Redox‐Polymer‐Wired [NiFeSe] Hydrogenase Variants with Enhanced O2 Stability for Triple‐Protected High‐Current‐Density H2‐Oxidation Bioanodes

Abstract Variants of the highly active [NiFeSe] hydrogenase from D. vulgaris Hildenborough that exhibit enhanced O2 tolerance were used as H2‐oxidation catalysts in H2/O2 biofuel cells. Two [NiFeSe] variants were electrically wired by means of low‐potential viologen‐modified redox polymers and evaluated with respect to H2‐oxidation and stability against O2 in the immobilized state. The two variants showed maximum current densities of (450±84) μA cm−2 for G491A and (476±172) μA cm−2 for variant G941S on glassy carbon electrodes and a higher O2 tolerance than the wild type. In addition, the polymer protected the enzyme from O2 damage and high‐potential inactivation, establishing a triple protection for the bioanode. The use of gas‐diffusion bioanodes provided current densities for H2‐oxidation of up to 6.3 mA cm−2. Combination of the gas‐diffusion bioanode with a bilirubin oxidase‐based gas‐diffusion O2‐reducing biocathode in a membrane‐free biofuel cell under anode‐limiting conditions showed unprecedented benchmark power densities of 4.4 mW cm−2 at 0.7 V and an open‐circuit voltage of 1.14 V even at moderate catalyst loadings, outperforming the previously reported system obtained with the [NiFeSe] wild type and the [NiFe] hydrogenase from D. vulgaris Miyazaki F.


Introduction
Bioelectrocatalysis has gained huge importance in the past decades and is considered ap romisingr esearch field for the de-velopment of novel sustainable energy conversion and storage systems [1] as well as for the green production of value-added chemicals and (solar) fuels. [1,2] In particular, the use of biocatalysts to produce sustainable H 2 /O 2 -powered biofuel cells with high powero utput has become am ajor research area in this context.A lthough the biocathodes used for such devices are mainly based on rather robust, stable, and easy-to-wire multicoppero xidases such as bilirubin oxidaseo rl accase, [3] the biocatalysta tt he bioanode, that is, highly active H 2 -oxidizing Ni/ Fe-based hydrogenases, typicallys uffers from pronouncedO 2 sensitivity and fast inactivation at high potentials especially under anode-limiting conditions. [4] Hence, the preparation of hydrogenase-based bioanodes requires the implementation of specific protection systems, for example, based on O 2 -reducing low-potential redox polymers, or the use of O 2 -tolerantb ut usually less activeh ydrogenases, such as, for instance, the hydrogenases from Escherichia coli, [5] Ralstonia eutropha, [6] or Aquifex aeolicus. [7] The introduction of O 2 -reducing redox polymers fore lectrical wiring and protection of air-sensitive hydrogenases was successfully demonstrated for varioush ydrogenases including [NiFe] hydrogenasef rom D. vulgaris Miyazaki F, [8] [FeFe] hydrogenase from Chlamydomonasr einhardtii, [9] and [NiFeSe] hydrogenase from D. vulgaris Hildenborough. [10] Moreover, effective protection was even observed for thin films [11] and polymer/enzyme-modified gas-diffusion electrodes (GDEs), which could be incorporated in membrane-free biofuel cells that exhibited benchmark powerd ensities for polymerbased systems. [12] In addition, the low-potential redox polymer Variants of the highly active [NiFeSe] hydrogenase from D. vulgaris Hildenborough that exhibit enhanced O 2 tolerance were used as H 2 -oxidation catalysts in H 2 /O 2 biofuelc ells. Two [NiFeSe] variants were electrically wired by meanso fl ow-potential viologen-modified redox polymers and evaluated with respectt oH 2 -oxidation and stability against O 2 in the immobilized state. The two variants showed maximum current densities of (450 AE 84) mAcm À2 for G491A and (476 AE 172) mAcm À2 for variant G941S on glassy carbon electrodes and ah igherO 2 tolerance than the wild type. In addition, the polymer protected the enzymef rom O 2 damage and high-potential inactiva-tion, establishing at riple protection for the bioanode. The use of gas-diffusion bioanodes provided current densities for H 2oxidation of up to 6.3 mA cm À2 .C ombination of the gas-diffusion bioanode withabilirubin oxidase-basedg as-diffusion O 2reducing biocathode in am embrane-free biofuel cell under anode-limiting conditions showedu nprecedented benchmark power densities of 4.4 mW cm À2 at 0. 7  KGaA. This is an openaccessarticleunder the termsoft he Creative Commons AttributionL icense, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. not only acts as an O 2 -quenching matrix but also as aN ernst buffer system for the biocatalyst, preventing inactivation at high potentials. [8] Hence, low-potential redox polymers provide ad oublep rotection system for such sensitive materials, which was even transposable to synthetic catalysts. [13] Moreover,h ydrogenases deactivated under aerobic conditions may be reactivated by the low-potential polymer matrix, as was shown previously for [NiFe] [14] and [NiFeSe] [10] hydrogenases.
Although O 2 -tolerant hydrogenases derived from the abovementioned microorganisms can be operated in the presence of distinct levels of O 2 , [15] inactivation at high potentials still remains an issue. [5][6][7] Furthermore, the overpotentialf or H 2 -oxidation is often at more positive values than those of O 2 -sensitive [NiFe] and [FeFe] hydrogenases. [5,15,16] Evidently,t his will limit the power output of ar elated biofuel cell because this value will directly affect the maximum open-circuit voltage (OCV). Hence, ac ombination of redox polymers (double protection shield) and hydrogenase variants with enhanced O 2 tolerance would allow the development of triple-protected bioanodes. We want to emphasize that althought he redox polymer requires as lightly more positive redoxp otential than the hydrogenasei tselft oe nsure successful electron exchange at oxidative conditions, the open-circuit potentialo ft he corresponding bioanode and, consequently,t he OCVo ft he cell will not be limited by the mid-point potential of the redox polymer owing to the pseudo-capacitive properties of the polymer matrix. [17] This effect was shown for glucoseo xidase/bilirubino xidase [18] as well as hydrogenase/bilirubin oxidase-basedb iofuel cells, [12] in which the anodic catalyst (glucose oxidase,h ydrogenase) was electrically wired by means of aredox polymer.
The properties of enzymes can be modulated by enzyme engineering. [19] For instance, glucose oxidase that uses O 2 as a natural electron acceptor could be turned into an O 2 -insensitive enzymeb ys ite-directed mutagenesis [20] andc ofactor redesign. [21] Artificial maturation of [FeFe] hydrogenasea llows for the fine tuning of the properties of thea ctive centero ft he enzyme. [22] The synthetic natureo ft he active site allows for the incorporation of specific ligandso rt he alteration of the overall ligand sphere to adjust the properties of the whole enzyme. [23,24] Following this approach, the stability of [FeFe] hydrogenases could be enhanced, andt he overall activity of the enzyme can be controlled. [23,24] Although the maturation process occurs spontaneously without any helper proteins or additional cofactors, [25] the preparation of the active site of a[ FeFe] hydrogenase requires distinct synthetic efforts, [26] and the maturationp rocess is an additional step in the preparation of the active enzyme. In contrast, once developed, variants of an enzymes howing altered properties are produced directly from the living organisms. The additional maturation step and complex synthesis is thus not required.
Recently,i tw as shown that variants (G491Aa nd G491S) of the [NiFeSe] hydrogenasef rom D. vulgaris Hildenborough with modification in as pecific amino acid close to the active site led to enhanced stability of the biocatalyst in the presenceo f O 2 while retaining ah igh activity for H 2 -oxidation[ G491A: up to (4080 AE 80) s À1 ;G 491S:u pt o( 2810 AE 150) s À1 ,w ild type: % (4850 AE 260) s À1 ]a nd ar edox potential that is still close to the H 2 /2 H + couple [approximately À450 mV vs. standard hydrogen electrode (SHE) at pH 7]. [27] The enhanced stabilityo f the altered proteins was attributed to ap hysicalb locking effect of the O 2 molecule in ah ydrophilic channel that connects the active site of the protein with the enzyme surface, thus preventing oxidation of as pecific active-site cysteine ligand (Figure 1a;f or am ore comprehensive description of the structuralc hanges inside the protein shell of the enzyme that lead to the desired O 2 -blocking effect,s ee Ref. [27]). This effect Figure 1. (a) Cartoon showing the mechanism of the different behavior of the wild type (1)and the variants (2)of[NiFeSe] hydrogenase when exposed to O 2 .Int he wild-type [NiFeSe] hydrogenase, oxygen reaches the active center of the enzyme( 1), whereas in the variants G491A and G491S, the pathway to the active center is partially blockedbya ltered amino acid residues, which hamper the access of O 2 (2). For ad etailedstructurala nd mechanistic description, see Ref. [27].(b) Schematic of the high-current-density carbon cloth gas-diffusion H 2 -oxidationb ioanodee quipped with ap olymer/ hydrogenase layer.For the immobilization oft he activeP (N 3 MA-BA-GMA)vio/hydrogenase layer,acarbon cloth-based gas-diffusionl ayer was first modified with an adhesion layer,t hat is, the redoxp olymer P(GMA-BA-PEGMA)-vio, which showsahigher hydrophobicm onomer content( for ad etailed description of the electrodearchitecture and electrochemicalgas diffusion cell, seeRef. [12]). By combiningthe low-potential redox polymer with am ore O 2 -tolerant hydrogenasevariant, as table high-current-density bioanode is obtained, which can be operated in am embrane-freeH 2 /O 2 biofuel cell. (a, b) The structureo ft he wt-[NiFeSe]from D. vulgaris Hildenborough (5JSH) [28] was useda sar epresentative enzyme model; not drawnt o scale. was evidencedb yp rotein film electrochemistry conducted in a direct electron transfer (DET) regime and in the presence and absence of O 2 . [27] However,t he DET mode does not provide any protectiona gainst high-potential inactivation or against high O 2 concentrationsa nd is hence impractical for potential applications. Nevertheless, it shows that variantso ft his typeo f hydrogenase can be prepared with enhanced O 2 stability.
In this contribution, we combined the advantages of the enhanced O 2 -stable variants of [NiFeSe] hydrogenase from D. vulgaris Hildenborough and the O 2 quenching and Nernst buffer properties of low-potential viologen-modified redox polymers to fabricateh igh-current-density bioanodes, whichc ould be successfully incorporated into H 2 /O 2 -powered biofuel cells exhibiting benchmarkp owerd ensities at moderate catalystl oadings.
Indeed,c yclic voltammograms of drop-cast P(N 3 MA-BA-GMA)-vio/G419A andP (N 3 MA-BA-GMA)-vio/G419S films measured under turnover conditions, that is, under H 2 atmosphere (Figure 2a,b,r ed curves), showed pronounced catalytic H 2 -oxidation waves with half-wavep otentials (a and b: approximately À0.32 Vv s. SHE), which matches closely the mid-point potential of the polymer-bound viologen units ( % 0.34 Vv s. SHE, black curves). The behavior is in line with the resultsm easured for the wild type ( Figure S2 in the Supporting Information). We hence conclude that the [NiFeSe] variants can also be productively wired throught he redoxp olymer P(N 3 MA-BA-GMA)-vio in am ediated electron-transfer regime. Moreover,l ong-term chronoamperometric measurements over 7h under continuous turnoverc onditions showed similaro perational stability for the wild type and the two variants ( Figure S3 in the Supporting Information).
The maximum current densities J max measuredf or the individual, freshly prepared electrodes were calculated as average values from three electrodes as (450 AE 84) mAcm À2 forG 491A and (476 AE 172) mAcm À2 for G941S.T he wild type shows a J max value of (752 AE 259) mAcm À2 (note that for the wild type and the variants, storage of the modified electrodes leads to ad ecrease in overall electrode activity). The results are in line with the measured activities of the used enzyme batches [G491A: (1918 AE 119) s À1 and G491S: (2416 AE 387)s À1 ;n ote that the standard deviations are overlapping],w hich are below the activity of the wild-type enzyme( 4850 AE 260 s À1 ) [27] and below the maximum values reportedp reviously for the variants (see values above). [27] The steady-state current observeda th igh potentials (> À0.2 Vv s. SHE, Figure 2, red lines) under turnoverc onditions indicates that the variants can also be effectively protected against high-potential inactivation in contrastt ot he operation under DETconditions for this type of hydrogenases. [29,30] Oxygen tolerance Because the low-potential redox polymer acts as an O 2quenching matrix and thus protects the enzyme from O 2damage, [8] the stabilityo ft he variantsa nd the wild type against oxygen in the immobilized sate was measured in the absence of H 2 in chronoamperometric experiments. Under these conditions, electrons from H 2 -oxidation are absenta nd cannotb eu sed by the polymer matrix to reducei ncoming O 2 . [8,31] Figure 3s hows chronoamperometric experiments at an applied potential( E appl )o f+ 160 mV (vs. SHE) under alternating gas-mixture atmospheres. The O 2 content in the gas feed was stepwisei ncreased after each H 2 cycle. To ensure that all H 2 had been removed beforet he O 2 was added to the gas feed, the cell was purged with argon.A fter the background current was reached, the film was exposed to an O 2 /Ar mixture with varying O 2 content (5, 10, and 15 %, gray shaded areas in Figure 3). The wild type shows as teady decrease of the H 2 -oxidation activity over all O 2 /Ar cycles. After exposure to 15 %O 2 , the electrode remainsi nactive when switching the gas feed back to H 2 .T his is consistent with af ast in-diffusion of O 2 to the active center of the enzyme (Figure 1a). In contrast, both variants show ar ather constantc urrent output after the 5a nd 10 %O 2 cycle. Moreover,e ven after exposure to 15 %O 2 ,b oth electrodes still show aremarkable activity towards H 2 -oxidation (G491A: % 20 %o ft he initial H 2 -oxidation current; G491S: % 35 %). The resultsd emonstrate that the variants indeed exhibit an increased O 2 tolerance compared with the wild-type enzyme owing to ap artial blockingo fm olecular oxygen (hampered access) based on the altered amino acid residues in the variants (Figure 1a)a nd, by this, that the variants providea n additional protection for the proposed H 2 -oxidation bioanodes. The electrochemical results obtained with the polymer/enzyme films are in line with the results reported for operating the same hydrogenases in the DET regime. [27] However,s trongv ariations in the residual currents wereo bserveda fter exposure to O 2 ,w hich is attributed to variations in film thickness and inhomogeneities of the catalytic layers, leading to different diffusion profiles of O 2 .H owever,i na ll experiments the variants showedahigherstabilityt owards O 2 .

Reactivation
For the wild-type [NiFeSe] hydrogenase, which wasdeactivated under aerobic conditions, reactivation is known to occur quickly at rathern egative potentials. [29,30] Moreover,t he reduced low-potential polymer matrix P(N 3 MA-BA-GMA)-vio is also able to reactivate the inactive [NiFeSe] hydrogenase. [10] To evaluate ap ossible reactivation behavior of the two [NiFeSe] hydrogenase variants, glassy carbon electrodes modified with P(N 3 MA-BA-GMA)-vio/G491A and P(N 3 MA-BA-GMA)vio/G491S films were exposed to O 2 until complete inactivation occurred, as evidenced by the current dropping back to background values. Application of an egative potential of À440 mV (vs. SHE;p olymer is fully reduced, inactive mediatorf orm) for 500 sl eads to reactivation of the enzyme (Figure4)a si ndicated by the oxidative currents, which were observed again when the potentialwas stepped back to + 160 mV (vs. SHE; t > 500 s, mediatorisoxidized, active form). Both potentials were applied under a9 0% H 2 /10% Ar gas feed. The wild type shows the same behavior (see Figure S4 in the SupportingI nformation and Ref. [10]).
For practical applications, this effect is highly desirable because ap otentially necessary exchange of ad eactivated electrode in ad evice can be prevented. Instead, as hort cathodic potentialp ulse might reactivate the electrode, and operation can be resumed.

Membrane-free H 2 /O 2 biofuel cells
Oxygen-tolerant hydrogenases typical display higher redox potentials,w hich will decrease the maximum OCV of ac orresponding biofuel cell compared with their O 2 -sensitive analogues. However, because the O 2 -tolerantv ariants G491A and G941S show similar H 2 -oxidation potentials as the wild type, [27] the electrical wiring is possible with the same polymer ( Figure 2). Hence, we expect similarO CV values for relatedb iofuel cells as for those based on the wild-type hydrogenase. To evaluatet he performance of the bioanodes in ab iofuel cell, polymer/hydrogenase-modified glassy carbon electrodes were combined with ag as-diffusion O 2 -reducing bilirubin oxidasebased biocathode. The use of ag as-diffusion system at the cathodes ide, in which mass transport is not limiting, ensures bioanode-limiting conditions. The biocathode was prepared with bilirubin oxidase from Bacillus pumilus (Bp-BOD), [32] a stable multi-copper oxidase used previously in biofuel cells, [18] by drop-casting a Bp-BOD stock solution (borate buffer, 50 mm,p H, pH 9, 54.75 mg mL À1 )o nto ac arbon cloth-based gas-diffusion layer equipped with ac onducting microporous Nafion/Teflon/carbon layer with enhanced surface area (for a detailedd escription of the immobilization process, see the Experimental Section). In cyclic voltammograms, maximum absolute currents for O 2 reductiono fa pproximately 180 mAw ere observedw hen the gas-diffusion electrode was exposed to air ( Figure S5 in the Supporting Information). These values are significantly highert han those obtained for the polymer/hydroge-nase-modified glassy carbon electrodes exhibiting maximum absolute currents < 80 mAfor all hydrogenases.
Membrane-free biofuelc ells prepared with the gas-diffusion Bp-BOD-based biocathode in combination with P(N 3 [8] and [FeFe] [9] hydrogenases. Recently,w es howedt hat the use of gas-diffusion layers modified with polymer/wt-[NiFeSe] and polymer/[NiFe] films displayed enhanced power outputo wing to an enhanced mass transport of the gaseous substrate H 2 towards the bioanode. [12] Currentd ensitiesf or the bioanode of close to 8mAcm À2 and power densities of 3.8 mW cm À2 for biofuel cells with ab ilirubin oxidase-modified gas-diffusion biocathode were observed. [12] To demonstrate the relevance of the O 2 -tolerant [NiFeSe] variants, carbonc loth-based gas-diffusion layers were first modified with P(GMA-BA-PEGMA)-vio [poly(glycidyl methacrylate-co-butyl acrylate-co-poly(ethylene glycol)methacrylate)-vio; for the structure and synthesis of this polymer, see Figure S1 in the Supporting Information and Ref. [12],r espectively] films followed by the immobilization of an active P(N 3  The less hydrophilic viologen-modified polymer P(GMA-BA-PEGMA)-vio acts as an adhesion layer between the hydrophilic active layer and the hydrophobic carbon cloth surface. [12] Moreover,t he underlying redox polymerl ayer prevents contribution from DET betweent he enzymea nd the porous electrode surface and excludes high-potential inactivation. [12] Under gas-diffusion conditions, the bioanode showed absolute H 2 -oxidation currents of approximately 0.8 mA (Figure 5a). The modified surfacea rea of the carbonc loth-based bioanode has ad iameter of approximately 4mm, which results in as urface area of the active layer of approximately 0.126 cm À2 ,a nd thus maximum current densities of 6.3 mA cm À2 were achieved. The values are similart op reviouslyr eported polymer-based gas-diffusion systemse quipped with wt-[NiFeSe] and [NiFe] hydrogenases (Table S1 in the Supporting Information). However, care must be taken when comparing current densities measured with porous electrodes. Because of the 3D structure of the electrodes, the real surfacei so ften unknown. Hence, the catalystl oading is ab etter value for comparison. For the hydrogenase, current densities of only 5.3 mA cm À2 were observed with as ubstantially higherc atalystl oading of 27.0 nmol cm À2 / 3.4 nmol electrode À1 as reportedi no ur previousw ork (see Ref. [12] and Ta ble S1 in the Supporting Information), which largely exceeds the values of the G491S variant at almosti dentical overall polymer loading [wt-[NiFeSe]: 230 mgelectrode À1 (previousw ork, Ref. [12]);G 491S:2 60 mgelectrode À1 ]. At a lower catalyst loading of 12.1 nmol cm À2 /1.53 nmol electrode À1 (polymer loading 230 mgelectrode À1 ), which is only slightly higher than the loading of the G491S enzyme, the wt-[NiFeSe] shows a J max value of only 3.6 mA cm À2 (see Ref. [12]). This effect might be relatedt oa ni mproved incorporation of the G941S variant in the polymer film when immobilizedo nt he rather hydrophobic carbon cloth-based electrodes. As tronger interaction prevents leachingo ft he enzymea nd thus ensures ah igher local concentration of the biocatalyst during the experiment. In addition, al oss of activity in the immobilizeds tate for the wild type may also contribute to ar educed electrode activity.A ne ffect of different polymer-to-enzyme ratios can be ruled out because almosti denticalp olymer loadings were used for all experiments. However,t he effect seems to be specific for the porous, hydrophobic carbon cloth electrodes because the wild type shows ah ighera ctivity on flat glassy carbon electrodes (see Figure2 and Figure S2 in the Supporting Information).
To evaluate the performance of the gas-diffusion P(GMA-BA-PEGMA)-vio//P(N 3 MA-BA-GMA)-vio/G491S electrode in an allgas-diffusion membrane-free H 2 /O 2 biofuel cell, the bioanode was combined with an O 2 -reducing biocathode modified with bilirubin oxidase from Myrothecium verrucaria (Mv-BOD,f or comparison purposes because it was used in our previously reported experiments [12] ). For the immobilization of Mv-BOD, the carbon cloth was first modified with 2-ABA (2-aminob enzoic acid) to ensureaproper orientationo fthe enzyme on the electrode surface. The modifier was anchored in an electrochemical graftingp rocess by applying an oxidative potential pulse. [33] The Mv-BOD wast hen immobilizedb ym eans of ac onventional drop-casting process and was operated in the DET regime. [12,33] Ah igh catalyst loading was used to ensure anodelimitingc onditions (nominal enzymel oading:1 .2 mg electrode À1 ). Absolute currents under gas-diffusion conditions (100 %O 2 )r eached approximately 2mA( Figure S7 in the Supporting Information), which largely outperforms the bioanode ( % 0.8 mA, Figure 5a). The fully assembled H 2 /O 2 biofuel cell (Figure 5b)s howeda nO CV of 1.14 V, which is slightly higher than the values obtained on glassy carbon electrodes (1.05-1.06 V);t his might be attributed to the slightly lower overpotential for O 2 reduction of Mv-BOD compared with Bp-BOD. [34] The maximum powerd ensity wasr eached at 0.7 Vand was estimated to be 4.4 mW cm À2 .T his value even outperforms our previously reported value for the [NiFe]-based biofuel cell (3.6 mW cm À2 ) [12] and-tot he best of our knowledge-sets a new benchmark for ab iofuel cell using redox-polymer-based bioanodes (Table S1 in the Supporting Information). Moreover, the catalyst loading is significantly lower than the [NiFe] system (31.8 nmol cm À2 /4 nmol electrode À1 ) [12] reported previously (Table S1 in the Supporting Information).
Cyclic voltammograms ( Figure S8 in the Supporting Information) measured before and after biofuel cell operation showed similarv alues for the bioanode, with the slightly higher currents after the biofuel cell test most likely as ar esult of changed diffusion properties inside the polymer/enzyme layer, for example, owing to swelling and/or slightly changed local pH values, which will affect the overall activity of the enzyme. In contrast, the current of the biocathode was slightly decreased after the biofuele valuation ( Figure S7 in the Supporting Information). This againh ighlights the high stabilityo ft he bioanode in am embrane-free biofuel cell under anode-limiting conditions. The operational stability of the biofuel cell was tested at aconstant load of 0.7 V (Figure 5c). After 10 hofcontinuous operation, 75 %o ft he initial current density remained. Cyclic voltammograms measured after the long-term experiment showeds ignificantly lower currents for the bioanode ( Figure S8 in the Supporting Information) and the biocathode  Figure S7 in the Supporting Information). We want to emphasize that the amplitudes of the polymer signals ( Figure S7 in the Supporting Information, dashed black curve)w ere also decreasedc ompared with the voltammograms measured with the freshly preparede lectrode. Thus, not only does deactivation/decompositiono ft he enzyme contribute to the decreased activity after long-term operation, but the loss of immobilization matrix may also have an effect. Nevertheless, the bioanode shows an outstanding performance and demonstrates the potentiala pplicability of G941S (enhanced O 2 tolerance) as ah ighly active and stable catalysti namembrane-free biofuel cell device. Moreover,t he proposed H 2 -oxidation bioanodes combinet he advantages of the protection matrix (O 2 quenching;n oh igh-potential deactivation)a nd the enhanced enzyme stabilityo ft he hydrogenasev ariants (blocking of O 2 access) in accordance with the mechanism depicted in Figure 1b and thus demonstrate at riple-protections ystem for the high-current-density H 2 -oxidation bioanodes.

Conclusions
The two [NiFeSe] variants show ah igher O 2 tolerance than the wild type in the immobilizeds tate, whichi si nl ine with results reportedf or the enzymes operated in ad irect electron transfer regime. [27] In combination with the redox polymer-based protection matrix, the proposedb ioanodes reveal at riple-protection system that ensures stable operation. Moreover,w ec ould demonstrate that the two [NiFeSe] variants show similarp erformance as the wild type and as [NiFe] as wella s[ FeFe] hydrogenases when incorporated into ac onventional redox-polymer-based biofuel cell.I na ddition, the use of gas-diffusion layers ensured high substrate transport towards the active polymer/enzyme layer,a llowing H 2 -oxidation currents of approximately 6.3 mA cm À2 for the G491S variant at comparatively low catalyst loadings. Combination of the gas-diffusion bioanode with ag as-diffusion O 2 -reducing biocathode allowed for the fabrication of aH 2 /O 2 -powered biofuelc ell with benchmark performance in am embrane-free configuration. We conclude that the novel O 2 -tolerant [NiFeSe] variants are promising candidates for biofuel cell applications,d emonstratingt hat enzyme engineeringi si ndeed ap owerful tool, which may be used to not only overcome sensitivity issues but also to further enhancethe activity and stability of biocatalysts.

Electrochemical experiments
All electrochemical experiments were conducted under the corresponding atmosphere (argon, hydrogen, oxygen, and their mixtures) and at room temperature by using aG amry Reference 600 potentiostat in at hree-electrode configuration with an Ag/AgCl/ 3 m KCl reference electrode. All potentials were rescaled to the standard hydrogen electrode (SHE) according to the equation E SHE = E Ag/AgCl/3 m KCl + 210 mV.P hosphate buffer (0.1 m,p H7.3) was used as electrolyte for all experiments. For cyclic voltammetric and chronoamperometric experiments, aP tc ounter electrode and modified glassy carbon disk working electrodes (3 mm) were used. The latter were polished by using, first, diamond particles (3 mm) followed by Al 2 O 3 powder (1 mm, then 0.3 mm) following standard protocols. Measurements under gas-diffusion conditions were performed in ah omemade glass cell [12] with carbon cloth-based gasdiffusione lectrodes [MTI, carbon foam sheet, porous C, 0.454 mm thick, % 10 mL cm À2 s À1 ,p orosity % 31 mmc oated on one side with ac onductive Nafion/Teflon-based microporous film (50 mm), carbon content 5mgcm À2 ,E Q-bcgdl-1400S-LD].T hermal mass flow controllers (GFC17, Aalborg Instruments and Controls) were used to adjust the desired atmosphere and gas mixtures with predefined compositions (for compositions of the gas feed, see the main text and figures). The back of the gas-diffusion electrode was exposed to the corresponding gas atmosphere (bioanode) or to air/ O 2 (biocathode). During the experiments in gas-breathing mode, the electrochemical cell/electrolyte was continuously purged with an argon stream to prevent permeation of O 2 into the bulk electrolyte. For characterization of the biofuel cells, power curves were measured by stepped potential chronoamperometric experiments to minimize contributions from capacitive charging currents. After each potential step, steady-state currents were used to calculate the corresponding power values. Modification of glassy carbon electrodes withhydrogenase/ polymerfilms All films were prepared by means of as tandard drop-casting process. For this, stock solutions of the polymer P(N 3 MA-BA-GMA)-vio and the corresponding hydrogenase variant were prepared:4mL of an aqueous P(N 3 MA-BA-GMA)-vio solution (7.3 mg mL À1 )w ere mixed with 3 mLo fp hosphate buffer (0.1 m,p H7.3), and the corresponding hydrogenase was added (wt-[NiFeSe]: 0.5 mL, 170 mm in Tris-HCl buffer,2 0mm,p H7.6;G 491A:1mL, 82.96 mm in Tris-HCl buffer;G 491S:1 .56 mL, 53 mm in Tris-HCl buffer). For electrode modification, 1.3 mLo ft he stock solution was drop-cast onto the 3mmg lassy carbon disk electrode. The modified electrodes were incubated overnight at 4 8Ca nd air dried for 1h prior to use. In a typical experiment, three electrodes were modified from the same stock solution.

Modification of carbonc lothelectrodes with hydrogenase/ polymerfilms
First, 20 mLo ft he polymer P(GMA-BA-PEGMA)-vio (7.5 mg mL À1 in water) was drop-cast on the microporous side of the carbon cloth electrode and dried overnight at room temperature. Subsequently, 20 mLo fG 491S (53 mm in Tris-HCl, 20 mm,p H7.6) was mixed with 15 mLo fP (N 3 MA-BA-GMA)-vio (7.3 mg mL À1 in water) and drop-cast onto the already existing polymer spot (diameter of % 4mm). The electrode was dried overnight at 4 8C.

Modification of carbonc lothelectrodes with Bp-BOD
The bare carbon cloth was pre-wetted with ethanol on both sides and rinsed with water.T hen, the microporous side of the gas-diffusion layer was modified with 20 mLo ft he Bp-BOx solution (54.75 mg mL À1 in borate buffer,5 0mm,p H9). The electrode was dried overnight at 4 8C.

Modification of carbonc lothelectrodes with Mv-BOD
For the preparation of the Mv-BOD-based cathode, the microporous side of the ethanol-treated carbon cloth gas-diffusion electrode was first modified with 2-amino benzoic acid (2-ABA) in an electrochemical grafting process in 0.1 m KCl/5 mm 2-ABA/water by applying ap otential pulse of + 0.8 Vv s. Ag/AgCl/3 m KCl for 60 s according to procedures reported in Refs. [12,33].T he modified electrode was rinsed with water and further modified with 120 mL of an aqueous Mv-BOD solution (10 mg mL À1 ,n ominal enzyme loading:1.2 mg electrode À1 )and dried at 4 8Co vernight.