Electrosynthesis of Commodity Chemicals by an AutotrophicMicrobial Community

Christopher W. Marshall,a Daniel E. Ross,b Erin B. Fichot,b R. Sean Norman,b and Harold D. Maya

Department of Microbiology and Immunology, Marine Biomedicine and Environmental Science Center, Medical University of South Carolina, Charleston, South Carolina,USA,a and Department of Environmental Health Sciences, University of South Carolina, Columbia, South Carolina, USAb

A microbial community originating from brewery waste produced methane, acetate, and hydrogen when selected on a granulargraphite cathode poised at �590 mV versus the standard hydrogen electrode (SHE) with CO2 as the only carbon source. This isthe first report on the simultaneous electrosynthesis of these commodity chemicals and the first description of electroacetogen-esis by a microbial community. Deep sequencing of the active community 16S rRNA revealed a dynamic microbial communitycomposed of an invariant Archaea population of Methanobacterium spp. and a shifting Bacteria population. Acetobacteriumspp. were the most abundant Bacteria on the cathode when acetogenesis dominated. Methane was generally the dominant prod-uct with rates increasing from <1 to 7 mM day�1 (per cathode liquid volume) and was concomitantly produced with acetate andhydrogen. Acetogenesis increased to >4 mM day�1 (accumulated to 28.5 mM over 12 days), and methanogenesis ceased follow-ing the addition of 2-bromoethanesulfonic acid. Traces of hydrogen accumulated during initial selection and subsequently accel-erated to >11 mM day�1 (versus 0.045 mM day�1 abiotic production). The hypothesis of electrosynthetic biocatalysis occurringat the microbe-electrode interface was supported by a catalytic wave (midpoint potential of �460 mV versus SHE) in cyclic volta-mmetry scans of the biocathode, the lack of redox active components in the medium, and the generation of comparatively highamounts of products (even after medium exchange). In addition, the volumetric production rates of these three commoditychemicals are marked improvements for electrosynthesis, advancing the process toward economic feasibility.

The U.S. economy is heavily reliant on the use of fossil-basedcarbon to produce many commodity chemicals and fuels.

However, due to supply difficulties, the inevitable decline of theseresources, increased world demand, and environmental concerns,a shift away from coal and oil to alternative energy sources such asnatural gas, solar, and wind is occurring. However, most of theseenergy sources are either limited by fluctuations in price and avail-ability or are nonrenewable, as in the case of natural gas. Thesefactors have encouraged research into the development of renew-able energy technologies powered by microbes. Of particular in-terest are microorganisms that can capture the global greenhousegas CO2 and convert it to a valuable commodity such as a fuel.

Bioelectrochemical systems (BESs) include microbial fuel cells(MFCs), microbial electrolysis cells (MECs), and electrosyntheticbiocathodes (4, 17, 18, 27). Of these, the bioanodes of MFCs andMECs have been the most intensively investigated. The newestand arguably most promising of these technologies is the genera-tion of valuable chemicals by electrosynthesis. Microbial elec-trosynthesis requires microorganisms to catalyze the reduction ofCO2 by consuming electrons on a cathode in a BES.

The purpose of the present study was to establish a sustainablebiocathode from a mixed microbial community that could fixCO2 and convert it to a commodity chemical. This was achievedby selecting cathodophilic microorganisms from an inoculum ob-tained from brewery wastewater. Acetate, methane, and eventu-ally hydrogen were the predominant products repeatedly formedat a cathode potential of �590 mV versus the standard hydrogenelectrode (SHE). Electrochemical evidence suggests that electrontransfer between the electrode and microbes in a biofilm is oper-ating in the absence of soluble redox active components in themedium. This is the first report of the concomitant production ofacetate, methane, and hydrogen through electrosynthesis. Theproduction rates of these compounds surpass what has been re-

ported in the literature and was accomplished at a comparativelyhigh cathode potential.

MATERIALS AND METHODSSource of microorganisms and initial screening. The biocatalysts de-scribed here were enriched from samples taken from a retention basin forbrewery wastewater at Palmetto Brewing Company in Charleston, SC. Toscreen for initial product formation, the brewery wastewater sludge wasused to inoculate 20-ml chambers of small BES reactors equipped withgraphite rod cathodes. The reactors were poised from �1,000 to �400mV versus SHE with the goal of selecting for the highest rate of productformation at the highest potential to limit energy input into the system.Products (acetate and methane) were detected after 28 days of incubationat �590 mV and again after the medium had been exchanged once. Con-trols without voltage applied were monitored for production due to fer-mentation of the wastewater. Once production free of fermentation wasindicated, inocula from these reactors were then transferred to largerthree-electrode BES reactors described below in order to further enrichand evaluate the electrosynthetic community.

Three-electrode bioelectrochemical systems. The BESs consisted oftwo identical custom designed glass chambers (Chemglass Life Sciences,Vineland, NJ) that had two crimp-seal, butyl rubber sampling ports, athreaded o-ring sealed port for the reference electrode, and a clampedo-ring junction for the membrane (see Fig. S1 in the supplemental mate-rial). The total volume of the glass chamber was 150 ml. The two glasschambers were separated by a proton exchange membrane (Nafion 117;

Received 31 July 2012 Accepted 18 September 2012

Published ahead of print 21 September 2012

Address correspondence to Harold D. May, mayh@musc.edu.

Supplemental material for this article may be found at http://aem.asm.org/.

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/AEM.02401-12

8412 aem.asm.org Applied and Environmental Microbiology p. 8412–8420 December 2012 Volume 78 Number 23

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

The Fuel Cell Store, Boulder, CO) and sealed with an o-ring and clamp.The reference electrode was Ag wire coated with AgCl and immersed in 3M KCl saturated with AgCl (�210 mV versus SHE). All potentials arereported versus SHE. Both glass chambers contained 30 g (dry weight) ofpretreated graphite granules of heterogeneous sizes ca. 10 by 5 by 3 mmand smaller (Showa Denko). Granules were on average 2 g/ml. A0.9525-cm diameter by 3-cm long pretreated graphite rod current collec-tor connected to a 0.065-cm titanium wire was buried in the graphitegranule bed. The graphite electrodes were pretreated by first sonication indeionized water and then washed with acetone, 1 M hydrochloric acid, 1M sodium hydroxide, and deionized water in succession to remove or-ganic and metal contamination.

The cathode chamber (biotic) was filled with 75 ml of freshwater me-dium containing per liter: 2.5 g of sodium bicarbonate, 0.6 g of sodiumphosphate monohydrate, 0.25 g of ammonium chloride, 0.212 g of mag-nesium chloride, 0.1 g of potassium chloride, 0.03 g of calcium chloride,20 ml of vitamin solution, and 20 ml of mineral solution. The vitaminsolution contained (per liter): 2 mg of biotin, 2 mg of folic acid, 10 mgof pyridoxine-HCl, 5 mg of thiamine-HCl·2H2O, 5 mg of riboflavin, 5 mgof nicotinic acid, 5 mg of D-Ca-pantothenate, 0.1 mg of vitamin B12, 5 mgof p-aminobenzoic acid, and 5 mg of lipoic acid. The mineral solutioncontained (per liter): 1.5 g of nitrilotriacetic acid, 3 g of MgSO4·7H2O, 0.5g of MnSO4·H2O, 1 g of NaCl, 0.1 g of FeSO4·7H2O, 0.152 g ofCoCl2·6H2O, 0.1 g of CaCl2·2H2O, 0.085 g of ZnCl2, 0.01 g ofCuSO4·5H2O, 0.02 g of KAl(SO4)2·12H2O, 0.01 g of H3BO3, 0.01 gof NaMoO4·2H2O, 0.03 g of NiCl2·6H2O, 0.3 mg of Na2SeO3·5H2O, and0.4 mg of Na2WO4·2H2O. The anode chamber (abiotic) contained a sim-ilar medium composition but without the vitamins or minerals and withincreased potassium chloride to 1 g/liter and sodium chloride to 2 g/liter.The medium was prepared under anaerobic conditions (80:20 [vol/vol]N2-CO2) and passed to the chambers of the BES in an anaerobic glove bag(Coy Laboratory Products). After transfer of the medium, the BESs wereremoved from the anaerobic chamber, and the headspace was flushedwith 80:20 (vol/vol) N2-CO2 before inoculation. The BESs were operatedin batch mode at 25 �2°C, and medium exchanges were accomplished bydecanting over 90% of the liquid volume, leaving only the granules andwhat liquid remained in the granular electrode bed. The medium ex-changes and subculturing were done in an anaerobic chamber by trans-ferring �10 ml of liquid and a small amount (1 to 5 g) of graphite granulesfrom the current-consuming, product-producing reactor into sterileBESs. Where noted, BESs were flushed with 100% CO2 by using a longneedle aseptically pierced through the stopper into the liquid and anothershort needle in the headspace as gas effluent. To inhibit methanogenicArchaea and enrich for acetogens, 10 mM 2-bromoethanesulfonic acidwas added at the time of a medium exchange to a reactor actively produc-ing methane and acetate by electrosynthesis.

Electrochemistry. During most of the experiments the cathode waspoised chronoamperometrically at �590 mV. On day 28 of biocathodeoperation, the replicate working electrodes (cathodes) shown in Fig. 2were subjected to cyclic voltammetry (CV). The scan range of the CV wasfrom �200 mV to �1,000 mV, and the scan rate was 1 mV/s. All electro-chemistry was performed using a VMP3 potentiostat (Bio-Logic USA).Coulombic efficiencies were calculated by dividing coulombs found in theproduct (Cp) by total coulombs consumed (CT). Cp � b·n·F, where b is thenumber of electrons in the product, n is the number of moles of product,and F is Faraday’s constant (96,485 C/mol). CT was calculated by integrat-ing the area under the current-versus-time curve (i-t curve).

Analytical methods. Fatty acids were measured using a high-pressureliquid chromatography (HPLC) apparatus (Shimadzu) equipped with aUV detector at 210 nm. The mobile phase was 0.005 M H2SO4 and had aflow rate of 0.55 ml/min through an Aminex HP-87H column (Bio-Rad,Hercules, CA). Methane and hydrogen were measured on a HP6890 gaschromatograph (GC) equipped with an HP-PLOT Molesieve 5A column(30 m by 530 �m by 25 �m) and a thermal conductivity detector. Theoven was held at 50°C for 2 min and then increased by 25°C/min to 170°C

and held for 0.2 min. The injector temperature was 120°C, and the detec-tor was temperature 250°C. Argon was the carrier gas.

Scanning electron microscopy (SEM). Graphite granules from thecathode were fixed in 2% glutaraldehyde in 0.1 M sodium cacodylatebuffer for 3 h. The granules then underwent a 2.5% osmium tetroxidepostfix wash for 1 h. The granules were then dehydrated by a series ofethanol washes (25, 50, 75, 95, and 100%). The samples were sputtercoated with gold and palladium with a 100-Å coating (Denton Vacuum).Images were taken with a JEOL JSM-5600LV scanning electron micro-scope.

RNA extraction. Samples for RNA extraction were either collecteddirectly into TRIzol (Invitrogen; for MEC granules) or concentrated ontoa Sterivex filter (Millipore; PES membrane, 0.22-�m pore size, for MECsupernatant), which was then stored in TRIzol. Samples in TRIzol wereincubated at room temperature for at least 15 min and then frozen at�80°C until further processing, as outlined in the supplemental material.

RT-PCR amplification and 16S rRNA sequencing. Reverse transcrip-tion (RT) was carried out with 100 ng of total RNA using random hexa-mers (SuperScript III; Life Technologies) according to manufacturer’sinstructions. PCR was performed with either universal bacterial or ar-chaeal primers for the V1-V3 or V2-V3 region of 16S rRNA (see Table S1in the supplemental material) with the following final concentrations: 1�Green GoTaq reaction buffer, 1 mM MgCl2, 0.2 mM deoxynucleosidetriphosphates, 0.2 �M forward primer mix (equal molar concentrationsof degenerate and less-degenerate primer), 0.2 �M reverse primer, 0.625U of Taq polymerase (Promega), and 0.5 �l of RT reaction per 25-�l PCRvolume. Two replicate PCRs were carried out with each of the two follow-ing cycling protocols (for a total of four replicates) to maximize primingcoverage. The first protocol consisted of an initial denaturing step (94°C,5 min), 10 amplification steps (45 s each of 94°C, 62°C [decreasing 0.5°Cper step], and 72°C), an additional 15 amplification steps (45 s each of 94,57, and 72°C), followed by a final 10 min extension at 72°C. The secondprotocol designed to target GC-rich templates (19) is the same as the first,except that all annealing steps were performed for 6 s instead of 45. AllPCR replicates were pooled (four total), cleaned (PCR cleanup kit;Qiagen), and quantified (NanoDrop). Amplicons were sequenced on aPacBio-RS sequencer (Engencore, LLC) using a 45-min run time andstandard protocols (11). Pacific Biosciences FASTAQ formatted circularconsensus sequences have been submitted to the GenBank sequence readarchive under SRA056302.

Taxonomic classification. Sequences were preprocessed and analyzedusing Mothur v.1.25 and v.1.27 (29, 30). Briefly, sequences with any of thefollowing features were removed: average quality score of �25, anoma-lous length (�300 or 615 bp), an ambiguous base (quality score � 1),8 homopolymers, or 1 mismatch to the barcode or primer. Remainingreads were dereplicated, grouped with similar fragments, and alignedagainst the Greengenes core database (7) using kmer searching (8mers)with Needleman-Wunsch global, pairwise alignment methods (20). Theprimers were then trimmed from each read: the B27f primer correspondsto Greengenes alignment positions 109 to 136, A109f corresponds to po-sitions 455 to 493, and U529r corresponds to positions 2232 to 2260.Resulting reads shorter than 300 bp or those likely due to sequence error(14) or chimeras (9) were removed. The reads were then classified using aBayesian approach and bootstrap cutoff of 80 (34) against the SILVAdatabase (26).

RESULTSEstablishing an autotrophic biocathode. A three-electrode BES(see Fig. S1 in the supplemental material) was inoculated from abrewery waste culture that was initially screened in a small two-electrode BES. The three-electrode BES was operated for 3 monthsat a fixed cathode potential of �590 mV. The electrode was themicrobial community’s only electron donor and CO2 its only car-bon source for growth throughout all experiments. During thefirst 10 days of incubation, the reactor generated 1.8 mM acetate,

Microbial Electrosynthesis of Commodity Chemicals

December 2012 Volume 78 Number 23 aem.asm.org 8413

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

followed by 2.6 mM methane, over 30 days as the main productsfrom CO2 fixation before the first exchange of the spent growthmedium (Fig. 1A). Production rates reached 0.18 mmol of acetateper liter of cathode liquid volume per day (mM day�1) and 0.12mM day�1 methane during this initial startup. Subsequently, aftersuccessive medium exchanges, methanogenesis became the dom-inant process and reached 0.78 mM day�1.

As CO2 was consumed and reduced to methane, the pH in thecathode chamber would frequently exceed 8 (Fig. 1A). To remedythis, 100% CO2 was flushed through the reactor for 30 min, whichthen lowered the pH of the medium to �6.5. Unexpectedly, thisCO2 flush also revived the production of acetate. The increase inacetogenic activity after CO2 flushing resulted in rates reaching1.02 mM day�1 with accumulation of 9 mM in the cathodechamber over 17 days. Methanogenesis also increased in responseto the flushing of CO2, reaching a rate of 1.58 mM day�1. Duringthe 17 days after the start of CO2 flushing the coulombic efficiencyreached 84% (Fig. 1B). To the best of our knowledge, this is the

first time the coproduction of acetate and methane has been dem-onstrated electrosynthetically.

Replication of the autotrophic biocathode. An importantquestion regarding microbial electrosynthesis resides in the abilityto generate sustainable and transferable production rates. After 92days of operation, supernatant and granules were transferredfrom the initial reactor into three replicate BESs poised at �590mV. After a lag period of �1 week, product formation began toincrease. Once again, acetate and methane were the predominantproducts in the replicates; however, the acetate production ratewas much lower than that of electromethanogenesis (Fig. 2A).Although acetogenesis did not disappear as it did early on in Fig. 1,the rates were not able to compete with methanogenesis, irrespec-tive of the periodic flushing of the cell with 100% CO2. Over a10-day period following the initial lag phase, acetate accumulatedto 1 mM and methane accumulated to 10 mM. The acetate pro-duction rate was 0.1 mM day�1, and the methane production ratereached 1.3 mM day�1 during this period. The coulombic effi-ciency of the replicates reached 60% (Fig. 2B).

Abiotic (sterile) reactors were also poised at �590 mV to de-termine whether the abiotic accumulation of hydrogen would besufficient to account for the methane and acetate observed underbiotic conditions (Fig. 2C). At the start of each experiment, �0.3mM hydrogen was immediately produced due to the initial polar-ization of the cathode. However, from that point forward theabiotic hydrogen production rate was observed at �0.045 mMday�1 over 20 days with a coulombic efficiency ranging from 53 to64%. Thus, this rate of production cannot account for the mMday�1 rates of methane and acetate production observed in any ofthe biotic BESs. Coulombs may have been lost in the biotic andabiotic BESs due to gas leakage through joints in the reactor, bub-bles trapped in the graphite bed, and in the case of the biotic BESselectrons accumulated into biomass. Despite the portion of elec-trons unaccounted for, the total coulombs consumed in the bioticreplicates far exceeded what was calculated in the abiotic BESs(Fig. 2B and D), indicating microbial catalysis that could not beexplained by abiotic hydrogen formation.

Increased rates of electrosynthesis. The rates of methane oracetate production could be increased by further enrichment ofthe electrosynthetic biocathodes or by adding a selective inhibitor.After 29 days of operation with repeated medium exchanges (be-ginning in Fig. 2), the rate of methanogenesis increased, the co-production of acetate continued, and eventually hydrogen (andoccasionally a small amount of formate) was produced (Fig. 3A).The rate of methanogenesis was consistently 1.6 mM day�1 andreached a maximum 7 mM day�1, accumulating to 1.5 mmol inthe headspace. The acetate production rate remained near thatobserved in the initial BES reactor (Fig. 1A), close to 1 mM day�1.Hydrogen did not accumulate to any significant degree until afterextended incubation in the experiments documented in Fig. 1 and2. This was also the case for the experiments presented in Fig. 3where the microbial community had been further enriched andhad experienced multiple medium exchanges. Electrohydrogen-esis again lagged behind methanogenesis but suddenly after 7 daysof reactor operation increased dramatically to more than 4 mMday�1, eventually reaching 11.8 mM day�1 and accumulated to1.5 mmol (Fig. 3A). Although hydrogen lagged behind methano-genesis, once it started it was produced concurrently with meth-ane. Also, after an extended lag, formate and acetate eventuallywere formed at rates of 1 mM day�1 in the methanogenic reactors.

FIG 1 Development of an electrosynthetic biocathode at �590 mV versusSHE. (A) Operation of a BES for 108 days. Complete replacement of the me-dium was completed on days 30, 36, 50, 57, and 91. The BES was flushed with100% CO2 for 30 min on the days marked with gray arrows. (B) Distribution ofcoulombs in products compared to total coulombs consumed after the firstflushing of CO2.

Marshall et al.

8414 aem.asm.org Applied and Environmental Microbiology

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

The electron recovery (coulombic efficiency) in methane, acetate,formate, and hydrogen was 54% (Fig. 3B). Subsequent transfercultures in replicate BESs of this community following the estab-lishment of hydrogen production have continued to perform sim-ilarly to what is presented in Fig. 3A, generating methane, hydro-gen, acetate, and formate.

Coproduction of acetate and methane was observed through-out the study (Fig. 1A, 2A, and 3A), but methanogenesis usuallyoutcompeted acetogenesis. This changed upon the addition of themethanogenic inhibitor 2-bromoethanesulfonic acid (Fig. 3C),which resulted in acetogenesis increasing to as high as 4 mMday�1. This rate of activity was sustained in the absence of methan-ogenesis with subsequent transfers of the treated culture to otherBESs. Acetate production started 2 days after medium exchangeand inhibitor addition and then increased over the next 10days, accumulating to 28.5 mM. After a lag of 7 days, hydrogenbegan to be produced by the community and was then gener-ated concomitantly with acetate (similar to what occurred inthe methanogenic reactor in Fig. 3A). The overall rate of hy-drogen production was 2 mM day�1 but reached rates of over 9mM day�1 and accumulated to 1.8 mmol in the headspace.Electron recovery in acetate and hydrogen from the 2-bromo-ethanesulfonic acid treated community was 67% (Fig. 3D). Thebiotic production of hydrogen in the reactor with the inhibitorand in the one without (Fig. 3A and C) exceeded abiotic pro-duction by at least 200-fold (Fig. 2C and D).

Electrochemical evaluation of the biocatalyst. Cyclic voltam-

metry (CV) was performed on the BESs in order to discern possi-ble redox active components associated with the biocathodes. Noredox peaks were detected in the abiotic (uninoculated) reactors,indicating a lack of electron shuttles in the medium (Fig. 4, blackline). Current production in the abiotic scan was very low at �590mV and consistent with the low rate of proton reduction observedat this potential over an extended time period (Fig. 2B and D). TheCV scan of the abiotic reactor stood in sharp contrast with thecatalytic wave seen in the three replicate BESs with live biocath-odes producing methane, acetate, and hydrogen (Fig. 4, red line).The onset of catalytic current during the reductive scan of a bio-cathode was at �340 mV and plateaued at �640 mV versus SHE.The midpoint potential of the catalytic wave was �460 mV, whichonly varied slightly (approximately �30 mV) between replicates.The current draw at the peak of the catalytic wave was � 5 mA. Inorder for the noncatalyzed abiotic BES to reach the same currentoutput, a potential of �900 mV or less was required. The300-mV discrepancy between peak current in the biotic scanstrongly supports microbial catalysis of electrode oxidation.

When supernatant (spent media) from the replicate BESs werefiltered and inserted into an abiotic reactor, no redox active peakswere observed (Fig. 4, green line). Since no redox active componentswere observed in the fresh medium or in the filtered supernatant, it isunlikely that a soluble mediator was responsible for electron transferfrom the electrode to the microorganisms at �590 mV.

Electrosynthetic microbial community composition. SEMwas used to visualize the prevalence of microorganisms attached

FIG 2 Replication of biocathodes at �590 mV. (A and B) Coproduction of acetate and methane (A) and coulombs consumed (B) after transfer of the brewerywaste biocathode. (C and D) Production of hydrogen (C) and coulombs consumed (D) at �590 mV in abiotic (sterile, uninoculated) control BESs. The BESswere flushed with 100% CO2 for 30 min on days 7 and 11. Error bars indicate the standard deviations (n � 3).

Microbial Electrosynthesis of Commodity Chemicals

December 2012 Volume 78 Number 23 aem.asm.org 8415

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

to the electrode. Biofilm formation was seen on the graphite gran-ule cathodes from untreated BESs producing acetate and methane(Fig. 5A). The dominant morphology was mostly of rod-shapedmicrobes varying in size from 2 to 5 �m long. Another, thicker rodshaped organism was also observed. These thicker rods were �1�m long and less prevalent. However, when the biocathode wastreated with 2-bromoethanesulfonic acid, these thicker rod-

shaped microbes were the dominant morphology on the electrode(Fig. 5B). The observation of these microorganisms on the cath-ode is consistent with the evidence from the CV, which is support-ive of microbial catalyst acting at the surface of the electrode.

To assess the composition of the active microbial populationwithin the electrosynthetic community, total RNA was extractedfrom samples taken from supernatant or graphite electrode gran-ules at day 91 when acetogenesis was predominant and day 108when methanogenesis was predominant as shown in Fig. 1A.Overall, in the culture supernatant, the predominant bacterialphyla were Bacteroidetes, Deferribacteres, Firmicutes, Proteobacte-ria, Spirochaetes, and Synergistetes (Fig. 6). At day 91, when aceto-genesis was the predominant activity, members of the Sulfurospi-rillum genus accounted for 62.3% of the bacterial reads sequencedin the supernatant with another 15.9% belonging to the genusWolinella. A modest change occurred on day 108, when methano-genesis was the predominant activity, with Sulfurospirillum spp.remaining as the most abundant but decreasing to 36.0%. Mem-bers of the genus Wolinella increased to 22.8% and members of thefamily Spirochaetaceae increased from 9.5 to 24.2%.

A more dramatic change in the active bacterial population wasobserved with the samples extracted off the graphite granule elec-trodes. Acetobacterium spp. were relatively minor members of thesupernatant community, but when acetate was the major product(day 91) the percentage of Acetobacterium on the electrode rose to60.3% (Fig. 6). When methane again dominated and acetate pro-duction was low (day 108), the Acetobacterium spp. decreased to

FIG 3 Increased rates of electrosynthesis. Two of the replicate BESs described in Fig. 2 were incubated further with two more medium exchanges, the last on day29. (A) Production of acetate, methane, hydrogen, and formate in one BES maintained without inhibitor. (B) Distribution of coulombs consumed and in allproducts observed in panel A. (C) Production of acetate and hydrogen in a second BES with 2-bromoethanesulfonic acid added. (D) Distribution of coulombsconsumed and in all products observed in panel C. The BESs were flushed with 100% CO2 on days 33 and 36.

FIG 4 Cyclic voltammetry on abiotic (black trace), cell-free supernatant(green trace), and biotic (red trace) BESs. A biotic scan performed on day 28 onreplicate reactors shown in Fig. 2. Scan rate, 1 mV/s.

Marshall et al.

8416 aem.asm.org Applied and Environmental Microbiology

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

4.7%. An unclassified family (WCHB1-69) from the Sphingobac-teriales represented 8.0% of the active population on the electrodeat day 91 but became the dominant bacteria at day 108 (37.7%). Incontrast, the abundance of WCHB1-69 was relatively constant atapproximately 4 to 7% in the supernatant at days 91 and 108. Alsofound on the cathode on day 91 were members of the family Rho-dobacteraceae (8.0%) and the genus Sulfurospirillum (7.4%). Ad-ditionally, on day 108, rRNAs of the Synergistaceae family (11.1%)and the Spirochaetaceae family (17.4%) were detected on thecathode.

The predominant archaeal sequences were from the genusMethanobacterium, constituting 93% of the total sequenced ar-chaeal reads, regardless of whether the supernatant or electrodeswere examined or when the samples were taken. It is important to

note that while acetogenesis was predominant at day 91, me-thanogenesis was also occurring at both the day 91 and the day 108time points. Methanobrevibacter represented �5% of the readsand unclassified sequences made up a low percentage of total ar-chaeal reads(�1%). See the supplemental material for additionaldetails of the phylogenetic results.

DISCUSSION

An autotrophic microbial community from brewery wastewaterwas selected on a cathode of a bioelectrochemical system for theproduction of valuable commodity chemicals. Methane, acetateand hydrogen were all sustainably and reproducibly generatedelectrosynthetically at a cathode potential of �590 mV versusSHE. Each of these products has been generated with microbial

FIG 5 Scanning electron micrographs of electrosynthetic cathode biofilms when primarily methanogenic after 148 days (electrode from the same reactor shownin Fig. 1) (A) and acetogenic after treatment with 2-bromoethanesulfonic acid (day 56, the electrode was from the same reactor shown in Fig. 3C) (B).

FIG 6 Percent abundance of 16S rRNA for Bacteria (A and B) and Archaea (C and D) from supernatants (s) and graphite cathodes (g) of the active microbialcommunity on days 91 and 108 (yellow arrows in Fig. 1A).

Microbial Electrosynthesis of Commodity Chemicals

December 2012 Volume 78 Number 23 aem.asm.org 8417

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

biocathodes, but this is the first study to demonstrate their simul-taneous production at rates higher than those reported in the lit-erature. Furthermore, it is the first report of the electrosynthesis ofacetate from CO2 by a mixed microbial community. Differencesin laboratory approaches can complicate the comparison of pro-duction rates, but sustained rates of methanogenesis and aceto-genesis based on cathode volume surpassed what has thus far beendiscovered for electrosynthesis of these compounds at potentialshigher than �700 mV (Table 1).

A distinguishing feature of the biocathodes examined here wasthe electrochemical evidence for direct electrode oxidation by themixed microbial community. Hydrogen production facilitated bythe microorganisms may shuttle electrons to the methanogenicand acetogenic microorganisms, but several pieces of evidenceindicate that direct electron transfer is also participating: the ex-pression of a catalytic wave observed by CV with an onset at �340mV and midpoint potential at �460 mV, the lack of similar peakswith sterile or spent media, biofilm formation on the electrode,delayed exponential production of hydrogen, and the recovery ofelectrons in all three products that exceeds the abiotic generationof hydrogen by several hundred fold.

Electrosynthesis of methane. Sustainable rates of methaneproduction above 1.5 mM day�1 were achieved and reached 7 mMday�1. Both of these volumetric rates are as high as or greater thanany reported in the literature with cathodes poised at potentialsabove �800 mV (Table 1). Pisciotta et al. recently reportedmethanogenesis (0.73 mM day�1) at �439 mV that unexpectedlydecreased as the potential was lowered to �539 mV, which led theauthors to discuss the possibility of organic substrates contribut-ing to the initial rates observed at �439mV (25). Cheng et al. andVillano et al. both demonstrated that lower potentials would sup-port higher methane productivity (5, 33). However, even withincreased inputs of energy, the volumetric rates were less thanreported here with a cathode potential of �590 mV. There couldbe numerous reasons for the higher rates observed with the brew-ery waste electrosynthetic community, including the source ofmicroorganisms, the selection and adaptation of microbes at thechosen cathode potential, and the design and material of the elec-

trode (graphite granules in this case). Regardless, the results pre-sented here clearly indicate that on a working volume basis therates of methanogenesis far surpass abiotic hydrogen production.Furthermore, we have demonstrated that elevated rates of sustain-able methane production may be achieved at potentials above�800 mV.

Electrosynthesis of acetate. Acetate production concomitantwith methane and hydrogen production in the initial BES reached1.02 mM day�1, a rate that is higher than what has been reportedfor electroacetogenesis. The first report of electroacetogenesisused pure cultures of Sporomusa ovata to produce 1 mmol ofacetate over 6 days (0.17 mM day�1) and trace amounts of2-oxobutyrate in a continuous flow reactor (22). A second reportby Nevin et al. demonstrated electroacetogenesis by several otherpure culture acetogens, but none matched the production rate ofS. ovata (21).

The rate of electroacetogenesis by the brewery waste communityincreased to 4 mM day�1 after the addition of 2-bromoethanesulfo-nic acid, an inhibitor of the methyl reductase of methanogens (12).This rate out paces reported rates for electroacetogenesis by S. ovataby20-fold. However, Nevin et al. demonstrated electroacetogenesisin a continuous flow system (batch systems were examined in thepresent study) over 6 days with S. ovata at a cathode potential (�400mV) substantially higher than what was used in the present study(22). Based on the CV analysis of the brewery waste electrosyntheticcommunity, the onset of the catalytic wave began at approximately�340 mV, indicating that rates of electroacetogenesis by the mixedcommunity could be similar to that of S. ovata at the higher poten-tials. From a productivity standpoint however, maintenance of themixed community at �590 mV supports a much higher rate ofeletroacetogenesis.

Production of hydrogen and possible mechanisms of elec-tron transfer from the cathode. With enough driving force, abiocathode will produce hydrogen at rates that exceed abiotic pro-duction from an electrode (Table 1). Aulenta et al. observed hy-drogen production of 8.0 mM day�1 by a graphite cathode poisedat �900 mV and inoculated with Desulfovibrio paquesii, which was�5-fold more than what was produced in abiotic controls (2).Sustained activity and growth of the organism with the electrodewas not determined. Rozendal et al. demonstrated that hydrogencould be produced with a mixed microbial community in a graph-ite cathode that was poised at �700 mV (28). Initially, the bio-cathode produced only methane, presumably hydrogenotrophi-cally due to abiotically produced hydrogen. Bicarbonate wasremoved from the medium to eliminate methanogenesis, and thisresulted in the production of up to 25.3 mM day�1 hydrogen(8-fold greater than abiotic production) and no methane for 1,000h. The removal of bicarbonate from the medium was not possiblefor the present study since the goal was the sustained electrosyn-thesis of organic compounds from CO2. Similar to what was ob-served by Rozendal et al. (28), hydrogen did not accumulate dur-ing the initial stages of the development of the brewery wastewatercommunity on a biocathode. Surprisingly, however, sustainableand transferable rates of hydrogenesis that were nearly half thatreported by Rozendal et al. (Table 1) eventually arose concomitantwith the production of methane or acetate while the cathode waspoised at �590 mV. Whereas the ratios of biotic to abiotic pro-duction ranged from 5 to 8 in the previous studies (2, 28), herewith the cathode poised at a higher potential the ratio increased to

TABLE 1 Comparison of reported rates of electrosynthesis

Product

Cathodepotential(mV vs SHE)

Maximum rate(mM day�1)

Microbial source(reference)

Hydrogen –700 25.3 (3.2 abiotic) Wastewater (28)–900 8.0 (1.5 abiotic) Desulfovibrio paquesii (2)–590 11.8 (0.045 abiotic) Brewery wastewater (this

study)

Methane –800 1.6 Wastewater (5)–800 0.4 Wastewater (33)–900 2.1 Wastewater (33)–439 0.73 Sediment (25)–539 0.54 Sediment (25)–590 7.0 Brewery wastewater (this

study)

Acetate –400 0.17a Sporomusa ovata (22)–590 4.0 Brewery wastewater (this

study)a This rate is based on the total liquid volume used in a continuous system for 6 days.

Marshall et al.

8418 aem.asm.org Applied and Environmental Microbiology

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

250 with several hundredfold more electron equivalents simul-taneously recovered in methane or acetate.

It is possible that electrons are being directly delivered from thecathode to the microorganisms producing methane, acetate, andhydrogen. It is also plausible that hydrogen could be serving as theelectron-carrying intermediate between the electrode and themethanogens and acetogens, but it is evident that such hydrogenmust be produced biotically at the cathode. It is clear that thebiology of the system is greatly facilitating the electrosyntheticprocess since the electron recovery in products is so high versuswhat is recovered abiotically. The catalytic wave detected by CV(Fig. 4), combined with the observation of a biofilm on the cath-ode and the delayed production of hydrogen concomitant withmethanogenesis and acetogenesis, is in agreement with the biolog-ical production of hydrogen being coupled to direct electrontransfer from the electrode to a microbe. The onset of currentdraw began at �340 mV (Fig. 4), a cathode potential that was300 mV higher than the onset of current draw in the abioticreactors, indicating that the microorganisms catalyzed electrontransfer from the electrode. Importantly, the plateau in current isa unique signature of microbial catalysis of electron transfer fromthe electrode because abiotic current draw would be continuouswith decreasing potentials. If the catalytic wave is expressed byproton-reducing bacteria, then the constant supply of electronsfrom the cathode in a proton-rich environment may enable thesemicrobes to extract energy in the form of ATP while generatinghydrogen. Although growth was not measured here, the evidenceof a biofilm and sustained and transferable activity suggests thatgrowth did occur. It is conceivable that a syntrophic relationshipbetween electrode-oxidizing proton reducers and acetogens andmethanogens may help support the growth of the entire commu-nity and result in faster production rates of all three products.Interestingly, however, methane and acetate continue to be pro-duced at fast rates even as hydrogen accumulation increases, indi-cating that hydrogen does not shutdown proton reduction underthese conditions. This is in agreement with what Aulenta et al.observed with D. paquesii producing hydrogen in an electrochem-ical cell (2). Therefore, either the methanogens and acetogens areunable to keep up with the microbes responsible for hydrogengeneration, or perhaps they do not use the free hydrogen anddirectly receive electrons from the electrode, possibly by directelectron transfer between species (32) or through an electron-carrying mediator other than H2. However, the lack of any redoxpeaks in the CV scan of the spent medium would suggest that themedium or the microbial community does not supply a solublemediator other than hydrogen.

Electrosynthetic microbial community. Microbial commu-nities are notorious for the intricate interactions between micro-organisms that frequently result in an efficient and productiveprocess. This is due to the natural selection of microorganismsthat will operate in stable consortia. Often it is desirable to selectfor such consortia to perform useful reactions, e.g., the synthesisof commodity chemicals, particularly when the growth and sur-vival of the microbial community is dependent on these reactions.Extended incubation in a BES with a poised potential and onlyCO2 as the carbon source served as the selection process for thisstudy. When a potential of �590 mV was applied, the result was acommunity that would electrosynthesize three commodity chem-icals: methane, acetate, and hydrogen. A diverse group of activemicroorganisms were detected on the cathodes with the bacterial

community shifting concomitantly with changes in the prevailingfunctional activity (acetogenesis, methanogenesis, and hydrogen-esis).

The data indicate that at least one member of the communitywill interact directly with the electrode. Acetobacterium spp. werethe most prevalent and active Bacteria on the electrode when ac-etate was produced. Previous attempts to electrosynthesize acetatewith Acetobacterium woodii failed, although it consumed H2 sup-plied to the cathode chamber (21). The Acetobacterium spp. de-tected here were strongly associated with the electrode and dom-inated that population (60.3%). Either these Acetobacterium spp.are quite different from A. woodii or the microbial community onthe electrode affords Acetobacterium with advantages unrecog-nized in the pure culture. The Sphingobacteriales that becamedominant as the community progressed have close sequence iden-tities to microorganisms found in electrode reducing biofilms andto hydrogen-producing communities. It is possible that microor-ganisms such as the Sphingobacteriales WCHB1 or Sulfurospiril-lum are oxidizing the electrode and generating hydrogen (similarto D. paquesii) that feeds the methanogens and acetogens; how-ever, this could not be proven at this time. Hydrogenotrophicmethanogens, Methanobacterium in particular (93%), dominatedthe Archaea detected on the electrode regardless of conditions,and the dominant microbial morphology observed on the elec-trode when methanogenic was a rod with the appearance ofMethanobacterium. Cheng et al. (5) reported a similar percentageof Methanobacterium in an electromethanogenic cathode. Allthree dominant members of the varying community discussedabove could potentially be responsible for electrode oxidation.

Implications for commodity chemical production. Methaneis the primary component of natural gas (NG), which is widelyused in automobiles and electricity generation (3, 10). It is also theprimary source of hydrogen for the production of nitrogen fertil-izers (1). No biofuel, including electrofuels at this time, couldcompete economically with the present low price of NG unlesssubsidized, but the cost of NG will rise as its use increases. Inaddition, even though a 100-year supply of NG has been estimated(13), it will eventually be consumed. Although it is by far thecleanest of the fossil fuels, its use still results in the release of cli-mate-changing CO2. Furthermore, the hydraulic fracturing pro-cess needed to extract shale gas requires large amounts of waterand risks groundwater contamination (23). Electromethane fromrenewable and sustainable sources of energy will have many of thesame benefits but none of these problems, and it could be devel-oped first to supplement NG with the goal of one day replacing it.As the present study helps to demonstrate, the rates of electro-methanogenesis can be improved. At 131 mol of methane pergallon of gasoline equivalent (GGE) (based on 114,000 Btu pergallon of gasoline, 1,011 Btu per cubic foot of CH4, and ideal gaslaw at 25°C), the 7 mM day�1 rate observed for electromethano-genesis would calculate to a 0.05 GGE day�1 m�3 reactor. Al-though still requiring improvement, increasing this rate by anorder of magnitude would conceivably produce 0.5 GGE each dayfrom a reactor the size of a kitchen appliance. As this technologyattracts more attention, rates may increase so that a renewablebiogas technology to replace NG may be developed.

Acetic acid is another valuable commodity chemical madefrom fossil fuels that is used in industrial processes to producevinyl acetate for paints and adhesives and to a smaller extent vin-egar (6). Production for human consumption, e.g., food and cos-

Microbial Electrosynthesis of Commodity Chemicals

December 2012 Volume 78 Number 23 aem.asm.org 8419

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

metics, requires a higher degree of purity, which is achieved bymicrobial fermentation (8, 24). Acetate is also a key intermediatein the production of biofuels, since it has been shown to be afeedstock for a microbial community to produce ethanol in BESsusing methyl viologen as an electron carrier (31). Any biosyntheticpathway that involves reducing CO2 to multicarbon compoundsmust first pass through acetyl coenzyme A (acetyl-CoA) and ace-tate can be readily converted to acetyl-CoA by microbes. Hence,electroacetate could be used as a precursor for fuel production orfor the production of high-purity foods and cosmetics. In addi-tion, a synthetic biology approach could be coupled with elec-troacetogenesis to produce commodity chemicals. A similar ap-proach was taken by Li et al. with formic acid as a feedstock tomake isobutanol (15).

Electrosynthesis potentially offers a revolutionary way of pro-ducing the chemicals needed to sustain our modern culture. Thecarbon source for the process, CO2, is plentiful and inexpensive,the electrons may be supplied from sustainable non-carbon-basedsources, and the land mass requirements are negligible and willnot compete with food crop production; moreover, strictly car-bon neutral electrosynthesis presents an attractive way to combatclimate change. Analogous to the field of microbial fuel cells whereintensive research has led to a better understanding of the processand exponential gains in current generation (16); here, it has beendemonstrated that the rates of production of multiple commoditychemicals by electrosynthesis can be further increased, therebyadvancing the technology closer to becoming competitive with thefossil-carbon based industries.

ACKNOWLEDGMENTS

We thank Palmetto Brewery of Charleston, SC, for the microorganisms,Steve Morton for help with the SEM, Kevin Huncik for GC and HPLCrepairs, Chanlan Chun for reactor design, and Edward LaBelle for helpfuldiscussions.

Funding was provided by the U.S. Department of Energy, AdvancedResearch Project Agency–Energy (award DE-AR0000089).

REFERENCES1. Abram A, Foster DL. 2005. A primer on ammonia, nitrogen fertilizers,

and natural gas markets. AEDE-RP-0053-05. Ohio State University, Co-lumbus, OH.

2. Aulenta F, Catapano L, Snip L, Villano M, Majone M. 2012. Linkingbacterial metabolism to graphite cathodes: electrochemical insights intothe H2-producing capability of Desulfovibrio sp. ChemSusChem 5:1080 –1085.

3. Balash PC, Kern KC. 2008. Natural gas and electricity costs and impactson industry. DOE/NETL-2008/1320. U.S. Department of Energy, Wash-ington, DC.

4. Cheng S, Logan BE. 2007. Sustainable and efficient biohydrogen produc-tion via electrohydrogenesis. Proc. Natl. Acad. Sci. U. S. A. 104:18871–18873.

5. Cheng S, Xing D, Call DF, Logan BE. 2009. Direct biological conversionof electrical current into methane by electromethanogenesis. Environ. Sci.Technol. 43:3953–3958.

6. Cheung H, Tanke RS, Torrence GP. 2012. Acetic acid, p 209 –237. InUllman’s encyclopedia of industrial chemistry. Wiley-VCH, Weinheim,Germany.

7. DeSantis TZ, et al. 2006. Greengenes: a chimera-checked 16S rRNA genedatabase and workbench compatible with ARB. Appl. Environ. Microbiol.72:5069 –5072.

8. Drake HL, Gossner AS, Daniel SL. 2008. Old acetogens, new light. Ann.N. Y. Acad. Sci. 1125:100 –128.

9. Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R. UCHIME

improves sensitivity and speed of chimera detection. Bioinformatics 27:2194 –2200.

10. EERE. 2010. Vehicle Technologies Program: natural gas basics. DOE/GO-102010-3068. EERE/U.S. Department of Energy, Washington, DC.

11. Eid J, et al. 2009. Real-time DNA sequencing from single polymerasemolecules. Science 323:133–138.

12. Gunsalus RP, Romesser JA, Wolfe RS. 1978. Preparation of coenzyme Manalogues and their activity in the methyl coenzyme M reductase system ofMethanobacterium thermoautotrophicum. Biochemistry 17:2374 –2376.

13. Hackett JT. 2011. Prudent development: realizing the potential of NorthAmerica’s abundant natural gas and oil resources. National PetroleumCouncil for the U.S. Department of Energy, Washington, DC.

14. Huse SM, Welch DM, Morrison HG, Sogin ML. 2010. Ironing out thewrinkles in the rare biosphere through improved OTU clustering. Envi-ron. Microbiol. 12:1889 –1898.

15. Li H, et al. 2012. Integrated electromicrobial conversion of CO2 to higheralcohols. Science 335:1596.

16. Logan BE. 2009. Exoelectrogenic bacteria that power microbial fuel cells.Nat. Rev. Microbiol. 7:375–381.

17. Logan BE, et al. 2006. Microbial fuel cells: methodology and technology.Environ. Sci. Technol. 40:5181–5192.

18. Lovley DR. 2006. Bug juice: harvesting electricity with microorganisms.Nat. Rev. Microbiol. 4:497–508.

19. Mamedov TG, et al. 2008. A fundamental study of the PCR amplificationof GC-rich DNA templates. Comput. Biol. Chem. 32:452– 457.

20. Needleman SB, Wunsch CD. 1970. A general method applicable to searchfor similarities in amino acid sequence of 2 proteins. J. Mol. Biol. 48:443-&.

21. Nevin KP, et al. 2011. Electrosynthesis of organic compounds from car-bon dioxide is catalyzed by a diversity of acetogenic microorganisms.Appl. Environ. Microbiol. 77:2882–2886.

22. Nevin KP, Woodard TL, Franks AE, Summers ZM, Lovley DR. 2010.Microbial electrosynthesis: feeding microbes electricity to convert carbondioxide and water to multicarbon extracellular organic compounds. mBio1:00103–10. doi:10.1128/mBio.00103-10.

23. Osborn SG, Vengosh A, Warner NR, Jackson RB. 2011. Methane con-tamination of drinking water accompanying gas-well drilling and hydrau-lic fracturing. Proc. Natl. Acad. Sci. U. S. A. 108:8172– 8176.

24. Parrondo J, Herrero M, Garcia LA, Diaz M. 2003. A note: production ofvinegar from whey. J. Inst. Brewing 109:356 –358.

25. Pisciotta JM, Zaybak Z, Call DF, Nam JY, Logan BE. 2012. Enrichmentof microbial electrolysis cell (MEC) biocathodes from sediment microbialfuel cell (sMFC) bioanodes. Appl. Environ. Microbiol. 78:5212–5219.

26. Pruesse E, et al. 2007. SILVA: a comprehensive online resource for qualitychecked and aligned rRNA sequence data compatible with ARB. NucleicAcids Res. 35:7188 –7196.

27. Rabaey K, Rozendal RA. 2010. Microbial electrosynthesis: revisiting theelectrical route for microbial production. Nat. Rev. Microbiol. 8:706 –716.

28. Rozendal RA, Jeremiasse AW, Hamelers HV, Buisman CJ. 2008. Hy-drogen production with a microbial biocathode. Environ. Sci. Technol.42:629 – 634.

29. Schloss PD, Westcott SL. 2011. Assessing and improving methods used inoperational taxonomic unit-based approaches for 16S rRNA gene se-quence analysis. Appl. Environ. Microbiol. 77:3219 –3226.

30. Schloss PD, et al. 2009. Introducing Mothur: open-source, platform-independent, community-supported software for describing and compar-ing microbial communities. Appl. Environ. Microbiol. 75:7537–7541.

31. Steinbusch KJ, Hamelers HV, Schaap JD, Kampman C, Buisman CJ.2010. Bioelectrochemical ethanol production through mediated acetatereduction by mixed cultures. Environ. Sci. Technol. 44:513–517.

32. Summers ZM, et al. 2010. Direct exchange of electrons within aggregatesof an evolved syntrophic coculture of anaerobic bacteria. Science 330:1413–1415.

33. Villano M, et al. 2010. Bioelectrochemical reduction of CO2 to CH4 viadirect and indirect extracellular electron transfer by a hydrogenophilicmethanogenic culture. Bioresour Technol. 101:3085–3090.

34. Wang Q, Garrity GM, Tiedje JM, Cole JR. 2007. Naive Bayesian classifierfor rapid assignment of rRNA sequences into the new bacterial taxonomy.Appl. Environ. Microbiol. 73:5261–5267.

Marshall et al.

8420 aem.asm.org Applied and Environmental Microbiology

on June 10, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from