Comparison of Anodic Community in Microbial Fuel Cells with Iron Oxide-Reducing Community Hiroshi Yokoyama, Mitsuyoshi Ishida, and Takahiro Yamashita*

J. Microbiol. Biotechnol. (2016), 26(4), 757–762 http://dx.doi.org/10.4014/jmb.1510.10037 Research Article Review jmb

Animal Waste Management and Environment Division, NARO Institute of Livestock and Grassland Science, Tsukuba 305-0901, Japan

Received: October 13, 2015 Revised: December 1, 2015 The group of Fe(III) oxide-reducing bacteria includes exoelectrogenic bacteria, and they Accepted: December 6, 2015 possess similar properties of transferring electrons to extracellular insoluble-electron acceptors. The exoelectrogenic bacteria can use the anode in microbial fuel cells (MFCs) as the terminal electron acceptor in anaerobic acetate oxidation. In the present study, the anodic

First published online community was compared with the community using Fe(III) oxide (ferrihydrite) as the January 15, 2016 electron acceptor coupled with acetate oxidation. To precisely analyze the structures, the

*Corresponding author community was established by enrichment cultures using the same inoculum used for the Phone: +81-29-838-8676; MFCs. High-throughput sequencing of the 16S rRNA gene revealed considerable differences Fax: +81-29-838-8606; between the structure of the anodic communities and that of the Fe(III) oxide-reducing E-mail: [email protected] community. Geobacter species were predominantly detected (>46%) in the anodic communities. In contrast, Pseudomonas (70%) and Desulfosporosinus (16%) were predominant in the Fe(III) oxide-reducing community. These results demonstrated that Geobacter species are the most specialized among Fe(III)-reducing bacteria for electron transfer to the anode in MFCs. In addition, the present study indicates the presence of a novel lineage of bacteria in the genus Pseudomonas that highly prefers ferrihydrite as the terminal electron acceptor in acetate oxidation. pISSN 1017-7825, eISSN 1738-8872

Introduction understood. Elucidating this mechanism would provide vital clues to improving the power output of MFCs. Microbial fuel cells (MFCs) are environmentally friendly In anoxic natural environments, bacteria decompose bioreactors that simultaneously perform sustainable acetate to CO2 using soluble and insoluble electron bioenergy production and wastewater treatment [11]. acceptors, such as sulfate, nitrate, and Fe(III) oxides. The Bacteria decompose the organic matter in wastewater to mechanism of electron transfer to the anode is suggested to

CO2 through redox reactions under anaerobic conditions in be similar to the transfer to insoluble Fe(III) oxides. In fact, the reactors. The electrons generated in these redox exoelectrogenic bacteria comprise many Fe(III) reducers, reactions are transferred to the anode by bacteria, where such as Geobacter species [2]. Given that the anode acts as a they react with electron acceptors such as O2 and substitute for Fe(III) oxide, it is predicted that the anodic

K3[Fe(CN)6] via a circuit in the cathodes. The anode serves community (AC) and Fe(III)-reducing community (FRC) as an insoluble extracellular electron acceptor in these would have similar structures. In the present study, to test redox reactions, called electrode respiration. Many bacteria, this hypothesis, ACs in three MFCs and a FRC were including Geobacter, Shewanella, Desulfuromonas, and established by enrichment cultures, and the structure of the Rhodopseudomonas, can mediate electron transfer to the bacterial communities was analyzed using high-throughput anode [13]. However, the mechanism of formation of an sequencing. For comparison, a sulfate-reducing community electrochemically active biofilm on the anode is not well (SRC) was also analyzed.

A 2016 ⎪ Vol. 26⎪ No. 0 758 Yokoyama et al.

Materials and Methods 16S rRNA gene fragment was amplified by polymerase chain reaction (PCR) with the 27F and 1492R primers using the KAPA Bacterial Culture HiFi HotStart ReadyMix PCR Kit (Kapa Biosystems, MA, USA). MFC32 was a two-chambered reactor comprising two glass Subsequently, the second round of PCR was conducted using bottles (500 ml) and a Nafion 117 membrane (Dupont Japan, 563F and 802R primers, including the Illumina overhang adapter Tokyo, Japan). The anode and cathode used were a carbon cloth sequences, according to the manufacturer’s instructions. The (5 cm × 5 cm) and a carbon rod (diameter, 0.5 cm; length, 15 cm), libraries were sequenced on a 300PE MiSeq run, and image respectively. The anolyte was a basal medium containing (per liter analysis, base calling, and data quality assessment were performed with the MiSeq Reporter software (Illumina). Paired-end read of distilled water) 0.98 g potassium acetate, 0.6 g NaH2PO4·2H2O, data exported in the FASTQ format were processed with the 2g NaHCO3, 2.9 g NaCl, 0.1 g KCl, 0.2 g NH4Cl, 1 mg resazurin, 41 mg 2-bromoethanesulfonate sodium salt (a methanogenesis Quantitative Insights Into Microbial Ecology (QIIME software ver. 1.8) pipeline [4]. The read sequences were joined, quality- inhibitor), 50 mg Na2S·9H2O, and 10 ml of trace mineral and vitamin checked, and clustered into operational taxonomic units (OTUs) solutions. The catholyte was composed of 16.5 g K3[Fe(CN)6], 0.6 g using the Uclust method [5]. Representative sequences were aligned NaH2PO4·2H2O, 2 g NaHCO3, 2.9 g NaCl, 0.1 g KCl, and 0.2 g using PyNAST [3], and a phylogenetic tree was constructed. After NH4Cl. MFC35 and MFC36 were cubic air-cathode single- chambered reactors (125 ml) fed with the basal medium. The a chimera check, the taxonomic classification and alpha and beta carbon-cloth anode (5 cm × 5 cm) was placed at the opposite side diversities were computed using the QIIME tool. The taxonomic to the carbon-cloth cathode containing 0.5 mg/cm2 of a Pt catalyst assignment of the major OTUs was checked using Classifier [18]. fused with the membrane. Activated sludge collected at the The beta diversity was calculated using an unweighted UniFrac NARO Institute of Livestock and Grassland Science, Tsukuba, distance matrix [15], and the result was visualized using a principal Japan, was inoculated into MFC32 and MFC35 as seed sludge, coordinate (PCo) plot. whereas cattle feces was inoculated into MFC36. The MFCs were The sequencing data were deposited in DDBJ under the accession connected to an external resistor and were operated at 30°C in the numbers LC071702-LC071715 and DRR040632-DRR040636 fed-batch mode. The external resistor was adjusted such that the (Sequence Read Archive). MFCs generated a voltage of 0.5–0.6 V. The FRC and SRC were established by enrichment culture in 15 ml test tubes containing a Results and Discussion gas phase of 100% N2. The test tubes were filled with 10 ml of the basal medium supplemented with poorly crystalline Fe(III) MFC Operation and Enrichment Cultures of FRC and SRC oxyhydroxide (ferrihydrite; 15 mM) or Na2SO4 (20 mM) for the To precisely compare the community structures, all FRC and SRC, respectively. Ferrihydrite was prepared by titrating communities, except MFC36, were cultured using the same a FeCl solution against 10% NaOH [19], and it was stored in a 3 inoculum (activated sludge) and the same medium containing glass bottle with a gas phase of nitrogen until use. Activated acetate as the sole carbon and energy source; cattle feces sludge (1 ml), identical to that used for the MFC, was inoculated were inoculated into MFC36. The MFCs were operated for into the test tubes, and the test tubes were statically incubated at 8 months. The profiles of electricity generation by the MFCs 30°C. The culture medium (1 ml) was transferred into fresh medium at an interval of 1–4 weeks, and the transfer was repeated are shown in Fig. 1. Usually, higher electricity generation is 10 times. Coulombic efficiency was estimated from the amount obtained with the use of K3[Fe(CN)6] than with O2 as an of electron flow and decrease in acetate concentration [14]. The electron acceptor in the cathode. Air-cathode single- Fe(II) concentration was determined photometrically using the chambered MFCs generally exhibit lower Coulombic phenanthroline method [6]. efficiency than double-chambered MFCs because of O2 intrusion from the membrane. Consistent with these High-Throughput Sequencing previous observations, MFC32 (two-chambered MFC with Next-generation sequencing was performed with the MiSeq K3[Fe(CN)6]) generated a higher current (1.0–1.5 mA) than Illumina sequencing platform (Illumina Inc., CA, USA) using the the air-cathode single-chambered MFC35 and MFC36 V4 region of the 16S rRNA gene [10]. The biofilms developed on (0.2–0.3 mA). The Coulombic efficiency of MFC32 was the anodes were extensively washed with distilled water for approximately 71–85%, which was higher than that of removing the bacteria loosely attached to the anodes. The genomes MFC35 (32–33%). The FRC decomposed acetate coupled were extracted from the washed biofilms with an UltraClean Soil DNA Isolation kit (MO BIO Laboratories, Carlsbad, CA, USA). To with reduction of Fe(III) oxide. The stoichiometry was prepare the genomes of the FRC and SRC, the cultured medium acetate:Fe(III) = 1:6.8, which was close to the expected ratio was centrifuged at 12,000 ×g for 15 min. The precipitates were of 1:8. The SRC consumed acetate and sulfate with the washed with distilled water, and the genomes were extracted stoichiometry of acetate:sulfate = 1:1.1. This value was also from the precipitates using the kit. Nearly the full length of the close to the expected ratio of 1:1. These values of stoichiometry

J. Microbiol. Biotechnol. Anodic Community in Microbial Fuel Cells 759

Fig. 2. Rarefaction curves (A) and PCo plot (B) showing the relationship between the anodic communities (MFC32, MFC35, and MFC36) and Fe(III)- and sulfate-reducing communities (FRC and SRC, respectively). Fig. 1. Time course of electricity generation by MFC32 (top), MFC35 (middle), and MFC36 (bottom). summarized in Table 1. The Chao1 richness was 12,000– 18,000 for the ACs and 10,000–11,000 for the FRC and SRC. verified the integrity of the FRC and SRC activities. Although Good’s coverage was more than 0.99 in all the communities, none of the rarefaction curves reached a Community Structure plateau (Fig. 2A). In the beta diversity analysis, the ACs Next-generation sequencing technology is a powerful formed a cluster in the PCo plot (Fig. 2B), suggesting that tool for analyzing the structure of microbial communities at the AC structures were similar to each other. Surprisingly, extremely high resolution. The numbers of reads sequenced the distance of the cluster to the FRC was almost identical and the OTUs and alpha diversity of the communities are to that to the SRC. These results show that the AC

Table 1. Number of reads and alpha diversity analyses. Bacterial community No. of reads No. of OTUs Chao1 richness Abundance-coverage estimator Good’s coverage MFC32 2,268,039 9,295 12,727 14,372 0.998 MFC35 1,850,091 11,582 15,756 17,738 0.997 MFC36 1,817,684 15,047 17,231 18,795 0.998 FRC 2,110,007 10,239 10,594 11,155 0.999 SRC 1,776,485 9,855 10,169 10,684 0.999 The OTUs and alpha diversity indices of the anodic communities (MFC32, MFC35, and MFC36) and Fe(III)- and sulfate-reducing communities (FRC and SRC, respectively) were calculated using QIIME.

April 2016 ⎪ Vol. 26⎪ No. 4 760 Yokoyama et al.

Fig. 3. Class (A) and genus (B) distributions of the anodic communities (MFC32, MFC35, and MFC36) and Fe(III)- and sulfate- reducing communities (FRC and SCR, respectively) based on the 16S rRNA gene sequences. structures were considerably different from that of the This inconsistency may have resulted from differences in FRC, and that the degree of the difference between the AC the type of inocula, medium composition, and ferrihydrite and FRC structures was similar to that between the AC and preparation. For example, Geobacter-predominant communities SRC structures. The physical and electrochemical properties (>50%) have been obtained by enrichment cultures with of the anode surface would differ from that of the ferrihydrite ferrihydrite and acetate using anoxic rice paddy soil [8] surface. and the sediments of a pond [12] as the inocula. Unexpectedly, Pseudomonas was the most predominant Taxonomic Assignment genus (70%) in the FRC. The high prevalence of Pseudomonas Deltaproteobacteria were detected at high frequency (50– in the FRC is a rare case. Pseudomonas comprises diverse 91%) in the ACs, whereas Gammaproteobacteria (71%) were species, including denitrifying bacteria. Although the predominant in the FRC (Fig. 3). Deltaproteobacteria (38%), Fe(III)-reducing or electricity-generating activity for most Bacteroidia (31%), and Gammaproteobacteria (23%) were Pseudomonas species has not been examined, Pseudomonas abundant in the SRC. At the genus level, Geobacter accounted sp. strain 200 has been reported to reduce Fe(III) [1]. for 46–90% of the ACs. Geobacter species were more OTU813-FRC was detected at the highest frequency (73%) abundant in the AC of MFC32 (90%) than in the AC of among the OTUs assigned to Pseudomonas. This implies MFC35 and MFC36 (46–67%) because the double-chambered that bacteria affiliated with OTU813-FRC efficiently obtain MFC32 was more anaerobic than the single-chambered energy to grow from ferrihydrite reduction coupled with MFC35 and MFC36. Approximately 90% of the reads acetate oxidation, as acetate and ferrihydrite were added as affiliated to Geobacter species in the ACs showed the the sole electron donor and acceptor, respectively, in the highest similarity to Geobacter chapellei among all the medium. OTU813-FRC displayed 93% identity to Pseudomonas Geobacter species (Fig. 4A). In the FRC, Geobacter species aeruginosa, which is known to produce electricity in an were detected at a very low frequency of 0.7%, which is MFC (Fig. 4B) [17]. Desulfitobacterium was the second inconsistent with the results of previous studies [8, 12]. predominant genus (16%) in the FRC. Desulfitobacterium

J. Microbiol. Biotechnol. Anodic Community in Microbial Fuel Cells 761

Fig. 4. Neighbor-joining phylogenetic trees showing the relationship between OTUs and Geobacter (A), Pseudomonas (B), or Desulfitobacterium (C) species. The percentages represent the number of reads assigned to the OTUs per number of reads assigned to the genus in the anodic communities in the MFCs or Fe(III)-reducing community (FRC). OTUs with a frequency higher than 1% are shown. Numbers on major branch points indicate bootstrap values. The scale bars represent a 0.5% difference in the DNA sequences.

hafniense, an exoelectrogen [16], was one of the most closely material that is not present in the natural environment. related species to the OTUs assigned to Desulfitobacterium Nevertheless, Geobacter species can use the anode as the (Fig. 4C). Some Desulfitobacterium species can reduce Fe(III) terminal electron acceptor. Bacteria likely possess the citrate and Mn(IV) oxide in addition to chlorinated plasticity to use various extracellular electron acceptors, compounds [7]. Desulfovibrio species were detected in both including the artificial material of the anode, to adapt to communities; the prevalence was 2.5% in the FRC and their environment. The present study has demonstrated 0.6–5.9% in the ACs (Fig. 3). Desulfovibrio contains the that Geobacter species are the most specialized among exoelectrogenic species Desulfovibrio desulfuricans [9]. Fe(III)-reducing bacteria for electron transfer to the anode Based on these observations, potent exoelectrogenic coupled with acetate oxidation. In addition, this study bacteria and Fe(III)-reducing bacteria are abundantly included indicates the presence of a novel lineage of bacteria in the FRC, although the community structure was markedly (OTU813-FRC) in the genus Pseudomonas that highly prefers different between the ACs and FRC. The anode is an artificial ferrihydrite as the terminal electron acceptor.

April 2016 ⎪ Vol. 26⎪ No. 4 762 Yokoyama et al.

Acknowledgments 267-278. 9. Kang CS, Eaktasang N, Kwon D-Y, Kim HS. 2014. Enhanced We thank Ms. K. Hirano and Ms. T. Ohmori for their current production by Desulfovibrio desulfuricans biofilm in a helpful assistance. We are also grateful to the members of mediator-less microbial fuel cell. Bioresour. Technol. 165: 27-30. Agrogenomics Research Center, National Institute of 10. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. 2013. Development of a dual-index sequencing strategy Agrobiological Sciences (NIAS), for the high-throughput and curation pipeline for analyzing amplicon sequence data sequencing. This work was supported in part by the Japan on the MiSeq illumina sequencing platform. Appl. Environ. Society for the Promotion of Science under KAKENHI (No. Microbiol. 79: 5112-5120. 26850177 and 40391370) and the genome-support grant 11. Kumar GG, Sarathi VGS, Nahm KS. 2013. Recent advances from NIAS. We have no conflicts of interest to declare. and challenges in the anode architecture and their modifications for the applications of microbial fuel cells. References Biosens. Bioelectron. 43: 461-475. 12. Lentini CJ, Wankel SD, Hansel CM. 2012. Enriched iron(III)- 1. Arnold RG, Dichristina TJ, Hoffmann MR. 1986. Inhibitor reducing bacterial communities are shaped by carbon studies of dissimilative Fe(III) reduction by Pseudomonas sp. substrate and iron oxide mineralogy. Front. Microbiol. 3: 404. strain 200 “Pseudomonas ferrireductans”. Appl. Environ. Microbiol. 13. Logan BE. 2009. Exoelectrogenic bacteria that power microbial 52: 281-289. fuel cells. Nat. Rev. Microbiol. 7: 375-381. 2. Bond DR, Lovley DR. 2003. Electricity production by 14. Logan BE, Hamelers B, Rozendal RA, Schrorder U, Keller J, Geobacter sulfurreducens attached to electrodes. Appl. Environ. Freguia S, et al. 2006. Microbial fuel cells: methodology and Microbiol. 69: 1548-1555. technology. Environ. Sci. Technol. 40: 5181-5192. 3. Caporaso JG, Bittinger K, Bushman FD, DeSantis TZ, 15. Lozupone C, Knight R. 2005. UniFrac: a new phylogenetic Andersen GL, Knight R. 2010. PyNAST: a flexible tool for method for comparing microbial communities. Appl. Environ. aligning sequences to a template alignment. Bioinformatics Microbiol. 71: 8228-8235. 26: 266-267. 16. Milliken CE, May HD. 2007. Sustained generation of 4. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, electricity by the spore-forming, gram-positive, Desulfitobacterium Bushman FD, Costello EK, et al. 2010. QIIME allows analysis hafniense strain DCB2. Appl. Microbiol. Biotechnol. 73: 1180- of high-throughput community sequencing data. Nat. Methods 1189. 7: 335-336. 17. Nor MHM, Mubarak MFM, Elmi HSA, Ibrahim N, Wahab 5. Edgar RC. 2010. Search and clustering orders of magnitude MFA, Ibrahim Z. 2015. Bioelectricity generation in microbial faster than BLAST. Bioinformatics 26: 2460-2461. fuel cell using natural microflora and isolated pure culture 6. Fadrus H, Maly J. 1975. Suppression of iron(III) interference bacteria from anaerobic palm oil mill effluent sludge. in the determination of iron(II) in water by the 1,10- Bioresour. Technol. 190: 458-465. phenanthroline method. Analyst 100: 549-554. 18. Wang Q, Garrity GM, Tiedje JM, Cole JR. 2007. Naive 7. Finneran KT, Forbush HM, VanPraagh CVG, Lovley DR. Bayesian classifier for rapid assignment of rRNA sequences 2002. Desulfitobacterium metallireducens sp nov., an anaerobic into the new bacterial taxonomy. Appl. Environ. Microbiol. bacterium that couples growth to the reduction of metals 73: 5261-5267. and humic acids as well as chlorinated compounds. Int. J. 19. Zavarzina DG, Kolganova TV, Boulygina ES, Kostrikina NA, Syst. Evol. Microbiol. 52: 1929-1935. Tourova TP, Zavarzin GA. 2006. Geoalkalibacter ferrihydriticus 8.Hori T, Mueller A, Igarashi Y, Conrad R, Friedrich MW. gen. nov sp nov., the first alkaliphilic representative of the 2010. Identification of iron-reducing microorganisms in family Geobacteraceae, isolated from a soda lake. Microbiology anoxic rice paddy soil by 13C-acetate probing. ISME J. 4: 75: 673-682.

J. Microbiol. Biotechnol.