|Home | About | Journals | Submit | Contact Us | Français|
CsrRS (or CovRS) is a two-component regulatory system that controls expression of multiple virulence factors in the important human pathogen group B Streptococcus (GBS). We now report global gene expression studies in GBS strains 2603V/R and 515 and their isogenic csrR and csrS mutants. Together with data reported previously for strain NEM316, the results reveal a conserved 39-gene CsrRS regulon. In vitro phosphorylation-dependent binding of recombinant CsrR to promoter regions of both positively and negatively regulated genes suggests that direct binding of CsrR can mediate activation as well as repression of target gene expression. Distinct patterns of gene regulation in csrR versus csrS mutants in strain 2603V/R compared to 515 were associated with different hierarchies of relative virulence of wild-type, csrR, and csrS mutants in murine models of systemic infection and septic arthritis. We conclude that CsrRS regulates a core group of genes including important virulence factors in diverse strains of GBS but also displays marked variability in the repertoire of regulated genes and in the relative effects of CsrS signaling on CsrR-mediated gene regulation. Such variation is likely to play an important role in strain-specific adaptation of GBS to particular host environments and pathogenic potential in susceptible hosts.
Many bacterial species utilize two-component systems (TCS) as a means to regulate gene expression in response to signals from the environment (1, 27). While several variations exist, the basic model of a TCS consists of a sensor histidine kinase that usually is positioned at the cell surface or periplasmic space to interact with external stimuli. Contact with an appropriate stimulus triggers a conformational change in the sensor protein that alters the autokinase activity of its cytoplasmic domain. Subsequent transfer of the phosphate group from the sensor to a cognate regulator component, in turn, modulates the activity of the regulator as a transcriptional repressor or activator of one or more target genes. Coordinate regulation of gene expression in response to environmental cues may be especially important for an organism like group B Streptococcus (GBS [S. agalactiae]) that exists in a commensal relationship with its human host as an asymptomatic colonizer of the genital and gastrointestinal tracts but has the potential to cause invasive infection during pregnancy and childbirth, in the colonized infant during the first weeks of life, or in the setting of concomitant chronic illness or advanced age (7, 26). Regulated changes in expression of virulence factors and metabolic pathways enhance the organism's adaptation to survive in the varied niches encountered in its existence as a commensal or as an invasive pathogen.
In keeping with the adaptation of GBS to diverse host environments, the genome sequences of GBS strains 2603V/R (hereafter referred to as 2603) and NEM316 revealed 17 and 20 predicted TCS, respectively (9, 30). Of these, the CsrRS (or CovRS) system has been investigated most thoroughly. Two independent studies reported that inactivation of csrR or csrRS resulted in increased expression of the cyl operon encoding the GBS β-hemolysin/cytolysin and a corresponding increase in hemolytic activity as well as a marked decrease in expression of cfb and its product, CAMP factor, that enhances the hemolytic activity of staphylococcal sphingomyelinase (12, 15). Both groups demonstrated, as well, that csrR mutants were attenuated in rodent models of GBS infection, a finding that supported the importance of CsrRS in pathogenesis.
Transcriptional profiling experiments using genomic macroarrays found evidence that CsrRS influenced expression of more than 100 genes in the type III GBS strain NEM316 (15). In contrast to the orthologous CsrRS (CovRS) system in Streptococcus pyogenes strain MGAS5005, in which CsrR is reported to act chiefly as a repressor, studies in GBS revealed similar numbers of activated and repressed genes (10, 15).
These initial studies provided important insights into the function of the CsrRS system as a global regulator of GBS gene expression that is likely to play a critical role in pathogenesis of GBS infection. However, many key features of this important regulatory system remain obscure. While the genes encoding CsrRS are highly conserved, the repertoires of genes regulated by the system appear not to be identical among GBS strains. For example, the cps operon that directs capsular polysaccharide biosynthesis was regulated to a modest degree in NEM316 (15). In contrast, measurement of transcripts of cpsE and of type-specific capsular polysaccharide revealed no significant difference between csrR mutants and their respective wild-type parent strains in either 2603 or 515 (12). A comparison of csrS and csrR mutants in strains 2603 and 515 revealed similar but less extreme changes in expression of three regulated genes in csrS mutants compared to csrR mutants, but whether both CsrS and CsrR have similar relative effects on expression of the entire repertoire of regulated genes has not been investigated previously (12). Finally, Lamy et al. demonstrated binding of the CsrR protein to the promoter regions of three genes whose expression is repressed by CsrR (15). However, it remains to be determined whether genes for which expression is activated by CsrR also are regulated by direct binding of CsrR or rather by repression of intermediate regulator(s).
In the present investigation, we used genomic microarrays to perform transcriptional profiling of csrR and csrS mutants in the background of strains 2603 and 515. We compared the repertoire of CsrRS-regulated genes in these strains with that described previously for strain NEM316 to determine the extent of conservation and diversity of the CsrRS regulon among GBS strains. We found considerable diversity in the CsrRS regulons and evidence for both activation and repression of target genes. Direct binding studies using recombinant CsrR revealed binding to both positively and negatively regulated promoters. Finally, we found that divergent patterns of regulation in csrS and csrR mutants were associated with strain-specific alterations in virulence. The results imply that variability in the CsrRS regulon may contribute to adaptation of particular GBS strains to specific host niches.
GBS strains used in this study included type Ia strain 515 (32) and type V strain 2603 (2603V/R) (30) and their derivative ΔcsrR and ΔcsrS mutants (12). GBS was grown in liquid culture in Todd-Hewitt broth (THB; Difco), on trypticase soy agar (TSA) supplemented with 5% defibrinated sheep blood (PML Microbiologicals), or on Todd-Hewitt agar supplemented with antibiotics and 5% defibrinated sheep blood. Escherichia coli DH5α and E. coli M15(pREP4) were grown in Luria-Bertani broth or on Luria-Bertani agar. When appropriate, antibiotics were added at the following concentrations: ampicillin, 100 μg/ml; and kanamycin, 25 μg/ml for E. coli. GBS was grown without shaking in liquid culture. E. coli was grown with shaking at 37°C. Plasmid pGEM-T (Promega) was used for the direct cloning of PCR products; plasmid pQE30 was used for the expression of recombinant His6-tagged CsrR (Qiagen).
GBS strains grown overnight on TSA blood agar plates were inoculated in 10 ml THB broth and collected by centrifugation (3,200 × g, 5 min) at mid-exponential-phase growth (optical density at 650 nm of 0.3). The pellet was resuspended in 0.5 ml 0.9% NaCl and 1 ml RNA Protect buffer (Qiagen) and kept at room temperature for 5 min. After centrifugation, the bacterial pellet was treated for 15 min at 37°C with 100 U mutanolysin (Sigma) and 15 mg/ml lysozyme (Sigma) in Tris-EDTA buffer, pH 8.0, in a final volume of 100 μl. Total bacterial RNA was then isolated using an RNeasy mini kit (Qiagen) according to the manufacturer's instructions. RNA samples were treated with DNase I (Invitrogen) for 30 min at 37°C to remove any contaminating DNA. The RNA concentration was adjusted to 100 ng/μl, and samples were stored at −80°C until use.
GBS amplicon microarrays were prepared as described previously (30) using DNA fragments of the annotated open reading frames (ORFs) from GBS strain 2603 (30). Additional primer pairs were designed to replace unresponsive 2603 amplicons and to include additional ORFs from strain A909, a serotype Ia strain (29). PCR primer pairs were designed with Primer3 (23) and locally developed Perl scripts. DNA fragments were amplified with these primers using a final concentration of 1× AmpliTaq buffer (Applied Biosystems, Blanchburg, NJ), 2.5 mM MgCl2, 0.8 mM deoxynucleoside triphosphate (dNTP) mix (Applied Biosystems), 1.25 U AmpliTaq DNA polymerase (Applied Biosystems), 0.15 μM each primer, and 20 ng/μl DNA with denaturation at 95°C for 5 min followed by amplification with 35 cycles at 95°C for 45 s, 55°C for 45 s, and 72°C for 45 s and a final elongation step at 72°C for 10 min. Amplicons were purified using Montage 96 well SEQ plates (Millipore, Billerica, MA) and spotted onto UltraGAPS aminosilane-coated glass slides (Corning, Corning, NY) using 50% dimethyl sulfoxide as the spotting buffer. Amplicons were bound to the slides by UV cross-linking at 25,000 μJ/cm2. Printed slides were stored until use in a benchtop desiccator. The final array contains amplicons for 2,086 genes from strain 2603 and 206 genes from strain A909, representing 96.4% and 97.0% of the annotated ORFs in the 2603 and A909 strains, respectively. The slide design has been deposited in ArrayExpress as A-TIGR-25, A-TIGR-26, and A-TIGR-27.
Total RNA (2 μg) from each sample was reverse transcribed into single-stranded cDNA using 1× first strand buffer (Invitrogen), 10 mM dithiothreitol (DTT), 6 μg random hexamers (Invitrogen), 0.5 mM dATP, 0.5 mM dCTP, 0.5 mM dGTP, 0.3 mM dTTP, 0.2 mM aminoallyl-dUTP (Invitrogen), and 400 U SuperScript II reverse transcription (RT) enzyme (Invitrogen). Cy dyes were chemically coupled to the incorporated aminoallyl-dUTP using Cy3- or Cy5-NHS-ester fluorescent dyes (Amersham-Pharmacia, Piscataway, NJ).
Slides were prehybridized in 8% goat serum or 1% bovine serum in 5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 0.1% sodium dodecyl sulfate (SDS) for 60 min at 42°C, washed in water and then isopropanol, and dried by centrifugation (20). Cy dye-labeled probes from mutant and wild-type strain RNAs were resuspended in hybridization buffer containing 50% formamide, 5× SSC, 0.1% SDS, and 0.6 μg/μl salmon sperm DNA (Applied Biosystems) and hybridized to the microarray slide at 42°C for 16 to 20 h in a sealed, humidified chamber (Corning) (20). Following hybridization, slides were sequentially washed in 2× SSC and 0.1% SDS for 10 min at 55°C, 0.1× SSC and 0.1% SDS for 10 min at room temperature, 0.1× SSC for 10 min at room temperature, and deionized water for 5 min at room temperature and then dried by centrifugation (20). RNA was separately isolated twice, and probes were prepared (see above) and hybridized to the array, resulting in two biological replicates for each experimental condition. Technical replication consisted of (i) ≥3-fold spotted replication on a single slide; (ii) hybridization of each RNA sample ≥4 or more times, including two dye-swap replicates; and (iii) duplicate or triplicate amplicons for a subset of genes. This level of replication resulted in ≥28 data points for each gene per experimental condition.
Slides were scanned using an Axon 4000B microarray scanner (Axon Instruments, Union City, CA) at 10-μm resolution. Data were saved as two independent 16-bit TIFF files corresponding to the Cy3 and Cy5 channels and were analyzed using TIGR Spotfinder to assess relative expression levels (24). Data from TIGR Spotfinder were stored in MAD, a relational database designed to effectively capture microarray data (24). The data from this study have been deposited in ArrayExpress (E-TIGR-131).
To adjust for differences in labeling and detection efficiencies of the fluorescent labels, the data were normalized using the MIDAS software tool (24). A low-intensity filter (100 K) was used to eliminate background fluorescence; data were normalized with iterative log mean centering (3). Ratios were calculated as the log2 (mutant/wild-type) for all spots with fluorescence above background. The mean and standard deviation were calculated from all of the log ratios for a given gene across all technical and biological replicates using local Perl scripts.
Chromosomal DNA sequences flanking CsrRS-regulated genes were inspected to identify potential operons on the basis of gene orientation and the presence of predicted Rho-independent terminators of transcription. Genes were added to the final list of regulated genes displayed in Table S1 in the supplemental material if they completed putative CsrRS-regulated operons but did not meet the threshold value to be classified as up- or down-regulated in the CsrR and/or CsrS mutant(s) or had fewer than 18 hybridization data values.
Intergenic regions upstream of CsrRS-regulated single genes or operons were searched for the presence of conserved motifs using AlignAce (http://atlas.med.harvard.edu/) and BEST (4). BEST (Binding-site Estimation Suite of Tools) runs, optimizes, and compares the results of four motif-finding programs: AlignAce, BioProspector, CONSENSUS, and MEME. Intergenic regions were submitted to the programs as follows: (i) all up- and down-regulated genes from each experiment, (ii) all up-regulated genes from each experiment, and (iii) all down-regulated genes from each experiment. Overrepresented motifs were identified and used to search the entire genome sequence of strains 2603 and 515. For each motif, the frequency of occurrence and location with regard to adjacent genes/operons were determined, as well as the extent of CsrRS regulation of genes downstream.
For real-time quantitative RT-PCR (qRT-PCR), cDNA was generated using 2 μg of total RNA, 2.5 ng random hexamers (Invitrogen), 0.5 mM dNTP mix, and 200 U SuperScript III reverse transcriptase (Invitrogen) in a 20-μl reaction mixture with RT at 25°C for 10 min followed by 50°C for 50 min and termination at 85°C for 5 min. Gene-specific primers (Operon Technologies) were designed to amplify ~150-bp fragments and to be 100% identical in both 2603 and 515 using Primer3 (23). Transcript levels were quantified using the Quantitect SYBR green PCR kit (Qiagen). Briefly, cDNA (2 μl of a 1:100 dilution of the above mixture) was used as template in a reaction containing 1× QuantiTect SYBR green mix (Qiagen) and gene-specific primers (see Table S2 in the supplemental material). The standard curve for each transcript was generated using a serial dilution of 515 csrS cDNA and gene-specific primers for SAG_0944. Reactions were run on an iCycler iQ (Bio-Rad). The reactions were denatured at 95°C for 15 min followed by amplification with 45 cycles at 94°C for 15 s, 55°C for 30 s, and 72°C for 30 s. Reactions were followed by a melt curve analysis with a disassociation step at 95°C for 1 min and 55°C for 1 min plus 0.5°C/cycle for 80 cycles.
A curated set of regulated data (see Table S1 in the supplemental material) was assembled from (i) genes ≥2-fold regulated in 2603 or 515 with n ≥ 18 and genes more than 2-fold regulated in NEM316; (ii) genes ≥2-fold regulated in at least two strains with n ≥ 10; and (iii) manual curation to add likely genes cotranscribed with those in the previous two conditions. The role category composition of this data set was compared to that of the population of amplicons on the microarray using Fisher's exact test with a Bonferroni stepdown correction for multiple experiments implemented in TIGR MeV (24). A P value of <10−4 was considered significant.
This subset and the complete data set were clustered using various algorithms in the TIGR MeV software to facilitate analysis of the data. Support trees were constructed using Euclidean distance and average linkage with resampling of experiment and sample trees by bootstrapping. K-means clustering was performed using Euclidean distance and calculated means. Ten clusters were selected and converged after eight iterations. Clustering of the subset and the complete data set resulted in similar conclusions.
Plasmid vector pQE-30 was used for GBS CsrR protein overexpression in E. coli. A 687-bp PCR-generated BamHI/HindIII DNA fragment corresponding to the CsrR coding sequence was amplified using specific primers #880 and #881 (see Table S2 in the supplemental material) and then cloned between the BamHI and HindIII sites of pQE-30 to give pQE30CsrR. The resulting clone was first transformed into E. coli DH5α for amplification and then isolated and introduced into the E. coli M15 bearing the plasmid encoding the lac repressor, pREP-4. The transformants were inoculated into LB medium containing both ampicillin (100 μl/ml) and kanamycin (25 μg/ml) with 500 μl of the overnight culture and grown at 37°C until A600 = 0.5 to 0.7. Expression of His6-CsrR was induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to 1 mM and then followed by 3 h of incubation at 37°C. Cells were lysed by sonication, and insoluble cell debris was removed by centrifugation (20,800 × g, 20 min). The cell lysate was then passed over an Ni2+-nitrilotriacetic acid agarose column for purification according to the manufacturer's instructions (Qiagen). His6-CsrR was suspended in 50% glycerol and stored at −20°C. The concentration of purified His6-CsrR was estimated by comparison with a bovine serum albumin standard using the bicinchoninic acid kit (Pierce).
For electrophoretic mobility-shift assays (EMSA), pairs of specific oligonucleotide primers were used for the PCR amplification of 250- to 305-bp DNA fragments representing the potential promoter regions of target genes cyl (β-hemolysin), cfb (CAMP factor), and scpB (C5a peptidase) (see Table S2 in the supplemental material). The purified PCR products were end labeled with γ-[S-32P]dATP in the presence of T4 polynucleotide kinase using the Promega gel shift assay system kit. Labeled probes were purified from free nucleotides on a G-25 spin column (Amersham). As a negative control, a similar size DNA fragment was also amplified from the promoter region of the capsule synthesis (cps) locus. His6-CsrR was serially diluted and added to 10 ng of probe DNA in binding buffer [20 mM Tris (pH 7.5), 1 mM CaCl2, 1 mM DTT, 10 μg/ml poly(dI-dC) and 100 μg/ml bovine serum albumin] in a total volume of 10 μl. DNA and protein were incubated for 15 min at room temperature. The reaction products were mixed with 2 μl of 50% glycerol and loaded onto a 5% Tris-borate-EDTA (TBE) polyacrylamide gel (Bio-Rad; TBE is 90 mM Tris-borate, 2 mM EDTA [pH 8.0]). After loading samples, the gel was run at room temperature in 0.5× TBE buffer at 350 V for 15 to 20 min and then dried and exposed to X-ray film. In some experiments, purified His6-CsrR was subjected to in vitro phosphorylation before addition to the assay mixture: 10 μg His6-CsrR protein was incubated for 90 min at 37°C with 32 mM acetyl phosphate in freshly made phosphorylation buffer (20 mM NaH2PO4, pH 8.0, 10 mM MgCl2, 1 mM DTT) in a total volume of 100 μl.
Adult outbred male CD-1 mice were obtained from Charles River Breeding Laboratories (Calco, Italy). The animals were 6 to 8 weeks of age at the beginning of each experiment. GBS strains were grown overnight at 37°C in THB (Oxoid, Ltd., Basingstoke, England), washed, and diluted in RPMI 1640 medium (Gibco Life Technologies, Milan, Italy). The inoculum size was confirmed by quantitative cultures. Mice were inoculated intravenously via the tail vein with different infecting doses of GBS in a volume of 0.5 ml. Control mice were injected by the same route with 0.5 ml of RPMI 1640 medium. Mortality was recorded at 24-h intervals for 30 days. The 50% lethal dose (LD50) was calculated by the method of Reed and Muench (22).
GBS-infected mice were evaluated for signs of arthritis and mortality. After challenge, mice were examined daily by two independent observers for 1 month to evaluate the presence of joint inflammation. Arthritis was defined as visible erythema and/or swelling of at least one joint. Clinical severity of arthritis was graded on a scale of 0 to 3 for each paw, according to changes in erythema and swelling (0 = no change; 1 point = mild swelling and/or erythema; 2 points = moderate swelling and erythema; 3 points = marked swelling, erythema, and/or ankylosis). Thus, a mouse could have a maximum score of 12. The arthritis index (mean ± standard deviation) was calculated by dividing the total score (cumulative value of all paws) by the number of animals in each experimental group.
The slide design for this study has been deposited in ArrayExpress as A-TIGR-25, A-TIGR-26, and A-TIGR-27. The microarray data from this study have been deposited in ArrayExpress (E-TIGR-131).
To investigate the function of the CsrRS TCS in GBS, we previously constructed mutants in the background of GBS type Ia strain 515 and type V strain 2603 (12). A nonpolar inactivating mutation was introduced into csrR (515ΔcsrR or 2603ΔcsrR) or csrS (515ΔcsrS or 2603ΔcsrS). Analysis of expression patterns of a limited number of genes revealed increased expression of cylE (β-hemolysin/cytolysin) and scpB (C5a peptidase) and reduced expression of cfb (CAMP factor) in the csrR mutants in both strain backgrounds and similar, though less extreme, changes in the csrS mutants (12). In the present study, we used GBS genomic microarrays as a more comprehensive means to investigate genome-wide changes in gene expression that result from inactivation of csrR or csrS in the same two GBS strain backgrounds.
Inactivation of csrR or of csrS was associated with altered expression of a large number of genes in both strain backgrounds (Table (Table1).1). Using as a threshold a twofold change in gene expression between the mutant and wild type, we found evidence of CsrRS regulation of 134 genes in strain 2603 and 80 genes in strain 515. One significant difference between the 2603 and 515 strains was the presence of 12 IS1381 transposase subunits found to be up-regulated in 2603ΔcsrR. A glycosyltransferase (SAG_1548, SAG_1551) in the up-regulated and cotranscribed SAG_1548 to -1555 is disrupted by an IS1381 transposase (SAG_1549, SAG_1550). Although the IS1381 transposase genes are on the opposite strand relative to the cotranscribed unit, the amplicon-based microarray queries both the transcript and its reverse complement. Therefore, up-regulation of genes SAG_1548 to -1555 results in what is likely to be artifactual “up-regulation” of the inserted transposase. Further, the high nucleotide identity of all the transposases results in positive results for all chromosomal locations. Therefore, the IS1381 transposases were removed from subsequent analyses, as it is unlikely that this apparent up-regulation is biologically relevant.
Bioinformatics software was used to search for putative CsrR-binding motifs associated with CsrRS-regulated individual genes and operons in the genome sequences of strains 2603 and 515. Among a number of candidate motifs, we failed to identify any that were both overrepresented in intergenic regions and preferentially located upstream of CsrRS-regulated genes. Therefore, this analysis did not support the earlier suggestion of a distinct CsrR binding sequence (15) but suggested rather that CsrR recognizes regions of DNA that are not readily identified by a canonical nucleotide sequence motif.
To confirm the changes in transcript abundance observed by microarray hybridization, we performed qRT-PCR for a subset of regulated genes using RNA samples from mutant and wild-type strains. Nine genes were selected for qRT-PCR testing using RNA samples from 515ΔcsrR, 515ΔcsrS, 2603ΔcsrR, and 2603ΔcsrS, and their respective wild-type parent strains. The change (fold) in gene expression by qRT-PCR correlated well with those calculated from the microarray experiments in each of the mutant strains (R2 = 0.94; see Fig. S1 and Table S3 in the supplemental material).
Transcriptional analysis using genomic microarrays confirmed the previously reported regulation of known or putative virulence factors including the cyl operon (SAG_0662 to -0673) encoding the GBS β-hemolysin/cytolysin, scpB (SAG_1236; C5a peptidase), and cfb (SAG_2043, CAMP factor), as well as a second gene transcriptionally linked to cfb (SAG_2042) that is predicted to encode a rhodanese-like protein of unknown function (see Table S1 in the supplemental material). A second locus encoding a predicted protein with 67% amino acid identity to C5a peptidase was also regulated by CsrRS (SAG_0416). As reported previously, and in contrast to the earlier analysis of GBS strain NEM316, we did not find a consistent pattern of regulation of the cps capsular polysaccharide synthetic operon in strain 2603. In 515ΔcsrR, there was a trend of down-regulation of the cps operon, but this trend only reached the twofold threshold for cps1aJ (SAL_1283) and cps1aH (SAL_1285) among the 16 genes in the cps operon.
Regulated genes in one or both strains encoded proteins involved in a wide range of cell functions including known or predicted virulence factors, transporters of amino acids, peptides, sugars, and metals, and proteins that mediate adaptation to environmental stresses (see Table S1 in the supplemental material). Among the various functional categories of regulated genes, transporters of amino acids, peptides, and amines were most significantly overrepresented in the study (P < 10−7), followed by transport and binding proteins (P < 10−5) and pathogenesis genes (P < 10−5). We found evidence for regulation of expression of several proteins that are predicted to be secreted or surface associated, including SAG_0297 (aminopeptidase C), SAG_1002 (putative protease), SAG_1890 (putative endopeptidase O), and two operons predicted to encode membrane proteins (SAG_0364 to -365 and SAG_0798 to -799).
Genes encoding proteins involved in transport of a variety of substrates constituted a large group of regulated genes. These included the oppA1-F operon (SAG_0148 to -0152), encoding an oligopeptide ABC transporter that has been implicated in modulating the attachment of GBS to host cells and the adc operon (SAG_0154 to -0156), homologs of which encode a zinc/manganese transporter in Streptococcus pneumoniae and a manganese acquisition and homeostasis system in Streptococcus gordonii (6, 16, 25). Also regulated by CsrRS are homologs of a system involved in iron transport (SAG_1007-1010).
In keeping with the inferred role of CsrRS in adaptation to changing environments, the system regulates several stress response mechanisms in GBS. These included homologs of AphC and AphF (SAG_1833 to -1834), two components of the alkyl hydroperoxide reductase of S. pyogenes. In that species, the alkyl hydroperoxide reductase system contributes to scavenging endogenous hydrogen peroxide and has been linked to virulence in a murine infection model (2). Homologs of the enterococcal and lactococcal general stress protein Gls24 (SAG_1135 and SAG_1137) were also regulated by CsrRS in GBS. Gls24 has been implicated in stress response and virulence in Enterococcus faecalis (18, 28). The GBS CsrRS also regulates expression of two separate operons predicted to encode components of a glycine/betaine osmoregulation system (SAG_1796 to -1797 and SAG_0241 to -0244), a system that mediates adaptation to osmotic stress in Bacillus subtilis and Lactococcus lactis (13, 14, 19, 31). Although their role in GBS is undefined, expression of two predicted transcriptional regulators is also controlled by CsrRS: SAG_0712 encoding a putative regulator of the OmpR family and SAG_0938 encoding a predicted GntR family transcriptional regulator (11, 27).
The availability of genome-wide transcriptional profiling data in strains 2603 and 515 together with the previously described results in strain NEM316 provided the opportunity to compare the CsrRS regulons in independent GBS isolates representing the three most important capsular serotypes in human infection, types Ia (strain 515), III (strain NEM316), and V (strain 2603). This analysis revealed a core group of 39 genes whose expression was changed as a result of inactivation of csrR and/or csrS in all three strain backgrounds (Fig. (Fig.11 and Table Table2).2). Two-way comparisons showed further overlap in the repertoire of CsrRS-regulated genes, with 16 genes regulated in both 2603 and 515, 18 in 2603 and NEM316, and 3 in 515 and NEM316. For each of the three strains, certain CsrRS-regulated genes were regulated in only one strain background. A higher number of uniquely regulated genes were identified in NEM316, but this may be a result of the differences in array platform and mutant design between the two studies. These combined results suggest that CsrRS regulates a conserved core group of genes in multiple GBS strains, including those coding for the important virulence factors β-hemolysin, C5a peptidase, and CAMP factor, as well as a large repertoire of genes whose regulation varies among GBS isolates.
Mutation of csrR in strain 2603 was associated with increased expression of 94 genes and reduced expression of 13 genes. This pattern suggests that CsrR acts predominantly as a transcriptional repressor but that it can also activate gene transcription, directly or indirectly. Inactivation of csrS resulted in expression changes of a smaller number of genes with 36 genes up-regulated in 2603ΔcsrS and 18 genes down-regulated (Table (Table1).1). For 27 genes, we observed altered expression in both 2603ΔcsrR and in 2603ΔcsrS (Fig. (Fig.2).2). In 23 of these 27 genes, gene expression was up-regulated in both 2603ΔcsrR and 2603ΔcsrS. For most regulated genes, mutation of csrR produced a greater effect than did mutation of csrS (see Fig. S2 in the supplemental material). We observed a change in expression in 2603ΔcsrR but not in 2603ΔcsrS for 80 genes (Fig. (Fig.2).2). For the majority of these genes, the change in expression in 2603ΔcsrS was in the same direction as that in 2603ΔcsrR, but it did not reach the twofold threshold. For two linked genes, SAG_1706 (hypothetical protein) and SAG_1707 (putative drug resistance transporter), expression was increased in 2603ΔcsrR and decreased in 2603ΔcsrS.
While the total number of CsrRS-regulated genes was somewhat lower in strain 515 than in 2603 (Table (Table1),1), we observed a similar pattern with respect to the relative effects of mutation in csrR versus csrS. That is, the predominant pattern was a greater effect on target gene expression in 515ΔcsrR than in 515ΔcsrS, but for several genes, the effect was greater in 515ΔcsrS than in 515ΔcsrR (Fig. (Fig.33).
An earlier investigation of GBS strain NEM316 used DNase I protection and EMSA to demonstrate direct binding of recombinant CsrR to a DNA segment upstream of the cyl operon (15). DNase I protection experiments also suggested direct binding to the promoter regions of two other genes whose expression, like that of the cyl operon, is repressed by CsrRS. To further characterize the interaction of CsrR with regulated promoters, we expressed CsrR as a His6 fusion in Escherichia coli and purified the recombinant protein by Ni2+-affinity chromatography. His6-CsrR was used in EMSA with DNA probes from strain 2603 that corresponded to the promoter regions of the cyl operon and scpB (C5a peptidase), two genes whose expression is repressed by CsrRS; and cfb (CAMP factor), a gene whose expression is activated by CsrRS. For all three promoters, band shifts were observed in the presence of His6-CsrR, indicating direct binding of CsrR to DNA sequences upstream of both CsrRS-repressed and CsrRS-activated genes (Fig. (Fig.4).4). No shift was observed after incubation of His6-CsrR with a DNA segment corresponding to the promoter region of the cps operon of strain 2603, a result that is consistent with the absence of CsrRS regulation of this locus in strain 2603 and that serves as a negative control for the specificity of CsrR binding to regulated promoters. Specificity of the binding interaction for each of the regulated promoters was also supported by competition with excess unlabeled probe, but not with excess unlabeled cps promoter sequences. These results indicate that CsrR binds directly to both positively and negatively regulated promoter sequences.
Signaling through TCS typically is transduced by phosphorylation or dephosphorylation of the regulator component in response to interaction of the sensor with an environmental stimulus. In keeping with this general model, the phosphorylation state of the regulator has been shown to change its affinity for target DNA sequences in several TCS in other species, including the homologous CsrRS system in Streptococcus pyogenes (5, 8, 17). To test the importance of CsrR phosphorylation for gene regulation in GBS, we incubated His6-CsrR with acetyl phosphate to phosphorylate the CsrR protein. EMSA using a probe for the cyl operon promoter revealed a minor increase in binding affinity (approximately twofold) for the phosphorylated compared to unphosphorylated His6-CsrR (Fig. (Fig.5).5). This modest effect of phosphorylation at the cyl promoter is consistent with the results of Lamy et al., who found no significant effect of phosphorylation on binding of CsrR to the cyl promoter (15). In contrast, phosphorylation had a marked effect on binding of CsrR to the scpB or cfb promoters, increasing binding affinity by approximately eightfold. Phosphorylation also appeared to enhance the formation of higher-molecular-size complexes, a result that suggests phosphorylation may promote oligomerization of CsrR. These results indicate that phosphorylation increases binding of CsrR to regulated promoters and that individual target promoters differ in their relative affinities for the phosphorylated versus unphosphorylated regulator protein.
The microarray analysis revealed both qualitative and quantitative differences in the relative effects on gene regulation of mutating csrR or csrS in the background of strain 2603 compared to that in 515 (Fig. (Fig.22 and and33 and Table Table1).1). Furthermore, opposite regulatory effects were observed for 2603ΔcsrS and 2603ΔcsrR (e.g., with SAG_1706 to -1707), whereas such divergent effects were not observed in the background of strain 515. To investigate whether such differential effects of CsrR compared to CsrS might be reflected in the relative pathogenic potential of the mutant strains, we tested the virulence of csrR and csrS mutants and wild-type strains 2603 and 515 in a murine model of systemic infection and septic arthritis. Mice were challenged intravenously with various doses of GBS and observed for development of signs of arthritis and for mortality. Wild-type strain 515 was the most virulent in these studies, with an LD50 of 7.2 × 104. The LD50 for 515ΔcsrR was 320-fold higher at 2.3 × 107, whereas 515ΔcsrS had an intermediate level of virulence (LD50 of 8.5 × 106). This hierarchy of relative virulence is the same as that reported previously for these strains in a murine intraperitoneal challenge model (12). Similarly, strain 2603ΔcsrR was attenuated in virulence (no deaths at challenge doses up to 108 CFU) relative to wild-type strain 2603 (LD50 of 2.8 × 107 CFU), as reported previously for the intraperitoneal challenge model (12). In striking contrast, strain 2603ΔcsrS was more virulent (LD50 of 2.4 × 106 CFU) than wild-type 2603. This virulence hierarchy was reflected not only in the relative lethality of the three strains in the 2603 background but also in severity of arthritis, whether scored by number of affected joints or by clinical severity index (Fig. (Fig.66).
These results demonstrate that differential patterns of regulation by CsrRS in strains 2603 and 515 are associated with striking differences in the overall relative virulence of csrS mutants in the two strain backgrounds. In particular, we observed increased expression of certain genes in 2603ΔcsrS, but not in 2603ΔcsrR or in either mutant in the 515 background. Of these, possible virulence genes include SAG_1135 and SAG_1137 that encode homologs of Gls24, a stress response protein shown to contribute to virulence in experimental enterococcal infection (18, 28). A similar pattern of regulation was noted for SAG_1796 and SAG_1797, which are predicted to encode a glycine/betaine osmoregulation system implicated in adaptation to osmotic stress in other species (13, 14, 19, 31). Differential CsrS-dependent regulation of these loci in the 2603 strain background may account for the unexpectedly high virulence of strain 2603ΔcsrS.
Results of the present investigation provide several new insights into the CsrRS TCS in GBS and its potential functions during infection. Transcriptional profiling studies using genomic microarrays yielded a comprehensive picture of global gene regulation by CsrRS in GBS strains 2603 and 515 in addition to that reported previously for strain NEM316. We found evidence for an extensive regulon in all three strains, including genes that encode products known or predicted to enhance bacterial survival under varied conditions and/or to contribute to pathogenicity in the human or animal host. Microarray hybridization experiments and qRT-PCR confirmed CsrRS regulation of three major virulence determinants identified in earlier studies: β-hemolysin, CAMP factor, and C5a-peptidase. There was also evidence of regulation of several predicted surface or secreted proteins of unknown function that may participate in GBS adherence or in modification of the local host environment.
The largest functional class of CsrRS-regulated genes encoded transport systems for various small molecules including peptides, amino acids, sugars, and metals. This broad regulation of small molecule transporters is consistent with the postulated role of CsrRS in mediating adaptation of GBS to varied environmental and nutritional circumstances encountered within the colonized or infected host. We also found evidence for CsrRS regulation of stress response systems such as the OpuA betaine uptake osmoprotection system. Adaptation to osmotic stress may be especially important for GBS survival at mucosal sites in which fluid and solute shifts result in changing osmotic conditions. Similarly, CsrRS appears to regulate expression of genes encoding alkyl hydroperoxide reductase, an enzyme implicated in resistance to endogenous hydrogen peroxide stress in S. pyogenes. Such resistance may be adaptive during interaction of GBS with host phagocytes or under other circumstances of oxidative stress.
Comparison of the repertoire of genes regulated by CsrRS in three different GBS strains suggests that there is a “core regulon” of genes that are regulated by CsrRS in multiple strains, but that there is substantial diversity in the remainder of the regulon. For the three strains studied to date, the core regulon consists of 39 genes, including the virulence factors described in earlier reports. Thirty-seven genes show evidence of CsrRS regulation in two of the three strains, while 155 genes are regulated in only one strain. Such variation in CsrRS regulons is likely to be a reflection of the overall genomic variability in this species (29). Heterogeneity in gene regulation among individual isolates may account, in part, for differences in adaptation to specific host environments and for the pathogenic potential of particular strains. The molecular basis for variability in CsrRS repertoire remains to be determined and may involve multiple factors, including strain-specific variation in promoter sequences of regulated genes and the presence or absence of additional interacting regulators.
By characterizing changes in gene regulation in csrS mutants as well as csrR mutants, we uncovered another level of complexity in CsrRS-mediated gene regulation. For most regulated genes, we observed a similar trend in gene expression in the csrS mutant as in the csrR mutant, but the change in gene expression was of a lower magnitude in the csrS mutant. However, in a significant minority of cases, mutation in csrS, but not in csrR, resulted in a change in gene expression, or the degree of change in expression was greater in the csrS mutant than in the csrR mutant. In two cases, we observed divergent effects on gene regulation in the csrS compared to csrR mutants. The most common pattern of regulation—a greater effect by inactivating CsrR than CsrS—is consistent with the basic TCS model: removal of the transcriptional regulator has a maximal effect, usually by derepressing target gene transcription. Inactivation of the sensor may have a similar, but often lesser, effect by preventing signaling that activates or inactivates the regulator. However, other patterns of responses are also possible. An equivalent effect of inactivating CsrS might be expected for genes whose promoter regions bind phospho-CsrR with much higher affinity than the unphosphorylated CsrR, assuming CsrR phosphorylation is dependent on CsrS signaling. In agreement with this formulation, we found heterogeneity in the importance of CsrR phosphorylation in determining the relative binding affinity of CsrR for different target promoters in vitro. Alternatively, or in addition, differential effects could be mediated by interaction of either CsrS or CsrR with other transcriptional regulators with various specificities for CsrRS-regulated genes. For example, evidence has been presented that the serine/threonine kinase Stk1 interacts with CsrR to regulate expression of β-hemolysin and CAMP factor in GBS (21).
Strain-to-strain variation in the CsrRS regulon and in the relative regulatory effects of the sensor and regulator components implies that GBS strains vary not only in gene content but also in their dynamic capacity to adapt to changing environments in the host. A particular repertoire of CsrRS gene regulation may confer a survival advantage or pathogenic potential in a specific microenvironment. This conclusion is supported by our finding that the virulence of csrR and csrS mutants relative to their respective parent strains differs in strains 2603 and 515. Thus, an individual strain may be better adapted for survival in a particular host site such as the bovine mammary gland or the human gastrointestinal or genitourinary tract. In this way, the species as a whole has enhanced adaptation to varied host niches. Together, these results provide further evidence that CsrRS serves as a global regulatory system in GBS that functions in both conserved and variable ways to enhance adaptation of this important pathogen for survival in the host.
We thank Dennis L. Kasper for helpful advice.
This work was supported in part by NIH grant AI59502.
Published ahead of print on 18 January 2008.
†Supplemental material for this article may be found at http://jb.asm.org/.