A Proteomic Analysis Reveals Differential Regulation of the S -Dependent yciGFE(katN) Locus by YncC and H-NS in Salmonella and Escherichia coli K-12

The stationary phase sigma factor (cid:1) S (RpoS) controls a regulon required for general stress resistance of the closely related enterobacteria Salmonella and Escherichia coli . The (cid:1) S -dependent yncC gene encodes a putative DNA binding regulatory protein. Application of the surface-enhanced laser desorption/ionization-time of flight (SELDI-TOF) ProteinChip technology for proteome profiling of wild-type and mutant strains of Salmonella enterica serovar Typhimurium revealed potential protein targets for YncC regulation, which were identified by mass spectrometry, and subsequently validated. These proteins are encoded by the (cid:1) S -dependent operon yciGFEkatN and regulation of their expression by YncC operates at the transcriptional level, as demonstrated by gene fusion analyses and by in vitro transcription and DNase I footprinting experiments with purified YncC. The yciGFE genes are present (without katN ) in E. coli K-12 but are poorly expressed, compared with the situation in Salmonella . We report that the yciGFE(katN ) locus is silenced by the histone-like protein H-NS in both species, but that (cid:1) S efficiently relieves silencing in Salmonella but not in E. coli K-12. In Salmonella , YncC acts in concert with (cid:1) S to activate transcription at the yciG promoter (p yciG ). When overproduced, YncC also activated (cid:1) S -dependent transcription at p yciG in E. coli K-12, but solely by countering the negative effect of H-NS. xenogeneic silencer H-NS.

In eubacteria, transcription depends on a multisubunit RNA polymerase (RNAP) consisting of a catalytically active core enzyme (E) with a subunit structure ␣ 2 ␤␤', that associates with any one of several factors to form different holoenzyme (E) species. The subunit is required for specific promoter binding, and different factors direct RNAP to different classes of promoters, thereby modulating gene expression patterns (1). The RNA polymerase holoenzyme containing the 70 subunit is responsible for the transcription of most genes during exponential growth (1). When cells enter stationary phase or are under specific stress conditions during exponential growth, S , encoded by the rpoS gene, becomes more abundant, associates with the core enzyme, and directs the transcription of genes essential for the general stress response (1)(2)(3). In the closely related Enterobacteria Salmonella and Escherichia coli, S is required for stationary phase survival, stress resistance, and biofilm formation. It is also involved in the virulence of Salmonella enterica serovar Typhimurium (S. Typhimurium) (4).
Transcriptome analyses in S. Typhimurium and E. coli K-12 have shown that rpoS controls more than 300 genes, 40% of which are of unknown function (3,5,6). A large fraction of S -controlled genes encode putative regulators and signal transducing factors, suggesting that S controls a complex network with regulatory cascades and signal input at levels downstream of S itself. We previously used a bank of S. Typhimurium mutants to identify S -regulated genes (7). One of these genes, the yncC gene (7), encoded a putative DNA binding protein of the GntR/FadR family of bacterial regulators (8 -10). To further investigate the function of the yncC gene, we decided to characterize the proteome of the Salmonella yncC mutant by the surface-enhanced laser desorption/ionization-time of flight (SELDI-TOF 1 ) ProteinChip technology.
The SELDI-TOF method is based on the selective protein retention on a solid-phase chromatographic chip surface and successive analysis by simple laser desorption/ionization mass spectrometry (11). Because of its high-throughput nature and experimental simplicity, this technology has been widely used for protein profiling of tissues and biomarker discovery (11) and unpublished work from our laboratory re-structed using primers YncC-E1 and YncC-E2 (Table II) to amplify the promoter-less yncC gene from ATCC14028 total DNA by PCR. EcoRI restriction sites were incorporated at its 5Ј and 3Ј ends. Following digestion with EcoRI, the fragment was inserted into the EcoRI site of pACK to give pACKyncC (the yncC and cat genes are in the same orientation and the yncC gene is likely transcribed from the cat promoter). The nucleotide sequence of the yncC insert in pACKyncC was verified by DNA sequencing. pyncC HIS , which expresses an N-terminal His 6 fusion to the yncC gene product under the control of the pQE30 IPTG-inducible promoter, was constructed as follows. Primers YncC-H3 and YncC-H5 (Table II) were used to amplify the yncC gene from ATCC14028 total DNA by PCR. BamHI and HindIII restriction sites were incorporated at its 5Ј and 3Ј ends, respectively. Following digestion with BamHI and HindIII, the PCR-amplified fragment was ligated into the BamHI and HindIII sites of pQE30. The nucleotide sequence of the yncC insert in pyncC HIS was verified. Construction of plasmid for in vitro transcription was a follows. The E. coli yciG fragment (extending from -227 to ϩ66 relative to the transcription start) was synthesized from primers M91 and M92 and the S. Typhimurium yciG fragment (-184 to ϩ48) from primers M47 and M48bis. The fragments were cleaved by EcoRI and BamHI and inserted into the pJCD01 vector cleaved by EcoRI and BamHI, leading to plasmids pJCDyciG and pJCDkatN.
Construction of the ⌬yncC ⌬hns and ⌬yci Mutants of S. Typhimurium-Chromosomal deletions in the yncC, hns and yciGFEkatN loci of Salmonella ATCC14028 were generated using PCR-generated linear DNA fragments and the Red recombination method as described by Datsenko and Wanner (30). Briefly, 63-66 nt primers with 43-46 nt homology with the gene of interest on the 5Ј end of the primer and 20 nt homology with the FLP recognition target flanked antibiotic resistance cassette of plasmid pKD3 at the 3Јend (sequences given in 30) were used. The primer pairs, YncC-P1 and YncC-P2, Hns-P1 and Hns-P2, YciG-P1 and KatN-P2 (Table II) were used for disruption of the yncC, hns, and yciGFEkatN operon, respectively. ATCC14028 containing the plasmid pKD46, which carries the recombination genes gam, bet, and exo under control of the araBAD promoter (30) was grown overnight at 30°C, diluted in LB carbenicillin containing L-arabinose 1 mM and grown to an OD600 of 0.5. Electrocompetent cells were prepared, transformed with the PCR-generated linear fragments and plated on LB containing chloramphenicol (15 g/ml) and incubated at 37°C. The resulting colonies were characterized using a combination of PCR reactions using locus-specific primers and common test primers (30). Finally, isogenic strains were constructed by P22 mediated transduction of the mutations into the appropriate strains. When required, the chloramphenicol resistance cassette was eliminated using a temperature-sensitive helper plasmid pCP20, which encodes the FLP recombinase (30). Because the hns mutants were very sick and might accumulate compensatory mutations, they were constructed freshly for each experiment.
Construction of a Chromosomal yciE-lacZ Transcriptional Fusion in E. coli K-12-A single copy yciE-lacZY transcriptional fusion was constructed from mutant MC4100yciE using conditional plasmids containing promoter-less lacZY genes and the FLP recognition target site as described (31). PCR assays were then used to ensure integration of the plasmids in the correct location and to determine the presence of multiple plasmid integrants (using common test primers, such as those described in (31). Locus-specific flanking primers were also used to amplify junction fragments that were subsequently analyzed by DNA sequencing. Isogenic strains were constructed by P1 mediated transduction of the mutations into the appropriate strains.
Protein Profiling by SELDI-TOF-MS-Bacteria were grown in LB for 18 h at 37°C. Cells were harvested and cell pellets obtained from 100 ml of culture were resuspended in 20 ml phosphate buffer 50 mM pH 7 and disrupted in a Cell Disrupter (Constant Systems, Daventry, UK). The cell debris were removed by centrifugation. Protein concentrations in the supernatants were determined using the DC Protein Assay kit (Bio-Rad) and adjusted to a concentration of 1 g/l in phosphate buffer 50 mM pH 7. These cytosol extracts were stored at Ϫ70°C. The SELDI analysis was performed using 10 g crude cytosol extracts of Salmonella and E. coli strains. For each strain, three independent cultures were used to prepare cytosol extracts and each extract was spotted in duplicate on the ProteinChip array. A strong anion exchange ProteinChip array (Q10, Bio-Rad), for which the complementary resin Q Ceramic HYPERD F (BioSepra-Pall, Cergy St Christophe, France) can be used for protein purification, was employed to capture negatively charged proteins. Before loading the cell extracts, the Q10 array was equilibrated once for 10 min with 150 l of buffer T50 (50 mM Tris-HCl pH 9), and once for 5 min with 100 l of buffer T50 containing 0.1% Triton, using a bioprocessor. Bacterial extracts (10 g) were diluted in 100 l of T50 containing 0.1% Triton and incubated on the chips for 1 h at room temperature under shaking. Then, the chips were washed: first for 10 min with 100 l of T50 containing 0.1% Triton, second, twice for 5 min with 150 l of T50 and finally twice with double-distilled H 2 O. The samples were then air-dried. Then, 0.7 l of sinapinic acid (Bio-Rad) saturated in freshly prepared 50% acetonitrile-0.5% trifluoroacetic acid was applied twice on each spot and the spots were air dried. Molecules retained on the surfaces were visualized by reading the spots of each array in a SELDI-TOF-MS reader (PSC4000; Ciphergen Biosystems, Copenhagen, Denmark). Spectra were generated by seven shots on 36 pixels at laser energy varying between 2600 and 4500 nJ and an accelerating voltage of 25 kV in positive mode with automatic data collection software 3.0 program. External mass calibration was performed on one spot of each array by using ubiquitin (8564. Raw spectra were processed and analyzed with the Ciphergen Express data manager software version 3.0 (CE; Ciphergen Biosystems). Spectra were externally calibrated with cytochrome C (bovine) (12,230.9 ϩ 1H), ␤-lactoglobulin A (bovine) (18,363.3 ϩ 1H) and horseradish peroxidase (43,240.0 ϩ 1H). The baseline was established using a smoothing of three points and a width of five times expecting peak width and spectral intensities were normalized by total ion current. Consistent peak sets of similar mass across the spectra were generated with CiphergenExpress Cluster Wizard. This Wizard operates in three passes across the spectra. The first pass performs peak detection at high signal-to-noise (s/n) ratio to pick out well-defined peaks as starting points for forming clusters. A second pass selects lower s/n ratio peaks, within a mass window defined around the first pass peaks. The algorithm completes the clusters in a third pass by creating artificial peaks where none were detected in the first two passes, at the exact center of clusters. In this analysis, unless otherwise specified, the first pass was performed with an s/n threshold of five, and the second pass with an s/n threshold of two, in a 0.5% width mass window. Clusters were assembled between 5000 and 40,000 Da. The cluster lists contained normalized peak intensity values for each sample within a group and p values were calculated between the medians of the peak intensities to detect significant differences in abundance for particular proteins. Proteome Fractionation on Q Ceramic HYPERD F-A volume of 80 l of Q Ceramic HYPERD F beads (BioSepra-Pall Corporation) were equilibrated three times in buffer T50, centrifuged, and resuspended in 10 ml of T50 containing 0.1% Triton. Bacterial cytosol extracts (8 mg in 4 ml) were incubated with the beads for 2 h at 4°C on a rotative shaker and centrifuged. Beads were washed twice in T50 containing 0.1% Triton, and then twice in T50 and finally once in Tris-HCl 5 mM pH 9.0. Proteins captured on the beads were eluted successively in 100 l HEPES 50 mM pH 8.0, 100 l phosphate buffer 50 mM pH 7.0, 100 l MES 50 mM pH 6.0, 100 l sodium acetate 50 mM pH 5.0, 100 l sodium acetate 50 mM pH 4.0, 100 l sodium acetate 50 mM pH 3.4, and finally 100 l sodium acetate 50 mM NaCl 1 M pH 3.4. Proteins in the eluted fractions (25 l) were separated by SDS-PAGE on a 12.5% acrylamide gel. The gel was stained with Coomassie blue.
Identification of the YciF and YciE Proteins by Mass Spectrometry-Mass spectrometry analyses have been conducted at the PF3 Proteomic platform (Abdelkader Namane, Institut Pasteur, France).
Sample Preparation-One-dimensional gel bands were excised from gels and collected in 96-well plate. Destaining, reduction, alkylation, trypsin digestion of the proteins followed by peptide extraction were carried out with the Progest Investigator (Genomic Solutions, Ann Arbor, MI). Following the desalting step (C18-ZipTip, Millipore) peptides were eluted directly using the ProMS Investigator, (Genomic Solutions) onto a 96-well stainless steel matrix-assisted laser desorption ionization target plate (Applied Biosystems/MDS SCIEX, Framingham, MA) with 0.5 l of CHCA matrix (5 mg/ml in 70% acetonitrile/ 30% H 2 O/0.1% trifluoroacetic acid).
MS and MS/MS Analysis-Raw data for protein identification were obtained on the 4800 Proteomics Analyzer (Applied Biosystems) and analyzed by GPS Explorer 2.0 software (Applied Biosystems/MDS SCIEX). For positive-ion reflector mode spectra 3000 laser shots were averaged. For MS calibration, autolysis peaks of trypsin ([MϩH] ϩ ϭ 842.5100 and 2211.1046) were used as internal calibrates. Monoisotopic peak masses were automatically determined within the mass range 800 -4000 Da with a signal to noise ratio minimum set to 20. Up to 12 of the most intense ion signals were selected as precursors for MS/MS acquisition excluding common trypsin autolysis peaks and matrix ion signals. In MS/MS positive ion mode, 4000 spectra were averaged, collision energy was 2 kV, collision gas was air and default calibration was set using the Glu 1 -Fibrino-peptide B ([MϩH] ϩ ϭ 1570.6696) spotted onto 14 positions of the matrix-assisted laser desorption ionization target. Combined PMF and MS/MS queries were performed using the MASCOT search engine 2.1.04 (Matrix Science, London, UK) embedded into GPS-Explorer Software 2.0 (Applied Biosystems/MDS SCIEX) on the National Center for Biotechnology Information database 20070518 (4927571 sequences, 17002359384 residues) with the following parameter settings: 50 ppm peptide mass accuracy, specific trypsin cleavage (K/R), one missed cleavage allowed, carbamidomethylation set as fixed modification, oxidation of methionines was allowed as variable modification, MS/MS fragment tolerance was set to 0.3 Da. Protein hits with MASCOT Protein score Ն 79 and a GPS Explorer Protein confidence index Ն 95% were used for further manual validation.
Enzymatic Assays-␤-galactosidase activity was measured as described by Miller (32) and is expressed in Miller units.
Electrophoresis and Immunoblot Analysis of Proteins-Whole-cell extracts were prepared and SDS-polyacrylamide gel electrophoresis was carried out as described by Silhavy et al. (28). The amount of protein in whole-cell lysates was determined using the DC Protein Assay kit (Bio-Rad). Equal amounts of protein were loaded in each slot. The molecular sizes of the proteins were estimated using molecular size standards (Fermentas, France). Rabbit antibodies against the S protein of S. enterica serovar Typhimurium were from Coynault et al. (33). Mouse monoclonal IgG1 penta-His antibody (Qiagen) was used to detect His-tagged proteins. Proteins were transferred to Amersham Biosciences Hybond P membranes (GE Healthcare) and incubated with the antibodies as previously described (33). Bound antibodies were detected using secondary anti-rabbit (for S detection) and anti-mouse (for His-tag detection) antibodies linked to peroxidase and the Amersham Biosciences ECL plus Western blotting detection system kit (GE Healthcare).
Overproduction and Purification of His 6 -YncC-A 500-ml culture of JM109 carrying pyncC HIS was grown in LB containing 100 g/ml carbenicillin at 28°C to an optical density of 0.6 and then supplemented with 1 mM isopropyl-␤-D-thiogalactoside. Following 4 h, cells were harvested, washed, resuspended in 10 ml of phosphate buffer (50 mM Na 2 HPO 4 pH 8) and lysed by high pressure cell disruption. The extract was supplemented with 300 mM NaCl and centrifuged at 15,000 ϫ g for 30 min at 4°C. The supernatant was added to 2.5 ml of Ni-NTA agarose (Qiagen) and gently mixed for 180 min. The slurry was packed into an Econo-Pac column (Bio-Rad), washed with 25 ml buffer A (50 mM Na 2 HPO 4 pH 8, 300 mM NaCl) containing 20 mM imidazole. His 6 -YncC was eluted with buffer A containing 250 mM imidazole, dialyzed against buffer B (20 mM Tris-HCl pH 8, 20 mM NaCl, 1 mM dithiotreitol), loaded onto a Hitrap Q anion exchangecolumn (1 ml; GE Healthcare). Following washing with 10 ml buffer B His 6 -YncC was eluted from the column with a linear gradient from 0.02 to 1 M NaCl between 0.15 M and 0.2 M NaCl. The pooled fractions were dialyzed against buffer C (10 mM Tris-HCl pH 7.9, 100 mM NaCl, 50% glycerol, 1 mM dithiotreitol). The purification yield was 3 mg of His 6 -YncC, as determined by Bradford assay.
Labeled DNA Fragments-Primers were 32 P-labeled at their 5Јends using phage T4 polynucleotide kinase and [␥-32 P]-ATP (800 Ci/mmol). The E. coli yciG promoter fragment was generated by PCR using MG1655 chromosomal DNA and primers M91 and M92. For nontemplate strand labeling, [ 32 P]-labeled primer M91 was used with unlabeled primer M92. The S. Typhimurium yciG fragments were generated by PCR using pJCDkatN as template (21) and [ 32 P]-labeled primer E7 and M48bis for nontemplate strand labeling and [ 32 P]labeled M48bis and E7 for template strand labeling. The fragments were then purified on a glass fiber column (High Pure PCR Product Purification Kit from Roche Diagnostics, Neuilly, France).
DNase I Footprinting-The [ 32 P]-labeled fragments were incubated for 20 min in a final volume of 15 l with increasing concentrations of His 6 -YncC in buffer C (40 mM Hepes pH 8.0, 10 mM MgCl 2 , 100 mM potassium glutamate, 5 mM dithiotreitol and 500 g/ml bovine serum albumin). 1.5 l of a 1 g/ml DNase I solution (Worthington) in buffer D (10 mM Tris-HCl pH 8.0, 10 mM MgCl 2 , 10 mM calcium chloride, 125 mM potassium chloride) were then added and incubated for 10 s when the labeled fragment was alone, or for 10 to 40 s depending on YncC concentrations (expressed in dimers). The reaction was stopped with phenol, as described (34), and loaded on a 7.5% sequencing polyacrylamide gel. Protected bands were identified by comparing the migration of the same fragment treated for the AϩG sequencing reaction (35). The E. coli H-NS protein was a gift from Sylvie Rimsky and its concentration is expressed in monomers. In the competitive binding assays with H-NS and His 6 -YncC, the proteins were first mixed and incubated with promoter fragments at 30°C for 20 min in buffer C ϫ 0.5 before DNase I attack.
In Vitro Transcription-The DNA fragments were prepared from pJCD01 derivatives containing the promoter fragments cloned upstream of the rrnB1 terminator using primers E7 and J7 (25). A shorter S. Typhimurium yciG fragment (-61 to ϩ48) that does not harbor the YncC binding site was generated using pJCDkatN and primers M57bis and J7 (21). The katE promoter fragment was described in Robbe-Saule et al. (21). E S RNA polymerase was reconstituted from core RNA polymerase either wild-type (Epicenter Biotechnologies, Madison, WI) or harboring ␣ subunits truncated in the C-terminal domains (36), and His-S prepared from S. Typhimurium (37). Different amounts of YncC were incubated for 20 min at 37°C with promoter fragments (at a final concentration of 10 nM) in buffer C before E S RNA polymerase addition (E: 15 nM, S : 60 nM) and incubation was prolonged for 20 min before addition of the heparin/XTP mixture (21). H-NS was first incubated with DNA templates (20 nM) for 20 min at 30°C in 5 l buffer A. A 5-l aliquot of a mixture containing RNA polymerase with or without YncC (E: 60 nM, S : 240 nM, YncC: 500 nM) was then added. Incubation was prolonged for 10 min before adding 5 l of a heparin/ XTP mixture (450 ⌴ ATP, CTP, and GTP and 45 ⌴ [␣-32 P]-UTP). The reaction was stopped following 10 min by adding 15 l of formamide containing 10 mM EDTA, 1.6% SDS, and 0.02% xylene cyanol blue. An aliquot was loaded on a 7% polyacrylamide sequencing gel.

Putative YncC targets revealed by Proteinchip SELDI-TOF
analyses-In a search for genes regulated by S in S. Typhimurium, we isolated a transposon insertion in the yncC gene (7, ATCC-F12) ( Table I). SELDI-TOF ProteinChip technology was used to capture and analyze proteins from clear lysates of the wild-type strain ATCC14028 and its mutant derivative ATCC-F12 grown to stationary phase in LB at 37°C. When spectra from ATCC14028 and ATCC-F12 were compared, three peaks were reproducibly detected with a higher intensity in ATCC14028 than in ATCC-F12 (peaks 1-3, data not shown). Mutant ATCCyncC, which contains a deletion of the yncC gene, was subsequently constructed (Table I) and compared with ATCC14028. The three peaks, (1, 2, and 3 of molecular sizes 18,641, 18,964, and 31,935 Da, respectively, Fig. 1A) were detected at higher intensity levels in ATCC14028 than in ATCCyncC (p value 0.004, 6 samples per strain as described under "Experimental Procedures"). Interestingly, these peaks were not detected at significant levels in the spectra of the ⌬rpoS mutant of ATCC14028 (ATCCrpoS, Fig.  1A). Therefore, these peaks might correspond to proteins encoded by genes regulated by S and YncC.
Identification of YncC Targets-Because peak 3 was detected in the wild-type extract with a low intensity, compared with the other two peaks (Fig. 1A), we first focused on the identification of peaks 1 and 2. To identify these proteins, a partial purification scheme was devised (Experimental Procedures) involving anion exchange chromatography of clear lysates from the wild-type strain and the rpoS and yncC mu-tants followed by separation on the basis of pI. In the pH 5 fraction, two bands corresponding to proteins of 20 and 17 kDa were detected at a higher intensity in the extract of the wild-type strain than in the yncC and rpoS mutant extracts (Fig. 1B, lanes 3, 5, and 6). The two bands (shown by stars on Fig. 1B) were cut from the gel for subsequent trypsin digestion and identification by mass spectrometry (Table III) (Experimental Procedures). The proteins were identified as the YciF and YciE proteins of Salmonella (Fig. 1B).
The YciF and YciE proteins are encoded by an operon, yciGFEkatN (12) (Fig. 1C). Their calculated molecular sizes (Fig. 1C) correspond to those predicted from SELDI-TOF. The elution of these proteins from the anion exchange resin at pH 5 ( Fig. 1B) is consistent with the calculated pI of YciF (5.24) and YciE (5.14). In addition, the 20 and 17 kDa proteins were not detected in ATCC-F1, which has a polar transposon insertion in yciF (Fig. 1B, lane 4).
The last gene in the operon, katN, encodes a protein of 31,848 Da that might correspond to peak 3 detected by SELDI-TOF (Fig. 1A). To check that peaks 1, 2, and 3 are encoded by the yciGFEkatN locus, the proteome of strain ATCC⌬yci, in which the entire yciGFEkatN operon is missing (Table I), was compared with that of the wild-type strain. The spectra of the two strains were similar, except that peaks 1, 2, and 3 were not detected in ATCC⌬yci (Fig. 1A), indicating that these peaks corresponded to YciF, YciE, and KatN. Peak 3, corresponding to KatN, has a molecular size (31,935 Da) that is slightly higher than that expected (31,848 Da). This difference might result from the presence of manganese in KatN (12) or from unknown posttranslational modification(s).
The relative abundance of the three proteins (YciFϾ YciEϾ KatN, Fig. 1A), is consistent with the position of the genes in the yciGFEkatN operon and mRNA levels (12). These proteins were not detected at significant levels in the ⌬rpoS mutant, in agreement with our previous finding that expression of the operon is highly dependent on S (12,21). The YciG protein (6 KDa), encoded by the operon, was not detected, likely because conditions used for SELDI-TOF analyses were optimized for accurate detection of proteins Ͼ 10 kDa and because the calculated pI of YciG (9.99) is too high for it to bind to the anion exchange array.
YncC is Required for Maximal Transcription of the yciGFEk-atN Operon in Salmonella-Results described earlier indicate that YciF, YciE, and KatN production is positively regulated by YncC. To determine whether this regulation operates at the transcriptional level, kinetics of expression of a katN-lacZ gene fusion in the wild-type strain and the ⌬yncC mutant were compared. Expression of the fusion was delayed during early growth stages and was reduced by the ⌬yncC mutation ( Fig.  2A). The yncC gene in pACKyncC complemented the ⌬yncC mutation for katN-lacZ expression, confirming that yncC activates katN transcription (Fig. 2B). In the wild-type strain, expression of katN-lacZ was induced earlier during growth and its expression level increased when yncC was ex-pressed in trans from pACKyncC, suggesting that YncC might be a limiting factor for katN-lacZ expression under these conditions (Fig. 2B). As expected (12), katN-lacZ was expressed at very low levels in the absence of S (Fig. 2F,  lanes 1 and 2). Neither the ⌬yncC mutation nor pACKyncC had any effect on the expression of lacZ fused to katE, a catalase encoding gene also regulated by S and used as a control (data not shown). Altogether, these results suggest that YncC exerts a positive effect on yciGFEkatN operon transcription.
YncC Binds Upstream of the yciGFEkatN Promoter-Plasmid pyncC HIS encodes a recombinant YncC protein contain- Typhimurium wild-type and mutant strains onto Q10 ProteinChip Array. Clear lysates from ATCC14028 (WT) and its mutant derivatives ATCCyncC, ATCCrpoS and ATCCyci were applied to the surface of a Q10 ProteinChip as described in the Experimental Procedures section. The captured proteins were detected using surface enhanced laser desorption/ionization (SELDI) time-of-flight mass spectrometry. The normalized mass (m/z) for each peak (in Da) is demonstrated on the x-axis, whereas intensity (A) is plotted on the y-axis. Extracts were examined several times at different laser energies, and energies of 4500 nJ and 3200 nJ were found to be optimal for detection of peaks 1, 2, and peak 3, respectively. To pick out peak 3, the first pass peak detection was performed with a s/n threshold of two instead of five  ing six histidine residues at its N terminus (His 6 -YncC), under the control of the IPTG-inducible promoter of pQE30. pync-C HIS , but not the pQE30 vector, complemented the ⌬yncC strain for katN-lacZ expression and increased katN-lacZ expression in the wild-type strain (Fig. 2F, lanes 9 and 10 and data not shown), indicating that His 6 -YncC is active. The recombinant protein was over-produced in E. coli, purified and used for DNase I footprinting experiments on both strands of the Salmonella yciG promoter region (yciG STM ) (Fig.  3). His 6 -YncC protected a 24 bp sequence centered on the -100 region (Fig. 3B) relative to the transcription start site (14). The YncC binding site is AT-rich and contains the inverted repeat AATATAT. As expected, a footprint was not detected in the katE upstream promoter region (data not shown).
YncC also Binds to the E. coli K-12 yciGFE Promoter Region-The ortholog of yncC in E. coli K-12 (named mcbR),

FIG. 2. Expression of a katN-lacZ fusion in Salmonella wild-type and mutant strains.
A-E, Kinetics of katN-lacZ expression relative to growth phase. Salmonella strains were grown in LB at 37°C. Exponential-phase cultures (optical density at 600 nm ϭ 0.5) were diluted into LB pre-warmed at 37°C to prolong the exponential phase. Aliquots were removed at various time intervals and ␤-galactosidase activity measured (lines) according to the method of Miller (32). The growth phase was determined by measuring the culture turbidity at an optical density of 600 nm (dashed lines). The measurements were repeated at least twice, and a representative experiment is shown. regulates colanic acid production by repressing expression of the ybiM gene (13). Deletion of mcbR in E. coli MG1655 elicited mucoidy and decreased biofilm formation because of overproduction of colanic acid (13). The yncC mutants of Salmonella, ATCC-F12 and ATCCyncC, were not mucoid (data not shown), consistent with the absence of ybiM from S. Typhimurium genome (http://www.ncbi.nlm.nih.gov/).
In both Salmonella and E. coli K-12, yncC/mcbR is located between the yncB and yncD genes that encode a putative oxidoreductase and a putative iron outer membrane transporter, respectively. However, the intergenic regions between yncB and either yncC in Salmonella or mcbR in E. coli K-12 MG1655, which likely contain the yncC/mcbR promoter, differ in length and sequence, suggesting that yncC and mcbR might be differentially regulated. In addition, the amino acid sequences of YncC and McbR diverge in the C-terminal domain (46% identity over amino acids 78 to 221), compared with the N-terminal domain (81% identity over amino acids 1 to 77), which contains the predicted DNA binding HTH domain (8,9). The C terminus of regulators of the GntR/FadR family contains an effector-binding and/or oligomerisation domain that influences the DNA-binding properties of the regulator (8 -10). Altogether, these findings suggested that YncC and McbR might have evolved to respond to different signals and/or perform different functions in E. coli K-12 and Salmonella.
The yciGFE locus is conserved in E. coli K-12, but katN is not (Fig. 1C). The -35 and -10 elements of the yciGFEkatN promoter in Salmonella (pyciG STM , 12) are conserved in E. coli K-12 (Fig. 1D). However, the DNA region upstream of the -35 element, including the YncC binding site identified upstream of pyciG STM , diverge in the two species. Nevertheless, the His 6 -YncC protein could bind to the yciG ECO promoter region (pyciG ECO ) (Fig. 3). The footprint containing two repeats of the AATATAT motif extended over 42 bp with two hypersensitive bands located on the nontemplate strand at the center of the protected region.
YncC belongs to the GntR subfamily of FadR (pfam007729), which binds as a dimer to its operator site via its winged helix domains (10). In the x-ray structure of the FadR-operator complex, the two recognition helices of each monomer project into a central major groove and the two ␤ ribbons of the wings into the flanking minor grooves resulting in specific DNA-protein interactions over 11 bp. Based on these data, YncC likely binds DNA as a dimer and we predict that at least two tandem YncC operator sites are present upstream of the pyciG ECO promoter. The protection footprint of about 20 bp at pyciG STM suggests the presence of a single YncC binding site. Interestingly the YncC binding sites are located closer to the transcription start site at pyciG ECO than at pyciG STM (Fig. 3B).
The E. coli K-12 His 6 -McbR protein and the Salmonella His 6 -YncC protein showed similar protection footprint patterns at the pyciG STM and pyciG ECO promoters (data not shown), a finding consistent with the high sequence conservation in the DNA binding domains of these proteins.
The yciGFE Locus is Poorly Expressed in E. coli K-12-To determine the functional relevance of YncC/McbR binding to the yciGFE promoter region in E. coli K-12, the ability of E. coli MG1655 wild-type strain and mutant derivatives to produce the YciF and YciE proteins was assessed by SELDI-TOF technology. However, the spectra obtained for the wild-type strain and the ⌬yciF and ⌬yciE mutants were similar (data not shown) and peaks corresponding to YciF and YciE could not be detected (Fig. 4A). To assess the expression of these genes in E. coli K-12 further, a chromosomal yciE-lacZ transcriptional fusion was constructed in two E. coli K-12 Lac Ϫ strains, MC4100 and MC1061. The fusion was expressed at very low levels in both strains and in their ⌬mcbR and ⌬rpoS mutant derivatives (Fig. 4B, lane 1 and data not shown). The His 6 -McbR and His 6 -YncC proteins were able to induce yciE-lacZ expression, although the levels of expression remained low (Fig. 4B, lanes 6 and 7 and 12 and 13, respectively). yciE-lacZ expression was induced to higher levels by His 6 -McbR than by His 6 -YncC (Fig. 4B, lanes 7 and 13), likely because the former is more abundant (Fig. 5A). Like the His 6 -YncC protein, the His 6 -McbR protein activated expression of the katN-lacZ fusion in the ⌬yncC and wild-type strains of Salmonella (Fig. 2C). Altogether, these results show that YncC and McbR are both able to induce expression of the yciGFE(katN) locus in Salmonella and E. coli K-12, and that this locus is expressed at drastically lower levels in E. coli K-12 than in Salmonella.
Gene Polymorphism at the trpA-yciGFE-ompW Locus in E. coli Strains-The low GC content of the yciF and yciE genes relative to the resident genome suggested that these genes were horizontally acquired in E. coli K-12 (38) and Salmonella (39,40). Interestingly, in nearly half of the 23 complete sequenced genomes of E. coli strains (http://www.  (7), pQE30 (12), and pyncC HIS (13); MC4100hns yciE-lacZ harboring pCABg (8), pmcbR HIS (9), pQE30 (14), and pyncC HIS (15); and MC4100hnsmcbR yciE-lacZ harboring pCABg (10), and pmcbR HIS (11). ncbi.nlm.nih.gov/), the trpA-yciGFE-ompW locus (Fig. 1C) is a site of DNA rearrangements, including deletions and/or insertions of phage related-and/or virulence associated-genes. In addition, in pathogenic E. coli O157:H7, a second copy of the yciGFE locus, including the katN gene, is located on a cryptic prophage (CP-933X) elsewhere in the genome. The sequence identity between the prophage-borne gene products in E. coli O157:H7 and corresponding gene products in E. coli K-12 and Salmonella is high (more than 80%). However, the noncoding sequence upstream of the prophageborne yciGFEkatN genes is different from those upstream of yciG ECO and yciG STM and does not contain the -10 and -35 promoter elements present in the two other loci (data not shown). The yciGFEkatN genes are absent from the closest E. coli relative, E. fergusonii (http://www.ncbi.nlm.nih.gov/). Altogether, these findings are consistent with horizontal acquisition of these genes in E. coli.
In contrast to the situation in E. coli, the yciGFEkatN locus belongs to the core genome in Salmonella (41), suggesting that Salmonella acquired these genes before the lineage divided into the two Salmonella species, S. enterica and S. bongori. The 479 pb sequence between trpA and yciG (Fig.  1C) is also conserved in the 16 complete sequenced genomes of S. enterica (http://www.ncbi.nlm.nih.gov/) and in S. bongori (http://www.sanger.ac.uk/Projects/Salmonella/). The sequence of the -10 and -35 elements is identical in all these Salmonella genomes. The sequence of the YncC binding region (in pyciG STM ) (Fig. 3B) is 100% identical in the genomes of S. enterica subsp. enterica and only one mismatch is found at the boundaries of this motif (Fig. 3B) in the two most ancestral groups (41), S. enterica subsp. arizonae (T/C at position -91) and S. bongori (G/A at position -110).
H-NS Silencing of yciGFEkatN is Relieved in Stationary Phase by S in Salmonella-H-NS is an abundant histone-like protein that binds preferentially to AT-rich DNA and subsequently oligomerizes along the DNA resulting in the formation of extended nucleoprotein complexes that cause gene repression (14 -16). Preferential binding of H-NS to sequences with higher AT-content than the resident genome allows H-NS to repress the expression of foreign DNA in a process known as "xenogeneic silencing" (15,16). Selective silencing of foreign DNA with low GC content in Salmonella by H-NS has been reported using chromatin immunoprecipitation and microarray analyses (39,40). Examination of these data revealed that H-NS binds the yciGFEkatN locus and represses its expression.
In agreement with these findings, expression of katN-lacZ in S. Typhimurium ATCC14028 (Fig. 2D) and SL1344 (data not shown) was increased by an hns mutation by more than 10-fold in exponential phase and by twofold in stationary phase. In the hns mutants, S was still required for katN-lacZ expression ( Fig. 2D and data not shown). The growth rate of Salmonella was highly affected by the hns mutation (Fig. 2D). H-NS is required for the normal proteolytic turnover of S (42) and, thus, levels of S in the exponential phase were higher in the hns strains than in the wild-type strains (Fig. 5D, lanes 5  and 7). As previously reported (39,40), the high S content in the hns strain likely impairs growth, because the growth defect of the hns mutant can be partially alleviated by deleting rpoS (Fig. 2D) or by replacing the wild-type rpoS allele by the rpoS LT2 allele (Fig. 2E). The rpoS LT2 allele contains a rare TTG start codon (instead of ATG). This mutation lowered the S level in ATCCrpoS LT2 and ATCCrpoS LT2 hns in exponential phase (Fig. 5D, lanes 5, 7, and 11, and lanes 7 and 13) and to a lesser extent in stationary phase (21) (Fig. 5C, lanes 5 and  11-14). In stationary phase, derepression of katN-lacZ expression by the hns mutation was stronger in ATCCrpoS LT2 than in ATCC14028 (12-and twofold respectively, Figs. 2D, E). However, S levels and katN-lacZ expression levels in stationary phase were not highly different in the two hns strains (twofold difference) (Figs. 2D, E and Fig. 5C lanes 13  and 14). These results suggested that the reduction in S level, because of the rpoS LT2 mutation, potentiated the magnitude of repression of katN-lacZ expression by H-NS.
S is not Efficient to Counter H-NS Silencing of yciGFE in E. coli K-12-Two peaks of 18,590 and 18,961 Da, likely corresponding to the YciF and YciE proteins, were detected by SELDI-TOF ProteinChip analyses in the ⌬hns mutant of MG1655 but not in the ⌬hns⌬yciF and ⌬hns⌬yciE mutants respectively (Fig. 4A). Detection of the YciE and YciF proteins in the hns mutant but not in the wild-type strain (Fig. 4A) suggested that expression of these proteins is silenced by H-NS in E. coli K-12, in agreement with a previous finding (43). Expression of the yciE-lacZ fusion was consistently derepressed in the hns mutant of MC4100 (Fig. 4B, lanes 1 and 2). Expression of the yciE-lacZ fusion in the hns strain was strongly affected by the rpoS mutation (Fig. 4B, lanes 2 and 5). These results indicated that S induces expression of yciGFE in E. coli K-12 only when H-NS mediated repression was relieved. Repression of yciGFE(katN) by H-NS is inversely correlated with the S levels (Figs. 2D, E and Figs. 5C, D), and, thus, one possible explanation for these results is that E. coli produces less S than Salmonella. This hypothesis was ruled out because similar levels of S were detected in MG1655, MC4100, MC1061, and ATCC14028 (Figs. 5B, C, D). In addition, similar levels of S were detected in the hns derivatives of these strains (Fig. 5B and data not shown).
YncC Directly Activates E S -dependent In Vitro Transcription at pyciG STM -Significant expression of the yciGFEkatN operon was not detected in the ⌬rpoS mutant of Salmonella (12) (Fig. 2F, lane 2), and katN-lacZ expression was very low in the rpoShns strain of Salmonella (Figs. 2D, E). These results suggested that pyciG STM is not efficiently transcribed in the absence of S . In agreement with this conclusion, a transcript initiating at pyciG STM was detected in vitro with the S -holoenzyme (E S ) but not with 70 -holoenzyme (E 70 ) (Fig. 6B).
In plasmids pACKyncC and pyncC HIS , yncC lacks its own promoter and is transcribed from the promoters in the vectors (Table I). pACKyncC and pyncC HIS did not induce expression of katN-lacZ in the exponential phase of growth ( Fig. 2B and data not shown). In addition, pyncC HIS did not induce expression of the fusion in the rpoS and rpoShns strains (Fig. 2F,  lanes 7 and 8 and data not shown). These results suggested that YncC and S act in concert to induce full expression of the operon. Consistent with these findings, YncC induced in vitro transcription by E S , but not by E 70 , at pyciG STM (Fig.  6B). YncC activation was moderate (2 Ϯ 0.6 at 100 -300 nM YncC) but reproducible in six independent experiments (Figs. 6B, C and data not shown). YncC was without effect when the fragment used for transcription lacked the YncC DNA binding region, (yciG STM short) (Figs. 6A, C). YncC was also without effect on transcription initiation at the S -dependent katE promoter (Fig. 7) (lanes 17 and 18) in agreement with in vivo data (not shown). Interestingly, transcription activation by YncC was not detected with a RNA polymerase in which the ␣ subunit lacks the C-terminal domain (Fig. 6C). These results suggested that transcription activation by YncC requires the ␣CTD domain.
On binding to high affinity "nucleation" sites, H-NS spreads along the DNA to lower affinity sites to occupy the promoter region, allowing the formation of higher order structures. An H-NS binding region contains several sites with variable affinity, and the number and the organization of binding sites determines the formation of a repressive nucleoprotein complex and modulates H-NS repression of gene expression (14 -16). DNase I footprinting experiments showed that H-NS binds to the pyciG ECO promoter region (Fig. 8, lanes 2-5). The binding pattern was marked by a series of protected and hypersensitive bands, indicating that H-NS changes the local topology of the promoter region. The polymerization of H-NS molecules along the promoter sequence would be expected to repress yciGFE expression by promoter occlusion or by antagonizing open complex formation. According to the 10-bp consensus binding sequence reported for H-NS nucleation sites (44), two predicted H-NS binding sites of high affinity are located in the upstream promoter region, precisely in the motifs recognized by YncC/McbR. The first site, located in the upstream YncC box AATaTATCcA (Fig. 3B), matches the consensus over 8 bp, whereas the other, AATaTATttt in the downstream YncC box (Fig. 3B), has 6 bp matches and, therefore, is expected to have lower affinity (note that the bases matching the H-NS consensus site are indicated in capital letters).
Interplay Between H-NS and YncC at the yciG STM Promoter-In vitro repression by H-NS was less marked at pyciG STM than at pyciG ECO (Fig. 7), a result consistent with in vivo data (Figs. 2F and 4B). The magnitude of H-NS repression was not affected by YncC (Fig. 7), and the magnitude of YncC activation was not affected by H-NS (Fig. 7). H-NS was able to bind to pyciG STM , protecting multiple sites along the DNA leading to the formation of a repression complex (Fig. 8, lanes 12 to 16). As for the pyciG ECO , two predicted H-NS binding sites (agaATAtATT centered at -98.5 and AtaTTATCTc centered at -93.5) in pyciG STM overlap the single YncC binding site (Fig.  3B). In the competitive footprinting assay between the two proteins, YncC occupied the pyciG STM site at a lower concentration than at pyciG ECO , but this did not appear to prevent H-NS binding at other sites along the promoter fragment (Fig.  8, lanes 20 -22). Indeed, some bands around -88, which are not protected by YncC or H-NS alone, are protected in the combined footprint, supporting the notion that H-NS and YncC did not compete for binding but rather bound simultaneously to pyciG STM fragment (Fig. 8, compare lanes 18 and  22). These results are in agreement with the in vitro transcription data (Fig. 7), and suggest that YncC activation and H-NS repression occur independently at pyciG STM .
The effect of the ⌬yncC mutation on katN-lacZ expression in Salmonella was attenuated in the absence of H-NS (Figs. 2A, D,  E). One possible explanation for this result is that the greater abundance of S in the hns strain compared with the wild-type strain (Fig. 5D) reduced the need for YncC in the absence of H-NS. Indeed, the impact of YncC on katN-lacZ expression was greatest at the entry to stationary phase (Figs. 2A, B), when S begins to accumulate in the cells (21), suggesting that the impact of YncC might be greatest at low S concentrations. Consistent with this hypothesis, katN-lacZ transcription activation by His 6 -YncC was higher in the presence of the rpoS LT2 allele (ninefold) (Fig. 2F lanes 5 and 6) than in the presence of the wild-type rpoS allele (3.5-fold) (Fig. 2F lanes 9 and 10). DISCUSSION In the present study, a proteomic method using ProteinChip arrays coupled with surface-enhanced laser desorption/ionisation time of flight mass spectrometry (SELDI-TOF-MS) was used for comparative proteomic profiling of cell extracts from Salmonella strains. These experiments revealed three proteins, subsequently identified as the yciF, yciE, and katN gene products that were produced in lower amounts in the yncC mutant than in the wild-type strain. Gene fusion analyses and in vitro transcription and DNase I footprinting experiments demonstrated that YncC controls production of these proteins at the transcriptional level and acts in concert with S . The S also controls expression of yncC (7), and thus, yciG-FEkatN is regulated by a S -dependent feed-forward regulatory loop. This dual role of S in the control of the operon, and the inverse correlation, observed in this study, between the S level and the magnitude of H-NS repression, might account for the strong sensitivity of yciGFEkatN expression to S levels and to activation by Crl (21), the S -chaperone that increases S activity. Like most S -dependent promoters, pyciG STM had a moderate activity in vitro, which was slightly stimulated by binding of YncC to pyciG STM (Figs. 3, 6, and 8), and required ␣CTD of RNAP (Fig. 6). The results suggested that YncC might act as a class I activator, making direct interactions with ␣CTD, thereby recruiting the rest of RNA polymerase (46). The location of activator binding in promoters subject to class I activation is variable, because of the flexibility of the linker between the N-and C-terminal domains of the ␣ subunit, but is usually near positions -61, -71, -81 or -91. The position of the YncC binding site, centered at -101 with respect to the transcription start site of yciG, is 10 bp upstream of the most distant transcription activators at simple 70 -dependent promoters. However the intrinsic DNA curvature found in many S -regulated promoters (including pyciG STM , data not shown) might facilitate protein-protein contacts between YncC and the CTD of the distal ␣ subunit of RNA polymerase, which is used preferentially by E S for activation (46 -48).
H-NS and YncC can bind DNA simultaneously to regulate in vitro transcription at pyciG STM (Figs. 7 and 8). In vivo however, the magnitude of YncC activation was reduced in the absence of H-NS (Figs. 2D, E). This might result from the high S levels in the hns strains, or from the involvement of an additional molecule that regulates yciGFEkatN expression. Alternatively, H-NS not only binds to the promoter region of yciGFEkatN ( Fig. 8) but also to coding regions (39, 40 and our unpublished results) and might form DNA bridges that contribute to transcription repression (14 -16). Significant binding of YncC to the coding sequences tested so far was not observed (data not shown). YncC might help relieve H-NS silencing indirectly, by increasing the transcription initiation rate at pyciG STM and, thus, the transcription elongation rate across the H-NS binding region, in line with the situation reported at the bgl promoter (49). Further experiments will evaluate the effect of downstream sequences in H-NS silencing of pyciG STM .
YncC/McbR and H-NS are also able to bind to the promoter region of the E. coli K-12 yciGFE genes. However, the sequence, the length, and the position of the McbR/YncC and H-NS binding regions, relative to the yciGFE(katN) promoter, are different in E. coli K-12 and Salmonella, resulting in differential mechanisms of regulation of these genes by YncC/ McbR and H-NS. It is remarkable that in E. coli K-12, regulation of yciGFE by YncC and by H-NS are intimately linked, whereas in Salmonella, YncC directly activates transcription, and thus, activation by YncC is, at least partly, disconnected from the H-NS network. One exciting possibility is that Salmonella has evolved the yciGFEkatN cis-regulatory sequences to integrate this locus into the RpoS network while maintaining its connection to the H-NS network, ultimately resulting in a more versatile but tightly controlled expression of this locus. The feed-forward regulatory loop mediated by YncC might allow signal input at levels downstream of S itself, through modulation of YncC activity or expression. Our data show that YncC production and/or activity is a limiting factor for yciGFEkatN expression and that the impact of YncC is major at low S concentrations. One hypothesis is that YncC induces yciGFEkatN expression under a specific environmental condition in the exponential phase of growth where S level is low. Our future experiments will assess environmental signals that might modulate YncC activation of yciG-FEkatN. These experiments might reveal putative cofactors that bind to the C-terminal domain of YncC and modulate its DNA binding activity.
The evolution of promoter architecture in closely related bacterial species might have important consequences for bacterial adaptation (50 -52). The physiological role of the yciGFE(katN) locus is unknown, and the significance, in the fitness of E. coli and Salmonella, of the differential regulation of these genes requires further investigation. Structural comparisons suggest a role for YciF in iron storage and/or protection against oxidative damage (53). KatN belongs to the family of manganese catalases but it does not play a major role in hydrogen peroxide resistance of Salmonella under standard growth conditions or in virulence in mice (12,54). This is because of the functional redundancy of the five hydrogen peroxide scavengers (three catalases and two alkyl hydroperoxide reductases) that contribute to Salmonella virulence and oxidative stress resistance (54). Nevertheless, overproduction of KatN increased resistance of Salmonella to hydrogen peroxide (12) and enhanced its virulence in NF-〉 pathway mutant Drosophila (55), suggesting that KatN might indeed contribute to Salmonella fitness. One could speculate on a correlation between gene polymorphism at the trpA-yciGFE-ompW locus in E. coli, silencing of yciGFE by H-NS, and the absence of katN. Investigation of the regulation and the role of the prophage-borne yciGFEkatN locus in the virulence and fitness of pathogenic E. coli O157:H7 strains might provide insight into the evolution and the function of this locus in closely related Enterobacteria.