The PerR regulon in peroxide resistance and virulence of Streptococcus pyogenes

Authors

  • Audrey Brenot,

    1. Department of Molecular Microbiology, Washington University School of Medicine, Box 8230, St. Louis, MO 63110-1093, USA.
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Katherine Y. King,

    1. Department of Molecular Microbiology, Washington University School of Medicine, Box 8230, St. Louis, MO 63110-1093, USA.
    Search for more papers by this author
    • Present address: Department of Pediatrics, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030, USA.

    • These authors contributed equally to this work.

  • Michael G. Caparon

    Corresponding author
    1. Department of Molecular Microbiology, Washington University School of Medicine, Box 8230, St. Louis, MO 63110-1093, USA.
      E-mail caparon@borcim.wustl.edu; Tel. (+1) 314 362 1485; Fax (+1) 314 362 3203.
    Search for more papers by this author

E-mail caparon@borcim.wustl.edu; Tel. (+1) 314 362 1485; Fax (+1) 314 362 3203.

Summary

Prior studies have shown that the catalase-deficient pathogen Streptococcus pyogenes (group A streptococcus) has a robust ability to resist oxidative stress that partially involves the transcriptional regulator PerR. However, the extent of the PerR regulon and the contribution of the members of this regulon to virulence are unknown. In this study, DNase I footprinting revealed that PerR binds specifically to a single site upstream of the promoter for the gene encoding alkyl hydroperoxide reductase (ahpC). However, analyses of transcript abundance revealed that while ahpC is regulated in response to growth phase, its regulation is independent of PerR. Instead, PerR regulates transcription of a divergent gene cluster that encodes a putative cold shock protein. The gene encoding the Dps-like peroxide resistance protein MrgA was repressed by PerR, consistent with the presence of a PerR binding site in its promoter. Phenotypic analyses of PerR, AhpC and MrgA mutants revealed that while AhpC is not essential for resistance to challenge with hydrogen peroxide in vitro, AhpC does contribute to scavenging of endogenous hydrogen peroxide and is required for virulence in a murine model of infection. In contrast, a MrgA mutant was hypersensitive to challenge with peroxide in vitro, but was fully virulent in all animal models tested. Finally, a PerR mutant was hyper-resistant to peroxide, yet was highly attenuated for virulence in all murine models. These data demonstrate that while a mutant's capacity to resist peroxide stress did not directly correlate with its ability to cause disease, the appropriate regulation of the peroxide stress response is critical for virulence.

Introduction

Like many developmental processes, establishment of infection requires a highly orchestrated series of events that are ordered both temporally and spatially. From the perspective of the microorganism, this ordering process requires sophisticated regulatory networks to co-ordinate changes in gene expression in response to the developmental stage of the infection. This concept implies that expression of certain genes at an inappropriate developmental stage will be deleterious to the infectious process. In this context, the analyses of bacterial stress responses are particularly interesting. These responses function to adapt the pathogen to a harsh environment that is often generated as a part of the host's attempt to clear the infection (Guiney, 1997; Cotter and Miller, 1998; Mahan et al., 2000). This modification of the environment often serves as a developmental cue to the pathogen for the regulation of genes required for pathogenesis (adhesins, toxins, etc.) though they themselves are not directly responsible for adaptation to stress. Even though these stress-regulated genes appear to enhance the fitness of the pathogen in the host, their expression is highly regulated (Guiney, 1997; Cotter and Miller, 1998; Mahan et al., 2000). This observation suggests that in the absence of stress, expression of stress-response genes may decrease fitness. Thus, analysis of stress responses in the context of the host–pathogen interaction can provide valuable insight into the developmental pathway of infection, as well as the strategies used by the host to eliminate infection.

Stress responses may play an important role in the regulation of virulence in the Gram-positive pathogen Streptococcus pyogenes (group A streptococcus). This organism is renowned for its ability to infect a large number of different tissues in the host, causing diseases which include pharyngitis (‘strep throat’), toxic-shock syndrome and necrotizing fasciitis (Cunningham, 2000). Adherence to host tissues likely plays a key role in the development of many of these diseases and at least one adhesin, known as protein F or Sfb, is regulated in response to increased concentrations of oxygen (VanHeyningen et al., 1993). Regulation occurs at the level of transcription by a mechanism that may be part of a stress response (Fogg et al., 1993; Gibson and Caparon, 1996; Granok et al., 2000). How S. pyogenes responds to oxidative stress in general is also an interesting question. As a member of the lactic acid family of bacteria, S. pyogenes does not produce catalase, an oxidoreductase often considered to be essential for resistance to the damaging effects of hydrogen peroxide and growth in aerobic environments. However, S. pyogenes can grow robustly in aerobic environments and S. pyogenes can even endogenously produce significant amounts of peroxide when grown under aerobic conditions (Gibson et al., 2000). Taken together, these observations suggest that an adaptation to oxidative stress plays an important role at some stage in the development of S. pyogenes infections.

Despite the absence of catalase, S. pyogenes has the capacity to mount an inducible stress response that increases its ability to survive in the presence of peroxide (King et al., 2000). Taking a functional genomic approach, inactivation of several genes whose products have homology to other known peroxidases revealed that alkyl hydroperoxide reductase (AhpC) and glutathione peroxidase (GpoA) did not affect the ability of S. pyogenes to grow under aerobic conditions or to mount the inducible peroxide protective stress response (King et al., 2000). In contrast, inactivation of an orthologue of perR that encodes a peroxide stress response regulator in Bacillus subtilis, resulted in constitutive expression of enhanced resistance to peroxide (King et al., 2000; Ricci et al., 2002). As a member of the Fur family of transcription repressors, PerR is a zinc-containing metalloprotein (Herbig and Helmann, 2001). The operator sequence, known as the Per box, is a highly conserved sequence located immediately upstream or downstream of the −35 and −10 elements of target promoters (Herbig and Helmann, 2001). In B. subtilis, Per boxes regulate the genes which encode alkyl hydroperoxide reductase ahpCF, a member of the Dps/PexB family of DNA-binding stress proteins known as MrgA (mrgA), catalase (katA), the haem biosynthesis operon (hemA) and the P-type metal-transporting ATPase (zosA) (Herbig and Helmann, 2001; Gaballa and Helmann, 2002; Helmann et al., 2003).

In the present study, we have examined the importance of regulation of genes involved in adaptation to peroxide stress for the pathogenesis of S. pyogenes infections. The analysis included characterization of the interaction of PerR with several target promoters, the definition of the S. pyogenes Per box, and the contribution of several regulated genes to the ability of S. pyogenes to resist peroxide stress in vitro. Also conducted was an evaluation of the contribution of PerR and several PerR regulated genes to the ability of S. pyogenes to cause disease in animal models representing several different host compartments, including subcutaneous and intraperitoneal infection of mice and intramuscular infection in a newly developed model of S. pyogenes infection that utilizes zebrafish (Neely et al., 2002). The data reveal several significant differences with how PerR interacts with promoters in S. pyogenes as compared to B. subtilis and that the requirement for any one component was dependent on the specific host compartment that was infected. It was also found that a mutant's capacity to resist peroxide challenge in vitro did not correlate with the ability to cause disease. These data suggest that while the PerR regulon is important for virulence, its influence may extend beyond adaptation to oxidative stress.

Results

Alkyl hydroperoxide reductase contributes to peroxide balance in S. pyogenes

Catalase and the alkylhydroperoxide reductase (AhpC) are members of the PerR regulon in B. subtilis (Bsat et al., 1996; Bsat et al., 1998). The fact that streptococcal species lack catalase has suggested an important role for AhpC in adaptation to peroxide stress and growth in aerobic environments. However, initial characterization of an AhpC mutant of S. pyogenes revealed that the mutant was not more sensitive to challenge with hydrogen peroxide, nor did the mutant lack a characteristic inducible protective response to peroxide that appears to require PerR (King et al., 2000). These data suggest that other gene products may be more important for protection from peroxide challenge, but offer little insight into the function of AhpC. However, like many lactic acid bacteria, S. pyogenes can produce peroxide endogenously (Muller, 1985; Gibson et al., 2000). Thus, maintenance of peroxide balance requires pathways for degradation of both external and internal peroxide. To determine if AhpC contributes to scavenging of endogenous hydrogen peroxide, the behaviour of mutants following growth on an indicator medium that is sensitive to the accumulation of hydrogen peroxide was evaluated. On this medium, colonies containing peroxide appear purple, while those lacking peroxide appear white (see Experimental procedures). Examination of colonies of Streptococcus gordonii, an oral streptococcal species known to accumulate large amounts of peroxide (Muller, 1985; Barnard and Stinson, 1999), revealed that as expected, they appeared purple (Fig. 1A). In contrast, colonies of the wild type S. pyogenes strain (HSC5) appeared white. However, an AhpC mutant of S. pyogenes produced colonies of an even darker purple colour than those of S. gordonii (Fig. 1A), indicating that the AhpC mutant accumulate larger amounts of endogenously produced peroxide that the wild type strain. Colonies of a glutathione peroxidase mutant (King et al., 2000) and PerR and MrgA (see below) mutants appeared identical to the wild type (not shown).

Figure 1.

Accumulation and degradation of hydrogen peroxide.
A. The accumulation of peroxide in S. gordonii, wild type (WT) and AhpC mutant (AhpC) S. pyogenes was evaluated following growth on peroxide indicator plates. Colonies containing peroxide appear purple, while those lacking peroxide appear white.
B. The rates at which cells from an AhpC mutant and mutants deficient in PerR (PerR) and MrgA (MrgA) could degrade exogenously added peroxide were compared to the wild type strain, uninoculated THY medium (media) and wild type bacteria that had been killed by heating (dead). Data presented report the average rate of degradation and standard error of the mean derived from two independent experiments. Significance was determined using Student's t-test for unpaired samples (P = 0.056 for the AhpC mutant, P > 0.05 for all other mutants).

The behaviour of AhpC mutants on indicator media suggests that AhpC contributes to the maintenance of peroxide balance in S. pyogenes. This conclusion was further supported by comparison of the rates of peroxide degradation between wild type and AhpC mutant cells. When hydrogen peroxide was added to a suspension of wild type S. pyogenes, peroxide was degraded at an appreciable rate (2 µM min−1) (Fig. 1B). In contrast, heat killed streptococci degraded peroxide at a negligible rate (0.4 µM min−1 or 0.2-fold relative to wide type) similar to the rate of degradation observed in medium alone (Fig. 1B). While not statistically significant, the rate of peroxide degradation was consistently higher than wild type in PerR and MrgA mutants (Fig. 1B). The AhpC mutant also degraded peroxide, but at a rate that was about 60% slower than that of wild type cells (Fig. 1B). Thus, despite its lack of catalase, S. pyogenes has a robust ability to degrade peroxide to which AhpC partially contributes.

MrgA mutants are hypersensitive to peroxide

The fact that AhpC mutants retain some ability to degrade peroxide and that AhpC mutants are not more sensitive to lethal peroxide challenge (King et al., 2000) indicates that the peroxide stress response is multifactorial. In B. subtilis and Staphylococcus aureus, a gene encoding a Dps-like DNA binding protein known as MrgA is regulated by PerR and plays a central role in protection from challenge by a lethal concentration of peroxide (Chen and Helmann, 1995). A MrgA orthologue (known as Dpr) is important for the ability of Streptococcus mutans to grow aerobically (Yamamoto et al., 2000). Examination of the S. pyogenes SF370 genome (Ferretti et al., 2001) revealed a single open reading frame highly similar (52% similar, 30% identical) to mrgA (Supplementary materialFig.S1). The S. pyogenes MrgA contains amino acid residues which are highly conserved among members of the Dps family including residues corresponding to H50, H62, D66, D77 and E81 that may be involved in binding iron (Grant et al., 1998). Replacement of mrgA with an allele containing an in frame deletion (see Experimental procedures) produced a MrgA-deficient mutant (Supplementary materialFig.S1).

In examining the phenotype of the MrgA mutant, it was found that unlike S. mutans (Yamamoto et al., 2000), the S. pyogenes mutant showed no defect in its ability to grow under aerobic conditions (data not shown) nor in its ability to degrade peroxide (Fig. 1B). As expected from previous data, a PerR mutant was approximately 100-fold more resistant to lethal challenge with hydrogen peroxide, while an AhpC mutant and a mutant in another putative peroxidase (GpoA) resembled the sensitivity of the wild type strain (Fig. 2). However, the S. pyogenes MrgA mutant was more than 100-times more sensitive to high concentrations of hydrogen peroxide than wild type (Fig. 2).

Figure 2.

Survival of mutants in presence of hydrogen peroxide. The ability of the indicated strains to survive challenge with 4 mM H2O2 was determined as described in the ‘Experimental procedures’. For comparison, the analysis included a strain defective for expression of glutathione peroxidase (GpoA). Other strain names are as defined in the previous figures. Data reported represent the average and standard error of the mean of at least three independent experiments.

PerR demonstrates unusual interactions with promoter regions

In B. subtilis, PerR binds to a 15 bp operator site known as a Per box in the ahpC and mrgA promoters that either overlaps or is located immediately adjacent to the −35 or −10 RNA polymerase recognition elements (Herbig and Helmann, 2001; Fuangthon et al., 2002). To characterize the interaction of S. pyogenes PerR with its target promoters, the promoters for ahpC and mrgA were defined by primer extension analysis. For each gene, a single extension product was obtained which identified a 5′-transcript end (Fig. 3A and B) that was consistent with the size of the transcripts observed in the Northern blot analysis (data not shown). Initial examination of the sequences flanking the promoters did not reveal any obvious candidates for a canonical Per box for either the ahpC or mrgA promoters. To determine whether PerR specifically interacts with the region upstream of ahpC, DNase I protection assays were conducted. For this analysis, a purified fusion of PerR with glutathione S-transferase (GST-Per, see Experimental procedures) and a 360 bp DNA fragment that included the region between ahpC and the next divergently transcribed open reading frame (Spy2077, see below) were used. The analysis revealed that GST-PerR, but not GST alone, protected a 29 bp region over a range of PerR concentrations (Fig. 3C). The protected region included a sequence similar to the canonical Per box described for the mrgA promoter of B. subtilis (Herbig and Helmann, 2001); however, this region lays approximately 100 bp upstream of the −10 and −35 recognition elements for ahpC(Fig. 4A).

Figure 3.

PerR interacts with the promoters of ahpC and mrgA. A primer extension analysis of the transcripts for ahpC (A) and mrgA (B) is shown. The extension products are shown in the lanes labeled ahpC and mrgA, respectively. Other lanes show the sequencing reaction products derived from plasmid templates and primers identical to those used to make the extension product. The asterisk at the left of the figure shows the identity of the 5′-terminal nucleotide for each transcript. C shows the ability of GST-PerR to protect sequences upstream of ahpC from degradation by DNase I. Reactions containing increasing concentrations of GST-PerR (left to right, 2.5–25 pmol) are compared with GST alone (GST) and a reaction that contained only the labeled template and DNase I (0). The protected region is highlighted by the brackets at the right of the figure which indicate the location of the region relative to the beginning of the ahpC transcript as determined by comparison with the Maxam-Gilbert sequencing reaction shown at the left of the figure.

Figure 4.

Organization of the promoters for ahpC and mrgA. A and B summarize the structure of the promoters for ahpC and mrgA, respectively. The bent arrow labeled +1 indicates transcription start sites, and the promoters elements for recognition by RNA polymerase are shown by the boxes labeled −10 and −35. The 29 bp region protected by GST-PerR is indicated by a bracket under the sequence and for both promoters a consensus Per box is underlined. A region of extended similarity that includes the Per box is shown by the arrow above and below the ahpC and mrgA promoters, and its orientation is indicated by the arrow. Putative ribosome binding sites are shown by the dashed lines under the sequence and the start codon and begin of the ahpC and mrgA coding regions are shown by the amino acid sequences under the figures. C compares the canonical Per box defined for Bacillus subtilis (B. s.) with the Per box determined for the ahpC region, and several other putative Per boxes identified in the available genomic information. Comparison between the S. pyogenes Per boxes allowed the definition of S. pyogenes (S. p.) consensus Per box. The orientations of the sequences relative to the indicated open reading frames are shown by the arrow at the right of the figure. Open reading frame numbers based on the SF370 genome are as follows: ahpC, Spy2079, mrgA, Spy1531, ykvW, Spy1434, fhuA, Spy0386. The open reading frame Spy1840 has no known homologue. The ykvW box is localized between Spy1434 (ykvW, heavy metal transporting ATPase) and Spy1436 (mf3, deoxyribonuclease) and the fhuA box, between Spy0386 (fhuA, ferrichrome ABC transporter) and Spy0388 (murE).

Recognizing that PerR may differ in its interactions with promoters relative to B. subtilis, the promoter for mrgA in S. pyogenes was analysed in greater detail. By searching for a sequence like the PerR binding site identified in the S. pyogenes region upstream of ahpC, a similar site was identified in the mrgA promoter just downstream of the −10 region (Fig. 4B). However, this site is oriented in the reverse direction as compared to the site found upstream of the ahpC promoter. Examination of available genomic resources revealed the presence of Per boxes in the potential regulatory regions of several other genes (Fig. 4C). The structure and location of these Per boxes is described in the Discussion (see below).

PerR is a transcriptional repressor of mrgA and a transcriptional activator of csp

Mutations in perR make S. pyogenes much more resistant to peroxide stress (King et al., 2000), consistent with the observation that PerR represses ahpC and mrgA in B. subtilis (Chen and Helmann, 1995). To determine if PerR regulates ahpC and mrgA in S. pyogenes, the relative abundance of each transcript was compared between wild type and PerR mutant strains using quantitative real time PCR. When compared in the mid-logarithmic phase of growth under normal growth conditions, the transcripts for mrgA were more abundant in the PerR mutant than in the wild type strain, as expected for PerR acting as a repressor of mrgA(Fig. 5A). However, while a statistically significant difference was observed for the ahpC transcript between the wild type and PerR mutant, the absolute magnitude of the difference was modest (approximately 1.3-fold), suggesting that PerR is not a major transcription regulator of ahpC. This result may not be surprising considering the atypical location and orientation of the PerR binding site relative to the ahpC promoter (see above). Taken together, these data suggest that this PerR binding site may regulate the cluster of genes that are divergently transcribed from ahpC that includes a tRNA gene and csp (Spy2077), a gene with similarity to those encoding cold shock proteins (Fig. 5B). Consistent with this hypothesis, comparison of transcript abundance during mid-logarithmic growth revealed that the csp transcript was present at levels approximately 2.8-fold lower in the PerR mutant relative to wild type (Fig. 5A), indicating that PerR can act as a transcriptional activator of csp. Analysis of the level of dnaK and groEL transcripts showed that the general stress response was not significantly induced in the PerR mutant strain (data not shown) compared to the wild type strain. In stationary phase, expression of both ahpC (Fig. 5C) and mrgA (Fig. 5D) was significantly increased relative to expression of during logarithmic phase, and for ahpC, this increase was independent of the presence of a functional PerR (Fig. 5C). Taken together, these data show: (i) that PerR represses expression of mrgA; (ii) that PerR is an activator of the csp gene cluster; and (iii) that expression of ahpC is regulated in response to growth phase and that this regulatory response does not involve PerR.

Figure 5.

PerR is a transcriptional regulator for mrgA and csp but not for ahpC. Relative abundance of ahpC, mrgA, and csp transcripts determined by quantitative real time PCR in RNA derived from WT and PerR mutant strains, in mid-logarithmic and stationary phase. Data shown are median transcript levels relative to the recA internal reference transcript for each strain.
A. The relative level of transcripts in the PerR strain (gray) compared to the WT strain (white) is shown in mid-logarithmic phase for mrgA, ahpC, and csp.
B. The location of ahpC region in the S. pyogenes genome is shown relative to its flanking gene. Numbering of open reading frames (shown below the figure) and gene identities of the flanking gene (shown above the figure) are based on the annotation of the genomic sequence of S. pyogenesSF370 (Ferretti et al., 2001). The rectangle represents the putative Per box identified in this region.
C and D. The relative level of transcripts in stationary phase (gray) compared to mid-logarithmic (white) phase is shown for ahpC (C) and mrgA (D) in the wild type (WT) and PerR mutant strain. Significance was determined by using paired Student's t-test (*P < 0.05 and **P < 0.004).

PerR and AhpC, but not MrgA mutants are attenuated in a murine model of cutaneous infection

Having characterized the behaviour of various mutants in vitro, it was of interest to determine the contribution of PerR-regulated genes to virulence in vivo. Because S. pyogenes is capable of infecting a wide-variety of different tissues, it was important to evaluate the contribution of PerR-regulated genes to virulence in models representative of a variety of host compartments. A murine cutaneous model involving subcutaneous injection of bacteria into hairless mice is representative of streptococcal soft tissue disease, and the development of disease requires the streptococci to survive in highly inflamed tissue for several days (Lukomski et al., 1999). At a dose of 107 cfu, the wild type strain produced a region of induration and erythema that was apparent by approximately 12–18 h. By 24–36 h, a region of central necrosis developed, which expanded in size over the course of the next 3–5 days. An eschar formed during this time, and after approximately 7 days, the necrotic ulcer began to heal and was completely resolved 10–14 days postinfection. The MrgA mutant, despite its extreme sensitivity to peroxide stress, was not attenuated in its ability to cause disease in cutaneous tissue as shown by the observation that the time course and sizes of the necrotic lesions produced did not differ from the wild type strain (Fig. 6). In contrast, the virulence of both the AhpC and PerR mutants was significantly attenuated. At 48 h after infection, when large ulcers had formed in mice infected by the wild type and MrgA mutant, the mice infected with PerR mutant showed no evidence of ulcer formation and the mice infected with AhpC mutant showed no ulcer or smaller ulcers compared to the mice injected with the wild type strain (Fig. 6). Ulcers eventually formed in mice infected by the PerR mutant, typically appearing by day 5; however, these never obtained the size of the lesions produced by the wild type and the MrgA mutant (data not shown).

Figure 6.

AhpC and PerR but not MrgA mutants are attenuated in a mouse subcutaneous infection model. Groups of 10 mice each received a subcutaneous injection of 1 × 107 cfu of the strains indicated in the figure and were observed for the development of a visible ulcer, whose area was determined as described in the Experimental procedures. At day 2, time of maximal lesion development for the wild type strain, lesions were measured for the mice injected with the wild type strain and the mutant strains. The wild type strain produced ulcers that were significantly larger than those produced the AhpC mutant (P < 0.01) and the PerR mutant (P < 0.005), but not the MrgA mutant. Each symbol represents the area of ulcer observed in an individual mouse and the horizontal bar indicates the mean value obtained for each group of 10 mice.

Examination of the number of cfu recovered from lesions revealed that the bacterial load for all the strains injected typically reached an average of at least 108 cfu per lesion between 12 and 18 h postinfection (Fig. 7). Recoverable cfu did not decrease significantly between 12 and 18 h for the lesions recovered from mice injected with the wild type strain, the AhpC or MrgA mutants and was always higher than 107 cfu. Thus, an enhanced rate of clearance over the first 18 h of infection cannot solely account for the reduced virulence of the AhpC mutant. In contrast, recoverable counts for the PerR mutant decreased from an average of 108 cfu at 12 h to 5 × 106 cfu at 18 h (Fig. 7).

Figure 7.

Persistence of S. pyogenes in skin. Mice were challenged with wild type, AhpC, MrgA and PerRS. pyogenes as described for Fig. 6. At 12 and 18 h following injection, the skin including underlying adipose tissue and abdominal muscle was excised at an additional radius extending at least 2 mm from the margin of the induration present at the site of infection, and the number of cfu in the tissue determined as described in the Experimental procedures. Each symbol represents the number of cfu of the indicated strains recovered from an individual mouse and the bar represents the mean value obtained for each group of mice. The data indicate that the persistence of AhpC and MrgA mutants was not significantly different from the persistence of the wild type strain.

PerR but not AhpC and MrgA mutants are attenuated following intraperitoneal infection of mice

Intraperitoneal infection of the wild type strain into C57Bl/6J mice resulted in the death of mice with an LD50 of 1.4 × 107 cfu. The mice typically died between 4 and 8 days after infection. When compared to wild type, PerR mutants were significantly attenuated. At a dose of 3 × 107 cfu, greater than 80% of PerR -infected mice remained alive after 8 days, while the majority of mice infected by the wild type, AhpC and MrgA mutant had died (Fig. 8). Similarly, greater than 90% of mice infected by the wild type and MrgA mutant developed abdominal perforations, while less than 20% of mice infected with PerR mutant bacteria developed perforations. Numbers of viable bacteria recovered by peritoneal lavage and from homogenized spleens at 12 and 18 h after infection revealed no significant differences in the numbers of recovered cfu between any of the wild type or mutant strains (data not shown).

Figure 8.

PerR mutants, but not AhpC and MrgA mutants are attenuated following intraperitoneal challenge. Groups of at least five mice each were challenged intraperitoneally with 3 × 107 of wild type (WT, closed diamond), AhpC (open square), MrgA(closed triangle), or PerR (open circles) mutants. Survival of mice was monitored daily for 9 days and the data presented as a Kaplan-Meier plot. The data indicate that PerR mutants were significantly less lethal than wild type (P < 0.01).

Neither AhpC, PerR nor MrgA mutants are attenuated in intramuscular infection of zebrafish

A newly developed model of streptococcal infection of zebrafish (Neely et al., 2002) was utilized to study the behaviour of mutants. In this model, intramuscular infection of zebrafish with the wild type S. pyogenes results in lethality after approximately 4 days with an LD50 of 1 × 104 cfu (Neely et al., 2002). Histopathology reveals extensive necrosis of muscle tissue and, in contrast to infection of murine cutaneous tissue, a marked absence of infiltration of inflammatory cells into the infected tissue (Neely et al., 2002). In the present study, at a dose of 1 × 106 cfu of the wild type strain, approximately 80% of the fish had died after 4 days, while all fish injected with fresh culture medium remained healthy. Infection by the PerR, AhpC and MrgA mutants demonstrated a course of infection nearly identical to wild type (data not shown).

Discussion

Studies of stress responses and their regulation in pathogenic bacteria can provide valuable insight into many aspects of pathogenesis including the environment the pathogen encounters in different host compartments and the mechanisms used by the host to eliminate the pathogen. It has long been thought that peroxide plays an important role in the host's response against S. pyogenes. However, in the present study, we found that the ability of S. pyogenes to resist peroxide stress does not always correlate with its ability to cause disease. Mutants lacking MrgA were highly sensitive to peroxide stress, but showed no defect in their ability to cause disease in three different animal models. In contrast, mutants in AhpC demonstrated no enhanced sensitivity to peroxide in vitro, but were attenuated in one of the murine models examined. In the absence of PerR, S. pyogenes is hyper-resistant to peroxide stress, yet is attenuated for virulence in the cutaneous tissues and peritoneum, but not muscle. Infection of the latter tissue is notable for its lack of recruitment of inflammatory cells (Neely et al., 2002) suggesting that despite the benefit of increased resistance to peroxide stress, inappropriate expression of PerR-regulated genes might detract from virulence in neutrophil-rich environments.

In B. subtilis, PerR recognizes a consensus 15 bp ‘Per box’ sequence organized in a 7-1-7 configuration that consists of a 7 bp inverted repeat that surrounds a single nucleotide (Herbig and Helmann, 2001; Baichoo and Helmann, 2002; Fuangthon et al., 2002), although many promoters, including the ahpC promoter, do not have perfect repeats (Herbig and Helmann, 2001). The Per box is typically located immediately downstream or immediately upstream of the −10 and −35 regions of regulated promoters (Herbig and Helmann, 2001). Analysis of S. aureus has revealed a similar organization of regulated promoters (Horsburgh et al., 2001). Based on its similarity to Fur, PerR may bind as two dimers that recognize opposite faces of the DNA helix (Baichoo and Helmann, 2002). When bound, PerR has an extended DNase I footprint of over 50 bp that overlaps the −10 and −35 regions of the promoter (Herbig and Helmann, 2001). The mechanism of PerR regulation of csp and mrgA in S. pyogenes was of interest because canonical Per boxes were not found in the usual positions in the promoters for these genes. Instead, DNase I footprinting analyses revealed that PerR protected a site located more than 800 bp from the csp open reading frame and included a sequence that was highly similar to the consensus B. subtilis 15 bp Per box, although it was not a perfect inverted repeat. Interestingly, a 24 bp sequence highly similar to a sequence overlapping the csp Per box was found in the promoter for mrgA. If PerR protects a region similar to that in the csp promoter (29 bp) then PerR would also not overlap the entire promoter like it does in B. subtilis (Herbig and Helmann, 2001), although it has been shown to protect additional sequences upon the addition of higher concentrations of PerR (Herbig and Helmann, 2001). Additional experimentation will be required to understand how PerR functions to regulate promoters in S. pyogenes.

Biochemical characterization of the PerR binding site in the region upstream of the ahpC gene facilitated a search for other PerR-binding sites in the S. pyogenes genome. Close matches exist at several chromosomal locations. The first of these is in the promoter region of an open reading frame putatively encoding a protein with no known homologue (Spy1840). A second is found in the region between two other divergent open reading frames encoding a putative heavy-metal transporting ATPase (Spy1434 or ykvW) and a putative deoxyribonuclease (Spy1436). The B. subtilis homologue of Spy1434 was shown to be regulated by PerR (Gaballa and Helmann, 2002) and the level of transcription of the ykvW gene in S. pyogenes was shown to be increased in mid-logarithmic phase in a PerR mutant compared to the wild type strain (A. Brenot and M. Caparon, unpublished data), indicating that this sequence is a functional Per box. A final potential Per box exists before the streptococcal FhuADBG operon, which is similar to genes encoding an iron-hydroxymate transporter. In S. aureus and B. subtilis the FhuBCDG operon is repressed by Fur, the ferric uptake repressor (Sebulksy et al., 2000). However, sequence analysis of the complete S. pyogenes SF370 genome suggests that S. pyogenes does not encode other members of the Fur family of transcriptional regulators including Fur, Zur, or Mnt (Ferretti et al., 2001). Thus, PerR may be responsible for some of the functions of Fur in this species. Alternatively, S. pyogenes may represent an ancestor of PerR- and Fur-containing bacteria in which the functions of the two proteins have not yet diverged. In this regard, it is interesting to note that Per boxes were not observed in the promoters for sodA, mtsABC or perR itself. The former two genes are involved in oxidative stress and iron transport and their regulation is altered in PerR mutant S. pyogenes strains (Ricci et al., 2002). In B. subtilis, expression of perR can be influenced by iron, and perR binds to a strong consensus Per box in its own promoter, although PerR does not seem to directly influence expression of the perR promoter (Fuangthon et al., 2002).

The streptococci lack catalase, a major oxidoreductase involved in peroxide resistance in most aerobic bacteria. They contain alternative peroxidases, including AhpC and a glutathione peroxidase (GpoA); however, deletion of the genes for both of these enzymes do not result in increased sensitivity to peroxide (King et al., 2000). In this study, we found that the DNA-binding protein MrgA is essential for resistance to hydrogen peroxide, consistent with previous studies showing that Dpr, the MrgA homologue in S. mutans, is a principle component of hydrogen peroxide resistance in that species (Yamamoto et al., 2000). The proteins from the Dps family are believed to protect the cells from the toxicity induced by hydrogen peroxide by binding free intracellular iron to prevent Fenton chemistry-catalysed formation of toxic hydroxyl radicals (Yamamoto et al., 2002; Pulliainen et al., 2003). However, in contrast to S. mutans Dpr mutants (Yamamoto et al., 2000), MrgA mutants of S. pyogenes are fully competent for growth under aerobic conditions. Similarly, Dpr mutants of Streptococcus suis are also peroxide sensitive but fully aerotolerant (Pulliainen et al., 2003). These data may suggest that some species of streptococci have additional mechanisms that promote aerotolerance. This assertion is further supported by the observation in this study that even AhpC mutant bacteria were able to degrade exogenously added peroxide.

For S. pyogenes, mutants in AhpC are attenuated for virulence in several animal models of streptococcal infection. In Escherichia coli, it appears that the function of ahpC is to maintain peroxide homeostasis via control of endogenously produced peroxide (Costa Seaver and Imlay, 2001) and the data from peroxide degradation assays in this study suggest a similar role for ahpC in S. pyogenes. Similar to other lactic acid bacteria, S. pyogenes has the capacity to produce significant quantities of peroxide (Gibson et al., 2000), although the reason why peroxide is generated is not clear. This may mean that the maintenance of internal peroxide balance is critical for the growth of S. pyogenes in a tissue environment. Alternatively, AhpC has a broad range of substrates that also include alkyl peroxides and peroxynitrites (Bryk et al., 2000). Further analysis of the importance of AhpC in streptococcal infection will provide valuable insight into the role of peroxide and possibly other pathways used by the host to respond to S. pyogenes in tissue.

Most genetic studies of bacterial virulence have focused on attenuation of virulence because of loss-of-function mutations. Very few studies have addressed the effects of gene misregulation on bacterial virulence (Akerley et al., 1995). As PerR mutants are more resistant to hydrogen peroxide in vitro, it was expected that these mutants would demonstrate a hypervirulent phenotype in mice. However, PerR mutants were attenuated indicating that inappropriate expression of PerR-regulated genes is not adaptive for the organism in some host compartments. Specifically, expression of some virulence factors at a particular time during the development of an infection may be critical to the further development of the infection, but inhibitory if expressed at other times. For example, expression of an adhesin may be essential for the initial establishment of an infection; although continued adhesin expression may hinder further bacterial spread and invasion (Hendrixson and St Geme, 1998).

These studies of S. pyogenes have elucidated the regulatory relationship between PerR, MrgA, Csp and AhpC, and highlighted their physiologic roles in responding to hydrogen peroxide in the environment. Furthermore, the findings presented here suggest that hydrogen peroxide is not abundant enough to affect survival of streptococci in the skin, peritoneum, or muscle. In contrast, molecular defences conferred by AhpC, most likely defence against reactive nitrogen intermediates, are crucial for innate immunity in these settings. Furthermore, inappropriate regulation of peroxide-response genes has proven to be maladaptive for the organism during infection. Thus, environmental regulation of the expression of bacterial products may be especially important for tissue-specific colonization and survival by pathogenic bacteria.

Experimental procedures

Strains, media and growth conditions

The construction of mutants containing in frame deletion mutations in ahpC and perR (HAHP and HPER, respectively) in a HSC5 wild type S. pyogenes background (Hanski, 1992) was described in a previous study(King et al., 2000). Other S. pyogenes strains included a mutant deficient in glutathione peroxidase (King et al., 2000). Molecular cloning experiments were performed using E. coli DH5α, while protein expression was conducted in E. coli BL21. Routine culture of S. pyogenes for in vitro peroxide sensitivity tests utilized Todd-Hewitt broth (BBL) supplemented with 0.2% yeast extract (Difco) (THY medium), and the conditions described previously (VanHeyningen et al., 1993). To produce solid media, Bacto agar (Difco) was added to THY medium to a final concentration of 1.4% and unless otherwise indicated, all solid cultures were incubated under the anaerobic conditions produced using a commercial gas generator (GasPack; BBL). Culture of streptococci for infection of animals followed established protocols, including growth in Brain–heart infusion medium (Difco) supplemented with 2% supplement B (Difco) (BHIB) (Husmann et al., 1997) for infection of mice and THY supplemented with 2% Proteose Peptone ♯3 (THY + P) for infection of zebrafish (Neely et al., 2002). Strains of E. coli were cultured in Luria–Bertani broth and tryptone agar as described (Scott, 1972). When appropriate, antibiotics were used at the following concentrations: ampicillin 100 µg ml−1 for E. coli; kanamycin, 25 µg ml−1 for E. coli and 500 µg ml−1 for S. pyogenes and chloramphenicol, 3 µg ml−1 for S. pyogenes.

DNA techniques

Plasmid DNA was isolated by standard techniques and used to transform E. coli by the method of Kushner (Kushner, 1978). Transformation of S. pyogenes was performed by electroporation as previously described (Caparon and Scott, 1991). Restriction endonucleases, ligases, kinases and polymerases were used according to the Manufacturers’ recommendations. Chromosomal DNA was purified from S. pyogenes as previously described (Caparon and Scott, 1991). Rapid PCR amplification of streptococcal DNA from crude lysates was performed by the method of Hynes et al. (1992). Fluorescently labelled dideoxynucleotides (Big DyeTM terminators, PE Applied Biosystems) were used according to the recommendations of the Manufacturer for the confirmation of the DNA sequences generated by PCR.

Computational analyses

Genome sequence databases for S. pyogenes (Ferretti et al., 2001) were searched by the program TBlastN (Altschul et al., 1997) using network services provided by the National Center for Biotechnology Information (http://www.ncbi.nih.gov/BLAST/). Sequences used as queries included mrgA of B. subtilis (GenBank accession ♯Z22928). Pattern-based searches for PerR binding sites in the available S. pyogenes genome information utilized the ‘findpatterns’ utility of the Wisconsin Package software (Genetics Computer Group).

Disruption of mrgA

An in frame deletion in the gene encoding MrgA (genomic locus Spy1531, see text) was constructed as follows: The PCR primers 5MrgUpBamHI and 3MrgDownHindIII (Supplementary materialTableS1) were used to amplify a region of the chromosome that included the entire mrgA open reading frame. The resulting fragment was inserted into a commercial vector (pCRII) by a TA-tail technique and the resulting plasmid (pKYK19) used as a template in an inverse PCR reaction with primers 5MrgIFD2 and 3MrgIFD2 (Supplementary materialTableS1). This PCR product was digested with SalI, subjected to ligation and then used to transform E. coli. The resulting plasmid contains an allele of mrgA lacking 414 bp internal to the open reading frame (see Supplementary materialFig.S1) and a unique SalI site inserted between the codons for T4 and D143. The deletion removed the region encoding three of the four alpha-helices that are predicted to make up the central core of the protein (Grant et al., 1998). This allele was introduced into the E. coli-streptococcal shuttle vector pJRS233 (Perez-Casal et al., 1993) to produce pKYK20, which was then used to replace the wild type allele in S. pyogenes HSC5 by a standard method (Ji et al., 1996).

Analysis of peroxide degradation and accumulation

The accumulation of peroxide in cultures of S. pyogenes was evaluated using peroxide indicator plates produced by adding azido-bis-thiosulfonate (ABTS, Sigma, 30 mg ml−1) and horseradish peroxidase (Sigma, 2 mg ml−1) to solid THY media. Indicator plates were incubated overnight in an anaerobic environment, and colour development assessed following 2 h of exposure to ambient air. For a quantitative analysis of the ability of S. pyogenes cells to degrade peroxide, growth from an overnight culture in THY medium was diluted 1:100 in fresh THY medium and incubated at 37°C for 5 h and then adjusted to an OD600 of 0.8. At this time (designated −15 min) a 400 µl aliquot was removed and the remaining culture challenged by the addition of exogenous peroxide to a final concentration of 10 mM. An additional aliquot was removed (0 min time point); the mixture was then incubated at 37°C and a 400 µl aliquot removed every 15 min. The cells were removed by centrifugation and the supernatant fluid (200 µl) was mixed with 50 µl of an indicator solution containing ABTS (30 mg ml−1) and horseradish peroxidase (2 mg ml−1). Following 60 min incubation at room temperature, the OD562 of the mixture was determined and compared to a standard curve generated using known concentrations of peroxide. Additional controls included analysis of a culture that had been killed by heating at 100°C for 60 min as was confirmed by plating on the appropriate medium. Each data point was determined in duplicate and each strain analysed a minimum of three times. Data presented report the average rate of degradation and standard error of the mean of two independent experiments.

Analysis of sensitivity to peroxide

Growth from overnight culture in THY was diluted 1:100 in fresh THY and incubated at 37°C for 5 h. An aliquot was removed and the remaining culture challenged by the addition of hydrogen peroxide to a final concentration of 4 mM. The mixture was incubated at 37°C for an additional 3 h, at which time a final aliquot was removed. The number of viable bacteria in various samples was determined by plating appropriate dilutions on solid THY media for determination of numbers of colonies forming units (cfu). The ability of various strains to survive peroxide challenge was then reported as the per cent survival calculated as the ratio of cfu ml−1 recovered following 3 h after challenge and the number of cfu ml−1 present immediately before challenge × 100%. Data reported represent the average and standard error of the mean of at least three independent experiments.

Primer extension analysis

Various streptococcal strains were isolated by diluting overnight growth in THY medium 1:100 in fresh THY medium followed by incubation at 37°C to mid-logarithmic phase (OD600 = 0.300) and early stationary phase (OD600 = 0.850). Total RNA was then isolated by the method of Cheung et al. (1994) using a commercial reagent (FastRNA BLUE, Bio101) and a high-speed reciprocating shaking device (FP-120, Savant Instruments). The concentration of RNA was determined by absorbance at 260 nm. The 5′ end of various messages was determined by primer extension as follows: Total RNA isolated from stationary phase cultures (16 µg) was incubated with 2 pmol of a primer that had been labelled with 32P using T4 kinase (Invitrogen). Primers for various genes included: ahpC, 3AhpPro2 (Supplementary materialTableS1); and mrgA, 3 MrgPro (Supplementary materialTableS1). Primers were allowed to anneal to templates for 20 min at 58°C, and then extended using AMV reverse transcriptase (Promega) for 75 min at 42°C. The 5′ end of each extension product was determined by comparison to DNA sequencing reactions generated using the same primer and a modified T7 polymerase (Sequenase 2.0, Amersham). Templates for the sequencing reactions were generated by PCR using the following primers: ahpC, 5AhpPro2 and 3AhpPro (Supplementary materialTableS1); and mrgA, 5MrgPro and 3MrgPro (Supplementary materialTableS1).

Quantitative real time PCR

RNA from various streptococcal strains was isolated as follows: Overnight growth in THY medium was diluted 1:100 in fresh THY medium followed by incubation at 37°C to mid-logarithmic phase (OD600 = 0.300) and early stationary phase (OD600 = 1.1) of growth. Total RNA was then isolated by a protocol adapted from the method of Cheung et al. (1994) using glass beads (Lysing Matrix A, Q/BioGene) and a high-speed reciprocating shaking device (FP-120, Savant Instruments). RNA was further purified (RNeasy Mini Kit for RNA clean up, Qiagen) and contaminating DNA was removed by DNase treatments according to the manufacturers instructions (RNase free DNase set, QIAGEN and DNase I, amplification grade, Invitrogen). Representative samples were assessed for RNA integrity by electrophoretic analysis with an Agilent 2100 Bioanalyzer (Agilent Technologies) and measurement of the A260/A280 ratio was used to determine the RNA concentration and purity (accepted if >1.8). Samples were rejected if PCR amplification preformed with RNA templates indicated the presence of contaminating DNA. For cDNA synthesis, 5 µg total RNA was treated with 200 U Superscript II Reverse transcriptase (Invitrogen) using 250 ng of random primer oligonucleotides (Invitrogen). Real time PCR of selected genes was performed using an iCycler thermocycler (Bio-Rad) using iQ SYBR Green Supermix (Bio-Rad). Internal regions (77–148 bp) of the selected genes were amplified using ORF-specific primers designed with PRIMER three software (Rozen and Skaletsky, 2000). The primers used were as followed (see Supplementary materialTableS1): AhpC-A1 and AhpC-A2 for the ahpC gene; MrgA-A1 and MrgA-A2 for the mrgA gene; AB33 and AB34 or AB33 and AB73 for the csp gene and JLP32 and JLP33 for the recA gene. The level of transcription of recA was not affected by perR inactivation or under a variety of in vitro experimental conditions (A. Brenot and M. Caparon, unpublished data); hence, expression of this gene was used to normalize expression data for each target gene. Each assay was performed in triplicate with at least two RNA templates prepared from bacteria from independent cultures on different days. Target gene expression in total RNA was calculated as 2–(meanΔCt), where Ct is the threshold cycle and ΔCt is equal to (meanCt(target)–meanCt(recA)). Differences in relative transcript abundance were tested for significance with a paired Student's t-test and the null hypothesis was rejected when P < 0.05.

Construction of GST-PerR

Glutathione S-Transferase (GST)-PerR was produced by PCR amplification of perR using primers 5PerBamHI and 3PerEcoRI (Supplementary materialTableS1), digesting the resulting product with BamHI and EcoRI and inserting the digested fragment between the BamHI and EcoRI sites of pGEX2b (Amersham). The resulting plasmid (pKYK22), which contains a fusion of the entire coding region of PerR to a gene for GST encoded on pGEX2b, was introduced into E. coli BL21 and GST-PerR expressed as follows: An overnight culture was diluted 1:30 in fresh media and incubated at 37°C for 3.5 h, at which time isopropyl-β-D-thiogalactoside (IPTG) was added to a final concentration of 1.0 mM. Following an additional 4 h of incubation at 37°C, cells were harvested by centrifugation, disrupted by sonication and the protein purified by glutathione sepharose chromatography by the method recommended by the Manufacturer (Amersham). The purified protein was of the expected molecular mass and was composed of > 95% GST-PerR as estimated from SDS-polyacrylamide gel electrophoresis and Coomassie blue R staining. The concentration of purified GST-PerR was determined by the bicinchoninic acid method using commercial reagents (Sigma) with bovine serum albumin as a standard.

DNase I protection assays

Substrates for DNase I protection assays of the ahpC promoter region were produced by PCR using the primers 5AhpPro2 and 3AhpPro (see above) and the products purified by anion exchange chromatography (ConcertTM, Invitrogen). The upper or lower strand of the substrate DNA relative to the ahpC promoter was labelled with 32P by treatment of the appropriate primer with T4 DNA kinase (Invitrogen) prior to PCR. Various concentrations of purified GST-PerR (2.5–25 pmol) were allowed to react with approximately 500 fmol of substrate for 20 min at room temperature in a 200 µl reaction containing 12 mM Hepes pH 7.5, 12% glycerol, 1 mM EDTA, 60 mM KCl, 5 mM MgCl2, 0.6 mM DTT, 0.1 mM CaCl2, 0.01 mM MnCl. The reaction was allowed to continue following the addition of DNase I (0.01 Units, Invitrogen) for 2 min at room temperature and was then quenched by adding 700 µl of ice-cold stop solution (7.1 µg ml−1 yeast tRNA, 71.4 µl ml−1 saturated ammonium acetate in ethanol). Following a 15 min incubation in a dry ice-ethanol bath, precipitated substrate was collected by centrifugation, washed once with an equal volume of 70% ethanol, mixed with loading buffer (80% formamide, 10 mM NaOH, 1 mM EDTA, 0.1% bromophenol blue, 0.1% xylene cyanol), and subjected to electrophoresis through a Tris-borate-EDTA-urea 6% polyacrylamide sequencing gel. The DNase I cleavage patterns were compared to a DNA sequencing reaction of the labelled substrate that was prepared by the method of Maxam and Gilbert (1980).

Animal infection models

The method of Bunce et al. (1992) as modified by others (Limbago et al., 2000; Schrager et al., 1996) was used to establish an infection of S. pyogenes in the subcutaneous tissue of mice as described in detail elsewhere (Brenot et al., 2004). Data presented are derived from two independent experiments. A lethal systemic infection of C57BL/6J mice was established by intraperitoneal injection of S. pyogenes as described (Brenot et al., 2004). Data presented represent the composite data from two independent experiments, each conducted with at least five mice per experimental group, and each with similar results. Intramuscular infection of zebrafish was conducted as previously described (Neely et al., 2002). Data presented represent two independent experiments, each of which was conducted with at least eight fish per experimental group. Numbers of cfu in cutaneous tissues were determined as described (Brenot et al., 2004).

Statistical analyses

Kaplan-Meier product limit estimates of survival curves were used to compare infection by wild type and mutant streptococci and differences were tested for significance by the log rank test (Glantz, 2002). The difference between the number of mice developing an ulcer following subcutaneous challenge with wild type and mutant was tested for significance by the chi-square test with Yate's correction (Glantz, 2002), and differences in the areas of the resulting ulcers were tested by the Mann–Whitney U-test (Glantz, 2002). For all test statistics, the null hypothesis was rejected when P < 0.05.

Acknowledgements

We would like to thank John Helmann and Kevin Francis for helpful discussions and providing reagents. This work was supported by Public Health Service Grant AI38273 from the National Institutes of Health.

Supplementary material

The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/mmi/mmi4370/mmi4370sm.htm

Table S1. Primers used in this study.

Fig. S1. In frame deletion of mrgA region.

Ancillary