Journal of Bacteriology, October 2003, p . 5722-5734, Vol . 185, No . 19

Listeria monocytogenes ^B Regulates Stress Response and Virulence Functions

Mark J . Kazmierczak, Sharon C . Mithoe, Kathryn J . Boor, and Martin Wiedmann^*

Department of Food Science, Cornell University, Ithaca, New York

Received 18 April 2003/ Accepted 15 July 2003

In L . monocytogenes, a food-borne pathogen capable of causing mild to severe infections in humans, ^B confers stress resistance (e.g., under acid stress and osmotic stress) and contributes to pathogenesis . To illustrate, an L . monocytogenes ^B null mutant survives less well than the wild-type parent at low pH (pH 2.5) and in a murine infection model (45, 67). ^B has been demonstrated to contribute to transcription of prfA, which encodes the central virulence gene regulator PrfA in L . monocytogenes (45) .

An increasing body of evidence suggests a broad role for ^B-dependent genes in the virulence of gram-positive bacteria . For example, ^B has been linked with regulation of virulence gene expression in S . aureus (6, 12) . Specifically, ^B contributes to transcriptional regulation of sarC in S . aureus . The sar locus partially controls expression of the agr locus; agr and sar are both global regulatory elements that control the synthesis of a variety of extracellular and cell surface proteins involved in the pathogenesis of S . aureus (12, 38) . A total of 23 proteins showed increased expression in the presence of ^B in a two-dimensional gel comparison of proteins isolated from both the wild-type S . aureus strain and its isogenic sigB null mutant (27). ^B also has been shown to contribute to virulence in Bacillus anthracis . Specifically, a B . anthracis sigB strain is less virulent than its wild-type parent (21) . Links between environmental stress responses and virulence in these bacteria suggest a central role for ^B in contributing to the ability of bacterial pathogens to survive environmental stress conditions, to direct expression of virulence genes, and to cause disease .

Identification of the genes regulated by ^B is the first step in deciphering specific mechanisms through which this alternative sigma factor confers resistance to multiple environmental stresses and contributes to virulence . Previous efforts at defining the ^B regulon have focused primarily on B . subtilis, a gram-positive, nonpathogenic model organism . Through a combination of in vitro transcription, reporter fusion transposon mutagenesis, two-dimensional protein gel electrophoresis, and RNA hybridizations (reviewed in reference 52), a total of 75 genes and proteins with ^B-dependent expression patterns were identified in B . subtilis . Recently, full genome macroarray analyses enabled two research groups to independently define over 120 B . subtilis genes that showed ^B-dependent expression profiles by comparing gene expression patterns of wild-type strains with those of isogenic sigB null mutants (51, 53) . In addition, use of a full-genome B . subtilis array allowed Helmann et al. (30) to identify 44 heat shock-induced genes preceded by previously unidentified potential ^B-dependent promoters . These transcriptome studies of the B . subtilis stress response illustrate the power of genome-wide, array-based expression studies .

The ^B regulon in B . subtilis, as determined by Price et al . (53) and Petersohn et al . (51), includes genes with a wide variety of functions . Approximately one quarter of the B . subtilis ^B-dependent genes encode proteins involved in metabolic functions, including glucose metabolism, protein degradation, and lipid metabolism . In addition, many genes regulated by ^B in B. subtilis encode transporters, e.g., solute transporters, permeases, and ATP-binding cassette (ABC) transport systems . The broad range of gene functions associated with the ^B regulon in B . subtilis suggests that, in addition to enhancement of fundamental cellular processes, ^B-mediated stress resistance mechanisms are also responsible for targeted action against specific stresses .

To better understand the role of ^B in stress resistance and virulence in L . monocytogenes, we combined promoter searches and DNA microarrays to identify genes directly activated by ^B from experimentally defined or predicted ^B-dependent promoter sequences . As some B . subtilis genes have shown apparent ^B-dependent induction in the absence of well-conserved ^B-dependent promoter sequences, some ^B-mediated effects may also occur through indirect regulatory mechanisms (e.g., possibly through ^B-dependent expression of additional transcriptional regulators [51]) .

For the purposes of this study, we chose to focus specifically on defining L . monocytogenes genes that are transcribed from ^B-dependent promoters . To that end, we first performed a hidden Markov model (HMM) similarity search for putative ^B-dependent promoters against the published complete L . monocytogenes genome sequence (28) . To allow a focused analysis of ^B-dependent gene expression, a microarray was constructed with the genes identified by the HMM search. To more fully explore the role of ^B in L. monocytogenes stress response and virulence, our microarray also included previously identified stress response and virulence genes that were not necessarily identified by HMM .

To be classified as being under direct ^B control, genes had to meet the following two independent criteria: (i) significantly higher expression in the wild-type strain compared to the sigB null mutant under the conditions used for the microarray experiments and (ii) the presence of an identifiable ^B-dependent promoter-like sequence upstream of the coding region . With these criteria, we identified 54 genes under direct ^B control . Although we refer to these genes as ^B dependent, it is important to recognize that the genes identified with this approach may not be exclusively ^B dependent but may also be regulated by ^B-independent mechanisms . Interestingly, while the genes that we found to be controlled by ^B include many that had been identified previously as part of the B . subtilis ^B regulon, several virulence and virulence-related genes were also identified (e.g., a gene encoding a bile salt hydrolase and genes encoding surface molecules with possible or confirmed roles in host cell attachment) . Our data provide further evidence in support of the contributions of ^B to L. monocytogenes stress resistance . Importantly, identification of multiple ^B-regulated virulence genes strongly suggests that this alternative sigma factor also contributes to the ability of L . monocytogenes to infect and survive within a host .

HMM searches. HMM searches were performed with HMMER (http://hmmer.wustl.edu) essentially as described by Price et al. (53) . The training alignments included 29 B . subtilis ^B-dependent promoters (described in reference 53) as well as four confirmed (prfA, rsbV, and opuCA) or likely (betL) L . monocytogenes ^B-dependent promoter sequences . The models were searched against the complete L . monocytogenes strain EGD-e genome sequence (28) . The output results were filtered, and only hits within 350 bp upstream of a coding region, as predicted by the ListiList web server (http://genolist.pasteur.fr/ListiList [44]), and with an E value of 0.006 were kept .

DNA and RNA isolation and treatment. RNA was extracted from L . monocytogenes either grown to stationary phase or exposed to osmotic stress (as described below) with the RNeasy Protect Bacteria midi kit (Qiagen, Valencia, Calif.) . The enzymatic and mechanical disruption protocol provided by the manufacturer was used, with the exception that cell lysis was performed with sonication at 21W three times for 20 s each (with cells iced between bursts) instead of bead beating . RNA was eluted from the column in RNase-free water and ethanol precipitated overnight at -20°C. Precipitated RNA was centrifuged, washed in 70% ethanol, and resuspended in RNase-free water . DNase treatment was performed with 30 U of RQ1 RNase-free DNase (Promega, Madison, Wis.) in the presence of 400 U of RNase inhibitor (RNasin; Promega) for 1 h at 37°C . The reaction mix was then extracted twice with an equal volume of 50% phenol-50% chloroform, followed by one extraction with an equal volume of 100% chloroform . RNA was ethanol precipitated from the aqueous layer and stored at -20°C in ethanol .

Cultures for RNA isolation were inoculated from single colonies and grown overnight at 37°C. The cultures were then diluted 1:200 in BHI and grown at 37°C to specified growth phases . Specifically, for harvest of stationary-phase cells, L . monocytogenes was grown for one additional hour after reaching an optical density at 600 nm (OD₆₀₀) of 0.8 . At this point, a 9-ml aliquot of culture was added to 18 ml of RNA Protect Bacteria reagent (Qiagen), and RNA isolation was performed as described above . For osmotic stress treatment, cells present in 40 ml of a culture grown to an OD₆₀₀ of 0.4 were collected by centrifugation and resuspended in 8.25 ml of warm (37°C) BHI broth, to which 1.25 ml of 4 M KCl was added to yield a final concentration of 0.5 M KCl. After 7 min of incubation at 37°C, corresponding to the approximate peak of ^B-induced transcriptional response reported in induction experiments in B . subtilis (64), 20 ml of RNA Protect Bacteria reagent was added, and RNA isolation was performed as described above .

Chromosomal DNA used for genomic DNA microarray control spots was isolated as described by Flamm et al. (20) from an overnight culture of L . monocytogenes . Briefly, cell walls were digested with lysozyme in 20% sucrose, followed by cell lysis with sodium dodecyl sulfate and proteinase K . DNA was purified by phenol-chloroform extractions and ethanol precipitated . DNA was then resuspended in spotting buffer consisting of 3x SSC (1x SSC is 0.15 M NaCl plus 0.015 M sodium citrate) plus 0.1% Sarkosyl) and quantified by UV spectroscopy .

Microarray construction. DNA microarrays were constructed to include 208 L . monocytogenes genes spotted in triplicate, 30 mouse ß-actin cDNA spots for nonhybridizing controls, and 60 spots of genomic DNA to aid in data normalization and analysis . The 208 L . monocytogenes genes included 166 of the 170 genes identified by HMM searches as including an upstream predicted ^B-dependent promoter (PCR amplification failed for four genes) as well as 36 genes previously reported to be involved in virulence or stress response and six negative control genes (i.e., housekeeping genes predicted not to show ^B-dependent expression) . All 208 target genes as well as the PCR primers used for their amplification are detailed in the supplemental material available at http://www.foodscience.cornell.edu/wiedmann/Mark%20supplemental%20table%20 (4-15).doc .

PCR primers were designed with PrimeArray software (56) to amplify the complete open reading frames (ORFs) of the selected genes . For ORFs of >1 kb, primers were selected to amplify a 1-kb fragment at the 5' end of each ORF . PCR was performed with AmpliTaq Gold (Applied Biosystems, Foster City, Calif.) with an L. monocytogenes 10403S cell lysate (prepared as described in reference 24) as template DNA . Appropriately sized products free of nonspecific amplification or extraneous bands, as determined by agarose gel electrophoresis, were purified by ethanol precipitation and reconstituted in spotting buffer (3x SSC-0.1% Sarkosyl) to a final concentration of 50 to 150 ng/µl . PCR was also used to generate a 1-kb PCR fragment of the mouse ß-actin gene with mouse ß-actin cDNA (Sigma, St . Louis, Mo.) as the template .

Arrays were spotted with a GMS 417 robotic arrayer (Affymetrix, Santa Clara, Calif.) on GAPS2 glass slides (Corning, Corning, N.Y.) . Immediately before use, slides were blocked in a solution of 1-methyl-2-pyrrolidinone containing 1.44% succinic anhydride and 0.02 M boric acid (pH 8.0), as directed by the slide manufacturer .

cDNA labeling and microarray hybridization. Precipitated, DNase-treated RNA was centrifuged, washed once in 70% ethanol and once in 100% ethanol, and resuspended in RNase-free water . The RNA was quantified on a UV spectrophotometer and checked for quality by A_260/280 ratio and formaldehyde-agarose gel electrophoresis . cDNA was generated from 10 µg of RNA with random hexamers and SuperScript II (Invitrogen, Carlsbad, Calif.) in the presence of indocarbocyanine (Cy3)-dUTP or indodicarbocyanine (Cy5)-dUTP (Amersham Biosciences, Piscataway, N.J.) . cDNA synthesis and labeling were performed with a dye-swapping design (sometimes referred to as reverse labeling [62]); each RNA sample was used for two separate labeling reactions, one with Cy3 and one with Cy5 . Completed reactions were incubated with 1.5 µl of 1 N NaOH at 65°C for 10 min to inactivate the enzyme and degrade the RNA . Reactions were then neutralized with 1.5 µl of 1 N HCl .

Wild-type and sigB mutant cDNA probes were mixed and purified with the QIAquick PCR purification kit (Qiagen) as described by the manufacturer, with the addition of a 750-µl 35% guanidine-HCl wash after binding . The purified probes were then dried and reconstituted in 20 µl of hybridization buffer (3.5x SSC, 0.25% sodium dodecyl sulfate, and 0.5 µg of salmon sperm DNA/µl; Invitrogen) . Probes were boiled for 2 min, applied to arrays, and hybridized overnight in a 60°C waterbath . Before scanning, slides were washed in 2x SSC-0.1% sodium dodecyl sulfate at 60°C for 5 min, followed by room temperature washes with 2x SSC-0.1% sodium dodecyl sulfate, 2x SSC, and 0.2x SSC for 5 min each . Slides were centrifuged to dry them and scanned with a Perkin Elmer Scan Array 5000 confocal laser scanner (Perkin Elmer, Wellesley, Mass.) .

Microarray replicates and data analysis. For each stress condition described above (osmotic stress or stationary phase), two independent RNA isolations (for both wild-type and sigB cells) were performed on separate days to provide true biological replicates . Each set of independent RNA samples was used to perform two microarray hybridizations with a dye-swapping design to correct for differences in Cy3 and Cy5 dye incorporation and to provide experimental duplicates . Briefly, both wild-type and sigB cell RNAs from each of the two independent RNA isolations were labeled separately with either Cy5 or Cy3 as described above . Labeled cDNA was used to perform two independent microarray hybridizations, one with Cy3-labeled wild-type and Cy5-labeled mutant cDNA and one with Cy5-labeled wild-type and Cy3-labeled mutant cDNA. Thus, data for four microarray repetitions were collected for each stress condition .

Raw TIFF images from the scanner were loaded into ScanAlyze (http://rana.lbl.gov/EisenSoftware.htm) for analysis . Spot grids were manually fitted to the microarray images. Spots were flagged and eliminated from analysis only in obvious instances of high background or stray fluorescent signals .

Because our microarray contained a relatively small number of genes, most of which were expected to show differential expression, proper normalization of the raw data was critical. Equalizing the mean intensities of both channels of an array, the most common method used for microarray normalization, would thus have provided severely skewed data on an array, with the majority of spots appearing to be unequally expressed . Therefore, we normalized our data using the mean intensities of spots containing genomic DNA . Data normalization was performed simultaneously on all four array data sets for a given stress condition, as follows . For each channel (Cy3 or Cy5), the mean intensity of all genomic DNA control spots was calculated independently . A correlation factor comparing genomic means among channels was calculated and used to proportionally adjust all spot intensities (experimental and control) so that the means of genomic spot intensities were equal across all arrays . From these normalized values, a floor was calculated as the average intensity plus 2 standard deviations of all ß-actin spots in a channel . Spots with values below the floor in both channels were eliminated from analysis; spots below the floor in only one channel had that channel intensity raised to the floor value, and the resulting data were included in the analyses (16) .

Floored and normalized channel intensities were analyzed with the Significance Analysis for Microarrays (SAM) program (63) . This statistical analysis involves factoring the change in expression of each gene relative to the standard deviation of all replicates for that gene. Therefore, genes with a low change will not be discounted if the ratios are consistent among repeats, effectively reducing false-negatives. False-positives are also avoided when genes have poor reproducibility between replicates . This method of statistical analysis maximizes both the quantity of genes found and the reliability of the results . Spot intensities for all channels were input in SAM as paired, unlogged values . All individual spots were considered repetitions, generating 24 data points for each gene (3 spots per gene x 4 arrays x 2 channels per array) . Delta values were chosen according to the lowest false discovery rate . Only genes with expression ratios of >1.5 were considered biologically significant .

RACE-PCR. Promoter regions were mapped with the 5' rapid amplification of cDNA ends (RACE) system (Invitrogen) according to the manufacturer's protocol . Briefly, gene-specific first-strand cDNA was tailed with dCTP by using terminal transferase. The products were then amplified with a nested gene-specific primer and a poly(G/I) primer in a touchdown PCR with AmpliTaq Gold (Applied Biosystems) . PCR products present in the wild-type L. monocytogenes but absent in the sigB strain were purified with the Qiagen QIAquick PCR purification kit and sequenced with the Big Dye terminator system and an Applied Biosystems 3700 sequencer at the Cornell University Bioresource Center .

Transcriptional analysis of potential ^B-dependent genes. Microarray analyses were performed with RNA isolated from wild-type L . monocytogenes and an isogenic sigB strain grown to stationary phase or exposed to osmotic stress, two conditions shown to activate ^B (4, 19, 22) . We identified 55 genes that showed significant expression ratios, with >1.5-fold expression in the wild-type strain over that in the mutant strain in at least one of the two stress conditions tested (Table 1) . Fifty-one of these genes displayed >2-fold induction; the highest relative induction was 27-fold .

 
Figure 1 shows scatter plots of wild-type versus sigB mutant spot intensities . As the microarray predominantly targeted genes predicted to be positively regulated by ^B, the distribution of the majority of the points above the diagonal, which indicates that a large proportion of the genes in the microarray were expressed more strongly in the wild type than in the mutant, was expected . As the presence of points falling far below the diagonal would signify increased gene expression in the absence of ^B, none to few were expected . For the purpose of this study and in accordance with previous studies (31), we applied a cutoff of 1.5-fold induction to signify biological significance . Two genes included in the microarray fell below this cutoff (showing statistically significant expression ratios between 1.2- and 1.5-fold) and are not discussed further here . As shown in Fig. 1, a few genes showed higher expression in the sigB mutant than in the wild-type strain; expression differences between strains for these genes were determined to be not statistically significant .

 
Overall, 44 of the 169 genes (26%) in our microarray that were either (i) predicted to be ^B dependent by HMM or (ii) members of operons previously identified as being regulated by a ^B-dependent promoter also showed ^B-dependent expression in our microarray experiments . There was no apparent correlation between a promoter's HMM E value and the relative expression level of a corresponding gene, as determined by microarray (r² < 0.01 for correlations between HMM E values and relative expression levels under either osmotic stress or stationary-phase conditions) . In addition, 11 of the 36 currently recognized stress response and virulence genes that were included in the array even though they did not show an HMM-predicted ^B-dependent consensus promoter displayed ^B-dependent expression patterns . Subsequent visual inspection of the upstream regions for these 11 genes identified putative ^B consensus promoter sequences for 10 of them (Table 2) . Although inlD showed ^B-dependent expression in our microarray experiments, no putative ^B consensus promoter sequence was identified for this gene .

 
Overall, we identified 54 ^B-dependent L . monocytogenes genes through combined application of HMM and microarray analyses . Based on identification of the predicted proteins, the genes were grouped into 10 functional groups (Table 2) . The two most highly represented categories were transport and metabolism proteins, which together comprised 20 of the 54 genes (Table 2) . Statistical analysis of microarray data (SAM [63]) showed that, in addition to high expression ratios, the putative oxidoreductase gene lmo0669 also had the highest significance score for differential expression of all genes represented in the microarray (Table 2) . The known ^B-dependent betaine/carnitine transporter operon opuC also showed ^B-dependent expression, along with eight other solute transport genes . The large proportion of metabolism and transport genes regulated by ^B underscores the importance of maintaining proper cellular functions during exposure to stress conditions and suggests that ^B provides protection by enhancing the operation of a specialized subset of basic cellular processes, which may help to produce a more stress-resistant state for the cell .

Two other notable functional categories of ^B-dependent genes are stress response and virulence genes (Table 2) . The stress response genes, which have apparent functions in bacterial stress resistance but do not encode transporters or proteins from other recognized categories included in Table 2, represented some of the most highly differentially expressed genes . For example, in the SAM output, which ranks differentially expressed genes based on the statistical significance associated with the expression ratios, four of the seven stress response genes ranked among the 16 genes with the highest significance values for differential expression, for RNA isolated from cells exposed to osmotic stress . The stress response genes identified include gadB, which was ranked as the third or fifth most significantly differentially expressed gene in cells grown to stationary phase or exposed to osmotic stress, respectively .

Three confirmed virulence genes were expressed at a higher level in the wild-type L . monocytogenes strain compared to the sigB mutant (Table 2) . bsh exhibited 20.1-fold higher mRNA levels in the wild type compared to the sigB strain under osmotic stress (Table 2); this expression ratio was determined to be highly significant by SAM analysis (ranked as the second most significant ratio for all the genes in the array) . The virulence genes inlA and inlB showed significantly higher expression under osmotic and/or stationary-phase stress in the wild type compared to the mutant strain, with expression ratios consistently >2.0 (Table 2) . In addition to these confirmed virulence genes, three other internalins (inlC2, inlD, and inlE) also showed significant expression ratios of >2.0 under osmotic and/or stationary-phase stress (Table 2) .

PCR mapping showed that the L . monocytogenes strain used in our study (10403S) bears an inlC2DE operon rather than an inlGHE operon, which is present in the L. monocytogenes EGD-e strain (57) . Each ORF of the inlC2DE gene cluster is potentially transcribed independently of the others (13) . We identified a ^B promoter consensus sequence upstream of both inlC2 and inlE but not inlD . Only one previously described internalin gene (inlC) that was included in our array and is present in strain 10403S did not show differential expression between the wild-type and the sigB mutant strains .

Microarray identification of operons and known ^B-dependent genes. For putative ^B-regulated genes that were identified only by HMM, only PCR products from the genes most proximal to given HMM-predicted promoters were spotted on the microarray . Therefore, downstream genes that may have been cotranscribed from a given promoter were not tested for ^B dependence in this array . However, the microarray did include genes comprising two operons (opuC and rsbV) that were known a priori to be ^B dependent (4, 22) . All genes in these two operons were expressed at significantly higher levels in the wild-type compared to the sigB strain . Furthermore, expression ratios for each gene within these operons were approximately equal (Fig . 2A and B) . The inlAB operon was not previously known to be regulated by ^B, but inlA and inlB were among the virulence genes included on the microarray . We found that both genes in this operon were expressed at higher levels in the presence of ^B and that inlA and inlB displayed similar expression ratios (Fig. 2C) .

 
Confirmation of predicted ^B-dependent promoters by RACE-PCR. To confirm that predicted ^B-dependent promoters were in fact responsible for the differential gene expression seen in the microarrays, we performed RACE-PCR on eight selected genes . These genes were chosen to represent highly differentially expressed genes and genes with high significance scores in SAM as well as known stress response or virulence genes . Gene-specific first-strand cDNA was generated for each gene, and a 3' poly(dC) tail was added to each cDNA product with terminal transferase . The tailed cDNA product was amplified by touchdown PCR with a poly(G/I) primer (Invitrogen) and a nested gene-specific primer .

PCR bands from wild-type cDNA reactions that met the following conditions were purified and sequenced (Fig. 3): (i) bands that must have been generated by reactions with wild-type cDNA and (ii) equivalent bands that must have been absent in sigB mutant cDNA reactions . For all genes tested, the transcriptional start site determined by RACE was 10 nucleotides (± 2 nucleotides) downstream of the ^B-dependent promoter -10 region that had been predicted by HMM or by visual inspection (Fig. 4) . The genes selected for promoter confirmation by RACE were bsh and inlA, both virulence genes; lmo0669, a putative metabolic gene; lmo1421 and opuCA (which encode putative and known compatible solute transporter proteins, respectively); and lmo2230, lmo1433, and gadB, all putative or proven stress response genes .

 
L . monocytogenes ^B promoter consensus sequence resembles that of B . subtilis. Successful identification of specific ^B-dependent genes in L . monocytogenes provided new information needed to facilitate further refinement of the predicted consensus sequence for ^B-dependent promoters in L . monocytogenes . The 54 ^B-dependent promoter sequences described above were aligned and used to generate a new consensus recognition site (Fig. 5) . The -35 region (GTTT) is separated from the -10 region (GGGWAT) by 13 to 17 nucleotides, most frequently by 15 or 16 . The guanidine in the -35 box is almost completely conserved (98%) . Overall, the L . monocytogenes ^B consensus sequence resembles the consensus sequence for B . subtilis ^B, which was reported by Petersohn et al. (50, 51) as GTTTAA-N_12-15-GGGWAW . This finding is not surprising, as the majority of the sequences used to train the HMM used in this study were obtained from B . subtilis genes . However, the predicted L. monocytogenes ^B consensus sequence does differ slightly from the B . subtilis consensus promoter sequence. Specifically, the two adenosines at the 3' end of the B. subtilis -35 region are not conserved in L. monocytogenes . The different consensus sequences suggest that ^B may vary in its ability to recognize certain promoters in these two species .

 
Effect of induction conditions on gene expression patterns. As ^B-dependent gene expression patterns were determined under two different stress conditions, we also analyzed whether the magnitude of ^B dependence for individual genes varied with the condition . Comparison of the microarray data generated under different stress conditions indicated that the relative magnitude of ^B-dependent expression was similar under the two conditions for the genes tested . The scatter plot in Fig. 6 shows the wild-type-to-sigB mutant gene expression ratios obtained for stationary-phase cells plotted against the gene expression ratios for osmotically stressed cells .

 
The majority (67%) of the ^B-dependent genes defined as described above showed significantly higher mRNA levels, with expression ratios of >1.5 in the wild-type L. monocytogenes strain under both stress conditions (Table 1) . For five genes, the expression ratios between the two stress conditions varied from each other by more than twofold; the largest difference was 2.5-fold (lmo0880, point A in Fig. 6) . However, the majority of the genes (39 out of 55) had expression ratios that differed by less than 1.5-fold under the two stress conditions . These results suggest that the majority of genes comprising the ^B regulon are induced to a similar extent in relation to each other regardless of the specific environmental stress . A subset of ^B-dependent genes (e.g., lmo0880) may also be subject to additional transcriptional regulation by other stress-specific mechanisms . This is not to say, however, that all stress conditions activate ^B to the same degree. For example, while our specific conditions resulted in similar induction levels among the targeted genes (slope of regression = 0.89), primer extension data suggest that induction of transcription from the ^B-dependent L. monocytogenes rsbV promoter may vary under the different stress conditions (4) .

Similar to the ^B regulon previously defined for B. subtilis, a gram-positive soil bacterium closely related to the genus Listeria, members of the L . monocytogenes ^B regulon were shown to encode a variety of protein functions involved in metabolic pathways, transport, and other fundamental cellular functions . As expected, many genes identified in this study encode proteins directly associated with stress resistance. Interestingly, several genes identified as ^B dependent represent putative or known virulence genes (e.g., inlE and bsh, respectively) or stress response genes that have been shown to contribute to the ability of L. monocytogenes to survive within an infected host (e.g., opuC [60, 66]) or under conditions similar to those encountered in an infected host (e.g., gadB [11]) . These results suggest that ^B, in addition to enhancing bacterial survival under stress conditions, also contributes to virulence in L . monocytogenes .

Role of ^B in L . monocytogenes stress resistance. Phenotypic characterization of L . monocytogenes strains lacking sigB has shown that ^B plays an important role in resistance to osmotic, oxidative, and acid stresses (4, 18, 19, 22, 67) . Similar studies with B . subtilis sigB mutants also provided evidence for the importance of a functional ^B for survival under ethanol, oxidative, osmotic, and acid stresses (2, 25, 65) . The ^B-dependent L . monocytogenes genes identified here provide additional insight into the functional bases of the reduced stress resistance in L . monocytogenes sigB mutants . For example, gadB, which encodes a glutamate decarboxylase important for acid stress survival in L . monocytogenes (11), was shown in this work to be part of the L . monocytogenes ^B regulon .

Interestingly, both the L . monocytogenes and the B . subtilis ^B regulons include multiple genes that have been shown experimentally to contribute to protection against osmotic stress, including genes encoding several solute transporters . The ABC transporter OpuC contributes to osmotic stress resistance by facilitating uptake of the osmoprotectants choline and glycine betaine in both L . monocytogenes and B. subtilis (1, 35, 36, 60), and expression of the encoding operon is ^B dependent in L. monocytogenes . Similarly, lmo1421 and the B. subtilis homologue opuB, each of which is a member of the ^B regulon in its species, both encode a putative choline transporter which may also contribute to osmotic stress resistance (22, 36) . In addition, ctc, which shows ^B-dependent expression in both L . monocytogenes (this study) and B . subtilis (29, 33), was recently shown to contribute to osmotolerance in L . monocytogenes (26) .

Finally, the ^B regulon in both L . monocytogenes and B . subtilis also includes genes that appear to contribute to oxidative stress resistance, consistent with the observation that an L . monocytogenes sigB mutant shows increased susceptibility to oxidative stress (18) . From our data, we propose that ^B-dependent resistance to oxidative damage in L . monocytogenes is likely due, at least in part, to glutathione reductase, which is encoded by lmo1433. Glutathione reductase is an enzyme that provides protection from oxidative stress by reducing glutathione disulfide to glutathione (9) . L. monocytogenes has been shown to accumulate glutathione, supporting the possible activity of this enzyme during oxidative stress (47) . While no glutathione reductase homologue was found in B . subtilis, oxidative stress protection in B . subtilis is dependent on the DNA-binding protein Dps, as well as on the trxA-encoded thioredoxin, both of which are ^B dependent (59) . Our HMM searches did not identify L . monocytogenes trxA as bearing a ^B-dependent promoter .

Role of ^B in virulence gene expression. Characterization of L. monocytogenes sigB mutants in both tissue culture and murine models of infection has provided initial evidence that ^B contributes to the ability of L. monocytogenes to cause infection . While it has been shown that transcription from one of three prfA promoters (specifically the P2 promoter) is abolished in a sigB mutant, no additional evidence for other functional contributions of ^B to virulence have been reported . Our results provide clear evidence that ^B in L. monocytogenes contributes to transcriptional activation of genes directly associated with virulence as well as those that likely contribute indirectly to virulence (58) .

Virulence genes defined as being part of the ^B regulon include bsh (encoding a bile salt hydrolase) as well as two genes from the internalin family (inlA and inlB). Although ^B-dependent transcription of inlA and bsh has also been independently verified by reverse transcription-PCR (D . Sue and M . Wiedmann, unpublished data), and while RACE-PCR confirmed the presence of functional ^B promoters for these genes, it is important to note that both genes also appear to be transcribed from ^B-independent promoters (14, 15) .

Interestingly, both bsh and inlA are regulated by PrfA (14, 15), a positive transcriptional regulator of virulence gene expression . While PrfA binding has been shown to activate transcription by binding to the -35 promoter region, none of the PrfA binding boxes upstream of the ^B-dependent virulence genes described here overlap the ^B consensus promoter -35 region . However, the ^B-dependent P2 promoter of prfA contains a sequence resembling a PrfA box (23), and Milohanic et al . (42) reported a gene, lmo0596, with a putative PrfA box located at the -35 region of a predicted ^B-dependent promoter . We conclude that transcription of a subset of L . monocytogenes virulence genes is regulated by a complex regulatory network that can include activation by both PrfA and ^B . We hypothesize that this PrfA/^B regulatory mechanism coordinates transcriptional activation of subsets of L. monocytogenes virulence genes in specific host environments, e.g., bsh and inlA have been shown to be critical for listerial pathogenesis in the intestine (15, 39) .

While bsh, inlA, and inlB represent experimentally verified virulence genes which were identified as being ^B dependent, we also defined as members of the ^B regulon additional putative L. monocytogenes virulence genes, including additional internalins. The L . monocytogenes internalins represent a diverse group of surface proteins with confirmed or putative virulence functions (39, 49, 57) . While internalin B does not display an LPXTG motif, the other internalins include this cell wall anchor domain . Proteins with LPXTG motifs are common among gram-positive bacteria, and their functions are broad in range and frequently affect virulence (reviewed in reference 46) . In addition to inlA, we identified four other L . monocytogenes genes encoding putative cell wall-anchored proteins displaying an LPXTG motif (including the internalin genes inlC2 and inlE) as being ^B dependent . While inlC2 and inlE, which are part of the inlC2DE operon in L. monocytogenes 10403S, have not been shown experimentally to contribute to virulence, members of the homologous inlGHE operon present in other L . monocytogenes strains (e.g., EGD-e) have been shown to contribute to host cell internalization (5, 57) . In addition to two other ^B-dependent genes encoding proteins with LPXTG motifs (lmo2085 and lmo0880), we also identified one ^B-dependent gene (lmo0994) that is unique to L . monocytogenes (i.e., absent from Listeria innocua and B . subtilis) . Further studies with appropriate null mutants will be necessary to determine the specific functions of these putative ^B-dependent virulence genes .

In addition to the established or putative virulence genes discussed above, we also identified ^B-dependent stress response genes that had been shown previously to contribute to L . monocytogenes virulence, infection, and intrahost survival . Examples include the ^B-dependent opuC operon; an L. monocytogenes LO28 opuC mutant showed reduced colonization of the mouse upper small intestine following peroral inoculation (60) and reduced numbers of bacteria in the spleens and livers of infected mice (66), although the phenotype appears to be strain specific (60) . The ^B-dependent gadB was previously shown to contribute to L . monocytogenes acid survival, including survival in synthetic gastric and ex vivo porcine stomach fluids (11) . In conjunction with our findings described above, these data support a broad role for ^B-dependent genes in L . monocytogenes virulence and intrahost survival . The majority of L. monocytogenes virulence studies thus far have used the mouse model of infection, but the mouse lacks the appropriate E-cadherin receptor, which is particularly critical for gastrointestinal invasion by L. monocytogenes (67). Use of more appropriate animal models to study L. monocytogenes pathogenesis (e.g., guinea pigs [39]) may help us to more clearly define the in vivo contributions of ^B to L . monocytogenes virulence .

Studies in other gram-positive pathogens have provided further evidence that ^B contributes to directing gene expression during host infection . In S . aureus, the virulence determinant SarA is expressed from a promoter that is strictly ^B dependent (6, 41) . SarA specifically activates transcription of agr, which encodes a positive regulator of extracellular virulence gene expression (43, 48) . In both S. aureus and S . epidermidis, ^B is also required for biofilm formation, a probable prerequisite for establishing infections (37, 54, 55). ^B contributes to virulence in Bacillus anthracis; a sigB mutation in B . anthracis severely attenuates virulence in a murine model of infection (21) . In addition to ^B, stress-responsive sigma factors may contribute broadly to bacterial virulence . For example, the gram-negative stress-responsive alternative sigma factor RpoS has also been shown to contribute to virulence in a variety of pathogens, including Yersinia spp., Salmonella spp., Pseudomonas aeruginosa, and Legionella pneumophila (3, 17, 34, 61) .

Comparative genomics of the ^B-dependent stress response. The definition of 54 ^B-dependent genes in L . monocytogenes provides a unique opportunity for a comparative evaluation of the predicted functions of the ^B-dependent stress response among low-GC-content, gram-positive bacteria . Overall, the range of protein functions associated with the L. monocytogenes ^B-dependent genes defined in this work appears to be similar to the range of functions encoded by the genes described for the B . subtilis ^B regulon (51, 53) . Through similarity searches, we determined that 31 of the 54 ^B-dependent genes in L . monocytogenes have homologous gene sequences in B . subtilis and 12 of these 31 genes show ^B-dependent expression in both L. monocytogenes (this work) and B . subtilis (51, 53) . These 12 genes include the stress response genes ctc, ltrC, and lmo1602, the rsbV operon, and metabolic genes (see Table 2) . These genome-scale comparisons of the L . monocytogenes and the B . subtilis ^B regulons thus suggest that the ^B-dependent stress response system has adapted in L . monocytogenes to facilitate pathogen-host interactions .

The analysis of the S . aureus ^B regulon performed by Gertz et al. (27) identified predominantly uncharacterized proteins . While 20 of the 27 ^B-dependent S . aureus proteins identified had homologues in B . subtilis, only 7 of those homologues were ^B dependent in B . subtilis . This finding parallels our own in that only a portion of the B . subtilis genes that are homologous to the ^B-dependent genes identified in L . monocytogenes are also ^B dependent in B . subtilis . Taken together, these observations suggest that the ^B regulon has evolved to serve different roles among these bacteria .

Towards defining the complete L . monocytogenes ^B regulon. Stress-responsive alternative sigma factors, including RpoS in gram-negative bacteria (32) and ^B in low-GC-content, gram-positive bacteria, contribute to regulation of large sets of genes, both directly and indirectly (51) . As commercial, full-genome microarrays become available, more genes than those found here can be identified as ^B dependent, including those not identified by the original HMM search . Still, defining all genes regulated by an alternative sigma factor represents a significant challenge, even with the availability of technologies such as two-dimensional gel electrophoresis and full-genome microarrays .

While our identification of 54 ^B-dependent genes likely represents about one-third of the L . monocytogenes ^B regulon, which is estimated to contain around 150 genes, the functional diversity represented by the proteins encoded by these genes provides valuable new insight into the specific functions of the L. monocytogenes ^B regulon . In addition to the 54 genes reported here, many additional members of the L. monocytogenes ^B regulon may have been correctly predicted by our HMM search, but their detection may have been masked by redundant regulation of transcription . Additional approaches such as in vitro transcription analyses with a purified L . monocytogenes ^B-RNAP complex (8) or transcriptional profiling of an L . monocytogenes strain with an inducible sigB (53) will likely be necessary to identify and confirm additional members of the ^B regulon . We have demonstrated, however, that a combination of HMM-based similarity searches and construction of a microarray as described here represents an efficient and economical approach for defining genes regulated by a specific mechanism, such as an alternative sigma factor .

This research was supported in part by the Cornell University Agricultural Experiment Station federal formula funds, project no . NYC-143422, received from the Cooperative State Research, Education, and Extension Service, U.S . Department of Agriculture . Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of the U.S . Department of Agriculture .

1. Angelidis, A . S., L . T . Smith, L . M . Hoffman, and G . M . Smith. 2002 . Identification of opuC as a chill-activated and osmotically activated carnitine transporter in Listeria monocytogenes . Appl . Environ. Microbiol. 68:2644-2650.
2. Antelmann, H., S . Engelmann, R . Schmid, and M . Hecker. 1996. General and oxidative stress responses in Bacillus subtilis: cloning, expression, and mutation of the alkyl hydroperoxide reductase operon . J . Bacteriol. 178:6571-6578.
3. Bachman, M . A., and M . S . Swanson. 2001. RpoS co-operates with other factors to induce Legionella pneumophila virulence in the stationary phase . Mol. Microbiol. 40:1201-1214.
4. Becker, L . A., M . S . Çetin, R . W . Hutkins, and A . K . Benson. 1998 . Identification of the gene encoding the alternative sigma factor ^B from Listeria monocytogenes and its role in osmotolerance.J . Bacteriol. 180:4547-4554.
5. Bergmann, B., D . Raffelsbauer, M . Kuhn, M . Goetz, S . Hom, and W . Goebel.2002 . InlA- but not InlB-mediated internalization of Listeria monocytogenes by non-phagocytic mammalian cells needs the support of other internalins . Mol . Microbiol. 43:557-570.
6. Bischoff, M., J . M . Entenza, and P . Giachino. 2001. Influence of a functional sigB operon on the global regulators sar and agr in Staphylococcus aureus.J . Bacteriol. 183:5171-5179.
7. Bishop, D . K., and D . J . Hinrichs. 1987. Adoptive transfer of immunity to Listeria monocytogenes. The influence of in vitro stimulation on lymphocyte subset requirements . J . Immunol. 139:2005-2009.
8. Cao, M., P . A . Kobel, M . M . Morshedi, M. F . W . Wu, C . Paddon, and J . D . Helmann.2002 . Defining the Bacillus subtilis ^W regulon: a comparative analysis of promoter consensus search, run-off transcription/macroarray analysis (ROMA), and transcriptional profiling approaches . J . Mol . Biol. 316:443-457.
9. Carmel-Harel, O., and G . Storz. 2000 . Roles of the glutathione- and thioredoxin-dependent reduction systems in the Escherichia coli and Saccharomyces cerevisiae responses to oxidative stress . Annu . Rev . Microbiol. 54:439-461.
10. Chan, P . F., S . J . Foster, E . Ingham, and M . O. Clements. 1998 . The Staphylococcus aureus alternative sigma factor ^B controls the environmental stress response but not starvation survival or pathogenicity in a mouse abscess model . J . Bacteriol. 180:6082-6089.
11. Cotter, P . D., C . G . Gahan, and C . Hill.2001 . A glutamate decarboxylase system protects Listeria monocytogenes in gastric fluid . Mol. Microbiol. 40:465-475.
12. Deora, R., T . Tseng, and T . K . Misra. 1997. Alternative transcription factor ^B of Staphylococcus aureus: characterization and role in transcription of the global regulatory locus sar . J. Bacteriol. 179:6355-6359.
13. Dramsi, S., P . Dehoux, M . Lebrun, P . L . Goossens, and P . Cossart.1997 . Identification of four new members of the internalin multigene family of Listeria monocytogenes EGD . Infect. Immun. 65:1615-1625.
14. Dramsi, S., C . Kocks, C . Forestier, and P . Cossart. 1993. Internalin-mediated invasion of epithelial cells by Listeria monocytogenes is regulated by the bacterial growth state, temperature and the pleiotropic activator PrfA . Mol. Microbiol. 9:931-941.
15. Dussurget, O., D . Cabanes, P . Dehoux, M . Lecuit, C . Buchrieser, P . Glaser, and P. Cossart. 2002 . Listeria monocytogenes bile salt hydrolase is a PrfA-regulated virulence factor involved in the intestinal and hepatic phases of listeriosis . Mol. Microbiol. 45:1095-1106.
16. Epstein, C . B., W . t . Hale, and R . A . Butow.2001 . Numerical methods for handling uncertainty in microarray data: an example analyzing perturbed mitochondrial function in yeast . Methods Cell Biol. 65:439-452.
17. Fang, F . C., S . J . Libby, N . A . Buchmeier, P . C . Loewen, J . Switala, J . Harwood, and D . G. Guiney. 1992 . The alternative sigma factor KatF (RpoS) regulates Salmonella virulence . Proc . Natl . Acad . Sci. USA 89:11978-11982.
18. Ferreira, A., C . P . O'Byrne, and K . J . Boor.2001 . Role of ^B in heat, ethanol, acid, and oxidative stress resistance and during carbon starvation in Listeria monocytogenes . Appl . Environ. Microbiol. 67:4454-4457.
19. Ferreira, A., D . Sue, C . P . O'Byrne, and K . J . Boor.2003 . Role of Listeria monocytogenes ^B in survival of lethal acidic conditions and in the acquired acid tolerance response . Appl . Environ. Microbiol. 69:2692-2698.
20. Flamm, R . K., D . J . Hinrichs, and M . F. Thomashow. 1984 . Introduction of pAMß1 into Listeria monocytogenes by conjugation and homology between native L . monocytogenes plasmids . Infect. Immun. 44:157-161.
21. Fouet, A., O . Namy, and G . Lambert. 2000 . Characterization of the operon encoding the alternative ^B factor from Bacillus anthracis and its role in virulence . J. Bacteriol. 182:5036-5045.
22. Fraser, K . R., D . Sue, M . Wiedmann, and K . J . Boor.2003 . The role of ^B in regulating the compatible solute uptake systems of Listeria monocytogenes: osmotic induction of opuC is ^B dependent.Appl . Environ . Microbiol. 69:2015-2022.
23. Freitag, N . E., and D . A . Portnoy. 1994. Dual promoters of the Listeria monocytogenes PrfA transcriptional activator appear essential in vitro but are redundant in vivo . Mol . Microbiol. 12:845-853.
24. Furrer, B., U . Candrian, C . Hoefelein, and J . Luethy. 1991. Detection and identification of Listeria monocytogenes in cooked sausage products and in milk by in vitro amplification of haemolysin gene fragments . J . Appl. Bacteriol. 70:372-379.
25. Gaidenko, T . A., and C . W . Price. 1998. General stress transcription factor ^B and sporulation transcription factor ^H each contribute to survival of Bacillus subtilis under extreme growth conditions . J . Bacteriol. 180:3730-3733.
26. Gardan, R., O . Duche, S . Leroy-Setrin, and J . Labadie.2003 . Role of ctc from Listeria monocytogenes in osmotolerance . Appl . Environ. Microbiol. 69:154-161.
27. Gertz, S., S . Engelmann, R . Schmid, A.-K . Ziebandt, K . Tischer, C . Scharf, J. Hacker, and M . Hecker. 2000 . Characterization of the ^B regulon in Staphylococcus aureus.J . Bacteriol. 182:6983-6991.
28. Glaser, P., L . Frangeul, C . Buchrieser, C . Rusniok, A . Amend, F . Baquero, P. Berche, H . Bloecker, P . Brandt, T . Chakraborty, A . Charbit, F. Chetouani, E . Couvé, A . d . Daruvar, P . Dehoux, E . Domann, G. Domínguez-Bernal, E . Duchaud, L . Durant, O . Dussurget, K.-D. Entian, H . Fsihi, F . G.-D . Portillo, P . Garrido, L . Gautier, W . Goebel, N . Gómez-López, T . Hain, J . Hauf, D . Jackson, L.-M . Jones, U . Kaerst, J . Kreft, M . Kuhn, F . Kunst, G . Kurapkat, E. Madueño, A . Maitournam, J . M . Vicente, E . Ng, H. Nedjari, G . Nordsiek, S . Novella, B . d . Pablos, J.-C . Pérez-Diaz, R . Purcell, B . Remmel, M . Rose, T . Schlueter, N . Simoes, A . Tierrez, J.-A . Vázquez-Boland, H . Voss, J . Wehland, and P . Cossart.2001 . C