3.1.2. Fructose Dehydrogenase

− Although not approaching the degree of detail associated with the simpler ET proteins cyt *c* and azurin, the molecular mechanistic mapping of several flavin dehydrogenases has reached an impressive level of detail over the last few years, represented by fructose dehydrogenase (FDH), CDH, and GDH in particular. The three enzymes display a common pattern, with a FAD catalytic center where fructose, cellobiose, and glucose, respectively, is oxidized, and an ET relays temporarily populated ET sites through which the liberated electrons are transmitted to the electrode surface. As the first FAD enzyme in our overview, *Gluconobacter* sp. FDH is a membrane-bound FAD-dependent oxidoreductase with a molecular mass around 140 kDa [25,30]. The protein holds three subunits: subunit I (67 kDa) contains a FAD cofactor, serving as the catalytic center for the two-electron oxidation of D-fructose to keto-D fructose. Subunit II (51 kDa) has three heme groups with the formal potentials of 0.15, 0.06, and −0.01 V vs. Ag/AgCl electrode (sat. KCl), respectively. Only two heme groups with the relatively lower redox potentials are proposed to participate in DET [30]. Subunit III (20 kDa) plays an important role in maintaining the structural integrity of the enzyme complex. A number of recent studies illustrate the employment of FDH for the development of biosensors and biofuel cells with high current density and operational stability [30,81,101,102].

The three-dimensional crystal structure, especially the enzyme surface properties, is essential for rational tuning of the redox enzyme immobilization. The detailed crystallography of FDH is still unclear, but homology models have helped to provide a clearer picture of intramolecular electron transfer (IET) [103]. Kano and associates constructed FDH variants with glutamine instead of the axial methionine ligands (M301, M450, or M578) of heme 1*c*, heme 2*c*, and heme 3*c*, respectively, illustrating that the ET pathway leads from M578 to M450 bypassing M301 [104]. Heme 1*c* with the highest formal potential of 0.15 V vs. Ag/AgCl electrode was evaluated not to be involved in DET, whereas heme 2*c* was identified as the ET bridge between FDH and the electrode surface. Heme 3*c* with the lowest formal potential of −0.01 V vs. Ag/AgCl (sat. KCl) was suggested as a bridge between FAD and heme 2*c* in the IET process (Figure 4) [30]. In addition, the catalytic current density of FDH was dramatically increased by deleting the amino acid residues on the N- or C-terminus of subunit II [30,105,106]. The deletion not only promotes enzyme loading but also provides more opportunities for favorable orientations on the electrode surface. Some researchers reported that hydrophobic anthracene groups anchored on single-walled carbon nanotubes are favorable for enhanced catalytic activity and stability via the hydrophobic C-terminal region of subunit II [101,107]. These studies suggest that orientation and enzyme loading are crucial for controlling the catalytic activity in DET-type bioelectrocatalysis. −

**Figure 4.** Proposed ET pathway from D-fructose to the electrode surface in the DET of FDH. The ET route involves FAD, heme 3*c*, heme 2*c*, but not heme 1*c*. Reproduced with permission from [30]. Copyright 2019, Elsevier.

Favorable FDH SAM immobilization rests on the following consideration: the isoelectric point (IEP) of FDH is 6.59, which means that FDH is overall positively charged in slightly acidic electrolyte [81]. Recent studies to improve the catalytic performance of FDH on various SAM-modified electrodes have been reported [25,41,44,81]. Bollella and associates reported extensive research on the catalytic activity of FDH on highly porous gold (h-PG) electrodes modified with 4-mercaptobenzoic acid (4-MBA), 4-mercaptophenol (4-MPh), and 4-aminothiophenol (4-APh) SAMs. The data showed that high bioelectrocatalytic activity and stability of FDH was only observed with -OH terminated SAMs [81], suggesting optimized enzyme orientation on this particular SAM. Negatively charged SAMs may help to favor FDH coverage due to preferred electrostatic interaction, but this is not necessarily the most favorable FDH orientation for DET. Murata and associates reached similar conclusions for FDH immobilized on 2-mercaptoethanol (MET) SAM-modified AuNPs [41]. Considering the importance of the gold nanostructure for the ET rate, the AuNP size is furthermore a crucial parameter. Kizling and associates synthesized 1.0 to 3.5 nm AuNP clusters functionalized with 1,6-hexanedithiol and 1-butanethiol for investigating the ET mechanism. A channel of "mediated" catalysis, i.e., electron "hopping", was observed for the smallest AuNP clusters around 1 nm with the half-wave potential close to the first oxidation potential of the AuNP at the edge of the HOMO-LUMO gap [44]. Such a mode accords with theoretical notions recently reported [92]. Siepenkoetter and associates reported the catalytic performance of covalently bonded FDH on NPG electrodes with varying pore sizes [25]. A large number of findings thus show that FDH has high affinity for well-defined SAM-modified electrodes with the polar but electrostatically neutral hydroxyl terminal group as the most efficient,

indicating the importance of hydrophilicity of the electrode surface towards the mixed surface charge distribution of the FDH target enzyme. However, details of the underlying mechanism remain unknown.3.1.3. Cellobiose Dehydrogenase

The second FAD enzyme, cellobiose dehydrogenase (CDH), is a versatile oxidoreductase for direct bioelectrocatalysis. CDH contains a catalytic dehydrogenase domain (DH) harboring a FAD as the redox center and an ET cytochrome *b* (CYT) domain to shuttle electrons from the FAD to the electrode surface [23,45,108]. The two domains are separated in the crystalline structure, but an integrated IET pathway can be opened through a flexible and hydrophilic amino acid linker between the catalytic and CYT domains [80]. This motif is encountered also for the molybdenum sulfite oxidases, cf. Section 3.1.4. CDH is extracted from the phyla *Basidiomycota* and *Ascomycota*, divided into class I, class II, and class III, respectively. Class III CDH from *Ascomycota* remains uncertain compared to class I and class II CDH [23].

The natural substrates of CDH mainly include cellulose, lactose, and glucose [108]. Class I CDH with short amino acid sequences shows direct, strongly pH-dependent catalytic activity only in solutions with pH below 5.5 [23]. Harreither and associates reported extensive studies on class II CDH from *Chaetomium attrobrunneum (Ca*CDH*)*, *Corynascus thermophiles (Ct*CDH*)*, *Dichomera saubinetii (Ds*CDH*)*, *Hypoxylon haematostroma (Hh*CDH*)*, *Neurospora crassa (Nc*CDH*)*, and *Stachybotrys bisbyi (Sb*CDH*)* [109]. pH-dependent catalytic activity was observed for neutral and slightly alkaline electrolytes with cyt *c* and 2,6-dichloroindophenol (DCIP) as electron acceptors. The different electrostatic environment in class I and II CDH reveals different optimal IET processes, with optimum IET in acidic electrolyte for class I CDH and neutral or slightly alkaline electrolyte for class II CDH [110]. Schulz and associates demonstrated turnover and non-turnover DET performance of CDH on polycrystalline gold modified with MUO or 6-MHO [23]. A clear catalytic current between the FAD and the electrode surface was reported. The midpoint potential of bound FAD cofactor was −163 mV vs. SCE at pH 3.0, approximately 130 mV less than that of the heme *b* relay, leading to a much lower onset potential for lactose oxidation. The authors concluded that the tunneling distance between the FAD cofactor and the electrode was around 12–15 Å, thereby allowing direct electrochemical communication. The authors also demonstrated that only class I CDH from *Trametes villosa (Tv*CDH*)* and *Phanerochaete sordida (Ps*CDH*)* displayed DET activity, whereas no DET signal was observed for class II CDH from reconstructed *Myriococcum thermophilum (recMt*CDH*)* and reconstructed *Corynascus thermophilus (recCt*CDH*)* at low pH.

The IEP of CDH (DH domain ~5, CYT domain ~3) gives a negatively charged surface around the ET path exit, indicating that positively charged SAMs are suitable for favorable DET [110]. Lamberg and associates anchored *Humicola insolens* CDH on various SAMs with different terminal groups for investigating the effects of charge and hydrophobicity [111] and found that hydrophilic SAMs were favorable for high catalytic activities, with lower enzyme orientation variations than for hydrophobic SAMs. Tavahodi and associates reported a DET-type lactose biosensor of *Ps*CDH on polyethyleneimine (PEI)-coated AuNP electrodes (PEI@AuNP) [112]. PEI with positively charged amino groups not only optimized the orientation of *Ps*CDH on the electrode to increase the IET rate but also gave high enzyme stability and sensitivity. Bollella and associates reported a lactose biosensor based on DET of *Ct*CDH on gold electrodes [113]. The authors demonstrated that BPDT SAMs with two thiol groups can be used to anchor metal NPs on a gold electrode surface by covalent bonding, showing the best ET rate on AuNPs/BPDT/Au electrode. A mediator-free *Hi*CDH/*Mv*BOD BFC using a positively charged MHP SAM for immobilization on AuNP-modified gold electrodes was also reported [114]. The half-life time, *i.e.,* the time duration over which the activity decreased to half the initial value, was 13 and 44 h with 5 mM glucose and 10 mM lactose in neutral buffer solution, respectively. Hirotoshi and associates found an efficient ET process of *Phanerochaete chrysosporium (Pc*CDH*)* on a mixed 11-AUT/MUO SAM-modified AuNP electrode [46]. Al-Lolage and coworkers addressed the catalytic performance and stability of *Mt*CDH by introducing cysteine mutants (E522 and T701) on the protein surface (Figure 5) [45]. The cysteine mutant with a surface thiol group made it possible to form stable thioether bonds with maleimide groups via click-chemistry, thereby controlling the orientations of the *Mt*CDH mutant on the

electrode surface and revealing efficient electrochemical glucose oxidation. Another example reported by Meneghello and associates clearly indicated that the DET-type bioelectrocatalysis of CDH is highly sensitive to the cysteine residues introduced at particular positions [115]. Experimental results also showed that divalent cations (*i.e.,* Ca2<sup>+</sup> and Mg2+) affect the IET kinetics [23,115].

**Figure 5.** (**a**) Structure of *Mt*CDH with cysteine E522 and T701 mutations shown in blue and green, respectively; FAD and the heme group are highlighted in yellow and red, respectively. The flexible chain linking the two domains is in blue; (**b**) the maleimide group anchored on an electrode surface can react with the mutant cysteine residue in the enzyme. Reproduced with permission from [45]. Copyright 2017, John Wiley and Sons.

#### 3.1.3. FAD-Dependent Glucose Dehydrogenase

Our final FAD enzyme target, glucose dehydrogenase (FAD-GDH), is one of the most widely known dehydrogenases with a tightly bound cofactor, making it different from NAD+-dependent GDH. FAD-GDH has been extensively studied as an emerging alternative to glucose oxidase (GOx) due to its favorable DET capability, insensitivity to dioxygen, and the fact that no hydrogen peroxide is generated [116–119]. The structure of GDH is analogous to that of FDH, comprising three subunits: an FAD-dependent catalytic subunit, an ET subunit with three heme groups, and a small "hitch-hiker" protein used for the flexibility of the catalytic subunit into the periplasm [116,120]. The catalytic subunit harbors a 3Fe-4S cluster close to the ET subunit protein surface, allowing efficient DET on the electrode without the need for mediators. Lee and associates demonstrated controlled DET of bacterial FAD-GDH from *Burkholderia cepacia* on three different SAM-modified electrodes, on which catalytic current density decreases with increasing SAM chain length [118]. A glucose biosensor based on FAD-GDH was reported recently by introducing a gold-binding peptide (GBP) for enzyme immobilization on a screen-printed electrode (SPE) [24]. GBP composed of 12 amino acids exhibiting a strong binding affinity to the gold electrode surface was fused to the enzyme terminus, thereby determining the enzyme orientation on the electrode. Around 10 times higher catalytic response toward 100 mM glucose was observed with FAD-GDH-GBP/Au compared with normal GDH/Au.

A DET-type glucose biosensor could also be fabricated coupled with electrochemical impedance spectroscopy (EIS) [120]. Three variable-length thiols, dithiobis(succinimidyl hexanoate) (DSH), dithiobis(succinimidyl octanoate) (DSO), and dithiobis(succinimidyl undecanoate) (DSU), were employed to modify the electrode surface. Charge transfer resistance (Rct) was a key parameter reflecting the DET efficiency, with the lowest resistance when FAD-GDH was immobilized on DSH SAMs because of the shortest tunneling distance. In addition, the steady-state catalytic current density of FAD-GDH was dramatically increased on AuNPs assembled on the gold electrode [73,121]. Ratautas and associates reported high glucose oxidation activities of FAD-GDH extracted from

*Ewingella Americana* without mediator [73,121]. They demonstrated that FAD-GDH immobilized on 4-ATP-modified AuNPs displayed higher catalytic activity and lower overpotential than on 4-MBA-modified electrodes. Notably, 4-ATP can be oxidized in neutral media and further converted to 4-mercapto-N-phenylquinone monoimine (MPQM), the quinone groups of which could bind covalently with primary amino groups of the enzyme, thereby enhancing the ET rate. The catalytic response was found to be positively correlated to the ratio of 4-ATP/4-MBA for electrodes modified by mixed thiol SAMs [121]. PQQ-dependent GDH is still another promising GOx candidate, widely utilized for glucose biosensors and biofuel cells [35,95,96]. Kim and coworkers prepared a glucose biosensor by immobilizing positively charged PQQ-GDH on MUA SAM-modified gold electrodes [122]. In this case, PQQ-GDH exhibited higher current density and detection sensitivity via electrostatic adsorption than via covalent bonding, due to less enzyme inactivation. Covalent bonding could thus enhance the interactions between enzyme and electrode surface, but the surface enzyme characteristics are the more critical factors that determine the electrochemical behavior.

#### 3.1.4. Sulfite Oxidase

Sulfite oxidase (SOx) is another heme-containing redox enzyme which has been studied for a long time. Two kinds of SOx are mainly employed in DET-type bioelectrocatalysis: chicken liver, *c*SOx [123], and human, *h*SOx [124]. SOx acts by a principle resembling that of the FAD enzymes, with a catalytic center and an electronic relay, but also with some important differences. As a metalloprotein, the catalytic domain is composed of a pyranopterin molybdenum (Mo) cofactor effecting two-electron oxidation of sulfite to sulfate and connected to a N-terminal cytochrome *b*<sup>5</sup> (cyt *b*5) domain by a flexible linker [123,125]. Although the crystallographic structure of *c*SOx shows that the distance between the catalytic Mo domain and the cyt *b*<sup>5</sup> domain is more than 32 Å, rapid internal ET in DET-type bioelectrocatalysis is still observed [126], effected by a conformational change on the electrode surface via the flexible tether. This "on-off" switch enables cyt *b*<sup>5</sup> to be either adjacent or remote from the Mo cofactor [70,127] in a gated ET mode, which is quite different from the rigidly bound ET relays in the FDH and GDH enzymes, but similar to the CDH operational mode.

The electrostatic surface charge distributions around both domains are complex and highlight the importance of subtle tuning of the electrode surfaces [124]. A positively charged SAM surface is favorable for immobilization of *h*SOx via the Mo domain, directing the smaller heme domain towards the electrode surface. As noted, a similar pattern with a flexible linker connecting the catalytic and ET domain applies to CDH [80]; cf. Section 3.1.3. Sezer and associates investigated the catalytic activity of *h*SOx on a mixed SAM-modified silver electrode, showing a significantly increased SERR signal when increasing the ionic strength. High ionic strength is favored to shorten the distance between the Mo cofactor and the heme domain, thereby facilitating both intramolecular and interfacial ET. Wollenberger and associates reported other studies of the electrochemical behavior of *h*SOx [127–130]. In situ scanning tunneling microscopy and spectroscopy to single-molecule resolution has, finally been reported quite recently and disclosed intriguing patterns of tunneling via the Mo- and heme group redox centers [130]. AuNPs with a diameter less than 10 nm were covalently bonded on the MUA/MUO SAM-modified gold electrode, giving a significant enhancement of the interfacial ET rate of *h*SOx [129]. The interfacial ET rate could be further increased by using SAMs with 3,3′ -dithiodipropionic acid di(N-hydroxysuccinimide ester) (DTSP) thinner than MUA/MUO SAMs [128]. They also reported that the introduction of BaSO<sup>4</sup> nanoparticles played a significant role in the DET-type bioelectrocatalysis of *h*SOx. The electrochemical communication between the active sites of *h*SOx and the electrode surface could be further enhanced by using a positively charged biopolymer. Kalimuthu and coworkers thus reported direct catalytic activity of *h*SOx on a chitosan-covered gold electrode [131]. Both non-catalytic redox signals corresponding to the heme group and catalytic signals in the presence of 4 mM sulfite, on chitosan-covered MPA-, MSA-, and 4-MBA-modified electrodes were observed. However, there were no catalytic signals on MUA-based electrodes despite an observed non-turnover feature, highlighting the importance of rational SAM selection.
