This Article Abstract Full Text (PDF) Supplemental material Other Versions of this Article: AEM.01085-07v1 73/24/7967    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Hu, Y. Articles by Boor, K. J. Search for Related Content PubMed PubMed Citation Articles by Hu, Y. Articles by Boor, K. J. Agricola Articles by Hu, Y. Articles by Boor, K. J.

 Previous Article  |  Next Article 

Applied and Environmental Microbiology, December 2007, p. 7967-7980, Vol. 73, No. 24
0099-2240/07/$08.00+0     doi:10.1128/AEM.01085-07
Copyright © 2007, American Society for Microbiology. All Rights Reserved.

Phenotypic and Transcriptomic Analyses Demonstrate Interactions between the Transcriptional Regulators CtsR and Sigma B in Listeria monocytogenes ^,

Department of Food Science,¹ Department of Microbiology and Immunology, Cornell University, Ithaca, New York²

Received 15 May 2007/ Accepted 7 October 2007

	ABSTRACT

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Listeria monocytogenes ^B positively regulates the transcription of class II stress response genes; CtsR negatively regulates class III stress response genes. To identify interactions between these two stress response systems, we constructed L. monocytogenes ctsR and ctsR sigB strains, as well as a ctsR strain expressing ctsR in trans under the control of an IPTG (isopropyl-β-D-thiogalactopyranoside)-inducible promoter. These strains, along with a parent and a sigB strain, were assayed for motility, heat resistance, and invasion of human intestinal epithelial cells, as well as by whole-genome transcriptomic and quantitative real-time PCR analyses. Both ctsR and ctsR sigB strains had significantly higher thermotolerances than the parent strain; however, full heat sensitivity was restored to the ctsR strain when ctsR was expressed in trans. Although log-phase ctsR was not reduced in its ability to infect human intestinal cells, the ctsR sigB strain showed significantly lower invasion efficiency than either the parent strain or the sigB strain, indicating that interactions between CtsR and ^B contribute to invasiveness. Statistical analyses also confirmed interactions between the ctsR and the sigB null mutations in both heat resistance and invasion phenotypes. Microarray transcriptomic analyses and promoter searches identified (i) 42 CtsR-repressed genes, (ii) 22 genes with lower transcript levels in the ctsR strain, and (iii) at least 40 genes coregulated by both CtsR and ^B, including genes encoding proteins with confirmed or plausible roles in virulence and stress response. Our data demonstrate that interactions between CtsR and ^B play an important role in L. monocytogenes stress resistance and virulence.

	INTRODUCTION

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Listeria monocytogenes is a gram-positive facultative intracellular pathogen that can cause severe invasive disease in humans, as well as in a number of different animal species. The vast majority of human listeriosis cases are caused by consumption of contaminated food products (44). The capacity of L. monocytogenes to survive and multiply under a wide range of environmental-stress conditions appears to be critical for food-borne transmission of the pathogen (18). Accordingly, a number of transcriptional regulators important for stress response and virulence gene expression have been identified in the organism (25, 40, 49). Positive regulatory factor A (PrfA), which activates the transcription of several L. monocytogenes virulence genes, was the first transcriptional regulator identified in the bacterium (40). Subsequently, a number of other L. monocytogenes transcriptional regulators have been identified, including four alternative sigma factors (^C appears to be present only in L. monocytogenes strains classified in lineage II, however) (31, 68), two negative regulators (HrcA and CtsR) involved in the regulation of heat shock genes (25, 49), and a number of two-component regulatory systems (64), as well as other regulators (e.g., CodY [2]). While many of these regulators appear to be important predominantly for transcription of stress response genes, null mutations in some stress response regulators have also been shown to result in reduced virulence or virulence-associated characteristics (6), thus providing evidence of mechanistic links between stress response and virulence in L. monocytogenes (29).

The stress-responsive alternative sigma factor ^B appears to have a central role in coordinating L. monocytogenes stress response and virulence gene expression (30, 32, 43, 48). ^B directly regulates the transcription of a large regulon in L. monocytogenes (30) by associating with the RNA polymerase core enzyme, thereby allowing the initiation of transcription of genes preceded by ^B-dependent promoters. L. monocytogenes ^B contributes to survival under a variety of conditions, including acid and oxidative stresses and during carbon starvation (1, 18, 47). In addition to regulating the expression of stress response genes, ^B also activates the transcription of virulence genes, such as inlAB and bsh (30). Some virulence genes (e.g., inlA and inlB) are coregulated by both PrfA and ^B (30, 32, 43, 46). Further, ^B directly regulates the transcription of prfA (48, 56, 57). As both ^B and PrfA also autoregulate their own transcription (45, 56), the regulatory network contributing to virulence gene expression involves at least two transcriptional regulators. While initial characterization of this regulatory network, including through array-based characterization of the ^B and PrfA regulons, has been performed (30, 32, 46), our understanding of the contributions of other regulators to this and other stress response and virulence gene transcription networks is limited. Initial evidence from our group suggested interactions between the transcriptional regulators CtsR and ^B in L. monocytogenes. Specifically, the CtsR-dependent clpC showed ^B-dependent transcription under energy stress, and a putative ^B-dependent promoter was identified upstream of the L. monocytogenes ctsR-mcsA-mcsB-clpC operon (8). In addition, clpC and clpP are regulated by both ^B and CtsR in Bacillus subtilis (22, 37), a close relative of L. monocytogenes.

While ^B is recognized as a positive regulator of transcription, CtsR (for class three stress gene repressor) is a transcriptional repressor. CtsR is active as a dimer and has three different functional domains, including a helix-turn-helix DNA-binding domain, a dimerization domain, and a putative heat-sensing domain (12). In L. monocytogenes, CtsR negatively regulates clp genes (clpP, clpE, clpB, and clpC) at the transcriptional level by binding directly to a heptanucleotide repeat sequence (A/GGTCAAANANA/GGTCAAA) (5, 49, 51); clp genes encode general stress proteins, with some genes possibly contributing to L. monocytogenes virulence (50). Characterization of L. monocytogenes strains with naturally occurring mutations in ctsR showed increased resistance of these strains to heat, H₂O₂, and high pressure, consistent with a negative-regulator role for CtsR (27-29) and suggesting that CtsR regulates the transcription of genes important for stress resistance. A naturally occurring ctsR mutant also showed decreased resistance to nisin (27). CtsR-dependent transcription of the heat shock gene clpB was reported in a ctsR null mutant (5), further confirming the importance of CtsR for heat shock. Characterization of another ctsR null mutant also showed reduced growth in tissue culture cells, suggesting CtsR contributions to virulence (6).

To examine the contributions of CtsR to stress response and virulence and to explore interactions between CtsR and ^B, we created three L. monocytogenes strains, including a ctsR strain, a ctsR sigB strain, and a ctsR strain with ctsR fused to an isopropyl-β-D-thiogalactopyranoside (IPTG)-inducible promoter and integrated at a tRNA^Arg locus. These strains, in combination with a sigB strain (63) and the parent strain, were used for phenotypic and microarray-based analyses.

	MATERIALS AND METHODS

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Bacterial strains and growth conditions.
L. monocytogenes 10403S and isogenic mutants of this parent strain were used in this study (Table 1). Stock cultures were stored at –80°C in brain heart infusion (BHI) (Difco, Sparks, MD) broth supplemented with 15% glycerol and streaked onto BHI agar plates prior to each experiment. For most experiments, overnight bacterial cultures were grown in BHI broth at 37°C with shaking (250 rpm), followed by inoculation into 5 ml of BHI broth (1:100 dilution) and growth at 37°C with shaking to an optical density at 600 nm (OD₆₀₀) of 0.4, followed by another 1:100 dilution into BHI broth with subsequent growth with shaking at 37°C to an OD₆₀₀ of 0.4 (representing log phase) unless otherwise specified. Multiple serial passages were used to generate cultures comprised of synchronized log-phase bacterial cells.

Construction of ctsR null mutants.

Swarming behavior.
To evaluate swarming behavior, single colonies of the parent or mutant strains were used to stab inoculate tryptic soy broth agar plates containing 0.4% agar, which were then incubated at room temperature for 48 h. Swarming ability was assessed by measuring the radius of the colony, which was then normalized to the colony radius for the parent strain (L. monocytogenes 10403S). Three independent assays were performed for each strain tested. An isogenic L. monocytogenes flaA mutant (55) was included as a negative control in each assay.

Heat survival experiments.
Heat survival experiments were conducted to evaluate the heat resistance of the parent strain, 10403S, as well as ctsR, sigB, ctsR sigB, and ctsR tRNA^Arg::pLIV2 ctsR-mcsA strains. The parent strain and the ctsR tRNA^Arg::pLIV2 ctsR-mcsA strain were also grown with and without the addition of IPTG. The heat survival experiments were performed using a continuous microflow apparatus, as previously described (41), to allow reproducible short-time exposure to heat stress. Briefly, log-phase cells (OD₆₀₀ = 0.4) were pumped from a 50-ml side arm flask through a microcoil that was submerged in an insulated 72°C oil bath (Virtis Research Equipment, Gardiner, NY). The cells were kept at 72°C for either 4 or 8 s, followed by collection of the heat-treated cells using a sterile vial. Bacterial numbers were determined by spread plating the bacteria on BHI agar prior to and after heat treatment. Heat survival was expressed as a log reduction value, which was calculated by subtracting the bacterial numbers (in log CFU/ml) after heat treatment from the bacterial numbers (in log CFU/ml) before heat treatment. Three independent experiments were performed.

Invasion assays.
Invasion efficiency in Caco-2 cells was determined for the parent strain, as well as for ctsR, sigB, and ctsR sigB strains that were either (i) grown to log phase (OD = 0.4) at 37°C or (ii) grown to log phase, followed by exposure to BHI with 0.3 M NaCl for 10 min at 37°C. Exposure to 0.3 M NaCl was performed to induce ^B activity (61), thus enabling enhanced detection of CtsR-^B interactions. Invasion assays were also performed with L. monocytogenes strains grown to early stationary phase (i.e., growth to an OD₆₀₀ of 1, followed by an additional 3 h of incubation) at 30°C, as L. monocytogenes grown at that temperature shows increased motility compared to bacteria grown at 37°C. Although many L. monocytogenes isolates are nonmotile at 37°C and motile at 30°C, strain 10403S is motile at 37°C, although at a reduced level compared to cells grown at 30°C (24, 62).

Invasion assays in Caco-2 cells (ATCC HTB-37) were performed as previously described (53). For all invasion assays, confluent Caco-2 monolayers grown in 24-well plates were inoculated with 10 µl L. monocytogenes cells (three wells/strain). After incubation for 30 min, the Caco-2 cells were washed three times with phosphate-buffered saline, and at 45 min postinoculation, fresh medium containing gentamicin (150 µg/ml) was added to kill extracellular bacteria. At 90 min postinoculation, the Caco-2 cells were washed three times with phosphate-buffered saline and then lysed by the addition of ice-cold sterile distilled water, followed by vigorous pipetting. Intracellular L. monocytogenes numbers were determined by spiral plating lysed Caco-2 cell suspensions on BHI agar using an Autoplate 4000 spiral plater (Spiral Biotech Inc., Norwood, MA). The invasion efficiency was calculated as the number of intracellular bacteria recovered (in CFU) relative to the bacterial numbers (in CFU) used for the inoculation; invasion efficiencies for mutant strains were then normalized to the 10403S parent strain invasion efficiency, which was set as 100%. Three independent invasion assays were performed for each L. monocytogenes strain tested.

TaqMan qRT-PCR.
RNA isolation for quantitative real-time (qRT) PCR was performed as previously described (42). qRT-PCR with previously described primers and probes (32, 34, 57, 61) was used to measure transcript levels for inlA, clpC, gadA, prfA, and plcA, as well as for two housekeeping genes (rpoB and gap). RNA was isolated from the parent strain, and selected mutants (ctsR, ctsR sigB, and sigB) that were either (i) grown to log phase (OD₆₀₀ = 0.4) at 37°C or (ii) grown to log phase, followed by exposure to BHI with 0.3 M NaCl for 10 min at 37°C. qRT-PCR was performed using the ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA), as detailed previously (42, 61). Reverse transcriptase negative control reactions, DNA standard curves, and analysis of qRT-PCR were performed essentially as described previously (42). Absolute cDNA copy numbers, calculated based on DNA standard curves, reflect mRNA levels for each target gene present in each RNA sample. Relative cDNA copy numbers were calculated as log cDNA copy numbers normalized to the geometric mean of cDNA copy numbers for the housekeeping genes rpoB and gap {i.e., log₁₀ target gene – [(log₁₀ rpoB + log₁₀ gap)/2]}). qRT-PCR was repeated three times using three independent RNA isolations from cells grown on three different days (42).

In addition, qRT-PCR primers and probes were designed for clpB and lmo1138 (see Table S1 in the supplemental material). qRT-PCR with these two primer/probe sets was performed using RNA isolated from the L. monocytogenes parent and ctsR strains (grown to log phase) to confirm CtsR-dependent repression of the two genes.

Microarray-based transcriptomic analyses.
To identify CtsR-dependent genes, two separate sets of microarray experiments were performed, including (i) one set of experiments comparing transcript levels between the parent and the ctsR strains, both grown to log phase, and (ii) one set of experiments comparing transcript levels between the ctsR tRNA^Arg::pLIV2 ctsR-mcsA strain (grown with 0.5 mM IPTG) and the ctsR strain, both grown to log phase. Microarray construction, RNA isolation and purification, cDNA labeling, and hybridization were performed as previously detailed (4). Briefly, RNA for microarrays was isolated using the RNAprotect bacterial reagent and the RNeasy Midi kit (Qiagen, Valencia, CA). DNase treatment of RNA was performed essentially as described previously (30), except that 40 U of DNase was used. The isolated RNA (in RNase-free water) was quantified and checked for purity using OD₂₆₀ and OD₂₈₀ measurements performed on a Nanodrop spectrophotometer (Nanodrop Technology Inc., Wilmington, DE). Agarose gel electrophoresis was used to verify RNA integrity.

cDNA was synthesized and differentially labeled using the SuperScript Plus Indirect cDNA Labeling System (Invitrogen, Carlsbad, CA). cDNA was generated from 10 µg of RNA using random primers in an overnight reverse transcription reaction at 42°C. The cDNA was purified using a Qiagen PCR purification kit prior to indirect labeling with Alexa Fluor 555 or Alexa Fluor 647 (performed overnight at room temperature). Labeled cDNA was purified using a Qiagen PCR purification kit to remove any unincorporated dye, and the labeled cDNA was quantified and checked for purity using OD₂₆₀ and OD₂₈₀ measurements.

Microarrays were constructed using 70-mer oligonucleotides targeting 2,857 L. monocytogenes ORFs (Qiagen Operon Array-Ready Oligo Sets) identified in the annotated genome sequence for L. monocytogenes EGD-e (23). Oligonucleotides (70-mer) targeting five Saccharomyces cerevisiae ORFS (act1, mfa1, mfa2, ras1, and ste3) were included on the microarray to serve as nonhybridizing controls, as described previously (67). Prior to hybridization, the slides were blocked and washed essentially as described by Chan et al. (4).

For each microarray, two target cDNAs (e.g., cDNAs from the parent and the ctsR strains) were combined in one tube, dried, and then resuspended in hybridization buffer containing sodium dodecyl sulfide (SDS), SSC (1x SSC is 0.15 M NaCl plus 0.015 M sodium citrate), salmon sperm DNA, dithiothreitol, and formamide. The combined cDNA target (in a 50-µl volume) was overlaid onto the microarray slides using mSeries LifterSlips (Erie Scientific, Portsmouth, NH). Following overnight hybridization at 42°C, the slides were washed in 2x SSC plus 0.1% SDS for 5 min at 42°C, followed by room temperature washes in 2x SSC plus 0.1% SDS, 2x SSC, and 0.2x SSC for 5 min each. The slides were centrifuged to dry and scanned with a GenePix 4000B Scanner (Molecular Devices, Sunnyvale, CA) at the Microarray Core Facility (Ithaca, NY).

Scanned microarray images were gridded using GenePix Pro 6.0 software. Raw image analysis data were preprocessed, and significant differences in gene expression patterns between strains were determined using LIMMA software (58) from R/BioConductor (21). Following background correction by the normexp method, within-array normalization (print-tip loess, a locally weighted linear regression model) and between-array normalization (scale) were used to correct for spatial and intensity biases and to make the results comparable across arrays. The LIMMA package was also used for differential expression analysis (59) to calculate moderated t and B statistics and P values (adjusted for multiple comparisons by controlling for the false-discovery rate). Genes with an adjusted P value of <0.05 were considered statistically significant, and a change of 1.5-fold was used as a cutoff for identification of differentially expressed genes. Genes that showed significantly different transcript levels either (i) between the parent and the ctsR strains or (ii) between the ctsR tRNA^Arg::pLIV2 ctsR-mcsA and the ctsR strains were considered putative CtsR-dependent genes.

HMM searches.
Potential CtsR-binding sites were determined using hidden Markov model (HMM) searches as previously described (30). The HMM training alignments included 33 CtsR-binding operators previously identified in gram-positive bacteria (11). The HMM model was searched against the template and nontemplate sequences for the L. monocytogenes EGD-e genome. Outputs were filtered, and only hits within 300 bp upstream of a start codon for an ORF, as annotated by ListiList (http://genolist.pasteur.fr/ListiList), and with an E value of 0.01 were considered meaningful.

Statistical analysis.
All statistical analyses were performed using SAS (SAS online Doc8, version 8; SAS, Inc., Cary, NC). qRT-PCR and heat survival were analyzed using one-way or two-way analysis of variance. For both swarming assays and invasion assays, the phenotypic results for the mutant strains (i.e., swarming ability and invasion efficiency) were normalized to the parent strain's swarming ability or invasion efficiency (which were set at 100%); comparisons between mutant and parent invasion efficiencies were performed using one-sample t tests; Bonferroni corrections to the P values were used to adjust for multiple comparisons. To test whether interactions caused by the deletion of two genes (i.e., sigB and ctsR) had a significant effect on a given phenotype (e.g., invasion efficiency), a linear model with two factors (gene 1 [ctsR] presence/absence and gene 2 [sigB] presence/absence) was used. For all tests, statistical significance was established at a P value of <0.05; significant P values are reported as the actual value, unless P was <0.001.

Microarray data accession number.
Raw and normalized microarray data in MIAME format are available at the NCBI Gene Expression Omnibus (GEO) data repository (15) under accession number GSE7514.

	RESULTS

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Loss of CtsR results in reduced swarming ability.
Since an L. monocytogenes strain with a naturally occurring 3-bp deletion in ctsR was reported as having reduced flaA transcription and protein expression (29), we examined the swarming ability of our ctsR mutant strain relative to that of the sigB and ctsR sigB strains at room temperature (Fig. 1). Both the ctsR and the ctsR sigB strains showed similar swarming, which was significantly lower than the swarming ability observed for the parent strain but also significantly higher than the swarming ability of the flaA strain (Fig. 1). The sigB strain showed no evidence of reduced swarming ability (Fig. 1).

View larger version (9K): [in this window] [in a new window]   FIG. 1. Swarming abilities of the L. monocytogenes 10403S parent strain and flaA, ctsR, sigB, and ctsR sigB strains. flaA was included as a negative control strain. Swarming ability was assessed by stabbing inocula into semisolid agar and then measuring colony growth radii after incubation at room temperature for 48 h; the growth radius for each mutant strain was normalized to the growth radius of the parent strain (10403S), which was set at 100%. An asterisk indicates that the growth radius of that mutant was significantly smaller than the growth radius for the parent strain (as determined by a one-sample t test; P values were adjusted for multiple testing using a Bonferroni correction). The ctsR and the ctsR sigB strains showed significantly smaller radii than the sigB strain and significantly larger radii than the flaA strain. The data shown represent the averages of three independent experiments; the error bars indicate standard deviations; no error bar is shown for the parent strain, since the parent strain radius was set at 100% for each experiment.

View larger version (15K): [in this window] [in a new window]   FIG. 2. Log reduction after exposure of the L. monocytogenes parent strain (10403S) and ctsR, sigB, and ctsR sigB strains to 72°C for 4 s (A) or 8 s (B) and ctsR tRNAArg::pLIV2 ctsR-mcsA strain grown with (+IPTG) and without (–IPTG) IPTG. The strains were grown to log phase before the heat survival experiments; survival experiments for the parent strain were performed with bacteria grown with (+IPTG) and without (–IPTG) IPTG to ensure that the addition of IPTG to the growth media did not affect heat survival. Strains with higher log reductions showed greater sensitivity to heat (e.g., L. monocytogenes 10403S showed an 8.2-log-unit reduction after 8 s of exposure to 72°C, while L. monocytogenes ctsR showed only a 4.2-log-unit reduction; the ctsR strain is thus less heat sensitive than the parent strain, 10403S). Log reduction was calculated by subtracting the bacterial numbers (in log CFU/ml) after heat treatment from the bacterial numbers (in log CFU/ml) before heat treatment. The data shown represent the averages of three independent experiments; the error bars indicate standard deviations. Tukey's multiple-comparison procedure was used to determine whether heat survival differed between specific strains; the bars labeled with different letters indicate log reduction values that differed significantly (P < 0.05), while the bars labeled with identical letters indicate log reduction values that did not differ significantly. Statistical analyses of log reduction values for the parent strain (10403S–IPTG) and the ctsR, sigB, and ctsR sigB strains using a linear model indicated a significant effect on heat survival of the ctsR deletion (P < 0.001 for survival after both 4 and 8 s), the sigB deletion (P = 0.0206 and P < 0.001 for survival after 4 and 8 s, respectively), and the interaction between the ctsR and sigB deletions (P = 0.0023 for survival after 8 s).

CtsR and ^B contribute to invasion efficiency of L. monocytogenes grown at 37°C.
Invasion efficiencies for Caco-2 cells, a human intestinal epithelial cell line, were initially determined for the L. monocytogenes parent strain, as well as for ctsR, sigB, and ctsR sigB strains (Fig. 3). For log-phase cells not exposed to salt stress, the parent strain and the ctsR strain had similar invasion efficiencies (Fig. 3), while the sigB strain showed significantly lower invasion efficiency than both the parent strain and the ctsR strain (Fig. 3). Interestingly, the ctsR sigB strain showed the lowest overall invasion efficiency, which was significantly lower than the invasion efficiency for the sigB strain (Fig. 3). The observation that the ctsR sigB strain had lower invasion efficiency than the sigB strain, despite the fact that the ctsR strain showed no reduced invasion efficiency (compared to the parent strain), indicated that interactions between the ctsR and sigB deletions may affect the invasion phenotype. Statistical analysis (using a linear model, as detailed in Materials and Methods) of the log-transformed invasion efficiencies confirmed this and showed a highly significant effect of the interactions between the ctsR and sigB deletions on the invasion efficiencies of log-phase cells (P = 0.0019).

View larger version (18K): [in this window] [in a new window]   FIG. 3. Invasion efficiency in the human intestinal epithelial cell line Caco-2 of the L. monocytogenes parent strain (10403S) and the ctsR, sigB, and ctsR sigB strains that were either (i) grown to log phase in BHI at 37°C (BHI) or (ii) grown to log phase, followed by exposure to 0.3 M NaCl for 10 min at 37°C (BHI-NaCl). Invasion efficiency was calculated as the number of intracellular bacteria recovered (in CFU) relative to the bacterial numbers (in CFU) used for inoculation; invasion efficiencies for mutant strains were then normalized to the parent strain invasion efficiency, which was set as 100%. The data shown represent the averages of three independent experiments; the error bars indicate standard deviations; no error bar is shown for the parent strain, since its invasion efficiency was set at 100% for each experiment. An asterisk indicates that the invasion efficiency of that mutant was significantly lower than the invasion efficiency for the parent strain (as determined by a one-sample t test; P values were adjusted for multiple testing using a Bonferroni correction). For bacteria grown in BHI, invasion efficiencies were significantly lower for (i) the sigB strain compared to the ctsR strain and (ii) the ctsR sigB strain compared to the ctsR or the sigB strain. Analyses of log-transformed nonnormalized invasion efficiency data for the parent strain (10403S) and the ctsR, sigB, and ctsR sigB strains using a linear model indicated a significant effect on invasion efficiency of the ctsR deletion (P < 0.001 and P = 0.0124 for BHI and BHI-NaCl, respectively), the sigB deletion (P = 0.0044 and P = 0.0172 for BHI and BHI-NaCl, respectively), and the interaction between the ctsR and sigB deletions (P = 0.0019 for BHI).

^B, and possibly CtsR, contributes to invasion efficiency of L. monocytogenes grown at 30°C.
Additional invasion experiments in Caco-2 cells were performed using the L. monocytogenes parent strain, as well as flaA, sigB, ctsR, and ctsR sigB strains, grown to stationary phase at 30°C (Fig. 4). Invasion assays with bacteria grown at 30°C were performed, in addition to the invasion assays with bacteria grown at 37°C, as L. monocytogenes strain 10403S shows increased motility when grown at 30°C and as our initial experiments (see above) showed that deletion of ctsR leads to reduced motility of L. monocytogenes 10403S. In these invasion assays with L. monocytogenes grown at 30°C, the flaA strain (included as a control) showed significantly lower invasion efficiency than the parent strain (Fig. 4), consistent with previous observations of the contributions of flagella to invasion (14, 55). While the ctsR strain showed numerically lower invasion efficiency (59.6%) than the parent strain (100%), these differences were only borderline significant (P = 0.0253 without a Bonferroni correction; P = 0.1 with a Bonferroni correction) (Fig. 4). The sigB strain showed reduced invasion efficiency compared to the parent strain, consistent with previous reports (20, 33), and the ctsR sigB mutant showed significantly lower invasion efficiency than the ctsR strain (P = 0.0005; t test) and numerically, but not significantly, lower invasion efficiency than the sigB strain (Fig. 4). Analysis of the log-transformed invasion efficiencies (using a linear model, as described in Materials and Methods) showed significant effects of both the ctsR and sigB deletions on the invasion efficiencies, further confirming the contributions of both CtsR and ^B to L. monocytogenes invasion efficiency.

View larger version (8K): [in this window] [in a new window]   FIG. 4. Invasion efficiencies in Caco-2 cells of the L. monocytogenes parent strain (10403S) and flaA, ctsR, sigB, and ctsR sigB strains that were grown to early stationary phase in BHI at 30°C. Invasion efficiency was calculated as the number of intracellular bacteria recovered (in CFU) relative to the bacterial numbers (in CFU) used for the inoculation; invasion efficiencies for mutant strains were then normalized to the parent strain invasion efficiency (which was set as 100%). The data shown represent the averages of three independent experiments; the error bars indicate standard deviations; no error bar is shown for the parent strain, since its invasion efficiency was set at 100% for each experiment. An asterisk indicates that the invasion efficiency of that mutant was significantly lower than the invasion efficiency for the parent strain (as determined by a one-sample t test; P values were adjusted for multiple testing using a Bonferroni correction). Statistical analyses of log-transformed nonnormalized invasion efficiency data for the parent strain (10403S) and the ctsR, sigB, and ctsR sigB strains using a linear model indicated significant effects on invasion efficiency of the ctsR deletion (P = 0.0306) and of the sigB deletion (P < 0.001).

qRT-PCR indicates CtsR-dependent clpC and ^B-dependent prfA, plcA, inlA, and gadA transcription but no evidence for CtsR and ^B interactions contributing to transcription of these genes.

View larger version (20K): [in this window] [in a new window]   FIG. 5. Transcript levels for clpC (A), gadA (B), prfA (C), plcA (D), and inlA (E) in the parent strain (10403S) and the ctsR, sigB, and ctsR sigB strains. Bacteria were either (i) grown to log phase in BHI at 37°C (BHI) or (ii) grown to log phase, followed by exposure to BHI with 0.3 M NaCl for 10 min at 37°C (BHI-NaCl). Transcript levels are expressed as log cDNA copy numbers normalized to the geometric mean of cDNA copy numbers for the housekeeping genes rpoB and gap {i.e., log10 target gene – [(log10 rpoB + log10 gap)/2]}; indicated as "log10 normalized copy no." on the y axes). The values shown represent the averages of qRT-PCR assays performed on three independent RNA collections; the error bars show standard deviations. NS indicates that neither the ctsR deletion, the sigB deletion, nor the interaction between the ctsR and sigB deletions showed a significant effect using a linear model for statistical analyses. If the linear model showed a significant effect of either the ctsR deletion or the sigB deletion on transcript levels for a given gene for cells grown under a given condition (BHI or BHI-NaCl), the respective P value is given on the graph (see Table S2 in the supplemental material for all results of these statistical analyses). Tukey's multiple-comparison procedure was used to determine whether transcript levels for a given gene differed between specific strains; the bars labeled with different letters (a and b) indicate transcript levels that differed significantly (P < 0.05), while the bars labeled with identical letters indicate transcript levels that did not differ significantly; if no letters are indicated, no significant differences in the individual comparisons were observed between strains. The linear model may reveal significant effects of a gene deletion, even if transcript levels for the four strains do not differ significantly in the individual comparisons, as the linear model accounts for the effects in the ctsR and sigB single mutants, as well as in the ctsR sigB double mutant.

View larger version (26K): [in this window] [in a new window]   FIG. 7. Genes and operons regulated by both CtsR and B. For both CtsR and B, any genes that showed significant and >1.5-fold differences in transcript levels were considered to be affected by a given regulator. (A) Venn diagram summarizing genes and operons that showed transcript levels that were found in microarray experiments to be significantly affected by either the ctsR deletion (this study) or the sigB deletion (55a) (microarray data are available under GEO database accession number GSE7492); and indicate genes that are up- or down-regulated by a given regulator. Genes preceded by a B consensus promoter or a consensus CtsR binding site are marked with SB or CT, respectively. (B) Selected genes and operons regulated by both CtsR and B. For genes that appear to be directly regulated by B and/or CtsR, putative B-dependent promoters and CtsR binding sites are indicated; no putative B-dependent promoters or CtsR binding sites are indicated for the gadC-gadB and the lmo0683-to-lmo0688 operons, which appear to be indirectly affected by both of these regulators. Putative terminators (shown as stem-loop structures) are also indicated. Transcript ratios determined in the microarray experiments are indicated below a given gene; differences (n-fold) in transcript levels between the parent strain (WT) and the sigB strain were determined either in log-phase cells exposed to salt [WT-sigB(NaCl)] or in stationary-phase cells [WT-sigB (STAT)]; positive numbers indicate that transcript levels were higher in the parent strain (indicating positive regulation by B), while negative numbers indicate that transcript levels were lower in the parent strain (indicating negative regulation by B). For CtsR, differences (n-fold) in transcript levels were determined between the parent strain (WT) and the ctsR strain grown to log phase (WT-ctsR) and between the ctsR tRNAArg::pLIV2 ctsR-mcsA and the ctsR strain grown to log phase in the presence of IPTG (ictsR-ctsR); negative numbers indicate that transcript levels were lower in the strain expressing CtsR (indicating negative regulation by CtsR), while positive numbers indicate that transcript levels were higher in the strain expressing CtsR (indicating [indirect] positive regulation by CtsR). Differences (n-fold) with adjusted P values of <0.05 are in boldface. The lmo0683 to lmo0688 genes shown represent part of a larger operon (lmo0675 to lmo0689).

Whole-genome microarray analysis identified 64 CtsR-dependent genes.
As phenotypic experiments provided evidence that interactions between CtsR and ^B contribute to both heat resistance and invasion efficiency, we used microarray experiments to characterize the L. monocytogenes CtsR regulon. In conjunction with recently completed characterization of the ^B regulon (55a) (microarray data are available under GEO database accession number GSE7492) with the same sigB strain used in all the experiments reported here, an additional goal was to identify genes coregulated by CtsR and ^B. Microarray experiments to identify the CtsR regulon used comparisons of transcript levels between (i) the parent strain and the ctsR strain and (ii) the ctsR tRNA^Arg::pLIV2 ctsR-mcsA strain expressing the ctsR-mcsA operon in trans (grown in the presence of IPTG) and the ctsR strain. Genes were classified as CtsR dependent if they showed 1.5-fold differences in transcript levels with an adjusted P value of <0.05 in either of the two comparisons. Using these criteria, we identified a total of 64 CtsR-dependent genes, including 42 genes negatively regulated by CtsR (i.e., genes that showed higher transcript levels in the ctsR strain) (Table 3) and 22 genes that showed lower transcript levels in the ctsR strain (Table 4). While all genes that showed lower transcript levels in the ctsR strain (indicating positive regulation by CtsR) are likely indirectly regulated by CtsR (as CtsR is a negative regulator), genes that showed higher transcript levels in the ctsR strain could be directly or indirectly regulated by CtsR. A total of at least 10 genes appear to be directly regulated by CtsR, as they show higher transcript levels in the CtsR strain, as well as putative or confirmed CtsR-binding sites (Table 5) upstream of a gene or operon. Consistent with a previous report (49), the ctsR-mscA-mscB-clpC operon (lmo0229 to lmo0232) was confirmed as directly CtsR repressed. In addition to this operon, four confirmed or putative heat shock genes were also directly repressed by CtsR, including the clpB-lmo2205 operon and clpP, which all encode traditional class III heat shock proteins, as well as lmo1138, which encodes a protein similar to ClpP (65.4% amino acid similarity to L. monocytogenes EGD-e clpP). In addition, the tatAC operon, which encodes a putative twin argenine translocase secretion system, was found to be directly repressed by CtsR; interestingly, tatAC appears to be absent in L. monocytogenes serotype 4b strains (13). Surprisingly, clpE, which previously has been reported to be CtsR repressed (49), was not found to be CtsR dependent in our microarray studies (clpE showed transcript level ratios of –1.34 and –1.24 with an adjusted P value of >0.05). CtsR-dependent repression of clpB and lmo1138 was also confirmed by qRT-PCR (Fig. 6).

View this table: [in this window] [in a new window]   TABLE 4. Genes identified with lower transcript levels in the ctsR strain by microarray analysisa

View larger version (9K): [in this window] [in a new window]   FIG. 6. Transcript levels for clpB and lmo1138 in the parent strain (10403S) and the ctsR strain grown to log phase in BHI at 37°C. Transcript levels are expressed as log cDNA copy numbers normalized to the geometric mean of cDNA copy numbers for the housekeeping genes rpoB and gap {i.e., log10 target gene – [(log10 rpoB + log10 gap)/2]}; indicated as "log10 normalized copy no." on the y axes). The values shown represent the averages of qRT-PCR assays performed on three independent RNA collections; the error bars show standard deviations. Within a given gene, bars labeled with different letters (a and b) indicate transcript levels that differed significantly (P < 0.05; t test).

The transcription of 40 genes is affected by both CtsR and ^B, including a number of heat shock genes that are directly regulated by both CtsR and ^B.
Comparison of the CtsR regulon with the L. monocytogenes ^B regulon (55a) (microarray data are available under GEO database accession number GSE7492) identified 40 of the CtsR-dependent genes reported here as also differentially regulated by ^B (Fig. 7). Nine coregulated genes were positively regulated by ^B and negatively regulated by CtsR, including seven genes that appear to be directly regulated by both regulators, as supported by identification of both ^B-dependent promoter sequences and CtsR binding sites upstream of the respective genes. Specifically, CtsR and ^B coregulate three genes and operons encoding heat shock proteins, including clpP and the clpC and clpB operons (Fig. 7 and 8). Genes and operons indirectly regulated by ^B and CtsR include a number of genes encoding proteins with confirmed or plausible roles in virulence and stress response. For example, the gadCB operon, which encodes a glutamate transporter and decarboxylase important for acid resistance (9), was found to be positively regulated by CtsR and negatively regulated by ^B (Fig. 7 and 8), and the tatAC operon was found to be negatively regulated by CtsR and negatively regulated by ^B (Fig. 7). Additionally, a number of genes positively regulated by CtsR and negatively regulated by ^B were located in the lmo0675-to-lmo0689 flagellar operon (e.g., cheR and motA) (Fig. 7); selected genes (i.e., lmo0699 and lmo716) in a downstream flagellar operon (lmo0691 to lmo0718) were also found to be positively regulated by CtsR and negatively regulated by ^B.

View larger version (19K): [in this window] [in a new window]   FIG. 8. Partial CtsR, B, and PrfA interaction network. The network is based on CtsR microarray data presented here, B microarray data (55a) (microarray data are available under GEO database accession number GSE7492), and PrfA macroarray data (46). Different symbols were used to represent stress response genes (squares), virulence genes (diamonds), motility genes (circles), stress response genes potentially involved in virulence (octagons), genes of unknown or other function (hexagons), and the genes encoding CtsR, B, and PrfA (orange rounded squares). Solid lines indicate direct regulation of the gene by a given regulator as determined by the presence of a CtsR operator site, B promoter, or PrfA box; dashed lines represent indirect regulation. Target arrows () indicate positive regulation by a given regulator (as indicated by higher transcript levels in the parent strain than in the mutant strain); target stops () indicate negative regulation by a given regulator (as indicated by lower transcript levels in the parent strain than in the mutant strain). Loops indicate autoregulation. Color coding was used to identify genes solely regulated by a given regulator (yellow), dually regulated (green), or regulated by all three regulators (blue). Genes arranged in vertical columns represent operons. The red arrow targeting CtsR indicates posttranslational regulation of CtsR by McsA, McsB, and ClpC, based on evidence reported for B. subtilis (36, 38).

	DISCUSSION

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A total of 64 genes are regulated by CtsR, either directly or indirectly, including genes important for motility, stress response, and virulence.
Transcriptomic analyses of the ctsR null mutant identified 64 genes as regulated by CtsR, including 10 genes directly repressed by CtsR. While previous studies generally identified small numbers of genes that are directly regulated by CtsR (49), we show that, in addition to genes that are directly repressed by CtsR, the protein also indirectly regulates the transcription of a number of L. monocytogenes genes. Identification of these newly recognized CtsR-dependent genes reveals some genetic mechanisms responsible for the phenotypes observed both here and in previous studies of L. monocytogenes ctsR strains (6, 27-29), including increased resistance to heat, high pressure, and oxidative stresses, as well as reduced motility, virulence, and tissue culture pathogenicity. Correlations between the CtsR-dependent genes identified here and the phenotypic characteristics of ctsR null mutants are further detailed below.

Increased heat resistance of L. monocytogenes ctsR can be explained by the observation that CtsR represses the transcription of a number of genes encoding class III heat shock stress response proteins, including clpC, clpP, and clpB (5, 19, 50), which were all identified as CtsR repressed in our microarray experiments. We also identified lmo1138 and the tatAC operon, both of which are directly repressed by CtsR, as novel members of the CtsR regulon. lmo1138 encodes a protein with high homology to other proteins in the large family of ClpP serine proteases, which are critical for the degradation of misfolded proteins. tatAC encodes two minimal translocases responsible for the twin arginine translocation (Tat) pathway, which is responsible for the export of fully folded proteins across the cytoplasmic bacterial membrane (13). Identification of these novel CtsR-repressed genes further confirms the importance of CtsR in regulating the transcription of genes that aid L. monocytogenes in response to stress conditions that lead to protein denaturation (e.g., heat stress and high pressure) and thus contributes to a better understanding of stress resistance in this important food-borne pathogen.

Interestingly, a number of genes encoding proteins with confirmed or plausible roles in virulence and stress response were also found to be regulated by CtsR. For example, CtsR was found to repress genes (i.e., qoxB and tpx) that may contribute to L. monocytogenes oxidative-stress response, consistent with the observation that L. monocytogenes Scott A strains bearing naturally occurring mutations in ctsR showed increased resistance to H₂O₂ (27, 29). The qoxABCD operon encodes a quinol oxidase, which is important for oxidative-stress response, as supported by observations that a B. subtilis strain with a mutation in this operon shows reduced aerobic growth (65). tpx encodes a thioperoxidase;