A Two-Component Regulatory System, CsrR-CsrS, Represses Expression of Three Streptococcus pyogenes Virulence Factors, Hyaluronic Acid Capsule, Streptolysin S, and Pyrogenic Exotoxin B

Abstract

Certain Tn916 insertions in the chromosome of an M1-type, nonmucoid Streptococcus pyogenes isolate (MGAS166) were previously shown to result in stable mucoidy with increased expression of the capsular synthetic genes. The transposon insertions in these strains are directly upstream of an apparent operon encoding a two-component regulatory system, designated csrR-csrS. Compared with MGAS166, these mucoid mutants are more hemolytic and cause significantly more tissue damage in a murine model of skin infection. To extend these observations, we constructed an in-frame deletion in the gene encoding the response regulator, csrR, and we evaluated the expression of other known S. pyogenes virulence factors. We discovered that csrR mutants have enhanced transcription of sagA, a gene associated with streptolysin S (SLS) and speB, the gene encoding pyrogenic exotoxin B (SpeB). The mutants also express substantially higher SLS activity and SpeB antigen in late-exponential-phase cultures. There is no change in expression of emm, scpA, sic, or cpa (genes encoding other S. pyogenes virulence factors). CsrR⁻ strains but not the wild-type parental strain produce necrotizing lesions in a mouse model of subcutaneous infection. A double mutant with deletions in both csrR and the capsular synthesis genes caused fewer and smaller necrotic skin lesions than the csrR mutants. However, this nonmucoid csrR strain was more likely than the wild type to yield necrotic lesions, suggesting that mucoidy contributes to virulence in this model of infection but that there are other csrR-regulated factors involved in the production of necrotic lesions.

Group A streptococci (GAS) may be carried asymptomatically or cause only mild self-limited disease on the skin or mucosal surfaces, but they may also invade deeper tissues and be extremely destructive. GAS infections involving the subcutaneous fascial planes (necrotizing fasciitis), muscles, deep pharyngeal spaces, lungs, or the bloodstream have dire and often fatal consequences. The hallmark of these invasive infections is aggressive spread through tissue and cellular necrosis. In many cases, systemic intoxication with a toxic shock syndrome accompanies local manifestations of infection. To produce these manifestations of invasive disease, GAS possess a large repertoire of known and suspected virulence factors, including M protein, hyaluronic acid capsule, pyrogenic exotoxins A, B (SpeB), and C, streptolysins S and O (SLS and SLO, respectively), streptococcal inhibitor of complement (SIC), C5a peptidase, proteins F and Cpa, streptokinase, hyaluronidase, DNAse, and many others.

With the application of genetic techniques to the analysis of GAS virulence in recent years, there has been considerable progress in understanding the roles that these factors play in streptococcal infection and disease. Some of these factors may be required for successful adherence and colonization of the host at various sites (e.g., lipoteichoic acid, capsule, and protein F). Some may protect GAS from constitutive host defenses (e.g., M protein, capsule, protein G, SIC, and C5a peptidase). Still others may function primarily to destroy tissues and allow streptococci to spread through the host (e.g., streptolysins and hyaluronidase). Thus, GAS possess a large and diverse armamentarium of defensive and offensive weapons which may be adaptive in certain situations. Given this diversity of function, it seems reasonable to assume that the bacteria must be able to sense their microenvironment and to deploy only those virulence factors that function to its advantage in that particular niche. In many pathogenic bacteria, this optimal deployment is accomplished by coordinately regulating the expression of sets of virulence-related factors.

The earliest coordinately regulated network of virulence factors described in GAS followed the discovery of a positive regulator of M protein expression (27, 31). This DNA-binding protein, now designated as Mga (33), contains sequences that are commonly conserved among the response regulators of two-component systems. It was subsequently shown that Mga also regulates the expression of SIC, C5a peptidase, and other potential virulence-associated proteins, but it does not affect the expression of hyaluronic acid capsule or streptokinase (9, 30). Another positive virulence regulator, designated rofA, was found to control expression of prtF, the gene encoding the fibronectin-binding adhesin, protein F (15). The rofA gene has limited homology with mga. Recently, we reported that a distinct, two-component regulatory system (designated mucRS) represses expression of the hyaluronic acid capsule in an M1 Streptococcus pyogenes strain (16a). Mutations that inactivate mucRS result in highly mucoid strains that are hypervirulent in a mouse model of soft tissue infection (5, 16). A virtually identical locus was identified in an M3 strain by Levin and Wessels; the DNA sequence was reported and designated csrRS (21). Mutations in csrRS also render mutant strains both hypermucoid and hypervirulent. Because of the striking degree of identity between mucRS and csrRS, we have adopted the gene designation of Levin and Wessels. Henceforth, the genes that we have characterized in the M1 background will also be designated as csrR and csrS.

In the present work, we demonstrate that csrRS also represses the expression of SLS and the cysteine protease, SpeB. We also demonstrate that although the increased mucoidy of csrRS mutants contributes to virulence, other factors regulated by this locus also contribute to the enhanced virulence of these mutants.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The streptococcal strains used for this study were derived from wild-type strain MGAS166 (Table 1). Strains were grown in Todd-Hewitt broth supplemented with 0.2% yeast extract (Difco, Detroit, Mich.) or on Todd-Hewitt agar plates (Difco). When antibiotic selection of streptococci was required, 0.02 μg of erythromycin per ml, 100 μg of streptomycin per ml, or 100 μg of spectinomycin (Sigma Laboratories, St. Louis, Mo.) per ml were added to the appropriate media. Escherichia coli JM109 or DH5α was used for cloning procedures. Ampicillin (50 μg/ml) or erythromycin (300 μg/ml) was added to Luria-Bertani media as needed for selection of clones.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Genotype or description	Reference
Strains
MGAS166	M1 SpeA2, streptomycin resistant	25
SBmuc5	Tn916::csrR	5
SBmuc7	Tn916::csrR (reverse orientation of SBmuc5)	5
UMAA2392	MGAS166 ΔcsrR csrS (single base pair change in start codon [ATG→ACG])	This study
NLB2226	MGAS166 (pNLB2223)	16
UMAA2497	MGAS166 ΔhasAB	This study
UMAA2526	UMAA2392 ΔhasAB	This study
UMAA2499	SBmuc5 (pASH2477)	This study
UMAA2353	SBmuc7 (pNLB2223)	This study
UMAA2561	UMAA2392 (pNLB2223)	This study
Plasmids
pJRS233	emR, pGHOST4+ oriV(Ts) (in streptococci)	28
pLZ12spec	aad9, shuttle vector	17
pJRS525	aad9, shuttle vector with MCS and lacZ	23
pNLB2223	pLZ12spec carrying csrR	16
pASH2477	pLZ12spec carrying csrRS	This study
pASH2233	pJRS233 carrying ΔcsrR	This study
pASH2468	pJRS233 carrying ΔhasAB	This study

The initial hypermucoid CsrR⁻ mutants of strain MGAS166 were isolated by transposon mutagenesis with Tn916 (Table 1). In-frame, site-directed deletions in MGAS166 were generated by using a method of plasmid insertional mutagenesis developed by Perez-Casal and colleagues (28). Briefly, in-frame deletions of target genes were constructed and inserted into pJRS233. This plasmid confers erythromycin resistance but depends on a temperature-sensitive origin of replication for maintenance in gram-positive organisms. Insertions of the plasmid onto the streptococcal chromosome were selected by growth at the nonpermissive temperature on erythromycin. After passage of cointegrates under permissive, nonselective conditions, colonies were screened for loss of the phenotypic changes associated with the targeted gene (e.g., change in mucoidy) or by PCR analysis when a phenotypic change was not easily detected. A deletion in the csrR gene was generated by ligating amplicons of PCR primer pairs 5′-GGCCGGCTGCAGTACTTGCTATTCCGCTACAGGTC-3′ and 5′-GCCGCCGAATTCTCTAACCCTTCACGACCATTGAC-3′ (upstream) and 5′-GGGCCCGAATTCTGAAGCCGTTGAGACTAATGTTGT-3′ and 5′-CCGCCGGGATCCGATAGGACCATGCAAGCCAGGAG-3′ (downstream) at an EcoRI site and cloning between the PstI and XbaI sites of pJRS233. Ligation of these two fragments generated a copy of csrR with a 441-bp in-frame deletion. In the translated protein, 147 residues are deleted from the native protein of 227 residues (from codons 40 to 187).

The principal capsular synthesis genes are encoded in GAS by the hasABC locus (1, 12, 13). A 1,248-bp deletion in hasAB was constructed by using the same method employed for the csrR deletion. Amplicons were generated using PCR primer pairs 5′-GGGCCCGAGCTCGTTATCGTTCACCGTTCCCTTGTC-3′ and 5′-GCCGCCGTCTAGAAGGCAACGATGGGATTAGA-3′ (upstream) and 5′-GGCCGGGTCTAGAGAGTCCCCAGTAAAAGTAGTCG-3′ and 5′-CCGCCGGCATGCCGCTTCTTCGACGATAAACTGG-3′ (downstream). These amplicons were ligated at a XbaI site and cloned into the HindIII site of pJRS233. The deletion fuses codon 297 of hasA to codon 305 of hasB in the translated product, deleting 99 residues of the C terminus of hasA. Plasmid clones containing the deleted genes were confirmed with a restriction enzyme. Allelic exchanges were confirmed by DNA sequencing of PCR products. Plasmids used for generation of site-directed deletions and complementation experiments are described in Table 1.

Uronic acid assay.

Isolates were grown in Todd-Hewitt broth (150 ml) to an absorbance at 600 nm (A₆₀₀) of 0.6 to 0.8. Aliquots of bacterial culture were removed hourly, washed once with sterile distilled water, and centrifuged. The pellet was resuspended in 0.5 ml of water. Chloroform (1.0 ml) was added, and the suspension was mixed vigorously at room temperature for 1 h. After centrifugation, the stain-all method of uronic acid quantitation was performed by using the aqueous phase (19). Hyaluronic acid from Streptococcus zooepidemicus (Sigma) was used as a standard. The peak uronic acid level obtained during the 6-h culture is reported.

Northern hybridization and immunoblotting.

RNA was isolated by using the method of Cheung and colleagues (10) and the FastPrep system (Bio101, Vista, Calif.). Bacteria were shaken with beads for 45 s at 6,000 rpm to lyse cells in CRSR 455 (chaotropic RNA stabilizing reagent 455). After extraction and precipitation of the fraction containing RNA, samples were loaded immediately onto a 2% formaldehyde gel, electrophoresed, and transferred to nylon membranes by capillary action. The membrane was hybridized overnight at 55°C in high sodium dodecyl sulfate (SDS) buffer with digoxigenin-labeled DNA probes (Genius system; Boehringer Mannheim). Hybridizing RNA was visualized with anti-digoxigenin alkaline phosphatase conjugate. Probes were generated with the following primer sets: 5′-CGTTTCTCTTGAGCTGCAACCTG-3′ and 5′-CAACATTAGTCTCAACGGCTTCATC-3′ for csrR, 5′-AATTGAGCTAGCCTTGTCCTTGTT-3′ and 5′-ATAACTTCCGCTACCACCTTGAGA-3′ for sagA, 5′-CCAATGTACCGTTAAAAGCAAATG-3′ and 5′-TGCATTTCCATACTAAGGTTTGA-3′ for speB, and 5′-TTTTTAATGATCTTCGCTTTGGTAAC-3′ and 5′-GTCATGTTTTCTTATCCTTATCGTGTG-3′ for cpa.

For immunoblotting, 0.5 ml of culture supernatant and 5 μl of β-mercaptoethanol were concentrated and washed twice in a Microcon-10 microconcentrator (molecular weight cutoff, 10 kDa). The final retentate was dissolved in 20 μl of 1× SDS-polyacrylamide gel electrophoresis loading buffer and loaded onto a SDS–15% polyacrylamide gel. After electrotransfer to membranes, blots were incubated with rabbit anti-SpeB (ToxinTech, Sarasota, Fla.) at a concentration of 1:1,000 and then with goat anti-rabbit immunoglobulin alkaline phosphatase conjugate and developed with a colorigenic substrate (7).

SLS assay.

Culture supernatants were collected at various times after inoculation and a series of twofold dilutions were added to equal volumes of 2% defibrinated, washed rabbit erythrocytes. After 60 min of incubation at 37°C, the samples were centrifuged and the absorbance at 540 nm was measured to determine the release of free hemoglobin. An equivalent amount of fully hemolyzed erythrocytes was used as a positive control. In some experiments, bacteria were initially grown in the presence of 10% horse serum to permit the release of cell-associated SLS. Cholesterol was added to some serum-free samples at a concentration of 10 μg/ml to inhibit SLO, and trypan blue was added to other samples to inhibit SLS.

Mouse infection model.

Virulence of GAS strains was determined by using a dermonecrotic mouse model as previously described (7). Briefly, streptococci were harvested at mid-log phase and concentrated to produce various inocula in a 200-μl bacterial suspension. This volume was injected subcutaneously into the right flank of hairless, 4-week-old male crl:SKH1(hrhr) Br mice (Charles River, Wilmington, Mass.). Mice were weighed prior to inoculation every 24 h. Necrotic wounds were measured daily using the following equation: area = π(L × W)/2, where L is the long axis and W is the short axis of the lesion.

RESULTS

Construction of csrR mutant strains.

To test the role of the csrRS locus in the regulation of virulence-associated genes, we constructed derivatives of a clinical isolate, MGAS166, which have a defective expression of csrR. Two mutants were obtained by screening transposon mutants for colonial hypermucoidy. SBmuc5 and SBmuc7 are derivatives of MGAS166 that have insertions of Tn916 upstream of csrR. The insertions are in opposite orientations in the two strains; both interrupt the −35 consensus sequence of a putative promoter.

Since the transposon insertions involved a regulatory sequence upstream of csrR rather than the structural gene, we also constructed strain UMAA2392, a strain which contains an in-frame csrR deletion in MGAS166 introduced by the in vivo recombination techniques described in Materials and Methods (Table 1). Fortuitously, the sequence of csrRS in this deletion strain also revealed a spontaneous, single base pair substitution in the csrS open reading frame that resulted in an alteration in its start codon (from ATG to ACG). To study independently the effects of hyaluronic acid capsule production in this mutant, we introduced in-frame deletions into the hasAB operon of both strains MGAS166 and UMAA2392 to generate strains UMAA2497 and UMAA2526, respectively. The capsular phenotype of each mutant strain was determined by the uronic acid assay (Table 2).

TABLE 2.

Levels of cell-associated uronic acids from GAS strain MGAS166 and derivative strains

Strain	Phenotype	Peak uronic acid level (μg/ml)
MGAS166	Wild type	1.0
SBmuc5	Tn916::csrR	5.2
SBmuc7	Tn916::csrR	5.5
UMAA2392	ΔcsrR csrS	4.3
UMAA2497	ΔhasAB	1.1
UMAA2526	ΔcsrR csrS ΔhasAB	0.9

Transcriptional analysis of CsrRS-regulated genes.

To assess the transcription of virulence-associated genes in csrR mutants, cultures of wild-type and mutant strains were grown to late exponential or early stationary phase for isolation of bacterial RNA. Northern hybridization analysis of MGAS166 and the three csrR mutants revealed that transcription of csrR was detected in the wild-type strain but not in the mutants (Fig. 1). csrR transcripts could be detected in the two Tn916 mutants after introduction of a plasmid carrying an intact csrR gene (pNLB2223) or the entire csrRS locus (pASH2477).

FIG. 1.

Northern hybridization analysis of virulence-associated genes in CsrR⁻ streptococcal strains. Bacterial RNAs from six strains were isolated in late exponential phase and hybridized with four different PCR-generated probes as indicated to the left of each panel. The source of each RNA sample is indicated above the respective lane. The fifth and sixth lanes in each panel contain RNA from derivatives of SBmuc5 and SBmuc7, respectively, into which a csrRS-bearing plasmid (pNLB2223) has been introduced. wt, wild type.

Using probes specific for an SLS-associated gene (sagA) and the gene encoding SpeB (speB), we found that transcripts of these genes in late exponential phase (A₆₀₀ = 1.0) were increased in the csrR mutants, compared to the wild-type strain, MGAS166 (Fig. 1). Reintroduction of csrR into the two csrR strains resulted in reduction of sagA and speB transcription to wild-type levels. These findings suggest that csrR mutations result in derepression of sagA and speB. In contrast, csrR expression did not affect the expression of cpa, an analog of streptococcal protein F from M1 strains. A report by Podbielski and colleagues describes the three cpa transcripts from this locus, which most likely result from alternative termination (29). Although we observe some variability in the relative abundance of the three transcripts among our strains, the total cpa message appears to be similar in all strains, suggesting that transcription from the cpa promoter is not altered in our mutants. Similarly, there was no effect of a csrR mutation on the expression of mga, emm, scpA, or sic (data not shown).

SpeB expression has been reported to increase in association with nutrient depletion, specifically at lower concentrations of glucose (8). We considered the possibility that a strain hyperexpressing capsule might deplete nutrients from the media faster than the wild-type strain. We reasoned that nutrients would not be more rapidly depleted in a CsrR⁻ strain that cannot make capsule. To assess whether excess capsule production contributes to the derepression of sagA or speB, we compared mRNAs from strain MGAS166, strain UMAA2392 (ΔcsrR csrS), and strain UMAA2526, a derivative of UMAA2392 that also possesses a deletion in hasAB. Since the depletion of glucose had been suggested as a signal for speB expression (8), we also included a sample of RNA from a culture of strain UMAA2392 to which glucose had been added 2 h after inoculation. Our results show that MGAS166 had minimal sagA and no detectable speB messages in these late exponential (A₆₀₀ = 1.0) samples (Fig. 2). Transcription of both messages was increased in UMAA2392, and there was no decrease in transcription either in the presence or absence of HasAB or with the addition of glucose. Message-encoding cpa was the same in all three strains grown under similar culture conditions. The addition of glucose reduced cpa transcription in strain UMAA2392, but the significance of this observation is unclear since the wild-type strain was not also examined under these growth conditions. We concluded that sagA is not indirectly regulated via a depletion of nutrients in the media, and speB expression is not repressed by these conditions. The increase in speB transcription in strain UMAA2526 (ΔcsrR csrS ΔhasAB) was reproducible in other experiments (data not shown). The cause of this increase is not known, but the enhanced transcription that we observed is contrary to the prediction of Chaussee et al. (8).

FIG. 2.

Northern hybridization analysis of virulence-associated genes in CsrR⁻ streptococcal strains. Samples from early-stationary-phase cultures of three strains were hybridized to probes specific for sagA, speB, and cpa. Samples included RNA from MGAS166, UMAA2392 (ΔcsrR csrS), UMAA2526 (ΔcsrR csrS ΔhasAB), and UMAA2392 with glucose added to a final concentration of 0.2% 2 h before RNA isolation.

Expression of SLS in CsrR mutants.

Betschel and colleagues demonstrated that a Tn916 insertion in sagA results in a complete loss of SLS activity (6). We therefore hypothesized that increased sagA expression observed in csrR mutants would be associated with increased SLS activity. To confirm this hypothesis, we performed hemolytic assays of various mutants. Assays were performed on supernatants of cultures sampled at various intervals between mid-exponential and stationary phase (Fig. 3A). Activities of SLS and SLO were distinguished by assaying aliquots in the presence of either cholesterol (to inhibit SLO) or trypan blue (to inhibit SLS). Assays for SLS were performed with horse serum, which acts as a carrier and enhances release of cell-associated toxin (3). Our experiments showed that SLS is secreted during stationary phase in all of the pathogenic strains studied. However, CsrR⁻ strain UMAA2392 secreted large quantities of SLS 2 h earlier in broth culture than its parental CsrR⁺ strain, MGAS166. In mid- to late exponential phase, SLS activity was fourfold greater in csrR mutants than in the wild-type strain (Fig. 3B). This observation is consistent with the enhanced transcription of sagA during this point in the growth curve, as reported above. In contrast, the activity of SLO was comparable in the wild-type and csrR mutants at all time points measured (Fig. 3C).

FIG. 3.

Hemolytic activities of streptococcal supernatants during 11 h of broth culture. (A) Absorbance of cultures from which aliquots were assayed. (B) Assays for SLS, performed in the presence of cholesterol to inhibit SLO. (C) Assays for SLO, performed in the presence of trypan blue to inhibit SLS. Note the different scales used in panels B and C. Data points for strain MGAS166 (wild type) are represented by the circles; those for strain UMAA2392 (ΔcsrR csrS) are represented by squares. White (open) and black (closed) data points indicate that assays were done in the absence or presence of horse serum, respectively. Assays were also performed without cholesterol and trypan blue and were within ±10% of the sum of the SLS and SLO activities (data not shown).

Expression of SpeB in CsrR mutants.

Just as the increased activity of SLS was shown to follow increased transcription of sagA, we wished to confirm that increased expression of SpeB would follow after derepression of speB. Expression of SpeB in the culture supernatants of wild-type and mutant strains was assessed by immunoblot analysis with a monospecific polyclonal antibody (Fig. 4). None of the strains tested produced detectable SpeB after 6 h of broth culture (mid-exponential phase). In contrast, by 12 h of culture (early stationary phase in this experiment), SpeB or precursor forms of SpeB were detected in the culture of the three csrR mutants but not in the wild type. Introduction of pNLB2333 (csrR⁺) into strains SBmuc5 and SBmuc7 abrogated expression of SpeB at this time point. Finally, by late stationary phase, all strains were expressing fully processed SpeB, with strain UMAA2392 (ΔcsrR csrS) expressing the greatest amount. The expression pattern of SpeB was therefore similar to that of SLS. Both proteins are expressed in the wild-type strain during stationary phase, but secretion of these proteins occurs earlier in the growth curve in CsrR⁻ strains.

FIG. 4.

Immunoblot analysis of streptococcal strains with a polyclonal antisera against SpeB. The three panels represent samples taken at different time intervals during growth in broth culture: mid-exponential (6 h of culture) (A), late exponential (12 h of culture) (B), and late stationary phase (48 h of culture) (C). Lanes were loaded with samples from the following strains: lane 1, MGAS166; lane 2, SBmuc5 (csrR::Tn916); lane 3, SBmuc7 (csrR::Tn916); lane 4, UMAA2392 (ΔcsrR csrS); lane 5, SBmuc5 (pNLB2223); and lane 6, SBmuc7 (pNLB2223). Lane M, molecular weight markers. The arrows indicate the mobility of 28-kDa proteins, corresponding to the molecular mass of fully processed SpeB. The slower-migrating bands that coexist in specimens containing specific SpeB bands are thought to represent unprocessed or complexed SpeB.

Virulence of CsrR mutants in mouse skin and soft tissue infection.

Virulence in the hairless mouse model was assessed by four criteria: weight loss following inoculation, frequency of lesion development, frequency of lesional necrosis, and size of necrotic lesions. Strain MGAS166 is known to produce necrotic lesions in these mice when injected with Cytodex beads (Sigma) (7); without beads, this strain causes minimal or no erythema at the injection site. The three csrR mutants were capable of producing abscess, skin necrosis, or both in the absence of Cytodex (Table 3). Using an inoculum of 2 × 10⁶ CFU grown under aerated conditions, strain MGAS166 produced no manifestations of infection in six mice, whereas strain UMAA2392 (ΔcsrR csrS) yielded dermonecrotic lesions in all animals (P = 0.002, Fisher’s exact test; experiment II [Table 3]). A similar inoculum of MGAS166 grown in static broth (a condition known to enhance the virulence of this strain) yielded subcutaneous abscesses in 4 of 12 animals but no necrotic lesions. In contrast, the ΔcsrR csrS mutant yielded necrotic lesions in all of 12 animals and was therefore significantly more likely to produce any lesion (P = 0.001) or a dermonecrotic lesion, specifically (P = 0.000007). At 10-fold lower inocula, the ΔcsrR csrS mutant produced necrotic lesions in half of the animals, regardless of the culture conditions used, although the difference from the less virulent wild-type strain was not statistically significant because of the small size of the experimental groups. However, these experiments suggest that the approximate dose of streptococci that produces lesional necrosis in 50% of the animals for the ΔcsrR csrS mutant is 2 × 10⁵ to 3 × 10⁵. The dose that produces lesional necrosis in 50% of the animals for MGAS166 is unknown, but it must be greater than 4 × 10⁶, at least 10-fold greater than the mutant.

TABLE 3.

Weight loss and lesion development in mice inoculated subcutaneously with GAS^a

Experiment (inoculum)	Strain or description	Mean 24-h weight gain ± SD	No. with lesions/no. inoculatedb	No. with necrosis/no. with lesionsb
I (3 × 105)	MGAS166 (wt)	1.3 ± 0.4	0/6	0/0
	csrR::Tn916 (muc5)	0.6 ± 0.5	1/6	0/1
	ΔcsrR csrS	−1.7 ± 1.6	4/6	3/4
	ΔcsrR csrS ΔhasAB	−0.1 ± 0.9	5/6	1/5
II (2 × 106)	Controlc	0.6 ± 0.07	0/6	0/0
	MGAS166 (wt)	−0.4 ± 1.2	0/6	0/0
	csrR::Tn916 (muc5)	−3.7 ± 0.8	6/6	6/6
	ΔcsrR csrS	−4.3 ± 0.6	6/6	6/6
	ΔcsrR csrS ΔhasAB	−3.4 ± 0.8	6/6	5/6
III (2 × 105)	Controld	0.5 ± 0.6	0/8	0/0
	MGAS166 (wt)	0.9 ± 0.4	0/6	0/0
	csrR::Tn916 (muc5)	0.2 ± 1.4	0/7	0/0
	ΔcsrR csrS	−2.1 ± 1.7	3/6	3/3
	ΔcsrR csrS ΔhasAB	0.4 ± 1.3	1/6	0/1
IV (4 × 106)	Controlc	1.2 ± 0.4	0/6	0/0
	MGAS166 (wt)	1.1 ± 0.3	4/12	0/4
	csrR::Tn916 (muc5)	−3.1 ± 0.6	6/6	6/6
	csrR::Tn916 (muc7)	−3.3 ± 0.6	6/6	6/6
	ΔcsrR csrS	−3.7 ± 0.4	12/12	12/12
	ΔcsrR csrS ΔhasAB	−1.6 ± 0.9	6/6	4/6

^aAll bacteria were grown in Todd-Hewitt broth at 37°C to mid-exponential phase (A₆₀₀ = 0.6 to 0.7). Culture conditions were as follows: experiment I, shaken and aerated; experiment II, shaken and aerated; experiment III, static and capped; experiment IV, static and capped. wt, wild type. 

^bCounts taken at 72 h. 

^cControl mice were uninoculated mice from the same cohort. 

^dControl mice received subcutaneous injections of THY broth. 

At higher inocula, the transposon mutants, SBmuc5 and SBmuc7, were as virulent as the ΔcsrR csrS mutant; however, SBmuc5 was less capable of producing lesions at the lower inocula (SBmuc7 was not tested). Strain UMAA2526, which has both the ΔcsrR csrS mutations and an in-frame deletion in hasAB, exhibits a nonmucoid phenotype but has enhanced expression of SLS and SpeB. Virulence of this strain was intermediate between MGAS166 and the ΔcsrR csrS mutant. Loss of capsule in the csrR mutant background resulted in a slightly lower frequency of lesional necrosis and less 24-h weight loss than in strain UMAA2392 (ΔcsrR csrS hasAB⁺), in some experiments. This observation suggests that elaboration of capsular polysaccharide is partly responsible for the hypervirulence of the csrR mutant but does not entirely account for enhanced virulence.

The difference in the virulence of encapsulated and nonencapsulated csrR mutants was also apparent by measuring the area of necrotic lesions in those animals that had them. The encapsulated strain produces larger lesions earlier in the course of infection than the unencapsulated strain (Fig. 5). The ΔcsrR csrS mutant also produces larger lesions than the transposon insertion mutant, SBmuc5.

FIG. 5.

Comparison of the size of necrotic lesions during the course of experiments II and IV (Table 3). In these experiments, necrotic lesions developed in all infected mice except in those inoculated with the wild-type strain, MGAS166 (no lesions), and in three of the mice inoculated with strain UMAA2526 (ΔcsrR csrS ΔhasAB). One of six mice inoculated with UMAA2526 in experiment II (A) and two of six mice in experiment IV (B) had no skin necrosis and are excluded from the analysis of lesion size in this figure. Lesions were measured, and the area was calculated 24, 48, and 72 h after inoculation. The infecting strains were SBmuc5 (solid black columns), SBmuc7 (white columns), UMAA2392 (cross-hatched columns), and UMAA2526 (shaded columns).

These findings suggest that csrR-regulated factors, other than hyaluronic acid capsule, also contribute to virulence in this model. To characterize the nature of these additional virulence factors, we conducted an experiment to determine whether their effect in pathogenesis is intrinsic or extrinsic to the bacteria. In this context, intrinsic factors are cell-associated factors that enhance the fitness only of those bacteria that express them (e.g., capsule and M protein), whereas extrinsic factors are those that act directly on tissue components and modify the microenvironment to favor the growth and spread of the bacteria (e.g., toxins). The capsule is a cell-associated factor that would provide little or no benefit to nonencapsulated bacteria in a mixed inoculum of mucoid and nonmucoid strains. However, if a mucoid CsrR mutant enhances the survival of nonmucoid wild-type strains in such an inoculum, we could conclude that extrinsic factors regulated by CsrR are responsible for this effect. We therefore devised an experiment in which MGAS166 was inoculated together with an equivalent number of strain SBmuc7 under conditions in which MGAS166 would normally be cleared from the tissues. The six mice inoculated with this mixture developed necrotic skin lesions, and they lost a mean of 2.9 g (±1.0 g) of weight. The lesions were removed, emulsified in a tissue grinder, and plated on Todd-Hewitt agar plates. A sample of each colony type was also replated on media containing tetracycline to confirm the presence of Tn916 in the mucoid (SBmuc7) but not the nonmucoid (MGAS166) colonies. No spontaneous hypermucoid variants of MGAS166 were detected in this experiment. We calculated that SBmuc7 increased ∼200-fold relative to the initial inoculum, whereas MGAS166 increased ∼5-fold. Since MGAS166 is typically cleared completely when inoculated at this level, we interpreted the increase in the mixed experiment as evidence of an extrinsic effect of the csrR mutant that facilitates growth of the wild type in tissue. This finding is consistent with the hypothesis that toxins, such as SLS and SpeB, may be responsible for the hypervirulence of csrR mutants.

DISCUSSION

We and others have previously shown that the csrRS locus represses the expression of hyaluronic acid capsule in GAS (16, 21). The data presented in this report show that csrRS also repress the expression of at least two other potential virulence factors, SLS and SpeB. Using a functional assay and an immunoassay, respectively, we were able to detect enhanced levels of both of these proteins in late exponential phase when csrR expression was disrupted. In addition, Northern hybridization analysis confirmed that csrR-mediated regulation occurs at least in part at the transcriptional level. Mutations in csrR had no effect on expression of cpa, mga, or the genes regulated by mga. Therefore, it appears that this two-component regulatory system, csrRS, controls expression of a subset of known virulence factors. While this paper was being reviewed, another laboratory published a description of a two-component regulatory system that was identified from the genomic sequence of GAS and was found to repress transcription of sagA, hasA, and the genes encoding streptokinase and mitogenic factor (14). This regulatory locus is the same gene tandem described in the present work.

Mutations affecting the expression of csrR and csrS also result in enhanced virulence in an animal model of skin and soft tissue infection. A strain with a deletion in csrR and a point mutation in the start codon of csrS was at least 10-fold more virulent than the parental strain. Like the transposon mutants that were previously described (16, 21), this mutant strain has increased quantities of uronic acid in broth-cultured bacteria and a mucoid appearance of colonies on agar plates. The mutant strain is more virulent, due in large part to the increased production of hyaluronic acid capsule. We demonstrated the importance of the capsule by infecting mice with a strain that contains both the ΔcsrR csrS mutations and a deletion of the capsular synthesis genes. The capacity of this strain to produce a necrotic skin lesion was intermediate between the wild type and the csrR deletion strain as measured by the occurrence of lesions, the frequency of necrosis, and the lesional size, when present. The csrR transposon mutants were not consistently as virulent as the strain carrying the csrR deletion, perhaps because these strains have insertions in the promoter of csrR rather than the open reading frame. The csrR mutation may be leaky in these strains.

Streptococcal capsule was first proposed to be a virulence factor by Kass and Seastone (18). Potential roles for the capsule in GAS virulence are suggested by studies demonstrating its capacity to prevent killing by phagocytes (24, 35), to prevent attachment of bacteria to macrophages (36), to shield the organism from oxygen metabolites (11), and, most recently, to act as a ligand for CD44 on human keratinocytes (32). The importance of capsule in the pathophysiology of streptococcal skin infection was recently demonstrated by Ashbaugh et al. (4). These authors demonstrated that GAS strains with deletions in either the capsular synthesis genes (hasABC) or the M protein gene (emm) cannot induce necrotic skin lesions in a murine infection model (4).

In our previous animal infection experiments with clinical strain MGAS166, it was necessary to include a suspension of Cytodex beads in the subcutaneous inoculum in order to induce necrotic skin lesions (7). The precise role of these particles in pathophysiology has never been clarified, but we have hypothesized that the beads may attenuate the effects of tissue-infiltrating phagocytic cells. In the present experiments, we found that csrR mutants induce necrotic skin lesions without the addition of Cytodex to the inoculum. This finding suggests that derepression of virulence genes by CsrR substitutes for the effects of Cytodex. We speculate that the augmented encapsulation in these strains may provide the protection from phagocytic cells afforded by Cytodex to the wild-type strain.

The partial attenuation of virulence in a csrR mutant by introduction of a hasAB deletion implies that capsule contributes to virulence in our animal model system. However, these experiments also suggest that csrR also controls important virulence factors other than capsule that contribute to necrotizing skin infection. SLS is one of the most toxic membrane-active cytolysins known (2). Owens et al. observed that a GAS strain that was SLS-negative after chemical mutagenesis was attenuated when given to mice by the intraperitoneal route (26). However, more concrete evidence for the role of SLS in infection was provided by Betschel et al., who reported that insertional inactivation of a gene designated sagA leads to a loss of SLS activity (6). It is quite likely that sagA actually encodes SLS. Its identity remains uncertain since the sequence of the SLS polypeptide is unknown and since sagA has not been cloned. However, a strain with a Tn916 insertion in sagA produces less weight loss and skin necrosis than the wild-type strain in murine infections (6). In invasive soft tissue infection, a potent cytolysin, such as SLS, may significantly impair inflammatory cells locally and facilitate the spread of streptococci in tissue.

In contrast to SLS, there is little direct evidence of a role for SpeB in necrotic skin infection. A proinflammatory effect has been attributed to SpeB in some experiments. For example, purified SpeB administered in conjunction with streptococcal cell wall can induce profound tissue inflammation in the lung (34). Two groups of investigators have reported that isogenic speB mutants are less virulent in intraperitoneal infection and after injection of air pouches in skin (20, 22). In contrast, Ashbaugh et al. found that there was no effect of a speB mutation on the occurrence of necrosis in a murine skin infection model (4). The specific role of SpeB in GAS virulence remains uncertain.

Our findings show that hyaluronic acid capsule is only partly responsible for the hypervirulence of csrR mutants. Although these mutants produce excessive amounts of SLS and SpeB, we have not proven that either of these toxins is responsible for the residual virulence observed when capsule production is eliminated. However, coinfection of the wild type and a csrR strain modified the tissue microenvironment in ways that enhanced the survival of the wild-type strain. This finding suggests that extrinsic factors, such as exotoxins, may play an important role in GAS virulence.

The elaboration of each of these factors may have more or less importance for colonization or disease, depending on the site and circumstances of infection. Therefore, it is reasonable to speculate that sets of GAS virulence factors may be coordinately regulated in response to specific environmental stimuli. mga is an example of one such regulatory locus; this locus controls the expression of M protein, SIC, and the C5 peptidase (9). These proteins are expressed during mid-exponential phase when the bacteria are growing vigorously. It is likely that this set of virulence factors is most important during rapid growth in the host. In contrast, CsrR appears to repress virulence factors during exponential growth. csrR mutants elaborate SLS and SpeB at the end of exponential phase, before the wild-type strain begins to express these proteins. We speculate that early secretion of these and perhaps other factors may confer an advantage to csrR mutant GAS in tissue. It is not known whether such mutants emerge spontaneously or whether they contribute to necrotizing fasciitis or other forms of rapidly invasive streptococcal disease. However, this is a testable hypothesis now that the control locus has been identified.