Fnr (EtrA) acts as a fine-tuning regulator of anaerobic metabolism in Shewanella oneidensis MR-1

Background EtrA in Shewanella oneidensis MR-1, a model organism for study of adaptation to varied redox niches, shares 73.6% and 50.8% amino acid sequence identity with the oxygen-sensing regulators Fnr in E. coli and Anr in Pseudomonas aeruginosa, respectively; however, its regulatory role of anaerobic metabolism in Shewanella spp. is complex and not well understood. Results The expression of the nap genes, nrfA, cymA and hcp was significantly reduced in etrA deletion mutant EtrA7-1; however, limited anaerobic growth and nitrate reduction occurred, suggesting that multiple regulators control nitrate reduction in this strain. Dimethyl sulfoxide (DMSO) and fumarate reductase gene expression was down-regulated at least 2-fold in the mutant, which, showed lower or no reduction of these electron acceptors when compared to the wild type, suggesting both respiratory pathways are under EtrA control. Transcript analysis further suggested a role of EtrA in prophage activation and down-regulation of genes implicated in aerobic metabolism. Conclusion In contrast to previous studies that attributed a minor regulatory role to EtrA in Shewanella spp., this study demonstrates that EtrA acts as a global transcriptional regulator and, in conjunction with other regulators, fine-tunes the expression of genes involved in anaerobic metabolism in S. oneidensis strain MR-1. Transcriptomic and sequence analyses of the genes differentially expressed showed that those mostly affected by the mutation belonged to the "Energy metabolism" category, while stress-related genes were indirectly regulated in the mutant possibly as a result of a secondary perturbation (e.g. oxidative stress, starvation). We also conclude based on sequence, physiological and expression analyses that this regulator is more appropriately termed Fnr and recommend this descriptor be used in future publications.


Background
Due to its respiratory versatility, Shewanella oneidensis strain MR-1 serves as a model organism for studying the regulation of aerobic and anaerobic growth [1][2][3]. In contrast to Escherichia coli, the regulatory systems that control transcription of genes responsible for different respiratory processes are poorly understood in environmentally relevant Shewanella spp. [4][5][6][7]. In E. coli, the transition from aerobic to anaerobic metabolism is primarily regulated by Fnr (fumarate and nitrate reduction regulator) and by the two-component regulatory system ArcAB (aerobic respiration control) [8][9][10][11]. A gene expression study in E. coli K12 indicated that one-third of its 4,290 genes were differentially expressed during aerobic versus anaerobic growth [12]. Among the differentially expressed genes, 712 (49%) genes were directly or indirectly affected by Fnr. Fnr possesses a [4Fe-4S] 2+ cluster that acts as an oxygen sensory domain [13]. Fnr in its active dimeric form binds to target DNA sequences inducing or repressing transcription [14,15]. Under aerobic conditions, or when oxygen levels increase, an Fe 2+ atom in the [4Fe-4S] 2+ cluster is oxidized resulting in the formation of a [2Fe-2S] 2+ cluster via a [3Fe-4S] 1+ intermediate. This oxidation causes a conformation change in Fnr, thus altering its affinity to DNA and regulatory control of transcription [14,15].
Studies with strain MR-1 mutants have identified three important regulators of anaerobic metabolism: EtrA (electron transport regulator protein), ArcA and CRP (cyclic AMP receptor protein) [4,5,[16][17][18][19]. Sequence alignment of the protein encoded by etrA reveal that the four cysteine residues that form the [4Fe-4S] 2+ cluster in Fnr are conserved in EtrA [16]. In a gene replacement study, etrA of strain MR-1 restored wild type physiology of an E. coli fnr deletion mutant [16]. EtrA shares 73.6% and 50.8% of amino acid sequence identity with Fnr in E. coli and Anr (arginine deaminase and nitrate reductase anaerobic regulator) in Pseudomonas aeruginosa, respectively. This high degree of similarity suggests that EtrA has a regulatory function in MR-1, possibly by sensing oxygen. Despite the lack of physiological evidence to support a regulatory role of EtrA in the anaerobic metabolism of strain MR-1 [7], a gene expression study using a partial microarray (691 ORFs) of the strain MR-1's genome suggested involvement of EtrA in the regulation of the transcription of genes associated with aerobic and anaerobic metabolism [6]. Growth experiments with an etrA deletion mutant in S. oneidensis strain DSP10 (a spontaneous rifampicin resistant mutant of MR-1) implicated EtrA in the regulation of genes related to aerobic and anaerobic metabolism, similar to what has been observed for Fnr in E. coli [12,20]. Unfortunately, the implications of these findings cannot be interpreted unambiguously since the rifampicin resistance of strain DSP10 influences electron transport [21].
To examine the regulatory role of EtrA in strain MR-1 in more detail, we generated an etrA knockout mutant EtrA7-1 in a wild type background. Growth and phenotypic characterization of this mutant combined with a whole genome transcriptome analysis confirms that EtrA regulates nitrate and fumarate reduction, plus provides experimental evidence for its positive regulatory role in DMSO reduction. Our genome-wide expression analysis shows differential expression of 612 genes for which sequence analysis recognized a EtrA motif for 72 of the operons encoding 118 genes, suggesting that their regulation is via direct interaction of EtrA with its promoters. Most of these genes are associated with metabolic functions.

Results
Genotypic and phenotypic characterization of a ΔetrA:: loxP mutant The growth of the etrA knockout mutant EtrA7-1 with nitrate was significantly impaired as cultures reached a maximum OD 600 of 0.02, at least 5-fold lower than the wild type strain (Figure 1). In addition, the doubling time for the mutant under these conditions was approximately 10 h compared to a doubling time of 2 h for the wild type. Plasmid pCCG03 carrying etrA, but not the parental pCM62 vector lacking etrA, restored near wild type growth to the EtrA7-1 mutant, which confirms that the observed phenotype was attributable to the deletion of etrA. After 10 h of incubation, nitrate was reduced in wild type and complemented EtrA7-1 cultures though less nitrate was reduced in the latter consistent with its slightly slower growth ( Figure 2). After 23 h, both the wild type and EtrA7-1 complement had completely converted nitrate and nitrite to ammonia, while nitrite was the major product in cultures of the EtrA7-1 mutant and the control strain harboring pCM62 ( Figure 2).
Anaerobic cultures of the mutant and the wild type strain were analyzed for the reduction of different electron acceptors with lactate as the electron donor. No growth of the EtrA7-1 mutant was observed with fumarate as electron acceptor whereas the wild type strain reached an OD 600 of 0.053 ± 0.005. Limited growth (approximately 50% lower OD 600 compared with the wild type cultures) was observed in mutant cultures amended with trimethylamine N-oxide (TMAO) or thiosulfate (data not shown). No OD increases with the mutant and the wild type were measured with DMSO provided as electron acceptor at 2 and 10 mM; however, HPLC analyses of cultures with 2 mM DMSO revealed that DMSO was completely consumed in wild type cultures, whereas no DMSO consumption was evident in the mutant cultures ( Figure 3 sulfite reduction rates were observed between the wild type and the EtrA7-1 deletion mutant ( Figure 3). Anaerobic cultures of the mutant and the wild type strains grown with pyruvate instead of lactate as electron donor showed similar results, i.e., the mutant showed limited or no growth with nitrate, fumarate and DMSO provided as electron acceptors compared to the wild type ( Figure 4). Similar to the lactate-amended cultures, the rates of nitrate, fumarate and DMSO reduction in wild type cultures exceeded those measured in cultures of the mutant strain (Table 1). Resting cell assays corroborated these findings and nitrate reduction and ammonium production occurred at higher rates in assays with wild type cells. Complete stoichiometric conversion to ammonium also occurred in the assays with mutant cells, although lower rates and a 3-fold longer incubation were required for complete reduction (i.e., 24 h for the EtrA7-1 versus 8 h for the wild type) ( Figure 5). A. B.

Effects of etrA deletion on transcription
The global transcriptome profile of mutant strain EtrA7-1 grown anaerobically with nitrate as the sole electron acceptor was compared to that of the wild type under the same growth conditions. A complete list of all the genes differentially expressed two-fold or higher is provided as supplemental information (Additional file 1). Out of 612 differentially transcribed genes in the EtrA7-1 mutant relative to the wild type, 289 were up-regulated and 323 were down-regulated. The differentially transcribed genes were classified in 19 functional "TIGR Role" categories (Additional file 2) based on the MR-1 genome annotation (GenBank accession number AE014299) [22]. Genes with unknown functions represented the largest category of up-regulated (14.8%) and the second most common category of down-regulated genes (17.3%). Genes associated with energy metabolism were the largest category (17.6%) of down-regulated genes (Additional file 2). Among the up-regulated genes, the "Protein synthesis" category ranked second (12.5%) and the "Other categories" ranked third (11.4%). This latter category included phage-, transposase-and plasmid-related genes. The "Energy metabolism" category represented 9.7%, ranking fourth.

Identification of putative EtrA binding sites
The promoters of the differentially expressed genes were examined for putative EtrA binding sites in order to identify those genes that were likely directly regulated by EtrA from the many genes whose expression changes were most likely due to secondary effects. For example, the up-regulation of phage-related genes is likely a response to stress, and not a direct result of the etrA deletion. Putative EtrA binding sites were identified for those genes that showed at least 2.5 fold change in the EtrA7-1 mutant relative to the wild type by extracting the upstream intergenic sequences and applying the Gibbs centroid sampler to identify a putative binding site motif. A motif was identified (Additional file 3) that displays similarities to the E. coli Fnr and Crp binding sites motifs; this motif was present upstream of 44 operons that encode a total of 78 genes. The largest proportion of these genes is in the "Energy metabolism" category ( Table 2 and 3, Additional file 2). Binding sites were detected upstream of an additional 28 operons when the detected motif (Additional file 3) was used to scan the upstream intergenic regions of all genes listed in Additional file 1.

Regulatory role of EtrA in energy metabolism
Since the "Energy metabolism" category contained the largest group of genes responsive to EtrA, these genes were analyzed in more detail. Up-regulated genes ( Table 2) in this group included genes encoding a cytochrome c oxidase (ccoPQN [SO2361-2362, SO2364]), proteins involved in gluconeogenesis such as PckA (SO0162), and nqrABCDEF-2 genes (SO1103-1108) encoding NADH:ubiquinone oxidoreductases. From this group, only the nqr gene clusters had a putative EtrA binding site. While the nqr-2 gene cluster was up-regulated in the etrA knockout mutant, the nqr-1 gene cluster (SO0903-0907) was down-regulated. Nqr is a Na + pump that during respiration generates a sodium motive force to mediate solute transport, flagellar motility and ATP synthesis [23]. Both nqr gene clusters had putative EtrA binding sites. The microarray data indicated that EtrA affects the transcription pattern of these genes differently. Similarly, the etrA deletion had a distinct effect on the expression of the fdh gene clusters encoding a formate dehydrogenase. The fdh-1 genes (SO4508-4511) were up-regulated whereas the fdh-2 gene cluster (SO4512-4515) was down-regulated. An EtrA binding site was only identified for the fdh-2 cluster and not for the fdh-1 cluster, indicating EtrA affects both clusters differently.
Other up-regulated genes in the "Energy metabolism" category included the succinate dehydrogenase gene sdhC (SO1927), the succinyl-CoA synthase operon Table 1 Comparison of reduction rates of several electron acceptors with pyruvate as electron donor by S. oneidensis MR-1 wild type strain and etrA knockout strain EtrA7-1
Other down-regulated genes grouped in different categories included genes encoding ABC transporters (cydCD [SO3779-3780], SO4446-4448), TonB-dependent receptors (nosA [SO0630]), and L-lactate permease (lldP [SO0827]) and a putative lactate permease (SO1522). The only gene directly down-regulated from this later group is lldP (SO0827), for which an EtrA binding site The relative expression is presented as the ratio of the dye intensity of the anaerobic cultures with 2 mM KNO 3 of EtrA7-1 to that of MR-1 (reference). b The standard deviation was calculated from six data points, which included three independent biological samples and two technical samples for each biological sample. c The arrows indicate that the gene is regulated by the binding site that follows. The direction of the arrow indicates the location of the gene. An arrow pointing down indicates the gene or operon is in the plus or sense strand and the arrow pointing up indicates the gene or operon is in the minus or anti-sense strand. The relative expression is presented as the ratio of the dye intensity of the anaerobic cultures with 2 mM KNO 3 of EtrA7-1 to that of MR-1 (reference). b The standard deviation was calculated from six data points, which included three independent biological samples and two technical samples for each biological sample. c The arrows indicate that the gene is regulated by the binding site that follows. The direction of the arrow indicates the location of the gene. An arrow pointing down indicates the gene or operon is in the plus or sense strand and the arrow pointing up indicates the gene or operon is in the minus or anti-sense strand.

Stress response caused by the etrA deletion
We detected induction of genes from various categories, which have been associated with stress response i.e., starvation, phage infection and oxidative stress, possibly due to accumulation of nitrogen oxide reactive species. Up-regulated genes (Additional file 1) were dominated by genes grouped in "Other categories". The majority of up-regulated genes were phage-related. For example, 25 genes of the LambdaSo phage (SO2940-2974), a gene encoding a viral capsid protein of the MuSo1 phage (SO0675), and genes of MuSo2 phage (SO2684-2685, SO2687, SO2702) were up-regulated. In contrast, the gene encoding the LambdaSo phage transcriptional regulator of the Cro/CI family (SO2990) was down-regulated (ratio 0.43). Transcriptional changes of most of these genes are likely indirect effects due to the deletion of the etrA gene and only for the LambdaSo phage genes S02957-2962 was an EtrA binding site predicted. The category "Transport and binding proteins" contains a large number of genes associated with stress response. Up-regulated genes in this group contain genes encoding heavy metal efflux pumps (SO0520, SO4596-4598, SOA0153) and genes for phosphate transport (SO1560, SO1723-1724 including pstB-1 [SO1725], pstB-2 [SO4289] and pstA [SO4290]). There was up-regulation of phoBR (SO1558-59) and phoU (SO1726) genes, which regulate the phosphate transporters genes during phosphate starvation [28][29][30][31][32].

Discussion
The results of this study demonstrate that EtrA positively regulates dissimilatory nitrate, fumarate and DMSO reduction pathways in S. oneidensis MR-1. The generation of etrA knockout mutant EtrA7-1 in the wild type strain MR-1 background eliminated any possible secondary effects on the phenotype, such as the electron transfer perturbation suspected with the rifampicin resistant DSP10 strain [6]. Similar to other etrA mutants of strain MR-1, EtrA7-1 retained its ability to reduce nitrate [6,7,16]; however, our results show that the anaerobic growth of the mutant was significantly impaired compared to the wild type when nitrate was the only electron acceptor. Likewise, the etrA deletion mutant lost its ability to reduce fumarate and DMSO with both lactate and pyruvate as electron donor. Regulation of DMSO reduction by EtrA in strain MR-1 was suggested previously [6] however this study provides physiological evidence that confirm its role. The ability of the EtrA7-1 mutant to reduce TMAO and thiosulfate also decreased; however the reduction of Fe(III) citrate, Mn(IV) and sulfite was not affected by the deletion. No differences in growth performance between the wild type and the mutant were observed under aerobic conditions (data not shown).
The transcriptome analysis provides a genome-wide expression profile of S. oneidensis MR-1 instead of the partial genome array that was previously evaluated (691 ORFs [6] vs 4,648 genes in this study). We observed in 612 (13%) differentially expressed genes represented though some are likely due to differences in growth rate between the mutant strain and the wild type strain. Nonetheless, the expression patterns of genes are consistent with the physiological data and with the transcription data reported for Fnr in E. coli [11,12,20] and with the more limited data by Beliaev et al. [6]. Genes involved in nitrate reduction (napDAHGB, nrfA, and hcp) were significantly down-regulated by the etrA deletion as well as those encoding the fumarate reduction (frdAB, fccA) and all the genes encoding for the DMSO reductases (dmsAB). All of these genes have been considered candidates for EtrA regulation in previous studies; however, results were not conclusive [5][6][7]16]. Sequence analysis of the regulatory regions of the differentially expressed genes, indicated possible EtrA recognition sites for most genes in the "Energy metabolism" category, some of which i.e. napDAHGB, nrfA, frdAB and dmsAB, confirms previous results [6] and further suggests that regulation of these genes is via direct interaction of EtrA with their promoters. Putative recognition sites for EtrA were also identified for the two nqr gene clusters, which had not been identified previously. Also, the regulatory regions for fdh gene clusters were evaluated and an EtrA binding site was recognized for only fdhA-1. The fdh-2 cluster does not possess an EtrA binding site, suggesting a different regulatory system.
Our data indicate that EtrA is a global regulator acting in cooperation with other regulatory proteins to control anaerobic metabolic processes in strain MR-1 [6,7,16], therefore, the expression of these genes cannot be expected to be under an "all or none" regulatory mechanism. Rather, these global regulators respond to multiple stimuli (e.g., oxygen levels, substrates) and finetune regulation via transcriptional control and interactions between regulatory proteins. Studies in S. oneidensis and in other Shewanella species that indicate the combined action of transcriptional regulators for the anaerobic metabolism in this organism [4,[17][18][19]. For example, recent studies showed that CRP, EtrA and the product of the cya genes act as expression regulators of several anaerobic respiratory systems, including nitrate reduction in S. oneidensis MR-1 and Shewanella sp. strain ANA-3 [4,[17][18][19]. In E. coli, Fnr and NarP positively regulate the nap and nrf genes [12,20,38,39]. MR-1 possesses the genes for a homolog of the two-component regulatory system in E. coli NarQ/NarP (SO3981-3982). The presence of alternate regulators that partially fulfill the function of EtrA can explain why nitrate reduction even though impaired, still occurred in the EtrA7-1 knockout mutant.
Down-regulation of genes for lactate transport was also observed. Since lactate was the source of reducing equivalents and carbon, a lack of electron donor and carbon may have contributed to the impaired growth of the EtrA7-1 mutant. Induction of transport proteins for carbon sources and electron acceptors has also been credited to Fnr in E. coli [12,20], and a putative EtrA binding site was predicted for the gene encoding a lactate permease (SO0827) in MR-1.
Impaired growth of EtrA7-1 could also be due to stress factors caused or enhanced by the deletion (e.g. accumulation of nitrogen oxide reactive species and starvation). The expression of phage-related genes induced in response to irradiation in strain MR-1 has been reported [40]. Up-regulation of the genes involved in activation of the strain MR-1 prophages LambdaSo, MuSo1 and MuSo2 in the EtrA7-1 mutant was observed, suggesting phage activity. Induction of bacterial genes (e.g., nusAG) required to stabilize the Lambda protein antitermination complex in E. coli was also shown [41,42]. Conversely, there is repression of the LambdaSo transcription regulator Cro/CI family, which represses the transcription of the Lambda genes in E. coli [43]. Also, the induction of genes associated with starvation, i.e., a condition that could activate the lytic cycle of prophages [43], was confirmed in the expression analysis.

Conclusion
The involvement of several regulatory controls has complicated the interpretation of gene expression patterns and functions in Shewanella spp. Results from the above etrA deletion mutant studies suggest a global regulatory role for EtrA, but one which works in conjunction with other regulators to fine-tune the expression of key genes in anaerobic metabolic pathways in S. oneidensis strain MR-1. Besides confirming and clarifying previous reports on Fnr regulation, we also provide experimental evidence for a positive regulatory role of EtrA in the DMSO reduction pathway of strain MR-1. Furthermore, our whole-genome transcriptional profile shows the effects of EtrA on the expression of genes not previously evaluated (e.g. nqr, fdh-1, phage-and stress-related genes), and differences in the expression pattern of genes previously analyzed (e.g. cydAB and sdhC) [6,12]. These observations are consistent with results obtained by Gralnick et al. [4] suggesting a distinctive regulatory system, although very similar to Fnr in E. coli. A stringent sequence analysis of the regulatory region of the genes affected by the mutation suggest direct interaction of EtrA to those in the "Energy metabolism" category, while stress-and phage-related genes are up-regulated indirectly as a consequence of a secondary perturbation. This and previous work taken together suggest that this regulator is more properly termed Fnr.

Bacterial strains and culture conditions
The bacterial strains, plasmids, primers and, probes used in this study are described in Table 4. S. oneidensis strain MR-1 and its mutant strains were grown in HEPES medium as described [44]. The medium was supplemented with 20 mM lactate and KNO 3 was added as electron acceptor in concentrations specified below. Oxygen was removed from the medium by boiling and purging with helium [45]. Cultures of E. coli strain β2155 (auxotroph of diaminopimelic acid [DAP]) were grown in Luria-Bertani (LB) medium supplemented with 100 μg/ml of DAP at 37°C. S. oneidensis strain MR-1 was cultivated in aerobic LB medium at 30°C during the mutagenesis process. Antibiotics used for the selection of MR-1 transformants were added in the following concentrations: 25 μg/ml of kanamycin, 7.5 μg/ml of gentamycin, and 10 μg/ml of tetracycline. Vessels that received no inoculum or no KNO 3 served as negative controls.
A deletion mutant of the etrA gene (SO2356) was constructed by allelic replacement as described [44]. Primers etrAN Rev (SacI) and etrAN Fwd (SacII) generated a 520 bp fragment containing about two-thirds of the upstream SO2357 gene, the SO2357-etrA intergenic region, and the first four base pairs of the 753-bp etrA gene (Table 4). Primers etrAC Rev (NdeII) and etrAC Fwd (EcoRI) generated a 526 bp fragment containing the last five base pairs of the etrA gene, the etrA-SO2355 intergenic region and about half of the downstream SO2355 gene ( Table 4). The resulting 'SO2357-etrA'-loxP-kan-loxP-'etrA-SO2355' assembly was cloned into the conditionally-replicating plasmid pKNOCK-Gm (Table 4), which encodes resistance to gentamycin, to generate plasmid pCCG02. S. oneidensis MR-1 Km r Gm s colonies were screened by PCR using primers etrAScreenout Fwd and etrAScreenout Rev (Table 4) to identify recombinants in which the etrA gene was replaced by the loxP-kan-loxP cassette.
The kan gene was subsequently removed from the mutant MR-1 genome by Cre recombination [44,46]. PCR and DNA sequence analyses confirmed the presence of the chromosomal ΔetrA::loxP allele. DNA sequencing was performed at the Genomics Technical Support Facility at Michigan State University.
ΔetrA::loxP mutant complementation Plasmid pCM62 (Table 4) was used as the vector for the expression of the etrA gene in a ΔetrA::loxP mutant (strain EtrA7-1). The etrA gene (SO2356) was PCR amplified from S. oneidensis MR-1 genomic DNA using the etrAcomp Fwd (BamHI) and etrAcomp Rev (EcoRI) ( Table 4). The amplicon was double digested with BamHI and EcoRI and ligated to the multiple cloning site in pCM62. This construct (pCCG03) was transformed into EtrA7-1 by conjugation from E. coli β2155. Ligation, electroporation into E. coli β2155, and conjugation in strain EtrA7-1 were performed as described [44]. Plasmid pCM62 was also transformed into EtrA7-1 via conjugation from E. coli β2155 and used as a control for any plasmid effects. Transformants were selected by streaking on LB plates with tetracycline. EtrA7-1 Tc r colonies were diagnosed by PCR using the etrAcomp primers (Table 4) and subsequently sequenced to verify the deletion of the etrA gene.
Phenotypic characterization of the ΔetrA::loxP mutant Cultures of the wild type, EtrA7-1, EtrA7-1 complement and EtrA7-1 harboring pCM62 were grown anaerobically with 3 mM KNO 3 in HEPES medium. Growth was monitored periodically by OD measurements at 600 nm. Samples (2 mL) were periodically withdrawn for analysis of nitrate, nitrite and ammonium concentrations as described [44,47].
To determine the effects of lactate on the reduction of DMSO, nitrate and fumarate, cultures of the wild type and the EtrA7-1 mutant strain were grown anaerobically with 20 mM sodium pyruvate as the electron donor and dimethyl sulfoxide (DMSO; 1 mM), fumarate (10 mM) or nitrate (2 mM) as electron acceptors. DMSO and fumarate reduction were monitored as mentioned above. Nitrate reduction was measured using a Dionex ICS-3000 ion chromatograph (Dionex Corp., Sunnyvale, CA) equipped with an IonPac AS14 Column.

Resting cell assays
Cells of the wild type strain MR-1 and the EtrA7-1 mutant were grown aerobically in HEPES medium with 20 mM lactate as the electron donor to OD 600 = 0.2. Cells were harvested by centrifuging for 10 min at 3,000 g, washed twice with 50 mM saline phosphate buffer (pH 7.0) [52,53], and resuspended in the same buffer to an OD 600 = 1.0 under anoxic conditions. The cells were incubated with 5 mM KNO 3 , and after 0, 1, 2, 4, 8 and 24 h were removed by centrifugation and three 1-mL replicate samples of the supernatant were assayed to determine nitrate and nitrite reduction rates. Assays with autoclaved wild type cells served as negative controls. Nitrate, nitrite and ammonium concentrations were determined as described [44].

Total RNA preparations
Total RNA was extracted from triplicate cultures of strains MR-1 and EtrA7-1 grown with 2 mM nitrate as the sole electron acceptor. The RNA was extracted with RNAwiz Solution following the instructions of the manufacturer (Ambion, Inc., Austin, TX). RNA samples were treated with RNase-free DNaseI (Roche Pharmaceuticals, Basel, Switzerland) and purified by phenol:chloroform (1:1) and chloroform extractions [54], and stored in ethanol at -80°C until use. Quality of the RNA was verified using the RNA 6000 Pico LabChip kit and the 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA).

Global expression analyses
A S. oneidensis strain MR-1 whole genome microarrays [55] were provided by Liyou Wu and Jizhong Zhou (Oak Ridge National Laboratory, Oak Ridge, TN). cDNA preparation and labeling were performed as described [56] using a 2:3 ratio of 5-(3-aminoallyl)-dUTP and dTTP. Hybridization and post-hybridization washes were done as described [57]. Three biological replicates per treatment were used for the hybridization of six microarray slides including technical duplicates (dyeswap). Data analysis was performed using the Gene-Spring 6.0 software (Silicon Genetics, Redwood City, CA). The data were normalized per chip and per gene (Lowess Normalization) and the spots with less than 55% pixel intensity above background plus two standard deviations were eliminated from the analyses [58]. The data were filtered using the Benjamini and Hochberg false discovery rate with 95% confidence and only those genes with a > 2-fold change in expression were considered significant.

Identification of putative EtrA binding sites
Regulatory motifs were predicted in the intergenic regions of differentially expressed genes using the Gibbs centroid sampler [59]. Intergenic regions were extracted, based on the S. oneidensis MR-1 genome annotation, that were at least 50 bp in length and upstream of differentially expressed genes or operons whose change in expression (average ± one standard deviation) was at least 2.5-fold. A total of 170 genes met these requirements: 55 had expression levels minus one standard deviation of ≥ 2.5 (up-regulated genes) and 115 had expression levels plus one standard deviation of ≤ 0.4 (down-regulated genes). For these genes, 118 upstream intergenic regions were extracted for analysis, after accounting for multiple genes within an operon. The parameters for the Gibbs centroid sampler used on these sequences were the following: up to two motif models were allowed, where each model was specified to be palindromic and 16-24 bases long, a maximum of three sites per intergenic was allowed, a position-specific background model [60] was employed, and centroid sampling was performed with 1000 burn-in iterations, 5000 sampling iterations and 10 random seeds. The results from four independent runs were compared, and the subset of 47 intergenic regions extracted that contained a predicted regulatory motif in at least one of those runs. These 47 intergenic sequences were analyzed with the Gibbs centroid sampler, using the same parameters as above, except that only one motif model was specified. Additional binding sites were detected using dscan (http://ccmbweb.ccv.brown.edu/cgi-bin/dscan.pl) to search the set of promoters for all the genes that exhibited ≥ 2-fold change in expression (Additional file 1). This set included a total of 424 intergenic regions.

Additional material
Additional file 1: Supplemental Table SI1. Genes differentially expressed in anaerobic cultures of MR-1 and Etra7-1 at different concentrations of KNO3. Complete list of genes differentially expressed including relative expression, standard deviation, "TIGR role" and predicted EtrA binding sites.
Additional file 2: Figure SI1. Distribution of differentially expressed genes (> 2-fold change) grouped in 19 functional categories in anaerobic cultures of EtrA7-1 compared to the wild type grown on lactate and nitrate. Additional file 3: Figure SI2. Sequence logo (http://weblogo. berkeley.edu/logo.cgi) of the identified EtrA binding site motif for S. oneidensis MR-1. The logo represents the palindromic model of the aligned sites, showing the relative frequency of each base at each position of the motif. The Y-axis indicates the information content measured in bits. All of the predicted sites that contribute to the model are in Table SI1 in the supplementary materials.