Quantification of Electron TransferRates to a Solid Phase ElectronAcceptor through the Stages ofBiofilm Formation from Single Cellsto Multicellular CommunitiesJ E F F R E Y S . M C L E A N , * , † G R E G W A N G E R , †

Y U R I A . G O R B Y , † M A R T I N W A I N S T E I N , †

J E F F M C Q U A I D , † S H U N ’ I C H I I S H I I , †

O R I A N N A B R E T S C H G E R , †

H A L U K B E Y E N A L , ‡ A N DK E N N E T H H . N E A L S O N † , §

The J. Craig Venter Institute, San Diego, CA, The Gene andLinda Voiland, School of Chemical Engineering andBioengineering and Center for Environmental, Sediment andAquatic Research, Washington State University, Pullman, WA,and University of Southern California, Los Angeles, CA

Received October 11, 2009. Revised manuscript receivedFebruary 8, 2010. Accepted February 13, 2010.

Microbial fuel cell (MFC) technology has enabled newinsights into the mechanisms of electron transfer fromdissimilatory metal reducing bacteria to a solid phase electronacceptor. Using solid electrodes as electron acceptorsenables quantitative real-time measurements of electrontransfer rates to these surfaces. We describe here an opticallyaccessible, dual anode, continuous flow MFC that enables real-time microscopic imaging of anode populations as theydevelop from single attached cells to a mature biofilms. Weused this system to characterize how differences in externalresistance affect cellular electron transfer rates on a per cellbasis and overall biofilm development in Shewanellaoneidensis strain MR-1. When a low external resistance (100Ω) was used, estimates of current per cell reached amaximum of 204 fA/cell (1.3 × 106 e- cell-1 sec-1), whilewhen a higher (1 MΩ) resistance was used, only 75 fA/cell(0.4 × 106 e- cell-1 sec-1) was produced. The 1 MΩ anodebiomass consistently developed into a mature thick biofilm withtower morphology (>50 µm thick), whereas only a thinbiofilm (<5 µm thick) was observed on the 100 Ω anode.These data suggest a link between the ability of a surface toaccept electrons and biofilm structure development.

IntroductionDissimilatory metal reducing bacteria (DMRB) can utilizesolid phase electron acceptors such as iron and manganeseoxides as electron acceptors to respire anaerobically via aprocess termed extracellular electron transport (EET). Thisability has important implications for metal cycling in theenvironment, the ecophysiology of these organisms, theirpotential use for bioremediation (1–3) and also microbialfuel cell applications (4, 5). Three different and perhaps

overlapping strategies for coordinating extracellular electrontransfer to solid phase electron acceptors are apparently used:direct cell-mineral contact with multiheme outer membranecytochromes serving as reductases (6), small molecular weightcompounds that function as electron shuttles (7, 8), orelectrically conductive bacterial pili or nanowires (9, 10).

Because of their EET abilities, DMRB are commonly usedin microbial fuel cells (MFCs) (4, 5), where they are capableof EET to MFC anodes as they anaerobically oxidize organicmatter. In particular, model organisms such as Shewanellaor Geobacter species have been used in efforts to unravel thebiological mechanisms involved with current production(6, 9–14). In such studies it is especially critical to quantifycell numbers as well as describe the depth and architecturesof biofilms as they develop on MFC electrodes. However,achieving this knowledge is difficult given typical MFC systemdesigns. Commonly used designs to maximize power gen-eration, cannot be used to accurately quantify the numberof cells per electrode surface area or the depth of the biofilm;often this can only be achieved at the terminus of theexperiment, which does not provide a direct link betweencell colonization/growth and power production. Further, thesampling methods chosen for such quantifications must bedelicately employed so that biomass or attachment is notaffected by mechanical disruption (15).

Destructive single end-point analyses such as total protein(14), and qualitative imaging using various microscopictechniques, render it impossible to study biofilm develop-ment, morphology, or architecture through time. Recentmathematical models of electron transfer in biofilms (16)indicate the importance of a conductive biofilm in the transferof electrons; therefore, biofilm parameters including depthmust be measured. Comparative studies of current generationmust be directly related to the number of cells on the electrodesurface and electron transfer rates. This is germane whenusing mutant strains, which may be altered in attachment,biofilm formation, and/or rates of EET. To determine therate of electron transfer per cell it is critical to first obtain anaccurate count of the number of cells on the anode. Theinability to accurately quantify attached cells and normalizefor them inhibits strict standardizations across samples,organisms, and MFC designs among the greater MFCcommunity and even within individual laboratories.

One solution to this dilemma is to merge the technologyused to measure electron transfer to solid surfaces with thetools for live, noninvasive imaging. A number of studies haveused optical flow cells and electrochemically controlledelectrode surfaces such as indium tin oxide (17) to investigatehow different potentials affect bacterial detachment (17),attachment, and/or growth rates (18). Recently Franks et al.(19) demonstrated the utility of a “mini-stack” MFC that wasmodified for nondestructive visualization of biofilm growth,a study that focused on changes of pH within anode biofilms.We describe here the construction and application of aparallel plate flow cell with incorporated fuel cell componentsthat accommodates real-time fluorescent microscopic imag-ing with real-time measurements of current production. Withthis system, we report for the first time, the electron transferrates on a per cell basis for Shewanella oneidensis MR-1interacting with anode electrodes; as well as the changes inanode potential through the development stages of a biofilmfrom single attached cells to a mature, multicellular, three-dimensional structure.

* Corresponding author e-mail: jmclean@jcvi.org.† The J. Craig Venter Institute.‡ Washington State University.§ University of Southern California.

Environ. Sci. Technol. XXXX, xxx, 000–000

10.1021/es903043p XXXX American Chemical Society VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 A

Materials and MethodsParallel Plate Flow Anode Chamber with Air Cathode. Theoptically accessible MFC was a modified, commerciallyavailable, biofilm flow cell (model no. FC-81; BiosurfaceTechnologies Inc.). These flow cells were modified byremoving the glass slide substrates and replacing them withelectrode materials. The anodes were made from superfineisomolded graphite plates (Graphite Store) that were shapedinto 5 × 25 × 1 mm electrodes fitted with 0.25 mm platinumwire leads and polished to a final mirror finish with 12000grit polishing paper. Silicone sealant was used to cover allsurfaces of the anodes except the polished top, yielding atotal accessible anode surface area of 1.25 cm2. The air-cathode consisted of Pt (5 g/m2) coated graphite felt (GDELT 120E-W, The Fuel Cell Store) coated with Nafion 117 film(Fluka) that was painted on to the surface and allowed toair-dry (20). By placing two anodes side by side in the samecompartment, we fit two MFCs into one chamber whichallowed us to compare their performances under identicaloperating conditions. Similar design was used by Dewan etal. (21), to compare power of many MFCs which had multipleanodes. The two anodes were connected to the shared air-cathode and cell voltage was measured individually across100 Ω and 1 MΩ resistors to determine current underoperating conditions (Iop) and current density (i) with a digitalmultimeter (Keithley 2700, Keithley Instruments Inc.). Theanode and cathode electrodes were separated from each otherby a nonconductive plastic slide. These MFCs were specif-ically designed to direct the growth medium across the anodesurfaces before flowing into the cathode compartment(Supporting Information (SI) Figure 1). At the operating flowrate of 200 µl/min the flow was calculated to be laminar(Reynolds number of 4.3). The chamber volume was 300 µLwith a calculated average flow velocity of 16 mm/mintranslating to a dilution rate of 0.67 min-1 or 1.5 volumechanges each minute.

Microorganism, Growth Medium, Buffer, and ElectrolyteSolution. Shewanella oneidensis strain MR-1 p519nGFP(constitutively expressed green fluorescent protein) was usedfor all experiments (22). For each experiment, cells from afrozen stock were streaked onto Tryptic-Soy Agar (Difco)plates augmented with Kanamycin and incubated for 18 hat 30 °C. These pregrown cells were transferred to 20 mL ofmodified M1 minimal medium (6) and incubated for 24 h at30 °C and at 150 rpm. To ensure an excess of the electrondonor lactate in all experiments the M1 medium wassupplemented with 60 mM lactate. Measurement of the MFCeffluent by high pressure liquid chromatography (HPLC)confirmed the excess of lactate (data not shown). To minimizethe concentration of oxygen in the system the influentmedium was continuously sparged with ultrahigh purity N2.Fresh medium was delivered into the MFC using a variablespeed peristaltic pump (404, Ismatec).

Flow Cell Operation. The MFCs where chemically steril-ized using a 10% peroxyacetic acid solution and flushed withsterile MFC buffer containing 50 mM PIPES, 100 mM NaClat pH 7.2. Prior to inoculation, background measurementswere taken in MFC buffer solutions, with and without lactate,as well as the minimal growth medium. Inoculation of theMFCs involved injecting the cells immediately upstream ofthe chamber and allowing them to flow into the chamberand attach to the graphite anodes for 10 min with the flowoff (batch mode). Following this, a high flow rate (i.e., 10 minat ∼2 mL/min) was set to flush out any unattached cells. Theflow rate was then decreased to the operational level of 200µL/min.

Measurements and Calculations. Potentiodynamic po-larizations were used to calculate power curves. These datawere used to determine the maximum current (Imax), maxi-mum current density (imax), and maximum power densities

(Pmax) of MFCs ((5)) based on the true surface area of theanode. The anode was the limiting electrode and cathodiccurrent was in excess. We used a potentiostat (Reference600, Gamry Instruments, Inc.) and a three-electrode systemwith the anode as the working electrode, the cathode as theauxiliary electrode and Ag/AgCl as the reference electrode.The anode potentials were scanned from the open circuitcell voltage (OCV) to a cell potential of 0 mV using a step sizeof 0.1 mV/s and a sampling time of 1 s. The cathode produceda stable cathodic potential of +230 mVAg/AgCl throughout theexperimental period as was measured before and after cellpolarization. The resulting increase in cell potential observedat open circuit indicates the anode potential becoming morenegative with respect to the cathode over time. A Student’st test was used for statistical comparisons of anode perfor-mance at 120 hrs, assuming a normal distribution and equalvariance. The internal resistance of the designed MFCs weremeasured with active biofilms using the ElectrochemicalImpedance Spectroscopy (EIS) method described by Dewanet al. (21), and found to be 40 Ω.

Microscopy (Epifluorescent, Confocal, Scanning Elec-tron). Epifluorescence. For real time imaging of the cells onthe anode surfaces a Zeiss Axio microscope with fluorescencecapabilities was used to image 10 random fields-of-view peranode surface at each set time interval to determine theaverage cell numbers. These counts were typically conductedwithout disruption of the operation of the MFC or flow ratesexcept for a brief interruption during the removal of the MFCfrom the microscope after the 18 h time point (Figure 1)resulting in a discharge of anode potential through a shortcircuit. Image processing was done by means of ImageJ andthe biofilm analysis package ISA 3D software (23). The effectof fluorescent imaging on the MFC’s was also conductedwith and without cells showing no change in the cell voltagedue to any of the excitation wavelengths used.

Confocal Microscopy. Microscopic observations wereperformed on a Leica TCSP5 confocal laser scanningmicroscope (CLSM) (Leica Microsystems). Image stacks wereobtained using a 10× objective. Fluorescence for GFP wascollected with 488 nm laser emission. Simulated 3-D imagesand sections (plan views) were generated using the softwareVolocity and the plan views with side profile slices usingIMARIS (Bitplane AG, Z|Aaurich, CH).

Scanning Electron Microscopy. At the termination of eachexperiment, anodes were fixed in a 2.5% solution of glut-araldehyde, processed through an ethanol dehydration series(i.e., 25, 50, 75, 100% v/v Ethanol, 0.5 h each treatment), thencritically point dried (815 Auto-Samdri, Tousimis) (10). Theanodes were mounted on aluminum stubs, coated withchromium and imaged in a Phillips XL30 Environmental SEM.

3.1. Results and DiscussionVariation in Biomass and Electrochemical Parameters withExternal Resistance. Using the dual anode real-time imagingcompatible MFC we successfully and repeatedly grew biofilmson the anode surface, monitored cell surface growth,quantified biomass and calculated operational current (Iop)with different external resistance. Experiments were veryreproducible, and we show here one representative data setfrom one typical experiment with fluorescent images (Figure1). The development of current over time followed a generaltrend for all repeated experiments with a lag in current forseveral hours followed by an exponential phase and arelatively stable current phase. The corresponding image data,showing how biomass changed during current production,revealed a consistent and striking difference developingbetween the biomass on the anodes. After 80 h, a thick andhighly structured biofilm had formed on the 1 MΩ anodewhile the biofilm on the 100 Ω remained as a thin layer ofcells. Using noninvasive CLSM imaging, the biofilm structure

B 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. xxx, NO. xx, XXXX

and thickness were quantified for both anodes. Thickerbiofilms with qualitatively more extracellular material wereobserved on 1 MΩ anode when compared to biofilms thatformed on the 100 Ω anode (Figure 2). The biofilms on the100 Ω anodes appeared as low mounds of cells rising only∼5 µm from the anode surface (Figure 2b) whereas the 1 MΩanodes produced a thick base layer with large “towers” (50-80

µm) (Figure 2a). This tower morphology is commonlyobserved in conventional continuous flow biofilm growthchambers where soluble compounds, such as oxygen orfumarate, serve as the terminal electron acceptor (22, 24).Images collected using SEM were consistent with thoseobtained using CLSM, despite an observed loss of biofilmbiomass during critical point drying for SEM analysis. These

FIGURE 2. Representative CLSM and SEM images of the anodes after 5 days of growth. CLSM images collected nondestructivelydemonstrate the biomass and structural differences between the 1 M Ω (A) and 100 Ω (B) anodes. Representative SEM of the 1 M Ω(C) and 100 Ω (D) processed by critical point drying confirm the structural differences in the biofilms as well as indicate that the 1M Ω colonies contain a greater amount of extracellular material between the attached cells within these colonies.

FIGURE 1. Simultaneous monitoring of current generation and biofilm development from single attached cells with epi-fluourescentimaging. Temporal current changes in the imaging compatible MFC for two different anodes with external resistances of 1 MΩ (A)and 100 Ω (B). C) Concurrent image data taken over the course of the experiment for both anodes. Differences in biofilmdevelopment were clearly visible by 88 h. Scale bars equal to 10 µm for 0 and 15 h and 100 µm for 18-130 h.

VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 C

data demonstrate that higher resistance (lower current flow)maximizes biofilm development or alternatively a lowerresistance(highercurrentflow)minimizesbiofilmdevelopment.

Current and Cell Density Correlation. Current produc-tion and cell counts were monitored from initial attachmentthrough the early stages of biofilm formation on the anode.Initial cell densities on the anodes averaged 2 × 105 cells/cm2 with virtually no change in numbers during the first 15 hof observation. During this initial period the anodes werecovered with only a single layer of cells but the current densityunder operation increased during this time (Figure 3),indicating an increase in specific anode respiration rate.Between 15 and 18 h, a rapid increase in cell numbers andcurrent was detected for both anodes. By 24 h cell densitiesreached 1.2 × 106 cells/cm2 with the 1 M Ω, and 8.8 × 105

cells/cm2 with the 100 Ω. Operational current density alsoincreased to 0.09 µA/cm2 (1 M Ω) and 0.08 µA/cm2 (100 Ω)over this period.

Specific Electron Transfer Rates. Current generation wascalculated on a per cell basis using the electrochemical andmicroscopic data obtained in real time (Figure 3). Thetemporal trends in these experiments indicate that the currentper cell maximum during operation was reached before thecells went into the rapid growth phase on the anode surface.The 100 Ω anode in the first run (Figure 1) increased initiallyto a maximum of 200 fA/cell (1.3 × 106 e- cell-1 sec-1) andthe 1 MΩ reached 80 fA/cell (0.4 × 106 e- cell-1 sec-1) (Figure3). These rates decreased to 63 fA/cell and 39 fA/cell,respectively, after the cell growth entered an exponentialphase. Replicate runs conducted over a longer time scale (75h) (SI Figure 2) demonstrated that the trends in cell growthand current generation were very consistent, although theabsolute values as well as the time required for rapid increasesin cell numbers and current densities showed some variability.

To our knowledge, these results represent the first timethe absolute values of current per cell over time have beenestimated from direct cell counts for a bacteria within anoperating MFC. These are also the first estimates of thespecific electron transfer rates to an anode surface forShewanella oneidensis strain MR-1. Previous studies haveestimated the current per cell from total protein concentra-

tions at selected time points during a batch operated MFCfor Geobacter sulfurreducens with values of 23 fmol/cell/day(14). At the values obtained for MR-1 on the anode operatingwith a 100 Ω resistor, we obtained ranges from 200 to 63fA/cell which corresponds to specific electron transfer ratesof 179 to 56 fmol/cell/day during continuous operation mode.

As noted above, the average cell concentration on theanodes during the first 15 h did not change significantly,whereas the current increased in a linear fashion during thisinitial period (Figure 3). To determine if, given enough time,the initial cells bound to the surface would produce detectiblecurrent without the components in the medium for growthand division, the experiment was modified such that afterinoculation and for the following 48 h, MFC buffer augmentedwith 60 mM lactate was substituted for the M1 growthmedium allowing respiration without growth. This experi-ment showed no observable increase in cell numbers or anydetectible increase in current during this time (data notshown). After 48 h the MFC buffer was switched to thestandard growth supporting M1 minimal medium and thecurrent and cell numbers began to increase as describedabove. This suggests that the increased current productionis dependent upon growth or perhaps newly synthesizedcellular components needed for extracellular electron transfersuch as redox active compounds, that is, more redox activecompounds per cell. While initial attachment on the twoanodes were in similar ranges, consistently more cellsattached to the 1 MΩ anode. At later time points, whendiscrete cells were no longer resolvable on the anode surface,the image analysis software package, ISA 3D (23), was usedto quantitatively compare the biomass differences anddetermine the correlation of colony size (average diffusiondistance) with current development (SI Figure 3).

Temporal Development of Anode Potential and Maxi-mum Current Density. Polarization measurements weremade at several times during the experiment, in an effort todetermine how the anode potential and maximum currentdensity (imax) correlated with biofilm development. For thesemeasurements (Figure 4), the system was placed in opencircuit for 2 h with uninterrupted flow, and the anodepotential monitored against an Ag/AgCl reference electrode(see SI Figure 1) to ensure the anode potential had stabilized.

The overall shape of the anode potential vs time curve(Figure 4) was observed to be similar between the two anodes.However, the 1 MΩ anode established a lower anode potentialand produced a higher imax more rapidly than the 100 Ωanode. The 100 Ω anode began to rapidly increase after 100 hand reached nearly the same potential as the 1 M Ω. However,neither polarization curve correlated with the timing of theIop increase trend and the period of exponential cell growth.The anode potential therefore became negative (reducing)and the imax increased in a linear fashion from 20 to 60 h withonly a monolayer of cells present on the anodes and did notcorrelate to the rapid current development period (30 to 60 h).At 20 h of growth with a single layer of cells and before rapidcell growth, the anode potential reached -150 mVAg/AgCl,consistent with the increasing current in the initial growthstages as shown in Figure 3. From 60 to 120 h the Iop of theMFC’s decreased slightly over time whereas the anodepotential continued to become more negative and theimaxincreased.

The higher cell potential and performance of the MFC atlater times is consistent with reported literature for batchstudies (6). Interestingly, in the early stages of the experiment,the potential and power densities are higher and areestablished sooner in the 1 MΩ anode than the 100 Ω anode,but by 120 h the 100 Ω anode approaches this value. Sincethe 100 Ω anode did not support a thick biofilm by 120 h,the development of its potential and Pmax may not necessarybe reliant on a thick biofilm suggesting that a limit on a

FIGURE 3. Representative timecourse measurements ofattached cell counts and current density during operation over24 h on the 1 M Ω and 100 Ω anode. A) Average cell countsper cm2 of anode area (left axis) and the current density duringoperation (right axis). B) Timecourse plot of the currentgenerated per cell based on direct in situ cell counts and thestable operating current.

D 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. xxx, NO. xx, XXXX

MFC’s power is reached prior to the maximum cell densityachieved on an anode.

Comparing MFC Performance Parameters with Pro-nounced Biomass Differences. At 120h, when the biofilmthickness differences between anodes were very pronounced,potentiodynamic polarization measurements were per-formed for five separate experiments (two anodes withineach chamber), and the mean of the data were used toquantify the differences in MFC performance with thedifferent external resistance (SI Figure 4). The stable operatingcurrent established after 120 h showed the largest measurabledifference between the two anodes, even while suffering fromhigh variability in the performance of the 100 Ω anode. The100 Ω anode statistically outperformed the 1 MΩ anode withmean stable operating currents of 1.2( 0.6 µA and 0.3( 0.03µA, respectively. The observed differences are consistent withstandard MFC theory that the maximum operating current,or power, is achieved when the fuel cell is operated underconditions where the internal and external resistances arenearly balanced (e.g., 100 Ω external and ∼40 Ω internal inthis system) (5).

The open circuit potential of the anodes at this later timepoint, although on average higher in the 1MΩ circuit, wasnot significantly different (p ) 0.061) at the 5% level (p )0.05), despite the clear difference in biofilm thickness (SIFigure 4). The maximum current density however, wassignificantly higher (p ) 0.027) on the 1 MΩ. The Pmax is onaverage higher in the 1 MΩ circuit (4.5 ( 1.0 mW/m2) thanthat of the 100 Ω (3.0 ( 0.9 mW/m2) though not significantlyso (p ) 0.084).

Overall, the current density and power density values fallinto a range of operating values for several reported microbialfuel cells using Shewanella and show our system is operatingas well as other designs using similar electrode materials.The most relevant comparison for a power measurementthat was believed to be predominately biofilm related wasreported by Biffinger et al. (25) at 0.52 mW/m2 for Shewanellastrain DSP10. They employed a Pt/C oxygen reductioncathode and attempted to control for a strict biofilmmeasurement with little influence of planktonic cells. Thepower densities reported by Biffinger et al. are comparableto those found in this study because both values werecalculated using the true surface area of the anode electrode,and not the underestimated “projected” surface area of aporous electrode, which can dramatically increase thecalculated power density (26).

DiscussionThe optical compatible dual anode system described herehas allowed, for the first time, real time, noninvasive estimatesof current per cell collected noninvasively over time in anoperating MFC. In the initial phase, the 100 Ω anode increasedto a maximum of 200 fA/cell (1.3 × 106 e- cell-1 sec-1) andthe 1 MΩ reached 80 fA/cell (0.4 × 106 e- cell-1 sec-1). Theserates decreased to 63 fA/cell and 39 fA/cell respectively afterthe cell growth entered a rapid phase of growth. The averagecurrent on a per cell basis over time for both anodes in theinitial stages fell within a range of values from 0.007 to 0.2pA per cell with the 100 Ω, which had a lower cell densityon the anode surface, outperforming the 1 M Ω. Onehypothesis as to why the current per cell decreases as the cellsurface growth enters exponential phase is that someelectrons are diverted to building cellular components ratherthan to respiration via the anode ((5)).

As cells were observed to populate the anode surface, arelatively consistent lag phase in current was seen beforerapid cell growth began and the stable current was attained.One possible explanation for the observed lag phase in currentproduction could be that initially the cells are not adaptedto transferring electrons to solid phase electron acceptorsand that upregulation of genes encoding specific proteinsand production of new components for this transfer arerequired. The cells from the liquid grown batch culture usedas the inoculum of the fuel cells are from the late exponentialphase and most likely the population is not homogeneousand can even include metabolically inactive cells that stillmaintain fluorescence from the long-lived GFP in this strain.Our cell counts would include the dead or inactive cells thatstill have intact membranes, which may or may not haveenzymatic activity and contribute to the overall current.Bretschger et al. (6) however, have previously shown thatmembrane fractions of cells alone do not produce current.Overall there are difficulties associated with obtaining a truevalue for the current per cell and therefore these datarepresent estimates that are relevant to the specific conditionsof our system. Using a different fluorescent construct thatharbors a short-lived GFP however, may allow an estimateof anode respiration rates for only the active cell population.

Previous studies using noninvasive NMR spectroscopyon live biofilms of S. oneidensis MR-1 under flowing condi-tions in similar minimal medium, quantified the anaerobicmetabolism rates with lactate and various soluble electronacceptors (i.e., fumarate, DMSO, NO3) (22) as well asmetabolite profiles within an MR-1 biofilm (27, 28). It was

FIGURE 4. Development of anode potential and performance over the course of biofilm development. Current vs time for (A) 1 M Ωand (B) 100 Ω anodes. The spikes in cell voltage observed in panels A and B represent the times where the fuel cell was in opencircuit and when the polarization data were collected. Results from polarization resistance from anodes in A and B over time; (C)development of anode potential. (D) development of maximum current density.

VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 E

found that even aerobically grown biofilms could immediatelyreduce fumarate, DMSO or NO3 without prior exposure tothese acceptors or adjusting the oxygen concentrations toanaerobic conditions. These reduction studies were con-ducted under bulk aerobic conditions where oxygen becomesrapidly limited at depth within a metabolically active biofilm.Here we have quantified the electron transfer to solid electronacceptor and the changing potentials from single attachedcells though to mature biofilms. Quantification of parameterssuch as the decreasing anode potential (becoming morenegative with respect to the reference electrode) over timeindicated that the cells can establish reducing conditionsearly in the growth phase while still as a single cell layer. Forboth the thick biofilms on the 1 MΩ anode as well as the thinbut active 100 Ω anode biofilm, the anode potential increasedrapidly when the circuit was opened, reaching reducingconditions within 2 h. Presumably this is due to the cell’soxidation of lactate coupled to the reduction of all availableredox active compounds within the cells and possibly anyextracellular redox components. It is very interesting thatcurrent was produced and anode potential was measurableabove background under constant flow with only a thindispersed monolayer on the anode. Given that this was underrelatively high laminar flow conditions and high dilution rates,nonadhered soluble redox active compounds, such aselectron shuttles, would be of minimal use with single cellson the surface. Of particular note is that the maximum rateof electron transfer to the anode occurred at cell densitiesthat did not completely saturate the available anode surfacearea. We suspect that a very low level of oxygen is also presentin solution since maturation of fluorescent proteins actuallyrequires the presence of molecular oxygen. Fluorescenceformation of GFP is prevented by rigorously anoxic conditions(<0.75 µM O2) but is readily detected at 3 µM O2 ((29)). Thissuggests that the cells (as a single cell layer) on the surfaceare limited for oxygen as the acceptor and/or prefer the anodeas the electron acceptor; an intriguing idea that warrantsfurther investigation.

MFC performance parameters, determined from powercurves, were averaged for five independent runs. The 120 htime point was chosen for the comparisons since that timeshowed the most pronounced difference in biomass andstable operating current (SI Figure 4) between the two anodes.These results may be of use in determining whether thick orthin biofilms are optimal for current generation in MFCapplications. The averages at 120 h show that the cellpotential, imax, and Pmax is higher in the 1MΩ circuit. However,at 120 h, only the imax, was shown to be significantly higherat the 5% level (p ) 0.05). Further study is needed to fullycharacterize if these parameters are significantly increasedwhen a larger biofilm is present on the anode.

Although we have documented that varied externalresistance reproducibly results in major biofilm differences,the explanation of these differences remains unknown.Previous work with MR-1 biofilms in flow cells (on inertsurfaces) showed their development into complex structuresin nutrient rich (24, 30) and in a defined minimal medium(22) grown with oxygen as the major electron acceptor. Thedevelopment and stability of these structures in MR-1 canvary depending on the availability of a soluble alternativeelectron acceptor such as fumarate (22). Noninvasive tem-poral development of MR-1 into biofilms from single attachedcells to mature biofilms has not been well documented orpreviously assessed on an anode of a microbial fuel cell orin correlation with electron transfer rates. It is possible thatthe availability of an electron acceptor (potential differencebetween the anode and cathode) or alternatively the rate atwhich electrons can be passed to an accepting surface(resistance) can drive the building of biomass. Increasingthe number of cells or redox active compounds to generate

a more reducing environment would be an effective strategyto more quickly reach conditions to enable reduction of solidphase oxides. The observed rapid biomass accumulation andthe more negative anode potential observed on the harderto reduce (higher resistance) surface supports this notion.

In summary, we have successfully demonstrated a newMFC design to couple imaging technology with microbialfuel cells enabling insights into the physiology of microbialcells as they develop into a biofilm especially as it pertainsto the movement of electrons from cells to a solid acceptor.The analogy of these systems to the interaction of microbesin the environment with solid mineral phases should not beoverlooked. Microbial fuel cells stand as a technology thatcan be used to investigate the physiology of the cells on asolid electron accepting surface and as a model system forquantifying cell-mineral electron transfer. Several modelshave been proposed detailing the mechanisms of electrontransport in MR-1 to solid metal oxides such as directattachment, electron shuttles and, the most recently dis-covered, microbial nanowires however the exact mechanismsand the relative contributions of each are unclear. Furtherstudies with optical compatible systems may help decipherthe role of various mechanisms in solid phase electrontransfer. In particular the use of real-time fluorescent genereporters of the various redox proteins, pili, and electronshuttles would help understand their temporal and spatial(depth) related expression profiles.

AcknowledgmentsThis research was supported by the J. Craig Venter Institute;the Legler-Benbough Foundation; and the Office of Science(BER), US Department of Energy with funds from theEnvironmental Remediation Science Program (grant DE-FG02-08ER64560).

Appendix A

NomenclatureIop current under operating conditions (A)i current density based on the true anode surface

area (A/m2)imax maximum current density based on the true surface

area of the anode (A/m2)Pmax maximum power density based on the true surface

area of the anode (watts/m2)

Supporting Information AvailableFour figures intended to supplement the material presentedin the manuscript. Diagram of the optically compatible MFC,replicate timecourse measurements for current per cellestimates, biofilm structure and current relationships, and5 day MFC performance parameter comparisons. Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) Fredrickson, J. K.; Romine, M. F.; Beliaev, A. S.; Auchtung, J. M.;

Driscoll, M. E.; Gardner, T. S.; Nealson, K. H.; Osterman, A. L.;Pinchuk, G.; Reed, J. L.; et al. Towards environmental systemsbiology of Shewanella. Nat. Rev. Microbiol. 2008, 6 (8), 592–603.

(2) Nealson, K. H.; Saffarini, D. Iron and manganese in anaerobicrespiration: environmental significance, physiology, and regu-lation. Annu. Rev. Microbiol. 1994, 48, 311–43.

(3) Lovley, D. R.; Coates, J. D. Bioremediation of metal contamina-tion. Curr. Opin. Biotechnol. 1997, 8 (3), 285–9.

(4) Logan, B. E. Exoelectrogenic bacteria that power microbial fuelcells. Nat. Rev. Microbiol. 2009, 7 (5), 375–81.

(5) Logan, B. E.; Hamelers, B.; Rozendal, R.; Schroder, U.; Keller,J.; Freguia, S.; Aelterman, P.; Verstraete, W.; Rabaey, K. Microbialfuel cells: Methodology and technology. Environ. Sci. Technol.2006, 40 (17), 5181–92.

F 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. xxx, NO. xx, XXXX

(6) Bretschger, O.; Obraztsova, A.; Sturm, C. A.; Chang, I. S.; Gorby,Y. A.; Reed, S. B.; Culley, D. E.; Reardon, C. L.; Barua, S.; Romine,M. F.; et al. Current production and metal oxide reduction byShewanella oneidensis MR-1 wild type and mutants. Appl.Environ. Microbiol. 2007, 73 (21), 7003–12.

(7) Marsili, E.; Baron, D. B.; Shikhare, I. D.; Coursolle, D.; Gralnick,J. A.; Bond, D. R. Shewanella secretes flavins that mediateextracellular electron transfer. Proc. Natl. Acad. Sci. U. S. A.2008, 105 (10), 3968–73.

(8) Lies, D. P.; Hernandez, M. E.; Kappler, A.; Mielke, R. E.; Gralnick,J. A.; Newman, D. K. Shewanella oneidensis MR-1 uses overlap-ping pathways for iron reduction at a distance and by directcontact under conditions relevant for Biofilms. Appl. Environ.Microbiol. 2005, 71 (8), 4414–26.

(9) Reguera, G.; McCarthy, K. D.; Mehta, T.; Nicoll, J. S.; Tuominen,M. T.; Lovley, D. R. Extracellular electron transfer via microbialnanowires. Nature 2005, 435 (7045), 1098–101.

(10) Gorby, Y. A.; Yanina, S.; McLean, J. S.; Rosso, K. M.; Moyles, D.;Dohnalkova, A.; Beveridge, T. J.; Chang, I. S.; Kim, B. H.; Kim,K. S.; et al. Electrically conductive bacterial nanowires producedby Shewanella oneidensis strain MR-1 and other microorgan-isms. Proc. Natl. Acad. Sci. U. S. A. 2006, 103 (30), 11358–63.

(11) Reguera, G.; Nevin, K. P.; Nicoll, J. S.; Covalla, S. F.; Woodard,T. L.; Lovley, D. R. Biofilm and nanowire production leads toincreased current in Geobacter sulfurreducens fuel cells. Appl.Environ. Microbiol. 2006, 72 (11), 7345–8.

(12) Bond, D. R.; Lovley, D. R. Electricity production by Geobactersulfurreducens attached to electrodes. Appl. Environ. Microbiol.2003, 69 (3), 1548–55.

(13) Srikanth, S.; Marsili, E.; Flickinger, M. C.; Bond, D. R. Electro-chemical characterization of Geobacter sulfurreducens cellsimmobilized on graphite paper electrodes. Biotechnol. Bioeng.2008, 99 (5), 1065–73.

(14) Ishii, S.; Watanabe, K.; Yabuki, S.; Logan, B. E.; Sekiguchi, Y.Comparison of electrode reduction activities of Geobactersulfurreducens and an enriched consortium in an air-cathodemicrobial fuel cell. Appl. Environ. Microbiol. 2008, 74 (23), 7348–55.

(15) Thormann, K. M.; Saville, R. M.; Shukla, S.; Spormann, A. M.Induction of rapid detachment in Shewanella oneidensis MR-1biofilms. J. Bacteriol. 2005, 187 (3), 1014–21.

(16) Torres, C. I.; Marcus, A. K.; Parameswaran, P.; Rittmann, B. E.Kinetic experiments for evaluating the Nernst-Monod modelfor anode-respiring bacteria (ARB) in a biofilm anode. Environ.Sci. Technol. 2008, 42 (17), 6593–7.

(17) Hong, S. H.; Jeong, J.; Shim, S.; Kang, H.; Kwon, S.; Ahn, K. H.;Yoon, J. Effect of electric currents on bacterial detachment andinactivation. Biotechnol. Bioeng. 2008, 100 (2), 379–86.

(18) Busalmen, J. P.; de Sanchez, S. R. Changes in the electrochemicalinterface as a result of the growth of Pseudomonas fluorescensbiofilms on gold. Biotechnol. Bioeng. 2003, 82 (5), 619–24.

(19) Franks, A. E.; Nevin, K. P.; Jia, H.; Izallalen, M.; Woodard, T. L.;Lovley, D. R. Novel strategy for three-dimensional real-timeimaging of microbial fuel cell communities: monitoring theinhibitory effects of proton accumulation within the anodebiofilm. Energy Environ. Sci. 2009, 2, 113–119.

(20) HaoYu, E.; Cheng, S.; Scott, K.; Logan, B. Microbial fuel cellperformance with non-Pt cathode catalysts. J. Power Sources2007, 171 (2), 275–281.

(21) Dewan, A.; Beyenal, H.; Lewandowski, Z. Scaling up microbialfuel cells. Environ. Sci. Technol. 2008, 42 (20), 7643–8.

(22) McLean, J. S.; Majors, P. D.; Reardon, C. L.; Bilskis, C. L.; Reed,S. B.; Romine, M. F.; Fredrickson, J. K. Investigations of structureand metabolism within Shewanella oneidensis MR-1 biofilms.J. Microbiol. Methods 2008, 74 (1), 47–56.

(23) Beyenal, H.; Donovan, C.; Lewandowski, Z.; Harkin, G. Three-dimensional biofilm structure quantification. J. Microbiol.Methods 2004, 59 (3), 395–413.

(24) Thormann, K. M.; Saville, R. M.; Shukla, S.; Pelletier, D. A.;Spormann, A. M. Initial phases of biofilm formation in Sh-ewanella oneidensis MR-1. J. Bacteriol. 2004, 186 (23), 8096–8104.

(25) Biffinger, J. C.; Pietron, J.; Ray, R.; Little, B.; Ringeisen, B. R. Abiofilm enhanced miniature microbial fuel cell using Shewanellaoneidensis DSP10 and oxygen reduction cathodes. Biosens.Bioelectron. 2007, 22 (8), 1672–9.

(26) Ringeisen, B. R.; Henderson, E.; Wu, P. K.; Pietron, J.; Ray, R.;Little, B.; Biffinger, J. C.; Jones-Meehan, J. M. High power densityfrom a miniature microbial fuel cell using Shewanella oneidensisDSP10. Environ. Sci. Technol. 2006, 40 (8), 2629–34.

(27) McLean, J. S.; Ona, O. N.; Majors, P. D. Correlated biofilmimaging, transport and metabolism measurements via com-bined nuclear magnetic resonance and confocal microscopy.ISME J. 2008, 2 (2), 121–31.

(28) Majors, P. D.; McLean, J. S.; Fredrickson, J. K.; Wind, R. A. HMRmethods for in-situ biofilm metabolism studies: Spatial andtemporal resolved measurements. Water Sci. Technol. 2005, 52(7), 7–12.

(29) Shaner, N. C.; Steinbach, P. A.; Tsien, R. Y. A guide to choosingfluorescent proteins. Nat. Methods 2005, 2 (12), 905–909.

(30) Teal, T. K.; Lies, D. P.; Wold, B. J.; Newman, D. K. Spatiometabolicstratification of Shewanella oneidensis biofilms. Appl. Environ.Microbiol. 2006, 72 (11), 7324–30.

ES903043P

VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 G