Electrochemical Measurement of Electron Transfer Kinetics by Shewanella oneidensis MR-1*

Shewanella oneidensis strain MR-1 can respire using carbon electrodes and metal oxyhydroxides as electron acceptors, requiring mechanisms for transferring electrons from the cell interior to surfaces located beyond the cell. Although purified outer membrane cytochromes will reduce both electrodes and metals, S. oneidensis also secretes flavins, which accelerate electron transfer to metals and electrodes. We developed techniques for detecting direct electron transfer by intact cells, using turnover and single turnover voltammetry. Metabolically active cells attached to graphite electrodes produced thin (submonolayer) films that demonstrated both catalytic and reversible electron transfer in the presence and absence of flavins. In the absence of soluble flavins, electron transfer occurred in a broad potential window centered at ∼0 V (versus standard hydrogen electrode), and was altered in single (ΔomcA, ΔmtrC) and double deletion (ΔomcA/ΔmtrC) mutants of outer membrane cytochromes. The addition of soluble flavins at physiological concentrations significantly accelerated electron transfer and allowed catalytic electron transfer to occur at lower applied potentials (−0.2 V). Scan rate analysis indicated that rate constants for direct electron transfer were slower than those reported for pure cytochromes (∼1 s−1). These observations indicated that anodic current in the higher (>0 V) window is due to activation of a direct transfer mechanism, whereas electron transfer at lower potentials is enabled by flavins. The electrochemical dissection of these activities in living cells into two systems with characteristic midpoint potentials and kinetic behaviors explains prior observations and demonstrates the complementary nature of S. oneidensis electron transfer strategies.

Respiratory electron flow typically occurs at the inner membrane, where oxidation and reduction can be easily linked to intracellular electron carriers and used to generate a membrane potential. However, when the electron acceptor or donor is insoluble, bacteria must possess a mechanism for transferring electrons beyond their inner membrane (1). This is especially true for Proteobcteria, which have an outer membrane that further insulates cytoplasmic and inner membrane processes from insoluble substrates. Metal oxides (such as Fe(III) and Mn(IV) oxyhydroxides) are well recognized naturally occurring electron acceptors that demand such an electron transfer strategy (2)(3)(4).
Shewanella oneidensis MR-1, a metabolically versatile member of the gammaproteobacteria (5), is capable of reducing insoluble metals, and this phenotype has been linked to a collection of interacting multiheme cytochromes spanning the inner membrane, periplasmic space, and outer membrane (6 -12). There is also evidence that some of these cytochromes decorate the surface of pili-like structures extending from the cell surface (13,14). Regardless of the ultimate location of the cytochromes, in all models of electron transfer, electrons must hop from these proteins to a solid surface or be transferred to a soluble mediator that can diffuse to a final destination (15,16). Although chelation of a metal oxide is a third option (17,18), the fact that Shewanella is able to transfer electrons to solid graphite electrodes (19 -23) underscores the need for a direct or diffusion-based electron transfer mechanism to link cellular proteins and surfaces.
Recent work has shown that Shewanella species secrete soluble flavins (FMN and riboflavin) that facilitate electron transfer to both metals and electrodes (23,24). For example, removal of accumulated soluble flavins decreases the rate of electron transfer by Shewanella biofilms to electrodes over 80%. Consistent with this observation, kinetic measurements with pure MtrC and OmcA (25) showed that direct reduction of solid metal oxides by these cytochromes was too slow to explain physiological rates of electron transfer, whereas turnover rates of these enzymes with soluble flavins were orders of magnitude larger. These studies suggest that the kinetics of electron transfer from cytochromes on the outer surface of Shewanella to electrodes will be significantly altered in the absence of diffusible mediators (9 -11, 26 -34).
Voltammetry has proven a useful technique for the analysis of electron transfer rates and pathways using purified proteins (35)(36)(37)(38)(39) and has recently been extended to the study of intact bacteria (23, 40 -42). In slow scan rate cyclic voltammetry, the rate of electron transfer from respiring Shewanella biofilms to electrodes rises sharply at the E°Ј of riboflavin and FMN (Ϫ0.2 V versus SHE) 2 (23). Such measurements relating thermodynamic driving force to turnover kinetics would be difficult with whole cell:Fe(III) oxide incubations, which do not allow fine control over the electron acceptor redox potential or real time recording of electron transfer rates. In addition, voltammetry provides tools to observe electron movement under single-turnover conditions (in the absence of electron donor), allowing reversible oxidation and reduction of proteins accessible to the electrode and study of kinetic behavior (43,44).
In this work, techniques of turnover (sustained electron transfer from cells to electrode in the presence of electron donor) and single turnover (reversible oxidation and reduction in the absence of electron donor) voltammetry were harnessed to investigate the role of outer membrane proteins in electron transfer from Shewanella to electrodes. In all of these studies, intact metabolically active cells were used, along with electrode surfaces known to act as acceptors for Shewanella. The results in the absence of soluble mediators provide evidence that electron transfer between MtrC and OmcA and surfaces requires a higher potential compared with when flavins are present to shuttle electrons to the surface. Mutant analysis also demonstrates that cells possessing different outer membrane cytochromes have differing abilities for direct and mediator-enabled electron transfer.

EXPERIMENTAL PROCEDURES
Strains and Growth Conditions-For each trial, S. oneidensis MR-1 (ATCC 70050) or mutant cultures were grown from a frozen stock, first aerobically in Shewanella basal medium (described below) with 30 mM lactate acting as the electron donor. After 12 h, the culture was transferred into 0.4 liters of anaerobic Shewanella basal medium, containing (per liter) 0.46 g of NH 4 Cl, 0.225 g of K 2 HPO 4 , 0.225 g of KH 2 PO 4 , 0.117 g of MgSO 4 ⅐7H 2 O, and 0.225 g of (NH4) 2 SO 4 . Prior to autoclaving, 5 ml of a mineral mix (containing per liter 1.5 g of nitrilotriacetic acid, 0.1 g of MnCl 2 ⅐4H 2 O, 0.3 g of FeSO 4 ⅐7H 2 O, 0.17 g of CoCl 2 ⅐6H 2 O, 0.1 g of ZnCl 2 , 0.04 g of CuSO 4 ⅐5H 2 O, 0.005 g of AlK(SO 4 ) 2 ⅐12H 2 O, 0.005 g of H 3 BO 3 , 0.09 g of Na 2 MoO 4 , 0.12 g of NiCl 2 , 0.02 g of NaWO 4 ⅐2H 2 O, and 0.10 g of Na 2 SeO 4 ) as well as 5 ml of a vitamin mix (containing per liter 0.002 g of biotin, 0.002 g of folic acid, 0.02 g of pyridoxine HCl, 0.005 of thiamine, 0.005 g of nicotinic acid, 0.005 g of pantothenic acid, 0.0001 g of B-12, 0.005 g of p-aminobenzoic acid, and 0.005 g of thioctic acid) were added. To buffer against pH changes, 100 mM HEPES was added, and 15 mM lactate and 40 mM fumarate were provided as the electron donor and electron acceptor, respectively. Medium was sparged with N 2 gas that had been passed over a heated copper column to remove oxygen, adjusted to pH 7.2, and autoclaved. After autoclaving, 0.05% (w/v) filter sterilized casamino acids were also added.
To prepare thin films of attached cells, bacteria were grown to stationary phase (A 600 between 0.5 and 0.6), which represented depletion of the electron donor. The anaerobic culture was then transferred to centrifuge tubes in an anaerobic chamber (CO 2 :N 2 :H 2 20:75:5) and centrifuged at 5000 rpm for 20 min. In an anaerobic chamber, the pellet was gently resuspended in 20 ml of Shewanella basal medium that did not contain electron donor, electron acceptor, or casamino acids. The cells were centrifuged and resuspended again and used to inoculate sterile, anaerobic electrochemical reactors in an anaerobic glove bag. The electrochemical cells were incubated at 30°C under a stream of nitrogen gas. Base-line voltammetry experiments were performed, and electrodes were poised at ϩ0.24 V versus SHE for at least 6 h to facilitate attachment to electrodes and further depletion of intracellular donors (see "Electrochemistry of Attached Films" below for details on analysis of films). When cells were allowed to colonize electrodes over the course of several days, similarly grown cultures were directly inoculated into the same electrochemical chambers without any washing steps, and electrodes were poised at ϩ0.24 V versus SHE continuously.
Deletion Mutants-S. oneidensis MR-1 mutant strains containing markerless deletion constructs for mtrC and omcA were created as described in Ref. 45. Briefly, upstream and downstream regions of a targeted gene were ligated in a pSMV-3 plasmid containing counter-selectable markers, which was subsequently transformed into an Escherichia coli WM3064 mating strain. Following conjugation between the mating strain and S. oneidensis MR-1 wild type cells, recombination replaced the targeted gene with the deletion construct. Incubation under conditions favoring loss of the insert produced strains with deletions in the desired region. Mutations were verified using PCR with oligonucleotide primers that spanned the deleted region, and proper expression of nearby genes was verified by quantitative PCR.
Electrochemical Techniques-5X-AQ carbon (Poco Graphite Company, Decatur, TX) was used as the working electrode. Electrodes machined to 0.5 cm ϫ 2 cm ϫ 1 mm were prepared by polishing with 400 grit paper, rinsed, and sonicated in deionized water. The electrodes were cleaned for 24 h in 1 M HCl and stored in deionized water. 0.22-mm-thick platinum wire was attached to the electrodes via a 3.5-mm-long nylon screw (Small Parts Inc., Miramar, FL) and a nylon nut. The platinum wire was soldered to a length of insulated copper wire within a 3-mm-diameter glass capillary tube to provide a sealed outlet extending from the electrochemical cell. The resistance between the carbon electrode and the copper wire was always less than 0.5 ⍀.
The electrochemical reactors used in this study were 25-ml glass cones (Bioanalytical Systems, West Lafayette, IN). A Teflon top was altered to contain a butyl rubber O-ring (limiting oxygen entry) and a series of holes to allow working, counter, reference electrodes, and gas lines to be introduced. A glass body saturated calomel electrode (Fisher) was used as the reference electrode. A 5-mm-diameter glass capillary tube was capped with a nanoporous vycor frit (Bioanalytical Systems, West Lafayette, IN) connected to a larger diameter tube filled with 1% agar in 0.1 M sodium sulfate solution as a salt bridge.
Electrochemistry of Attached Films-After placing Shewanella suspensions in anaerobic reactors with electrodes, the working electrode was poised at ϩ0.24 V versus SHE, using a Bio-Logic VMP Multichannel Potentiostat. During growth and during catalytic voltammetry, reactors were stirred with a magnetic stir bar (Ax-Man, St. Paul, MN). Chronoamperometry was monitored until current levels fell to near background (Ͻ1 A/cm 2 ) to ensure that cells attached and were depleted of electron donors. At various time points, chronoamperometry was halted, and cyclic voltammetry (CV) was performed, until peaks in CV analysis became consistent. CV sweeps were performed from Ϫ0.56 to 0.45 V (versus SHE). The scan direction was then reversed and swept to the original value of Ϫ0.56 V. CV was performed at a scan rate of 1.0 mV/s. When peak heights had stabilized (typically within 6 h), working electrodes were washed to remove loosely attached cells by replacing the medium in the reactor. The medium was removed with a sterile nitrogen-flushed syringe and replaced with 5 ml of fresh anaerobic medium without electron donor or acceptor. After a brief period of stirring, the replacement process with fresh medium was performed again. Preliminary CV analysis after each wash showed that no changes in peak heights were observed after a total of five washes, and flavin peaks were also undetectable at this point. Three complete cycles of single turnover slow scan rate CV data were collected at 1 mV/s for each film. Scan rate analysis was then performed on cell films with a Gamry PCI4 Femtostat (Gamry Instruments, Warminster, PA). All of the sweeps were performed from Ϫ0.56 to 0.8 V (versus SHE) with a scan rate of 1-1000 mV/s with two sweeps/ scan rate. Following scan rate analysis, the electrodes were discarded. The data were analyzed (e.g. base-line removal, derivative analysis) using SOAS software (recently described in Ref. 46). Rate constants were also estimated using the program Jellyfit (courtesy of L. Jeuken).
Microscopy-Electrodes at various stages were placed in amber 1.5-ml centrifuge tubes containing 1 ml of medium with 60 M propidium iodide and 10 M Syto 9 from a Live/Dead BacLight bacterial viability kit (Invitrogen). The film was stained for 20 min, rinsed, and placed in the center of a microscope slide surrounded by a gasket slightly thicker than the electrode. The basin was filled with 50 l of medium, and a large coverslip was sealed in place (ColorStay, Revlon Cosmetics, Los Angeles, CA). The stained film was viewed with a Nikon Eclipse C1 confocal microscope using 488-and 561-nm filters. The morphology and coverage of the biofilm were evaluated at several locations, and representative sites were chosen for the collection of three-dimensional images using 0.5-m intervals, beginning 20 m above the top of the film and continuing until 20 m beyond the electrode/biofilm interface. The resulting slices were rendered using EZ-C1 3.20 FreeViewer software and ImageJ 1.34s.
A S3500N scanning electron microscope (Hitachi, Japan) was also used to image freshly polished electrodes and electrodes with attached cells. The electrodes were fixed in 2.5% glutaraldehyde and dehydrated for 5 min in 25, 50, 75, 95, and 100% ethanol. The electrodes then were dried in a Denton DV-502 vacuum evaporator (Moorestown, New Jersey), sputter-coated (Fullam EffaCoater, Clifton Park, New York), and fixed through graphite tape to a sample stage. The samples were placed directly into the scanning electron microscope vacuum compartment and examined at accelerating voltage of 5 kV.

Evidence for Two Potential Ranges for Electron Transfer in
Growing Shewanella Biofilms-It was recently discovered that Shewanella strains secrete redox-active flavins (24), which can accelerate the rate that electrons are transferred from cells to metals and electrodes (23,24). However, other observations showing interactions between purified cytochromes and solid electron acceptors suggested that a mechanism independent of flavins could also be functional. To characterize electron transfer across a range of applied potentials by S. oneidensis MR-1 under growth conditions, cultures were first grown on 2-cm 2 planar electrodes held at oxidizing potentials (ϩ0.2 V versus SHE) and analyzed by cyclic voltammetry (Fig. 1).
The concentration of soluble flavins in a S. oneidensis MR-1 culture can vary from 0.2 to 1 M, because of factors such as cell density and the length of incubation (23,24). To initiate growth in the presence of varying flavin concentrations, biofilms were amended with increasing levels of flavins. For example, when late exponential phase cultures of S. oneidensis MR-1 were used, flavin concentrations in the bioreactor were 0.1-0.2 M (FMN plus riboflavin). Over the next 96 h, anodic current rose (doubling nearly every 24 h), and leveled off at 35.6 A Ϯ 6.1 (n ϭ 3). When flavins were supplemented at inoculation, the rate of increase was faster (doubling time approaching 12 h for 1 M added flavins), and current stabilized at levels over 3-fold higher (110 A Ϯ 5.1, n ϭ 3).
Representative cyclic voltammetry data for biofilms grown in the presence of increasing levels of flavins are shown in Fig. 1A, with the derivative of each voltammogram shown in Fig. 1B, to allow visualization of midpoint potentials and wave symmetry in catalytic data. Voltammetry of all biofilms was dominated by a catalytic wave that initiated by Ϫ0.25 V and rose steeply at a midpoint potential of Ϫ0.2 V. The magnitude of this response increased with flavin level (23).
Interfacial electron transfer increases exponentially with the potential difference between donors and acceptors. Thus, small changes near the midpoint potential have a large effect on electron transfer rate, but as potential rises, reactions responsible for bringing electrons to the interface approach a limit. The sharp rise in the rate of electron transfer, centered at the mid- However, voltammograms always demonstrated a secondary response to applied potential, as shown by the slow, nearly linear rise in anodic current at potentials above Ϫ0.1 V in all experiments (Fig. 1A). The magnitude of this boost was similar, because current rose 10 -15 A as potential was increased from Ϫ0.1 to ϩ0.25 V. Although flavins altered electron transfer at the lower potential window (near Ϫ0.2 V), the higher potential increase always remained and was not sensitive to flavin concentrations. This suggested a second mechanism that was active at higher potentials, was not dependent upon flavins, and was broader in its response to potential.
To further characterize the effect flavins exerted on reaction rates in whole cells, Fig. 2 shows the relationship between added riboflavin and specific rates of electron transfer per unit of attached biomass (as A/g protein, 1 A ϭ nearly 10 pmol of electrons/s). Soluble mediators increased the amount of cell biomass that was able to colonize electrodes but also increased the specific rate each unit of biomass could transfer electrons to the surface. This further demonstrated that Shewanella possessed an excess of lactate oxidation activity and that higher specific rates of electron transfer per cell were achieved when riboflavin was also present to shuttle electrons to the surface.
Adsorption of Cells to Electrode Surfaces for the Study of Direct Transfer-The remainder of the experiments were aimed at explaining the origin of the higher potential electron transfer behavior, which was not affected by the level of soluble flavins. To more precisely study wild type and mutant strains, a method was developed to obtain repeatable electrodes interfaced with cell films. Identical electrodes were exposed to S. oneidensis that had been grown under anaerobic conditions. When an electrode (poised at ϩ0.24 versus SHE) was immersed in a cell suspension of S. oneidensis MR-1 lacking electron acceptors, cyclic voltammetry detected reversible peaks that stabilized in terms of height and location within 6 h. When similar experiments were performed with chloramphenicoltreated cells, identical peak development over time was observed (data not shown), ruling out synthesis of new proteins as a significant cause of the new voltammetry features.
Confocal microscopy of electrodes revealed a dense, 1-3m-thick layer of cells initially attached to surfaces (Fig. 3). However, when electrodes were washed five times (via five changes of anaerobic cell-free electrolyte), voltammetry features became lower in magnitude but more defined. Confocal microscopy of these washed electrodes revealed a thinning of the adsorbed cell layer, with individual cells and microcolonies (Fig. 3). Scanning electron microscopy also verified that electrodes retained less than a monolayer of intact cells, which often displayed thin cellular structures (Fig. 3). Thus, this protocol removed all but the most firmly attached cells, limited the heterogeneity of the cell-electrode interface, and maximized diffusion to the electrode surface.
A series of control experiments were also conducted, including exposure of electrodes to sonicated cells, sonicated cell extracts (supernatants or pellet fractions), and supernatants from cell suspensions that had preincubated for 6 h at 30°C to allow lysis or secretion of redox-active components to occur. These treatments tested the hypothesis that observed peaks were due to secreted proteins, compounds shed from cells, or products of cell lysis. None of these extracts from lysed, sonicated, or aged cultures were able to reproduce peaks or redox behavior observed with whole cells. These observations indicated that intact cells attached to electrodes were required to produce the electrochemical activity.
Metabolic Activity of Films on Electrodes-To confirm that these submonolayers of cells were metabolically active, lactate was added, whereas electrodes were held at an oxidizing potential (ϩ0.24 V versus SHE). The addition of 20 mM lactate immediately produced an anodic current, which stabilized within 0.5 h. At this point, cyclic voltammetry of electrodes revealed a catalytic response (Fig. 4, A), unlike those typically observed for Shewanella (such as those shown in Fig. 1A). Instead of a rapid onset centered at Ϫ0.2 V, current rose across a broad higher potential window.
When 1 M riboflavin was added to these lactate-oxidizing cell films, anodic current immediately increased. In the presence of riboflavin, cyclic voltammetry returned to the familiar pattern of low onset potential, with a steep increase in current centered at Ϫ0.2 V. However, as observed in natural biofilms, the broad linear rise in anodic current at higher potentials still remained. Thus, the behavior seen in naturally grown biofilms could be recreated in thin films, but the films produced a more homogeneous population of independent cells, without demanding growth of the cells (allowing study of mutants), all of which could also be washed to remove electron donors or flavins.
Cells lacking the outer membrane decaheme cytochromes OmcA and/or MtrC could attach to electrodes in a similar fashion and be tested for their ability to transfer electrons to electrodes (Fig. 4, B and C). Strains containing deletions in either of these cytochromes were severely impaired in rates of sustained electron transfer, most dramatically when MtrC was deleted. Because similar levels of attached cells were detected on all electrodes (14 g of cell protein/electrode of wild type and ⌬MtrC, 11 g of cell protein/electrode for ⌬OmcA), a failure to adhere could not explain these large differences in rates of electron flux. Single Turnover Voltammetry of Wild Type MR-1-Because films of live cells and mutants could be attached to electrodes, in the absence of both electron donors and electron shuttles, the reversible exchange of electrons between electrodes and proteins on the outer surface of cells could be probed for the first time. All of the following experiments were performed with at least three freshly attached thin films of S. oneidensis MR-1, in which the medium was exchanged repeatedly to remove loosely bound cells and soluble electron donors. In addition, all cells were incubated with electrodes poised at oxidizing potentials prior to analysis to deplete internal electron donor pools and facilitate attachment. Thin films were used once (e.g. for scan rate analysis) and discarded; all replicates were performed with freshly polished electrodes incubated with newly adsorbed cells.
Films were used to make three measurements: the estimated total number of accessible redox-active centers, the potential range of reversible electron transfer, and interfacial rate constants. The moles of electroactive redox centers/cm 2 discharged during a single sweep (coverage) was estimated by Equation 1, where Q is the total charge, in Coulombs, obtained by integration of peaks (dQ/dt, to adjust for scan rate), and n ϭ 1 (38). At slow scan rates (1 mV/s) the total sum of discharged electrons during a single sweep provided values as high as 3,000 pmol of electrons/cm 2 (for wild type). However, as scan rates exceeded 10 mV/s, total discharge of the film converged on a much lower value, approaching 250 pmol/cm 2 . As the electrode-interacting cytochromes were embedded in live cells, which still produced a small background level of anodic current, this larger peak area at slow scan rates could be due to additional electrons flowing in and out of periplasmic or quinone pools. However, the rapid diminishing of this "reservoir" of electrons at moderate scan rates (as shown by the stable integrated peak areas in Fig. 6A) prevented this pool from contributing to the higher scan rate data used to infer kinetics. Based on calculations of cytochromes packing onto a surface (such as the 85-kDa MtrC) and the rougher polishing treatment used in our experiments, coverage levels on the order of 50 pmol of electrons/cm 2 (accounting for the decaheme nature of MtrC) would be expected if the electrode were only in communication with proteins able to physically bind the electrode surface (11,35,47). The fact that our values at lower scan rates exceeded this range further suggested that proteins on the cell surface could rapidly exchange electrons from sites inside the periplasm or on the cell membrane, independent of the time scale of the experiment.
At slow scan rates (1 mV/s), reversible peaks were separated by a value of ϳ150 mV (Fig. 5). Such peak separation is common for adsorbed proteins and cytochromes (48). For films of wild type MR-1, discharge of the entire film spanned a window of at least 400 mV, which is similar to what has been observed in purified multiheme outer membrane cytochromes from Shewanella (11,29). However, the potential at which electron transfer from the cell surface occurred was centered nearly 0.1-0.2 V higher than what has been reported for most purified Shewanella cytochromes (11,29), suggesting further heterogeneity or differences in protein orientation when incorporated in the cell membrane.
An important observation throughout all of these experiments was that the height of base line-corrected peaks was proportional to scan rate (Fig. 6B). This relationship is the classical demonstration of thin-film behavior (49,50), as described by Equation 2.  OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42

Direct Electron Transfer by Shewanella
Stated differently, the entire pool of redox-active sites, ⌫, discharged completely, causing i p to vary as a function of . This was consistent with the rate of any electron delivery to the cellelectrode interface (e.g. from the cell interior to outer membrane cytochromes) being faster than the rate of interfacial electron transfer itself (the final hop to electrodes). With no diffusional limitations confounding the analyses, analysis at higher scan rates could be used to explore interfacial electron transfer behavior (49). Although this analysis has been applied to adsorbed compounds (49) and isolated proteins (35,51), these films of cells provided an opportunity to infer the same information from intact bacteria for the first time.
At scan rates as slow as 50 mV/s, peak separation became evident, as shown in Fig. 7. This peak separation was used to estimate a range for interfacial electron transfer rate constants between electrodes and surface-attached proteins, using both the Laviron approach, and the fitting program Jellyfit (29,49). As noted by others (29), because the complex multiheme proteins responsible for this electron transfer likely demonstrate dispersion and may represent averages from a population of  The data from ⌬mtrC is in red, that from ⌬omcA is in blue, and that from ⌬mtrC/⌬omcA is in green. The current vales are normalized to represent all electrodes containing 1 g of protein/cm 2 . The two similarly colored traces show data from two independent replicates, showing variability between films. Two sweeps were performed for each CV, and the second is shown. WT, wild type. redox centers, the values should be interpreted as ranges rather than as precise constants.
Significant peak separation always occurred at scan rates lower than those observed for pure proteins (Fig. 7). This indicated that the electron transfer pathway of whole cells behaved as a system with a k 0 more on the order of 1 s Ϫ1 . This peak separation behavior and estimated k 0 window (estimates ranging from 0.5 to 3 s Ϫ1 were obtained) were at least 2 orders of magnitude slower than what has been observed for isolated Shewanella cytochromes on electrodes but were similar to turnover rates reported for cytochromes in membrane preparations (as well as in whole cells) interacting with goethite (25).
Electron Transfer Behavior of MR-1 Mutants-As shown in Fig. 4, cultures containing deletions in omcA or mtrC were able to attach to electrodes but had different abilities to sustain electron transfer from cells. Because these mutants could not respire sufficiently to external acceptors to establish biofilms in long term electrode growth experiments (such as those shown in Fig. 1), the thin film procedure provided a new means to compare mutant constructs.
For example, when strains containing a deletion in the gene encoding the outer membrane decaheme cytochrome MtrC were adsorbed to electrodes, reversible peaks could still be detected, although the absolute height and integrated area under them was lower. Because the data in Fig. 5 has been normalized to account for slight differences in cell attachment, this indicated less accessible redox centers per cell in the mutant. The shape of the anodic peak was altered slightly and was centered at a potential 50 mV higher than the wild type. Similarly, when a mutant lacking the outer membrane decaheme cytochrome OmcA was adsorbed to electrodes, the total amount of redox activity decreased further, and the potential window shifted again (Fig. 5). Little difference was observed in peak separation data at higher scan rates, indicating that the k 0 trends for each cytochrome mutant were similar.
A double mutant lacking both outer membrane decaheme cytochromes (MtrC and OmcA) demonstrated much lower peak heights (ϳ10-fold lower than wild type or single outer membrane cytochrome deletion mutant, when adjusted for attached protein). The fact that this double mutant was still capable of a small amount of reversible electron transfer suggested that additional outer membrane redox active proteins remain to be identified (Fig. 5). Because of the small peak height, kinetic measurements were not made for this mutant.

DISCUSSION
Biochemical, genetic, and genomic analyses have shown that Shewanella strains share a conserved suite of cytochromes responsible for transfer of electrons from the inner to the outer membrane, where they are accessible to metals, mediators, or electrodes (7, 9, 12, 26 -29, 31, 52). Incubations of outer membrane cytochromes isolated from Shewanella have demonstrated that purified versions of these proteins will reduce metals (9 -11, 26 -34). The recent finding that Shewanella also secretes soluble flavins that accelerate metal or surface reduction showed that a soluble agent could help solve the problem of the last "hop" from proteins to surfaces (23,24). In this work, we have provided observations of electron transfer from intact cells without the apparent aid of flavins, shown that this electron transfer occurs at specific potentials, and provided estimates of rate constants for the process in living cells. Thus, both mechanisms of electron transfer may be operational, with the relative contributions determined by flavin concentrations, diffusional distances (for mediated transfer), and driving force (for direct transfer).
When enzymes (35,36,53) or bacteria (40,41,54,55) are linked to electrodes via pathways with rapid interfacial rates of electron transfer, an increase in applied potential causes an identical increase in the potential of attached enzyme redox center(s). This creates the classic sigmoidal i-V wave, which rises at a single redox potential and plateaus at a limiting current reflecting a limiting enzymatic turnover rate (42). In contrast, the more linear dependence of electron transfer rate with voltage observed in higher potential regions with Shewanella ( Figs. 1 and 4) suggested an interaction with slower or mixed interfacial kinetics. The response to soluble flavins in both naturally growing cells and cell films further illustrated that there was an excess of enzyme activity at the electrode but a bottleneck in the electron transfer pathway that could be alleviated by the presence of a soluble redox mediator.
The results obtained with intact cells of Shewanella allow some comparison with those obtained in vitro with recombinant proteins in two key parameters; apparent midpoint potentials, and estimated electron transfer rates. It is important to note that recombinant proteins are typically purified with affinity tags, expressed without lipid-binding domains, incubated in the presence of detergents, studied with soluble iron ligands, and/or are bound to electrodes with the aid of agents such as polymixin (10,27,29,30,56). Such treatments could alter the exposed surface of the protein, increase flexibility, allow closer contact, or facilitate more rapid turnover.
For example, MtrC has been shown in two separate studies to be reversibly oxidized at basal plane graphite electrodes across a wide potential range, similar to what was observed in our study with whole cells (of over 400 mV). Films of purified MtrC produced broad redox peaks centered at Ϫ138 mV (29) and Ϫ100 mV (11). Using cytochromes attached to scanning tunneling microscope tips, an equally wide potential window cen-  OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 28871 tered near Ϫ200 mV was observed for MtrC (30). Films of OmcA have also been reported to have potentials in the Ϫ150 to Ϫ100 mV range, using metal oxide or graphite electrodes (29,56).

Direct Electron Transfer by Shewanella
In contrast, direct electron transfer from intact Shewanella films to graphite electrodes described here required potentials near 0 V, or at least 100 -200 mV more positive than previously reported. Data from Shewanella sp. strain MR-4 biofilms also detected reversible electron transfer reactions in the same 0 V range (23). In this work, deletion of specific cytochromes altered the redox behavior of whole cells, which supported the hypothesis that we were observing electron transfer as catalyzed by these well studied outer membrane (or pili-attached) proteins.
A second comparison is related to rates of electron transfer. Estimates for interfacial rate constants in excess of 100 s Ϫ1 have been reported for purified MtrC (11,29). Such rates (in the 100 -1000 s Ϫ1 range) are highly typical of adsorbed enzymes studied by film voltammetry (38,57). Rates for cytochromes interacting with solid electron acceptors are more variable and are often based on measurements during electron transfer to solid substrates (expressed as k cat or rates observed under moderate driving force).
Using purified OmcA, both Eggleston et al. (56) and Xiong et al. (58) observed electron transfer to Fe(III) nanoparticles or oxide-coated electrodes on the order of 10 Ϫ13 mol e Ϫ cm Ϫ2 s Ϫ1 , equivalent to ϳ0.1 s Ϫ1 /cytochrome. Incubations with whole Shewanella putrefaciens and Fe(III) nanoparticles produced estimates of 0.001 s Ϫ1 (59), although only a percentage of cytochromes were likely reacting with Fe(III), causing this latter rate to represent a lower boundary. Similarly low rates of singleprotein turnover to goethite of 0.5-5 s Ϫ1 were recently provided by Ross et al. (25). Because an overpotential of at least 100 mV is present to estimate k cat , k o values for intact cells could be expected to be at least this low.
Our scan rate analysis data from whole cells provided independent estimates of interfacial electron transfer to external acceptors. Although these values represent averages of all electron transfer reactions involved in direct transfer for an intact Shewanella cell, they also represent a glimpse into the real world conditions experienced by a cell attempting utilize its intact electron transfer system to reduce this surface, with proteins confined to complexes within the membrane. All of our measurements were consistent with rates on the order of 1 s Ϫ1 or slower, which were 10 -100-fold lower than k 0 values observed for pure proteins adsorbed to electrodes (11,29).
When the maximum number of redox centers able to form a cell-electrode connection and rate constants from our scan rate analysis are combined, a prediction of the exchange current density (j 0 ) for the nonmediated pathway can be made. Assuming that the final step of electron transfer occurs through cytochromes with a size similar to OmcA or MtrC, the most electron transfer sites (regardless of whether cytochromes are on the cell surface or on cell pili) that can possibly be in physical contact with any electrode is on the order of 10 Ϫ12 mol/cm 2 (11,60). Based on estimated k 0 values for this pathway (ϳ1 s Ϫ1 ), this is equivalent to an exchange current between cells and electrodes of only ϳ1 ϫ 10 Ϫ7 A/cm 2 .
Values in this low j 0 range require significant overpotentials (of at least ϳ100 mV) simply to achieve a forward anodic current of the magnitude attributed to the non-flavin-mediated pathway (ϳ10 A/cm 2 ) (50). This requirement for overpotential is compounded by the broad potential window of the cytochromes, meaning that at a potential Ͻ0 V, only a fraction of redox sites are experiencing conditions that enable forward electron transfer. These calculations are also consistent with estimates by others (11,58), where films of pure OmcA subjected to moderate driving force supported electron transfer rates equivalent to only ϳ1.6 A/cm 2 .
Taken together, the boost in electron transfer in naturally grown biofilms above 0 V (Fig. 1) and the electron transfer observed in the same potential range with adsorbed Shewanella in the absence of flavins (Fig. 4) are likely due to cytochromeelectrode interactions. The relationship between applied potential and electron transfer at higher potentials is consistent with a step in the direct pathway being relatively slow, compared with reactions that supply electrons to this interface. This is also supported by the thin-film behavior relationship, which would be expected if reaction(s) feeding electrons to the interface were faster than the interface reaction itself.
In contrast, even small amounts of flavins (when present) can facilitate the transfer of electrons to electrodes at lower potentials and cycle back to Shewanella cell surface cytochromes. Diffusion limits the contribution of electron transfer by this mechanism, as shown by the strong relationship between soluble flavin concentration and anodic current (Figs. 1 and 4) and current/protein ratios (Fig. 2).
Implications for Shewanella-These results show that redoxactive proteins can be present on the outer surface of cells but not interfaced properly with the cell interior (e.g. as seen in MtrC mutants that produce peaks in Fig. 5 but no catalytic activity in Fig. 4). In addition, the cytochromes present on the outer surface appear to experience a bottleneck in electron transfer to solid phase electron acceptors, a conclusion also reached in recent experiments with purified OmcA and MtrC by Ross et al. (25).
Molecular computations of multiheme cytochromes (33) have shown that subtle variations in the orientation of hemes relative to a surface can slow rates of electron transfer by the same protein from nearly 100 to 0.01 s Ϫ1 . Because changes in the form of chelated Fe(III) (27), the nature of oxide surfaces (56), and even alterations to the orientation of hemes at a surface (33) can dramatically alter electron transfer rates, surfaces other than carbon electrodes may interact differently with the direct transfer pathway of Shewanella.
In the environment, the diversity of reducible oxide surfaces may select for flavin secretion as a means for cells to gain access to wider range of acceptors, without requiring the organism to express many different cytochromes tuned to a variety of unpredictable surfaces. Thus, flavins could act as a kind of "universal translator," allowing Shewanella to gain access to a diverse array of redox-active substrates, using only a single cytochrome-based pathway.
These observations also argue that, under conditions such as microbial fuel cells where anodes stabilize at low (Ϫ0.1 V) potentials, Shewanella would not compete with organisms that are capable of higher rates of respiration in the Ϫ0.1 V range (40,41,61) unless mediators accumulated to significantly facilitate electron transfer. Alternatively, if an anode was maintained at a higher potential (via use of a lower external resistance or potentiostat control), these observations provide a thermodynamic basis for enrichment of different populations of bacteria, based on the differing responses to potential in this relatively narrow range.