INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In Staphylococcus aureus, the expression of many virulence factors is coordinately regulated. The genetics of this regulation is largely understood in terms of the function of agr and sar (30). Together, the components of these loci form part of a complex pathway that leads to decreased transcription of select cell surface virulence factor genes and increased transcription of regulated extracellular toxins and enzymes (see Fig. 1.) (6, 16, 17, 21, 31). Translational control of alpha-toxin by the agr-sar system has also been demonstrated (28, 32). In agr or sar mutant strains, regulated cell surface proteins (e.g., coagulase, fibronectin binding protein, and protein A) are produced throughout the exponential and postexponential phases of growth. This is in contrast to wild-type strains that only produce these proteins during exponential growth. Furthermore, in agr and sar mutant strains, many extracellular toxins and enzymes that are normally present in postexponential phase cultures (e.g., alpha-toxin, metalloprotease, and serine protease) are reduced to as low as 5% of their normal levels (17, 32).

The agr (accessory gene regulator) locus encodes a self-inducing, pheromone-sensing, signal transduction circuit (Fig. 1A). One of two divergent agr messages is transcribed from a promoter designated P2 (33). This message, RNAII, encodes four proteins, AgrA, AgrB, AgrC, and AgrD. Two of these agr-encoded proteins share sequence homology with elements of other bacterial two-component signal transduction systems: AgrC and AgrA function as sensor and regulator proteins, respectively. The activating signal of the agr system is a pheromone encoded within the prepeptide protein AgrD. AgrB is believed to be the enzyme responsible for the maturation and/or secretion of the 8-amino-acid peptide pheromone (18, 19).

View larger version (17K): [in this window] [in a new window]   FIG. 1.   (A) Model of the function of the agr-sar system; (B) hypothetical model of the function of the rot gene product. Details can be found in the text. Chromosomal DNA is depicted as a thick black line. Promoters (P) are numbered. Genes (boxes) and their translation product are identically shaded. Phosphorylated proteins are associated with a circled letter P. The straight arrows and squiggly lines represent mRNA. Translation of hla mRNA is illustrated by the addition of a ribosome (black circles) to the message. The curved arrows show the relationships among the components of the system. Positive and negative effects are marked with (+) and (), respectively.

The agr system up-regulates its function when AgrC binds the AgrD-derived signal (30). Like other bacterial sensor proteins, the binding of the signal results in the autophosphorylation of AgrC and, presumably, a concomitant activating conformational change (25). The phosphate group on AgrC is then thought to be transferred to the regulator protein AgrA, resulting in the activation of AgrA. Unlike other bacterial signal transduction systems in which the activated regulator protein directly initiates the transcription of target promoters, AgrA functions with the translation product of an unlinked locus named sar (staphylococcal accessory protein regulator) (7). The sar product, SarA, binds a region of DNA between the two agr promoters, and in conjunction with activated AgrA it up-regulates transcription of the agr messages (6, 16, 29). The result of the increased transcription is an amplification of the circuit encoded by RNAII and high-level production of a 510-ribonucleotide message known as RNAIII (17, 29, 32).

The current understanding is that RNAIII is involved in both repressing the transcription of cell surface protein genes and activating the transcription of extracellular protein genes (32, 33). In the case of alpha-toxin, a direct interaction between RNAIII and the alpha-toxin message is required for full translation (28, 29). When synthesized from a heterologous promoter in an agr-null mutant, RNAIII returns a wild-type pattern of virulence factor messages and translation products (28, 39; our unpublished data). Mutational analysis of RNAIII has shown that delta-toxin does not play a role in the regulation of virulence factor gene expression, suggesting that the message is the effector molecule of the agr-sar system (19, 32, 33).

An additional layer of complexity in this model is added by the observation that SarA is transcribed on three different overlapping messages known from largest to smallest as A, C, and B (3). These messages are initiated from distinct upstream promoters named P1, P3, and P2, respectively. Each message ends at a common terminator downstream of the SarA open reading frame. The P1 and P2 promoters are dependent on ^SA, the primary sigma factor in S. aureus, while the P3 promoter is dependent on ^SB, a multiple-stress-responsive sigma factor (23, 39). These data can be linked to the observation that virulence factor production is regulated by environmental conditions. For example, SarA message and levels of alpha-toxin are both elevated when the bacteria are exposed to oxidative stress (6). Additionally, SarA, its messages, or unidentified translation products encoded by the messages may play a direct role in virulence factor regulation. This hypothesis is based on the observation that decreases in transcription of the gene encoding cell protein A requires a different trans-encoded sar message in sar as compared to agr mutants of S. aureus (8).

Despite the progress made in understanding the agr-sar system, the best available evidence suggests that additional regulatory factors are required for virulence factor production (2, 11, 12, 39). One example of this difference is seen with alpha-toxin. This hemolysin is concomitantly transcribed, translated, and secreted 2 h after the appearance of RNAIII; however, RNAIII remains elevated while alpha-toxin production falls within 1 h of reaching peak production (39). In addition, unidentified regulatory molecules have also been invoked to explain the decrease in alpha-toxin message seen when S. aureus is treated with protein synthesis inhibitors (2). In the present study, we used transposon Tn917 mutagenesis to identify a gene that encodes a previously undescribed regulator of alpha-toxin.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, phage, plasmids, media, growth conditions, and virulence factor assays. Bacteria, bacteriophage, and plasmids used in this study are described in Table 1. S. aureus was cultivated in tryptic soy broth (TSB) (Difco Laboratories, Detroit, Mich.) and incubated at 37°C with rotary agitation at 200 rpm or grown on tryptic soy agar plates (TSA). Escherichia coli was grown at 37°C in Luria-Bertani broth with agitation or on Luria-Bertani agar. Antibiotic-resistant staphylococci were selected and maintained in tetracycline (10 µg ml¹) or erythromycin or chloramphenicol (5 µg ml¹). Resistant E. coli were grown in media augmented with 100 µg of ampicillin ml¹. The method used for quantitative measurement of alpha-toxin has been previously described (15, 27). Assays for coagulase and protease have been described by Hart et al. (15).

DNA isolation. Chromosomal DNA was isolated from S. aureus using the method of Dyer and Iandolo (13). Staphylococcal plasmid DNA was purified using a Qiagen (Chatsworth, Calif.) Plasmid Mini Kit. The plasmid isolation procedure was modified by incubating the cell suspension in P1 buffer containing 100 µg of recombinant lysostaphin ml¹ (AMBI UK, Trowbridge, United Kingdom) for 30 min at 37°C. The procedure was further modified by removal of the precipitate formed after the addition of neutralization buffer by centrifugation for 30 min. Routine procedures were used to isolate DNA from E. coli (1).

Recombinant techniques. Plasmids were constructed and amplified in E. coli strain DH5 using standard recombinant DNA techniques (1). Restriction endonucleases, DNA modification enzymes, and polymerases were obtained from Promega (Madison, Wis.) and used as recommended by the manufacturer.

1

1

Construction of S. aureus strain PM466. To create PM466, the agr locus was deleted from strain RN6390 by allelic exchange using plasmid pJM48. Plasmid pJM48 was constructed in multiple steps. Initially, a 3.3-kb ClaI-HpaI Agr-encoding fragment was removed from plasmid pRN6650, creating pJM33. The 2.8-kb EcoRI-HindIII fragment from pJM33 that contains agr-flanking DNA was then cloned into similar sites in pBC SK (Stratagene, La Jolla, Calif.). This fragment was removed from pBC SK by digestion with PstI and transferred into similar sites in the temperature-sensitive shuttle vector pSPT181(ts), creating pJM48.

Transposon Tn917 mutagenesis and phenotypic screens. Strain PM466 was subjected to mutagenesis with transposon Tn917 carried on plasmid pTV1 (41). To overcome the low transformation efficiency of S. aureus, a colony of PM466 harboring pTV1 was grown at 32°C, the permissive temperature, on TSA containing chloramphenicol, to create a pool of bacteria with pTV1. Mutant bacteria were selected at 42°C and screened for protease activity on nutrient agar (38) with 5% skim milk (Difco Laboratories) and hemolytic activity on blood agar base (Difco Laboratories) supplemented with 5% rabbit blood.

Southern and Northern hybridization. Digested staphylococcal chromosomal DNA was subjected to electrophoresis through 0.7% agarose gels, transferred to nylon membranes (MagnaGraph; Micron Separators Inc., Westborough, Mass.), and probed using the Genius system (Boehringer Mannheim, Indianapolis, Ind.) as instructed by the manufacturer. Hybridizations used a randomly primed digoxigenin-labeled 6.1-kb BamHI fragment from pRN6650 that contains agr plus flanking DNA or a 1.2-kb HindIII probe from pIM36 that encodes the transposase of transposon Tn551, standard buffer plus 50% formamide for prehybridization and hybridization, and stringent washes performed at 68°C (27). Detection used the chemiluminescent substrate disodium 3-(4-methoxyspiro{1,2-dioxetane-3,2'-(5'-chloro)tricyclo[3.3.2.2³⁷]decan}-4-yl) phenyl phosphate (CSPD) (Boehringer Mannheim).

Inverse PCR and nucleotide sequencing. Inverse PCRs contained chromosomal DNA from strains PM614, PM615, or PM616 digested with either EcoRI or PstI and self-ligated at a concentration of 5 ng of DNA µl¹. The Tn917-specific outward facing primers were 5'-GAGCATATCCACTTTTCTTGGAG-3' and 5'-CACAATAGAGAGATGTCACGTC-3' (GenBank accession no. M11180). DNA was amplified by the method of Coen (10). The nucleotide sequence for rot was obtained using an Applied Biosystems 373A or 377 DNA Sequencer with dye terminator cycle sequencing chemistry (Perkin-Elmer, Foster City, Calif.) on Qiagen purified DNA. Template DNA consisted of a pool of three independently amplified PCR products. Sequencing primers were designed to extend the newly acquired sequence. Additional S. aureus sequence data were obtained from The Institute for Genomic Research (website at http://www.tigr.org). Data were analyzed using the Wisconsin Genetics Computer Group sequence analysis software package Version 9.1.

Construction of S. aureus strain PM720. PM720 was created by allelic exchange using S. aureus PM614 and plasmid pJM202. Plasmid pJM202 is plasmid pSPT181(ts) with a 1.3-kb PCR fragment generated from the wild-type S. aureus strain RN6390 using primers that correspond to sequence upstream and downstream of rot, 5'-CAAAGCCTGACACGACAATCC-3' and 5'-CTGAAAGATGAGACAGTAGATG-3', respectively. To construct pJM202, the rot-containing PCR fragment was cloned into plasmid pCRII (+) (Invitrogen, Carlsbad, Calif.) and verified by restriction endonuclease and sequence analysis. The rot fragment in pCRII was removed by digestion with EcoRI and moved into a similar site within the multiple-cloning site of pSPT181(ts).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of S. aureus strain PM466. Strain PM466 is a new agr-null derivative of S. aureus RN6390 created by allelic exchange using plasmid pJM48. The deletion in PM466 encompasses the entire agr P2 operon and the first 379-bp of the P3 transcript. The expected chromosomal deletion was confirmed in strain PM466 by Southern analysis (Fig. 2). Measurements of virulence factor activity demonstrated that post-exponential-phase culture supernatant fluids from PM466 had less than 5% of the protease and alpha-toxin activities associated with RN6390. Coagulase activity was approximately 10-fold higher in PM466 than in the wild-type control. RNAIII in PM466 could not be detected by Northern analysis (data not shown).

View larger version (31K): [in this window] [in a new window]   FIG. 2.   (A) Restriction map of the wild-type allele of agr in RN6390 (top) and the agr-null allele (agr) in PM466 (bottom). The agr P2 operon and P3 transcript are represented by the striped and shaded boxes, respectively. Chromosomal DNA on the 3' end of the P2 operon and 3' end of the P3 transcript are represented by black and gray lines, respectively. Abbreviations: C, ClaI; E, EcoRI; H, HincII; H/C, ligation of a HincII site with a T4 polymerase blunt-ended ClaI site. (B) Chromosomal DNA from RN6390 (lane 1) and PM466 (lane 2) digested with EcoRI and analyzed by Southern hybridization using agr and flanking single-stranded DNA as the probe. DNA hybridizing to agr-flanking regions is marked with arrows.

Transposon Tn917 mutagenesis and transductional analysis. Strain PM466 was subjected to mutagenesis with transposon Tn917. Approximately 2 × 10⁴ bacteria with chromosomal insertions of the transposon were screened for proteolytic activity on skim milk agar, both with and without erythromycin. Eleven protease-positive strains were isolated. To rule out mutations in genes that only activate protease expression, the erythromycin-resistant, protease-positive strains were screened for hemolytic activity on rabbit blood agar plates. Nine of the original eleven isolates had alpha-toxin activity. The loss of plasmid pTV1 from these nine strains was confirmed by testing for vector-encoded antibiotic resistance on TSA supplemented with the MIC of chloramphenicol. The lack of growth of the nine strains in this medium suggested that the erythromycin resistance was mediated by a chromosomal insertion of the transposon.

View larger version (14K): [in this window] [in a new window]   FIG. 3.   BLASTP alignment of the rot gene product (Rot) and the S. aureus regulatory proteins AgrA (AgrA e value = 1.9; SarA e value = 5.9) and S. epidermidis AgrA (e value = 0.49). Rot was used as the query sequence against the nonredundant database at the National Center for Biotechnology Information. Identities are shown, + denotes similarity, and numbers at right refer to amino acids in the respective proteins.

Verification and initial characterization of the rot mutation. As viewed on indicator plates, inactivation of rot restores a post-exponential-phase protease- and alpha-toxin-positive phenotype to the agr-null strain of S. aureus PM466. To quantify the effect of the rot mutation on virulence factor production, we compared alpha-toxin activity in culture supernatant fluids from strains RN6390, PM466, PM614, and PM720 (Fig. 4). PM466, the agr-null strain, has approximately 4% of the activity associated with RN6390, its wild-type parental strain. Compared to the activity seen with PM466, the rot mutation in PM614 results in a 40-fold increase in alpha-toxin activity. This level is approximately half that associated with a wild-type strain. Similar results were seen with PM615 (data not shown).

View larger version (35K): [in this window] [in a new window]   FIG. 4.   Quantitative measurements of alpha-toxin activity in supernatant fluids from post-exponential-phase (10-h) cultures of S. aureus strains RN6390 (wild-type, gray bar), PM466 (vertically striped bar), PM614 (white bar), PM720 (diagonally striped bar), and PM702 (black bar). Relevant genotypes are shown (wt, wild-type; agr, agr-null allele, rot+, wild-type rot allele; rot::Tn917, insertionally inactivated rot).

View larger version (52K): [in this window] [in a new window]   FIG. 5.   Northern analysis of the alpha-toxin message in RNA from post-exponential-phase cultures of S. aureus strains (A) PM466, (B) PM614, and (C) PM720. The leftmost lane contains 30 µg of total RNA serially diluted (1:2) in subsequent lanes. The transcript was identified by hybridization using digoxigenin-labeled probe specific for the gene encoding alpha-toxin with chemiluminescent chemistry.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have identified a locus in S. aureus that encodes a regulator of virulence factors. This locus was named rot because our data suggests that the predicted gene product acts as a repressor of toxins. In an agr-null background, a mutation in rot increases the expression of protease and alpha-toxin. In addition, we showed increased transcription of the gene encoding alpha-toxin in an agr rot double mutant strain.

Based upon the fact that protein synthesis inhibitors can down-regulate transcription of virulence factor genes, mimicking RNAIII mutations, Balaban and Novick postulated that intermediary factors are required for transcription of virulence factor genes (2). Negative regulators of virulence factor genes constitute one possible class of these molecules. To identify loci that encode repressors, we subjected strain PM466 to mutagenesis with transposon Tn917 and screened for the restoration of two extracellular virulence factor activities.

S. aureus PM466, a genetically defined agr-null strain, was constructed for this study. PM466 is a derivative of RN6390, the wild-type strain used to define the molecular genetics of agr and sar (6, 34). In contrast to S. aureus RN6911, the published RN6390-derived agr-null mutant strain, PM466 has a specific deletion rather than an antibiotic marker and an accompanying deletion of unknown extent (33). Despite this genetic difference, quantitative measurements of cell surface and extracellular proteins suggest that the two agr-null strains have a common phenotype (data not shown). This observation supports previous findings in which the changes seen in RN6911 were interpreted as being solely due to the inactivation of agr.

Initially, we screened transposon Tn917 mutants for restored protease activity. Although the protein or proteins responsible for the zone of proteolysis surrounding single colonies of bacteria on skim milk agar have not been definitively identified, this activity has been shown to be RNAIII dependent (9). Wild-type strains produce clear zones of proteolysis on indicator plates, while the agr-null strains RN6911 and PM466 lack this activity (our unpublished data). To rule out mutations in genes that only up-regulate protease expression in the agr genetic background, the protease-positive strains were screened for hemolytic activity. S. aureus produces four different hemolysins (alpha-, beta-, delta- and gamma-toxins); however, rabbit erythrocytes suspended in agar are only susceptible to the action of alpha- and delta-toxins (14). Since the agr deletion in PM466 encompasses the gene encoding delta-toxin, the hemolytic activity associated with mutants created in the PM466 background is due to alpha-toxin (34). Despite the fact that RNAIII has been reported to be required for alpha-toxin translation, several of the proteolytic mutants displayed a hemolytic phenotype (28). Cotransductional analysis of the proteolytic- and alpha-toxin-positive mutants was used to verify genetic linkage between the extracellular protein phenotype and the erythromycin resistance encoded by the transposon. Finally, the phenotype associated with the rot allele in the agr-null strains was confirmed by demonstrating that wild-type rot is sufficient to restore an Agr phenotype to PM614.

The rot locus has limited amino acid sequence homology with the known staphylococcal transcriptional regulators AgrA and SarA. The lack of structure and function data for AgrA and SarA precludes assigning a specific biological significance to the homologous region. Genetic experiments have suggested that AgrA is an activator of agr; however, AgrA has not been shown to bind DNA within the agr promoter region (29). Recent evidence suggests that SarA is a DNA-binding protein that activates both agr and individual virulence factor genes (4, 6). While it is not possible to exclude the hypothesis that the rot gene product is a component of a protein complex or the activator of a repressor, the limited homology among the regulatory genes and the phenotype conferred by the rot::Tn917 mutation suggests that the rot-encoded protein is a transcriptional repressor. In addition to those encoded by sar and agr, two uncharacterized proteins were identified by searches that used the predicted rot translation product. We have cloned the genes for these proteins and have produced knockout mutants in S. aureus.

Quantitative measurements of alpha-toxin activity and Northern analysis of the corresponding message were used to verify rot and, in part, to define its activity. Measurements of alpha-toxin activity indicated that restoration of rot in PM614 completely represses alpha-toxin to agr levels. Moreover, rot mutations were found to only partially restore alpha-toxin activity to agr-null strains. This observation may be explained by the translational effect of RNAIII on the alpha-toxin message, although the mirroring of alpha-toxin activity and message in PM614 suggests that regulation occurs at the level of transcription. Therefore, it is possible that the rot-encoded protein may up-regulate an activator that is necessary for full alpha-toxin expression.

Collectively, our data suggest that rot encodes a repressor of extracellular virulence factor transcription. We are testing a direct model (Fig. 1B) that predicts that the rot gene product (Rot) binds within the promoter region of regulated genes during the lag and exponential phase of bacterial growth, blocking their transcription. Transcription of Rot-regulated promoters occurs when levels of the bound repressor are decreased, thus exposing the promoter and allowing the binding of transcriptional activators and RNA polymerase. This model is analogous to the H-NS-DsrA-RNA pathway of E. coli (37). In the E. coli system, DrsA-RNA is part of a complex that binds the histone-like protein (H-NS), thus relieving DNA secondary structure that inhibits the transcription of regulated genes (24). A competing hypothesis is that rot and agr encode components of independent, yet partially redundant, pathways. Under this scenario, the rot translation product may act as either a repressor or an activator of factors necessary for virulence factor synthesis. In either case, the rot-associated activity appears to be altered by an agr product or factors that are regulated by agr, because the rot mutation does not alter alpha-toxin expression found in culture supernatant fluids from stationary cultures of wild-type strains.