Electron acceptor redox potential globally regulates transcriptomic profiling in Shewanella decolorationis S12

Electron acceptor redox potential (EARP) was presumed to be a determining factor for microbial metabolism in many natural and engineered processes. However, little is known about the potentially global effects of EARP on bacteria. In this study, we compared the physiological and transcriptomic properties of Shewanella decolorationis S12 respiring with different EARPs in microbial electrochemical systems to avoid the effects caused by the other physicochemical properties of real electron acceptor. Results showed that the metabolic activities of strain S12 were nonlinear responses to EARP. The tricarboxylic acid cycle for central carbon metabolism was down-regulated while glyoxylate shunt was up-regulated at 0.8 V compared to 0.2 and −0.2 V, which suggested that EARP is an important but not the only determinant for metabolic pathways of strain S12. Moreover, few cytochrome c genes were differentially expressed at different EARPs. The energy intensive flagella assembly and assimilatory sulfur metabolism pathways were significantly enriched at 0.8 V, which suggested strain S12 had stronger electrokinesis behavior and oxidative stress-response at high EARP. This study provides the first global information of EARP regulations on microbial metabolism, which will be helpful for understanding microorganism respiration.

accurately poised redox potential in MES provides an ideal tool to observe the regulation of EARP on microorganisms 9,10 . In recent years, the influence of anode potential on power generation, microbial community and contaminants degradation have been intensively studied for MES optimization. However, many studies showed the biofilm growth, current generation and substrate consumption were nonlinear responses to EARP, which is inconsistent with the thermodynamics theory 6,11 . Microbial respiration with electrode is a central driving force for the complex bioelectrochemical functions in MES, while only few studies paid attention on the regulation of EARP on the intracellular processes, or only focused on several EARP-specific enzymes involved in TCA cycle and the electron transfer process 10,12 . Based on previous reports 6,7,10,12 , we hypothesized that, in addition to genes encoding the components in TCA cycle and electron transfer chain, some important but ignored energy intensive metabolic pathways may be enriched at more positive potentials which may be responsible for the nonlinear responses. In order to verify our hypothesis, global and comprehensive information of EARP regulation on microorganism metabolism is essential.
Shewanella genus has both direct and indirect electron transfer pathways to electrodes and thus was used as one of the model electrode-respiring bacteria [13][14][15] . In this study, the physiological performances and transcriptome of S.decolorationis S12 respiring with electrodes with different redox potentials (− 0.2, 0.2 and 0.8 V vs standard hydrogen potential, comparable to the EARP of natural common electron acceptors such as goethite (− 0.18 V), sulfate (− 0.21 V), S 3 O 6 2− (0.22 V), O 2 (0.81 V)) were comparatively analyzed 7 . This study provides new insights into our understanding of bacterial responses to EARP and offers important information on the regulation of bacterial metabolism.
Confocal laser scanning microscopy (CLSM) results showed that only scarce cell clusters were observed on the anode surfaces in all MES which can be attributed to the short cultivation time (8 h, Supplementary Fig. S1). Anodes in MES −0.2 and MES 0.8 showed denser cell cluster and larger biofilm-covered area than anode in MES 0.2 , which is in line with the total biomass quantification results (Fig. 1B). However, more unviable cells (with red fluorescence) were observed on the 0.8 V electrode than on the others, indicating an unfavorable biofilm growth status on the electrode surface in MES 0.8 .
As a key coenzyme, NADH participates in a number of metabolic reactions and the ratio of NAD + /NADH has been suggested to be a key factor in determining cell metabolism and can reflect the extracellular redox status 16 . However, in our study the NAD + /NADH ratios showed no significant difference among MES −0.2 , MES 0.8 and MES 0.2 ( Supplementary Fig. S2), which is in line with a recent study with Geobacter sulfurreducens that the NAD + /NADH ratio showed no difference under four potentials of 110, 10, − 90, and − 190 mV 17 .
Global gene expression pattern. Sequencing resulted in a total number of ~10.8 to 15.2 million reads per sample and more than 93.8% of the reads could be mapped to the annotated strain S12 genome (accession number AXZL00000000). A total of 4215 unigenes (4219 unigenes of reference genome) had mapped reads and 50.91% (2255/4215) of the unigenes could be annotated in KEGG database. Mean FPKM value (Fragment per kb per million mapped reads) was used for calculating fold change. Differentially expressed unigenes were identified with FDR (False discovery rate) ≤ 0.05 and |log 2 fold change| ≥ 1. A total of 369 genes were differentially transcripted among different samples, of which 282 genes in log 2 (FPKM 0.8 V /FPKM 0.2V ), 255 genes in log 2 (FPKM 0.8 V /FPKM −0.2V ) and 45 genes in log 2 (FPKM 0.2 V /FPKM −0.2V ). Less transcriptome differences were observed between MES −0.2 and MES 0.2 relative to those with MES 0.8 , suggesting that strain S12 respiring with − 0.2 and 0.2 V electrode performed more similar metabolic profile compared to cells with 0.8 V electrode. All those 369 differentially transcripted genes were divided into 20 categories according to COG classification (Fig. 2). Furthermore, metabolic pathway enrichment analysis exhibited flagellar assembly (corrected P-value = 0.001) and sulfur metabolism (corrected P-value = 0.001) were differently enriched in MES 0.8 compared with MES 0.2 and MES −0.2 (Fig. 2). Several noteworthy genes involved in central carbon metabolism and electron transfer process were highlighted in the enrichment analysis.
Central carbon metabolism. Carbon metabolism is one of the most important pathways determining the substrate uptake, energy generation and cell growth, in which TCA cycle plays a central role 12 . In this study, several genes encoding key enzymes in TCA pathway were significantly up-regulated in MES 0.2 and MES −0.2 compared with those in MES 0.8 (Fig. 3), including aconitate hydratase (converting citrate to isocitrate), 2-oxoglutarate dehydrogenase E2 component and 2-oxoglutarate dehydrogenase E1 component (converting 2-oxoglutarate to succinate), succinyl-CoA synthetase (catalyzing the redox reaction between fumarate and succinate). Moreover, 2-methylcitratesynthase gene (prp C) which contributes to the anaerobic acetyl-CoA metabolism, the gate in TCA cycle, also showed significantly higher transcription level under the two lower EARPs. These results indicated that the TCA pathway of S. decolorationis S12 was enriched in MES 0.2 and MES −0.2 compared with MES 0.8 . Remarkably, formate dehydrogenase and the key enzyme of malate synthase in glyoxylate shunt were significantly up-regulated in MES 0.8 relative to MES 0.2 and MES −0.2 (Fig. 3).
Respiration and energy generation. The capacity that bacteria transfer electrons to extracellular acceptors (e.g. mineral oxides, humics and electrodes) is fundamentally important for various biogeochemical processes and the development of MES. Therefore, it is one of the most interesting issue to investigate whether and how the bacteria regulate the electron transfer pathway in response to EARP. By analyzing 53 differentially transcripted genes attributed to electron transfer and energy generation processes, the transcription pattern of those genes under − 0.2 and 0.2 V were similar and significantly distinguished from that under 0.8 V ( Supplementary  Fig. S3). Among those, 6 genes encoding different types of cytochromes were significantly up-regulated at 0.8 V while 3 were up-regulated at 0.2 or − 0.2 V ( Table 1), suggesting that some electron transfer components in S. decolorationis S12 are EARP-specific. The outer membrane cytochrome c (e.g. MtrC, OmcA) and electron mediator flavins have been proven to be essential for Shewanella extracellular electron transfer, however no significant difference was detected for these redox components 13 .
Flagellar assembly and sulfur metabolism pathways. Flagellum is a key cellular appendage for bacteria motility, movement toward solid surfaces and bacterial electrode respiration 18 . The energy intensive flagella synthesis and functioning are tightly related to cell energy conservation and consumption 19 . Our results showed that flagellar assembly pathway was significantly enriched (corrected P-value = 0.001) in MES 0.8 (Table 2). Moreover, genes encoding two-component system components (MCP, FliC, CheV and CheR), for regulating flagellum switch, were also upregulated at 0.8 V. It has been reported that Shewanella biofilm cells showed depressed flagellum assemble compared with planktonic cells as flagella were no more needed after cells were embeded in biofilms 20 . Assimilatory sulfur metabolism was one of the most significantly enriched pathways in MES 0.8 (corrected P-value = 0.001) (Fig. 4). Twelve genes involved in this pathway were significantly up-regulated which highlighted a completely activated sulfur assimilation pathway including extracellular sulfate transport, sulfite and sulfide . Central carbon metabolism of strain S12 at different EARPs. The red solid arrow lines denote TCA cycle key steps and the green ones denote key steps of glyoxylate pathway. The black ★ and longitudinal double ★ respectively denote the expression is significant (FDR ≤ 0.05 and |log 2 FC| ≥ 1) and extremely significant (FDR ≤ 0.01 and |log 2 FC| ≥ 2) differences with comparing to 0.8 V. Name of enzyme  Table 1. The significant differentially expressed cytochrome-associated genes. "-" denotes no significant difference; "a" denotes no available COG number.
generation and cysteine synthesis. Especially, the sulfite reductase (NADPH) alpha subunit, which is the key gene in sulfur metabolism pathway, was significantly upregulated in MES 0.8 (23 and 9.5 fold higher than that in MES 0.2 and MES-0.2 , respectively) 21 . It was suggested that strain S12 converted lots of energy and electrons or performed cysteine synthesis in MES 0.8 . Cysteine is an essential amino acid for many oxidoreductases (e.g. cytochromes, thioredoxins and Fe-S proteins). The enrichment of cysteine synthesis pathway was consistent with the results that more genes encoding cytochromes or some other cysteine-containing enzymes involved in energy generation were upregulated under 0.8 V. Moreover, cysteine itself can function as electron mediator in bacterial electron transfer 22 .

Discussion
Bacterial respiration is a central driving force in various environmental and engineering processes. The EARP has been considered to play a key role in determining the bacteria energy conservation, cell growth and activities, and become one of the key factors in understanding biogeochemical processes and optimizing microbial reactors.
In this study, a significant partial of genes involved in central carbon metabolism, sulfur metabolism, cell motility and stress-response showed specific responses to EARP by analyzing the global transcriptomic profiling of S. decolorationis S12.
On the central carbon metabolism, our results suggested a higher TCA cycle activity with lower EARP. Consistently, Matsuda et al. reported that the TCA pathway of S. oneidensis MR-1can be activated by decreasing the EARP from 0.4 to 0 V or lower, under which the expressions of rate-limiting enzymes in TCA pathway were up-regulated 10 . However, Gobbler et al. suggested a higher-activated TCA cycle of S. oneidensis MR-1 and more biofilm cells under higher EARP 12 . Regarding the discrepancy, it should be noted that our transcriptome results mainly come from the planktonic cells rather than biofilms, because planktonic cells play a dominant role in both electricity generation (via secreting electron shuttles) and biomass yield in Shewanella MES 15 Table 2. The differentially expressed flagellar genes. "-" denotes no significant difference; "a" denotes no available COG number.
reported the protein expression profile of Shewanella biofilm cells was significantly different with planktonic cells 20 . It is possible that the central carbon metabolism between biofilm and planktonic cells of Shewanella has different responses to EARP. Unlike the other model electrode respiring bacteria genus Geobacter, Shewanella species have anapleurotic reactions through glyoxylate shunt which can reduce the oxidation of carbon substrate by TCA cycle and save carbon sources for biosynthesis 24,25 . The up-regulated glyoxylate shunt indicated an incomplete TCA cycle under higher EARP (0.8 V vs 0.2 V and − 0.2 V). Therefore, it seems that the central carbon metabolism of strain S12 switched from complete TCA cycle to glyoxylate shunt along with the increase EARP. Moreover, it was reported that formate oxidation has a contribution on the growth rate and yield of Shewanella oneidensis strain MR-1 under anaerobic conditions 26 . The formate dehydrogenase being upregulated significantly in MES 0.8 could be responsible for the higher lactate degradation and growth rate. It has been reported that Shewanella species including S. decolorationis S12 can utilize lactate as carbon resource and electron donor in oxygen respiration though complete TCA cycle but low active or incomplete TCA cycle under anaerobic conditions 4,5 . Although the EARP of MES 0.8 is close to oxygen (0.81 V vs SHE), strain S12 in MES 0.8 performed an up-regulated glyoxylate shunt and incomplete TCA cycle pathway. It can be seen that EARP has significant impact on microbial metabolism pathways, but it is not the only determining factor. The other factors of an electron acceptor such as the solubility, molecular structure may also play essential roles in regulating bacterial metabolism. It should be noted that, different with the natural electron acceptors, the inherent electrochemical processes such as electrode kinetics, the overpotential of electroactive materials and extremely large electrostatic interaction in the electric double layer in MES may also have interactions with bacteria cells which should be considered in further studies.
The different transcriptomic profile of several cytochromes which participated in electron transfer supported the previously reported electrochemical phenomena, that electrode respiring bacteria showed different voltammetry behaviors on electrodes with different redox potentials which indicated different redox-species were generated on the electrode surface 6,7 . A recent report showed an ImcH/CbcL system in Geobacter for sensing the EARP switch below or above 0.1 V 6 , which also indicated possible different redox sensing strategy was used between Shewanella and Geobacter. Despite the transcription levels of S. decolorationis S12 outer membrane cytochrome c showed no difference here, the conformation flexibility of the outer membrane cytochrome c can facilitate Shewanella to adapt to transfer electrons to electron acceptors at different EARPs. Therefore, the self-regulation of Shewanella to electron acceptors may be a complex processes including inter-and intra-molecular changes.
It is interesting to note that the higher EARP did not resulted proportional biomass increase in MES 0.8 compared with MES 0.2 and MES −0.2 which can be explained by the high oxidative stress and energy-inefficiency redundancy pathways (e.g. flagellar assembly and assimilatory sulfur metabolism) 19,27 . Planktonic cells play a dominant role in electrode respiration in Shewanella MESs, especially before a mass of biofilms generated on electrode surface 23,28 . Electrokinesis (i.e. bacteria rapid swim around solid electron acceptor) has been demonstrated to be a crucial process in the electron transfer of planktonic cells to electrodes. Higher electrode potential could stimulate Shewanella electrokinesis near the electrode surface by which the planktonic cells could transfer electrons to electrode at high frequency via a "touch and go" model 28 . The higher transcription level of flagella coding genes at high EARP is in line with the electrokinesis theory in Shewanella extracellular respiration and consistent with the less biofilm formation on the 0.8 V electrode surface ( Supplementary Fig. S1). However, it remains mystery that why S. decolorationis cells preferred to swim rather than be sessile on electrode surface at 0.8 V, even though flagella synthesis and cell movement is highly energy-cost and no electron transporting moiety is included in those appendages 14 .
Furthermore, high EARP may cause oxidative stress to bacterial cells by generating reactive oxygen species (ROS) 29 , which may explain that more biofilm cells with impaired membrane (with red fluorescence) can be founded on the 0.8 V electrode. Synthesis of glutathione or other low molecular weight thiols is a key strategy for bacteria addressing oxidative stress 30 . Cysteine is an essential precursor for generating or endowing those compounds with ability to remove ROS. It has been reported that cysteine synthesis of Shewanella or some other bacteria were increased under H 2 O 2 or alkaline stress 31 . Therefore, it was presumed that the enrichment of assimilatory sulfur metabolism may be a strategy of Shewanella to address the oxidative stress caused by the high electrode potential.
All results in this study indicated that EARP, a key but not the only determinant for central carbon metabolism, performed global regulation on the transcriptomic profile of strain S12. Briefly, S. decolorationis S12 showed an EARP-specific central carbon metabolism and electron transfer pathway. The energy intensive flagella assembly and assimilatory sulfur metabolism pathway were significantly enriched under higher EARP as strategies for electrokinesis and anti-oxidative stress, respectively. The results not only extend our understanding on microbial respiration and metabolism processes but also provide implications for the developments of MES and some other microbial technologies.

Materials and Methods
Bacterial strains and MESs operation. S. decolorationis strain S12 was isolated from activated sludge and preserved in our laboratory 32 . The strain can respire with various electron acceptors, such as Fe (III), fumarate, nitrate, azo dyes and electrode [33][34][35] . A single colony of strain S12 was cultured aerobically in 100 ml LB at 30 °C overnight. The cells were centrifuged and washed three times with lactate medium (LM, 10 mM, pH7.5) to be used as inoculums for MES.
Dual-chamber with potentiostat (CHI 1040C) MES were assembled and sterilized as previously reported 15 . Briefly, the anode chamber of MES contained a graphite anode, a reference electrode (AgCl/Ag, 0.195 V vs standard hydrogen electrode (SHE)) and 100 mL of lactate medium (LM). The cathode chamber contained a graphite cathode and 100 mL of PBS with 50 mM potassium ferricyanide. Anode potentials in different MESs were poised at − 0.2 V, 0.2 V and 0.8 V. All samples had three repetitions and all mentioned potentials had been converted to vs SHE. Initial bacterial density in anode culture was 0.05 (OD 600 , optical density at a wavelength of 600 nm).
Chemical and physiological analysis. Lactate concentration in culture medium and protein-based biomass volume of the planktonic cells were by HPLC and a protein quantification assay, as previously reported 36 . SigmaPlot 11.0 was used for data (mean and standard deviation) statistical analysis. CLSM was used to analyze the development and structure of the biofilms attached on anode surfaces 15 . Before CLSM observation, biofilm samples were stained with LIVE/DEAD BacLight staining kit (Molecular Probes) which can distinguished viable (with green fluorescence) or unviable or stressed (with red fluorescence) bacterial cells. Randomly sampled view fields were observed and analyzed for each anode biofilm. To obtain three-dimensional structure information, the biofilm sample was observed under the "z-Stack" model of the Zen software (Zeiss) 15 .

RNA extraction and transcriptome analysis.
For transcriptome analysis, S. decolorationis S12 cells cultured in LM with different EARPs were collected for RNA extraction as described by Holmes et al. 37 . Planktonic cells were included in transcriptome analysis due to their significant role in electricity generation with electron mediator 23,38 . RNA protect was immediately added to the cell deposits and then stored at − 80 °C before RNA extraction. RNA was extracted using the commercial RNeasy Mini Kit (Qiagen, Germany) with DNase (RNase-free, Takara) treatment according to the manufacturer's instruction. The extracted RNA was quantified and evaluated by Agilent 2100 bioanalyzer and Caliper Labchip GX. Amplified fragments were sequenced using Illumina HiSeq ™ 2500.
Raw sequence data were filtered to remove those containing adapter and reads with more than 10% unknown nucleotides, and reads with more than 50% of low quality base (value ≤ 5). Clean reads were mapped into the transcriptome reference database using Tophat 2.0.1 and Samtools 0.1.18.0. Less than 3 mismatch bases were permitted, and unique mapped reads were obtained. Cufflinks 2.0.0 software was used for calculating the FPKM value and difference analysis. P-value was used to evaluate the difference of gene transcription and FDR was used to determine the threshold of P-value 39 .
Genes were annotated in COG (Cluster of Orthologous Groups of protein) database for identification of orthologous proteins and KEGG (Kyoto Encyclopedia of Genes and Genomes) database for annotation and metabolic pathway analysis. Enrichment metabolic pathways were analyzed by both KOBAS 2.0 and Galaxy LEfSe, which set P-value < 0.05 as statistically significant.