CovS Inactivates CovR and Is Required for Growth under Conditions of General Stress in Streptococcus pyogenes

Abstract

The gram-positive human pathogen Streptococcus pyogenes (group A streptococcus [GAS]) causes diseases ranging from mild and often self-limiting infections of the skin or throat to invasive and life-threatening illnesses. To cause such diverse types of disease, the GAS must be able to sense adverse environments and regulate its gene expression accordingly. The CovR/S two-component signal transduction regulatory system in GAS represses about 15% of the GAS genome, including many genes involved in virulence, in response to the environment. We report that CovR is still able to repress transcription from several promoters in the absence of the putative histidine kinase sensor for this system, CovS. We also show that a phosphorylation site mutant (D53A) of CovR is unable to repress gene expression. In addition, we report that a strain with a nonpolar mutation in CovS does not grow at a low pH, elevated temperature, or high osmolarity. The stress-related phenotypes of the CovS mutant were complemented by expression of covS from a plasmid. Selection for growth of a CovS mutant under stress conditions resulted in isolation of second-site mutations that inactivated covR, indicating that CovR and CovS act in the same pathway. Also, at 40°C in the wild-type strain, CovR appeared to be less active on the promoter tested, which is consistent with the hypothesis that it was partially inactivated by CovS. We suggest that under mild stress conditions, CovS inactivates CovR, either directly or indirectly, and that this inactivation relieves repression of many GAS genes, including the genes needed for growth of GAS under stress conditions and some genes that are necessary for virulence. Growth of many gram-positive bacteria under multiple-stress conditions requires alteration of promoter recognition produced by RNA polymerase association with the general stress response sigma factor, σ^B. We provide evidence that for GAS, which lacks a sigB ortholog, growth under stress conditions requires the CovR/S two-component regulatory system instead. This two-component system in GAS thus appears to perform a function for which other gram-positive bacteria utilize an alternative sigma factor.

Streptococccus pyogenes (group A streptococcus [GAS]) is a gram-positive pathogen that causes a wide range of diseases in humans. The most common diseases caused by GAS are pharyngitis and pyoderma, which are mild, noninvasive infections of the throat and skin, respectively. However, GAS also has the ability to spread deeper into tissues, which leads to more serious, invasive diseases, such as necrotizing fasciitis, septicemia, and streptococcal toxic shock syndrome (reviewed in reference 10). In the past, GAS was also a much-feared common cause of puerperal sepsis (37). The ability of single GAS strains to produce such diverse diseases suggests that there is a complex system that regulates the gene expression of the bacterium in response to its microenvironment.

Bacteria often sense and respond to specific environmental changes by using two-component signal transduction regulatory systems (26, 43). In general, these systems rely on phosphotransfer between two proteins, a membrane-located histidine kinase sensor and a DNA-binding response regulator. A specific environmental signal sensed by the kinase leads to autophosphorylation of a particular histidine in a conserved domain of this class of proteins. The phosphoryl group is subsequently transferred to an aspartate residue of the receiver domain of the cognate response regulator, which induces a conformational change that allows the regulator to bind DNA. Such binding leads to activation or repression of a nearby promoter. Bifunctional histidine kinases also have phosphatase activity that enables them to catalyze the release of the phosphoryl group from the response regulator under certain conditions, and the regulator is then expected to release the DNA. This type of regulation permits rapid and reversible changes in gene expression in response to the local environment. Such two-component systems are often involved in stress responses by leading to the production of the alternative sigma factor(s) required for transcription of genes needed for growth or survival under stress conditions (25, 33, 36).

One of the 13 predicted two-component regulatory systems in the GAS genome (18) has been studied extensively. The response regulator CovR (also called CsrR [31]), which is a member of the OmpR family, represses about 15% of the genes in this organism either directly or indirectly (19). Because many of these genes are virulence factors of GAS, the regulator was designated Cov (control of virulence) (16). CovR has been shown to bind the promoter for the hyaluronic acid capsule operon (has) in vitro and to repress this promoter directly in vivo (17, 35). Additional virulence genes whose promoter regions are bound by CovR in vitro include ska (streptokinase), speMF/sda (streptococcal DNase), sagA (streptolysin S), and covR itself (16, 22, 35). Because covS is cotranscribed with covR and the protein exhibits significant homology to histidine kinases (52), it is assumed that CovS is the sensor for this two-component system, although this has not been demonstrated directly. It has recently been shown that repression of Phas by CovR is greater at high Mg²⁺ concentrations and that an insertion in covS prevents this effect (20). Although the covS mutation may be polar on the expression of downstream genes, this observation suggests that CovR and CovS are in the same pathway.

Since GAS strains are able to grow in diverse conditions in the host, it seems likely that they have a mechanism(s) for survival in adverse environments, like the high temperatures resulting from fever and the low pH of the vagina and abscesses. Bacteria respond to stress by altering gene expression to produce proteins needed for growth in the new conditions. In addition to specific responses induced by individual types of stress, a general, global, or universal type of response has also been distinguished (3, 23, 24, 53). The latter response results in increased production of the same group of general stress proteins in response to many different types of stress conditions. In most bacteria, this global stress response is mediated by one or more alternative sigma factors that alter the affinity of RNA polymerase to allow transcription of genes not transcribed under normal laboratory growth conditions. In Escherichia coli, σ^S is required for a response to general stress, in Bacillus subtilis σ^B serves this purpose, and in Streptomyces coelicolor nine paralogous general stress response sigma factors have been identified (23, 24, 55). However, none of the three completed GAS genomes encodes a recognizable ortholog of such alternative sigma factors (5, 18, 49).

In this work, we found that in the presence of CovR, CovS is required for growth of the GAS under different mild stress conditions. From this we concluded that CovR and CovS interact and are required for the GAS response to several stress conditions. We propose that this two-component system is an alternative to the use of a minor sigma factor for survival under stress conditions in the GAS.

MATERIALS AND METHODS

Bacterial growth conditions and media.

Unless indicated otherwise, GAS strains were grown at 37°C without agitation in Todd-Hewitt broth supplemented with 0.2% yeast extract (THY). THY-H agar was made by adjusting the pH of THY to 7.5 with NaOH and adding 100 mM HEPES (pH 7.5) (Sigma) prior to autoclaving. THY-M agar was made by adjusting the pH of THY to 6.0 with HCl and adding 100 mM MES pH 6.0 (Sigma) prior to autoclaving. For high-salt-concentration experiments, 0.65 M NaCl was added to THY-H prior to autoclaving. Antibiotics were used at the following concentrations: chloramphenicol, 5 μg/ml for GAS and 20 μg/ml for E. coli; erythromycin, 0.5 μg/ml for GAS and 500 μg/ml for E. coli; spectinomycin, 100 μg/ml for both GAS and E. coli; and kanamycin, 200 μg/ml for GAS and 50 μg/ml for E. coli.

Construction of covS mutant.

Overlapping PCR was performed as previously described (29). An overlapping PCR product containing a promoterless chloramphenicol resistance cassette (cat) followed by a ribosome-binding site (to avoid polar effects on the downstream gene) was used to replace covS. The promoterless cat gene was amplified from pLZ12 (13) by using primers Cat-L (gtcacggatcctgactaacTAGGAGGCATATCAAATGAAC) and Cat-R (cagcggatcccatctaggcctcCTCATATTATAAAAGCCAGTC) (lowercase underlined letters indicate restriction sites incorporated into primers, boldface type indicates ribosome-binding sites, and uppercase letters indicate regions complementary to cat). The product was then digested with BamHI and ligated into BamHI-digested pUC19 (58) to construct pEU7050 (Tariq Perwez, unpublished data). In the following primer sequences, uppercase letters indicate regions complementary to the template used in the PCR, and lowercase letters indicate bases at the ends of the primers that were incorporated during the PCR to form overlapping regions between two PCR products. Primers covRS2 (ATTAGGAGAAGATGATGTTAGC) and delcovS2 (tgcctcctagttagtcagTCTTCTGTTTTTGTTTCTGATTTTC) were used to amplify the 5′ region upstream of covS from the JRS4 chromosome and to introduce a region that overlapped the cat sequence. Primers delcovS5 (gaaaatcagaaacaaaaacagaagaCTGACTAACTAGGAGGCA) and delcovS6B (cgattacgtgatttatccgGGATCCCATCTAGGCCTCCTC) were used to amplify the promoterless cat cassette from plasmid pEU7050. Primers delcovS3B (gaggaggcctagatgggatccCGGATAAATCACGTAATCG) and mucORF3-A1 (GACCCAATTTCGCCATGGTT) were used to amplify the 3′ region downstream of covS from the JRS4 chromosome and to introduce regions overlapping the cat sequence. The gel-purified PCR products of the region with 5′ homology to the CovRS operon and cat were combined in a PCR in order to fuse the two together by using nested primers delcovSN1 (CCGAAATCAGAAAACACCCAGAC) and delcovSN3 (GGCCTATCTGACAATTCCTGAATA). The same procedure was used to fuse the PCR products of cat to the 3′ region of the CovRS operon by using nested primers delcovSN2 (ATCTTTTCCATCTAGTCACCCCC) and delcovSN4 (GATTTAGACAATTGGAAGAGAAAAGAG). The two resulting overlapping PCR products (5′Cov-cat and cat-3′Cov) were gel purified and combined in another PCR by using nested primers delcovSN5 (GGTCTCTCGTTTGATCGTTTATGTGATG) and delcovSN6 (CCTAGGTTTTCTAACCTCTTCAAAGC). The resulting 4-kb fusion product was confirmed by digestion with NheI and XhoI. Approximately 15 μg of the linear PCR product was transformed into JRS4 by electroporation, and chloramphenicol-resistant transformants were selected.

Allelic replacement of covS was confirmed by PCR by using primers covSS2 (ATGGAAAATCAGAAACAAAAAC) and covSA2 (TACTCTAACTCTCTTTAGACT). As expected, the mutant produced a band that was ∼700 bp smaller than the band produced by the wild type. For confirmation, the 5′ and 3′ junctions were amplified by using primers covSA2 and delcovSN4 (5′ junction) and primers covSS2 and delcovSN3 (3′ junction). The 5′ junction between the CovR/S region and cat and the complete covR gene were sequenced to confirm this construct.

RNA hybridization.

RNA hybridization was performed as described by Biswas and Scott (7). DNA probes were prepared by PCR amplification by using JRS4 chromosomal DNA as the template. The following primer pairs were used in this study: for rpsl, spn-rpsl1 (gccgaattcGAATGTAGATGCCTACAATTAACCA) and spn-rpsl2 (cccaagcttTTTACGACTCATTTCTCTTTATCCC); for sagA, SagAL (GGAGGTAAACCTTATGTTAA) and SagAR (AGATTATTTACCTGGCGTAT); and for sda, SpeMF-A1 (CAGAAGATTGCATTGATACC) and SpeMF-S1 (ATGAATCTACTTGGATCAAG).

Construction of covR-complementing plasmids pJRS996 and pJRS329.

Primers R1-Cov/orf1 (ggaattcTCTGGTATTAGTTTTAGACAAAGACGC) and MucR-R (gagagaccggaattcATGACTTATTTCTCACGAAT) were used to amplify covR and 500 bp of the promoter upstream region of covR from JRS4 chromosomal DNA (underlined lowercase letters indicate the sequences of the restriction sites incorporated into the PCR product). The PCR product was cloned into the pCR-blunt II Topo vector (Invitrogen catalog no. 45-0245) to construct pEU7013 (Charlotte Denis and Michael Federle, unpublished data). pEU7013 was digested with EcoRI, and the covR-containing fragment was ligated into EcoRI-digested pLZ12 (13) to construct pJRS996 (Denis and Federle, unpublished). QuikChange site-directed mutagenesis (Stratagene catalog no. 200518-5) was performed by using pJRS996 as a template to incorporate the D53A amino acid change in CovR with the mutagenic primers D53A-S1 (GTTTGATTTAATCCTGCTTGCCTTAATGTTACCAGAGATGGATGGTTTTG) and D53A-A1 (CAAAACCATCCATCTCTGGTAACATTAAGGCAAGCAGGATTAAATCAAAC) (boldface type indicates the change that was made). The resulting plasmid was pJRS329. The D53A amino acid change was confirmed by DNA sequence analysis.

Both pJRS996 and pJRS329 were propagated in E. coli XL1-Blue cells (Stratagene) and subsequently transformed into the previously constructed strain JRS969 (17) and selected for chloramphenicol resistance.

Gus assays.

Gus (β-glucuronidase) assays were performed as described by Biswas and Scott (7). One unit of Gus activity was defined as the activity which liberated 1 μg of phenolphthalein (from phenolphthalein glucuronide)/h/mg of protein in a GAS lysate at 37°C and pH 6.8. The protein concentration was determined with a Micro BCA protein assay kit (Pierce catalog no. 23235) standardized with bovine serum albumin (Sigma catalog no. P-0914).

Construction of covS-complementing plasmid pJRS325.

A PCR fragment containing the coding region of covS and its ribosome-binding site was amplified from JRS4 by using primers CovS-F-BamHI (ccgggatccTACGTTATTCGTGAGAAATA) and CovS-A2-PstI (ccgctgcagACTCTAACTCTCTTTAGACT) (underlined lowercase letters indicate the sequences of the restriction sites incorporated into the PCR products). The resulting 1.5-kb fragment was digested with BamHI and PstI and ligated into BamHI/PstI-digested pOri23 (42), an erythromycin-resistant shuttle vector that is able to replicate in GAS and allows covS to be expressed from the constitutive lactococcal promoter P23. The resulting plasmid, pJRS325, was propagated in E. coli Top10 cells (Invitrogen) before transformation into JRS331.

Construction and labeling of probes used for RNase protection assays.

Internal regions of specific genes were amplified by PCR from JRS4 chromosomal DNA by using Herculase (Stratagene catalog no. 600260) to obtain blunt ends and were subsequently cloned into the pCR-blunt II Topo vector (Invitrogen catalog no. 45-0245), and the orientation was confirmed by PCR. The following primer pairs were used to make specific probes: for hasA, inthas-S4 (cccacccAGCTTCAATGATGAGACAGTTTATG) and inthas-A4 (gggtgggAGGAGGAATTCACCTAGGAATGTTTGATTTT); and for gusA, intgus-S1 (cccacccAAGCCAGACAGAGTGTGATA) and intgus-A1 (gggtgggTAAGGGTAATGCGAGGTAC). The specific gene fragments were then amplified from their pCR-bluntII derivative plasmids (pEU7229-hasA and pEU7230-gusA) by using primers M13F (GTAAAACGACGGCCAG) and M13R (CAGGAAACAGCTATGA). A MAXIscript Sp6/T7 in vitro transcription kit (Ambion catalog no. 1322) was used as directed by the manufacturer to create [α-³²P]UTP-labeled antisense RNA from the resulting PCR products. SP6 polymerase was used in the in vitro transcription reactions for both probes. Labeled probes were extracted from a 6% acrylamide denaturing gel.

RNase protection assays.

Total RNA was harvested from GAS in the late exponential phase and 2 h into stationary phase of growth by sedimentation through 5.7 M CsCl as previously described (7). RNA was treated with DNase I (Ambion catalog no. 1906) for 3 h at 37°C. RNase protection assays were then performed by using an RPAIII kit (Ambion catalog no. 1414) and RNase T₁ to digest nonhybridized RNA.

RESULTS

Gene expression is affected differently in a covS mutant and a covR mutant.

To study the interaction between CovR and CovS in GAS, we constructed a covS mutant in a JRS4 background by allelic replacement. Overlapping PCR (29) was used to form a linear DNA product consisting of a promoterless cat cassette followed by a ribosome-binding site that was flanked by 5′ and 3′ regions of homology to the CovR/S operon. This PCR product was transformed directly into the M type 6 GAS strain JRS4 (46). Homologous recombination led to replacement of 1,388 bp of the covS open reading frame by cat. The resulting recombinant, strain JRS331, was shown by PCR to have the expected junctions between the introduced marker and the chromosome, and this was confirmed by DNA sequence analysis. The gene downstream of covS was still transcribed in this insertion mutant, as demonstrated by reverse transcription-PCR (data not shown).

In the absence of CovR, colonies appear mucoid due to expression of the hyaluronic acid capsule operon, whose synthetic genes are repressed by CovR (16, 22, 31). However, strain JRS331, the covS mutant, was not mucoid on THY plates (Fig. 1). This indicates that CovS is not required for CovR to repress the has promoter under standard laboratory growth conditions. Since it had previously been reported that a covS mutant, strain D471ΔcsrS, has the same phenotype as a covR mutant (6), we determined the DNA sequence of the covR and covS genes in strain D471ΔcsrS. In addition to the deletion mutation constructed in covS, we found that this strain has a 1-base deletion in covR that results in premature termination of the protein at amino acid 92.

FIG. 1.

CovS mutant is not mucoid. Colonies of JRS4 (wild type), JRS331 (ΔcovS), and JRS948 (ΔcovR) were grown on THY at 37°C overnight.

To determine whether a mutation in covS has an effect on the levels of expression of other CovR-regulated genes, transcript levels for two CovR-repressed genes that are maximally transcribed in the stationary phase were assessed by RNA hybridization. RNA extracted 2 h into the stationary phase from a wild-type strain (JRS4), a CovR mutant (JRS948) (16), and a CovS mutant (JRS331) were hybridized to DNA probes for sagA and sda (also called speMF). As expected, in covR mutant JRS948, the levels of the sagA and sda transcripts were elevated relative to the levels in wild-type strain JRS4. However, no differences in the transcript levels were detected between the CovS mutant strain, JRS331, and the wild-type strain (Fig. 2). Thus, under these growth conditions, CovS did not affect the ability of CovR to repress these virulence genes.

FIG. 2.

Transcription of CovR-regulated genes in the absence of CovS. RNA was harvested 2 h into the stationary phase of growth and hybridized to specific DNA probes. On each filter, the left column contained 4 μg of RNA, and the right column contained 0.5 μg of RNA. Duplicates are arranged vertically. A probe for rpsL (which encodes a ribosomal protein that is not CovR regulated) was used to ensure that equal amounts of RNA were loaded. The data are representative of hybridizations resulting from at least two independent RNA isolations. Densitometry analysis of the blot indicated that the sagA transcript was upregulated 8.9-fold and the sda transcript was upregulated 5.3-fold in the absence of covR. There were no measurable changes in the sagA and sda transcript levels in the covS mutant compared to the wild type. W.T., wild type.

The phosphorylated aspartate residue is essential for CovR repression of Phas in vivo.

Since a CovS mutation does not affect the ability of CovR to repress gene expression under normal laboratory conditions, we wanted to test whether phosphorylation of CovR is necessary for the ability of this molecule to repress transcription. CovR is a member of the OmpR/PhoP family of response regulator proteins, which contain an aspartate residue in the N terminus that is essential for phosphorylation and activity. If this essential aspartate is changed to an alanine, the response regulator cannot be activated to alter gene expression (9, 11, 59). By aligning CovR with other OmpR family response regulator proteins that are inactivated by the aspartate-to-alanine substitution, we identified the aspartate residue in CovR that should be required for phosphorylation of the molecule (Fig. 3). For these experiments, we used strain JRS969, in which covR is deleted and which contains Phas fused to a gus reporter gene at an ectopic chromosomal location (17). Thus, Gus activity could be used to measure repression by CovR at the has promoter. Site-directed mutagenesis was used to insert the D53A substitution codon in a copy of covR that had been cloned into plasmid pLZ12 to obtain pJRS329. A wild-type covR-containing pLZ12 derivative, pJRS996 (Denis, Federle, and Scott, unpublished), was used as a control. Cells were grown in THY-H with 10 mM MgCl₂ to maximize CovR repression of Phas (20). When the cells reached the late exponential phase, they were assayed for Gus activity. Strain JRS969/pJRS996, which had the wild-type covR gene on the plasmid and formed nonmucoid colonies on THY plates containing chloramphenicol, exhibited 10-fold less Gus activity than strain JRS969 exhibited. The level of activity was similar to the level of Gus activity in the CovR⁺ parent strain JRS964 (17), and the results showed that Phas was repressed by the wild-type covR gene on the plasmid. Strain JRS969/pJRS329, containing the covR mutant D53A allele, had Gus levels similar to those of CovR⁻ parent strain JRS969, and colonies were mucoid on THY plates containing chloramphenicol. Thus, a CovR mutant was complemented by wild-type CovR but not by a mutated CovR in which the phosphorylation site had been altered. This supports the hypothesis that phosphorylation of CovR is required for activity in vivo, although CovS is not essential for this activity.

FIG. 3.

Alignment of CovR with homologous response regulators. CovR of S. pyogenes MGAS8232 (GenBank accession number AAL97083.1) and PhoP of B. subtilis 168 (GenBank accession number AAL97083.1CAB14871.1), as well as OmpR, PhoB, and CheY of E. coli K-12 strain MG1655 (GenBank accession numbers AAC76430.1, AAC73502.1, and AAC74952.1, respectively), were aligned by using T-COFFEE (38) and were shaded based on homology by using BOXSHADE. Residues that are not shaded have no identity or similarity to the consensus residues; residues with a black background are identical to the consensus residues, and residues with a gray background are similar, but not identical, to the consensus residues. The aspartate residue essential for phosphorylation is indicated by a vertical arrow.

CovS is required for growth of GAS at low pH.

Based on the strong homology that CovS exhibits with other sensor histidine kinases, we predicted that CovS might be active only under appropriate growth conditions. Thus, we assayed the ability of the covR and covS mutant strains to form colonies at low pH. As expected, both the wild-type and covR mutant strains grew almost as well on plates buffered at pH 6.0 as on THY plates buffered at pH 7.5 (Fig. 4A). However, the covS mutant was unable to form colonies at pH 6.0. In order to be sure that the covS mutation was responsible for this dramatic phenotype, we placed covS under control of the constitutively expressed P23 lactococcal promoter (plasmid pJRS325) and transformed it into covS mutant strain JRS331. The ability to grow at pH 6.0 was fully complemented by pJRS325 in strain JRS331, while JRS331 carrying only the vector (pORI23) (42) was not able to grow (Fig. 4A).

FIG. 4.

Overnight cultures grown at 37°C in THY were plated and incubated for up to 48 h. The results are expressed as the number of CFU per milliliter under stress conditions divided by the number of CFU per milliliter under nonstress conditions and are the means of at least two experiments; the error bars indicate standard deviations. The asterisk indicates that the pH 6.0 survivor colonies were mucoid; when sequenced, they were found to have second-site mutations in covR (see text). The paragraph sign indicates that colonies which were mucoid appeared in only 8 of 21 cultures. E.O.P., efficiency of plating; w.t., wild type; DΔC, D471ΔcsrS.

To determine whether the effect of a nonpolar covS mutation was specific to the M type 6 strain JRS4, the same covS deletion was introduced into strain AM3 (M type 3) (51). The resulting covS mutant was also unable to form colonies at pH 6.0 (data not shown).

When JRS331 was plated at pH 6.0, rare survivors were sometimes seen (Fig. 4A). All of these survivors were mucoid, suggesting that they might have mutations in covR. The sequence of covR was determined for 10 mucoid survivors isolated from independent overnight cultures of JRS331. Each survivor contained a mutated copy of covR, and each mutation was different (Table 1). However, when two survivor colonies from a single pH 6.0 overnight culture were examined, they both had the same covR mutation. This and the low but variable frequency of survivors on the pH 6.0 plates suggest that the covR mutation occurred during overnight growth in standard THY medium and that the mutants were selected by plating the covS strain at pH 6.0.

TABLE 1.

CovR mutants isolated from JRS331 following plating at pH 6.0

Isolate	Type of mutation in covRa	Predicted consequence of mutation
JRS384	R203S (missense)	Reduced DNA bindingb
JRS388	R203H (missense)	Reduced DNA bindingb
JRS1800	G222E (missense)	Reduced DNA bindingb
JRS1804	M1I (missense)	No protein made
JRS398	V18F (missense)	Unknown
JRS1807	V18F (missense)	Unknown
JRS390	E152 (nonsense)c	Early termination
JRS396	K45 (nonsense)c	Early termination
JRS392	Insertion of TATT at nucleotide 24	Early termination
JRS1802	Deletion of nucleotides 112 to 121	Early termination

^aThe designations indicate the wild-type amino acid, its position relative to the start codon, and the amino acid resulting from the mutation.

^bBased on PhoB homology (8).

^cTermination codon.

To distinguish whether the covR covS double mutants grew because they had adapted physiologically to low pH or because of the mutation that occurred in covR, one pH 6.0 survivor with a second-site mutation in covR (JRS396) was tested for growth at pH 6.0. The efficiency of plating of JRS396 was close to 1, indicating that the change that it had acquired was inherited (Fig. 4A). As a confirmation of this, covR covS strain D471ΔcsrS also had an efficiency of plating close to 1 at pH 6.0.

Other environmental stresses inhibit the growth of JRS331.

Two types of stress response in other bacteria have been described. One type is a response to a single specific type of stress, and the other is initiated by many different types of stress signals, such as pH, temperature, high osmolarity, etc. To investigate whether the CovS-regulated stress response was the latter global type, we assayed colony formation at 40°C and on plates with a high salt concentration (0.65 M NaCl instead of 0.03 M NaCl, the normal concentration). We found that both the elevated temperature and the increased salt concentration prevented colony formation by covS mutant JRS331, but both wild-type strain JRS4 and covR mutant strain JRS948 grew efficiently under both stress conditions (Fig. 4B and C). We also tested two covR covS double mutants, and, as expected, they formed colonies efficiently at both 40°C and in the presence of a high salt concentration (Fig. 4B and C). As in the case of acid stress, the plating defects of JRS331 were complemented by the covS wild-type gene on plasmid pJRS325 (Fig. 4B and C). These results indicate that in the presence of CovR, CovS is needed for growth under several stress conditions.

As observed at low pH, the frequency of mucoid colony formation by JRS331 at 40°C varied with the overnight culture used. This was expected since the survivors appeared to be mutants that arose during growth before the culture was plated under stress conditions. Analysis of the DNA sequences of eight of the very rare mucoid colonies that grew at 40°C showed that all of them contained second-site mutations in covR, like the colonies isolated at low pH (Table 2).

TABLE 2.

CovR mutants isolated from JRS331 following plating at 40°C

Isolate	Type of mutation in covRa	Predicted consequence of mutation
JRS342	R203C (missense)	Reduced DNA bindingb
JRS374	T192I (missense)	Interaction with RNA-polymerase disruptedb
JRS340	M1I (missense)	No protein made
JRS372	W184 (nonsense)c	Early termination
JRS370	Y214 (insertion of 1 nucleotide)c	Early termination
JRS376	Deletion of 1 nucleotide	Early termination (following amino acid 49)
JRS1806	Deletion of 1 nucleotide	Early termination (following amino acid 91)
JRS368	Deletion of 356 nucleotides	Lacks promoter and first 21 amino acids of CovR

^aThe designations indicate the wild-type amino acid, its position relative to the start codon, and the amino acid resulting from the mutation.

^bBased on PhoB homology (8).

^cTermination codon.

Survivors were also sometimes seen when JRS331 was plated on medium with a high salt concentration. In this case, only some of the survivors were mucoid. The mucoid survivors grew at a low pH, high temperature, and high osmolarity (data not shown), as observed with other mucoid survivors. Two of the mucoid survivors were sequenced, and like the survivors identified under other conditions, they had a mutation in covR (M1I and L202F). On the other hand, the nonmucoid survivors grew at an efficiency of plating close to 1 when they were plated again on high-salt medium, but they were still sensitive to both low pH and high temperature (data not shown). The covR gene in the three nonmucoid survivors sequenced was the wild-type gene, as expected. This suggests that in addition to the pathway for general stress response, there is an alternative pathway available for a covS mutant to grow under salt stress conditions.

JRS331 is not killed at 40°C.

Because the assay described above required overnight incubation, it did not indicate either the kinetics of growth inhibition of the covS mutant under stress conditions or whether the stressed cells remained viable. To investigate this, cultures of JRS331 and JRS4 were grown in liquid at 37°C, and aliquots were diluted into prewarmed 40°C THY at different times. Growth was monitored with a colorimeter (Fig. 5). When JRS331 cells in the early or mid-log phase were diluted and the temperature was changed from 37 to 40°C, the cell density doubled only once before growth ceased (Fig. 5A); in contrast, all samples of the wild-type JRS4 control incubated at 40°C grew with a slightly greater doubling time than the initial culture at 37°C, but they reached the same final optical density (Fig. 5B). To determine whether the nongrowing covS mutant cells incubated at 40°C were viable, the culture was plated at 37°C at the end of the experiment. The viable counts for each culture were the counts expected for the optical density measured. Thus, this mild stress caused growth inhibition of the covS mutant but did not cause death within the 6 h investigated.

FIG. 5.

Growth of JRS331 and JRS4 cultures shifted from 37 to 40°C. Cultures grown in THY at 37°C (•) were diluted and shifted to 40°C at the early log phase (○) and at the mid-log phase (▴). Growth curves were constructed four times for JRS331 and twice for JRS4.

Phas is derepressed at 40°C.

We found that a covS mutant did not grow under three stress conditions tested in the presence of wild-type covR. This suggests that CovR represses genes required for growth under stress conditions and that CovS inactivates CovR to relieve repression of these genes. If CovS is activated by stress to inactivate CovR, there should be more Phas transcript at 40°C than at 37°C, and this should be CovR dependent. For these experiments, a strain that contained a Phas derivative fused to a gus reporter gene at an ectopic chromosomal location (JRS968) and a covR deletion derivative of this strain (JRS973) were used (17). At 37°C, there was 3.4- ± 0.5-fold more Gus activity in the covR deletion strain than in the wild-type parent, indicating that CovR repressed Phas about threefold. We found that CovR repression of Phas in the wild-type strain was largely alleviated at 40°C: JRS968 had 2.5- ± 0.1-fold more Gus activity at 40°C than at 37°C. This temperature effect on Phas transcription was CovR dependent since there was no difference between the Gus activity of covR deletion strain JRS973 grown at 40°C and the Gus activity of this strain grown at 37°C (1.2- ± 0.2-fold). These results suggest that CovS is activated under stress conditions to inactivate CovR.

To confirm the Gus reporter results, we assayed has transcript levels directly. Total RNA from JRS968 grown at 37 and 40°C was assayed by using RNase protection. As expected, there was more of both the has transcript and the gus transcript at 40°C than at 37°C (5.0- and 3.4-fold increases, respectively) (Fig. 6). Thus, activation of CovS by heat stress led to derepression of stress survival genes and at least one other CovR-regulated gene, has.

FIG. 6.

RNase protection assays. Fifty micrograms of RNA from JRS968 grown to the late exponential phase at either 37 or 40°C was hybridized to [α-³²P]UTP-labeled has and gus probes.

DISCUSSION

CovS is not needed for CovR activation.

The sequences of the genes for the GAS global regulatory system, covR and covS, indicate that their protein products are members of the OmpR/EnvZ two-component signal transduction family. In agreement with this, purified CovR binds DNA within the promoter regions of genes repressed by CovR, and phosphorylation increases the affinity of CovR binding at these specific promoters (17, 35). Because covS lies downstream of covR and the two genes are cotranscribed, it has been assumed that CovS is required for phosphorylation of CovR. Here we provide evidence that CovS is not essential for activation of CovR, since repression of CovR-regulated promoters occurs under normal laboratory growth conditions in the absence of CovS. We also show that a D53A phosphorylation site mutant of CovR cannot be activated to repress Phas, suggesting that phosphorylation is essential for CovR activity in vivo. Therefore, in the absence of the cognate histidine kinase, CovS, CovR must be phosphorylated by something else in the cell. In other two-component signal transduction systems, the response regulator may be activated by phosphorylation in vivo either by noncognate histidine kinases (cross talk) or by acetyl phosphate. For example, OmpR, which is highly homologous to CovR, can interact with DNA in the absence of its cognate sensor kinase, EnvZ. OmpR is phosphorylated by acetyl phosphate and possibly by an unidentified sensor (32). The GAS genome encodes 13 two-component systems (18), only 1 of which has been found to be essential for growth (16; K. S. McIver, personal communication). If activation of CovR in the absence of CovS requires a single alternative kinase that is not essential for growth of GAS, when the covS mutant was plated at a low pH or high temperature, some survivors should have had mutations in the gene encoding this kinase. Instead, we found that all such colonies had covR mutations. Therefore, it seems likely that no single alternative dispensable kinase replaces CovS. This increases the probability that acetyl phosphate has the ability to phosphorylate CovR.

CovS is required for growth under stress conditions.

In this work, we found that CovS is required for growth of GAS under three different conditions of mild stress: 40°C, pH 6.0, and increased salt concentration. During infection of the human host, GAS is known to grow under at least the first two of the conditions which we tested. A fever of 40°C is not uncommon in GAS infections, and GAS grows in the vagina, where the pH is lower than 6.0 (10, 37). Furthermore, during invasive GAS infections, the pH may fall to 6.0 or less following the formation of necrotic lesions or abscesses (47, 48). Increased salt concentrations may play a role in GAS infections of the skin, where the release of perspiration results in increased osmolarity and a decreased pH. However, the fact that the CovS-mediated response is triggered by apparently unrelated stress conditions indicates that CovS mediates a general or global stress response of GAS. In other bacteria, conditions like low oxygen tension, some types of starvation, and sublethal concentrations of antibiotics also trigger the general stress response (2, 4, 41). It seems likely that during infection, GAS strains encounter types of stress which we have not tested that also require the CovS-mediated general stress response for growth.

As in most bacteria, the shared signal that triggers the general stress response is not apparent. What do elevated temperature, lower pH, and high osmolarity have in common that can be recognized by CovS? One possibility is that all of these stresses affect membrane structure and fluidity. In the case of the CpxRA two-component system of E. coli, signal transduction is initiated by various envelope stresses, including accumulation of misfolded proteins, alkaline pH, and disruption of lipopolysaccharide by EDTA (14). In this system, the physical state of membrane lipids also seems to be important (34). Also, in addition to shrinking the cell volume, increased osmotic pressure alters the cytoplasmic membrane, which might affect the structure of a membrane protein like CovS.

This work suggested that the GAS response to a stress signal causes CovS to inactivate CovR, since inactivation of covR by mutation allows growth of the covS mutant. CovR is a global repressor that decreases expression of 15% of the GAS genome (19). It appears, therefore, that among the genes repressed by CovR are genes needed for growth under different conditions of environmental stress. Although the stress response regulon in the GAS has not been investigated yet, some of the genes repressed by CovR show very strong homology to genes in other bacteria required for growth under specific stress conditions. These genes include opuA, encoding the transporter for the glycine-betaine/proline osmoprotectants in such bacteria as Lactococcus lactis and B. subtilis (28, 54, 57), the chaperone genes dnaJ, dnaK, and groEL, and the ATPase genes clpB, clpC, and clpE, which are involved in the heat shock response of many bacteria (3, 27), as well as genes encoding transcriptional regulators, including homologs of the negative regulators of the class I and class III stress genes (12, 30).

CovR also represses genes encoding virulence factors whose selective value for the GAS is not clear. Presumably, the GAS produces these factors to facilitate its own growth, not to harm the human host. Since they are likely to be derepressed in response to environmental stress, it is possible that some of these virulence factors protect the GAS from adverse environmental conditions or allow it to escape from inhospitable local environments. For example, CovR-repressed virulence factors include proteins likely to facilitate escape of the GAS from a local environment by promoting invasion of deeper tissues (e.g., streptokinase and DNase) (10, 50) and the hyaluronic acid capsule which provides a physical buffer that interferes with the entry of salts and H⁺ ions into the GAS. Another group of CovR-regulated virulence factors interacts with host cells in a way that might alter the microenvironment. These factors include streptolysin S, a potent cytolysin, and the proteases encoded by speB and mac (also called ideS) (10, 15, 56). Thus, some of the CovR-repressed virulence factors produced by the GAS may have evolved to allow the GAS to grow in harsh environments within the human host.

Mechanism of CovS response to general stress.

In vitro, CovR binds more efficiently to DNA when it is phosphorylated (17, 35). We show here that in vivo, the D53 residue in CovR that is predicted to be phosphorylated is required for CovR repression of Phas. We therefore propose that stress-induced inactivation of CovR by CovS results from dephosphorylation. We suggest that stress activates the phosphatase activity of CovS and that this results in dephosphorylation of CovR. Dephosphorylated CovR would then dissociate from the promoters to which it is bound, leading to derepression of the promoters. In support of this interpretation, we found increased expression of a CovR-repressed promoter when wild-type GAS cells were grown at 40°C, a general stress response condition.

Two functional classes of sensor kinases in bacteria have been described; the monophasic proteins have kinase activity but no phosphatase activity, and the biphasic proteins have both activities. Recently, partial crystal structure information for sensor kinases led to a description of differences between the two protein classes based on the folding of the catalytic ATP-binding domain (1). The ATP-binding domain of sensor kinases contains a structure known as the ATP lid, which encloses or releases the nucleotide. Although the ATP lid is mostly alpha helical, in bifunctional kinases (EnvZ) this region includes a loop (T loop) that is not alpha helical (Fig. 7A), while the lid of monofunctional kinases (CheA) contains two loops that are not alpha helical separated by a small alpha helical region at this location (Fig. 7B). Homology modeling of the ATP lid region of CovS has shown that its structure is very closely related to the EnvZ structure (21, 40, 45). This model predicts that CovS has a nonhelical T loop within its presumed ATP lid (Fig. 7C). Thus, CovS is predicted to fall into the bifunctional histidine kinase class.

FIG. 7.

CovS is homologous to bifunctional histidine kinases. In panels A and B, alpha helices are orange, and β-sheets are green. (A) ATP binding lid of EnvZ from E. coli (PDB 1BXD), which includes the T loop that is not alpha helical that is characteristic of bifunctional histidine kinases (1). (B) ATP binding lid of CheA from Thermotoga maritima (PDB 1B3Q), which includes the two small non-alpha-helical regions characteristic of the ATP lid of monofunctional kinases. (C) Homology modeling of CovS (blue) with its closest relative, EnvZ (gray) (21, 40, 45).

Distinctive features of the CovR/S stress response.

Instead of the positive regulation of the general stress regulon found in most bacteria, in the CovR/S system of GAS there is a double-negative regulatory mechanism. Transcription of the covRS operon is itself repressed by CovR (16), so as long as CovR is active, very little new CovR protein is made. Therefore, another consequence of stress-induced inactivation of CovR is derepression of CovR synthesis. It can be expected that under stress conditions, more inactive CovR accumulates so that when stress is relieved and CovR becomes active, there is a rapid repression response. The design of this system thus makes it very sensitive to environmental change. Since CovR-regulated promoters may vary in their degrees of sensitivity to CovR phosphorylation, CovR-regulated genes may exhibit differential expression under both normal and stress conditions and may respond to various degrees when the GAS encounters stress conditions. At the capsule gene promoter, Phas, CovR must bind to at least four sites to give full repression in vivo (17). Other promoters, which have fewer CovR binding sites, may show different sensitivities to phosphorylation of CovR. The number, location, and orientation of CovR binding sites at each promoter might also affect the degree of the response to environmental change.

In most bacteria, transcription of the general stress regulon genes requires an alternative sigma factor. The only alternative sigma factor in GAS is sigX. However, we found that inactivation of both copies of sigX in the JRS4-derived M6 GAS strain JOS21 (39) had no effect on growth at 40°C (data not shown). Instead of requiring an alternative sigma factor for transcription, the stress response genes of GAS are repressed by bound active CovR, and stress leads to inactivation of this repressor. This mechanism may allow a greater range of responses of stress-regulated gene expression than an alternative sigma factor could provide. In bacteria with a stress response sigma factor, the absence of the factor would be expected to prevent recognition of the regulated promoter by RNA polymerase, resulting in a complete lack of transcription of the stress regulon. However, most promoters controlled by a repressor show a low constitutive level of transcription even when the repressor is active. Thus, for CovR-repressed genes, even in the absence of stress, there should be a low level of stress gene products. This may facilitate growth of the GAS when it first encounters stressful conditions.

Although most bacteria respond to environmental change by altering the specificity of RNA polymerase through synthesis of alternative sigma factors, both Streptococcus pneumoniae and Streptococcus mutans are like the GAS in that their genomes do not encode an alternative sigma factor. However, both of these organisms have genes with significant homology to covR and covS. In S. mutans, the gcrR gene product exhibits the greatest homology to CovR (84% similarity and 75% identity), and, like CovR, GcrR is inactivated by several stress conditions (44). Although the mechanism for this and the extent of the GcrR regulon are unknown, the Gcr system may be similar to CovR/S. We predict that two-component systems similar to CovR/S are used by both S. pneumoniae and S. mutans and possibly by other organisms to mediate the general stress response.