Transcriptome analysis reveals response regulator SO2426-mediated gene expression in Shewanella oneidensis MR-1 under chromate challenge

Background Shewanella oneidensis MR-1 exhibits diverse metal ion-reducing capabilities and thus is of potential utility as a bioremediation agent. Knowledge of the molecular components and regulatory mechanisms dictating cellular responses to heavy metal stress, however, remains incomplete. In a previous work, the S. oneidensis so2426 gene, annotated as a DNA-binding response regulator, was demonstrated to be specifically responsive at both the transcript and protein levels to acute chromate [Cr(VI)] challenge. To delineate the cellular function of SO2426 and its contribution to metal stress response, we integrated genetic and physiological approaches with a genome-wide screen for target gene candidates comprising the SO2426 regulon. Results Inactivation of so2426 by an in-frame deletion resulted in enhanced chromate sensitivity and a reduced capacity to remove extracellular Cr(VI) relative to the parental strain. Time-resolved microarray analysis was used to compare transcriptomic profiles of wild-type and SO2426-deficient mutant S. oneidensis under conditions of chromate exposure. In total, 841 genes (18% of the arrayed genome) were up- or downregulated at least twofold in the Δso2426 mutant for at least one of six time-point conditions. Hierarchical cluster analysis of temporal transcriptional profiles identified a distinct cluster (n = 46) comprised of co-ordinately regulated genes exhibiting significant downregulated expression (p < 0.05) over time. Thirteen of these genes encoded proteins associated with transport and binding functions, particularly those involved in Fe transport and homeostasis (e.g., siderophore biosynthetic enzymes, TonB-dependent receptors, and the iron-storage protein ferritin). A conserved hypothetical operon (so1188-so1189-so1190), previously identified as a potential target of Fur-mediated repression, as well as a putative bicyclomycin resistance gene (so2280) and cation efflux family protein gene (so2045) also were repressed in the so2426 deletion mutant. Furthermore, the temporal expression profiles of four regulatory genes including a cpxR homolog were perturbed in the chromate-challenged mutant. Conclusion Our findings suggest a previously unrecognized functional role for the response regulator SO2426 in the activation of genes required for siderophore-mediated Fe acquisition, Fe storage, and other cation transport mechanisms. SO2426 regulatory function is involved at a fundamental molecular level in the linkage between Fe homeostasis and the cellular response to chromate-induced stress in S. oneidensis.


Background
Shewanella oneidensis MR-1, a facultatively anaerobic member of the γ-class of Proteobacteria, can respire anaerobically by reducing a wide variety of alternative electron acceptors including fumarate, nitrate, elemental sulfur, and such oxidized metals as Fe(III), Mn(IV), and U(VI) [1][2][3][4][5]. In addition, MR-1 is capable of reducing hexavalent chromium, or chromate [Cr(VI)], to less toxic and sparingly soluble trivalent chromium [Cr(III)] under both aerobic and anaerobic conditions [6][7][8]; however, there have been no reports to date of the ability of MR-1 to generate energy for growth using Cr(VI) as the sole terminal electron acceptor for anaerobic respiration [9]. This plasticity in the use of alternative electron acceptors for anaerobic respiration has engendered interest in S. oneidensis MR-1 as a model environmental organism with potential utility in the bioremediation of dissolved metal ions, and as a consequence, the complete MR-1 genome was sequenced to advance understanding of Shewanella biology [10].
To expand knowledge of metal stress responses in particular, we previously initiated genome-based studies focused on identifying the molecular components involved in the S. oneidensis response to chromate [11][12][13], an anthropogenic pollutant widely distributed in the environment due to its prevalent use in manufacturing and military industries [14,15]. Chromate toxicity is associated with the generation of reactive oxygen species during the intracellular partial reduction of Cr(VI) to the highly reactive radical Cr(V) by various in vivo nonspecific reductants or cellular one-electron reductases [16]. Our previous work employing DNA microarrays and multidimensional liquid chromatography-tandem mass spectrometry (LC-MS/MS) demonstrated that a functionally unknown DNA-binding response regulator (designated SO2426) in MR-1 was reproducibly and significantly upregulated at both the transcript and protein levels in response to acute chromate challenge [11,13].
In accordance with its ability to adapt to various environmental conditions, the S. oneidensis MR-1 genome encodes a relatively large repertoire of transcriptional regulators, including 88 predicted two-component regulatory system proteins consisting of 23 histidine protein kinase (HK) genes, 57 response regulators (RR), and 8 HK-RR hybrids [10]. Prototypical two-component systems, which constitute the predominant mechanism used by bacteria for coupling environmental signals to specific adaptive responses, comprise a sensor histidine kinase and a cognate response regulator [17][18][19]. Fundamental to signal transduction is a phosphotransfer mechanism that forms the core of both mechanistically simple as well as more complex signaling pathways. In this scheme, a membrane-bound sensor kinase catalyzes ATP-dependent autophosphorylation at a conserved histidine residue in the HK cytoplasmic autokinase domain in response to a specific environmental cue. The signal is transduced through transfer of the phosphoryl group from the phos-phoHis of the HK to a conserved aspartate residue within the receiver module of a cytosolic RR. Phosphorylation of the RR alters its affinity for DNA, leading to the transcriptional activation or repression of specific genes. An essential feature of two-component systems is that specific histidine kinases and response regulators function as cognate pairs and are often arranged within the same operon. The S. oneidensis SO2426 protein appears to be an orphan RR as defined by the fact that there is no obvious cognate histidine kinase flanking the so2426 gene or proximally located based on the genome annotation [10].
The primary aim of the present study was to investigate the function of the uncharacterized SO2426 regulator within the context of chromate stress by performing a genome-wide screen for potential target genes of this RR in S. oneidensis MR-1. Toward this end, we created an so2426 in-frame deletion mutant and compared its global transcriptional profiles with that of wild-type MR-1 in response to chromate challenge using time-resolved microarray experiments. Comparative transcriptome analysis suggested that SO2426 was associated with the activation of genes involved in siderophore-mediated Fe uptake, Fe storage, and possibly other cation transport processes.
The deduced SO2426 protein consists of an N-terminal CheY-like receiver domain and a C-terminal winged-helix DNA-binding domain. The N-terminal portion of SO2426 (amino acids 13-124; see Figure 1) shares homology with receiver domains and contains the highly conserved signature residues that are thought to constitute the phosphorylated acid-pocket active site [19,21]. In CheY, these residues correspond to Asp12, Asp13, and Asp57, with Thr87 and Lys109 completing the cluster of conserved residues surrounding the phosphorylation site [19]. Thr87 and Lys109, although not absolutely required for phosphorylation, have been implicated in contributing to the phosphorylation-induced conformational change [22][23][24]. S. oneidensis SO2426 contains three contiguous aspartate residues near its N-terminus at positions 18, 19 and 20, and there is a conserved substitution to serine at the equivalent position of 87. The phosphateaccepting aspartic residue Asp57 and Lys109 (relative to CheY residue numbering) are retained at equivalent positions in the SO2426 primary sequence. It is presently not known whether a cognate histidine kinase interacts with SO2426 or whether phosphorylation plays a regulatory role in the cellular activity of SO2426. The fact that SO2426 contains the highly conserved residues constituting the active phosphorylation pocket (D18, D19, D62, and K109) suggests that the protein might be differentially controlled by phosphorylation. Furthermore, it Sequence alignment of S. oneidensis SO2426 with other two-component response regulators Figure 1 Sequence alignment of S. oneidensis SO2426 with other two-component response regulators. ClustalW [20] was used to perform a multiple sequence alignment consisting of S. oneidensis SO2426 (GenBank ID 24348419), CpxR from E. coli (GenBank ID 16131752) and Vibrio cholerae (GenBank ID 147675245), and OmpR from E. coli (GenBank ID 16131282) and V. cholerae (GenBank ID 147673571). The region underlined with "=" is the aligned regulator receiver domain with predicted domain (SO2426: positions 13-124), and the region denoted with "~" is the aligned C-terminal domain containing the wHTH DNA-binding motif (SO2426: positions 158-235). Boldface letters highlighted in grey indicate conserved signature residues of receiver domains [19]. Residue D62 is predicted as 4-aspartylphosphate, the putative phosphorylation site (highlighted in yellow). The star, colon, and dot notations rank the sequence conservation from high to low.
The C-terminal portion of SO2426 (amino acids 158-235; see Figure 1) contains a predicted winged helix-turnhelix (wHTH) motif indicative of the DNA-binding domains of response regulators in the OmpR/PhoB subfamily [25][26][27]. The basic structure of the carboxy-terminal domains of OmpR/PhoB subfamily response regulators is characterized by an amino-terminal four-stranded β sheet, a central three-helical bundle, and a C-terminal β-strand hairpin (the wing), which provides an additional interface for DNA contact [26].

The so2426 gene is co-transcribed in an operon
Based on the MR-1 genome annotation, SO2426 is an apparent orphan response regulator given that the so2426 locus is not linked with a gene that encodes a potential cognate histidine kinase. As schematically represented in Figure 2A, so2426 is tightly clustered with four functionally unknown downstream genes, all possessing the same transcriptional polarity. Located immediately downstream of so2426 with a short intergenic spacer of 10 bp is a small ORF (132 bp) predicted to encode a soluble hypothetical protein (SO2425), which was shown previously to be induced at the transcriptional level in response to acute chromate challenge [11]. The gene cluster is also characterized by ORFs encoding a zinc carboxypeptidase domain protein (SO2424), a hypothetical protein (SO2423), and a conserved hypothetical protein (SO2422) (Figure 2A). The so2423 ORF overlaps so2422 by 3 bp based on the MR-1 genome annotation (http:// www.tigr.org). The so2426 region also includes a gene encoding a putative TonB-dependent receptor (SO2427) positioned approximately 170 bp upstream of so2426.
Reverse transcription-PCR (RT-PCR) experiments were performed to determine whether the so2427-so2422 gene cluster is transcribed as a polycistronic mRNA. Primers P1-P9 ( Figure 2A) used in the PCR reactions following reverse transcription with random hexamers were designed to amplify regions spanning two or more ORFs in order to establish the transcriptional organization. To assess the reproducibility of the results, multiple independent RT-PCR experiments were carried out using different preparations of total RNA isolated from untreated or chromate-treated MR-1 cells. Figure 2B shows a representative gel of the RT-PCR results using the entire suite of primers. Each PCR product was confirmed by DNA sequencing. No products were amplified when using either total RNA as the template for PCR amplification ( Figure 2B) or primers complementary to an intergenic region located outside of the so2427-so2422 cluster (results not shown), thus indicating the absence of contaminating genomic DNA in the total RNA preparations. S. oneidensis MR-1 genomic DNA was used as a positive control for the PCR conditions (results not shown).
Amplicons of the expected sizes, i.e., 0.85 kb for primers P1/P5, 1.9 kb for primers P1/P6, 2.3 kb for primers P1/P7, 2.5 kb for primers P2/P8, and 3.3 kb for primers P9/P3, were obtained ( Figure 2B), indicating that the so2427-so2422 region is co-transcribed as a polycistronic mRNA under no-metal and chromate-amended conditions. However, given the presence of a putative ρ-independent terminator immediately downstream of so2427, we cannot rule out the possibility that the product generated from the so2427-so2426 transcript might be due to readthrough transcription.

Characterization of the so2426 promoter region
The transcription start site of the so2426 gene was localized using 5' RACE analysis of mRNA transcripts isolated from wild-type cells grown in either the absence or presence of chromate. Two clear 5' termini, both mapping to A residues and corresponding to positions 25 and 27 bp upstream of a second in-frame ATG (amino acid residue M11) (Figure 3), were obtained by DNA sequencing of the 5' RACE amplicons. These sites are located at positions 4 and 6 bp downstream of the initial ATG (M1) start codon, indicating that the original sequence annotation for this gene is incorrect and that the second ATG (M11) most likely constitutes the translation start codon for SO2426. The so2426 transcription start point indicated potential -10 and -35 basal promoter elements, 5'-TAatAT and 5'-gTGACA, respectively, with a 15-bp spacing between them. The putative -10 promoter sequence (5'-TAatAT) has a 4/6 match to the consensus σ 70 -dependent -10 hexamer element (5'-TATAAT) from E. coli. Interestingly, located one base upstream of the -10 hexamer is the extended -10 element 5'-TaTG-3' (Figure 3), identified by Mitchell et al. [29] and shown to be an important determinant for promoter activity. Finally, the identified so2426 promoter region contains the putative -35 element 5'-gTGACA, which is a strong match (5/6) to the E. coli consensus -35 hexamer (5'-TTGACA).
Further analysis of the so2426 upstream regulatory region revealed the presence of a putative recognition site for the ferric uptake regulator (Fur), the iron-binding global transcriptional regulator that serves as the dominant sensor of Fe availability in both gram-positive and gram-negative bacteria (for a review, see [30]). A putative palindromic Fur box, 5'-AAATGAtATTgATTcTCgTTT-3', which closely matched our previously derived predicted Fur-binding motif for S. oneidensis MR-1 [31], was identified at positions -5 to +16 immediately downstream of the -10 pro-moter element and overlapped the mapped transcription start sites for so2426 ( Figure 3). Identification of a Fur box is consistent with our previous microarray studies demonstrating increased so2426 mRNA levels in a S. oneidensis MR-1 fur knockout strain [31] and suggests that expression of the SO2426 response regulator might be subject to Fur-mediated repression.

Growth and chromate reduction kinetics of the so2426 deletion mutant
A cre-lox recombination method [32,33] (17)  A B -77 to +645 (S. oneidensis MR-1 genome coordinates 2534690 and 2533969) from the annotated translational start site, leaving only 69 nucleotides of the so2426 gene intact at the C-terminal end. The growth behavior of the Δso2426 mutant in response to Cr(VI) was examined using a Bioscreen C reader to monitor culture turbidity (OD 600 ) at 30-min intervals under aerobic conditions over 72 h. As shown in Figure 4A, growth of the S. oneidensis Δso2426 strain was comparable to the wild type in LB broth without added chromate. By contrast, growth of the Δso2426 mutant was inhibited to a greater extent than the wild type in the presence of 0.3 mM chromate as evident from the longer time interval required before the commencement of exponential growth.
To determine whether deletion of so2426 affected the ability of MR-1 to remove extracellular Cr(VI) from the medium, the wild-type and mutant strains were challenged with 0.3 mM chromate (the same final concentration used for the transcriptomic studies) when cells reached mid-log phase (OD 600 , 0.5). Residual hexavalent chromium was measured spectrophotometrically at 3-h Nucleotide sequence of the so2426 promoter region and N-terminal coding region Figure 3 Nucleotide sequence of the so2426 promoter region and N-terminal coding region. The transcriptional start site (+1) is indicated by arrows, and putative -10 and -35 basal promoter elements are presented in boldface and underlined. Closed circles indicate conserved residues matching a predicted Fur-binding recognition site. The annotated (original) translation start codon (M1 residue) of so2426 is shown in boldface. 5'-RACE results described here indicate that the annotated translation start codon is incorrect, and residue M11 constitutes the actual start codon. The 5'-RACE primers used to identify the transcription start site are shown in italics and underlined. [34]. The chromate disappearance profiles in Figure 4B show that the parental MR-1 strain removed 100% of the external Cr(VI) within 9 h, while the Δso2426 mutant transformed approximately 50% within the same time period. No abiotic conversion of chromate was observed for the cell-free LB control (data not shown).
Complementation experiments were performed to ascertain whether the enhanced chromate sensitivity and slower chromate transformation rate for the so2426 null mutant were due to the deletion of the response regulator gene so2426 and not to secondary-site mutations introduced during the mutagenesis procedure. A complemented Δso2426 strain was constructed by cloning the full-length so2426 wild-type ORF plus 100 bp of contiguous upstream sequence into the broad-host-range vector pBBR2426 as described in the Methods. Transformation of pBBR2426 into the Δso2426 mutant resulted in a complemented strain, designated Δso2426(so2426 + ). Genetic complementation of the so2426 gene in the mutant strain restored growth (data not shown) and Cr(VI) reduction capacity to near wild-type levels ( Figure 4B), thus reaffirming the integral role of SO2426 in the cellular response to chromate.  Quantitative real-time RT-PCR (qRT-PCR) analysis was employed to assess the general quality of the microarray data by providing an independent measurement of gene expression for a small subset of genes. The five genes (so0404, so2426, so3670, so1826, and so3585) selected for comparative qRT-PCR displayed downregulated or unaltered expression patterns at the 180-min time point of chromate exposure based on microarray hybridization. Comparison of gene expression levels as determined by microarray and qRT-PCR analyses indicated a high degree of concordance, with a Pearson's correlation coefficient (r) value of 0.99 (data not shown). For the mutant strain, both microarray hybridization and qRT-PCR indicated that so2426 expression levels were not detectable above background, thus providing additional confirmation of deletion of the so2426 gene.

Functional classification of genes differentially expressed
in the Δso2426 mutant Pairwise complete-linkage clustering was used to identify groups of potentially co-ordinately regulated genes among the 841 ORFs (18% of the arrayed genome) showing a twofold or greater statistically significant (p < 0.05) change in expression for at least one time point during Cr(VI) exposure. One particularly noteworthy cluster of similarly expressed genes (Cluster A in Figure 5) comprises 46 genes with functions distributed across the following role categories: cellular processes (2 genes), energy metabolism (9 genes), hypothetical proteins (17 genes), protein synthesis (1 gene), regulatory functions/signal transduction (4 genes), and transport and binding proteins (13 genes). These clustered genes displayed a distinct pattern of downregulated expression in the Δso2426 mutant relative to the wild-type MR-1 strain, particularly at the earlier time points (5, 30, 60, and 90 min) of chromate exposure (see Table 1). This is consistent with our previous studies demonstrating that the wild-type so2426 gene was induced at the transcriptional level early in Cr(VI) exposure, i.e. 90 min or less, but not at 24 h posttreatment [11,12]. Closer inspection of the 46 potentially co-regulated genes (Table 1) revealed that many of the transport/binding, hypothetical, and regulatory genes were shown previously to be upregulated in wild-type MR-1 cells in response to chromate treatment [11,13], thus suggesting that these ORFs are candidate gene targets of positive control by SO2426. The genes constituting Cluster A are discussed in greater detail in the sections below.
Cluster B ( Figure 5) includes 20 genes, three of which encode AcrB/AcrD/AcrF family proteins (SO1882, SO3484, SO4692) presumably involved in drug efflux and resistance. All of these 20 genes showed decreased expression in the Δso2426 mutant except at the 24-h time point, where mRNA levels increased ( Figure 5 and Addi-tional file 1). Notably, Cluster B also contains genes so0392 (putative lipoprotein) and so4688 (glycosyl transferase, group 2 family protein), both with annotated functions associated with the cell envelope, as well as cpxA (so4478). In E. coli, CpxA is the sensory inner membrane kinase that functions with its cognate partner, a cytoplasmic response regulator (CpxR), in a prototypical twocomponent system to sense and respond to perturbations in the bacterial cell envelope (reviewed in [35]). S. oneidensis cpxR also displayed downregulated temporal expression profiles in the Δso2426 mutant under chromate challenge but grouped with genes in Cluster A.

Genes encoding transport and binding functions
Thirteen of the genes in Cluster A have annotated functions in metal transport and binding, in particular Fe acquisition and homeostasis: ftn (ferritin), so1580 (TonBdependent heme receptor), so1771 (GntP family permease), so2045 (cation efflux family protein), alcA (siderophore biosynthesis protein), so3031 and so3032 (both putative siderophore biosynthesis proteins encoded immediately downstream of alcA), so3033 (ferric alcaligin siderophore receptor), so3063 (sodium:alanine symporter family protein), so4150 (putative transporter), viuA (ferric vibriobactin receptor), so4712 (putative ABC transporter, ATP-binding protein), and so4743 (putative TonBdependent receptor). In contrast to the TonB-dependent receptor genes so1580 and so4743, either no significant change or a maximal twofold decrease at the 180-min time point was observed for so2427 (putative TonBdependent receptor), located directly upstream of the so2426 response regulator gene. Of the 13 genes with transport and binding functions in Cluster A, ten ORFs (ftn, so1580, so2045, alcA-so3031-so3032, so3033, so3063, viuA, so4743) were shown in an earlier study to be induced in chromate-challenged wild-type MR-1 cells compared to untreated cells [11,13]. Temporal expression patterns for these genes demonstrated that they were downregulated 2-to 20-fold in the Δso2426 strain over the 180-min time course (Table 1), suggesting that SO2426 acts as a direct or indirect positive regulator of a subset of Fe uptake and storage genes in S. oneidensis.
The alcA-so3031-so3032 operon encodes proteins required for siderophore biosynthesis in MR-1 [36], and the product of gene so3033 is predicted to allow for the cellular utilization of the structurally undetermined MR-1 siderophore. The differential profiles for these four genes were characterized by a peak in down-regulated expression (ranging from ~7-to 17-fold) at the 180-min time interval post chromate addition, followed by no significant change in expression at 24 h ( Table 1). The first gene in the siderophore biosynthetic operon shows 48% sequence identity to Bordetella pertussis alcA, which is required for alcaligin production [37], while the two  Complete linkage clustering analysis of genes with altered expression profiles in the Δso2426 mutant Cluster A Cluster B accumulation in the Δso2426 mutant. In the presence of 0.3 mM chromate, replicate cultures of wild-type MR-1 exhibited increasing levels in relative siderophore production over time (Figure 6), which was consistent with previous microarray expression data showing induction of siderophore biosynthesis genes in response to chromate exposure [11]. By contrast, the temporal profile for the Δso2426 mutant showed an initial moderate increase and then a dramatic reduction in siderophore excretion compared to the wild type ( Figure 6). Collectively, these data support the hypothesis that SO2426 is a positive regulator of siderophore-mediated Fe uptake.
Cluster A also includes an S. oneidensis gene that is involved in Fe storage, i.e., the ftn gene (so0139) encoding ferritin. The ftn mRNA expression levels were down-regulated 4-to 20-fold in the SO2426 null mutant across the entire temporal range, with the peak in repression occurring at the 180-min time point (Table 1). Although iron is an essential micronutrient for most organisms, free excess Fe in the presence of oxygen is potentially toxic because of its tendency to participate in Fenton chemistry reactions, which generate cell-damaging reactive oxygen intermediates. By scavenging and storing intracellular free Fe in a remobilizable form, Ftn protects against Fe(II)-mediated formation of hydroxyl radicals [39,40]. Studies have indicated that ferritins make important contributions to both Fe storage as well as cellular protection against oxidative stress [41,42] and that ftn transcription is co-ordinately regulated by iron and redox stress in some bacteria [43]. S. oneidensis ftn was shown previously to be induced at a low level (~2-fold) in chromate-exposed wild-type MR-1 cells in contrast to untreated cells [11]. The induction of ftn in S. oneidensis might act, at least partially, to mitigate chromate-generated oxidative stress by sequestering excess free Fe available for the catalysis of Fenton reactions and the formation of additional reactive hydroxyl radicals.

Genes of unknown function
The majority of coordinately repressed genes in Cluster A belong to the functional category of hypothetical and conserved hypothetical proteins (Table 1). Most notably, genes comprising the conserved hypothetical so1188-so1189-so1190 operon exhibited the highest degree of down-regulated expression in the chromate-exposed Δso2426 mutant relative to the wild type, with decreases in mRNA ratios ranging from 45-to > 300-fold over the 180min period. No significant changes in expression, however, were observed for these genes at the 24-h time point. In addition, the hypothetical gene so2425, which is cotranscribed with so2426 in a polycistronic transcript (shown in this study), was repressed as much as 11-fold over the 180-min period, pointing to the possibility that the so2426 operon may be under positive autoregulatory control. Surprisingly, the functionally unknown genes so2424, so2423, and so2422, which also form part of the so2426 operon, showed either a less than 2-fold change or no significant differential expression over the time course.

Regulatory genes
In addition to metal transport and hypothetical genes described above, the temporal expression profiles for four transcriptional regulatory genes (so0916, so0544, cpxR, so4567) were altered in the Δso2426 mutant. The function of the regulators encoded by so0916, so0544, and so4567 in S. oneidensis as well as their target genes are not known. However, gene so4477 encodes a homolog of CpxR, the response regulator component of the CpxAR envelope stress response system, which has been well characterized in E. coli (reviewed in [35,44]). The annotated cpxR gene in S. oneidensis MR-1 was downregulated as much as threefold in response to the so2426 deletion under chromate stress conditions (Table 1). This is in contrast to the twoto three-fold induction measured for cpxR in chromatetreated wild-type MR-1 cells [11]. Located immediately downstream of cpxR in an apparent operon is cpxA, which was downregulated as much as 3-fold in the mutant but grouped in a separate cluster (Cluster B in Figure 5).
Relative siderophore production by wild-type MR-1 and Δso2426 mutant cells under chromate conditions Figure 6 Relative siderophore production by wild-type MR-1 and Δso2426 mutant cells under chromate conditions. A final chromate concentration of 0.3 mM was added to midlog-phase (OD 600 , 0.5) wild-type MR-1 (•) and Δso2426 mutant (❍) cultures cultivated in LB broth, and relative siderophore synthesis by each culture was monitored over time using the CAS assay. The relative siderophore production was calculated by subtracting the supernatant A 630 for the wild type or mutant from the uninoculated control and then determining the ratio of corrected supernatant A 630 to control A 630 . The graphed values represent mean ratios ± standard errors (bars) for three replicate CAS measurements. In E. coli, the CpxAR two-component pair constitutes a stress response system the main function of which is to sense and respond to cell envelope distress, principally misfolded periplasmic proteins, by activating genes encoding proteases and folding catalysts [35,45,46]. However, the complexity, extent, and response overlap of the Cpx system have become progressively apparent, as evident from its involvement in outer membrane porin expression [47], stationary-phase survival [48], and the response to high pH stress [49] to name a few. Although the specific cellular role of the CpxR homolog in S. oneidensis has not been established, it is intriguing to consider the possibility that interplay between CpxR and SO2426 may occur in response to metal stress. It is conceivable that the CpxA-CpxR two-component system may respond to oxidative damage imposed on biomolecules as a consequence of reactive oxygen species generated during intracellular partial Cr(VI) reduction.

Other differentially regulated genes in the so2426 deletion mutant
While hypothetical and metal transport/binding genes are dominant in Cluster A (representing 65%), other notable genes with ≥ 2-fold-repressed mRNA levels in the Δso2426 mutant include ORFs codifying a putative bicyclomycin resistance (Bcr) protein (SO2280) and a ferric iron reductase protein (SO3034). Bcr proteins are members of the major facilitator superfamily class of membrane efflux pumps and play important roles in drug resistance [50]. In a previous work, the so2280 gene in a wild-type MR-1 genetic background exhibited 2-to 4-fold upregulated expression at the transcript level in response to short-term chromate exposure, i.e., up to 90 min [11], but was shown in this study to be repressed as much as 4-fold in Cr(VI)challenged Δso2426 cells (Table 1), suggesting that the so2280 gene is a direct or indirect target of SO2426-mediated activation under chromate conditions. Although the cellular function of SO2280 is presently unknown, its coordinately regulated expression with the siderophore biosynthesis operon (alcA-so3031-so3032) and gene so3033, encoding a ferric alcaligin siderophore receptor, is intriguing. Studies in other bacterial systems have shown that certain proteins resembling Bcr at the amino acid level function in siderophore export from the cell [51].
Downregulated expression of gene so3034 (a putative ferric iron reductase) peaked at the 90-min time point with a 4-fold decrease in mRNA levels (Table 1). Ferric iron reductases are thought to be involved in reducing siderophore-complexed Fe(III) to Fe(II), thus enabling dissociation of the high-affinity Fe 3+ -chelating siderophore from iron [30]. The entire so3030-so3034 gene cluster, which is predicted to be involved in siderophore production and utilization, exhibits similar temporal expression profiles in the mutant, suggesting a role for the SO2426 response regulator in coordinating these functions. It is not clear whether the other differentially expressed energy metabolism genes (fdhG, fdnH, fdhE, so4503, so4506, so4509) contribute directly to chromate stress response or show altered transcription as a result of secondary effects associated with the so2426 deletion. These genes were not observed to be induced in response to chromate exposure in our previous transcriptomic studies using wild-type S. oneidensis MR-1 [11].

Conclusion
S. oneidensis SO2426 is an apparent orphan response regulator with an N-terminal domain that contains an aspartate residue at the canonical receiver phosphorylation site and a predicted C-terminal wHTH motif indicative of response regulators in the OmpR subfamily. Although previous work in this laboratory demonstrated enhanced mRNA levels for the so2426 gene in a S. oneidensis Fur-deficient strain [31] and in chromate-challenged wild-type MR-1 cells [11], the basic molecular function of this predicted transcriptional regulator remained undefined. In this study, by integrating functional genomics and genetic approaches, we have shown that SO2426 is a direct or indirect positive regulator of siderophore-mediated Fe transport systems, the Fe storage gene ftn, and other as-yet uncharacterized cation transport genes. At this time, however, we cannot discount the possibility of negative regulation by SO2426 as well. Clustering analysis of temporal transcriptome data revealed a distinctive subset of 46 coordinately expressed genes that were consistently downregulated in the so2426 deletion mutant during the initial 180-min period of chromate exposure. Many of these genes -namely, ftn, the alcA-so3031-so3032 siderophore biosynthesis operon, so1580 (TonB-dependent heme receptor), so4743 (TonB-dependent receptor), viuA, so2280 (bcr), as well as a number of hypothetical genes -constituted the predominant molecular response to acute chromate stress in wild-type MR-1 [11]. The transcriptomic analyses described here identified a number of possible directly SO2426-regulated genes, including those of unknown function, that may serve as targets for future studies.
One possible hypothesis that can be generated from the collective data described here is that S. oneidensis MR-1 employs at least two control mechanisms for regulating Fe acquisition via siderophores under conditions of external Fe insufficiency: derepression as a result of decreased Fur activity and activation by SO2426 to enhance needed expression of Fe(III) transport systems. CAS siderophore assays clearly demonstrated that the Δso2426 mutant was deficient in siderophore production compared to wildtype MR-1 cultures in the presence and absence of chromate, indicating that SO2426 plays a positive role in coordinating the regulation of siderophore-dependent Fe uptake mechanisms in S. oneidensis. Furthermore, the impaired chromate reduction rate of the so2426 deletion strain points to a connection between Fe transport and chromate stress response that requires further delineation at the molecular level. Our previous global studies characterizing the transcriptome and proteome of wild-type MR-1 in response to sub-lethal acute doses of chromate demonstrated a dramatic and unexpected induction of genes involved in Fe sequestration and uptake in contrast to unchallenged cells [11,13]. Two possible explanations for this observation are (i) Fe limitation under external chromate conditions, and (ii) direct or indirect displacement of the Fe(II) cofactor from Fur-binding sites under elevated Cr(VI), resulting in decreased DNA-binding activity of the Fur repressor and subsequent derepression of target genes. Exposure to other metals has been shown to perturb intracellular Fe pools and hence influence Fur-mediated repression [52]. Future studies will focus on experimentally confirming direct gene targets of SO2426 control and further elucidating the linkage between regulation of Fe homeostasis and chromate stress response at molecular and physiological levels.

Bacterial strains, plasmids, and growth conditions
All bacterial strains and plasmids used in this study are listed in Table 2. S. oneidensis strain MR-1 [4,53] was used as the wild type. The Δso2426 strain, a derivative of MR-1, contains an in-frame deletion of the so2426 locus on the chromosome (see below for a detailed description of mutant construction). For growth studies, wild-type and mutant strains were cultivated aerobically with shaking (250 rpm) at 30°C in Luria-Bertani (LB) medium (pH 7.2) alone or in media amended with 0.3 mM (final concentration) of K 2 CrO 4 . Optical density was monitored at a wavelength of 600 nm (OD 600 ) in triplicate using either a Spectronic 20D+ spectrophotometer (Thermo Electron Cooperation, Waltham, MA) or the Bioscreen C microbiological culture system (Growth Curves USA, Piscataway, NJ) as described elsewhere [36].
For transcriptome analyses, three separate batch cultures each of wild-type MR-1 and the Δso2426 mutant were grown to mid-exponential phase (OD 600 , 0.5) in 100 ml of LB broth, followed by addition of 2 M K 2 CrO 4 to a final concentration of 0.3 mM. The two different sets of chromate-challenged cultures (wild type and mutant), each consisting of three biological replicates, were grown aerobically in parallel at 30°C. Cells were harvested for RNA extraction at 5, 30, 60, 90, 180 min, and 24 h post-chromate exposure. For RT-PCR and 5' RACE experiments, wild-type S. oneidensis MR-1 cells were grown to mid-log phase (OD 600 , 0.5) in LB broth at 30°C and then exposed to chromate (final concentration, 1 mM) for 30 min before harvesting cells for RNA extraction. Untreated MR-1 cells were grown in parallel and used as the control.

Chromate reduction and CAS siderophore assays
Disappearance of extracellular Cr(VI) was quantified spectrophotometrically using the 1,5-diphenylcarbazide (DPC) method as described elsewhere [34]. Cultures were assayed at different time points ( The chrome azurol S (CAS) assay for detection of siderophore production was performed essentially as  [54]. Siderophore biosynthesis and excretion by the Δso2426 mutant was compared with those by the wild-type MR-1 strain in LB broth with or without the addition of 50 μM FeCl 3 or 0.3 mM K 2 CrO 4 . Cultures of these strains were grown aerobically to stationary phase (OD 600 > 1.0) at 30°C for 24 h. Cellfree supernatants of S. oneidensis cultures were mixed 1:1 with the CAS assay solution and equilibrated at room temperature for 2 h before the absorbance at 630 nm was measured. The relative siderophore production was calculated as the ratio of the control (uninoculated LB medium) A 630 to the wild-type or mutant strain supernatant A 630 . All CAS measurements were performed in triplicate, and at least two independent determinations were conducted.

Construction of an S. oneidensis so2426 deletion strain and complementation
An S. oneidensis MR-1 in-frame deletion mutant lacking the so2426 locus was generated using the cre-lox recombination system as described elsewhere [32,33]. The application of this mutagenesis strategy to targeted S. oneidensis MR-1 genes has been described previously [55]. The primer sequences used in the generation and verification of the Δso2426 mutant are listed in Table 3. PCR was used to amplify a 900-bp region upstream and a 840-bp region downstream of the so2426 gene using primer pairs Del900-F/Del900-R and Del840-F/Del840-R, respectively. The amplified regions were cloned into the kanamycin-resistant (Km r ) plasmid pJK100 [32]. The resulting plasmid, pJK102, carrying a correct construct was introduced into the E. coli strain WM3064 [32], a diaminopimelic acid (DAP) auxotroph, and then subsequently moved into MR-1 via conjugation with WM3064. The resultant MR-1 strain was Km r /Tet s and contained the loxP-Km r -loxP cassette. Removal of the kanamycin cassette and the helper plasmid was performed as described elsewhere [33]. A cre-recombinase enzyme-producing plasmid, pCM157/Tet r , was introduced into MR-1 to resolve the loxP-Km r -loxP cassette through conjugation. The Tet r plasmid pCM157 was cured from MR-1 by continuous culturing in non-selective LB medium. The in-frame deletion of so2426 was confirmed by PCR amplification using several sets of primers (ISU2426/ISD2426, OP840, OP900, kanF/kanR in Table 3) and DNA sequencing.
The Δso2426 mutant was complemented by reintroduction of the wild-type so2426 allele on a low-copy-number  (Table 3). The PCR product was gel purified, digested, and ligated into the BamH1 and HindIII sites of the plasmid pBBR1MCS-5 [56], which contains a gentamicin resistance (Gm r ) cassette. The resultant construct harboring the complete S. oneidensis so2426 gene with its endogenous promoter was designated pBB2426. The pBB2426 plasmid was electroporated into E. coli strain WM3064 and then introduced into the S. oneidensis Δso2426 mutant and wild-type MR-1 strains via conjugation. In addition, plasmid pBBR1MCS5 without the insert (empty vector) was also transferred into the mutant and wild-type strains via conjugation with WM3064. Gm r Shewanella colonies were verified for the presence of the so2426 gene by PCR analysis and DNA sequencing.

RNA isolation
For microarray profiling and real-time RT-PCR experiments, total cellular RNA was isolated from S. oneidensis cultures using the TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. RNA preparations were treated with RNase-free DNase I (Ambion Applied Biosystems, Foster City, CA) to remove residual genomic DNA and subsequently further purified using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer's RNA cleanup protocol. Total RNA was quantitated as described previously [36]. For RT-PCR and 5'-RACE analyses, total RNA was extracted from S. oneidensis cultures using the RNeasy Mini Kit (Qiagen). Chromosomal DNA contamination was removed by incubating total purified RNA with 3 U of RNase-free DNase (Ambion Applied Biosystems) for 30 min at 37°C. The DNA digestion reaction was performed twice for each RNA sample. The quantity and purity of the RNA was assessed as described previously [36]. The integrity of all RNA samples was assessed visually using 1% agarose gel electrophoresis and ethidium bromide staining.

Microarray hybridizations and data analysis
Transcriptome analyses were performed with a microarray containing 4,197 PCR amplicons and 451 specific 50-mer oligonucleotides, covering approximately 94% of the total predicted gene content of S. oneidensis MR-1. Fabrication of the MR-1 arrays has been described in detail elsewhere [57]. Synthesis of the two differentially labeled cDNA pools (wild-type MR-1 and the Δso2426 mutant) to be compared, microarray prehybridization and hybridization, and post-hybridization washings were performed as described previously [36]. Temporal gene expression analysis was performed using six independent microarray hybridizations (three biological replicates × two dye-swap reactions) for each of six time points, with each slide containing two spots representing each gene at different array locations for a total of 12 signal intensity measurements per gene per time point. Image quantification, data normalization, and analysis of gene expression data for statistical significance were conducted as described by Brown et al. [36]. The time-series microarray expression profiles of the Δso2426 mutant relative to the parental strain were clustered using Hierarchical Clustering Explorer (HCE) [58]. During the clustering process, only genes with an expression value of at least ≥ 2-fold or ≤ 0.5-fold in one or more of 6 expression profiling time points were included in the analyses. As a result, a dataset of 841 genes was clustered based on average linkage using Euclidean distance.

Access to microarray data
The microarray data reported in this study have been deposited in MIAME-compliant format at Gene Expression Omnibus on the NCBI website [59] under series accession number GSE12129. The statistically analyzed microarray output is provided in Additional file 1.

Quantitative real-time RT-PCR
Reverse transcriptase, quantitative real-time PCR (qRT-PCR) was used to provide an independent assessment of gene expression for five selected genes (so0404, so2426, so3670, so1826, and so3585), which exhibited different expression patterns (i.e., down-regulated or no change). These genes are predicted to encode a hypothetical protein (SO0404), a DNA-binding response regulator (SO2426), proteins involved in transport and binding of cations (SO3670 and SO1826), and a putative azoreductase (SO3585). Relative expression patterns for each selected gene were independently confirmed using the following primer pairs, which were designed using the program Primer3 http://www-genome.wi.mit.edu/cgi-bin/ primer/primer3_www.cgi: SO0404, 5'-AGTATAAC-CAAGCGCCAGTA and 3'-GCATCGGTATTAACTTGCTC; SO2426, 5'-GCAGAAGGATTTAGGTCGAT and 3'-CGGT-GTTGATTAAAGTACGC; SO3670, 5'-TCTAAACAGTCG-CAGGAGCA and 3'-GCGCCATATTGCTATCCATT; SO1826, 5'-GGGTGTCCCAAGCTAGT CAA and 3'-GAG-CATTACTCGTCCCCTGA; SO3585, 5'-CGAGGCTATC-CATC ACTTAG and 3'-TGGAAAACACGATAAAGACC. Reverse transcription was performed on the 180-min time point samples with 3 μg of total cellular RNA and 2.5 mM random hexamers using Superscript™ II RNase H -Reverse Transcriptase (Invitrogen) as described previously [31]. Quantitative PCR was carried out in an iCycler iQ ® realtime PCR system (BioRad, Hercules, CA) in 50-μl reaction mixtures containing 1 μl cDNA, 600 nM forward and reverse primers, and iQ SYBR green supermix (BioRad) according to the manufacturer's instructions and conducted under the following conditions: 30 sec at 95°C, followed by 40 cycles of 15 sec at 95°C, 30 sec at the specific annealing temperature, and 30 sec at 72°C. The qRT-PCR reactions were performed in triplicate for each of the three biological replicates tested. Standards for each gene of interest were included in the analysis as described previously [31]. The final Δso2426/WT ratio for each target gene and the standard error were calculated as described elsewhere [31], and the linear correlation between the qRT-PCR and microarray data was determined based on the log mean values using Sigma-Plot version 9.0 (SPSS Inc., Chicago, IL).

RT-PCR and 5'-RACE analyses
Reverse transcription-PCR (RT-PCR) was performed to investigate the transcription of the so2427-so2422 gene cluster on the MR-1 chromosome. First-strand cDNA was synthesized from 1 μg of DNase-treated total RNA isolated from MR-1 cultures grown in LB broth under either non-stress (no chromate added) or chromate stress (1 mM, 30-min exposure) conditions. Random hexamers (250 ng), dNTPs (10 mM), 5× First-Strand buffer (4 μl), 0.1 M DTT (1 μl), RNaseOUT (1 μl), recombinant RNase inhibitor (40 U), and Superscript™ II Reverse Transcriptase (200 U; Invitrogen) were added to the reaction mixture, which was incubated at 25°C for 10 min, 42°C for 50 min, and then 70°C for 15 min in accordance with the recommendations of the supplier. Synthesized cDNA was used as the template in subsequent PCR amplification reactions with gene-specific oligonucleotide primers P1-P9 (Table 3) spanning the so2427-so2422 gene region. The TripleMaster PCR System (Eppendorf, Westbury, NY) was used for amplification, with primers added to each reaction mixture at a final concentration of 250 nmol. The reaction mixtures were subjected to 30 cycles of denaturation at 94°C for 60 s, annealing at 44-52°C (depending on the primer T m ) for 60 s, and an extension at 72°C for 1-2 min in an Eppendorf Mastercycler ep gradient thermocycler. Each amplified product was analyzed by 1% agarose gel electrophoresis with ethidium bromide staining. In addition, each PCR product obtained was cloned into the pGEM-T vector (Promega, Madison, WI) and sequenced at Purdue University's Low-Throughput DNA Sequencing Laboratory to confirm that they were the expected loci of each respective gene region.