Genomic Barcode-Based Analysis of Exoelectrogens in Wastewater Biofilms Grown on Anode Surfaces

Abstract

The most energy-demanding step of wastewater treatment is the aeration-dependent elimination of organic carbon. Microbial fuel cells (MFCs) offer an alternative strategy in which carbon elimination is conducted by anaerobic microorganisms that transport respiratory electrons originating from carbon oxidation to an anode. Hence, chemical energy is directly transformed into electrical energy. In this study, the use and stability of barcode-containing exoelectrogenic model biofilms under non-axenic wastewater treatment conditions are described. Genomic barcodes were integrated in Shewanella oneidensis, Geobacter sulfurreducens, and G. metallireducens. These barcodes are unique for each strain and allow distinction between those cells and naturally occurring wild types as well as quantification of the amount of cells in a biofilm via multiplex qPCR. MFCs were pre-incubated with these three strains, and after 6 days the anodes were transferred into MFCs containing synthetic wastewater with 1% wastewater sludge. Over time, the system stabilized and the coulomb efficiency was constant. Overall, the initial synthetic biofilm community represented half of the anodic population at the end of the experimental timeline. The part of the community that contained a barcode was dominated by G. sulfurreducens cells (61.5%), while S. oneidensis and G. metallireducens cells comprised 10.5% and 17.9%, respectively. To the best of our knowledge, this is the first study to describe the stability of a synthetic exoelectrogenic consortium under non-axenic conditions. The observed stability offers new possibilities for the application of synthetic biofilms and synthetically engineered organisms fed with non-sterile waste streams.

Introduction

The application of microbial fuel cells (MFCs) in a wastewater treatment plant is one of many relevant techniques that can be applied to advance the already 100-year-old wastewater treatment process [23]. Carbon elimination in wastewater treatment plants is conducted via aerobic organisms that oxidize the organic carbon sources in a respiratory step. The wastewater aeration is the most energy-consuming step during the treatment process. To reduce costs, an MFC could be implemented in place of the aerobic oxidation step. Here, microorganisms use a solid anode as an electron acceptor for anaerobic respiration. These organisms catalyze the efficient conversion of chemical energy stored in the organic carbon sources to electrical energy if the anode is connected to an electrical load and a cathode.

Previously, it was shown that MFCs used with wastewater can produce energy while reducing the amount of organic carbon [4]. The anodic community was analyzed in a few studies. Of the detected organisms, some were closely related to microorganisms that were shown to transfer respiratory electrons onto an anode [13]. Deltaproteobacteria comprising the genus Geobacter seem to play a particularly key role in many communities, while other reports describe wastewater-fed communities on anodes that were dominated by members of the Bacteroidetes, Betaproteobacteria, and Gammaproteobacteria [14]. It is largely unknown how these organisms contribute to the electron transfer process at the anode. Moreover, the anode serves as a nonspecific surface for biofilm growth that can be occupied by microorganisms that do not contribute to the transfer of electrons. To improve the performance of an MFC, the anodic microbial community must be selected and monitored [10]. This will be a major challenge for future fuel cell research because tools have not been established to assemble tailored biofilms under non-axenic conditions or even to understand the functional role of all major contributors of anode biofilm communities [22].

Models for extracellular respiratory organisms, also called exoelectrogens, are Shewanella spp. and Geobacter spp. Whereas Shewanella oneidensis is a facultative anaerobic microorganism that can use a wide variety of electron acceptors [31], not all strains of Geobacter can grow in the presence of oxygen. Geobacter sulfurreducens can tolerate up to 10% oxygen [16], whereas Geobacter metallireducens is strongly inhibited by oxygen. However, G. metallireducens is extremely versatile in terms of the electron donors it can use [1]. Many studies have analyzed the abilities of S. oneidensis and G. sulfurreducens in terms of biofilm growth on anodes. Whereas S. oneidensis seems to form rather thin films, G. sulfurreducens can form several-micrometer-thick biofilms that are conductive [20].

Hence, the three organisms have characteristics that would, in combination, form an organism that is extremely robust and can use a multitude of electron donors and acceptors. Instead of combining the characteristics in one organism, the combination of their abilities in a synthetic three-organism biofilm would potentially be almost as efficient. Moreover, all three strains are genetically tractable. Hence, the stable cultivation of synthetically engineered Geobacter and Shewanella strains under non-axenic conditions makes it possible to use available waste streams for biotechnological processes without prior costly deactivation.

It was the aim of this study to analyze how stable a community of model organisms would be under wastewater treatment conditions. To monitor the three exoelectrogens and distinguish them from naturally occurring wild types in wastewater sludge, barcodes were integrated into their genomes. A barcode is a synthetic and unique DNA sequence that is integrated in a non-coding region so that the phenotype does not change. Microbial electrochemical systems were started with all three strains at the same time. After the current density was stabilized, the anodes were transferred into new setups containing synthetic wastewater with 1% wastewater sludge. The current density, total amount of organic carbon (TOC), and coulomb efficiency were measured over time.

 

Materials and Methods

Strain Preparations

The barcodes were integrated into a non-coding region of the genomes of S. oneidensis, G. sulfurreducens, G. metallireducens, and E. coli, with a space of at least 100 bp to adjacent coding regions. The synthetic sequences were designed with GeneDesign software to exclude the formation of secondary structures and ensure binding of qPCR primers to the designated sequences only [25]. An overview of all strains and plasmids used in this study is shown in Tables S1 and S2, and the primers are shown in Table 1. Barcoded strains are referred to with a subscript bc on each strain name. An overview of the insertion site of the barcoded strains is shown in Fig. S1.

Table 1.The primer numbers are used in the Materials and Methods section to indicate for which experiments the individual oligonucleotides were used.

Integration of the barcode into the S. oneidensis genome was conducted using the suicide vector pMQ150. The plasmid was linearized with BamHI and SalI (New England Biolabs, Frankfurt am Main, Germany). Primers 1 and 4 were designed with a homology of 35 nt to the ends of the linearized plasmid. Primer sets 1-2 and 3-4 were used to amplify a region of 500 bp upstream and downstream, respectively, of the insertion site. The two fragments were assembled using a subsequent PCR step with primers 1 and 4. This was possible because primers 2 and 3 overlapped with 25 nt. These primers further contained the barcode sequence. The entire fragment was cloned into the linear plasmid according to the isothermal assembly method described by Gibson et al. [8]. Afterwards, the plasmid was transformed in Escherichia coli WM3064 and subsequently transferred into S. oneidensis MR-1 using conjugation; integration was described elsewhere [27].

Integration of the barcode into G. sulfurreducens was performed using 602- and 601-bp-long fragments upstream and downstream of the insertion site, amplified using primer sets 5-6 and 7-8. The primers 9 and 10 were used for the amplification of a kanamycin resistance cassette from the plasmid pSCVAM. Primers 6 and 9 as well as 7 and 10 had a 25 bp overlap for assembly in a consecutive PCR. The overlapping region of primers 7 and 10 was part of the barcode sequence. The three PCR fragments were assembled via PCR and cloned in the plasmid pJET 1.2 using the CloneJET PCR Cloning Kit according to the manufacturer s instructions (Fermentas, St. Leon-Rot, Germany). The 2.2 kbp insert was amplified with primers 5 and 8 and transformed into G. sulfurreducens as described by Coppi et al. [5]. As a minor modification, agar plates containing 35 μg/ml instead of 25 μg/ml kanamycin were used.

Integration into G. metallireducens was similar, but an upstream fragment of 598 nt (primers 11 and 12), the kanamycin resistance sequence (primers 13 and 14), and the 599 nt downstream sequence (primers 15 and 16) were amplified via PCR. The entire construct was ligated via isothermal assembly into the HindIII (New England Biolabs) linearized plasmid pk18mob_amp. The transformation of G. metallireducens was performed according to Oberender et al. [21]. As a minor modification, anoxic aqua bidest was used for resuspension of the cells.

Insertion of the barcode into E. coli was conducted using the vector system described by Haldimann and Wanner [9]. The synthetic sequence was synthesized by GenScript (Piscataway, NJ, USA) and amplified with primers 17 and 18. The primers also contained homologous regions to the vector pAH162 that was cleaved with BamHI and SalI (New England Biolabs). The amplicon was ligated into the linearized plasmid pAH162 via isothermal in vitro ligation [8]. The plasmid was inserted into the E. coli genome as described previously [9]. A schematic overview of the strain preparations is depicted in Fig. S2.

Bacterial Strains and Growth Conditions

Prior to microbial electrochemical cell (MEC) experiments, microbial strains were pre-grown in the minimal medium adapted from Dolch et al. [6] with minor changes. The final concentrations of yeast extract, cysteine, and CaCl2 were 0.2% (w/v), 2.5 mM, and 0.4 mM, respectively; 0.02% peptone was added. The medium was supplemented with 20 mM sodium lactate and 40 mM disodium fumarate for S. oneidensis and with 20 mM sodium lactate, 10 mM acetate, and 40 mM disodium fumarate for G. sulfurreducens; whereas. 10 mM acetate, 4.4 mM propionate, and 10 mM sodium nitrate was used for G. metallireducens. No electron acceptors in addition to the anode were added to the medium in the MEC experiments. The minimal medium contained 12.5 mM sodium lactate and 5 mM sodium propionate as electron donors. E. coli strains were cultured in lysogeny broth medium at 37℃.

MEC Setup and Electrochemical Measurements

MECs were operated in a two-chamber setup as described elsewhere [30]. The working electrode material was graphite felt GFD2 EA (2 mm thick) from SGL Group, Carbon Company (Meitingen, Germany) [15]. Prior to use, the electrodes were first rinsed with isopropanol and thereafter installed in the MEC before filling it with deionized water and autoclaving. The electrodes were connected via platinum wires (0.1 mm; Chempur, Karlsruhe, Germany) to a potentiostat (Pine Instruments, Grove City, PA, USA). A saturated calomel electrode (Sensortechnik Meinsberg GmbH, Ziegra-Knobelsdorf, Germany) was used as the reference.

The anode chamber was filled with 2 L of minimal medium that did not contain any electron acceptor. It was continuously flushed with 80/20 N2/CO2 to maintain a constant anoxic environment with a stable pH. The optical densities of the bacterial cultures were measured at 655 nm, and the starting cell density of each strain in the anodic chamber was 0.04. All MEC experiments were started in independent sextets at a constant temperature of 30℃. The anodes were poised to a constant potential of 0.241 V vs. the normal hydrogen electrode (NHE).

Measurements were stopped after 6 days. Three anodes were immediately transferred to new setups containing anoxic synthetic wastewater according to ISO 11733 [11] (0.02% (w/v) peptone, 0.01% (w/v) meat extract, 0.1 mM glucose-monohydrate, 0.4 mM NH4Cl, 0.1 mM KH2PO4, 0.2 mM Na2HPO4·2H2O, 3.5 mM NaHCO3, 1.0 mM NaCl, and 0.148 mM FeCl3·6H2O). As a minor medium modification, deionized water was used instead of potable water because of the high amount of carbonate in potable water from Karlsruhe. At the same time, the constant gas flow through the system was stopped, and the fuel cells were slowly stirred instead. As an additional inoculum, 1% of a mixture of aerobic and anaerobic wastewater sludge from the wastewater treatment plant in Karlsruhe, Germany was added. The standard measurement protocol consisted of measuring the current at 0.241 V vs. NHE for 14 days. The three remaining MFCs were used for cell quantification and fluorescent in situ hybridization (FISH) analysis (Fig. 1).

Fig. 1.Work flow. In phase I, the three strains were incubated at 241 mV vs. NHE in a bicarbonate-buffered medium for 6 days. Afterwards, one MEC was used for FISH, two for qPCR, and three for transferring the anodes into new MECs and starting the second phase. In phase II, the MECs were run with synthetic wastewater and 1% wastewater sludge. After 14 days, one MEC was used for FISH and two for qPCR.

Fluorescent in situ Hybridization

Fluorescent in situ hybridization analysis was conducted according to Dolch et al. [6]. All fluorescent probes that were employed for the analysis are listed in Table 2.

Table 2.Fluorescently labeled oligonucleotide probes and helper oligonucleotides used for fluorescent in situ hybridization experiments.

Images were taken with a Leica DM 5500 B microscope using a 63× water immersion lens and a DFC 300 FX digital color camera from Leica (Wetzlar, Germany). The filter sets L5 (excitation filter 480/40 and suppression filter 527/30), Y3 (545/30 and 610/75), and A4 (360/40 and 470/40) were used for FITC, Cy3, and DAPI, respectively. Picture stacks were assembled using ImageJ [24].

Extraction of DNA

After completion of the MEC program, DNA was extracted using the innuPREP Stool DNA Kit from Analytic Jena (Jena, Germany) according to the manufacturer’s instructions. Whole anodes were used for the lysis step after addition of E. colibc cultures as an internal standard. Standard curves were compiled using serial dilutions of pure cultures after counting in an improved Neubauer counting chamber (Marienfeld, Lauda-Königshofen, Germany). The individual dilution steps from all three strains were pooled in one vial prior to DNA extraction. From each of the six dilution steps, DNA was extracted in triplicate according to the manufacturer’s instructions. It was also isolated from the wastewater sludge as a negative control for the qPCR experiment.

An additional goal was to estimate the percentage of the starter community consisting of the three barcode strains within the anode consortium after further inoculation with wastewater sludge. Therefore, the amount of DNA resulting from the addition of E. colibc cells was subtracted from the amount of total isolated DNA. By comparing the DNA concentrations prior to and after addition of wastewater sludge, we could calculate the percentage of residual barcode-containing cells. The DNA concentration was measured using a NanoDrop 2000 Spectrophotometer (ThermoScientific, Schwerte, Germany).

Quantitative PCR

Strain-specific primers and hydrolysis probes were designed with the software tool Beacon Designer (Premier Biosoft, Palo Alto, CA, USA) to quantify the abundance of S. oneidensisbc, G. sulfurreducensbc, and G. metallireducensbc in the MFCs via qPCR (Table 3). All qPCRs were conducted in a CFX96 Cycler (Bio-Rad, Munich, Germany) using white polypropylene thin-walled plates (4titude, Wotton, UK) and adhesive qPCR seal (Sarstedt, Nümbrecht, Germany). A reaction volume of 20 μl was chosen as suggested by the DyNAmo Flash qPCR Kit manual (Biozym, Hessisch Oldendorf, Germany). The primer (Sigma-Aldrich, Steinheim, Germany) concentration was 0.5 μM, the hydrolysis probe (Biomers, Ulm, Germany) concentration was 0.25 μM, and 1 μl of template-DNA was added. The optimal annealing temperatures as well as the efficiencies were determined using isolated DNA of the pure strains and temperature gradient qPCR with the DyNAmo Flash SYBR Green qPCR Kit (Biozym, Hessisch Oldendorf, Germany). The efficiencies at 60℃ are shown in Table 3. Primer pairing or primer-dimer formation was not detectable. All experiments were accompanied with no-template controls, and the isolated DNA from the wastewater sludge was used as a negative control.

Table 3.Sequences of primers and probes used for real-time PCR experiments.

Standard curves were established using biological triplicates and applied on each qPCR plate. On the basis of standard curves, cell counts of each isolated DNA sample were determined. Cell quantifications of the anode samples were based on three separate DNA samples from two independent MFCs. Quantitative PCR experiments with these samples were conducted in technical triplicates. The cell count per anode was calculated by multiplying the mean of the samples by the amount of lysis solution.

Electrical Conductivity

The conductivities of the anode media and the synthetic wastewater were measured using an HI 99300 EC/TDS meter (HANNA Instruments, Kehl am Rhein, Germany).

Analytical Measurements and Statistical Analysis

Samples were taken daily for quantification of TOC using a multi N/C 2100S and TOC-Gas-Generator TG 600 from Analytik Jena (Jena, Germany). It was assumed that each consumed molecule of organic carbon would have a redox state of 0 (following the simplified formula of organic carbon as (CH2O)n). Therefore, the oxidation of one molecule of organic carbon to CO2 would lead to the release of four electrons. The coulomb efficiency was calculated by dividing the number of electrons transferred to the anode by the theoretical value. Statistical analysis was conducted by calculating the median and the normalized median of absolute deviation (MADN) [3] for the MFC data. The average and standard deviation were calculated for cell number analysis.

 

Results

Establishing a Biofilm on the Anodes

The electrical conductivity of the medium was 8.1 mS/cm. During the 6-day initial monitoring period (phase I, Fig. 1), there was a steep increase in the current during the initial hours followed by a plateau. After 6 days, the current density was stable and reached 392.39 ± 209.73 μA/cm2 (Fig. S3).

Current Production of the Pre-Incubated Anodes in Wastewater

The electrical conductivity after phase II was 0.6 mS/cm. The same potential, 241 mV vs. NHE, was applied, and the current density was measured. There was immediate current production without a detectable lag phase. When the current density dropped below the starting value, an identical dose of organic carbon was added. This procedure was repeated twice. Each time, an immediate increase in current density occurred; however, the measured current was always lower than in phase I (Fig. 2A).

Fig. 2.Analysis of current density, TOC concentration, and Coulomb efficiency in phase II of the experiment. The solid lines show the median of three independent MECs and the dashed lines the MADN. The dashed vertical lines indicate the addition of the synthetic wastewater. (A) Current density of three independent MECs in phase II. Anodes with exoelectrogens were transferred into anoxic synthetic media containing 1% wastewater sludge. (B) Amount of total organic carbon (TOC) in the MECs in phase II. (C) Coulomb efficiency of the MECs in phase II.

Reduction of Organic Carbon and Coulomb Efficiency

Before each addition of carbon, the TOC dropped to a level similar to the original value (Fig. 2B). In the first period, this took almost 5 days. After subsequent additions, the carbon consumption stabilized to approximately 50 mg organic carbon per day.

To evaluate the efficiency of the bioelectrical systems with respect to reducing the organic carbon load in wastewater and thereby producing electricity, the coulomb efficiency was used as a nominal value. In Fig. 2C, the coulomb efficiency is shown over time. In the first 5 days, it increased steadily until a new carbon source was added. Thereafter, it stabilized and did not change after feeding. After 14 days, the coulomb efficiency was 7.3% ± 1.2%.

Microbial Population Analysis

It was shown that the biofilm on the pre-incubated anodes together with wastewater sludge could produce current by oxidizing organic carbon in artificial wastewater. To analyze how the biofilm developed between the end of phases I and II, anodes were fixed and used for FISH analyses. At the end of phase I, all graphite felt fibers were covered with a uniform and dense biofilm of bacteria except between fiber crossings where more biomass production could be observed (Fig. 3A). The anode had thicker biofilms around the fibers (almost 50 μm radius) after 14 days in wastewater (Fig. 3B). Surprisingly, the biofilm consisted only of bacteria. Therefore, the supernatant of phase II was also stained for bacteria and archaea. Most of the cells were grouped into flocks. In those flocks, some archaea (green) were detected (Figs. 3C and 3D). Additionally, the planktonic organisms in phase II were further stained with probe GEO3 to quantify the amount of Geobacteraceae in the planktonic phase. However, no fluorescence signal was detectable, suggesting that Geobacteraceae were localized to the anode surface only (Fig. 3D).

Fig. 3.Fluorescent in situ hybridization images of the two phases. DNA was stained nonspecifically with DAPI (blue). Biofilm on the anode after (A) phase I and (B) phase II. Bacteria (EUB388-I) were stained in red and archaea (ARCH915) in green. Planktonic cells after phase II (C) stained with EUB388-I (red) and ARCH915 (green) and (D) GEO3 (red) and ARCH915 (green).

Survival of the Three Exoelectrogenic Strains

First, the limit of detection (LOD) was determined for each strain separately and was 144 ≥ LOD ≥ 14.4 cells per reaction for S. oneidensisbc, 154 ≥ LOD ≥ 30.8 for G. sulfurreducensbc, and 404 ≥ LOD ≥ 40.4 for G. metallireducensbc. The Cq values were used as internal controls and did not differ in the qPCR experiments by more than 1.7%. This assured that the solution was homogenous and that the quantification procedure allowed comparison of values between different bioelectrochemical setups. Prior to all fuel cell experiments, the barcoded strains were characterized in growth experiments using nitrate or fumarate as electron acceptor as described in the Materials and Methods section. During these tests, no differences in growth characteristics could be observed, indicating that the integration of the barcode did not change the cell physiology (data not shown).

Even though pre-incubation was started with cell suspensions of equal OD values, the cell numbers differed. This is most probably due to a difference in cell size. Hence the starting inoculum was composed of 28.9 ± 0.4% G. sulfurreducensbc cells, 45.5 ± 2.7% G. metallireducensbc cells, and 25.6 ± 3.2% S. oneidensisbc cells. G. sulfurreducensbc was dominant after 6 days of incubation both in the planktonic and in the sessile phase. We found 40.2 ± 7.1% of the planktonic cells were S. oneidensisbc in phase I, whereas only 6.7 ± 2.5% of the sessile cells accounted for this organism. The quantities of G. metallireducensbc cells in both fractions were smaller (Fig. 4). The total number of sessile cells was 8.81 × 1010 ± 4.79 × 1010. Hence, the average DNA content per cell was 10.8 ± 0.7 fg.

Fig. 4.Distribution of the three barcoded strains at the end of phases I and II. The amount of cells of S. oneidensisbc (green), G. sulfurreducensbc (red), and G. metallireducensbc (purple) was quantified by applying multiplex qPCR. A, Distribution of barcoded strains in the inoculum. B, Distribution of barcoded strains in the planktonic phase at the end of phase I. C, Distribution of barcoded strains on the working electrode at the end of phase I. D, Distribution of barcoded strains on the working electrode at the end of phase II. In the planktonic samplesfrom phase II, the amounts of cells were below the LOD value for each strain. The error bar indicates the deviation of two independent MECs from each phase.

After transferring the anodes into synthetic wastewater containing 1% sludge and further incubation for 14 days, 1.1 ± 0.1 mg DNA was isolated from the anodes. Using the above-quantified DNA content per cell and estimating that this value can be, to some extent, expanded to all bacterial cells, 1.04 × 1011 ± 9.66 × 109 cells were expected to be part of the anodic biofilm. The amount of sessile barcode cells decreased to 5.02 × 1010 ± 1.18 × 1010. Therefore, half (48.8 ± 11.3%) of the cells on the anode contained a barcode. Among the barcoded strains, the dominance of G. sulfurreducensbc was even more pronounced because it constituted 97.9 ± 2.6% of all barcoded cells. Versus the initial biofilm composition at the end of phase I, 61.5% of all G. sulfurreducensbc, 10.5% of all S. oneidensisbc, and 17.9% of all G. metallireducensbc cells were retained on the anode surface. However, none of the three strains was detected in the planktonic part of the bioelectrochemical systems.

 

Discussion

Establishing Traceable Microorganisms

In this work, we developed four strains with synthetic and unique DNA sequences (barcodes) inserted in their genomes. Instead of deleting a gene with the barcode sequence [7], the barcode was integrated in a non-coding region of the genome. Therefore, the phenotypes of the organisms remained unchanged. The barcode sequences were used for cell quantification. By applying a standard curve of known cell counts for each strain, a multiplex qPCR protocol was developed. This was a faster approach than staining with FISH probes and performing fluorescence-activated cell sorting. However the most prominent advantage was that these barcodes distinguished these strains from naturally occurring phylogenetically related strains. Hence, this barcode system can be used to analyze the influence of synthetic biofilms on the performance of an MFC or to monitor the prevalence of synthetically introduced production strains on anodes in waste streams. Because we lack an understanding of the ecology of microbial communities on anodes [22], this system further is a platform for studying microbial communities by monitoring their development under different applied potentials or process conditions. Of note, using this method, it is not possible to distinguish between live and dead cells. Moreover, it could be possible that extracellular DNA could influence the measurement, which would lead to an overestimation of the cell number. Still, DNA is a very good carbon source that can be used by a variety of organisms. Therefore, an accumulation of extracellular DNA in a non-axenic system will most likely not occur.

Performance of the MFCs

We have shown that the three exoelectrogenic strains could establish a stable current-producing biofilm on the anode in a bicarbonate-buffered medium in phase I. In phase II, these anodes were transferred to a new setup containing anoxic synthetic wastewater with 1% wastewater sludge. Here, the current densities were always lower than in phase I. One reasonable explanation could be the lower electrical conductivity of the synthetic wastewater. The current density dropped over time, although the TOC values showed that residual carbon was present. This potentially indicates that only a certain fraction of the TOC was consumed by the community in the bioelectrochemical cells. After approximately 8 days, the coulomb efficiency stabilized. The lower efficiency compared with phase I is most probably due to the formation of gaseous end-products like hydrogen or methane by fermentative bacteria and methanogens, respectively. This is also in line with the detection of archaea in the planktonic phase after phase II. The pre-incubated anodes could degrade 50 mg/l TOC per day, while anodes with only wastewater sludge accumulated organic carbon (data not shown). This shows that although pre-incubated anodes might not be quite sufficient to produce energy from wastewater, they reduced more organic carbon.

Our focus on the established biofilm and how it developed led us to determine that pre-incubation resulted in a uniform and thin biofilm covering the graphite felt fibers. This biofilm became thicker in wastewater. Geobacteraceae are known to be enriched in MFCs [13]. Therefore, the planktonic cells were stained with GEO3 – a probe for the Geobacter cluster. Quite surprisingly, cells belonging to this cluster were absent in the planktonic phase nor were they detectable in the wastewater sludge. The amount of sessile exoelectrogens declined in phase II relative to that found at the end of phase I. However, the barcoded organisms represented half of the anodic population. It is known that G. sulfurreducens needs bicarbonate to grow [19]. Here, we showed that if Geobacter was pre-grown in a bicarbonate-buffered condition as part of a biofilm, it can survive and most probably produce electricity in an anodic biofilm in wastewater without CO2 as a buffer. This agrees with the results of Jung and Regan [12], who found Geobacter spp. on an anode in a non-bicarbonate-buffered system.

The highest stability was that of G. sulfurreducensbc. Roughly 60% of the initial community size could be detected at the end of phase II, while 20% of G. metallireducensbc and 10% of S. oneidensisbc cell counts after phase I could be quantified after phase II. The high stability of G. sulfurreducens cells can likely be explained by its ability to build dense multilayer biofilms [20].

This study showed that it is possible to dictate the microbial community of biofilms in non-axenic systems to a surprisingly high degree. Certainly, the amount of cells that can be retained will depend on the operating conditions of the MEC. This will be the basis of further studies to analyze factors that positively influence this process. Nevertheless, the integration of synthetic biofilms with certain abilities into waste streams not only offers the possibility to improve a carbon elimination process but also enables the biotechnological use of waste streams without prior deactivation of the natural community.