Discussion on Direct Electron Transfer-Type Bioelectrocatalysis of Downsized and Axial-Ligand Exchanged Variants of d-Fructose Dehydrogenase

D-Fructose dehydrogenase (FDH) gives a clear direct electron transfer (DET)-type bioelectrocatalytic wave even at planar gold (Au) electrodes. The recombinant (native) FDH (r_FDH) has three hemes c in subunit II (1c, 2c, and 3c from N-terminus). With a view to downsize the enzyme and shorten the distance between an electrode-active site and an electrode, we constructed a variant that lacked 143 amino acid residues involving the heme 1c moiety (Δ1cFDH) and a variant that lacked 199 amino acid residues involving the heme 1c and 2c moieties (Δ1c2cFDH). In order to shift the redox potential of heme 2c of Δ1cFDH to the negative direction, the M450 residue as the axial 6th ligand of heme 2c was also replaced with glutamine (M450QΔ1cFDH). The DET-type catalytic properties of r_FDH and the three variants at planar Au electrodes were compared with each other, and the steady-state waves were analyzed on a random orientation model. The orientation of the enzymes on the electrode was also discussed. In addition, in order to examine the electron transfer pathway in the DETtype reaction of Δ1c2cFDH, ESR measurements and inhibition of DET-type reaction by cyanide ion were performed. © The Author(s) 2020. Published by ECSJ. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium provided the original work is properly cited. [DOI: 10.5796/electrochemistry.20-00029]. Uploading "PDF file created by publishers" to institutional repositories or public websites is not permitted by the copyright


Introduction
Bioelectrocatalysis, in which the electrode reaction and the catalytic reaction of a redox enzyme are coupled with each other, have attracted increasing attention from viewpoints of environment, energy, and health thanks to enzymatic properties of highly selective and active catalysts under the mild conditions. There are two types of bioelectrocatalysis: direct electron transfer (DET) and mediated electron transfer (MET) types. Especially, DET-type bioelectrocatalysis, in which the electrode and enzymatic reactions are directly coupled, plays an significant role in the construction of mediatorfree and simple bioelectrochemical-devices for biosensing (biosensors) and biochemical electricity production (biofuel cells) with minimum overpotential in theory. [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] In order to improve or investigate the DET-type bioelectrocatalysis, several protein-engineering methods have been examined: the point mutation around the active site to change its catalytic characteristics, [17][18][19][20] the deglycosylation to shorten the distance between an electrode and the redox center buried in the enzyme and improve the interfacial electron transfer kinetics, 18,21,22 and the insertion of a tag sequence or cysteine residue(s) to control the orientation of the redox enzyme on gold (Au) electrodes. [23][24][25] D-Fructose dehydrogenase from Gluconobacter japonicus NBRC3260 (FDH; EC 1.1.99.11) is a heterotrimeric membrane protein consisting of subunits I (67 kDa), II (51 kDa), and III (20 kDa). Subunit I contains a covalently bound flavin adenine dinucleotide (FAD). Subunit II comprises three heme c moieties that we call hemes 1c, 2c, and 3c from its N-terminus. 26,27 Subunit III plays an important role in the expression of FDH. 27 This enzyme facilitate a 2-electron oxidation of D-fructose to 5-keto-D-fructose. The physiological electron acceptors in living bodies are ubiquinones in cell membranes (UQs, Fig. S1). 27,28 The activity of the DET-type bioelectrocatalysis of FDH is quite high. 29 In previous works by our group, it has been proposed that electrons are transferred from the substrate, to FAD, heme 3c, heme 2c, and an electrode, and that heme 1c is not involved in the electron transfer of the DET-type bioelectrocatalysis; heme 2c is the electrode-active site in the DET-type bioelectrocatalytic reaction of FDH. 17 In contrast, however, heme 1c is the electron-donating site to UQs in the living cell and its presence is essential to complete the respiratory chain ( Fig. S1), since the cells expressing an FDH variant in which 143 amino acid residues involving the heme 1c binding site was largely deleted (¦1cFDH (Fig. S2(A))) showed no respiration activity. 30 Interestingly, ¦1cFDH modified electrode showed higher limiting current density of the DET-type bioelectrocatalysis. 30 We considered that the increment of the limiting current density was presumably due to the ¦1c-downsizing of FDH, which increases the surface concentration of ¦1cFDH.
We also constructed an FDH variant that lacked 199 amino acid residues including the heme 1c and 2c moieties (¦1c2cFDH, Fig. S2(B)) in order to increase the surface concentrations due to the ¦1c2c-downsizing and realize a short-cut electron transfer from heme 3c to electrodes directly. 31 According to the above two expectations, the variant may lead to an increase in the limiting current density and the decrease in the overpotential, respectively, in the DET-type bioelectrocatalytic oxidation of D-fructose. The expected short-cut electron transfer from heme 3c to a planar Au electrode seems to be realized, 31 though further study will be required to verify the proposed electron transfer pathway. However, contrast to our expectation, its limiting catalytic current density was lower than that of ¦1cFDH. 31 On the other hand, another ¦1cFDH variant that had glutamine in place of methionine as the native 6th axial ligand of heme 2c at the 450th amino acid residue (M450Q¦1cFDH (Fig. S2(A))) showed a drastic decrease of the overpotential in the oxidation of fructose due to a negative shift in the formal potential of heme 2c as the electron-donating site of the variant. 32 In this paper, we believe that it is very important to discuss unified effects of the mutations on the DET-type bioelectrocatalysis by re-examination of the DET-type bioelectrocatalysis of the recombinant (native) FDH (r_FDH) and the three variants: ¦1cFDH, ¦1c2cFDH, and M450Q¦1cFDH. Planar Au electrodes were intentionally used as scaffolds of the enzymes in order to eliminate (or minimize) the effects of the roughness factors of the electrode surface. Data obtained by electrochemical measurements at each enzyme-adsorbed Au electrode were analyzed based on a random orientation model and the obtained parameters were compared with each other. The orientation of the variants on the electrode was also discussed. In addition, in order to examine the pathway of electrons in the DET-type reaction of ¦1c2cFDH, electron spin resonance (ESR) measurements and inhibition of DET-type reaction by cyanide ion were performed.

Materials
DNA ligase, Herculase II fusion DNA polymerase, and restriction endnucleases were purchased from Toyobo (Japan), Agilent Technologies (Santa Clara, CA), and Takara Shuzo (Japan), respectively. Other chemicals were obtained from Wako Pure Chemical Industries (Japan).

Electrochemical measurements
The working electrodes were planar Au disk electrodes (3-mm diameter, BAS Inc., Japan), which were polished with Al 2 O 3 powder (0.05-µm particle size) to a mirror-like finish, rinsed with distilled water, and sonicated in distilled water. The reference electrode was handmade Ag « AgCl « sat.KCl and the working electrode was Pt wire. A 3 µL aliquot of each enzyme solution was added to a McB solution (pH 4.5) in an electrochemical cell for measurements of bioelectrocatalytic currents (L = dm 3 ). As a result, the final concentration of Triton X-100 and 2-mercaptoethanol were 3 ppm (w/v) and 6 µM, respectively. In this paper, all the potentials are referred to the reference electrode and all electrochemical measurements were performed under anaerobic conditions. Cyclic voltammograms were recorded by using an ALS1000 electrochemical analyzer at 25°C. In order to record the rotating disk voltammetry, a rotating disk electrode (RDE-1, BAS Inc., Japan) was employed.

Electron spin resonance spectroscopy
The ¦1c2cFDH solution was concentrated to 20 µM for the measurements. The sample solution was placed into a glass capillary cell with an inner diameter of 0.5 mm. ESR spectra were recorded on a JES-FA100 spectrometer (JEOL, Japan). The microwave power was set to 2 mW.

Other analytical methods
The enzyme activity was spectrophotometrically measured using potassium ferricyanide (as an electron acceptor) and the ferric dupanol reagent, as described in the literature. 26
Usually, mesoporous electrodes are very useful (or sometime essential) to observe DET-type bioelectrocatalytic waves in order to increase the possibility of the productive orientation of redox enzymes by curvature effects. 34 However, as described in our previous papers, 30-32 clear steady-state waves of D-fructose oxidation were observed at planar Au electrodes on which r_FDH, ¦1cFDH, M450Q¦1cFDH, or ¦1c2cFDH was adsorbed ( Fig. S3(A)). Since the scan rate (from 1 to 50 mV s ¹1 ) and the rotating speed (from 0 to 4000 rpm) did not affect the catalytic waves, the catalytic currents were independent of the mass transfer of the substrate, and were controlled by the interfacial electron transfer kinetics or the enzyme kinetics. [35][36][37][38][39] The steady-state catalytic current density ( j s ) under such conditions for enzymes with single orientation is given by: 40 where k c is the catalytic rate constant of the enzyme reaction, but is not necessarily proportional to k cat (sol). The parameters k f and k b correspond to the rate constants of the forward (anodic) and the reverse (cathodic) electrode reactions, respectively (Eqs. (S2) and (S3)). j lim s is the limiting steady-state current density (Eq. (S3)). Detailed description on Eq. (1) is given in Appendix S1. Equation (1) predicts sigmoidal waves with a limiting current region giving j lim s . Such ideal sigmoidal waves were observed at the M450Q¦1cFDH-and ¦1c2cFDH-adsorbed Au electrodes ( Fig. S3(A)). The exponentially increasing part on the waves was clearly observed, indicating fast electron transfer between each enzyme and the electrode. It can be considered that M450Q¦1cFDH and ¦1c2cFDH adsorbed in rather ordered orientations on the planar Au electrode and that the orientation is suitable for the DET-type bioelectrocatalysis. The ¦1c2c-downsizing may induce an attractive interaction between the electrode and the electrode-active site (heme 3c as described later). Since the orientation of M450Q¦1cFDH is more suitable to DET than ¦1cFDH, the M450Q mutation might cause some change in the enzyme surface conditions near the heme c moiety.
In contrast, the r_FDH-and ¦1cFDH-adsorbed electrodes gave a linearly increasing region on the catalytic wave ( Fig. S3(A)). This is called residual slope 35,36 and is ascribed to the random orientation of enzymes with an eccentrically located electrode-active site (Appendix S1). j lim s values were not clearly defined in these cases within the potential window of the measurements. At potentials more positive than 0.5 V, the DET-type reaction is inhibited by gold oxide formed at such high potentials. 17,39 It can be considered that r_FDH and ¦1cFDH adsorbed randomly on the planar Au electrode. 31,32 Unfortunately, we could not find clear explanation on a question why the M450Q mutation could improve the orientation. However, similar improvement in the catalytic wave was observed by the M450Q mutation in r_FDH. 17 The observed CVs were normalized against the current density at 0.5 V ( j 0.5 V ). The normalized waves at the r_FDH-and ¦1cFDHadsorbed electrodes were almost identical with each other (Fig. S3(B)). This is the evidence that the electrode-active site of Electrochemistry, (in press) r_FDH is identical with that of ¦1cFDH (that is, heme 2c), as reported in previous papers. 17,30 The half-wave potential of the normalized CVs at the M450Q¦1cFDH-adsorbed electrode was approximately 0.08 V and were 0.19 V more negative than that of the wave of the ¦1cFDH-adsorbed electrode. 32 This negative shift in the half-wave potential of the catalytic wave at the M450Q¦1cFDH-adsorbed electrode is ascribed to the negative shift in the formal potential of the electrode-active redox center (heme 2c) in the enzyme (E o0 E ) by replacing methionine as the native 6th axial ligand of heme 2c at the 450th amino acid residue with glutamine with an electro-donating property. 17,32 The result also supports that heme 2c is the electrodeactive site of M450Q¦1cFDH.
Notably, the half-wave potential of the normalized CVs at the ¦1c2cFDH-adsorbed electrode was almost identical with that at the M450Q¦1cFDH-adsorbed electrode. The FAD-containing subunit without the heme-containing subunit loses the DET-type activity, 41 and then the most possible candidate of the electrode-active site in ¦1c2cFDH is heme 3c with the most negative formal potential among the three hemes c in r_FDH. 42 In order to clarify the interfacial electron transfer pathway of ¦1c2cFDH on electrodes, ESR measurements were carried out under D-fructose-reducing conditions. The FAD in the variant is 2electron reduced by the hydride transfer from D-fructose. The twoelectron-reduced FAD is ESR-silent, but the substrate-reduced ¦1c2cFDH yielded a strong isotropic ESR signal at g µ 2 ( Fig. 1). This signal is assigned to its FAD semiquinone radical in the enzyme, obviously indicating that one of the two electrons in the reduced FAD is transferred to heme 3c to yield the semiquinone radical in the enzyme. 31 The result supports our proposed pathway of the electron transfer from heme 3c in ¦1c2cFDH to electrodes.
Moreover, potassium cyanide (KCN) was used to shift the formal potential of heme 3c and to verify the pathway of the electron transfer in the DET-type bioelectrocatalysis of the ¦1c2cFDHadsorbed electrode. CN ¹ is coordinated to the iron atom of heme 3c at the axial position. Such a coordination of CN ¹ on heme c causes a shift of the formal potential of the heme by approximately 0.4 V to the negative potential direction. 43 The addition of KCN to the electrochemical reaction buffer at a final concentration of 1 mM decreased the catalytic current density drastically (Fig. 2). We considered that the drastic decrement of the catalytic current is ascribed to an uphill intramolecular electron transfer from the substrate-reduced FAD to the CN ¹ -coordinated heme 3c; the electron transfer became difficult due to the negative shift in the formal potential of heme 3c. The result also supports our proposed pathway in the electron transfer; heme 3c is the electrode-active site of ¦1c2cFDH. 31 However, one question arises why heme 3c in r_FDH, ¦1cFDH, and M450Q¦1cFDH does not work as an electrode-active site. In these variants, the intramolecular electron transfer from heme 3c to heme 2c might be much faster than a conceivable direct and heterogeneous electron transfer from heme 3c to an electrode. The peptide chain involving the heme 2c moiety that is deleted in the ¦2c-downsizing might inhibit the interfacial electron transfer from heme 3c to an electrode.
When we focused on the j 0.5 V value, M450Q¦1cFDH gave the largest value among them (Fig. S3(A)). ¦1cFDH also gave rather large value of j 0.5 V . Since j 0.5 V is considered to be close to j lim s , the increase in j 0.5 V is ascribed to an increase in the surface concentration of the enzyme (! E ), as expected from Eq. (S4) by assuming that k c remained almost unchanged by the mutations.
Comparing the values of j 0.5 V of M450Q¦1cFDH-and ¦1cFDHadsorbed electrode to those of other mutants adsorbed electrode, the ¦1c-downsizing is effective to increase ! E . As described before, M450Q¦1cFDH adsorbed rather orderly, while ¦1cFDH adsorbed randomly on the electrode. Therefore, the ¦1c-downsizing effect is more evident for M450Q¦1cFDH compared with that ¦1cFDH.
Contrary to our expectation, the j 0.5 V value at the ¦1c2cFDHadsorbed electrode was smaller than that of ¦1cFDH (and M450Q¦1cFDH) and almost same as that at the r_FDH-adsorbed electrode (Fig. S3(A)). The ¦1c2c-downsizing caused a decrease in the surface concentration of the enzyme (! E ). This may probably due to a decrease in the hydrophobic property of subunit II by the ¦1c2c-downsizing. Such change in the hydrophobic property seems to be very significant in FDH enzymes, since Triton X-100 is essential to solubilize the enzymes. Triton X-100 may competitively prevent the adsorption of ¦1c2cFDH to some extent. The hydrophobicity of enzyme seems to play an important role in the adsorption of such membrane-bound enzymes.
All explanations described above are qualitative. In order to discuss quantitatively, the recorded voltammograms were analyzed on the basis of the random orientation model for the DET-type bioelectrocatalysis. A steady-state catalytic current without the concentration polarization of the substrate was used, and Eq. (S7) 35,37 was fitted to the steady-state waves of the forward scan at the ¦1c2cFDHand M450Q¦1cFDH-adsorbed electrodes using nonlinear regression analysis by Gnuplot μ with k°m ax /k c , ¢¦d, and E°B E as adjustable parameters by setting j lim s ¼ j 0:5 V and the transfer coefficient (A) = 0.5 ( Fig. S4(C, D)), where k°m ax is the standard rate constant at the closest approach when the enzyme is most suitable orientation for the electron transfer reaction, ¦d is the difference of the distance of electrode-active center from the electrode surface between the closest and farthest approach of the Electrochemistry, (in press) enzyme, and ¢ is the coefficient in the long range electron transfer the transfer coefficient. The physical meanings of the other parameters are given in Appendix S1. The catalytic current of the forward scan at the r_FDH-and ¦1cFDH-adsorbed electrodes were also analyzed in the same manner ( Fig. S4(A, B)), but k c ! was also employed as an adjustable parameter, because clear j lim s could not be determined by the voltammogram.
The evaluated values of the fitting parameters of each enzymeadsorbed electrode are summarized in Table 1. The E°B E value for M450Q¦1cFDH was evaluated to be 18 « 1 mV, which is more negative than those of the r_FDH (54 « 1 mV) 31 and ¦1cFDH (34 « 1 mV). 31 These data indicate that M450Q¦1cFDH can transfer the electrons at a more negative potential than r_FDH due to its negatively shifted electron donating site, heme 2c, by the M450Q mutation. The E°B E values evaluated for ¦1c2cFDH (20 « 1 mV) 31 was also more negative than those of the r_FDH and ¦1cFDH and is in good agreement with the potential of the noncatalytic wave measured at the r_FDH-immobilized glassy-carbon electrode on which anthracene-modified single-walled carbon nanotubes were deposited. 45 This result supports that heme 3c is the electrode-active site in ¦1c2cFDH, while heme 2c in r_FDH and ¦1cFDH variants. 31 The E°B E value for r_FDH was almost the same as the value for heme 2c of r_FDH determined by a mediated spectroelectrochemical titration (60 « 8 mV at pH 5.0), 42 while those values of ¦1c2cFDH and ¦1cFDH were slightly more positive than that of heme 3c spectroelectrochemically determined (¹10 « 4 mV). In these variants, some change in the environment (especially in water accessibility) around the heme c moieties may occur by the mutation.
The evaluated values of k c ! support the above discussion; the ¦1c-downsizing is effective to increase !, while ¦1c2c-downsizing causes a decrease in !. The k°m ax /k c values of the M450Q¦1cFDHand ¦1c2cFDH-adsorbed electrodes were notably larger than those of the r_FDH-and ¦1cFDH-adsorbed electrodes. This situation is related to our discussions that M450Q¦1cFDH and ¦1c2cFDH adsorbed on the planar Au electrode rather homogeneously in orientations suitable for the DET-type reaction. The distance between the electrode and the electrode-active site in M450Q¦1cFDH (heme 2c) and ¦1c2cFDH (heme 3c) was shortened by the ¦1cand ¦1c2c-downsizings, respectively, and thus k°m ax increased. Figure 3 shows predicted orientations suitable for the DET-type reaction of r_FDH, ¦1cFDH, and ¦1c2cFDH, in which the orientations were set so as to minimize the distance between the electrode-active site and the electrode. The heme 3c moiety in ¦1c2cFDH is located relatively near the surface of the enzyme. 31 Therefore, the orientation of the variant in Fig. 3(C) may be the most possible one to realize the fast heterogeneous electron transfer. The occupied area of the variant on the surface is small in this orientation. However, the adsorption affinity of the variant (to compete Triton X-100) does not seem to be enough in this orientation. This may responsible for a decrease in ! for ¦1c2cFDH.
r_FDH and ¦1cFDH adsorb in random orientation. Therefore, Fig. 3(A) and (B) are one of the possible orientations for r_FDH and ¦1cFDH. However, M450Q¦1cFDH adsorbs rather ordered orientation suitable for the DET-type reaction, as described above. The orientation in Fig. 3(B) may be a possible one to minimize the distance between heme 2c and the electrode. However, the distance does not seem to be sufficiently short to realize the fast heterogeneous electron transfer. Some other suitable orientation or some conformational change by the M450Q mutation may occur.
FDH can do MET-type bioelectrocatalytic reaction in the presence of a suitable mediator. In the MET-type reaction, the mediator can shuttle electrons between the enzyme and an electrode. The movement of the mediator near the electrode surface may affected by the orientation of the adsorbed enzyme. Therefore, we measured the MET-type bioelectrocatalytic wave under the given conditions of the DET-type reaction of the enzymes. On the addition of ferrocene dimethanol as a mediator, the MET-type catalytic currents were added on the DET-type catalytic currents, as shown in Fig. S5. The concentrations of the mediator were set to 2 µM and 4 µM to maintain a linear response of the MET-type catalytic current to the concentration of the mediator. Unfortunately, the MET-type  Figure 3. Schematic of predicted orientations suitable for DET-type reaction of (A) r_FDH, (B) ¦1cFDH, and (C) ¦1c2cFDH. The structures of subunit I (green) and subunit II (cyan) were prepared by FAD-glucose dehydrogenase from Aspergillus flavus (PDB 4YNT) and thiosulfate dehydrogenase from Marichromatium purpuratum (PDB 5LO9) as templates, respectively, in the homology modeling. 46 Since the structural information of similar proteins is not obtained, the subunit III was not shown in the structures.
Electrochemistry, (in press) catalytic current was almost the same for all of the four FDH variants. This was contrary to our expectation that the MET-type catalytic current for ¦1c2cFDH and M450Q¦1cFDH would be smaller than those of the other FDH variants due to the suitable orientation at a decreased distance between the electrode-active site and the electrode. Change in ¡-orders of the distance obtained by the fitting might not affect the actual movement of the mediator. The stability of each FDH variant was also investigated by measuring j 0.5 V values every hour, and the values were normalized against j 0.5 V at the start of the measurement (t = 0 h) in the presence of D-fructose (Fig. S6). Unfortunately, the ¦1cand ¦1c2c-downsizings caused a decrease in the stabilization. However, the M450Q mutation seems to increase the stability as judged from the comparison between M450Q¦1cFDH and ¦1cFDH. The reason was not clear.

Conclusions
We re-examined the properties of DET-type bioelectrocatalysis of r_FDH and the effects of the ¦1cand ¦1c2c-downsizings and the M450Q mutation by comparison of the DET-type bioelectrocatalytic waves and their wave analyses. Heme 3c plays as the electrodeactive site in the DET-type bioelectrocatalysis of the ¦1c2cFDH variant, while heme 2c is the electrode-active site in r_FDH, ¦1cFDH, and M450Q¦1cFDH. The electron transfer pathway of ¦1c2cFDH was verified by the ESR measurements and the KCN inhibition experiment. As a result, the ¦1c2c-downsizing successfully decreased the overpotential in the bioelectrocatalytic oxidation of fructose. The ¦1c2c-downsizing was also useful to improve the orientations suitable for the DET reaction and then to improve the heterogeneous electron transfer kinetics. The positive effects were comparable with those observed in the M450Q mutation in ¦1cFDH. On the other hand, the ¦1c-downsizing was effective to increase the surface concentration of the enzyme, while ¦1c2cdownsizing caused a decrease in the surface concentration. The decrease seems to be ascribed to a decrease in the hydrophobicity of the membrane-bound enzyme in the presence of Triton X-100. Most of qualitative interpretations of the catalytic waves were supported by solid analysis of the waves. Additional mutations might be required in subunit II to increase the hydrophobicity around heme 3c in ¦1c2cFDH.

Supporting Information
The Supporting Information is available on the website at DOI: https://doi.org/10.5796/electrochemistry.20-00029.