The mxd operon in Shewanella oneidensis MR-1 is induced in response to starvation and regulated by ArcS/ArcA and BarA/UvrY

Background S. oneidensis MR-1 is a dissimilatory metal-reducing bacterium. Under anoxic conditions S. oneidensis MR-1 attaches to and uses insoluble minerals such as Fe(III) and Mn(IV) oxides as electron acceptors. In the laboratory, S. oneidensis MR-1 forms biofilms under hydrodynamic flow conditions on a borosilicate glass surface; formation of biofilms was previously found to be dependent on the mxd gene cluster (mxdABCD). Results This study revealed environmental and genetic factors regulating expression of the mxd genes in S. oneidensis MR-1. Physiological experiments conducted with a S. oneidensis MR-1 strain carrying a transcriptional lacZ fusion to the mxd promoter identified electron donor starvation as a key factor inducing mxd gene expression. Tn5 mutagenesis identified the ArcS/ArcA two-component signaling system as a repressor of mxd expression in S. oneidensis MR-1 under planktonic conditions. Biofilms of ∆arcS and ∆arcA strains carrying a transcriptional gfp -reporter fused to the mxd promoter revealed a reduced mxd expression, suggesting that ArcS/ArcA are necessary for activation of mxd expression under biofilm conditions. Biofilms of ∆arcS and ∆arcA mutants were unable to form a compact three-dimensional structure consistent with a low level of mxd expression. In addition, BarA/UvrY was identified as a major regulator of mxd expression under planktonic conditions. Interestingly, biofilms of ∆barA and ∆uvrY mutants were able to form three-dimensional structures that were, however, less compact compared to wild type biofilms. Conclusions We have shown here that the mxd genes in S. oneidensis MR-1 are controlled transcriptionally in response to carbon starvation and by the ArcS/ArcA and the BarA/UvrY signaling system. BarA might function as a sensor to assess the metabolic state of the cell, including carbon starvation, leading to expression of the mxd operon and therefore control biofilm formation.


Background
Shewanella oneidensis MR-1 is a dissimilatory metalreducing bacterium [1] and can use under anoxic conditions insoluble Fe(III) and Mn(IV) oxide minerals as electron acceptors [2,3]. In the laboratory, S. oneidensis MR-1 forms biofilms under hydrodynamic flow conditions on a borosilicate glass surface, where biofilm formation is mediated by a set of complementary molecular machineries, comprised of the type IV MSHA pilus and a putative exopolysaccharide biosynthesis (EPS) gene cluster (mxdABCD) [4,5]. The first gene of this cluster is mxdA, which is predicted to encode for a gene with unknown function; however, MxdA was recently shown to control indirectly cellular levels of c-di-GMP in S. oneidensis MR-1 [6]. MxdB has homology to a membrane-bound type II glycosyl transferase and was thought to be involved in the transport of extracellular material involved in forming the matrix of S. oneidensis MR-1 biofilms. This hypothesis was supported by genetic analysis revealing that ΔmxdB mutants were unable to transition from a cell monolayer to a three dimensional biofilm structure [4]. MxdC shares homology with an efflux pump and mxdD was annotated as a conserved hypothetical protein with no known homology. ΔmshAΔmxdB double mutants were entirely deficient in initial attachment and biofilm formation [5]. Expression of adhesion factors such as EPS are regulated in Vibrio cholerae, Escherichia coli and Pseudomonas aeruginosa in response to environmental factors. The vps gene cluster in V. cholerae, for example, was shown to be controlled in a cell-density dependent manner [7][8][9][10] involving several two-component signaling systems (TCS).
Recently, it was shown that phoshorylation of ArcA by ArcS requires the presence of HptA, a separate phosphotransfer domain [14]. HptA of S. oneidensis MR-1 shares homology with the N-terminal domain of ArcB, the sensor histidine kinase of the E. coli ArcB/ArcA system, but does not share significant homology with ArcS from S. oneidensis MR-1. ArcS/HptA have been shown to functionally complement an E. coli ΔArcB mutant [13]. In E. coli, ArcA is part of the ArcB/ArcA (anaerobic respiration control) two-component regulatory system, a major regulator of gene expression involved in aerobic/anaerobic respiration and fermentative metabolism [15][16][17][18][19][20]. A recent study investigated the domain structure of ArcS in S. oneidensis MR-1 and revealed significant differences when compared to E. coli ArcB [21]. It was shown that in the N-terminal part, ArcS possesses a CaChesensing domain, two cytoplasmic PAS-sensing and two receiver domains. Due to the expanded sensory region, ArcS of Shewanella species might be able to respond to a wider array of environmental signals and is not restricted to changing redox conditions.
ArcA has been previously shown to play a role in biofilm formation in S. oneidensis MR-1. S. oneidensis MR-1 ΔarcA mutants form biofilms with about 70% less biomass on a borosilicate glass surface under hydrodynamic flow conditions and are unable to mature into a highly three-dimensional biofilm structure when compared to wild type [22].
In this study, we investigated physiological and genetic factors involved in the regulation of the mxd operon in S. oneidensis MR-1. We found that mxd expression was induced by carbon starvation. The TCS ArcS/ArcA was discovered to constitute a major activator of the mxd genes under biofilm conditions, and to repress mxd expression under planktonic conditions. BarA/UvrY was identified as a major inducer of mxd expression under planktonic conditions and appeared to have a minor role in biofilm formation.

Results
ΔmxdA and ΔmxdB mutant cells are deficient in cell-cell aggregation when grown planktonically under minimal medium conditions Wild type S. oneidensis MR-1 cells, when grown for 16 h in a liquid minimal medium, formed a thick biofilm ring at the air-liquid interface on the borosilicate surface of a test tube ( Figure 1A). Stationary phase cultures (OD 6003 .2) aggregated in a rotating culture test tube and quickly settled to the bottom of the tube when rotation was arrested for 10 minutes ( Figure 1A). We took advantage of this aggregation phenotype and developed a quantitative aggregation assay by calculating the ratio of the optical density, measured at 600 nm, of cells before and after dispersion by rigorously vortexing ( Figure 1B). Analyzing wild type and mutants by this assay, we found ΔmxdA and ΔmxdB mutant cultures to be deficient in aggregation ( Figure 1). Consistent with this observation, the biomass of biofilms of these strains that formed at the air-liquid interface on the borosilicate glass test tube surface was dramatically reduced relative to wild type. Notably, the described aggregation and adhesion phenotypes were not observed under LB medium conditions.
These data indicate a possible role for mxdA and mxdB in cell-surface adhesion when growing in minimal medium. When comparing growth rates in LB to minimal medium, we found no correlation between growth rate and mxd expression, suggesting that a low growth rate, as found under starvation conditions in minimal medium, was most likely not responsible for mxd induction (data not shown).
We therefore hypothesized that limitation for essential nutrients or accumulation of metabolites might be involved in mxd induction, and specifically tested whether carbon or nitrogen limitation induced mxd expression. For this purpose we constructed a wild type mxd::lacZ reporter strain (AS832) (see Table 1 and 2). This strain was grown in LB medium to an OD 600 =0.3. Cells were pelleted, resuspended in minimal medium amended with 50 mM sodium lactate, incubated for 120 minutes at 30°C and subsequently assayed for specific βgalactosidase activity. Similarly, cells were also exposed to minimal medium without carbon or nitrogen source. As a control, cells were resuspended in the same LB culture medium. As shown in Figure 2 no increase in mxd expression was observed when cells were incubated in the LB culture medium for 120 minutes ( Figure 2) and compared to the same sample at t=0 minutes. Similarly, cells exposed to minimal medium void of a nitrogen source also did not show any increase in mxd expression. Cells exposed to minimal medium supplemented with lactate led to minor mxd induction. However, shifting cells to minimal medium void of a carbon source led to significant mxd induction (~400 MU). Thus, starvation for carbon appears to be important for mxd expression in S. oneidensis MR-1.
ArcS/ArcA functions as a repressor of the mxd operon in planktonic cells Tn5 mutagenesis was performed to identify genes regulating mxd expression. We subjected the wild type mxd:: lacZ reporter strain (AS832) to four independent rounds of Tn5 transposon mutagenesis. A total of 12,000 Tn5 insertion mutants were qualitatively screened for deregulated mxd expression by visually comparing colours of Kan-resistant colonies plated on X-gal plates relative to the parental strain. 48 out of 12,000 Tn5 insertion mutants were identified either as a loss-or gain-of-function mutants, respectively. After quantitative confirmation of the Tn5 mutant phenotypes by β-galactosidase assays (data not shown), Tn5 insertion sites were mapped. Among the selected Tn5 mutants, we found in two independent mutageneses insertions in the response regulator ArcA and its cognate histidine sensor kinase ArcS associated with a gain-of-function phenotype. In order to exclude polar effects due to the Tn5 insertions, we constructed in a wild type background marker-less in-frame deletions of arcS (AS841) and arcA (AS839), respectively (see Table 1 and 2). We then introduced the mxd::lacZ construct into these strains to generate strains AS860 and AS863, respectively, and examined mxd expression in these mutants when grown under LB medium conditions. As data in Figure 3 (top) show, a 12 times higher mxd expression in exponentially growing cells and about 1.5 times higher mxd expression in stationary phase cells was observed relative to wild type. Our data show that ArcS/ArcA is a major transcriptional repressor of mxd under planktonic conditions, and represses the mxd operon primarily in exponentially growing cells.
Further support for a direct role of the ArcS/ArcA system in control of mxd expression comes from a mxd promoter deletion analysis. The mxd transcription start site (+1) was experimentally determined by primer extension analysis and mapped at −150 bp (data not shown and Figure 4A). In silico analysis of the mxd promoter predicted putative ArcA binding sites at −29 bp, -86 bp and −112 bp upstream of the mxd transcription start site ( Figure 4A and B). In order to experimentally test these predictions, we created truncations in the putative mxd promoter region, and transcriptionally fused the truncated promoters to lacZ, yielding strains AS832-835 ( Figure 4B) (see Table 1 and 2). All strains were grown in LB medium, and cells from early exponential phase (2 h) through late stationary phase (24 h) were harvested and assayed for β-galactosidase activity ( Figure 4B). Interestingly, when deleting the region upstream of −100 bp from the transcriptional start site (AS834), expression was increased about eightfold during exponential growth phase (> 6 h) compared to reporter strains carrying mxd upstream regions deleted to −150 bp (AS833) and −300 bp (AS832) ( Figure 4B). As the ArcA binding sites were predicted at −29 bp, -86 bp and −112 bp upstream of the mxd transcriptional start site, the predicted −112 bp ArcA binding site is deleted in the −100 bp reporter strain (AS834), thus abolishing putative ArcA binding. Collectively, the observed data are consistent with the hypothesis that ArcS/ArcA is a major transcriptional repressor of the mxd operon under planktonic conditions.
BarA/UvrY is a major activator of mxd expression in planktonic cells In the above reported transposon mutageneses, we also identified uvrY (SO1860) to transcriptionally control mxd.
Recently biochemical evidence showed that BarA  Table 2 Primers used in this study Mxd expression in S. oneidensis MR-1 wild type. Mxd expression in S. oneidensis MR-1 wild type cells exposed to carbon and nitrogen starvation conditions. Wild type cells carrying the mxd promoter transcriptionally fused to lacZ were grown under complex medium conditions to mid-exponential phase. Cells were pelleted and the supernatant was discarded. Subsequently, cells were either resuspended in minimal medium containing no carbon source (MM minus carbon, or in minimal medium amended with 50 mM sodium lactate (MM plus carbon), or in minimal medium with the nitrogen source omitted (MM minus nitrogen). A set of control samples (black bars) was pelleted and resuspended in the same medium. Samples were assayed for β-galactosidase activity. Data represent an average of three independent experiments.
is the cognate sensor histidine kinase of UvrY, and that BarA/UvrY in S. oneidensis MR-1 constitute a functional two-component regulatory system [23]. We therefore constructed markerless in-frame deletions of barA (AS845) and uvrY (AS843), inserted the mxd::lacZ construct into each strain, and determined mxd expression in strains AS861 (AS845 background) and AS862 (AS843 background) under LB medium conditions (see Table 1 and 2). As evident from Figure 5 (top), both ΔbarA and ΔuvrY mutants showed drastically reduced mxd expression primarily in stationary phase. Furthermore, we observed that ΔbarA and ΔuvrY mutant strains, when grown for 24 h under minimal medium conditions, failed to aggregate under planktonic conditions, similar to a ΔmxdB (AS831) mutant ( Figure 1A and Figure 5). These data provide genetic evidence that BarA/UvrY might function as an activator of the mxd operon under planktonic growth conditions. This conclusion is further supported by the observation that ΔbarA and ΔuvrY mutants exhibit a ΔmxdB phenotype when grown planktonically in minimal medium.
ArcS/ArcA and BarA/UvrY regulate formation of hydrodynamically-grown biofilms The above data showed that ArcS and ArcA act as repressors of mxd expression, whereas BarA and UvrY strongly activate mxd expression under planktonic growth conditions. We next examined whether these regulators have a function under biofilm conditions. Biofilms of wild type, ΔarcS, and ΔarcA mutants were grown under hydrodynamic biofilm conditions, and biofilms were imaged by CLSM at 24 h and 48 h postinoculation. Interestingly, both ΔarcS and ΔarcA mutant biofilms were unable to form a three-dimensional biofilm structure, and their biofilms were of similar structure as mxd mutant biofilms ( Figure 6). As this finding was opposite to what we had expected based on the ΔarcS and ΔarcA mutant phenotypes in planktonic cells, we examined whether the biofilm phenotype of ΔarcS (AS842) and ΔarcA (AS840) mutants was indeed due to down-regulation of mxd. A transcriptional P mxd ::gfp reporter strain was constructed and introduced into wild type (AS837), ΔarcS (AS856) and ΔarcA (AS855), respectively. Biofilms of wild type (AS837), ΔarcS (AS856) and ΔarcA (AS855) carrying the P mxd ::gfp reporter were grown for 24 h in LM medium, harvested from the flow chamber and analyzed by flow cytometry for GFP fluorescence intensity (see Table 1 and 2). To account for non-specific background signals, a wild type strain carrying a promoterless gfp -reporter construct (AS838) was used as a control. While on average about 40% of the cells derived from a wild type biofilm showed P mxddependent GFP fluorescence above background, only about 1% of the cells from ΔarcS and ΔarcA biofilms did so (Additional file 1: Figure S1), consistent with the previously observed biofilm defect. These data provide evidence at the level of global biofilm structure as well as of single cell gene expression that ArcS/ArcA is necessary for significant positive activation of mxd expression in S. oneidensis MR-1.
ΔbarA and ΔuvrY mutants formed well-developed three-dimensional structures that were less compact compared to wild type ( Figure 6). These data therefore suggest that BarA/UvrY plays only a minor regulatory role under biofilm conditions.

Discussion
Carbon starvation induces mxd gene expression in S. oneidensis MR-1 While investigating physiological factors inducing mxd expression in S. oneidensis MR-1, we discovered that expression of the mxd genes in S. oneidensis MR-1 were regulated differentially depending on whether carbon starvation conditions prevailed under planktonic or biofilm conditions (Figure 7). The data showed furthermore that arcA/arcS as well as barA/uvrY are important regulators of mxd expression although under different conditions (Figure 7). Consistent with our data, earlier findings in P. aeruginosa and E. coli had shown that nutrient-depletion enhanced biofilm formation, while high concentrations of nutrients repress the formation of biofilms [24,25]. In nature, accessible organic carbon is often scarce and can be found sorbed to surfaces such as organic-rich flocculates of marine snow and fecal pellets. Being able to sense and respond to changing carbon concentrations in these environments is crucial to the survival of bacteria. While starvation for carbon generally leads to a decrease in growth rate and metabolic activity in bacteria, our data suggest that S. oneidensis MR-1 cells activate production of adhesion factors responsible for biofilm formation under these conditions. This acclimation strategy could potentially confer an ecological advantage for S. oneidensis MR-1 cells interacting with a carbon rich surface, as they are already primed for adhesion.
BarA/UvrY functions as an activator of the mxd genes under planktonic growth conditions and has a role in the regulation of biofilm formation We showed here that BarA/UvrY activates mxd expression under organic rich medium conditions when planktonic cells entered stationary phase (Figure 7). BarA/ UvrY is highly conserved in Gram-negative bacteria, and controls a variety of physiological functions including carbon storage [26][27][28][29][30]. In carbon storage regulation (Csr) BarA/UvrY regulates small RNAs controlling elements of this pathway, which are major posttranscriptional regulators of biofilm formation in E. coli [31]. The stimuli for the BarA sensor histidine kinase in E. coli are aliphatic carboxylic acids, such as formate, acetate, propionate and others, providing a physiological signal reflecting the metabolic state of cells and thereby linking posttranscriptional control by the Csr system with central metabolism [30].
Interestingly, S. oneidensis MR-1 biofilms of both ΔbarA and ΔuvrY mutants formed less compact biofilms when grown under hydrodynamic flow conditions. Based on these data and the above discussed findings that low carbon concentration induces mxd expression, we hypothesize that BarA might function as a sensor for carbon starvation, e.g., at high cell density when nutrients become growth limiting in planktonic culture. We hypothesize that under these conditions starvationsensing BarA signals to UvrY, which, in return, directly or indirectly activates mxd expression and, by this cascade, controls biofilm formation. Homologous of BarA/UvrY have been shown to control secondary metabolism, including the excretion of biofilm exopolysaccacharides in other γ-proteobacteria [32][33][34][35][36]. In the closely related bacterium Pseudomonas fluorescens production of several antibiotic-like secondary metabolites is regulated by the orthologs GacA/GacS and via the small RNAs RsmXYZ [37]. In P. fluorescens expression of these small RNAs was found to be positively controlled by GacS/GacA at high cell density and intermediates of central metabolism such as 2-oxoglutarate, succinate and fumarate which may be present at elevated intracellular concentration under conditions when cells are electron acceptor-limited [37]. It is conceivable that S. oneidensis MR-1, similar to P. fluorescens, senses its metabolic state at the level of primary metabolites, and uses the level to control aspects of secondary metabolism including biofilm formation. The BarA/UvrY system and its components have been studied to some extent in S. oneidensis MR-1 [23]. It was found to contain all major components of the BarA/UvrY/Csr pathway. UvrY in S. oneidensis MR-1 positively regulates the two small RNAs, csrB1 and csrB2 and a corresponding CsrA ortholog was also identified. The transcriptional fusion construct used in our study, which contains the mxd Shine Dalgarno sequence, is able to reflect control in mxd expression on the transcriptional and posttranscriptional level. Therefore, the decreased mxd expression detected in the barA and uvrY mutants might be a result of transcriptional regulation by uvrY which directly or indirectly interacts with the mxd promoter or a posttranscriptional control possibly via CsrA or both.
Interestingly, S. oneidensis MR-1 biofilms of ΔbarA and ΔuvrY mutants were only partially defective ( Figure 6). These biofilm defects might be a consequence of the idiosyncrasy of a biofilm environment: microbial biofilms are nutrient-stratified environments where cells at the surface of the biofilm have better access to nutrients, including oxygen, whereas cells in the layers distant from the planktonic interface become increasingly nutrient limited. If the BarA/UvrY system responds to lower concentrations of organic substrates, this regulator might be activated in the deeper, nutrient-deprived layers of the biofilm. Consequently, in the absence of BarA or UvrY part of the biofilm population would not express the mxd genes and confer adhesion, leading to a loosely structured biofilm such as observed in ΔbarA and ΔuvrY mutants.

The ArcS/ArcA TCS functions as a repressor of the mxd genes under planktonic growth conditions and activates the mxd operon in a biofilm
We identified and showed here that the ArcS/ArcA system controls mxd expression in S. oneidensis MR-1. Even though a role for ArcA in S. oneidensis MR-1 biofilm formation was previously introduced, no mechanistic explanation was provided. Our data show that ArcS/ ArcA act as a repressor of the mxd genes under planktonic conditions (Figure 7, left) while it activates mxd expression in the biofilm (Figure 7, right).
The two different modes of action under planktonic and biofilm conditions could be explained as a consequence of additional mxd regulation at the transcriptional level. Unidentified transcriptional regulators could alter the transcriptional mxd output we observe in ΔarcS and ΔarcA mutants under planktonic and biofilm conditions. Due to the ecological differences that cells experience in planktonic culture and in a biofilm, the response in terms of mxd expression would then be very different. A further possibility is that ArcA receives signal inputs from other sensor kinases in addition to ArcS. Lassak et al. provided biochemical evidence showing that the ArcS/ArcA TCS in S. oneidensis MR-1 is only functional in the presence of a phosphotransfer domain HptA [14]. The function of phosphotransfer domains is not entirely clear, but they are thought to serve as a means to integrate signal inputs from several sensor kinases and relay that information to the cognate response regulator. Depending on whether a cell experiences planktonic growth conditions or is part of a structured biofilm, the input signals can vary greatly, and, as a consequence, mxd expression can be very different in these environments. Further investigation will be availability by sensing the redox state of the quinone pool. Based on the homology of the two Arc systems, it is possible that Arc has a similar function in S. oneidensis MR-1. To test whether expression of the mxd operon was regulated in response to metabolic changes, and more specifically to redox changes (oxic/anoxic), via the Arc system, experiments with S. oneidensis MR-1 wild type strains carrying a copy of lacZ fused to the mxd promoter under controlled chemostat-like conditions had been conducted. Strains were cultivated in a batch fermenter in LB medium or LB medium amended with 50 mM sodium fumarate and grown aerobically (dissolved oxygen was monitored during the entire experiment) to exponential phase and then shifted to anoxic growth conditions by depleting oxygen. βgalactosidase activity in these strains was monitored before and up to 12 hours after the shift. No change in mxd expression was observed upon oxygen depletion (data not shown). This led us to the conclusion that a change in redox conditions and metabolic activity per se (induced by electron acceptor starvation) did not play a role in Arc mediated mxd regulation. Based on recently published data, revealing that Shewanella ArcS possesses additional sensory regions when compared to ArcB in E. coli, the Arc system in Shewanella species might also be able to sense other unknown environmental signals [28].

Conclusions
The presented data show that carbon starvation is the dominant environmental cue triggering mxd induction in S. oneidensis MR-1, and that the mxd genes are controlled transcriptionally by ArcS/ArcA and BarA/UvrY. Interestingly, BarA/UvrY appears to be a major regulator of the mxd genes and is primarily responsible for induction in cells that have entered stationary phase and are exposed to starvation conditions while ArcS/ArcA appears to control mxd expression independent of growth phase. Although the signal for the BarA sensor histidine kinase has not been identified in S. oneidensis MR-1, it is reasonable to speculate that it is of similar molecular nature as the recently identified metabolites for E. coli BarA. However, considering that E. coli and S. oneidensis MR-1 inhabit different ecological niches, it is also conceivable that the signal input might be different. Thus, we hypothesize that based on our data carbon starvation could be the physiological signal sensed by BarA directly or indirectly. Both ΔbarA and ΔuvrY mutants were unable to induce mxd expression when cells entered stationary phase. In fact, mxd expression in both mutants resembles the expression level observed in logarithmically growing wild type cells, indicating a possible role for BarA/ UvrY in starvation response.

Strains and media
Strains used in this study are listed in Table 1. E. coli strains were grown at 37°C in lysogeny broth (LB) medium. Where necessary medium was solidified by 1.5% (w/v) agar and supplemented with 50 μg/mL kanamycin or 100 μg/mL ampicillin. S. oneidensis MR-1 strains were grown at 30°C in LB medium, lactate Construction of mxd transcriptional reporter strains S. oneidensis MR-1 mxd reporter strains were constructed by transcriptionally fusing various-length fragments of the mxd upstream region to lacZ and gfp. A promoterless copy of either lacZ or gfp in the appropriate vector served as a control.

LacZ -reporter strains
To obtain a strain reporting on the transcriptional activity of mxd, a 450 bp fragment upstream of the mxdA translation initiation site was amplified with primers Pmxd-fw-SphI and Pmxd-rv-XbaI (Table 2) using S. oneidensis MR-1 genomic DNA as template.

Gfp -reporter strains
To construct a strain reporting on the transcriptional activity of mxd, a 450 bp fragment upstream of the mxdA translation initiation site was amplified with primers Pmxd-fw-HindIII and Pmxd-rv-XbaI (Table 2) using S. oneidensis MR-1 genomic DNA as template. The PCR product was purified from an agarose gel, restriction digested with HindIII and XbaI and ligated into a HindIII and XbaI restriction digested pProbe NT vector yielding pJM6. All reporter constructs were introduced into E. coli S17-λ pir by standard procedures. Plasmid was then prepared from positive clones and introduced into S. oneidensis MR-1 wild type or mutant strains by electroporation.
Quantitative cell aggregation assay S. oneidensis MR-1 wild type and mutant cells were grown in test tubes on a roller drum to exponential (OD 600 = 0.3) and stationary phase (OD 600 = 2.0) in minimal medium amended with 50 mM sodium lactate. Immediately after removing test tubes from the roller drum, one milliliter samples were taken and OD 600 was determined. Further samples were taken after 15 minutes and 30 minutes. After measuring the optical density, cells were vigorously vortexed for 20 seconds and the optical density measurement was repeated. The ratio of OD 600 before and OD 600 after dispersion was calculated and used as an approximation to estimate the extend of cell aggregation in the different strains.

Construction of gene deletions
S. oneidensis MR-1 in-frame deletions were constructed by homologous recombination. The deletion constructs were created by amplifying the regions flanking the target gene. The fragment length was optimized to about 750 bp. The primers for the 5'-end fragment were 5-O (outside) and 5-I (inside) and the primers for the 3'-end fragment were 3-I (inside) and 3-O (outside). Subsequent to amplification, the flanking regions were fused via a complementary tag that was added to the 5'-end of each inner primer. The fusion product was inserted into the cloning vector pDS3.1 and the mobilizing strain E. coli S17-λ pir [38] was transformed with this sucicide vector. Functionality of the sacB gene was verified before transferring the deletion vector by conjugation into the S. oneidensis MR-1 target strain. Single crossover events were selected for on LB plates containing gentamycine and confirmed by using two primer combinations: 1) primer X-F and primer 3-O and 2) primer X-R and primer 5-O, whereas primer X-F and primer X-R will bind upstream and downstream of the flanking regions, respectively. The functionality of the sacB gene was verified in S. oneidensis MR-1 strains that tested positive for a single crossover event. Resolution of the integrated vector by a second crossover event was performed with a positive strain. This strain was grown in LB medium without selection and plated onto solid LB medium containing 10% sucrose. Deletion events were verified by PCR using primer X-F and primer X-R, where a successful deletion resulted in a PCR product with a size of the wild type product minus the size of the target gene.

Construction of strains constitutively expressing GFP
Construction of S. oneidensis MR-1 strains constitutively expressing GFP was carried out using a Tn7 based delivery system [39]. GFP-labeling was performed by biparental mating. Cultures of S. oneidensis MR-1, AS262 and AS392 were grown in LB broth overnight. 0.5 mL of each culture containing about 10 8 cells was washed twice in one culture volume of phosphate buffered saline (PBS). S. oneidensis MR-1 and AS262 cells were combined and resuspended in 250 μL PBS. AS392 cells were resupended in 250 μL PBS. 50 μL of the mixed S. oneidensis MR-1/AS262 cell suspension was combined with 50 μL AS392 cell suspension and spotted onto dry solidified LB medium. Petri dishes were incubated upright for 8 h at 30°C. The cell mass was then resuspended in PBS and spread onto LB agar supplemented with 10 μg/mL gentamycine to select for S. oneidensis MR-1 carrying a chromosomal insertion of the gfp-carrying Tn7. PCR was used to map the site of insertion in the S. oneidensis MR-1 genome.

Tn5 mutagenesis and screen for mxd -deregulated mutants
Transposon mutagenesis was performed by mating AS536 with the donor strain E. coli BW20767 (AS259) harbouring suicide plasmid pRL27, which carries a hyperactive transposase and a Tn5-mini transposon with a kanamycin resistance cassette and a R6K origin of replication [40]. The mating was performed at a 1:1 donor-recipient ratio at room temperature for 6 h. Transconjugants were plated onto solid LB medium containing kanamycin, tetracycline and X-gal to qualitatively screen for deregulated mxd mutants. Mutants were identified based on the intenstity of their blue colony color compared to the non-mutagenized control strain AS536. The mutant phenotypes were quantitatively confirmed by β -galactosidase assay in liquid culture. The location of a Tn5 insertion was mapped by arbitrary primed PCR [4]. Chromosomal DNA was prepared from the mutants and two rounds of amplification were used to specifically amplify and enrich for the DNA flanking the insertion site. In the first round primer tpnRL 17-1-O or tpnRL 13-2-O, which are unique to one end of the transposon, and two different arbitrary primers ARB1 and ARB6 [4] were used for amplification. Among the many possible amplified regions from the first round of PCR were products primed from the transposon and flanking chromosomal DNA. Products flanking the transposon were specifically amplified in the second round of PCR with primers tpnRL17-1 or tpnRL13-2 [4] and ARB2. After the second round of PCR the obtained PCR products were purified and subsequently subjected to DNA sequence analysis using primers tpnRL17-1 or tpnRL13-2. To identify the location of the transposon insertion, the resulting nucleotide sequences were compared with the S. oneidensis MR-1 sequence database by BLAST search: (http://blast.ncbi.nlm.nih.gov).

β -galactosidase assay
For β -galactosidase assays, S. oneidensis MR-1 strains were grown at 30°C in 250 mL flasks containing 25 mL of either LB medium or LM or 4M medium amended with 50 mM lactate. Specific activities were determined by a modified Miller method [41]. Briefly, cells were harvested during different growth stages and resuspended in Z-Buffer to an OD 600 of 0.5-0.7. Samples were prepared in triplicates by adding 100 μL of cell suspension to 900 μL Z-buffer with 0.27% (v/v) βmercaptoethanol, 50 μL chloroform and 100 μL 0.1% SDS and vortexing for 10 seconds. After equilibration at 28°C for 10 minutes, the reaction was started by addition of 0.2 mL o-nitrophenyl-D-galactoside (ONPG) [4 mg * mL -1 ] and incubating the samples at 28°C. The reactions were stopped with 0.5 mL Na 2 CO 3 [1M] when samples developed a yellowish color. Samples were centrifuged for 5 minutes at 13,000 rpm and OD 420 was recorded. Specific activities were expressed as Miller Units and calculated as follows: 1 Miller Unit = 1000 * (OD 420 )/(t * V * OD 600 ), where t = time V= volume OD= optical density

Biofilm cultivation
Biofilms were grown at 30°C in three-channel flow cells as decribed previously [12]. Briefly, LB overnight cultures of the relevant S. oneidensis MR-1 strains were diluted 1/100 in LB and grown to early stationary phase. Then the optical density at 600 nm was adjusted to 0.01 in 4M MM or LM without carbon source. 1 mL of the OD 600 = 0.01 cell suspension was injected into each flow channel while the medium flow was stopped. The flow cells were inverted (glass slide facing bottom) and incubated for 40 min at 30°C. After incubation flow cells were reverted and medium was pumped through the flow cell at a constant velocity of 0.3 mm/s per channel by a Watson-Marlow Bredel (Cornwall, United Kingdom) 205S peristaltic pump. Biofilm studies were carried out in triplicate in at least two independent experiments.

Biofilm image acquisition and processing
Microscopic visualization of biofilms was performed using an upright Leica TCS SP2 AOBS confocal laser scanning microscope (CLSM; Leica Microsystems, Wetzlar, Germany) using the following objectives: HCX PL APO 63X/1.2 W CORR CS and HC PL FLUORTAR 20X/0.5. For three-dimensional reconstruction of biofilm images, CLSM images were processed with the IMARIS software package (Bitplane AG, Zuerich, Switzerland) and Adobe Photoshop.

Flow cytometry
24 h old LM grown biofilm of S. oneidensis MR-1 wild type and mutant cells carrying a P mxd ::gfp reporter construct were harvested from the flow chamber, passed 50 times through a 25 gauge needle to suspend any cell aggregates and fixed in 2% paraformaldehyde. Flow cytometry data were obtained using a BD FACSCalibur flow cytometer (BD Biosciences, San Jose, CA). Samples were analysed using the 488 nm excitation from an argon-ion LASER at 15 mW. Detector voltages were set at defined values [800 V for the fluorescence channel (FL1) and both the FL1 and forward scatter channel amp gain were set to logarithmic scale] prior to the experimental analysis in which samples were run in succession on the same day. A control sample of wild type strain MR-1 biofilm cells carrying promoterless gfp (AS838) was used for background subtraction. Data acquisition and analysis was performed with CellQuest (BD Biosciences) software.

Additional file
Additional file 1: Figure S1. Expression of mxd in S. oneidensis MR-1 wild type and ΔarcS and ΔarcA mutant biofilms. GFP fluorescence intensities of S. oneidensis MR-1 wild type, ΔarcS and ΔarcA biofilm mutant cells measured by flow cytometry. All strains carried a P mxd ::gfp reporter and were grown in LM in a hydrodynamic flow chamber for 24 h. Biofilm cells of wild type strain MR-1 carrying promoterless gfp were used as a control for background subtraction. Fluorescence intensities were calculated as a percentage of the total cell population after background subtraction. Data represent one of two performed experiments with similar trends.