Photoreduction of Shewanella oneidensis Extracellular Cytochromes by Organic Chromophores and Dye‐Sensitized TiO2

Abstract The transfer of photoenergized electrons from extracellular photosensitizers across a bacterial cell envelope to drive intracellular chemical transformations represents an attractive way to harness nature's catalytic machinery for solar‐assisted chemical synthesis. In Shewanella oneidensis MR‐1 (MR‐1), trans‐outer‐membrane electron transfer is performed by the extracellular cytochromes MtrC and OmcA acting together with the outer‐membrane‐spanning porin⋅cytochrome complex (MtrAB). Here we demonstrate photoreduction of solutions of MtrC, OmcA, and the MtrCAB complex by soluble photosensitizers: namely, eosin Y, fluorescein, proflavine, flavin, and adenine dinucleotide, as well as by riboflavin and flavin mononucleotide, two compounds secreted by MR‐1. We show photoreduction of MtrC and OmcA adsorbed on RuII‐dye‐sensitized TiO2 nanoparticles and that these protein‐coated particles perform photocatalytic reduction of solutions of MtrC, OmcA, and MtrCAB. These findings provide a framework for informed development of strategies for using the outer‐membrane‐associated cytochromes of MR‐1 for solar‐driven microbial synthesis in natural and engineered bacteria.


Introduction
Certain specieso fb acteria conduct electrons acrosst he cell envelopei namanner that couples intra-and extracellular redox reactions. In nature,t his extracellular electron transfer (EET) offers ac ompetitive advantage in anaerobic habitats because electronsr eleased by intracellular energy-conserving pathways cross the cell membrane to reduce extracellular ter-minal electron acceptors that include particles containing Fe III and Mn IV . [1] However,E ET can also exchange electrons between electrodes and bacteria known as electrotrophs. Electricity is generated when respiratory electrons from the oxidation of waste-water-derived electron donors are delivered to the anode of am icrobial fuel cell. [2] Microbial electrosynthesis is performed when the pathway of this respiratory electron transfer is reversed such that cathode-derived electrons are delivered to intracellular enzymes. [2b, c, 3] This approacha ffords strategies for tappingi nto the catalytic diversity and selectivity of enzymes fors ustainable molecular syntheses that extend beyondH 2 production and CO 2 reduction while at the same time negating the need for time-consuminga nd costly enzyme purification.I ndeed, the prospect of using EET to couple robust and efficient extracellular light-harvesting systems to intracellular catalysis represents aparticularly attractive approacht os ustainable solar-assistedp roduction of chemicals. [3a, 4] Shewanella oneidensis MR-1 (MR-1) provides am odel for the biochemistry and biophysics of EET [5] and as ac onsequence ap latform for the rational design of strategies for microbial electro-and photosynthesis. EET in this Gram-negative electrotroph is underpinned by arrays of closely packed, proteinbound heme cofactors that conduct electrons within and between proteins.T he MR-1 outer membrane is spanned by por-in·cytochrome complexest hat conducte lectrons between the periplasma nd externalm aterials. Foremost amongst these is at ight 1:1c omplex of two proteins:M trA and MtrB (MtrAB, Figure 1A). It is proposed that the decahemec ytochrome MtrA The transfer of photoenergized electrons from extracellular photosensitizers across ab acterial cell envelope to drive intracellular chemical transformations represents an attractive way to harness nature's catalytic machinery for solar-assistedc hemical synthesis. In Shewanella oneidensis MR-1 (MR-1), transouter-membrane electron transfer is performed by the extracellular cytochromesM trC and OmcA acting togetherw ith the outer-membrane-spanning porin·cytochrome complex (MtrAB).
Here we demonstrate photoreduction of solutions of MtrC, OmcA,a nd the MtrCAB complex by soluble photosensitizers: namely,e osin Y, fluorescein, proflavine, flavin, and adenined inucleotide, as wella sb yr iboflavin and flavin mononucleotide, two compounds secreted by MR-1. We show photoreduction of MtrC and OmcA adsorbed on Ru II -dye-sensitized TiO 2 nanoparticles and that these protein-coated particles perform photocatalytic reduction of solutions of MtrC, OmcA, and MtrCAB. These findings provide af ramework for informed development of strategies for using the outer-membrane-associated cytochromes of MR-1 fors olar-driven microbial synthesis in natural and engineered bacteria.
conducts electrons across the outer membrane by virtue of its insertionw ithin ap orin formed by MtrB. [6] MtrAB forms at ight complexw ith the extracellularc ytochrome MtrC, with whichi t exchanges electrons ( Figure 1A). During EET as econd extracellular cytochrome, OmcA,c an bind to, and exchange electrons with, MtrC. [7] X-ray diffraction has shown that MtrC and OmcA are structuralh omologues [8] with ten heme groups bound in a staggered cross constellation ( Figure 1B). In addition, MtrC and OmcA were shown to possess one and two disulfide bonds, respectively.I nv itro reduction of these bonds triggered tight binding of flavin mononucleotide (FMN) or riboflavin (RF) to both proteins; [8b] this might be significant because cellular studies have suggested that MtrC and OmcA act as flavocytochromes during anaerobic respiration. [9] Several mechanisms by which MtrC and OmcA might facilitate electron exchange between the MR-1 cell surfacea nd extracellular materials have been proposed. Direct electron exchange might occur between these materials and the cofactors of MtrC and OmcA. [9][10] Electronsm ight be shuttled between these materials and the cell-surface cytochromes through the diffusion of extracellular,r edox-active mediators that include flavins and low-molecular-weight Fe complexes. [11] In addition, outer-membrane extensions, coated with MtrC and OmcA and sometimes termed nanowires, have been implicated in mechanisms for electron exchange with remote materials across distances exceeding the cell dimensions. [12] We have explored strategies to effect the photoreduction of MR-1 extracellularc ytochromes with the aid of water-compatible light-harvesting systems, because we envisage this as a route by which to facilitate solar-assisted microbial production of chemicals by delivering photoexcited electrons to intracellular enzymes. We recently described how am onolayer of MtrC supported light-driven charget ransport to an underlying ultraflat gold electrode when coated with 3,4-dihydroxybenzoicacid-capped TiO 2 nanocrystals (diameter % 7nm) that had been photosensitized with ap hosphonated Ru II -tris(bipyridine) dye. [13] Here we report the photoreduction of solutions of MtrC, OmcA, and the MtrCAB complex by both biotic and abiotic photosensitizers that include organic dyes and transitionmetal complexes (Figure2A). We demonstrate photoreduction of MtrC and OmcA adsorbed on widely availableT iO 2 nanoparticles sensitized with aR ud ye ( Figure 2B). In addition, we show that MtrC or OmcA adsorbed on TiO 2 particles serve as electron relays in the photoreduction of solutionso fM trC, OmcA,a nd the MtrCAB complex ( Figure 2C). These results extend the framework from which informeda pproaches to artificial microbial photosynthesis can be developed for strains of native and engineered [14] bacteria that support EET through the action of multiheme cytochromesf rom, or homologous to those of,MR-1.

Results and Discussion
Photoreduction of MtrC, OmcA, and MtrCABb ys oluble photosensitizers Evidence for visible-light-driven reduction of MtrC by soluble photosensitizers (Figure 2A)w as sought through electronic absorbance spectroscopy.O xidized (ferric) heme groups contribute as ingle broad feature to spectra between 500 and 600 nm whereas two sharper and more intense features with maxima at 523 and 552 nm are indicative of reduced (ferrous) heme groups. [15] Experiments were performed with eight photosensitizers,t he structuresa nd key photochemical properties of which are provided in Table S1 in the Supporting Information. FMN, RF,a nd flavin adenined inucleotide (FAD) are naturally occurring flavins. Thefirst two are secreted by MR-1 and participate in EET. [11a-d] Proflavine, fluorescein, and eosin Ya re wellstudied light-harvesting analogues of redox-active molecules that serve as electron shuttles to enhance the performance of microbial fuel cells. [16] [Ru(2,2'-bpy) 3 ]Cl 2 (bpy = 2,2'-bipyridine) and [Ru(bpy) 2 {4,4'-(PO 3 H 2 ) 2 bpy}]Br 2 (RuP) are robust light-harvestinganalogues of Fe III chelates used as extracellularterminal electron acceptors by MR-1. [1a, b, 5a] Triethanolamine( TEOA) and HEPESw erei ncluded unless stated otherwise because these tertiarya mines can serve as pH buffers and sacrificial electron donors during photoreduction. [17] Anaerobic solutionso f3 2mm MtrC displayed spectralf eatures that are typical of the oxidized protein and unchanged by the addition of 10 mm FMN (e.g.,F igure 3A,b lack line). Illuminationo ft his sample (l > 390 nm, power % 400 Wm À2 ;s ee the Experimental Section) resulted in the appearance of peaks with maximaa t5 23 and 552 nm indicative of ferrous heme ( Figure 3A,g ray lines). Them agnitudes of these peaks, which represent the extent of MtrC reduction, increased to am aximum during 90 min illumination and were unchanged by 30 min furtheri llumination ( Figure 3A,b lue line). As imilar experiment in the absence of FMN provided no evidencef or heme reduction (Figure S2 A). Thus, FMN was revealed to be an effective photosensitizer for the visible-light-driven reduction of heme groups within MtrC.
The extent of heme photoreduction was quantified by addition of excesss odium dithionite (e.g.,F igure 3A,d ashed red line). Dithionite (S 2 O 4 2À )h as ar eduction potential of about À500 mV (all potentials quoted vs. SHE) under the conditions of our study, [18] and when present in excessi tr educes all ten MtrC heme groups. [6,15,19] As ac onsequence, the electronic ab-sorbance of fully reduced MtrC at 552 nm was comparedw ith that of the MtrC generated by FMN-dependentp hotoreduction, and for the example shown in Figure 3A photoreduction was found to proceed to 56 %. Similar resultsw ere obtained in repeat experiments and for ratios of FMN to MtrC that ranged from approximately1 :0.005 to 1:1.5 with 10 or 100 mm MtrC (Table 1). Thus, the extent of MtrC photoreduction was independent of whether the concentration of FMN exceeded that of MtrC or was catalytic (sub-stoichiometric) with respect to the protein.
Results very similar to those described above were obtained in parallel experiments with MtrC replaced by OmcA,a nd also when FMN was replaced by RF or FADw ith either protein (Table 1). However,1 00 %p hotoreduction of MtrC and OmcA was triggered by 90 min illumination of samples that contained eosin Y, proflavine, or fluorescein( e.g.,F igure 3B and Ta ble 1). As in the cases of FMN, RF,a nd FAD, during the 90 min as teady-state level of cytochrome photoreduction was reached; this was independent of whether eosin Y, proflavine,  or fluorescein were present either in catalytic quantities or in excess. In contrast, less than 15 %p hotoreduction of the cytochromes was observed during9 0min illumination with either [Ru(bpy) 3 ] 2 + or RuP present even at % 200-fold excess over protein, and the photoreduction failed to reach steady-state conditions (Table 1).
To gain insighti nto the extent of photoreduction observed after 90 min illumination with each photosensitizer (PS) it was of interest to identify the corresponding photocatalytic cycle(s). Photoreduction of ap rotein can in principleb ea ssociated with oxidative and reductivequenching [20] of aphotoexcited state (PS*, Figure4A). The protein substrate would be directly reduced when PS* is oxidativelyq uenched. The product (PS + )w ould then be regenerated to the ground state (PS 0 ) through oxidation of as acrificial electron donor (SED). During reductiveq uenching, PS* would first oxidize the SED and form PS À .T hisw ould then reduce the protein to recover the ground state (PS 0 )a nd complete the photocatalytic cycle (Figure 4A). The feasibility of thesep athways is determined by the (photo)reduction potentials of the photosensitizer relative to those of the SED and protein (e.g.,F igure 4B and Table S1). In practicet he rates of these reactions and those of competing processes,p roductiveo rn ot with respect to reduction of the protein, will define the extent of photoreduction under any given conditions. [20] Oxidative and reductiveq uenching of organic photosensitizers typically produce radical species, and as ac onsequence their photochemical behavior often extends beyondt hat illustrated in Figure 4A. Table 1. Photoreduction of MR-1 cytochromes by the indicated photosensitizers.
[ c] Photosensitizer/protein ratios from 0.5:0.005 to 0.5:1.5 with 10 mm and 100 mm photosensitizer. With 20 and 10 heme groups per MtrCABa nd MtrC(OmcA), respectively,t his ensuredcomparable optical densities for the experimentsw ith each protein.   In order to assess the contributions made by TEOA and HEPES as SEDs during photoreduction of the extracellular cytochromes, OmcA (0.5 mm)w as illuminated for 90 min with a2 0fold excesso fagiven photosensitizer in a5 0mm phosphate solution and, in separate experiments, with TEOA or HEPES. Phosphate is inactive as aS ED, and we chose to restrict these studies to OmcA, because MtrC and OmcA have very similar structures, thermodynamic properties, [8b, 19] and, as we have shown here, photochemical behavior (Table 1). Eosin Yw ith TEOA (or HEPES) supported 100 %p hotoreduction of OmcA ( Figure 5). However, no photoreduction was detected in the absence of the tertiary amines,w hich suggests that OmcA reductioni sc oupled to reductiveq uenching of the eosin Yphotoexcited state by the SED. Under conditions comparable to those used here the corresponding PS 0/À couple has ar eduction potential (E m )o f%À580 mV. [21] The E m values for the OmcA heme groups span from approximately + 50 to À450 mV at the neutral pH of thesee xperiments and they are relativelye venly distributed across this potentialw indow. [19] The same is true of the corresponding values for the MtrC heme groups. [6,13] Thus, the observed 100 %p hotoreduction of OmcA,a nd of MtrC, by eosin Ys upported by TEOA and/orH EPES is consistent with the relevant (photo)reduction potentials ( Figure 4B). Similarly,t he pattern of fluorescein-dependent OmcA photoreduction indicates that reductive, but not oxidative, quenching operates ( Figure 5). Reductiveq uenching of fluorescein under conditions comparable to those in our studies produces ar eductant [21] with sufficient driving force to reduce OmcA and MtrC completely (Figure 4B). That this occurs with TEOA but not with HEPES indicates additional complexity in the latter system.
Previous studies have revealed complex photochemistry of proflavine under conditions comparable to those employed here. [20a, b] In anaerobic aqueous solutions reductive quenching predominates when aS ED is present. However,i nt he absence of aSED the corresponding PS* decays by triplet-triplet annihilation and photoionization to produce as olvated electron. The proflavine-dependentp hotoreduction of OmcAp roceeded more effectively when TEOA or HEPES were present ( Figure 5) and presumably occurred through reductiveq uenching. The PS 0/À couple relevant to this pathway has as ufficiently negative reduction potential [20c] to account for the complete photoreduction of OmcA and MtrC observed when the SEDs are present ( Figure 4B).
For RF,F MN, andF AD it is generally accepted that the corresponding PS* is reductively quenched to yield as emiquinone that undergoes rapid disproportionation to generateahydroquinone, this being the active photoreductant. [20d, e] Both interand intramoleculare lectron transfer reactions lead to reductive quenching [20d, e] and this is consistentw ith the cytochrome photoreduction that was observed in the absence of TEOA or HEPES as SED in our experiments( Figure 5). The hydroquinone forms of RF,F MN, and FADh ave E m values in the middle of the range spanned by the heme groups of OmcA and MtrC (Figure 4B). Thus, in the presence of TEOA and HEPES the steadystate levels of cytochrome photoreduction ( % 60 %, Ta ble 1) produced by these photosensitizers in comparison with the complete reduction produced by eosin Y, fluorescein,o rp roflavine were consistent with consideration of the available driving forces. However,t his was not the case for [Ru(bpy) 3 ] 2 + and RuP,f or which significantly less photoreduction was observed. Oxidativea nd reductive quenching of the corresponding photoexcited states would generate stronger reductants [20f, 22] than the photocatalyticc ycles operative with anyo ft he organic photosensitizers studied here ( Figure 4B). Because negligible photoreduction of the extracellular cytochromes was induced by the Ru II dyes their effectiveness as photosensitizers relative to the organic dyes must be compromised, by,f or example, slow net electron exchange with the cytochromesa nd SED and/or nonproductive side reactions. Indeed, greater levels of OmcA and MtrC photoreduction were observed over 5h illumination(e.g., Figure S2 Ba nd C).
Twoa dditional series of experiments were performedt o quantifyt he behavior of MtrC under conditions that might be relevant to those on the surfaceo fM R-1. The first series of experiments quantified photoreduction of the heme groups in detergent-solubilized MtrCAB complex purified from the MR-1 outer membrane( Figure 1A). MtrCAB is as table 1:1:1c omplex of the MtrC, MtrA, and MtrB proteinst hat contains 20 heme groups with E m values spanning ap otential window similart o those of MtrC and OmcA. [6,19] Suspensions of MtrCABi lluminated with each of the photosensitizers discussed above, TEOA,  (Table 1). As ac onsequence, the response of MtrCAB to these photosensitizers is most likely determined by the same factorsa sf or the extracellular cytochromes alone. Finally,i nv iew of recent reports that MtrC might exist as af lavocytochromeo nt he surface of MR-1, [9] it was of interestt oc haracterize the response of such ap rotein to illumination with visiblel ight.F lavocytochrome, composed of FMN tightlyb ound to MtrC, was prepared as described previously,b ya naerobic incubation of MtrC with glutathionea nd FMN, followed by gel filtration to separate the flavocytochromef rom free FMN. [8b] Electronic absorbance spectroscopy established that the heme groups and FMN remained oxidized throughout these processes, which were performed in the dark, and also after 90 min illumination ( Figure S3). Previous studies have established that the fluorescenceo fF MN is partially quenched on binding to the glutathione-reduced MtrC and that this fluorescencei sr ecovered when the FMN is released by oxidation of the flavocytochrome. [8b] This quenching most likely occurs through FRET due to the proximity of the bound FMN to heme. We suggest that bound FMN does not support photoreduction of the flavocytochrome because the correspondinge xcited state is quenched more rapidlyb y energy transfer than by electron transfer.I ti sp roposed that the observed photoreduction of MtrC by solutions of FMN, and presumably also RF or FAD( Ta ble 1), occurs through electron transfer at as ite other than that occupied by the flavin in the flavocytochrome.

Photoreduction of MtrC and OmcA adsorbedond ye-sensitizedTiO 2 nanoparticles
Ap revious study reported negligible intermoleculart ransfer of photoenergized electrons from RuP to am olecular cobalt catalyst in ap H-neutral aqueous solution in the presence of a SED. [23] However,c o-adsorption of this dye and catalyst on TiO 2 nanoparticles promoted efficient light-driven H 2 evolution. Charge separation from photoexcited RuP was enhanced through rapid oxidative quenching by TiO 2 ,a nd photoenergized electrons in the conduction band were readilyt ransferred to the catalyst. The TiO 2 conduction band has areductive potentialo fa pproximately À620 mV at pH 6 [24] such that complete photoreduction of MtrC and OmcA by RuP-coatedT iO 2 particles (RuP·TiO 2 ), is thermodynamically feasible ( Figure 4B). As ac onsequence it was attractive to establish whether MtrC or OmcA would adsorbo nR uP·TiO 2 in an electroactive configuration ( Figure 2B).
RuP forms as table linkagew ith TiO 2 at slightly acidic pH by virtue of its phosphonic acid groups. [25] The ability of OmcA or MtrC to adsorb on P25 TiO 2 nanoparticles under these conditions was assessed when each protein (1 mm)i n2 00 mLo f 150 mm SED MES at pH 6w as incubatedw ith 0.5 mg mL À1 particles. After 30 min incubation with occasional inversion the samples were centrifuged to pellet the particles and any adsorbed cytochrome. The white particles were found to have taken on ap ink-red color after incubation with the proteins ( Figure S4). The amount of cytochromet hat remained in the supernatant was quantified by electronic absorbance spectroscopy and found to be significantly less than that in the protein solution prior to incubation with the particles (Figure 6A,B).
Both observations were consistent with protein adsorption on the TiO 2 particles.
The integrity of the adsorbed proteins was confirmed by electronic absorbance of the protein-coated (TiO 2 ·MtrC(OmcA)) particles measured with an integrating-sphere spectrophotometer (e.g.,F igure 6C,t op). Use of the integrating sphere mitigated against loss of incident light due to scattering by the particles and facilitated the resolution of spectral features from the adsorbed proteins. The corresponding spectra displayed features typical of the oxidized proteins:n amely,astrong absorbance in the Soret region with am aximum at 420 nm and ab roader,l ess intense feature in the ab region between 500 and 600 nm. These features disappearedo na ddition of excess particles( 0.5 mg mL À1 )in 150 mm MES (pH 6) followed by centrifugation to pellet the particlesa nd retrieval of the supernatant(red). The protein-coated particles were resuspended in 200 mLo f2 5mm phosphate, 150 mm MES (pH 6) and incubated for 30 min to release the boundp rotein.T he particles werepelleted by centrifugation, and spectroscopyo ft he supernatant(blue)q uantifiedthe releasedp rotein.C )Electronic absorbance recorded with an integrating-sphere spectrophotometerf or suspensions (0.055 mg mL À1 )of( RuP·)TiO 2 ·MtrC (OmcA) particlesa si ndicated. Spectra are presented aftersubtraction of the responsef rom an equivalent solution of unmodified particles and offset on the y-axis for clarity;T iO 2 ·MtrCparticles before( black) and after( red) addition of excess sodium dithionite (top), RuP·TiO 2 ·MtrC particles before (black) and after (red)2 0min illumination (middle), RuP·TiO 2 ·OmcA particles before (black) and after (red) 10 min illumination (bottom). Experiments performed in anaerobic 150 mm MES (pH 6), 20 8C. Illumination l > 390 nm, power % 400 Wm À2 . dithionite, when peaks with maximaa t4 20 and 552 nm characteristic of the fully reduced proteins werer evealed (e.g.,F igure 6C top). Thus, the adsorbed proteins retained the spectral properties and redox activities of their soluble counterparts.
To assess the longevity of protein adsorption, TiO 2 ·MtrC(OmcA) particles were suspended in af resh 200 mL solution of 150 mm MES (pH 6) and incubated with occasional inversion for 20 ha t48C. The particles were then pelleted by centrifugation, and electronic absorbance spectroscopy of the supernatant provided no evidence for protein desorption. In contrast, protein was clearly present in supernatant recovered from centrifugation of TiO 2 ·MtrC(OmcA) particles incubated for 30 min at 4 8Ci n2 00 mLo f2 5mm phosphate, 150 mm MES, pH 6. Recoveredp articles were white. Electronic absorbance spectra of the desorbed proteins provided no evidencefor perturbation of protein structure by adsorption( Figure 6A,B, blue). It was concludedt hat adsorption of MtrC and OmcA in their native states was tight and essentially irreversible in 150 mm MES at pH 6b ut that the binding was reversed under conditions of competitive phosphate binding.
The maximum extentso fp rotein adsorption were estimated from the differences in electronic absorbance of protein solutions before and after incubation with TiO 2 particles, and from the electronic absorbance of the protein released from the particles on exposure to the phosphate-containing buffer (Figure 6A,B). Both methods quantified the amounto fM trC and OmcA that had been adsorbed as approximately 1.8 and 1.4 nmol, respectively,p er mg of TiO 2 nanoparticles. Taking the surfacea rea of the particles as 50 m 2 g À1 and the dimensions of MtrC and OmcA to be approximately 4 6 8nma nd 5 6 10 nm, respectively, [8] indicated that both proteins adsorbed at close to monolayer coverage.
The stable adsorption of MtrC(OmcA) on TiO 2 having been established, the possibility of visible-light-driven reduction of the adsorbed cytochromes by RuP-sensitizationo ft he TiO 2 ( Figure 2B)w as assessed. In the desired configuration,a dsorbed RuP should pass photoexcited electrons to adsorbed protein through the conduction band of the TiO 2 .R uP was consequently first adsorbed on the P25 TiO 2 particles to 20 % of its maximal coverage as described in the Experimental Section. The RuP·TiO 2 particles were then exposed to sufficient MtrC(OmcA) to saturatet he sites that werea vailablef or protein adsorption. Washed RuP·TiO 2 ·MtrC(OmcA) particlesw ere then resuspendedi na naerobic 150 mm MES (pH 6) and illuminated (l > 390 nm, 400 mW m À2 ). After this illumination the electronic absorbance spectra showed ac lear red shift of the Soret maximum and peaks with maximaa t4 20 and 552 nm (e.g.,F igure 6C,m iddle, bottom). These spectralc hanges revealed visible-light-driven reduction of MtrC(OmcA) adsorbed on the TiO 2 particles sensitized with RuP.

Photocatalytic reduction of solutions of MtrC, OmcA and MtrCAB by RuP·TiO 2 ·MtrC(OmcA) particles
Am otivation for this work was to inform strategies that might allow extracellular photosensitizers to generate photoenergized electrons that can enter MR-1 with the aid of extracellu-lar and outer-membrane-associated cytochromes ( Figure 1A), in order to drive reductive catalysis by intracellular enzymes. For the RuP·TiO 2 ·MtrC(OmcA) particles this requires that the adsorbed proteins be able to pass electrons to redoxp artner proteins in ap hotocatalytic process ( Figure 2C). Given that the coverage of the TiO 2 by MtrC(OmcA) approachedt hat predicted for am onolayer,a nd that no desorption was detected over 20 hi n1 50 mm MES (pH 6) (see above), we reasoned that the protein-coated particles offered very little opportunity for direct electron exchange to occur from the surface of the RuP·TiO 2 particles to proteins in solution. As ac onsequence, stirred,a naerobic suspensions of RuP·TiO 2 ·MtrC particles (0.037 mg mL À1 )w ere added to solutions of MtrC, OmcA, or MtrCAB ( % 6.2 mm heme) and illuminated to seek evidence for electron transfer through the adsorbed proteins to protein molecules in solution. Electronic absorbance spectroscopy revealed 75-85 %h eme reduction in each solution over 60 min illumination ( Figure 7, Table 1, and Figure S5). Particles recovered by centrifugationa tthe end of the experiment had ar ed color,a nd no reduction was observed when the experiment was repeated withouti llumination. It was concluded that the RuP·TiO 2 ·MtrC particles performed photocatalytic reduction of the cytochromes olutions( ca. one RuP particlep er 21 heme groups).P arallele xperiments establishedt hat RuP·TiO 2 ·OmcA particles performed photocatalytic reduction of MtrC, OmcA, and MtrCAB( Ta ble 1a nd Figure S7).
In principle, electron transfer from RuP to MtrC(OmcA) on the particles could occur by two pathways: [20f] through the conduction band of the TiO 2 or by direct RuP-to-protein electron transfer.R esultsf rom af inal series of experiments were consistentw ith electron transfer through the TiO 2 conduction band. No evidence forphotoreduction was observedw hen solutions of RuP and protein were illuminated in the absence of TiO 2 particles [RuP in 175-fold excesso fM trC, RuP with MtrC (OmcA or MtrCAB) at concentrations equivalent to those in the particle-containing suspensions, Figure S6].S imilarly,n op hotoreduction was detected when 25 mm phosphate was included in suspensionst hat contained RuP,T iO 2 ,a nd MtrC because phosphate at these concentrationsa dsorbed on the TiO 2 in preference to RuP and MtrC (see above). Direct TiO 2 -to-protein electron transfer was demonstrated when TiO 2 ·MtrC particles were seen to catalyze photoreduction of MtrC solutions when illuminated by UV light that excited electrons across the TiO 2 band gap ( Figure S8). These particles failed to catalyzep hotoreduction when illuminated by visible light (Figure S8), due to the absence of RuP that can be photoexcited by d-to-p* metal-to-ligand charge-transfer transition. [23][24][25][26] Taken together these results are consistent with facile visible-light-driven reductiono fM trC(OmcA)a fter chargei njectioni nto the conduction band of TiO 2 initiatedb yp hotoexcitation of the RuP.I n agreement with this conclusion, use of the integrating-sphere spectrophotometer showedn oe vidence for photoreduction of the MtrC adsorbed on TiO 2 particles in the absence of RuP during 60 min illumination with l > 390 nm ( Figure S9). This experiment also revealed at ime-dependent decreasei nt he apparent absorbance through the Soret region (< 450 nm) that is mostl ikely to arise from changes in light scattering by the particles. This behavior,r ather than damage to the protein, can account for the spectrald ifferences displayed by suspensions of (RuP·)TiO 2 ·MtrC particles in relation to solutions of MtrC, Figures 6C and 7.

Prospects for light-driven microbial synthesis
Enzymes are excellent catalysts for processes very relevant to developing the productiono fs olar fuels and chemicals. [4,[26][27] However,t ime-consuming and costly purification procedures, together with limited stability of the isolated proteins, often present bottlenecks to their effective utilization. By performing catalysis insideb acteria the need for enzymep urification is removed and catalyst self-repair and regenerationm ight be possible. [4,27] Furthermore, there are opportunities to improve on the efficiencies of naturalp hotosynthetic processes [28] by employing extracellularp hotosensitizers designed, for example, to absorb light across the solar spectruma nd to deliver photoenergized electrons to intracellular enzymes through EET pathways. Af ramework for informed development of light-driven microbial synthesis in MR-1 and heterologous hosts [14] is provided by the photoreduction of the MR-1 extracellulara nd outer-membrane-spanning cytochromes demonstrated here.
MtrCAB, MtrC, and OmcAa re redox-active overs imilar windows of electrochemical potential, and their complete photoreduction is triggered by eosin Y, fluorescein, and proflavine when combined with an appropriate SED. Comparable reduction of MtrCAB is achievedb yd ithionite, [6] and when MtrCAB spans the bilayer of al ipid vesicle it couples oxidation of external dithionite to reduction of internalized methyl viologen. [10] As ac onsequence it is thermodynamically feasible that the eosin-Y-, proflavine-, and fluorescein-dependent photoreduction of MR-1 extracellular cytochromesw ould support strongly endoergic intracellular reactions, because reduced methyl viol-ogen drives the reduction of CO 2 and water in well-established biochemical assays. The same thermodynamic predictions are made for the RuP·TiO 2 -dependent cytochrome reductions because these particles catalyze light-driven proton and CO 2 reductionb ym olecular catalysts and metalloenzymes. [26] Facile intracellular electrosynthesis of succinate by MR-1 [29] is driven by electron transfer from MtrCAB to ap eriplasmic fumarate reductase( E fumarate/succinate % 20 mV at pH 7). If the effective lightdriven microbial synthesis of additional products is to be achieved similarly,f acile electront ransfer to appropriate enzymes through natural or engineered [14] pathways will be required.
Clearly,t he effective translation of information gained in these studies into solar microbial synthesis by MR-1, or by bacteria engineered to contain the MR-1 outer-membrane-associated cytochromes, requires ac onsideration of many factors, not least the viability of the bacteria in the presence of photocatalytic concentrationso ft he photosensitizers. In this regard it is significant that FMN and RF are secreted by MR-1 and enhance EET. [11] It is also of note that MR-1 retains the ability to grow and to secrete RF in the presence of P25T iO 2 particles. [30] As ac onsequence, experimentst hat explore the possibilities of employing the photosensitizers described heref or visible-lightdriven synthesis by MR-1 are ongoing in our laboratories. Given that the outer-membrane cytochromeso fM R-1 evolved to deliver electrons to extracellular Fe III -containing mineral particles, an alternative strategyf or their photoreduction could employ nanocrystalline,s emiconductive Fe III oxide particles [31] as photosensitizers. It will be of interestt oe stablish whether such particles with appropriate conduction band energies and opticalp roperties can transfer electrons to MR-1 outer-membrane cytochromes.
Proteins:S oluble forms of MtrC and OmcA from S. oneidensis MR-1w ere purified as previously described. [8] Then, because both proteins were expected to contain a4 5-residue C-terminal extension that included aH is tag, the samples were passed through aN i-NTA column in HEPES (pH 7.6, 20 mm)/NaCl (100 mm). Most protein in samples of MtrC or OmcA failed to bind to the column, and that material, after buffer exchange, was used for the experiments reported here. LC-MS revealed molecular weights for the proteins that were consistent with the absence of the His tag, most likely due to proteolytic cleavage  [33] each as as uspension in Triton X-100 (2 %, v/v), were prepared as previously described. Purified proteins were stored at À80 8C, and their concentrations were defined by use of electronic absorbance spectroscopy of the oxidized (airequilibrated) proteins at 410 nm with extinction coefficients [mm À1 cm À1 ]o f1 000, 1000, and 2000 for MtrC, OmcA, and MtrCAB respectively.
Experiments with solutions of soluble photosensitizers:A naerobic solutions containing the desired protein with eosin Y, proflavine, fluorescein, RF,F MN, or FADw ere prepared to explore the possibility of protein photoreduction by soluble photosensitizers (Figure 2A). Experiments were performed in aN 2 -filled chamber (Belle Te chnology,c lear acrylic chamber with atmospheric O 2 < 10 ppm), and samples were illuminated from outside the chamber with aK L5125 Cold 150 Wl ight source with high-quality UV filter quartz glass (Krüss) fitted with a1 50 W( 15 V) halogen lamp (Osram). As ac onsequence the walls of the N 2 chamber served as an additional filter for light of wavelengths < 400 nm. Light intensity at the sample was calibrated by use of aS OLAR-100 Amprobe solar power meter (sensor wavelength 400-1100 nm). Stirred solutions of soluble photosensitizers and cytochromes at 5 mm heme concentration were illuminated in 1cml ight path cuvettes (Starna Scientific, special optical glass with > 75 %t ransmission above 320 nm). The high optical densities and quantities of proteins required for experiments with > 100 mm total heme precluded easy access to higher volume approaches. Instead these experiments were performed in unstirred solutions with 1mmp ath length cuvettes (Starna Scientific, special optical glass). Experiments that explored the mechanism(s) of OmcA photoreduction were performed at pH 7i np hosphate (50 mm)o rT EOA (50 mm)/phosphate (50 mm)o rH EPES (50 mm)/CaCl 2 (2 mm)/KCl (10 mm)a si ndicated. Electronic absorbance spectra were measured with aB iochrom WPAB iowave II diode array spectrophotometer located inside the N 2 -filled chamber.
Preparation and characterization of RuP-sensitized P25 TiO 2 particles coated with MtrC or OmcA:As tock dispersion of P25 TiO 2 nanoparticles [2 mg mL À1 in MES (pH 6, 150 mm), stored at 4 8C] was prepared as follows to ensure minimal aggregation in the wash steps. Particles (10 mg) were suspended in MilliQ water (1 mL), sonicated for 1min, and recovered as ap ellet after 5min centrifugation (9000 g,48C). The particles were then subjected to five rounds of resuspension, sonication, and centrifugation as above, in which the resuspensions were into MilliQ water (1 mL, rounds 1a nd 2), then MES (pH 6, 150 mm,1mL, rounds 3a nd 4), and finally MES (pH 6, 150 mm,5mL). Particles saturated with adsorbed RuP were prepared essentially as previously described. [34] In short, the stock dispersion of particles was sonicated for 10 min, and an aliquot was removed and incubated with excess RuP for 30 min on ice with occasional inversion [TiO 2 (0.1 mg) in RuP (33 mm,2 00 mL)/MES (pH 6, 150 mm)].T he RuP-coated particles were recovered by 5min centrifugation (9000 g,48C), and electronic absorbance of the supernatant quantified non-adsorbed RuP with use of the extinction coefficient 10.2 mm À1 cm À1 at 455 nm ( Figure S1). The adsorbed RuP was released by incubation of the particles in phosphate/MES (pH 6, 25 mm and 150 mm,r espectively) and quantified through the electronic absorbance of the supernatant after centrifugation to recover the white particles. These procedures confirmed the upper limit of adsorption as approximately 40 nmol RuP per mg TiO 2 under our conditions. Maximum adsorption of MtrC(OmcA) onto the particles was defined in as imilar manner after the desired mass of freshly sonicated particles had been incubated with protein (1 mm,2 00 mL)/MES (pH 6, 150 mm) (see the Results). The spectral integrity and redox activity of the adsorbed MtrC(OmcA) was assessed by electronic absorbance spectroscopy with aH itachi U4100 integrating-sphere spectrophotometer.
Particles coated with RuP and MtrC(OmcA) ( Figure 2B)w ere prepared by incubation with sufficient RuP to achieve 20 %m aximal coverage, and the resulting particles were exposed to sufficient MtrC(OmcA) to saturate the available binding sites. Freshly sonicated particles (1 mg) were pelleted by centrifugation (9000 g,4 8C, 5min) and resuspended in RuP (1 mL, 8 mm)i nM ES (pH 6, 150 mm). After 30 min incubation on ice with occasional inversion, particles with adsorbed RuP were recovered by centrifugation (2200 g,4 8C, 5min). Electronic absorbance spectroscopy of the supernatant confirmed adsorption of all the previously detectable RuP.P elleted particles were resuspended in MtrC(OmcA) (2 mm, 1mL)/MES (pH 6, 150 mm)a nd incubated for 30 min on ice with occasional inversion. Particles were recovered by centrifugation (2200 g,4 8C, 5min). Electronic absorbance of the supernatants confirmed that protein adsorption had saturated the available binding sites (typically 90-95 %o ft he maximum adsorption seen in the absence of RuP). For photoreduction experiments, particles coated with RuP and MtrC(OmcA) were washed by resuspension in MES (pH 6, 150 mm), recovered by centrifugation (2200 g,4 8C, 5min), and transferred to aN 2 -filled chamber (as above). Photoreduction of the protein co-adsorbed with RuP on TiO 2 particles ( Figure 2B)w as assessed with stirred anaerobic suspensions [0.1 mg mL À1 in MES (150 mm,p H6)] in sealed quartz cuvettes (1 cm path length) illuminated as described above. Cuvettes were removed from the N 2 -filled chamber for measurements of electronic absorbance with aH itachi U4100 integrating-sphere spectrophotometer.T he ability of particles coated with RuP and MtrC(OmcA) to pass electrons to non-adsorbed proteins ( Figure 2C)w as assessed with the particles (1 mg mL À1 ,3 5 mL) resuspended in an anaerobic solution (950 mL) of the desired concentration of MtrC, OmcA, or MtrCAB [detergent in the last case (Triton X-100, approximately 0.2 %, v/v)] and MES (pH 6, 150 mm). The stirred samples in sealed glass cuvettes (1 cm path length) were illuminated, and their electronic absorbances were measured inside the N 2 -filled chamber as described above.