Structure, activity and evolution of the group I thiolactone peptide quorum-sensing system of Staphylococcus aureus

Authors


Abstract

In Staphylococcus aureus, the agr locus is responsible for controlling virulence gene expression via quorum sensing. As the blockade of quorum sensing offers a novel strategy for attenuating infection, we sought to gain novel insights into the structure, activity and turnover of the secreted staphylococcal autoinducing peptide (AIP) signal molecules. A series of analogues (including the l-alanine and d-amino acid scanned peptides) was synthesized to determine the functionally critical residues within the S. aureus group I AIP. As a consequence, we established that (i) the group I AIP is inactivated in culture supernatants by the formation of the corresponding methionyl sulphoxide; and (ii) the group I AIP lactam analogue retains the capacity to activate agr, suggesting that covalent modification of the AgrC receptor is not a necessary prerequisite for agr activation. Although each of the d-amino acid scanned AIP analogues retained activity, replacement of the endocyclic amino acid residue (aspartate) located C-terminally to the central cysteine with alanine converted the group I AIP from an activator to a potent inhibitor. The screening of clinical S. aureus isolates for novel AIP groups revealed a variant that differed from the group I AIP by a single amino acid residue (aspartate to tyrosine) in the same position defined as critical by alanine scanning. Although this AIP inhibits group I S. aureus strains, the producer strains possess a functional agr locus dependent on the endogenous peptide and, as such, constitute a fourth S. aureus AIP pheromone group (group IV). The addition of exogenous synthetic AIPs to S. aureus inhibited the production of toxic shock syndrome toxin (TSST-1) and enterotoxin C3, confirming the potential of quorum-sensing blockade as a therapeutic strategy.

Introduction

Quorum sensing is a bacterial cell-to-cell communication mechanism, which relies on the interaction of a diffusible signal molecule (a ‘pheromone’ or ‘autoinducer’) with a sensor or transcriptional activator to couple gene expression with cell population density. (for reviews, see Dunny and Winans, 1999; Williams et al., 2000). In Gram-positive bacteria, oligopeptides are often used as quorum-sensing signal molecules and are involved in controlling the development of genetic competence in Bacillus subtilis and Streptococcus pneumoniae and of virulence in Staphylococcus aureus (Grossman, 1995; Havarstein et al., 1995; Ji et al., 1995; 1997;Pestova et al., 1996; Kleerebezem et al., 1997).

Staphylococcus aureus is a primary pathogen that causes both pyogenic and toxin-mediated infections and poses a serious health threat especially in the hospital environment as a consequence of multiantibiotic resistance (Crossley and Archer, 1997). In S. aureus, the global regulatory locus, agr, is involved in regulating the expression of diverse cell surface proteins (e.g. protein A, coagulase, fibronectin-binding proteins) and exoproteins [e.g. proteases, haemolysins, toxic shock syndrome toxin 1 (TSST-1) and enterotoxin B] in concert with cell density (Novick, 1999). As S. aureus reaches the stationary phase, agr represses genes coding for cell surface proteins and activates the expression of the secreted exoproduct genes. In a number of experimental animal models, agr mutants exhibit significantly reduced virulence, highlighting the key role of this regulatory locus (Abdelnour et al., 1993; Cheung et al., 1994; Booth et al., 1995;Gillaspy et al., 1995).

The agr locus consists of two divergent transcription units, RNA II and RNA III, controlled by two promoters, P2 and P3 (Morfeldt et al., 1988; 1996a; Novick et al., 1995). Although RNA III itself is the effector for the agr response (Novick et al., 1993), the P2 operon consists of four genes, agrBDCA. AgrA and AgrC appear to constitute a classical two-component system, in which AgrC is a histidine protein kinase sensor, and AgrA is a response regulator; however, the DNA target for AgrA is not known, and it does not appear to bind either the P2 or the P3 promoter (Morfeldt et al., 1996a; 1996b). The functions of AgrB and AgrD have only been uncovered recently by Ji et al. (1995; 1997), who isolated a secreted autoinducing peptide (AIP) pheromone from spent culture supernatants, which activates transcription of both the P2 and the P3 promoters. This AIP is derived from an internal fragment of the agrD coding sequence, and its synthesis requires the product of the agrB gene (Ji et al., 1997). The pheromone receptor has been identified as AgrC, which is autophosphorylated on a histidine in response to spent culture supernatants or purified pheromone (Lina et al., 1998). Furthermore, S. aureus strains can be subdivided into at least three groups on the basis of the ability of their peptide pheromones to cross-activate or -inhibit agr expression (Ji et al., 1997). In addition, other staphylococcal species, such as Staphylococcus epidermidis, produce AIPs that can block agr-mediated S. aureus virulence gene expression (Otto et al., 1998; 1999).

The linear peptide sequences of these staphylococcal pheromones have been reported to be inactive by some groups (Ji et al., 1995; 1997; Otto et al., 1999) but not others (Balaban et al., 1998). Furthermore, although the primary peptide sequences for each of the staphylococcal pheromones so far described are different, they share a common central cysteine that is located five amino acid residues from the C-terminal amino acid. Mass spectrometry and chemical synthesis have indicated that these molecules are in fact cyclic thioesters, in which the α-carboxyl group of the C-terminal amino acid residue is linked to the thiol of the central cysteine residue (Ji et al., 1995; Mayville et al., 1999).

As the blockade of quorum sensing offers a novel strategy for attenuating infection as a result of multiple antibiotic-resistant staphylococci, we sought to gain further insights into the structure, activity and turnover of these staphylococcal pheromones. In particular, we focused on the group I S. aureus peptide and used both alanine and d-amino acid residue scanning to determine the functionally critical residues for agr activation and inhibition.

Results

The group I S. aureus AIP is oxidized to an inactive methionyl sulphoxide derivative in spent culture supernatants

To confirm the structure of the group I AIP reported by Mayville et al. (1999), we developed a modified growth medium (RCD) to enhance the yield (compared with CYGP medium) of the group I AIP produced by S. aureus RN6309B. After solid-phase extraction and preparative reverse phase (RP) high-performance liquid chromatography (HPLC), we located a single peak of activity using the S. aureus agrP3-blaZ reporter assay (Ji et al., 1995). Electrospray (ES) mass spectrometry (MS) indicated the presence of a single molecule displaying a molecular ion peak at m/z 961.613 (data not shown). This is consistent with the cyclic thioester structure shown in Fig. 1A, which was synthesized and shown to co-elute with the natural product on RP-HPLC. In contrast to the synthetic linear octapeptide, the cyclic thioester was active in the agrP3 reporter assay with an ED50 of 19 nM, a value in the same range as that obtained by Mayville et al. (1999).

Figure 1.

Structures of the group I AIP (A), group I methionyl sulphoxide AIP (B), group I AIP analogue in which the C-terminal methionine is replaced by norleucine (C), group I AIP cyclic N-acetyl pentapeptide (D) and group II cyclic N-acetyl pentapeptide (E).

RP-HPLC also revealed the presence of a major peak eluting at 10.4 min. When analysed by ES-MS, this peak displayed a molecular ion at m/z 977.626, a mass difference of 16.13 compared with the active cyclic thioester. This compound was tentatively assigned the chemical structure shown in Fig. 1B, a methionyl sulphoxide peptide, as S-oxidation of the methionine thioether side-chain is a side-reaction that occurs readily under appropriate conditions. Confirmation was obtained from the observation that the isolated methionyl sulphoxide peptide could be chemically reduced back to the active AIP (see Experimental procedures). Furthermore, the concentrations of the group I AIP and its sulphoxide derivative in RN6390B culture supernatants grown in RCD for 14 h to stationary phase were calculated to be ≈ 10 µM and 14 µM respectively. The methionyl sulphoxide AIP was unable to either activate or inhibit the agr P3-blaZ reporter to any significant extent (data not shown). The importance of the C-terminal methionyl sulphur was confirmed by synthesizing the norleucine analogue in which the sulphur is replaced by methylene (Fig. 1C). This group I AIP analogue was unable to either activate or inhibit the agrP3 reporter. In addition, replacement of the terminal methionine residue with serine, glutamic acid, lysine or proline all abolished the activity of the group I AIP.

Inhibition of agr and modulation of virulence factor production by linear and cyclic peptide pheromones

Three different S. aureus AIP pheromone groups have been identified on the basis of the ability to cross-activate or -inhibit (Ji et al., 1997). In addition, Balaban et al. (1998) reported that a linear peptide termed RIP (for RNA III-inhibiting peptide; primary sequence YSPWTNF) inhibited the agr response. To establish the agrP3-blaZ inhibition assays, we synthesized RIP and the group II and III linear and cyclic peptides and evaluated their capacity to activate or inhibit the agr response in a group I strain. In agreement with Ji et al. (1997) and Mayville et al. (1999), only the cyclic peptides were inhibitory. Neither RIP nor any of the linear peptides exhibited any agonist or antagonist activities against the group I reporter strain. Interestingly, the group I methionyl sulphoxide AIP also possessed no inhibitory activity (data not shown). Furthermore, removal of the N-terminal amino acids external to the macrocyclic template generated the corresponding cyclic N-acetylated pentapeptides for both the group I and II AIPs (Fig. 1D and E). These structural modifications converted the group I pheromone from an activator to an inhibitor (IC50 8 µM) but substantially reduced the inhibitory activity of the group II AIP pheromone (from an IC50 of 2 nM to 4 µM).

As S. aureus secretes multiple exotoxins, we sought to determine whether the inhibitory AIPs and cyclic pentapeptides influenced exoproduct production. The SDS–polyacrylamide gel shown in Fig. 2 reveals that both group II and group III AIPs block the production of at least five secreted proteins in the toxic shock syndrome toxin (TSST-1) producer, S. aureus strain KH1187A. Western blot and N-terminal sequence analyses identified three of the proteins as TSST-1, the C3 enterotoxin (SWISSPROT accession no. P23313), and lipase (SWISSPROT accession no. 10335). Similar results were obtained with the group II cyclic pentapeptide (data not shown). Conversely, the production of the cell wall protein, protein A, in a group I strain was enhanced by the addition of the exogenous group III AIP at the time of inoculation but abolished by the cognate group I AIP (data not shown).

Figure 2.

SDS–PAGE showing the influence of group II and group III AIPs on the exoprotein profiles of cell-free supernatants prepared from S. aureus group I strain KH1187A grown to early stationary phase (OD540 2.6). Lanes 1 and 5 show molecular mass markers; lanes 2 and 6, purified TSST-1; lanes 3 and 8, control supernatants; lane 4, group II AIP (25 µM); and lane 7, group II AIP (25 µM). The arrows on the righthand side indicate the positions of the exoproteins downregulated by the addition of exogenous AIP. These were identified from Western blots and N-terminal sequencing as TSST-1 (1), enterotoxin type C3 (2) and lipase (3). No SWISSPROT database matches were found for the N-terminal sequences obtained for bands (4) and (5).

Structure–activity analysis of the group I AIP

To gain further insights into the structural requirements for activation and inhibition of the agr response, we synthesized a series of analogues of the group I AIP in which: (i) the ring S was replaced with N; (ii) each amino acid residue was replaced, in turn, with the corresponding d-isomer; and (iii) each amino acid residue, apart from the central cysteine, was replaced with alanine.

As it has been speculated that the AIP thiolactone serves as an acyl donor for the covalent modification of the sensor kinase, AgrC (Mayville et al., 1999), we synthesized the lactam analogue of the group I peptide in which the central cysteine is replaced with diaminopropionic acid residue; thus, the ring S is substituted by N (Fig. 3A). If acylation of AgrC is a necessary prerequisite for activation of the agr response, then the lactam should be inactive. However, Fig. 3B reveals that the lactam is capable of activating the agr response (ED50 24 µM assuming saturation has been reached), albeit with an approximately 1000-fold lower activity than the group I AIP (ED50 19 nM).

Figure 3.

A, Structure of the S. aureus group I AIP lactam analogue in which the central cysteine has been replaced by diaminopropionic acid (DPR) to introduce an amide bond between the DPR and C-terminal methionine.

B. Dose–response curve showing the activation of the agr response in a group I S. aureus strain by group I AIP lactam analogue. Error bars represent standard deviations where n = 3. The ED50 for the lactam analogue was calculated as 24 µM estimated as ≈ 24 µM assuming that saturation had been reached.

To explore the importance of amino acid stereoconfiguration in the group I AIP, we synthesized a series of analogues in which each l-amino acid residue in turn was replaced by the corresponding d-isomer. Figure 4A reveals that, although six out of the eight analogues exhibit markedly reduced activity, replacement of neither the ring phenylalanine nor the terminal methionine with the corresponding d-isomer had any significant effect. In contrast, analogues containing d-cysteine and d-serine significantly inhibited activation of the agrP3 promoter by the cognate group I AIP (Fig. 4B).

Figure 4.

Activation (A) and inhibition (B) of the agr response in the group I S. aureus agrP3 reporter strain by analogues of the group I AIP in which each l-amino acid has been replaced in turn by the corresponding d-isomer.

Alanine scanning is a useful strategy for establishing the importance of individual amino acid residues within a peptide. We therefore synthesized a series of group I AIP analogues in which each residue in turn apart from the central cysteine (which is required for ring formation) was replaced by l-alanine. Each AIP analogue was assayed for the ability to activate or inhibit the agr response in a group I S. aureus strain.

Figure 5 shows that five of the seven alanine substitutions resulted in substantially reduced activity compared with the cognate group I AIP. Replacement of three out of four amino acid residues within the ring resulted in a 480- to 600-fold reduction in EC50, as opposed to a nine- to 13-fold reduction for the two N-terminal amino acid residues and a twofold increase in activity for the alanine-2 substitution. However, replacement of the endocyclic aspartic acid with an alanine residue converted the group I AIP from a group I strain activator (EC50 19 nM) to a potent inhibitor (IC50 33 nM). The alanine methyl moiety appears to make an important contribution to the inhibitory activity of the molecule, as substitution of the alanine-5 residue with glycine reduced the IC50 from 33 nM to 1.4 µM (data not shown).

Figure 5.

Activation and inhibition of the agr response in the group I S. aureus agrP3 reporter strain by analogues of the group I AIP in which each amino acid, except the central cysteine, has been replaced in turn by l-alanine. ED50s and IC50s were calculated as described in Experimental procedures.

Identification of a new S. aureus AIP group

Using the agrP3 reporter assay to screen for novel agrD sequences and AIPs in a collection of S. aureus clinical isolates, we identified five S. aureus strains that produced AIPs that inhibited the agr response of the S. aureus group I but did not contain agrD sequences corresponding to either group II or group III. Four of the strains were isolated from scalded skin syndrome patients in four different UK hospitals (O'Reilly et al., 1981) and one (strain 317) from peritonitis in a patient undergoing continuous ambulatory peritoneal dialysis. Although each strain contained an agrD sequence that was highly homologous to the group I agrD (85% identity at the protein level; see GenBank accession no. AF255950), comparison of the two sets of sequences revealed that each of the five strains contained a single base change in the internal agrD group I AIP coding region, such that the corresponding AIP contained a single amino acid change (aspartate to tyrosine). Interestingly, this is the same position revealed by the alanine scanning data as crucial in determining whether the AIP functioned as an activator or an inhibitor.

These data raised the question as to whether the five S. aureus strains producing the modified group I AIP have evolved a functional agr system, i.e. have the agrC and agrB genes involved in AgrD processing and AIP sensing, respectively, co-evolved with the point mutation in agrD such that the agr system in the strains can be activated by the cognate AIP. In this case, these strains would constitute a new AIP group (i.e. group IV). Northern and dot-blot analysis (data not shown) revealed that RNA III is expressed in these strains and can be induced prematurely by the addition of spent culture supernatant from the same strain, implying that their agr system is fully functional. Figure 6 shows that the synthetic group IV AIP is an inhibitor of the agr response in group I S. aureus strains (IC50 7 µM). Conversely, the group I AIP inhibited RNA III expression in a group IV strain (data not shown).

Figure 6.

A. Structure of the group IV AIP.

B. Dose–response curve for the inhibition of the group I S. aureus agrP3 by the group IV AIP. Error bars represent standard deviations where n = 3. The IC50 for the group IV AIP (7 µM) was extracted from the sigmoidal dose–response curve as described in Experimental procedures.

Discussion

The capacity of staphylococci to produce quorum-sensing signal molecules that inhibit virulence gene expression in other staphylococcal strains has been suggested to correlate with the ability of a strain to compete with others for sites of colonization or infection (Ji et al., 1997; Otto et al., 1999). Such observations offer opportunities to design novel anti-infective agents capable of attenuating virulence through the blockade of quorum sensing. Recently, a population genetic study of methicillin-resistant (MRSA) and methicillin-sensitive (MSSA) strains of S. aureus revealed a high degree of conservation within the agrD gene with more than 92% of the 192 strains examined falling into group I and the remainder belonging to groups II and III, suggesting that group I is the dominant AIP group (van Leeuwen et al., 2000). In the present paper, we sought to gain some novel insights into structural requirements for the activation and inhibition of the agr response by the group I AIP. By purifying the native group I S. aureus AIP from spent culture supernatant, we identified two related AIPs, the predicted natural product together with a methionyl sulphoxide derivative. Both compounds were present in concentrations in excess of 10 µM, but only the cognate group I AIP displayed any biological activity. As the methionyl sulphoxide peptide lacked the capacity to either activate or inhibit, it is tempting to speculate that the formation of the sulphoxide constitutes a mechanism for modulating the agr response. Interestingly, using signature-tagged mutagenesis, Mei et al. (1997) identified a Tn917 mutant in the putative peptide methionine sulphoxide reductase, which was attenuated in a mouse model of S. aureus bacteraemia. It is therefore tempting to speculate that this gene product could conceivably play a role in the turnover of the group I and group IV AIPs. Further evidence for the importance of the C-terminal methionine sulphur was obtained by replacing methionine with norleucine to provide an AIP with a similar linear hydrocarbon side-chain but lacking the sulphur. This group I AIP analogue, in common with the methionyl sulphoxide derivative, was completely inactive (as were compounds in which the methionine was replaced with serine, glutamic acid, lysine or proline), further emphasizing the critical role of the C-terminal thioether side-chain. However, replacement of l-methionine with the d-isomer had no effect on the ability of the molecule to activate agr, emphasizing the importance of the possession rather than the configuration of the C-terminal methionine side-chain. Inactivation of AIPs by oxidation of the C-terminal methionine is only likely to be important for the group I and group IV strains, as neither S. aureus groups II or III nor the three different AIPs known to be produced by the coagulase negative staphylococci (Otto et al., 1998;Novick, 1999) possess a C-terminal methionine residue.

Apart from the staphylococci, no bacterial quorum-sensing signal molecules containing thiolactones have been described in other organisms. Given the potential reactivity of the thioester-containing staphylococcal AIPs, it has been suggested that activation, but not inhibition, of the agr response depends upon covalent modification of the AgrC receptor via a trans-acylation reaction (Mayville et al., 1999). Such reactions occur in the complement system where an unstable internal thioester present in, for example, complement C3 permits covalent modification of adjacent proteins and carbohydrates (Joiner, 1988). Although no direct data have been obtained for such a modification, we have presented data to show that a more stable lactam derivative of the group I AIP, in which the macrocyclic sulphur is replaced by nitrogen (i.e. the thiolactone is replaced by an amide), is still capable of activating agr. This strongly suggests that trans-acylation is not necessary for AgrC activation, although the possibility that the group I lactam is acting at a site other than AgrC cannot be completely ruled out. Our finding does, however, contrast with that of Mayville et al. (1999) and Otto et al. (1999), who reported that neither the lactam derivative of the group II S. aureus AIP nor a lactam based on the Staphylococcus epidermidis AIP was capable of activating agr in the cognate strain. The group II S. aureus lactam was, however, capable of inhibiting group I S. aureus strains, although it showed no inhibitory activity against the group II strain (Mayville et al., 1999). These data appear to highlight differences between the various AIP/AgrC systems, as alanine scanning of the group II AIP yielded no compounds capable of inhibiting the group II strains (Mayville et al., 1999). In contrast, the alanine scanning of the group I AIP presented here yielded a highly potent inhibitor of group I strains. Further work is therefore required to establish the detailed molecular mechanism by which AgrC is activated or inhibited in the different AIP groups.

Using solid-phase peptide synthesis, we undertook detailed structure–activity analysis of the group I S. aureus AIP, in which we removed the three N-terminal amino acids to generate the corresponding N-acylated cyclic pentapeptide and undertook both d-amino acid and l-alanine scanning. Analogues were obtained that (i) exhibited reduced activity as activators; (ii) were converted to inhibitors; or (iii) possessed no activity as either activators or inhibitors of a group I agr strain. Interestingly, only one compound with greater potency as an activator than the cognate AIP was identified, i.e. the replacement of the group I AIP serine with alanine reduced the ED50 from 19 to 9 nM. However, in common with previous studies (Mayville et al., 1999; Lyon et al., 2000), compounds capable of antagonizing the activity of the natural AIP ligand were more readily identified. These compounds are all likely to function as competitive inhibitors of the cognate AIP–AgrC interaction. When the results of the alanine scanning of the group I (this paper) and group II (Mayville et al., 1999) AIPs are compared in the context of activation of the cognate strain, it is interesting to note that, although replacement of the endocyclic amino acid residue (aspartate) adjacent to the central cysteine in the group I AIP with alanine converts the molecule to a potent self-group inhibitor, the corresponding group II AIP analogue (in which serine-6 is replaced by alanine) exhibits no inhibitory activity against group II strains (Mayville et al., 1999). Although this alanine-6 variant of the group II AIP is around 10-fold less active as an activator of group II strains, it remains a potent cross-group inhibitor (Mayville et al., 1999).

While screening a collection of S. aureus strains to determine their AIP group, we identified five strains, each of which possessed the same point mutation that resulted in the substitution of tyrosine for aspartic acid in the group I AIP. From the alanine scanning data, substitution at position 5 converted the group I AIP from a self-activator to a self-inhibitor. The tyrosine-5-containing group I AIP (i.e. the putative group IV AIP) was synthesized and shown to inhibit group I strains while stimulating RNA III expression in the producer strains. Conversely, agr expression in group IV strains is inhibited by the group I AIP. Thus, it follows that evolution of the group IV strains must have involved changes in the AgrC receptor, such that it is activated by the group IV AIP. During the revision of this manuscript, Jarraud et al. (2000) and Otto et al. (2001) both reported the independent discovery of the same (fourth) group of S. aureus AIPs, in which the endocyclic tyrosine of the group I AIP is replaced by aspartate. These group IV strains were all producers of exfoliatin, the toxin responsible for scalded skin syndrome, and belong to phage type 2 (Jarraud et al., 2000), as do the five group IV S. aureus strains identified in this study (unpublished data). In addition, although it is clear that the agr system of these group IV strains is fully functional, RNA III is expressed some 2–3 h earlier in the growth phase than in other agr groups (Jarraud et al., 2000). In contrast to the results presented in this paper, Jarraud et al. (2000) reported that spent culture supernatants from the group IV strains activated rather than inhibited agr in group I strains. However, this study used the purified synthetic group IV AIP, and it is possible that the spent supernatants used by Jarraud et al. (2000) contained other components capable of influencing agr expression.

From the data presented by Mayville et al. (1999) and by Otto et al. (1999) and in the present paper, it is clear that the thiolactone moiety is essential for both activation and inhibition of the agr response. Using solid-phase peptide chemistry, AIPs can be generated with different primary amino acid sequences for evaluation as quorum-sensing blockers. In the group I S. aureus strain KH 1187A, the production of α-haemolysin, TSST-1, enterotoxin C3 and other exoproducts can be abolished by the provision of the exogenous group II or III AIP or the corresponding cyclic pentapeptides, illustrating the potential of quorum-sensing blockade in modulating virulence factor production in vitro. Indeed, Mayville et al. (1999) have shown that the group II AIP is capable of attenuating skin abscess formation by a group I S. aureus strain. During the revision of this manuscript, Lyon et al. (2000) reported that intragroup activation and intergroup inhibition are both mediated by the same group-specific receptors and that the truncated group II cyclic pentapeptide is a potent cross-group inhibitor. The latter finding is in agreement with our observations on the inhibitory properties of the cyclic pentapeptides, although there is an apparent 20-fold difference in the IC50 values obtained for the group II cyclic pentapeptide (0.2 µM versus 4 µM). The reason(s) for this discrepancy are not yet clear, but may be a consequence of differences in the assay conditions and reporter strain used to determine the IC50s.

Experimental procedures

Bacterial strains and growth media

S. aureus strains RN6390B (agr+) and RN6390B containing the agrP3-blaZ fusion plasmid pRN6683, which functions as a sensor for the group I S. aureus octapeptide pheromone (Ji et al., 1995), were kindly provided by Dr Richard Novick (New York University Medical Centre, New York, USA). S. aureus strain KH 1187A is a toxic shock syndrome toxin (TSST-1)-producing strain obtained from Professor K. Holland (Department of Microbiology, University of Leeds, UK). Clinical isolates of S. aureus were obtained from Dr Carol Webster and Professor Roger Finch, Division of Microbiology, School of Clinical Laboratory Sciences, University Hospital, Nottingham, and included four scalded skin syndrome-associated strains isolated from the UK (O'Reilly et al., 1981). Strain 317 was isolated from peritoneal dialysate from a patient undergoing continuous peritoneal dialysis. Staphylococci were grown either in CYGP broth (Ji et al., 1995) or, for the purification of the cyclic octapeptide pheromone, in a defined medium (RCD) consisting of serum-free RPMI cell culture medium (Sigma) supplemented with 6% (w/v) casamino acids and 2% (w/v) glucose.

Purification of the group I AIP

Staphylococcus aureus RN6390B was grown overnight in RCD broth (1 l) and, after the removal of bacterial cells by centrifugation, the supernatant was filtered through a 0.22 µm membrane filter followed by a 5 kDa size exclusion filter (Pall Gelman Sciences). The filtrate obtained was lyophilized, reconstituted in water and the peptide pheromone isolated using a solid-phase extraction procedure on Sep-pak Plus C18 cartridges (Millipore) with 5%, 20%, 40% and 60% (w/w) acetonitrile as eluants in 0.05% aqueous trifluoroacetic acid (TFA). The fractions obtained were assayed for biological activity, and the active fractions were combined and rechromatographed on preparative RP-HPLC (Kromasil 100-5 C8, 10 × 150 mm; Hichrom) using a linear gradient of 20–80% eluant B (eluant A: 0.05% aqueous TFA; eluant B: 0.05% TFA in MeCN–H2O, 9:1) for 30 min at a flow rate of 4.0 ml min−1, and the effluent was monitored at 220 nm. After bioassay, the combined active fractions (Rt 10–20 min) were rechromatographed on RP-HPLC and eluted with a linear gradient of 20–45% eluant B for 13 min, followed by an isocratic mobile phase of 45% eluant B for 5 min at a flow rate of 4.0 ml min−1. The yield of natural product (as the TFA salt) was determined by weight.

Peptide synthesis

Stepwise assembly of the linear-protected group I AIP was carried out on an automated PepSynthesizer 9050 (PerSeptive Biosystems), using standard Fmoc/t-Bu chemistry (Chan and White, 2000) and 2-chlorotrityl chloride polystyrene (1% DVB, 200–400 mesh; Novabiochem, CN Biosciences) (Barlos et al., 1989a; 1989b; 1991). The linear, partially protected peptide Boc-Tyr(t-Bu)-Ser(t-Bu)-Thr(t-Bu)-Cys-Asp(Ot-Bu)-Phe-Ile-Met-OH was obtained from the resin (260 mg) using 1.0% TFA and triethylsilane (16 µl) in dichloromethane (10 ml). Cyclization was achieved by adding 1-hydroxy-7-azabenzotriazole (13.6 mg; PerSeptive Biosystems), 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (58 mg; Aldrich Chemical), N,N-diisopropylethylamine (DIPEA; 17.4 µl) and 4-(N,N-dimethylamino)pyridine (1.2 mg; Aldrich) to a solution containing 130 mg of the partially protected linear peptide in dichloromethane (100 ml). The resulting mixture was stirred at ambient temperature under a nitrogen atmosphere for 30 h, filtered, and the organic solution was evaporated in vacuo to dryness to leave the protected cyclized peptide. This material was treated with TFA–H2O–Et3SiH–EtSMe (90:5:3:2) for 2 h at ambient temperature and then evaporated in vacuo to dryness. The procedure yielded the desired cyclic peptide (Fig. 1A) together with the methionyl sulphoxide derivative (Fig. 1B) as a major side-product. The latter (12 mg) was reduced by treatment with ethyl methyl sulphide (25 µl) for 1 h at room temperature in a solution of TFA (12 ml) containing NH4I (30 mg) (Nicolás et al., 1995; Vilaseca et al., 1998; Ferrer et al., 1999), and the desired cyclic peptide WAS isolated by preparative RP-HPLC as described above. A similar approach to that described above was also taken to synthesize the linear and cyclic S. aureus group II, III and IV AIPs and each of the other group I AIP analogues described. The chemical identities of all synthetic peptides were established using mass spectrometry and 1H nuclear magnetic resonance (NMR) spectroscopy.

AgrP3 reporter assays

These were carried out using S. aureus strain RN6390B [pRN6683] essentially as described by Ji et al. (1995) using the chromogenic cephalosporin, nitrocefin, as substrate. ED50 and IC50 values were extracted from the sigmoidal dose–response curves using the prism2 program (Graphpad). All assays were carried out at least in triplicate.

PCR analysis of S. aureus agrD genes

To obtain strains belonging to each of the three S. aureus AIP groups, a collection of clinical isolates was screened by PCR using the primers agrD123F (5′-TATGCACCTGCAGCTACTAA-3′) and agrD123R (5′-TCATGACGGAACTTGCGC-3′) based on conserved regions of the agrD sequences corresponding to group I (GenBank accession no. AF001782), group II (GenBank accession no. AF001782) and group III (AF001783). PCR products were subjected to automated DNA sequencing, and the DNA sequences obtained were translated and aligned using the seqeditor and megalign programmes (DNAstar).

RNA III analysis

The level of RNA III transcript was monitored by either Northern or dot-blot hybridization with an RNA III probe (Sambrook et al., 1989). RNA was extracted from the cells using the SV total RNA isolation system (Promega). The RNA III probe was labelled and detected using the digoxigenin (DIG) non-radioactive system (Roche Diagnostics) together with a CDP-Star luminescent substrate. Blots were analysed and quantified using a Lumingraph LB980 photon camera (EG and G Berthold).

Phenotypic analysis

Exoprotein production was analysed by SDS–PAGE and Western blotting of filter-sterilized stationary phase culture supernatants. Samples plus a purified preparation of TSST-1 as a control (kindly provided by Dr Howard Tranter, CAMR, Porton Down, UK) were subjected to SDS–PAGE (12% separating gel) and either stained with Coomassie blue or electroblotted onto nitrocellulose or polyvinylidene difluoride (PVDF; Novagen). Immunoblots were probed with a polyclonal rabbit antibody raised against TSST-1, and PVDF membranes were subjected to N-terminal sequencing (Biopolymer Synthesis and Analysis Unit, University of Nottingham). Sequence similarity searches were performed using the blastx facility of the NCBI website (http://www.blast.genome.ad.jp). To examine the influence of exogenous peptides on cell wall protein A production, staphylococci digested with lysostaphin (100 µg ml−1) were subjected to Western blotting and probed with horseradish peroxidase-conjugated rabbit anti-rat IgG immunoglobulins (Dakopatts).

Acknowledgements

We thank Dr Richard Novick for provision of the agrP3-lacZ reporter strain. This work was supported by grants funded by the MRC and by studentships funded by MRC (to C.R.), BBSRC (to Z.A. and S.J.W.) and Oxoid plc UK (to P.M.), which are gratefully acknowledged.

Ancillary