Regulation of Energy Metabolism by the Extracytoplasmic Function (ECF) σ Factors of Arcobacter butzleri

The extracytoplasmic function (ECF) σ factors are fundamental for bacterial adaptation to distinct environments and for survival under different stress conditions. The emerging pathogen Arcobacter butzleri possesses seven putative pairs of σ/anti-σ factors belonging to the ECF family. Here, we report the identification of the genes regulated by five out of the seven A. butzleri ECF σ factors. Three of the ECF σ factors play an apparent role in transport, energy generation and the maintenance of redox balance. Several genes like the nap, sox and tct genes are regulated by more than one ECF σ factor, indicating that the A. butzleri ECF σ factors form a network of overlapping regulons. In contrast to other eubacteria, these A. butzleri ECF regulons appear to primarily regulate responses to changing environments in order to meet metabolic needs instead of an obvious role in stress adaptation.

In 2007 the complete genome of the human clinical isolate A. butzleri RM4018 was sequenced [15]. The bacterium appears to have a large number of signal transduction systems, indicating that this organism is able to respond to many different environmental signals. Apart from having one of the highest densities of two component systems per megabase, A. butzleri RM4018, unlike Campylobacter spp. or Helicobacter spp., is also predicted to contain seven s/anti-s factor pairs belonging to the ECF family (Extracytoplasmic function) of sigma factors. The seven A. butzleri RM4018 ECF s factor pairs are genetically unlinked but for each pair, the s factor gene is positioned adjacent to and in the same orientation as the putative cognate anti-s factor gene. The amino acid identity between any two of the seven ECF s factors varies between 30 and 49%, suggesting that each ECF pair plays a different role in the biology of strain RM4018 and responds perhaps to a different suite of external signals.
ECF s factors have been identified in both Gram-negative and Gram-positive bacteria and they are fundamental for bacterial adaptation to different environments and for survival under different stress conditions [16,17]. Their role in cellular physiology is highly variable and includes adaptation to membrane-affecting compounds, extreme temperatures and pH, light, high pressure, carbon, nitrogen and iron starvation, oxidative and osmotic stress, and the regulation of virulence in pathogenic organisms [16,18]. Despite their diverse effects, ECF s factors share common features. In most cases ECF s factors are co-transcribed with their cognate anti-s factor. The anti-s factor, which often spans the inner membrane, sequesters and inactivates the cognate s factor at the inner face of the cytoplasmic membrane. Upon a specific external stimulus, the ECF anti-s factor is inactivated by protein degradation and the s factor is released from the membrane, where it associates with RNA polymerase and mediates transcriptional activation of its target genes [19]. The ECF anti-s factor can also be released from the ECF s factor by a phosphorylated response regulator NepR, this partner-switching mechanism has recently be discovered in alphaproteobacteria [20]. Some ECF s/ anti-s factors form together with an outer membrane TonBdependent receptor, a trans-envelope signal transduction pathway [31]. These TonB-dependent receptors are involved in the uptake of specific molecules but also they sense and transmit, via the antis factor, extracellular signals, which lead to the activation of a specific ECF s factor.
In the present study, we investigated the role of the A. butzleri RM4018 ECF s factors. We developed the genetic tools to manipulate A. butzleri and determined the regulons of most ECF s factors using a combination of microarray-based transcriptome analysis and functional assays.

Bacterial Strains, Plasmids and Growth Conditions
Bacterial strains and plasmids used in this work are listed in Table 1. A. butzleri strains were routinely grown at 30uC in Brain Heart Infusion (BHI) broth (Oxoid) or on Mueller Hinton (MH) agar (Oxoid) supplemented with 5% sheep blood (Biotrading). E. coli strains were routinely grown at 37uC in Luria-Bertani (LB) broth or on LB agar plates (Biotrading) supplemented with ampicillin (100 mg/ml) or kanamycin (50 mg/ml) when needed.

Construction of A. Butzleri ECF s Factor and ECF Anti-s Factor Mutants
The genes encoding the seven ECF s factors and their cognate anti-s factors were amplified using the primer pairs AB0986F/ AB0987R, AB1044F/AB1044R, AB1437F2/AB1437R2, AB1460F/AB1460R, AB1576F/1576R, AB2151F/AB2151R and AB2300F/2300R (Table 2), the proofreading enzyme Pfu (Promega) and A. butzleri RM4018 chromosomal DNA as template. The PCR products were tailed with a 59-A nucleotide using Taq polymerase (Invitrogen) and ligated into the pGEM-T Easy vector (Promega) to obtain pGEMab0983-0984, pGE-Mab1040-1041, pGEMab1429-1430, pGEMab1452-1453, pGEMab1567-1568, pGEMab2164-2165 and pGEMab2315-2316 (Table 1). Inverse PCR was performed on the s/anti-s plasmids to delete a large part of both the s and anti-s factor (s/ anti-s knock-out plasmids) or of only the anti-s factor encoding genes (anti-s knock-out plasmids). Unique BamHI restriction sites were introduced at the same time. The plasmid pGEMDab1429-1430 was obtained by digesting plasmid pGEMDab1429 with BamHI and BclI. The knockout constructs were created by digestion of the inverse PCR products with BamHI and ligation to a 1.4-kb BamHI fragment containing a kanamycin resistance gene (aph(39)-III) from pMW2 [21].
The ECF s factor genes were inactivated by marker exchange mutagenesis. First, the knock-out plasmids were introduced into A. butzleri RM4018 by electrotransformation. To obtain electrocompetent A. butzleri RM4018, a 5 ml overnight culture was diluted 20 times in 100 ml fresh BHI medium and incubated at 30uC on a shaking platform (150 rpm). Bacteria were harvested by centrifugation (4,5006g, 1 h, 4uC) when the optical density at 550 nm had reached values between 0.2 and 0.6. The bacteria were washed twice in 5 ml of ice-cold sucrose-glycerol solution (15% glycerol; 272 mM sucrose in water), resuspended in 0.5 ml ice-cold sucroseglycerol solution and aliquoted into 50 ml solutions containing approximately 325610 9 CFU. One mg of each plasmid was added to 50 ml (approximately 325610 9 CFU) of competent A. butzleri RM4018 and incubated for 3 min on ice. The cells were transferred to a 0.2 cm electroporation cuvette (Bio-Rad) and electroporated using a Bio-Rad Gene Pulser set at 2.25 kV, 400 V and 25 mF. Bacteria were recovered in 1 ml of BHI broth for 10 min at room temperature, transferred to 2 ml of pre-warmed BHI and incubated for 3 hours at 30uC on a shaking platform (150 rpm). Mutants were selected by growing the cells for two to five days at 30uC aerobically on kanamycin-containing MH plates. Homologous recombination resulting in double-crossover events was verified by PCR.

Phenotype Characterization
Growth curves were generated by diluting pre-cultures grown overnight in BHI to a starting optical density (OD 550 ) of 0.05 in 30 ml of BHI. The 30 ml cultures were grown in conical flask under aerobic conditions, 150 rpm at 30uC. Bacterial growth and cell density were monitored by measuring the absorbance at 550 nm at different time intervals. The exponential growth rate was calculated from four separated growth experiments. Several stress conditions which limited the growth of the wildtype Arcobacter strain were tested. To measure the influence of extreme temperatures, the starting cultures were incubated at 4uC or 60uC for 15 min prior to their incubation at 30uC. To study the effect of the pH on the growth of the strains, the pH of the BHI was adjusted to pH 5 or 9. Osmotic stress was tested by adding 0.35 M NaCl to the BHI. In additional experiments, other stressinducing chemicals were added to the BHI to concentrations which limited the growth of the wildtype Arcobacter strain. These chemicals included: ethanol (5%); SDS (1%); the iron chelator 2,2dipyridyl (300 mM); the oxidative stress-inducing chemicals H 2 O 2 (0.04%) and diamide (2 mM); or antimicrobial compounds such as penicillin G and polymyxin B (25 mg/ml). Motility assays were performed by stabbing the strains with a pipette tip into semisolid medium (thioglycolate medium containing 0.4% agar, Difco), followed by incubations under aerobic conditions at 30uC or 37uC for 48 hours.

RNA Isolation
Overnight grown cultures of A. butzleri were diluted to an OD 550 of 0.1 in BHI and incubated at 30uC on a shaking platform set at 150 rpm. RNA was isolated from 5 ml of mid-logarithmic phase cultures (OD 550 of approximately 0.5), using the RNA-Bee TM kit (Tel-Test, Inc) following the manufacturers specifications.

Microarray Hybridization and Analysis
For expression profiling, an indirect comparison of gene expression levels was performed [22,23]. In this microarray experimental design, each labeled cDNA was combined with labeled genomic DNA from A. butzleri RM4018. Mixtures were hybridized to a previously designed and manufactured A. butzleri DNA array [15]. Labeling of RNA and DNA, hybridization procedure and microarray data analysis were performed as previously described [24]. Details of the microarray have been deposited in the NCBI GEO repository (http://www.ncbi.nlm. nih.gov/geo/) under platform accession number GPL14948. The microarray data set has been deposited in the NCBI Gene Expression Omnibus (GEO) repository (http://www.ncbi.nlm.nih. gov/geo/) under accession number GSE34089.

Real-time RT-PCR
Primers used in this assay were designed using the Primer Express software (Applied Biosystems) and are listed in Table 2. Prior to amplification, RNA samples were treated with RNase-free DNase I (Invitrogen). RT-PCR was performed on 0.2 mg of DNase I treated RNA with 1 mM of primers and the SYBRH Green I kit (Eurogentec) using a LightCyclerH 480 Real-Time PCR System (Roche). The PCR parameters were 30 min at 48uC;  [25]. Each sample was examined in four replicates and was repeated with at least two independent preparations of RNA. Standard deviations were calculated and displayed as error bars.

Nitrate/nitrite Assay
The nitrate reductase activity was determined by measuring the production of nitrite from nitrate as previously described [26]. Briefly, strains were grown overnight (16 h) under aerobic conditions in BHI broth containing 20 mM of sodium nitrate. Nitrite accumulation in the supernatant was detected by mixing 50 ml of the culture supernatants with 850 ml of 1% (w/v) sulphanilamide dissolved in 1 M of HCl and 100 ml of 0.02% (w/v) naphthylethylenediamine solution. After 15 minutes the formation of p-sulfobenzene-azo a-naphthylamine was measured at 540 nm. The amount of nitrite present in 50 ml culture supernatant was estimated using a nitrite standard curve. The nitrite production was adjusted to the total bacterial proteins present in the culture as estimated by using the BCA protein assay kit. To determine the nitrate concentration in Arcobacter culture supernatants the supernatants were 10 times diluted in BHI. The diluted culture supernatants were incubated with sulphanilamide and naphthylethylenediamine as described for the nitrite detection. Next a trace amount of zinc dust was added to the mixture. The zinc dust reduces nitrate to nitrite and a red color develops due to the formation of p-sulfobenzene-azo a-naphthylamine. After 2 minutes the zinc has settled on the bottom and 900 ml of the supernatant was transferred to a new tube and the absorbance of the red color was measured at 540 nm. The amount of nitrate present in 50 ml culture supernatant was estimated using a nitrate standard curve which was also treated with zinc dust. Reactions were measured four times and repeated with two independent cultures.

Sox Enzyme Activity Assay
Overnight cultures in BHI were washed once with Arcobacter Minimal 1 (AM1) medium (4.2 mM Na 2 HPO 4 , 2.4 mM KH 2 PO 4 , 9 mM NaCL, 19 mM NH 4 Cl 16M9 salts, 1 mM of MgCl 2 , 0.2% of sodium pyruvate, 0.02% of FeNH 4 citrate; adjusted to pH 7.5) amended with 20 mM of sodium sulfite, and 10 mg of phenol red per milliliter. The bacteria were subsequently diluted in sodium sulfite-, phenol red-amended AM1 medium to an OD 550 of 0.05 and allowed to grow in completely filled capped 5 cm glass vials. The strains which were able to oxidize sulfur caused a drop in the pH that was visible as a color change from red to yellow. After 6 hour of growth at 30uC, absorbance at 564 nm was measured and the results obtained for the wild-type strain were set at 100%.

Citrate Uptake Assay
The ability of the strains to utilize citrate was determined using Simmons citrate agar (Oxoid). Strains grown overnight on MH agar supplemented with 5% sheep blood were streaked onto Simmons Citrate Agar supplemented with 0.01% sodium pyruvate. Cells were incubated at 30uC under aerobic conditions for 48 hours. The strains which were able to use the citrate as carbon source increased the pH of medium, resulting in a color change of the medium from green to blue.

Statistical Analyses
Analysis of variance (ANOVA) was used to identify if any of the mutations significantly changed the growth rate of the Arcobacter strains. Statistical analyses were performed using the SPSS 19.0 statistical package program (SPSS Inc., Chicago, IL).

Mutagenesis of the ECF s Factor and Anti-s Factor Encoding Genes
For each ECF s/anti-s pair, mutational inactivation of the anti-s factor results in constitutive induction of the s response, while mutational inactivation of the s factor results in the downregulation of ECF-dependent transcription [19,27]. This allowed us to study the function of the ECFs in A. butzleri strain RM4018 without knowing the specific signal(s) that activate them. In order to address the role of the seven ECF s factors in A. butzleri RM4018, we inactivated either both the s and the anti-s factor, or the anti-s factor alone. Since no suitable Arcobacter antibiotic resistance cassettes were available and the previously-used chloramphenicol cassette of Campylobacter coli could not be used since A. butzleri RM4018 is resistant to chloramphenicol, we replaced a large part of the ECF coding regions with a kanamycin resistance gene (aph(39)-III) of C. coli [21]. After modifying the Arcobacter mutagenesis protocol described by Ho et al. [28], we were able to replace the ECF s/anti-sfactor genes AB0986/0987 (s 1 , As 1 ), AB1044/1045 (s 2 , As 2 ), AB1460/1461 (s 4 , As 4 ), AB1576/ 1577 (As 5 , s 5 ) and AB2300/2301 (As 7 , s 7 ) with the Km r cassette. We also obtained single mutants in the anti-s factor genes As 1 , As 2 , As 4 , As 5 and As 7 . All mutants contained the Km r cassette in the same orientation as the ECF genes. As no suitable Arcobacter shuttle plasmids exist and no other antibiotic cassettes could be used, we were unable to perform complementation studies.

Phenotype Characterization
Comparison of the various mutants with the parent strain RM4018 revealed neither differences in bacterial shape or colony formation on sheep blood plates nor in growth rate in BHI medium as statistically estimated by Anova F = 1,777 (P.0.05) ( Figure 1A and B). A common role of ECF s factors is to protect bacteria against external stress [16]. Exposure of the A. butzleri mutant strains to extreme temperatures (e.g., 4uC or 60uC), pH 5

Identification of ECFs Factor-regulated Genes
To identify the genes regulated by the RM4018 ECF s factors, RNA was isolated from wild-type and mutant strains grown to mid logarithmic phase and subjected to microarray-based transcriptome analysis. To obtain maximal ECF s factor-dependent transcript differences, transcripts of each s/As mutant and its cognate As mutant were compared. Genes showing more than a fourfold change in transcript levels were considered as ECF sdependent. Based on this criterion, the ECF s factors s 1 , s 2 , s 4 , s 5 and s 7 were found to regulate 3, 65, 14, 42 and 72 genes, respectively (Tables S1 to S5). None of the ECF s factors appeared to regulate its own transcription. Interestingly, more than thirty genes were regulated by two or more ECF s factors (Table 3). In agreement with the phenotypic characterization, the identified ECF s-dependent genes indicate that the ECF s factors of A. butzleri regulate other genes than ECF s factors in other bacterial species.

Genes Regulated by AB0986 (ECF s 1 )
The transcriptome analysis revealed that s 1 activates the transcription of AB0988, encoding a putative TonB-dependent receptor protein, and down-regulates the transcription of AB1593 and AB0033, that encode a putative sodium: alanine symporter and a L-lactate permease, respectively (Table S1).

Genes Regulated by AB1044 (ECF s 2 )
Based on the microarray data, ECF s 2 activates the transcription of 35 genes and caused a down-regulation of the expression of 30 genes (Table S2). All annotated gene products represent putative proteins. The sodium: solute symporter AB0504 is the most strongly up-regulated gene product. Apart from this, ECF s 2 is predicted to control the transcription of the genetic loci: AB0345-AB0346, encoding a Nrf-type nitrite reductase; AB0353 to AB0359, coding for a Nap-type nitrate reductase and a C 4dicarboxylate transport system; AB0376-0377, encoding an aldehyde dehydrogenase and an uncharacterized protein; AB0494-AB0495, encoding an acetate kinase and a phosphate

Genes Regulated by AB1460 (ECF s 4 )
According to the transcriptomics data, s 4 activates 11 genes and caused a down-regulation of the expression of 3 genes (Table  S3). The main up-regulated gene product is the putative TonBdependent receptor protein (AB1462) encoded by the same locus encoding the s 4 /anti-s 4 factors. Among the other genes upregulated by ECF s 4 are: napHGA (AB0354 to AB0356), aceEF (AB1480-1481), frdA (AB0297), nrfA (AB0345) and fdhA1 (AB1507) encoding, respectively, a part of a Nap-type nitrate reductase, two pyruvate dehydrogenase components, a fumarate reductase subunit, a cytochrome c 552 nitrate reductase and a formate dehydrogenase subunit. A methyl-accepting chemotaxis protein (AB0602) is the most strongly down-regulated gene product.

Genes Regulated by AB1577 (ECF s 5 )
Identification of ECF s 5 -dependent genes by transcriptomics analysis revealed 16 up-regulated and 26 down-regulated genes ( Table S4). The closely-linked TonB-dependent receptor gene AB1573, was the most strongly up-regulated. The putative TctABC transport proteins AB0102-AB0104, the carbonic anhydrase AB0107, the putative ammonia monooxygenase AB0108, and the ubiquinolcytochrome c oxidoreductase encoded by the petAB (AB2054-AB2055) genes are among the gene products up-regulated by s 5 . The genes fur2 and cspA, and the genetic locus AB1062-1066, which includes NADPH quinonereductase, NADPH:flavodoxinoxidoreductase, and NADPH nitroreductase, were found to be repressed.

Genes Regulated by AB2301 (ECF s 7 )
The transcriptome analysis revealed that s 7 activates 37 genes and represses another 35 (Table S5). Almost all annotated genes repressed by s 7 were activated by s 2 . The genes encoding the putative TctABC transport proteins (AB0103-0105) are the most up-regulated genes by s 7 , while the sox operon and AB0494-AB0495, encoding an acetate kinase and a phosphate acetyltrans- The A. butzleri ECF s Factors Regulate Parts of the Electron Transport Chain Unlike the tonB genes which also appear to be regulated by ECF sigma factors in other bacterial species, we found that a large number of genes regulated by A. butzleri ECF s factors are involved in the electron transport chain. To verify that the ECF s factors indeed influence the electron transport chain, we first performed real-time RT-PCR on the ECF-dependent tonB, nap, sox and tct genes. We could confirm that the tonB genes AB0988, AB1462 and AB1573 are indeed up-regulated by ECF s 1 , s 4 and s 5 respectively (Figure 2a, c and d). ECF s 2 activates the nap and sox genes (Figure 2b) of which the latter are repressed by s 7 (Figure 2e). The real-time RT-PCR results also confirmed that the tct genes are down-regulated by ECF s 2 and up-regulated by s 5 and s 7 (Figure 2b, d and e).
To prove that the differences in nap, sox and tct transcripts result in phenotypic differences, we determined the nitrate reductase activity, the ability to oxidize sulfate, and the utilization of citrate in ECF mutants compared to the wild-type. As expected, none of the strains except A. butzleri As 2 ::Km produced nitrite from the supplied nitrate in the presence of oxygen (Figure 3a). The constitutive expression of s 2 in the case of As 2 ::Km mutant led to nitrite production regardless of the presence of oxygen, indicating that ECF s 2 up-regulates the Nap-type nitrate reductase. To investigate whether the altered transcription levels of the sox genes resulted in a change in Sox enzyme activity, all the strains were grown for 6 hours in AM1 medium supplemented with sulfite and the pH indicator phenol red. The mutants DAs2 and Ds7/As7 further induced the change in medium pH observed during growth of the parental strain, indicating that these mutants have an increased ability to oxidize reduced sulfur compounds ( Figure 3b). These results confirmed that ECF s 2 and s 7 regulate the Sox sulfur oxidation proteins in an opposite way. Finally, we investigated whether these strains are able to utilize citrate, which causes an increase in the pH and a color change of the medium when these bacteria are grown on Simmons agar. In agreement with microarray and real-time RT PCR results, all strains except the mutants DAs2 and Ds7/As7 caused a color change of the media, indicating that the DAs2 and Ds7/As7mutants could not metabolize citrate (Figure 3c).

Discussion
Although Arcobacter has been classified as an emerging pathogen (ICMSF, 2002), knowledge on this organism is still limited. The genome sequence of A. butzleri strain RM4018 [15] revealed that, unlike other characterized members of the epsilon proteobacteria, this strain possesses seven putative extracytoplasmic function (ECF) s factors. These s factors are often involved in the regulation of virulence or stress-related genes in other bacteria. In the present study, we developed genetic tools to manipulate A. butzleri and used these tools to investigate the function of the putative RM4018 ECF s factors. A. butzleri is resistant to many different antibiotics [15,29,30]; therefore, it was difficult to find a suitable antibiotic cassette to inactivate the ECF s factor-encoding genes. Only a kanamycin resistance gene from C. coli [31] appeared to be functional in strain RM4018. By electroporation we were able to generate 10 different mutants and were able to inactivate 5 of the 7 ECF s factors and to identify the corresponding regulons. Despite repeated attempts, we were unable to mutate the ECF s factors AB1430 (s 3 ) and AB2165 (s 6 ), which may indicate that these factors are essential for bacterial growth under the conditions employed.
ECF s factors are involved in a wide range of environmental responses in many bacterial species [18]. They are mainly divided into stress response ECF s factors and iron-starvation ECF s factors [32]. Our results indicate that the ECF s factors of the strain RM4018 are neither involved in iron-starvation nor in the stress response. Furthermore, we found no evidence of positive regulation of their own transcription, as often found for other ECF s factors [27]. Except for the tonB genes, the RM4018 ECF s factors regulate completely different sets of genes. This seemingly atypical gene regulation in Arcobacter has also been noted for the regulation of the flagellar genes as Arcobacter lacks the s factors FliA and RpoN, which in many bacterial species regulate flagellar biosynthesis [15]. So, in contrast to other eubacteria, many conserved genes in A. butzleri appear to be regulated by a s factor belonging to a different s factor class; therefore, knowledge on transcription regulation of conserved genes described in other bacterial species cannot be simply extrapolated to A. butzleri.
In A. butzleri RM4018, 5 out the 7 ECF s/anti s-factor pairs are flanked by genes encoding putative TonB-receptor proteins. In other species, the ECF s/anti s factors form, together with an outer membrane TonB-dependent receptor, a cell-surface signaling (CSS) system [33]. Transcriptome analyses showed that s 1 , s 4 and s 5 regulate the expression of the TonB-dependent receptors encoded by the same loci, suggesting the putative presence of three potential trans-envelope signal transduction systems in A. butzleri RM4018. Most of the described CSS systems are involved in iron signaling and transport [34]. However, other functions have been described including for the Ralstonia solanacearum Prha-PrhIR system, which senses the presence of a plant cell-wall structure and initiates a regulatory cascade that induces the hypersensitive response and the transcription of pathogenicity genes [35]. The role of the three putative CSS systems in A. butzleri RM4018 remains to be elucidated, as the growth of mutants in these systems are not affected by iron limitation.
Transcriptome analysis, as well as functional assays, showed that A. butzleri ECF s factors also regulate a number of genes involved in the binding and transport of specific compounds, energy metabolism and sulfur oxidation. We showed that Arcobacter does not reduce nitrate under aerobic conditions ( Figure 3A), although it does under anaerobic conditions (data not shown). The ECF s 2 is involved in this environmental adaptation as the As 2 ::Km mutant produces nitrite under aerobic conditions. This may indicate that oxygen stress has an effect on ECF s 2 activity. Several genes must be indirectly regulated by the ECF s factors as they were up-regulated in the ECF s mutants. Interestingly and also seen in Bacillus subtilis [36], a number of these genes are regulated by more than one ECF s factor, which indicates that they may be needed under different growth conditions. For example, s 2 and s 7 regulate the dctPQM and sox genes in an opposite way. The DctPQM proteins form a tripartite ATPindependent periplasmic transporter, which catalyzes the uptake of C 4 -dicarboxylates like malate, fumarate and succinate in many aerobic bacteria [37]. The Sox proteins which are not found in other members of the Campylobacteraceae are involved in the oxidation of reduced sulfur compounds in sulfur, photo-and chemo-lithotrophic bacteria [38]. Similarly, we showed that s 5 and s 7 activate, while s 2 caused a down-regulation of the expression of the tctABC genes encoding a tricarboxylic transport system. Despite this, only s 7 and s 2 appeared to have a significant effect on citrate utilization. In many bacterial species a twocomponent signal transduction system activated by citrate is responsible for the activation of the tctABC genes [39]. In A. butzleri, the putative TctABC system may also depend on the twocomponent system (AB0105-0106) located directly downstream of the putative tctABC genes. The expression of AB0105-0106 was not dramatically affected by s 5 and this may explain why citrate utilization was distinct between Ds5/As5 and Ds7/As7. All together, the genes regulated by the different A. butzleri ECF s factors indicate that the ECF s factors form a complex network of regulons that have a major role in regulating bacterial metabolism.
In conclusion, we have shown in this initial study that the ECF s factors of A. butzleri control the transcription of a complex network of regulons that contain genes that are not commonly regulated by ECF s factors family of proteins. These genes are mainly involved in the energy metabolism. In contrast to other eubacteria, many conserved genes in A. butzleri appear to be regulated by a s factor belonging to a different s factor class.

Supporting Information
Table S1 Genes identified by micro-array analyses which are more than fourfold up or down regulated by A. butzleri ECF sigma 1.