an adaptation of Zusman′s equations. This approach leads to a reasonable qualitative description of the experiments, but relies upon approximations that are not fulfilled when reactants are attached to the electrode surface.[52]

Recently, Matyushov[52] introduced a novel phenomenological formalism where the distance of the immobilized protein to the electrode is described by overdamped oscillations of the protein in a soft harmonic potential with force constant κ, such that the average electronic coupling ⟨Δ⟩ can be expressed asand the crossover parameter isThe effective relaxation time τeff is a function of the characteristic times τ and τR, with τR = (β2DR)−1 and DR an effective diffusion coefficient that accounts for the oscillatory motions of the protein at the interface.

This theory has been recently contrasted with experimental results obtained for Cyt-c, azurin, and CuA centers.[50,52] The combination of eqs , 2 and 5, with Δ replaced by ⟨Δ⟩, allowed for quantitative fitting of the observed distance dependencies of kET0 in the entire range, yielding τeff and τR values in the microsecond and nanosecond timescales, respectively.[50,52] Consistently, τeff and τR values were found to be protein specific and to decrease with the flexibility of the scaffold and with the diffusion coefficient, respectively.[50] The data suggest that the amplitude of the oscillation is an effective parameter that represents a convolution of linear motion and angular reorientation coupled with interfacial water dynamics, in agreement with previous molecular dynamics simulations and time-resolved surface-enhanced resonance Raman investigations of Cyt-c on SAM-coated electrodes.[33,47,48,54]

The model predicts higher ET activation energies in the frictional control regime, Ea(FC), compared to the nonadiabatic regime, Ea (NA), in agreement with the experiments (Figure C)Moreover, incorporation of the experimentally verified power law dependencies of kET(26,31,32,37,50,51) and τeff[50] with medium viscosity (η) into the model, allows for the definition of the crossover parameter g in terms of protein-specific empirical parameters that reflect the sensitivity to η of relevant protein motions.[50]

An important outcome of these studies is that τ and τR have similar magnitudes, which results in frictional control even at electron tunneling distances as long as 19 Å. Moreover, the frictional control is strongly enhanced by the medium viscosity. These findings suggest that frictional control is likely to be an important feature in long-range intra- and intermolecular ET reactions in vivo, particularly in mitochondrial respiration as intramitochondrial viscosities vary between ca. 40 and 400 cP, depending on the physiological conditions. The sensitivity to the viscosity, and therefore the crossover parameter, is dictated by the specific dynamical features of each metalloprotein.

Spectroelectrochemical Case Studies

Cytochrome c (Cyt–c)

This small monohemic protein mediates ET from complex III to complex IV in the mitochondrial respiratory chains and has a number of alternative proapoptotic functions such as cardiolipin (CL) peroxidation, apoptosome assembly, and histone chaperone inhibition.[25,55] The heme iron of the native Cyt-c has a hexacoordinate low-spin (6cLS) configuration with Met80 and His18 as axial ligands (horse heart Cyt-c numbering; Figure A).

Figure 2

(A) Crystallographic structure of ferric horse heart Cyt-c (PDB 1HRC). Ω-loops 20–35, 40–57, and 70–85 are represented in ocher, lime, and orange, respectively. (B) Per residue root-mean-square fluctuations (RMSF) of different Cyt-c variants. Red: Y67F, blue: NO2Cyt-c, orange: WT Cyt-c, and green: E66Q. Adapted with permission from ref (31). Copyright 2020, Elsevier.

Cyt-c has been the subject of an enormous amount of fundamental and applied research,[25] including a number of SERR and SEIRA spectroelectrochemical investigations starting with the seminal work by Hildebrandt and Stockburger in 1989.[56] Part of this work has been summarized in previous review articles.[9,16,17,57] In most (spectro)electrochemical studies, the protein is electrostatically adsorbed on SAM-coated electrodes containing carboxylate functional groups,[9,16,17,31,33,47,58−63] although adsorption,[36,64−67] cross-linking,[54] and wiring[32] to other types of SAMs have also been reported. SERR and SEIRA experiments, in combination with molecular dynamics simulations, show that adsorption of Cyt-c to negatively charged SAMs involves the patch of positively charged lysine residues that surround the partially exposed heme edge and that constitutes the binding site to natural partner biomolecules.[33,36,48,65,68] Increasing electric fields at the SAM/protein interface have been shown to induce conformational transitions of the adsorbed Cyt-c that do not necessarily involve significant alteration of the secondary structure but a motion of the flexible Ω-loops (Figure A), which in turn leads to the detachment of the labile sixth ligand Met80, particularly in the ferric state. This axial position may either remain vacant, leading to a five-coordinate high spin (5cHS) heme or be occupied by two alternative His residues (His33 or His26) leading to a bis-His 6cLS configuration in equilibrium with variable proportions of the native and 5cHS forms. These conformational changes can be modulated and eventually suppressed by controlling the interfacial electric field, which in turn depends on the electrode material and applied potential, on the thickness and functional groups of the SAMs, and on the pH and ionic strength of the solution.[9,16,17] Interestingly, the interaction of Cyt-c with the biologically relevant lipid cardiolipin (CL), both in solution[69] and upon adsorption of the Cyt-c/CL complex on SAM-coated electrodes, leads to similar conformational changes.[64] The alternative 5cHS conformation presents enhanced peroxidase and nitrite reductase activity.[64,66,67,69] The magnitude of the electric field operating at the SAM/protein interface was assessed through SEIRA measurements of the vibrational Stark effect experienced by nitrile reporters incorporated as head groups of the SAMs[70] and into the protein,[60] affording values comparable to those found at biological membranes.

Even under conditions that do not lead to conformational changes, the local electric fields play a crucial role in modulating the dynamics of Cyt-c in the electrostatic complexes, which in turn determines the electronic coupling, as verified by time-resolved SERR[33,47,48,54] and SEIRA[58,71] spectroelectrochemical experiments (TR-SERR and TR-SEIRA, respectively), as well as by electrochemically controlled plasmonic detection.[72] TR-SERR experiments also show that relatively weak electrostatic SAM/Cyt-c interactions that do not induce ligand exchange produce subtle deformations of the flexible Ω-loops that affect the H-bonding network and, more specifically, disrupt the Met80-Tyr67 H-bond.[59] The effect of this electrostatic perturbation, which implicates the same binding domain involved in interactions of Cyt-c with partner biomolecules, is a twofold decrease of the reorganization energy that facilitates ET. Thus, electrostatic interactions of Cyt-c crucially determine ET parameters and transitions to alternative conformations able to perform alternative functions. To shed more light on this topic, Oviedo-Rouco et al.[31,63] studied the heterogeneous ET reactions and conformational transitions of different Cyt-c variants adsorbed on SAM-coated electrodes by TR-SERR spectroelectrochemistry. The variants include the wild-type protein, the mutants Y67F and E66Q, and the WT protein nitrated at the Tyr74 residue (NO2-Cyt-c). All of these proteins have a conserved Met/His axial coordination and largely superimposable structures, but differ in the flexibility of some specific regions (Figure B), mainly at the level of the Ω-loops 20–35, 40–57, and 70–85. Interestingly, the reduction potentials (E°′) show a clear correlation with the root-mean-square fluctuations (RMSF) of the loop 40–57 (Figure A) and weaker dependencies with other structural elements (not shown).[31]

Figure 3

Correlations found for redox and conformational parameters of four Cyt-c variants: WT Cyt-c, Y67F, E66Q, and NO2-Cyt-c. (A) Fluctuations of the Ω-loops 40–57 (red) and 70–85 (blue) as functions of the E°′ and pKa values, respectively. (B) ET activation energies measured at thin (blue) and thick (red) SAMs plotted as functions of the pKa and E°′ values, respectively. The lines are included to guide the eye. Adapted with permission from ref (31). Copyright 2020, Elsevier.

To assess whether these differences in protein flexibility also affect the conformational transitions, the pKa of the so-called alkaline transition was determined for the same set of Cyt-c variants. This transition consists of the pH-induced replacement of the Met80 axial ligand by either Lys73 or Lys79 to yield Lys/His axial coordination.[25,63] The pKa value of the transition is 9.4 for WT Cyt-c, and it shifts downwards and upwards for the other variants. As shown in Figure A, pKa values tend to increase with the RMSF of the loop 70–85, but show no correlation with the flexibilities of the other two loops. Thus, the thermodynamic parameters that characterize the ET function and the transition to alternative conformations (E°′ and pKa, respectively) are modulated by the conformational flexibility of the protein, but each magnitude is fine-tuned by the flexibility of different structural elements.

Consistent with the results presented in the previous section, the ET activation energies of the four Cyt-c variants are higher when measured at thin SAMs, i.e., in the friction-controlled regime (Ea(FC)), than at thick SAMs, i.e., in the nonadiabatic regime (Ea(NA)). Moreover, Ea(FC) increases with the pKa′s of the alkaline transitions, while Ea(NA) shows an inverse dependency with E°′ (Figure B). These cross-correlations suggest that Ea(FC) and pKa are largely determined by the flexibility of the loop 70–85 while Ea(NA) and E°′ are mostly influenced by the flexibility of the loop 40–57. As previously shown, the loop 70–85 contains most of the residues that constitute the binding site of Cyt-c to the SAMs (and to natural partners)[48] and, therefore, its flexibility is expected to modulate Ea(FC) through κ and Es (eq ). Moreover, this loop contains the Lys residues involved in the alkaline transitions and, therefore, its flexibility is expected to affect pKa. On the other hand, the decrease of E°′ values with increasing flexibility of the loop 40–57 is consistent with a higher accessibility of water molecules to the redox center, which is known to downshift E°′ due to preferential stabilization of the Fe3+ form.[25] The parallel decrease of Ea (NA) is more difficult to rationalize as discrimination of inner-sphere (λin) and outer-sphere (λout) contributions is not trivial, but it has been argued that most likely it is related to a decrease of λout with increasing loop flexibility, rather than to variations of λin.[31]

One should note that in all of these studies of immobilized proteins, the measured quantities represent an average of all electroactive orientations. In the case of electrostatically adsorbed Cyt-c, this distribution of orientations is rather narrow as shown by MD simulations, SEIRA, and SERR experiments.[33,36,47,48,54,58,65]

In summary, TR-SERR spectroelectrochemical experiments demonstrate that for Cyt-c at interfaces of relatively low local electric fields the flexibility of the Ω-loops critically modulates the redox parameters of the native Cyt-c conformation and the transition to alternative conformations with a different functionality. These results are consistent with solution experiments showing that the pathogenic mutations G41S, Y48H, and A51V found in humans, and other modifications at the level of the loop 40–57, result in enhanced flexibility of both Ω-loops compared to the WT protein, which leads to higher peroxidase activity due to partial detachment of the axial ligand Met80, and lower pKa and E°′ values.[73−78]

Respiratory Complex IV

This integral membrane protein complex, also termed cytochrome c oxidase (CcO), is the terminal enzyme in aerobic respiration. It contains a CuA center that accepts electrons from Cyt-c, which are subsequently transferred to a heme a and from there to the catalytic binuclear heme a3-CuB site where O2 is reduced to H2O. The process is coupled to proton translocation across the membrane. Steininger et al.[79] used TR-SEIRA to investigate the conformational changes of R. sphaeroides CcO occurring in the ms timescale during the enzymatic turnover, i.e., in the presence of O2, and compared these results with previous studies under anaerobic conditions.[80] For these experiments, the enzyme containing an engineered His-tag was bound to a nitrile triacetic acid (NTA)-modified electrode followed by reconstitution into a lipid bilayer (Figure ), as previously described.[81,82]

Figure 4

Top: Schematic representation of CcO immobilized via an engineered His-tag on a Ni–NTA-coated Au electrode and reconstituted into a lipid bilayer. Bottom: TR-SEIRA spectra of the amide I band recorded under aerobic conditions after a change of potentials from −0.8 V to the open-circuit potential. Adapted with permission from ref (70). Copyright 2016, Elsevier.

The His-tag was introduced into subunit II of the enzyme, which contains the CuA electron entry site, to allow for functional stepwise heterogeneous electron transfer from and to the electrode.

The immobilized enzyme was first electrochemically fully reduced at −800 mV. Thereafter, the potential was switched to the open-circuit value and the enzyme was allowed to perform enzymatic reoxidation in the presence of molecular oxygen. TR-SEIRA spectra were acquired using the step-scan mode, triggered by periodic potential pulses from −800 mV to the open-circuit potential. Changes of the protein secondary structure associated with the enzymatic turnover were assessed by phase sensitive detection of the amide I region (Figure ). This analysis indicates a high flexibility of the secondary structure including the helices surrounding the catalytic center, which is higher for β-sheet elements compared to α-helices both under aerobic and anaerobic conditions. Conformational changes are faster in the case of reoxidation by oxygen compared to electrochemical reoxidation and reflect the full turnover of CcO.[79]

Following a different strategy, Sezer et al.[83] adsorbed detergent solubilized CcO from R. sphaeroides on Ag electrodes coated with NH2-terminated SAMs that are meant to mimic the electrostatic binding of the enzyme to its electron donor Cyt-c. The integrity and electrochemical reduction of the enzyme was monitored by SERR. As in a previous work,[84] reduction of both hemes was monitored from the downshift of the ν4 vibrational band, while the redox state of the heme a3 was selectively assessed from the position and intensity of the isolated bands that arise from the vibrations of the formyl group. Under anaerobic conditions both hemes could be successfully reduced at potentials very close to those measured in solution, thus indicating the integrity of the centers. However, a significant population of solely heme a was found to undergo a conformational change in the adsorbed state that implies the loss of one His ligand, as revealed by SERR, and a concomitant downshift of the reduction potential.[83] Using the same methodology, in combination with H2O/D2O experiments and QM–MM calculations, the authors were able to identify the resonant enhanced CH2 twisting modes of the propionates of the individual hemes a and a3, and their responses to the protonation state.[85] These studies reveal that at least three of the four heme propionates are protonated in the fully reduced enzyme, whereas for an intermediate redox state with the reduced heme a and oxidized heme a3 only one heme a3 propionate is protonated, thus supporting the involvement of this group in the proton pathway.[85] The PMIRRAS technique was also applied to monitor the adsorption of CcO on SAM-coated electrodes and to assess the integrity and orientation of the immobilized enzyme.[86]

A few other terminal O2-reductases, such as the aa3 quinol oxidase from Acidianus ambivalens,[87] the cbb3 enzyme from Bradyrhizobium japonicum, [88] and the bo3 ubiquinol oxidase from Escherichia coli [89] have also been investigated using SERR and SEIRA spectroelectrochemical methodologies.

Respiratory Complex I (CpI)

CpI is a very large membrane enzyme catalyzing the first step of electron transport chains. It oxidizes NADH transferring electrons to membrane soluble ubiquinone, for which it is endowed with an internal chain of several redox groups including FMN and iron–sulfur clusters. The ET steps in the CpI drive proton translocation across the membrane. Gutiérrez-Sanz et al. used SEIRA to investigate the CpI from R. marinus in a biomimetic construct.[90] For this purpose CpI was incorporated into two types of unilamellar liposomes and the resulting proteoliposomes were deposited onto the Au-coated ATR crystal functionalized with a SAM of 4-aminothiophenol (Figure ).

Figure 5

Schematic representation of the biomimetic construct of a CpI-lipid bilayer system adsorbed on a Au-coated ATR crystal functionalized with 4-aminothiophenol. Proton translocation leads to either acidification (A) or alkalinization (C) of the SAM surface, depending on CpI orientation. Electrochemical quinine reoxidation leads to acidification in both cases (B). Reproduced from ref (81). Copyright 2018, American Chemical Society.

Changes of the amide I/amide II band intensity ratio reveal conformational changes during catalysis, which may involve movements of transmembrane helices or other secondary structural elements. Moreover, the 4-aminothiophenol SAM is shown to behave as a SEIRA reporter of pH that responds to the local changes that result from the proton translocation activity of CpI (Figure ). Santos Seica et al.[91] moved one step forward in monitoring conformational movements during the CpI turnover. To that end the authors introduced a small and highly flexible nitrile infrared label that allows to monitor intramolecular movements of CpI segments by SEIRA. Specifically, they labeled two residues belonging to the amphipathic helix across the membrane arm of CpI and anchored the protein to the Au-coated ATR crystal using the His-tag/Ni–NTA technique. SEIRA experiments show that the labeled residues move to a more hydrophobic environment upon NADH reduction of the enzyme, likely as a response to the reorganization of the antiporter-like subunits in the membrane arm.

Hydrogenases

[NiFe]-hydrogenases are a family of enzymes that contain a [NiFe] bimetallic active site in a large subunit where H2 can be oxidized to H+ and vice versa. A smaller ET subunit contains a chain of iron–sulfur clusters. In addition to terminal and bridging cysteine ligands, the [NiFe] center contains CN– and CO coordinated to the iron. These unusual ligands are strongly IR-active and represent sensitive markers of the electronic state of the center, which has driven a large number of IR-based spectroscopic and spectroelectrochemical studies.[92,93] Spectroelectrochemical studies of electrode-confined hydrogenases are motivated both for structural and mechanistic fundamental aspects as well as for the technological interest of H2 splitting as an energy source. For example, Zebger and co-workers[94−99] performed SEIRA studies of a variety of hydrogenases from different organisms immobilized on Au-coated ATR crystals using various immobilization procedures. This includes the attachment to a Ni–NTA-modified Au surface of the Ralstonia eutropha enzyme containing a C-terminus His-tag,[94] the electrostatic adsorption to SAMs of amino-terminated mercaptans of the enzymes from Desulfovibrio vulgaris [95−97] and R. eutropha,[98] as well as electrostatic adsorption of the R. eutropha enzyme to films of aromatic amines obtained by electrochemical reduction of diazonium salts.[99] The D. vulgaris enzyme has also been studied by SEIRA with its lipid tail inserted into a phospholipid bilayer supported on a 4-aminothiophenol SAM.[100] SEIRA characterization shows that the enzymes can be immobilized with preservation of the active site structure and with relatively uniform orientations that enable either direct or mediated ET and electrocatalytic H2 oxidation. Voltammetric cycling is found to reversibly activate–inactivate the adsorbed enzyme, which could be related to specific intermediate-state redox couples and to a potential-induced enzyme reorientation.[95,96,98]

The alternative PFIRE approach introduced by Vincent and co-workers[19,101−103] consists of adsorbing the enzyme on a high surface area carbon electrode constructed from carbon black deposited onto a Si internal reflection element. The layer of enzyme-modified carbon particles is finally covered with carbon paper and a graphite connector. This approach allowed for the identification of catalytic intermediates in wild-type and single mutant hydrogenases, thus contributing to the elucidation of the catalytic mechanism.[102,103] For example, PFIRE experiments demonstrated that the so-called Ni–L state is an active catalytic intermediate of [NiFe] hydrogenases that for the E. coli enzyme can be generated reversibly in the dark at room temperature in a pH-dependent process.[19,101]

Multicopper Oxidases (MCO)

This group of enzymes oxidize a variety of substrates by accepting electrons at a mononuclear copper center and transferring them to a trinuclear copper site where O2 is reduced to H2O. As oxygen reduction is the limiting step in fuel cells, substantial efforts are in progress to develop efficient biocathodes based on MCO such as laccases (LAC) and bilirubin oxidases (BOD). For example, Su et al.[104] studied adsorbed LAC on carbon nanotubes functionalized with naphthyl residues previously deposited onto a Au-coated ATR crystal for SEIRA characterization. Potential-dependent SEIRA spectra were analyzed using Fourier self-deconvolution and two-dimensional correlation spectroscopy to extract structural information from the amide I band. This analysis demonstrated that the adsorbed LAC retains the native structure in both oxidation states and, furthermore, reveals redox-linked conformational changes similar to the protein in solution. In agreement with these spectroscopic observations, the immobilized LAC shows excellent oxygen reduction activity.[104] Hitaishi et al.[20] used PMIRRAS to characterize the orientation and structural integrity of BOD in biocathodes constructed by electrostatic adsorption of the enzyme on SAM-coated Au electrodes at variable pH. They found out that the protein dipole moment and the charge in the vicinity of the electron entry Cu site drive the enzyme orientation. For weak electrostatic interactions, local pH variation affects the electron transfer rate as a result of protein mobility on the surface, while stronger electrostatic interactions destabilize the protein structure. Macedo et al.[22] developed a setup for in situ XAS and applied it to probe ET in BOD-based biocathodes during the oxygen reduction regime. The enzyme was directly adsorbed on carbon cloths oxidatively functionalized and covered with carbon nanoparticles to create a mesoporous structure. The electrochemical reduction of the BOD active site was monitored by following the attenuation of the XAS signal at 8997 eV attributed to Cu2+ and the rising of a signal at 8983 eV attributed to Cu+. In the presence of molecular oxygen, BOD reduction requires an overpotential of about 150 mV compared to an inert atmosphere. These experiments provide evidence that the copper ions act as a tridimensional redox-active electronic bridge for the electron transfer reaction.

Blue Copper Proteins

These small soluble proteins function as electron shuttles using a mononuclear T1 copper center. Davis, Canters, and co-workers[21] introduced the concept of monitoring protein interfacial ET at optically transparent electrodes by means of redox-state-dependent fluorescence changes. The strategy consists of binding an organic fluorophore to the protein surface, to establish FRET between the dye and the redox site. The initial work was carried out with Cy5-labeled blue copper azurin adsorbed on alkyl-terminated SAMs using total internal reflection fluorescence microscopy (TIRF). This method, which was extended to other T1 blue copper and heme proteins,[105] demonstrated a high sensitivity that allowed for the spectroelectrochemical detection of a few hundred molecules in the earlier studies,[21,105,106] down to the direct and mediated single-molecule ET level in more recent developments by Akkilic et al.[107] and Pradhan et al.,[108] respectively. In the latter case, azurin was labeled with the dye ATTO647N and with biotin. The protein was immobilized on glasses previously functionalized with biotin and incubated with NeutrAvidin (Figure ) and was subjected to a fixed applied potential using a ferricyanide redox mediator.

Figure 6

Single azurin (Azu) imaging and intensity traces at different potentials. (A) Spectroelectrochemical setup. (B) Schematic of the immobilization on PEG-passivated glass through NeutrAvidin–biotin binding. (C and D) Confocal images of the same area recorded at oxidizing and reducing potentials, respectively. (E) Azu structure containing reduced and oxidized Cu and the dye (yellow) in the fluorescence (top) and quenched (bottom) states, respectively. (F) Time traces of the same single Azu molecule recorded at different applied potentials. Reprinted with permission from ref (99). Copyright 2020, The Royal Society of Chemistry.

Analysis of the fluorescence time traces unveils significant fluctuations of the ET rates and reduction potentials that indicate dynamical heterogeneity. The observed changes are ascribed mainly to variations in complex association constants and/or structural changes, as well as to changes in the reduction potential and electronic coupling to a lesser extent.[108]

Alcohol Dehydrogenase (ADH)

NAD-dependent alcohol dehydrogenases (ADHs) catalyze the oxidation of alcohols to aldehydes and ketones with the concomitant reduction of NAD to NADH. Crespilho et al. developed setups for studying the electrocatalytic performance of enzymes adsorbed onto electrodes with either DMES[24] or in situ EPR[23] detection and applied them to immobilized ADH. The DEMS setup utilizes a flexible carbon fiber (FCF) working electrode, which facilitates the coupling of protein film voltammetry for monitoring NADH formation with mass spectrometry for detecting volatile intermediates and final products formed during enzymatic alcohol oxidation (Figure ). The results confirm dissociation of NADH as the rate-limiting step, thus favoring an ordered sequential Bi–Bi mechanism.[24] For EPR detection, the carbon fibers were functionalized with quinones, wherein the quantity of unpaired electron spins can be measured. The experiments show an increasing number of free unpaired electrons with increasing applied overpotentials and NADH oxidation, which suggests that the quinone groups on the carbon material electrocatalyze the oxidation of NADH to NAD+.[23]

Figure 7

Schematic representation of the DEMS electrochemical cell. CE: counter-electrode, WE: working electrode, RE: reference electrode, and FCF: flexible carbon fiber array modified with ADH. The interface is constituted by a PTFE membrane over a steel frit. Reprinted with permission from ref (24). Copyright 2017, The Royal Society of Chemistry.

Other Redox Proteins and Enzymes

Moe et al.[109] used a combination of SEIRA and CV to study endonuclease III, a DNA glycosylase that removes oxidized pyrimidines from DNA, adsorbed on Au electrodes coated with SAMs of carboxyl-terminated alkanethiols and of normal and damaged DNA. The study shows that electrostatic interactions required for the redox activation of the enzyme may result in high electric fields that alter the structure and thermodynamic properties of the enzyme. It also indicates that the ET is modulated by subtle differences in the protein–DNA complex.

Kato et al.[110] used a combination of SEIRA and CV to determine the reduction potentials of the iron cofactors of P. aeruginosa NO-reductase. With the aid of CO as the vibrational probe, the researchers were able to show that the reduction of the heme b3 initiates the enzymatic NO reduction.

Salewski et al.[111] studied E. coli Type II NADH:quinone oxidoreductase adsorbed on SAM-coated electrodes using SEIRA in combination with other spectroscopic and electrochemical methods. The study revealed two distinct substrate binding sites for NADH and the quinone and a bound semiprotonated quinol as a catalytic intermediate.

Todorovic et al. used SERR spectroelectrochemistry to characterize novel dye decolorizing peroxidases (DyP) adsorbed on SAM-coated electrodes. The studies reveal a well-preserved structure of the heme pocket for the adsorbed enzymes, albeit with altered spin populations with respect to the proteins in solution,[112] and allowed for the identification of a catalytic intermediate, most likely compound I.[113] Moreover, they found the immobilized DyP from Pseudomonas putida to outperform the broadly used HRP as the electrocatalytic biosensor for H2O2.[114]

Aiming to assess its potential for constructing biosensor devices, Silveira et al.[115] studied by SERR the periplasmic cytochrome PccH from Geobacter sulfurreducens adsorbed on SAMS with different functionalities. The structural and thermodynamic features of PccH are preserved upon attachment on mixed −NH2/–CH3 SAMs, while adsorption on single component −OH, −NH2, and −COOH-functionalized SAMs leads to a distribution of native and non-native heme spin configurations.

Kielb et al.[116] used SERR for characterizing the redox properties of the hexameric tyrosine-coordinated heme protein (HTHP) from Silicibacter pomeroyi, which is known to exhibit peroxidase- and catalase-like activity. The experiments reveal two redox transitions at −0.17 and −0.54 V, which are attributed to different orientations of the protein on the SAM-coated Ag electrodes. Reduction is found to lead to partial loss of heme cofactors.

Future Outcomes and Challenges

The immobilization of electron-transferring proteins and redox-active enzymes on electrode surfaces gives rise to significant interest because it enables electrochemical probing of their function in a simple way. It also sets the stage for the utilization of these macromolecules as building blocks for a variety of devices, including biosensors, biofuel cells, bioelectrocatalytic reactors, and bio-photovoltaic devices.

Protein immobilization, on the other hand, imposes conditions that may be quite distinct from solution in terms of local dielectric constants and electrostatic potentials, local ion concentrations, specific and unspecific interactions, and mobility. Therefore, immobilization strategies should be carefully designed to fulfill biomimicry or biocompatibility criteria, depending on the specific application. A meaningful in-depth assessment of the functional features of the immobilized proteins and enzymes is a significant challenge as it requires in situ or in operando investigation of the structure, orientation, conformational changes and dynamics, heterogeneous and intraprotein ET, enzymatic activity, catalytic cycle, and product generation. This can be achieved by hyphenating the electrochemical setup with a variety of spectroscopic methods, such as SERR, SEIRA, PFIRE, PMIRRAS, FRET, XAS, EPR, and DMES. In this perspective article, I have shown the value of these spectroelectrochemical techniques for elucidating complex aspects of ET and of redox-coupled reactions of a variety of immobilized proteins, including small heme and copper electron shuttling proteins, large respiratory complexes, hydrogenases, multicopper oxidases, alcohol dehydrogenases, endonucleases, NO-reductases, and dye decolorizing peroxidases, among others. The different methods provide largely complementary information, thus highlighting the advantages of multispectroscopic detection. In spite of that, most researchers base their studies on a single spectroelectrochemical technique, often due to restricted access to other methods and also due to inherent limitations of each method; not all spectroelectrochemical methods are suitable for all protein/electrode systems. For example, SERR detection is particularly useful for investigating the structure, electronic properties, and orientation of the active site of heme proteins, as attaining simultaneous resonance of the laser probe with the strong Soret absorption band of the heme and with the surface plasmons of Ag nanostructures is relatively simple. Although SERR spectroelectrochemistry is nowadays a well-established technology, there is still plenty of room for improvement in the reproducible nanofabrication of tunable SERR substrates, thus facilitating quantitative analysis and expansion to other redox proteins. IR-based methods, on the other hand, are sensitive to the protein orientation and secondary structure, as well as to conformational details, H-bonding, protonation of the peptide backbone, amino acid side chains, cofactors, and internal water molecules.[117] A drawback is their relatively low instrumental sensitivity, partially due to the low intensity of conventional light sources. The implementation of more intense light sources in routine setups, such as quantum cascade lasers, is a foreseeable development that may greatly expand the applicability of IR spectroelectrochemical techniques.[3] Other technical advancements that can be anticipated include the reproducible and cost-effective nanofabrication of SEIRA-active arrays, the expansion of DMES methods to the palette of available ionization methods, and improvements in the design of cells for magnetic and X-ray based spectroelectrochemical methods. Equally important is the development of advanced mathematical and computational tools for the reliable treatment of the large data sets that characterize stationary and time-resolved spectroelectrochemical experiments, particularly for the identification of unknown intermediate or transient species. Some of the most promising technologies in this respect include machine learning, neural networks, and wavelets.[4] The availability of highly sensitive spectroelectrochemical methods along with more powerful data processing tools will possibly guide the development of more comprehensive theoretical frameworks for the treatment of heterogeneous protein ET kinetics, will foster our fundamental understanding of redox proteins and enzymes and will contribute to the rational design of efficient protein-based technological devices.