A role for type I signal peptidase in Staphylococcus aureus quorum sensing

Authors


E-mail alex-horswill@uiowa.edu; Tel. (+1) 319 335 7783; Fax (+1) 319 335 8228.

Summary

The Staphylococcus aureus Agr quorum-sensing system modulates the expression of extracellular virulence factors. The Agr system is controlled by an autoinducing peptide (AIP) molecule that is secreted during growth. In the AIP biosynthetic pathway, two proteolytic events are required to remove the leader and tail segments of AgrD, the peptide precursor of AIP. The only protein known to be involved in this pathway is AgrB, a membrane endopeptidase that removes the AgrD carboxy-tail. We designed a synthetic peptide substrate and developed an assay to detect peptidases that can remove the N-terminal leader of AIP. Several peptidase activities were detected in S. aureus extracts and these activities were present in both wild-type and agr mutant strains. Only one of these peptidases cleaved in the correct position and all properties of this enzyme were consistent with type I signal peptidase. Subsequent cloning and purification of the two known S. aureus signal peptidases, SpsA and SpsB, demonstrated that only SpsB catalysed this activity in vitro. To investigate the role of SpsB in AIP biosynthesis, SpsB peptide inhibitors were designed and characterized. The most effective inhibitor blocked SpsB activity in vitro and showed antibacterial activity against S. aureus. Importantly, the inhibitor reduced expression of an Agr-dependent reporter and inhibited AIP production in S. aureus, indicating a role for SpsB in quorum sensing.

Introduction

Staphylococcus aureus is a Gram-positive bacterium that is the causative agent of numerous acute and chronic infections. S. aureus is known to secrete an impressive array of toxins, haemolysins and tissue-degrading enzymes that are required for pathogenesis (Lowy, 1998). One of the primary mechanisms of controlling the secretion of these virulence factors is a quorum-sensing system called the accessory gene regulator (Agr). The Agr system monitors the extracellular concentration of a post-translationally modified peptide that is produced and secreted by S. aureus. The peptides are seven to nine amino acids in length with the C-terminal five amino acids constrained as a cyclic thiolactone through a cysteine side-chain (Ji et al., 1997; Novick, 2003). At a critical concentration, the peptides trigger the quorum-sensing cascade and induce their own biosynthesis, hence they are referred to as autoinducing peptides (AIPs).

The molecular details of the Agr locus have been investigated in detail and the chromosomal region is known to contain two divergent transcripts named RNAII and RNAIII (Novick, 2003). The RNAII transcript is an operon of four genes, agrBDCA, that encode factors required to synthesize AIP and activate the regulatory cascade. Briefly, AgrD is the precursor peptide of AIP (Ji et al., 1997), AgrB is a membrane endopeptidase involved in generating AIP (Zhang et al., 2002; Qiu et al., 2005), AgrC is a histidine kinase that is activated by binding AIP (Lina et al., 1998), and AgrA is a response regulator that induces transcription of both RNAII and RNAIII (Koenig et al., 2004). The RNAIII transcript yields a regulatory RNA molecule that acts as the primary effector of the Agr system by upregulating extracellular virulence factors and downregulating cell surface proteins (Novick et al., 1993a). A portion of the RNAIII transcript also encodes the δ-haemolysin (Janzon et al., 1989), a small amphipathic peptide with surfactant properties that may be important in biofilm development (Kong et al., 2006). Recently, a role for the RNAIII transcript in controlling the translation of the transcription factor Rot was demonstrated (Geisinger et al., 2006), which may explain the ability of RNAIII to regulate virulence factor expression.

Among isolates of S. aureus, there are four different classes of Agr systems each recognizing a unique AIP signal (Fig. 1). All of the AIP signals retain the basic thiolactone structure, but besides the fixed cysteine, the other residues vary among the four classes (Novick, 2003). Two of the AIP structures, type I and IV, differ by only one amino acid and function interchangeably (Jarraud et al., 2000), while AIP type II and type III are divergent, giving three different groupings of AIP signals. Through an intriguing mechanism of cell-to-cell communication, these three AIP groups (type I/IV, II and III) cross-inhibit each other's functions with surprising potency in a mechanism called bacterial interference (Ji et al., 1997). Recent in vivo competition experiments have suggested Agr interference may give the producing strain a competitive advantage (Fleming et al., 2006), and inhibitory AIPs have been shown to block development of an acute infection (Wright et al., 2005). Interestingly, there is also a correlation between Agr systems and disease epidemiology (Jarraud et al., 2002).

Figure 1.

AgrD peptide sequences of all four S. aureus Agr groups. AgrD is shown as three parts: N-terminal amphipathic leader, AIP and C-terminal charged tail. The type I AgrD residues are numbered, and this same numbering is used in the text. Conserved residues are shown with asterisks. The −1 and −3 residues from the N-terminal cleavage site are shown in bold, and the reported consensus cleavage site is A–X–A (Sibbald et al., 2006). The sequences of the P+1 and NIF inhibitors are shown at the bottom.

Despite extensive studies on the Agr quorum-sensing system, the AIP biosynthetic pathway has not been fully defined. The only protein known to be involved in this process is AgrB, an integral membrane protein encoded in the Agr locus. Through a series of studies, AgrB was demonstrated to have six transmembrane spanning domains and possess endopeptidase activity that targets the 46-residue AgrD peptide (Saenz et al., 2000; Zhang et al., 2002). AgrD is the AIP precursor and is composed of three parts, starting from the N-terminus: a 24-residue amphipathic leader, followed by eight residues constituting the AIP, and lastly 14 C-terminal residues that are predominantly charged (Fig. 1). Based on previous reports, AgrB can remove the charged C-terminal tail of AgrD (Zhang et al., 2002; Qiu et al., 2005). From site-directed mutagenesis studies, AgrB has essential histidine (H77) and cysteine (C84) residues on the cytoplasmic side, suggesting that catalysis occurs on this face and that AgrB could be a cysteine protease, although efforts to inhibit activity with cysteine protease inhibitors have not been successful (Qiu et al., 2005). Assuming AgrB catalyses removal of the AgrD C-terminal tail, at least three additional steps are required to complete the AIP biosynthetic pathway, all of which remain undefined. In no specific order, the thiolactone ring has to be formed, the amphipathic leader has to be removed and the AIP signal must be exported to complete the pathway. Whether or not AgrB is involved in these other steps of AIP biosynthesis remains unclear. No other proteins are encoded in the Agr locus with activities that have been attributed to AIP production.

Considering the complexity of the AIP biosynthetic pathway, we hypothesize that additional proteins could be involved. Indeed, similar suggestions have been made in other reports. Efforts to reconstitute AIP biosynthesis in Staphylococcus epidermidis yielded low levels of signal (Saenz et al., 2000), signifying additional proteins may be required for the pathway. Reconstituting the pathway in Escherichia coli did not yield functional AIP signal (Zhang et al., 2002), nor did it yield a cleavage event to remove the AgrD N-terminal leader (Qiu et al., 2005). Other proteins, such as SvrA (Garvis et al., 2002), have been suggested as potential candidates for the pathway (Novick, 2003), implying that additional proteins may be required to produce AIP. There have even been attempts to identify AgrB protein partners through cross-linking experiments (Qiu et al., 2005).

It is possible that AgrB can catalyse most of the activities required for AIP production and secretion. In addition to removing the AgrD tail, the essential cysteine of AgrB could catalyse ring formation through thioester exchange, and AgrB could facilitate the transport of AgrD to the outer face of the membrane. However, with all of these activities, it seems unlikely that AgrB can catalyse yet another peptidase activity to remove the N-terminal amphipathic leader, a cleavage site with different properties from the C-terminus. Without other proteins in the Agr locus, we speculated that S. aureus has recruited another protein to remove the N-terminal leader, likely a peptidase conserved among bacteria generating AIP-like molecules. To address this question, we designed a peptide substrate that mimics the N-terminal cleavage site and endeavoured to identify enzymes capable of cleaving the substrate. As outlined in this report, the results did not lead to AgrB, but in fact suggest that type I signal peptidase is responsible for removing the N-terminal leader of AgrD.

Results

Development of a peptidase assay

We postulated that a peptidase may exist in S. aureus to remove the N-terminal leader from AgrD during AIP biosynthesis. With proteases often possessing overlapping cleavage specificities, genetic screens for protease activities can be challenging (Stephenson et al., 2003). To identify peptidases that remove the AgrD leader, we pursued a biochemical approach using the Agr type I system, the best studied of the Agr subgroups as a model system (unless indicated, all references to AgrD and AIP refer to Agr type I). As the final AIP structure has been determined by mass spectrometry (Ji et al., 1997; Kalkum et al., 2003), the N-terminal cleavage site is known to be NIAA↓YSTCDFIM, with NIAA being the end of the amphipathic leader and YSTCDFIM being the linear AIP peptide sequence (Fig. 1). Knowing this cleavage information allows the design of synthetic substrates for monitoring peptidase activity. Using strategies gleaned from other peptidase studies (Dassanayake et al., 2004), we designed the following synthetic substrate: Fluorescein-linker-NIAAYST-amide (Fig. 2A). Throughout this report, the fluorescein peptide substrate will be referred to hereafter as ‘Pep1’ (Table 1). When Pep1 is cleaved at various positions, the end-products can be separated by size and charge using agarose gel electrophoresis, or alternatively by other chromatographic means. The fluorescein moiety will fluoresce under UV light, allowing convenient monitoring of cleavage events, and the cleavage sites can be mapped using control peptides. The amino acids and fluorescein group combine to give Pep1 a net negative charge of −1. The N-terminal cleavage products will have a net negative charge of −2 with the added carboxyl group, aiding their separation from the starting substrate, while the C-terminal cleavage products are no longer observed. To follow peptidase activities, three control peptides were designed: Fluorescein-linker-NIAA (AAf), Fluorescein-linker-NIA (Af) and Fluorescein-linker-N (Nf). Each control peptide represents potential cleavage events on the AgrD precursor. The Pep1 and control peptides were all synthesized and the sizes confirmed by mass spectrometry.

Figure 2.

Schematic of peptidase assay.
A. Fluorescein peptide substrate (Pep1) mimicking the AgrD N-terminal cleavage event. The shaded residues represent the end of the AgrD amphipathic leader and the unshaded residues are part of AIP. The products of cleavage at the correct position are shown.
B. Four per cent TAE agarose gel with 5 μM Pep1 or controls (Nf, Af, AAf) separated by gel electrophoresis and exposed to ultraviolet light. The structures of Pep1 and controls loaded in each lane are shown.

Table 1.  List of peptides used in this report.
PeptideAbbreviated nameNotes
  • a. 

    The linker is amino-hexanoic acid.

Fluorescein-linka-NIAAYST-amidePep1Assay substrate mimicking AgrD N-terminal cleavage, substrate for SpsB
Fluorescein-link-NIAAAAfControl for NIAA↓YST cleavage
Fluorescein-link-NIAAfControl for NIA↓AYST cleavage
Fluorescein-link-NNfControl for N↓IAAYST cleavage
Fluorescein-link-NIAAPST-amideP+1fAssay substrate with Y changed to P
NIAAPST-amideP+1P+1 peptide without fluorescein
APST-amideAPSTTruncated P+1 peptide control
AIFNPST-amideAIFInhibitor without fluorescein
NIFKPST-amideNIFImproved inhibitor without fluorescein

The separation and visualization of the fluorescein-containing peptides can be achieved using techniques similar to the treatment of small-molecular-weight DNA fragments. In a typical experiment, a 4–4.5% Tris-acetate-EDTA (TAE) agarose gel is prepared and a few nanomoles of fluorescein peptide are loaded from a stock solution or enzyme reaction. The peptides are separated at voltages similar to DNA gel separations (i.e. 100–125 V), and the fluorescein-containing bands can be visualized by direct placement on a UV box. As shown in Fig. 2B, Pep1 is easily resolved from the three controls. The AAf and Af control peptides run close together, but they can be reproducibly separated in longer runs.

While the assay is easy and convenient, there are certain limitations. The fluorescein group is hydrophobic, limiting the solubility of the peptide. Due to the solubility problem, reactions were performed at a fixed concentration of 150 μM Pep1, which may be below the Km for some peptidases. Additionally, a significant quantity of Pep1 must be cleaved to visualize the band on an agarose gel, requiring a considerable amount of peptidase turnover. These complications necessitated the use of long incubation times in order to see the reaction products. The visualization of reactions with agarose gels also limits the quantitative nature of the assay. However, trends can still be observed, and for all peptidases investigated, we see a strict dependence on time and amount of enzyme added.

Peptidase activities in S. aureus

Using the developed assay, S. aureus samples were prepared to screen for peptidases involved in removing the N-terminal leader of AgrD. Two strains were selected for these studies, wild-type strain SH1000 and strain SH1001, an agr deletion mutant of SH1000 (Horsburgh et al., 2002). Three different samples were prepared from each strain and tested in the peptidase assay. In case the peptidase is secreted, the spent medium of each strain was collected and assayed to check for activity. To test cell-associated activities, such as cell wall or membrane-bound enzymes, whole cells were washed and assayed. Finally, to detect cleavage events that occur in the cytoplasm, lysates were prepared of each strain and assayed. As shown in Fig. 3A, peptidase activities were detectable in all three fractions. Notably, there was little, if any, difference between the wild-type and agr mutant strains, indicating the peptidase activities are not catalysed by proteins in the agr operon, such as AgrB.

Figure 3.

Peptidase activities in S. aureus.
A. Activities in strain SH1000 and SH1001 (agr mutant) that can cleave Pep1. Samples of whole cells (Wc), media extract (Med) and cytoplasm (Cyt) were incubated with Pep1 and separated on a 4.5% TAE agarose gel. Peptide standards (Pep1, AAf, Af, Nf) were added as a reference.
B. The same experiment as in gel (A) except EDTA was added to all reactions.
C. Longer electrophoresis of an agarose gel to show the separation of the Af and AAf bands and the alignment of the AAf band with the whole-cell activity. The activity in the media extract and the effect of adding EDTA is also shown.

The secreted fraction of each strain yielded the smallest product, and by comparison to control peptides, the unidentified peptidase cleaved between residues N21 and I22 of AgrD (Fig. 3A), which is several residues removed from the expected cleavage position (see Fig. 1 for residue numbering). Through further testing, this secreted protease is sensitive to EDTA (Fig. 3B), suggesting the enzyme is a metalloprotease. In the whole cell fraction, both strains displayed strong peptidase activity that cleaved between residues A24 and Y25 and this activity was resistant to EDTA treatment (Fig. 3A and B). Confirmation of this cleavage position was obtained through longer agarose gel runs (Fig. 3C). The activity appeared stronger in the agr mutant, but in additional experiments, this difference appeared minimal (data not shown). Importantly, this cleavage event is in the correct position to release the AIP signal from the N-terminal amphipathic leader of AgrD. As the activity is observed in whole cells, it is likely to be a cell wall-bound, membrane-associated or membrane-bound peptidase.

A third peptidase activity was detected in the cytoplasmic fraction of S. aureus (Fig. 3A). This peptidase cleaved Pep1 between the alanine residues (A23 and A24), which was verified through additional gel electrophoresis experiments (data not shown). Notably, the cytoplasmic peptidase was sensitive to the presence of EDTA (Fig. 3B), suggesting another metalloprotease. Surprisingly, a similar peptidase activity that cleaved in the same position and displayed EDTA sensitivity was identified in a cell wall fraction of S. aureus (data not shown). Whether or not these activities are the same is not clear, as some level of cross-contamination in samples is possible. Alternatively, the two activities could be separate enzymes with overlapping cleavage specificities. Further study of either enzyme that cleaved between the alanine residues was not pursued.

Identification of the secreted metalloprotease

With our new strategy to identify peptidases in S. aureus, it is important to demonstrate the feasibility of the new peptidase assay. As this section will outline, peptidases can be detected with this approach, profiled with inhibitors, purified and finally identified, setting the stage to investigate other peptidase activities of S. aureus. Considering the wealth of knowledge about secreted metalloproteases in S. aureus (Dubin, 2002; 2003; Shaw et al., 2004), this activity was the most straightforward to identify for initial studies. Additionally, it is important to note that the steps of AIP biosynthesis are not defined. It is possible that AIP is released as a longer precursor, such as cleavage between AgrD residues N21 and I22, and then converted to the active form by cleavage between residues A24 and Y25.

The metalloprotease activity was purified from the S. aureus conditioned media using ammonium sulphate precipitation and anion exchange chromatography. Active fractions were identified using the peptidase assay. A protein band of ∼40 KDa was purified to greater than > 90% homogeneity by SDS-PAGE, and the presence of this band coincided with the ability to cleave Pep1 (data not shown). As expected, the activity was completely inhibited by chelators EDTA and EGTA. Importantly, the protein was functional in zymography analysis demonstrating it has protease activity (data not shown). Based on the size of the protease, inhibition profile, and cleavage between asparagine and isoleucine residues, we hypothesized that the enzyme is aureolysin. Aureolysin is the only known secreted metalloprotease in S. aureus (Arvidson, 1973; Drapeau, 1978; Dubin, 2002; Shaw et al., 2004). The preferred cleavage site is at the N-terminal side of bulky hydrophobic residues, such as isoleucine, which is consistent with the activity in our assay, and after zymogen cleavage, the active size of aureolysin is 40 KDa (Shaw et al., 2004). To confirm this prediction, the aur gene was inactivated by plasmid integration. As anticipated, the secreted metalloprotease activity was absent in the peptidase assay of the aur mutant (Fig. 4A).

Figure 4.

Testing of peptidase activity and AIP production in S. aureus aureolysin (aur) mutant.
A. Peptidase assay on the media extracts from wild-type and aur mutant strains. Samples were incubated with and without EDTA as indicated, and separated on a 4% TAE agarose gel. Control peptides (Pep1, Nf, Af AAf) were run as a reference.
B. Inhibition assay to detect the presence of AIP. The fluorescence of S. aureus reporter strain AH430 (Agr type II) was measured at 23 h growth in the presence of TSB (reporter control, white bar), supernatant from agr mutant strain SH1001 (agraur+, dark shaded bars), supernatant from wild-type strain MZ100 (agr+aur+, black bars), and aur mutant AH320 (agr+aur-, light shaded bars). As indicated, supernatants were diluted 10-fold and 100-fold into the inhibition assay.

To explore the role of aureolysin in S. aureus quorum sensing, supernatants of the aur mutant were tested for AIP levels. The presence of AIP can be detected using a bioassay that reports on the activation of the Agr P3 promoter. To develop this bioassay, we exploited an intrinsic property of S. aureus termed ‘bacterial interference’, where different classes of AIP molecules cross-inhibit quorum sensing in other strains (Ji et al., 1997). For testing the aur mutant, supernatants of the mutant and an aur+ strain were tested for inhibition of quorum sensing in an Agr type II strain (SA502A). Quorum sensing is monitored with a reporter plasmid containing the P3 promoter of the RNAIII transcript driving GFP expression (plasmid pDB59). Normally, the P3 promoter will induce GFP in late log to early stationary phase, but the presence of AIP from an Agr type I or type III strain inhibits quorum sensing by competing for the AgrC surface receptor and preventing expression of GFP (Ji et al., 1997; Lyon and Novick, 2004). When the supernatants of the aur– and aur+ strains were added to SA502A with plasmid pDB59, they behaved identically at every dilution tested (Fig. 4B). Similarly, when strain SH1000 is grown with the reporter in the presence of EGTA, there is no change in the quorum-sensing response (data not shown). These results provide convincing evidence that aureolysin is not necessary for AIP biosynthesis, or in some other aspect of quorum sensing. It is possible that another peptidase can substitute for aureolysin during the production, but clearly, the enzyme is not essential for the biosynthetic pathway.

Studies on the cell-associated peptidase

The only peptidase that cleaved in the correct position on the AgrD precursor was the cell-associated activity (Fig. 3). To gain insight on this enzyme, protoplasts were prepared, separated from cell wall components and tested in the peptidase assay, and the correct activity was identified in the protoplast fraction (data not shown). When the protoplasts were treated with detergent, such as Triton X-100, the activity was extracted from the protoplasts, suggesting the enzyme is membrane bound or an integral membrane protein. To confirm this observation, membranes were purified from S. aureus and tested in the peptidase assay. Again, the correct EDTA-resistant activity was observed in both SH1000 and SH1001 (Fig. 5), consistent with the whole-cell assay, providing convincing evidence that the peptidase is membrane associated. Again, no Agr dependence was observed, suggesting the activity is chromosomally encoded outside the agr locus.

Figure 5.

Pep1 cleavage activity in membranes. Reactions were run on 4.5% TAE agarose gels with controls (Pep1, AAf, Af). Membranes were purified from SH1000 (Agr+) and SH1001 (Agr–), incubated with Pep1 with and without EDTA (1 mM), and separated by gel electrophoresis.

A common strategy to classify a peptidase is through inhibitor profiling. An entire panel of cysteine and serine protease inhibitors, in combination with chelators EDTA and EGTA, were exposed to the purified membranes and then tested in the peptidase assay (see Experimental procedures). Surprisingly, the activity was resistant to every inhibitor tested. The experiment was repeated using whole cells as the source of the protease with the same result (data not shown).

The compilation of all these results suggests the unknown activity is a type I signal peptidase, hereafter referred to as ‘signal peptidase’. Signal peptidases are essential, membrane-anchored enzymes that release proteins passing through the Sec or Tat secretion systems (Paetzel et al., 2002; van Roosmalen et al., 2004). The active site is on the outer face of the S. aureus membrane and cleavage of Pep1 also occurred at the outer face of the membrane using whole cells. Consistent with the inhibitor profiling results, signal peptidases are resistant to all common protease inhibitors (Paetzel et al., 2002). Additionally, these enzymes are known to cut at cleavage sites analogous to the one in AgrD. The consensus cleavage site follows an A–X–A motif in a signal sequence with significant wobble allowed at the −3 position. Small and branched chain residues (A, S, T, G, I, V, L) are allowed at this residue position, including isoleucine, which is present in the AgrD sequence (Fig. 1). Moreover, signal peptidases are house-keeping enzymes that are highly conserved in bacteria and thus not part of the agr locus. Based on these enzyme properties and our observations, we hypothesized that S. aureus utilizes signal peptidase in the AIP biosynthetic pathway.

Purification of signal peptidases from S. aureus

Staphylococcus aureus has two signal peptidase genes called spsA and spsB. The spsA gene encodes a signal peptidase that is apparently inactive, as two of the conserved catalytic residues are not present (Cregg et al., 1996). The spsB gene is essential and is known to complement an E. coli strain with a temperature-sensitive mutation in the lepB gene, encoding the only known signal peptidase of E. coli. All available evidence suggests SpsB is the primary signal peptidase (Cregg et al., 1996). Although it remains to be demonstrated, the two genes appear to be co-regulated in an operon with another open reading frame (ORF) (designated SAOUHSC_00901 in the NCTC8325 genome) with the following molecular organization: ORF-spsA-spsB. The protein encoded by the ORF appears to be a membrane protein of unknown function.

To address the potential role of the S. aureus signal peptidases in removing the N-terminal leader of AgrD, both the SpsA and SpsB proteins were purified. Although SpsA is reportedly inactive as a signal peptidase, the enzyme has never been investigated and could potentially perform the cleavage event. Both the spsA and spsB genes were cloned into a Histidine-tag overexpression vector for affinity purification of the proteins. To improve solubility, the N-terminal membrane anchors of both proteins, residues 2–22, were removed in a strategy similar to other signal peptidase purifications (van Roosmalen et al., 2001). The tagged spsA and spsB genes were overexpressed in E. coli, and the proteins were purified by denaturing affinity chromatography and refolded by stepwise removal of urea (Fig. 6A). When the proteins were incubated with the fluorescein peptide substrate, only SpsB displayed activity, consistent with the prediction that SpsA is inactive (Fig. 6B). As expected, the activity was dependent on the incubation time and amount of protein added (data not shown). SpsB was also active with other known fluorescent peptide substrates for signal peptidase (Zhong and Benkovic, 1998), further indicating that the purified enzyme has the correct activity. The results provided the first biochemical demonstration that SpsB signal peptidase can catalyse the removal of the amphipathic leader of AgrD.

Figure 6.

Purification of signal peptidases and activity assay.
A. Typical purification of SpsB using denaturing nickel affinity chromatography. Results with SpsA were similar. As indicated, addition of IPTG induced the expression of SpsB. Lane abbreviations: SM, size marker; CFE, cell-free extract; FT, column flow-through; lanes 1–4, column washes; lanes 5 and 6, elution fractions.
B. Peptidase assay with purified SpsA and SpsB protein. SpsA and SpsB were incubated with Pep1 and reactions were separated on 4% TAE agarose gels. Controls (Pep1, Nf, Af, AAf) were also run as a reference.

Development of a SpsB inhibitor

Although SpsB can catalyse the cleavage of the peptide substrate in vitro, is this activity the same as observed in whole cells and membranes of S. aureus? As the spsB gene cannot be inactivated, one way to address this question is through SpsB inhibition. The panel of protease inhibitors used with whole cells and membranes was exposed to purified SpsB, and consistent with published reports on signal peptidases, SpsB was resistant to all the inhibitors (data not shown). Without a commercially available inhibitor, we endeavoured to generate a synthetic peptide inhibitor of this protein. It is known that the presence of a proline residue in the −3 to +1 region of a signal peptidase cleavage site will block the cleavage event (Barkocy-Gallagher and Bassford, 1992). We designed a linear peptide based on our assay substrate with a proline residue at the +1 position in place of tyrosine possessing the following sequence: NIAAPST-amide (called ‘P+1’; see Fig. 1, Table 1). We also prepared a fluorescein derivative of this proline-containing peptide (Fluorescein-linker-NIAAPST-amide, called ‘P+1f’) to determine whether it was a substrate for SpsB. When P+1f was incubated with SpsB, it could not be cleaved (Fig. 7A), which confirms the prediction that SpsB will not cleave peptides with prolines in the −3 to +1 region. Importantly, when the same proline peptide, lacking the fluorescein moiety (P+1), was added to the peptidase assay with Pep1, it inhibited SpsB activity in a dose-dependent manner (Fig. 7B). Considering 100–150× molar excess is needed to visualize inhibition, P+1 peptide is not a potent inhibitor, but the working range was adequate for these studies. As a control, a truncated version of the P+1 inhibitor was synthesized that contained only four residues, APST, of the original inhibitor (Table 1). When the APST peptide was incubated with SpsB, it did not inhibit activity at similar concentrations (Fig. 7B). These results provided evidence that P+1 could be used to block SpsB activity, which would allow the role of SpsB in AIP biosynthesis to be explored.

Figure 7.

Testing of P+1 peptide as a substrate and inhibitor.
A. Peptidase assay testing P+1f as a substrate for SpsB. TAE agarose gel (4.5%) of controls (Pep1, P+1f, AAf; see Table 1) and SpsB reactions with Pep1 and P+1f as substrates.
B. Agarose gel of controls (Pep1, AAf) and SpsB reactions with P+1 or APST peptides added at concentrations 1.5 mM, 15 mM and 22.5 mM, representing 10×, 100× and 150× molar excess over Pep1.
C and D. Agarose gel of reactions of Pep1 and SH1000 membranes (C) or Pep1 and SH1000 washed cells (D) were incubated with P+1 at concentrations of 0 mM, 3.75 mM, 15 mM, 30 mM, representing 25×, 100× and 200× molar excess over Pep1. As a control, a reaction with 30 mM APST was also included on each gel.

P+1 inhibition of peptidase activity in membranes and whole cells

Based on the properties of the peptidase identified in membranes and whole cells (Figs 3 and 5), we have assumed that this enzyme is SpsB, although it has not been confirmed. With the availability of the P+1 inhibitor for SpsB protein, we can test the prediction that the identified activity in the samples is indeed SpsB. Membranes were purified from S. aureus strain SH1000 (agr+) and SH1001 (agr–), reactions were prepared with Pep1 and the membrane samples, and P+1 inhibitor and APST control were added to the reactions at concentrations ranging from 3 to 30 mM (Fig. 7C). Similarly, washed cells of SH1000 and SH1001 were prepared and substituted for the membranes in the reactions (Fig. 7D). As anticipated, P+1 inhibited the peptidase activity in a dose-dependent manner in both the membrane and whole-cell samples, while the APST control peptide had no effect. These results provided further evidence that the peptidase detected in whole cells (Fig. 3) and membranes (Fig. 5) is indeed SpsB.

Inhibition of β-lactamase secretion

β-Lactamase can be used as an indicator of the in vivo activity of SpsB in S. aureus. β-Lactamases are well-characterized enzymes that are known to be secreted through the Sec system and released by action of type I signal peptidase. Measurement of β-lactamase activity is a common way of gauging the signal peptidase activity in the presence of secretion inhibitors (Kulanthaivel et al., 2004). BlaZ is a well-studied S. aureusβ-lactamase encoded on the pI258 multidrug resistance plasmid and has been used extensively as a reporter (Novick, 1991; Wang and Novick, 1987).

The blaZ gene was cloned from plasmid pI258 (strain RN23) into the pEPSA5 expression vector, placing blaZ expression under control of a xylose-inducible promoter. The resulting plasmid was transformed into strain SH1000 and supernatants were tested for β-lactamase activity. Using a nitrocefin assay, BlaZ activity was detected in a xylose-dependent manner in S. aureus supernatants (data not shown). To test inhibition of SpsB activity, the P+1 and APST peptides were added in increasing concentrations to strain AH488 (SH1000 with pEPSA5-blaZ) and BlaZ activity was measured at various intervals. After 6 h, P+1 caused significant inhibition of BlaZ activity (Fig. 8), with 60% and 80% decreases in activity observed at P+1 concentrations of 5 and 10 mM respectively. In contrast, similar concentrations of APST had no effect on the level of BlaZ activity. The inhibitory effects of P+1 were even more striking after 10 h of incubation (data not shown). No β-lactamase activity was detected in strain AH445, which has plasmid pEPSA5 without an insert. The effects on BlaZ activity do not appear growth-related, as all the conditions resulted in similar optical densities. Altogether, these experiments support the other results with P+1 inhibitor, demonstrating that P+1 can block SpsB activity and thus reduce secretion of proteins requiring this activity. Importantly, these experiments indicate that P+1 inhibitor can be used to probe the role of SpsB in S. aureus quorum sensing.

Figure 8.

Peptide inhibition of SpsB activity monitored with β-lactamase assays. APST control peptide (white bars) or P+1 inhibitor (black bars) were added to strain SH1000 with pEPSA5-blaZ and incubated for 6 h. BlaZ activity was monitored with a nitrocefin assay and reported as μmol min−1 ml−1 with standard deviations. Growth of strains in all conditions was similar.

P+1 inhibition of quorum sensing

Quorum sensing can be monitored in S. aureus strains using reporter plasmids with the P3 promoter driving GFP expression (Yarwood et al., 2004). As this induction requires AIP production, the reporter plasmids are useful indicator of the ability of strains to generate AIP molecules. When the P+1 inhibitor was added directly to strain AH462 (SH1000 with pDB59), a dose-dependent inhibition of quorum sensing was observed (Fig. 9A). Additionally, the amounts of P+1 used were all at levels that did not inhibit S. aureus growth (1–10 mM), eliminating growth effects as a reason for the reduced P3 promoter induction. As a control, the APST peptide was incubated at the same concentrations with strain AH462, and APST showed significantly reduced effects on quorum sensing (Fig. 9A). All of these observations are consistent with the experiments monitoring the P+1 inhibition of SpsB activity (Figs 5 and 7). The results provided the first evidence supporting a role for SpsB in AIP biosynthesis and quorum sensing.

Figure 9.

Peptide inhibition of quorum sensing in S. aureus.
A. Reporter strain AH462 (SH1000 with pDB59).
B. Reporter strain AH430 (SA502A with pDB59).
To monitor quorum-sensing activation, fluorescence was monitored over time with cells growing in TSB supplemented with peptides, and the plots shown reflect a time point at 18 h. Conditions tested were no peptide (shaded bars), APST control peptide (white bars) and P+1 inhibitor (black bars). Growth of strains in all conditions was similar.

Through further testing, we observed that P+1 could block quorum sensing in the Agr type II strain AH430 (SA502A with pDB59, Fig. 9B), suggesting the requirement for SpsB may extend to AIP biosynthesis in other S. aureus strains. Considering the variability in the AgrC surface receptor, this observation also suggests P+1 is not operating through AgrC competitive inhibition. Bacterial interference is known to require the cyclic moiety in the AIP structure (Lyon and Novick, 2004), further supporting the notion that linear P+1 is inhibiting through a SpsB-dependent mechanism. Additionally, the APST peptide had little effect on quorum sensing in strain SA502A (Fig. 9B), while retaining the serine and threonine residues conserved in the type I AIP, an AIP molecule known to be a potent inhibitor of the type II Agr system (Ji et al., 1997).

The availability of the P+1 inhibitor also allowed the role of SpsB in AIP biosynthesis to be investigated. For this experiment, we added increasing concentrations of P+1 to early log phase SH1000 cells, grew the cultures and collected supernatants at various optical densities, and tested the supernatants for AIP levels using the Agr inhibition bioassay with reporter strain AH430. While P+1 did attenuate the Agr response (Fig. 9), the effect on AIP production was less pronounced at P+1 concentrations up to 10 mM (data not shown). We postulated that the metabolic instability of P+1 was complicating the AIP bioassay, and the in vivo tests with P+1 support this argument. In experiments with a fluorescein-attached version of P+1 (P+1f), substantial degradation of the peptide was observed in the presence of SH1000 cells. Except for SpsB peptidase activity, the same activities that cut the Pep1 substrate (Fig. 3) also cut P+1f, with the dominant cleavage activity occurring between the alanine residues. In additional experiments, a P+1 concentration of over 10 mM is required to show a minor SH1000 growth defect, even though SpsB is an essential enzyme. Further, P+1 antibacterial activity was not observed at concentrations up to 30 mM, indicating the minimal inhibitory concentration (MIC) is over 30 mM (data not shown). These observations suggest that the P+1 inhibitor is not stable in the presence of SH1000 cells.

Design and characterization of an improved SpsB inhibitor

We reasoned that changing the alanine residues within P+1 may maintain SpsB inhibition, while improving metabolic stability and allowing the AIP biosynthesis tests. Using P+1 as a guide, two peptides were designed and synthesized with changes to the alanine residues, and these peptides were called ‘AIF’ and ‘NIF’ (Table 1). For the inhibitor design, the isoleucine was kept at the −3 position to maintain contact with SpsB, while the alanines were changed to larger residues, with a goal of slowing protease cleavage. Both peptides inhibited SpsB activity in the Pep1 peptidase assay, and the level of inhibition was nearly identical to the P+1 results shown in Fig. 7B. NIF was also compared with P+1 using a quantitative fluorescence assay for SpsB activity. Again at each concentration tested, the level of NIF inhibition was similar to that of P+1 (data not shown). From these in vitro studies with the new peptides, they were as effective as P+1 at inhibiting SpsB activity.

Given that P+1 is metabolized by S. aureus, a stable SpsB inhibitor should function better in the in vivo experiments. As expected of a SpsB inhibitor, NIF showed improved antibacterial activity against SH1000. At concentrations of 1–5 mM, P+1 had no effect on SH1000 growth in repeated tests (Fig. 10A), but NIF introduced a noticeable growth lag at 2.5 mM and killed SH1000 at 5 mM (Fig. 10B). This experiment suggested the MIC value for NIF is in the 2.5–5 mM range, much lower than MIC value for the P+1, which is estimated at over 30 mM (data not shown). The AIF inhibitor behaved similarly to P+1 and did not display any antibacterial activity in the 1–10 mM range (data not shown). The reason AIF did not show improved in vivo activity is not clear, and the peptide was not characterized further.

Figure 10.

Characterization of the NIF inhibitor.
A and B. Growth of strain SH1000 in the presence of SpsB inhibitors. An early log phase culture of SH1000 was transferred to a microtitre plate and incubated with shaking at 37°C, and the SpsB inhibitors P+1 (A) or NIF (B) were added at concentrations of 0, 1, 2.5 and 5 mM, as indicated. Growth was monitored by following optical densities at 595 nm.
C. NIF inhibition of quorum sensing in SH1000. Reporter strain AH462 was grown in TSB supplemented with peptides and fluorescence was monitored over time. The plot shown reflects a 14 h time point. Bars correspond to no peptide (shaded bar), APST control (white bars), P+1 (black bars) and NIF (hatched bars).
D. NIF inhibition of AIP biosynthesis. SH1000 was grown with either no peptide or the indicated concentrations of NIF and APST. Filtered supernatants were prepared, added to reporter strain AH430, and fluorescence was monitored over time. The plot shown reflects a 15 h time point. As a control, GFP fluorescence of the AH430 reporter strain was monitored and is shown as a black bar. For the other conditions, bars correspond to no peptide (shaded bar), APST control peptide (white bars) and NIF inhibitor (hatched bars).

The effect of the NIF inhibitor on quorum sensing was also tested. At concentration of 0.5 mM, P+1 had no effect on the SH1000 Agr system versus the APST control, while NIF reduced the response by 67% (Fig. 10C). At higher concentrations, NIF continued to exhibit a more potent Agr inhibition than P+1 until a concentration of 2.5 mM, where NIF began to introduce a growth lag (Fig. 10B). Thus, the improved bioactivity of NIF correlated with a stronger inhibition of the Agr response. Overall, these studies indicated the properties of NIF were superior to P+1 and further supported the link between SpsB activity and Agr quorum sensing.

NIF inhibition of AIP biosynthesis

With the improved NIF inhibitor, the requirement for SpsB in AIP biosynthesis could be determined. For the experiment, SH1000 was grown to early stationary phase with various concentrations of the NIF and APST peptides, filtered supernatants were prepared, and the amount of AIP was tested in Agr inhibition bioassays with the AH430 reporter strain. At 0.25 mM NIF, there was little effect on AIP production compared with APST and the no peptide control. However, at 0.5 mM NIF, the AIP levels decreased substantially, as evidenced by a 50% increase in GFP expression versus the 0.5 mM APST control (Fig. 10D). At NIF concentrations of 1 mM, the AIP levels decreased to a point where GFP expression was within error of the AH430 reporter control, while again APST had no effect. In control experiments, the amount of carry-over inhibition from NIF on the AH430 reporter strain was negligible. At 2 mM NIF, strong inhibition of AIP production was also observed, but the results were more variable, possibly due to some carry-over inhibition (data not shown). When the experiments were repeated with P+1, this inhibitor had no effect on AIP biosynthesis in the same 0.5–2 mM concentration range (data not shown). Altogether, NIF inhibition of SpsB reduced the amount of AIP produced, demonstrating a role for SpsB in AIP biosynthesis.

Discussion

In this report, we screened for peptidases involved in the biosynthesis of AIP. Our findings demonstrate that SpsB, a type I signal peptidase, has a role in the removal of the N-terminal leader of AgrD.

The basis of this study was the development of a new peptidase assay to monitor AgrD cleavage, which led to the detection of three S. aureus peptidase activities. All of the activities were observed in an agr deletion strain, demonstrating the enzymes are encoded outside the agr locus. Two of these activities were identified as Aur and SpsB, and as outlined herein, SpsB is the primary enzyme involved in AIP biosynthesis. In many ways, the discovery of a new role for signal peptidase in quorum sensing is perhaps not surprising. Signal peptidase is a house-keeping enzyme that is anchored to the outer face of the membrane and facilitates protein secretion through the Sec and Tat export systems (Paetzel et al., 2002; van Roosmalen et al., 2004). With its cellular location and dedicated role in secretion, the recruitment of signal peptidase to release AIP from its leader sequence parallels the normal function of this enzyme. As signal peptidases also have proposed roles in the generation of Bacillus signalling peptides and Enterococcus faecalis peptide pheromones (Lazazzera, 2001; Slamti and Lereclus, 2002; Chandler and Dunny, 2004), there are similar trends among these Gram-positive peptide regulatory systems.

Considering the requirement for signal peptidase, the lack of a typical signal peptide in AgrD might seem surprising. However, Ji and co-workers demonstrated that switching the N-terminal leader to a known signal peptide blocked AIP production and the amphipathic nature of the leader sequence targets AgrD to the membrane (Zhang et al., 2004). Further, essential residues in AgrB were identified on the cytoplasmic face of the enzyme through site-directed mutagenesis experiments (Qiu et al., 2005). These observations suggest that AgrB must bind AgrD and perform catalysis before getting the AIP precursor to the outer face of the membrane. With a signal peptide, AgrD would be targeted to a secretion system, bypassing the necessary AgrB-processing events. Bearing in mind all these observations, the reason for having an amphipathic leader on AgrD, instead of a signal sequence, is becoming more evident.

With the assignment of a role to SpsB, it is possible to revisit the previously proposed model for AIP biosynthesis (Zhang et al., 2002). Based on our findings and other recent published work (Qiu et al., 2005), we postulate the following sequence of events: following translation, the amphipathic leader targets AgrD to the membrane and specific residues within AgrD, perhaps those of the C-terminus (Zhang et al., 2004), promote an interaction with AgrB. Next, the endopeptidase activity of AgrB removes the AgrD C-terminal tail (Qiu et al., 2005), allowing the formation of a peptide–AgrB intermediate. The cysteine residue of the remaining AgrD peptide catalyses thioester exchange, promoting the formation of a thiolactone ring and displacing the peptide from AgrB. Through an unknown mechanism, the AIP precursor is transported to the outer face of the membrane, and finally, the amphipathic leader is removed by SpsB. The only significant change in this revised model is that N-terminal cleavage is catalysed by SpsB. The actual order of events and requirement for other enzymes in this pathway remain to be demonstrated.

While a role for SpsB has been outlined for Agr type I quorum-sensing system, is the enzyme required to generate AIP molecules in type II, III and IV strains? SpsB is highly conserved among S. aureus strains so it will be available for these biosynthetic pathways. However, as shown in Fig. 1, there is very little conservation throughout the N-terminal leader of AgrD, except for the isoleucine (I19) and glycine (G20) residues. These residues could serve to present the substrate to SpsB, as glycine is a known helix breaker. Traditional signal peptides often have proline or glycine residues five to six positions upstream of the signal peptidase cleavage site to facilitate presentation to the active site (Paetzel et al., 2002; van Roosmalen et al., 2004). As glycine and serine are allowed at the −1 position, the AgrD type II and IV peptides have possible cleavage sites based on signal peptidase wobble rules (Fig. 1), and additionally, cleavage at these sites would release the correct AIP molecule based on mass spectrometry determination of the structures (Kalkum et al., 2003). Similarly at the −3 position, alanine is preferred but an even broader range is allowed, including branched chain amino acids found in AgrD type I, II and IV leaders (Paetzel et al., 2002; van Roosmalen et al., 2004; Sibbald et al., 2006). In support of these sequence observations, the P+1 inhibitor blocked quorum sensing in a type II strain (Fig. 9B), suggesting that SpsB may be required for AIP biosynthesis in this strain. Meanwhile, the AgrD type III does not contain a typical signal peptidase site with tyrosine present at the −1 position (Fig. 1). This observation suggests either that a S. aureus type III strain has an unusual signal peptidase enzyme or that the AIP biosynthetic pathway is altered in some fashion.

While we have outlined a role for signal peptidase in S. aureus, the enzyme is also postulated to be involved in the generation of linear regulatory peptides from various Bacillus species (Lazazzera, 2001; Slamti and Lereclus, 2002). In Bacillus subtilis, these peptides regulate competence and sporulation, and they have been proposed to function as quorum-sensing molecules or alternatively as an internal timing device (Lazazzera et al., 1997; Perego and Brannigan, 2001). On the B. subtilis chromosome, there are seven precursor genes encoding the peptide precursors, each with a type I signal peptide. Surprisingly, purified signal peptidase did not cleave the B. subtilis peptide precursors and attempts to demonstrate a requirement for the enzyme using chromosomal mutants were not successful (Stephenson et al., 2003). These observations contrast our findings with AgrD from S. aureus, which does not have a typical signal peptide yet is cleaved by signal peptidase. Considering the abundance of secreted and membrane proteases in Gram-positives, biosynthesis of some peptide regulatory signals may be performed by committee, whatever protease catalyses the function might vary by environmental condition or strain. Such a possibility would explain why it has been challenging to define roles for proteases in these biosynthetic pathways using genetic approaches (Stephenson et al., 2003).

Across all S. aureus strains and Agr systems, whether or not signal peptidase is the only enzyme involved in removing the N-terminus of AgrD remains to be determined. While our data indicate a role for SpsB in quorum sensing, the enzyme is essential for protein secretion and thus cannot be deleted to demonstrate the newly assigned function. In our experiments, SpsB was the only detectable enzyme that could cleave between AgrD residues A24 and Y25, releasing the active AIP molecule. During the biosynthetic pathway, it is possible that a longer AIP precursor is generated and secreted, and the actual role for SpsB is to cut the precursor to the correct size of the functional AIP molecule. Similarly, it is feasible that inhibition of SpsB blocks the secretion of a protein required for AIP biosynthesis. In regards to these alternative explanations, Aur was the only detectable secreted enzyme that cleaves Pep1 (Fig. 4A), and AIP is produced in an aur mutant (Fig. 4B), demonstrating that Aur is not required for the biosynthetic pathway. Further, the other major extracellular proteases of S. aureus, ScpA, SspA and SspB, are thought to be activated through a post-translational cascade that initiates with Aur (Shaw et al., 2004), findings that suggest these secreted proteases are also not involved in AIP biosynthesis. Even if other enzymes are involved in removing the AgrD leader, the studies with the P+1 and NIF inhibitors demonstrate that SpsB catalytic activity has a role in the AIP biosynthetic pathway.

A peculiar feature of protein secretion in S. aureus is the presence of a second signal peptidase, SpsA, whose catalytic serine and lysine residues are missing (Cregg et al., 1996). Surprisingly, the spsA gene is conserved in all S. aureus genomes to date (Sibbald et al., 2006), suggesting its retention may provide some unknown function for the cell. It is not unusual to possess multiple signal peptidases, as a number of Gram-positive bacteria, such as B. subtilis and Listeria monocytogenes, produce more than one. However, contrary to S. aureus, all of these enzymes are thought to be active (van Roosmalen et al., 2004). SpsA was tested in our peptidase assay and did not display any activity, consistent with predictions that the enzyme is not active (Cregg et al., 1996; van Roosmalen et al., 2004; Sibbald et al., 2006). Interestingly, the spsA gene appears to be co-transcribed with spsB and another uncharacterized ORF on the S. aureus chromosome. Why S. aureus has evolved and retained this molecular arrangement is not clear, and the possibility that SpsA has some undefined role in quorum sensing remains to be determined.

The biosynthetic pathways for producing the peptide quorum-sensing molecules have remained largely unexplored in Gram-positive bacteria. While the thiolactone-containing AIPs have been detected and structurally characterized in a few types of Gram-positives (Ji et al., 1997; Nakayama et al., 2001; Kalkum et al., 2003; Sturme et al., 2005), many other bacteria possess chromosomal loci with strong similarities to the Agr locus. Using the assays and strategies outlined in this report, the identities of the enzymes involved in these other biosynthetic pathways could be determined. It is possible that signal peptidase plays a role in the production of many of these other AIP-like molecules. Considering the requirement for peptide regulatory systems in pathogenesis (Autret et al., 2003; Novick, 2003; Bourgogne et al., 2006), it is important to explore how the peptide signals are generated as these systems could be attractive drug targets (Muir, 2003; Chan et al., 2004). As signal peptidase is essential and surface exposed, it has already been the focus of numerous drug development efforts (Allsop et al., 1996; Bruton et al., 2003; Kulanthaivel et al., 2004). The definition of a new role for this enzyme in S. aureus quorum sensing heightens the significance of discovering signal peptidase inhibitors.

Experimental procedures

Culture media and growth conditions

A list of strains and plasmids used and their genotypes is provided in Table 2. E. coli cultures were maintained Luria–Bertani (LB) broth and S. aureus strains were maintained in tryptic soy broth (TSB). E. coli antibiotic concentrations were (in μg ml−1): ampicillin (Amp), 100; chloramphenicol (Cam), 30; kanamycin (Kan), 50. S. aureus antibiotic concentrations were (in μg ml−1): chloramphenicol (Cam), 10; erythromycin (Erm), 10; tetracycline (Tet), 10. All reagents were purchased from Fisher Scientific (Pittsburg, PA) and Sigma (St Louis, MO) unless otherwise indicated.

Table 2.  Strain and plasmid list.
Strain or plasmidGenotypeResistanceSource or reference
Escherichia coli
 DH5α-ECloning strainNoneInvitrogen
 ER2566Overexpression strainNoneNew England Biolabs
 AH486ER2566/pET28a-spsAKanThis work
 AH487ER2566/pET28a-spsBKanThis work
Staphylococcus aureus
 MZ100Type I Agr wild typeNoneShanks et al. (2005)
 RN23pI258MultidrugS. Silver
 RN6911Δagr::tetTetNovick et al. (1993b)
 SH1000Type I Agr wild typeNoneHorsburgh et al. (2002)
 SH1001SH1000/Δagr::tetTetHorsburgh et al. (2002)
 SA502AType II Agr wild typeNoneD. Bartels
 AH320MZ100/ΔaurNoneThis work
 AH430SA502A/pDB59CamThis work
 AH445SH1000/pEPSA5CamThis work
 AH462SH1000/pDB59CamThis work
 AH488SH1000/pEPSA5-blaZCamThis work
Plasmids
 pDB59P3-GFP reporterAmp, CamYarwood et al. (2004)
 pEPSA5Expression vectorAmp, CamForsyth et al. (2002)
 pEPSA5-blaZβ-Lactamase expression vectorAmp, CamThis work
 pET28aOverexpression vectorKanNovagen
 pET28a-spsASpsA (Δ2-22) expression vectorKanThis work
 pET28a-spsBSpsB (Δ2-22) expression vectorKanThis work
 pKOR1Knockout vectorAmp, CamBae and Schneewind (2006)
 pKOR1-Δauraur knockout vectorAmp, CamThis work
 pSU20Cloning vectorCamBartolome et al. (1991)

Recombinant DNA Techniques

Restriction and modification enzymes were purchased from New England Biolabs (Beverly, MA), and were used according to manufacturer's instructions. All DNA manipulations were performed in E. coli DH5α-E (Invitrogen, Carlsbad, CA). All oligonucleotides were synthesized at Integrated DNA Technologies (Coralville, IA). Plasmids were transformed into E. coli by CaCl2 heat-shock as described (Inoue et al., 1990). Non-radioactive sequencing was performed at the DNA sequencing facility at the University of Iowa.

Construction of signal peptidase and blaZ overexpression plasmids

The spsA and spsB genes were PCR amplified from MZ100 genomic DNA with oligonucleotides incorporating NdeI and XhoI sites (spsA: for 5′-GTTGTTCATATGACTTTTGTAATAGTTGGTCATGTC-3′, rev 5′-GTTGTTCTCGAGTTAAGATTTGAACTGAACAGTCCA-3′; spsB: for 5′-GTTGTTCATATGGGTAAATTTATTGTTACGCCATATAC-3′, rev 5′-GTTGTTCTCGAGTTAATTTTTAGTATTTTCAGGATTG-3′). The PCR products were digested with NdeI and XhoI and cloned into pET28a (Novagen, Madison, WI) cut with the same enzymes. The blaZ gene was PCR amplified from RN23 genomic DNA using oligonucleotides incorporating a sarA ribosome binding site and KpnI site on the 5′-end (for 5′-GTTGTTGGTACCAGGGAGAGGTTTTTATTATGAAAAAGTTAATATTTTTAATTG-3′) and an EcoRI site on the 3′-end (rev 5′-GTTGTTGAATTCGAATATTAAAATTCCTTCATTACACTC-3′). The PCR fragment was cloned into the pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA) by TA cloning. The blaZ gene was removed with an EcoRI and KpnI digest and cloned into pEPSA5 cut with the same enzymes. All constructs were verified by DNA sequencing.

Construction of an aureolysin deletion

An approximately 500 bp DNA region 5′ of the aur coding sequence was PCR amplified from MZ100 genomic DNA using the following primers (for 5′-GTTGTTGGATCCGAAAATGGTAAAACAAAGGAAGCTG-3′, rev 5′-GTTGTTGGTACCCCTAGGCATTCCTCCTGAAATCTTAAAAAC-3′), digested with BamHI and KpnI, and cloned into vector pSU20. A similar-size DNA region on the 3′ side of the aur gene was amplified using the following primers (for 5′-GTTGTTGGTACCACGCGTCAAGAAGAAGTAATGTTAAACAC-3′, rev 5′-GTTGTTGAATTCGATAGGGACACGATATATATTTAG-3′), digested with KpnI and EcoRI, and cloned in tandem with the other DNA fragment on pSU20. The combined aur deletion construct was amplified with PCR using attB-tailed oligonucleotides and cloned into the pKOR1 knockout vector with Gateway technology (Invitrogen). A markerless aur deletion was constructed using the pKOR1-Δaur plasmid as described elsewhere (Bae and Schneewind, 2006).

Peptides

Synthetic peptides used in this study were synthesized at the Macromolecular Core Facility at Pennsylvania State University (Table 1). The Core Facility confirmed the correct size of each peptide by mass spectrometry. All peptides were dissolved in sterile water and filter sterilized using 0.22 μm spin filters. Fluorescein-labelled peptides (Pep1, AAf, Af, Nf and P+1f) were dissolved at a concentration of 75 mM, and unlabelled peptides (P+1 and APST) used for inhibition studies were dissolved at a concentration of 200 mM. Attempts to use solvents, such as DMSO, improved peptide solubility, but led to the appearance of unidentified bands on the agarose gel, complicating experimental analysis (data not shown).

Fluorescein peptide-based peptidase assay

All assays were buffered with 20 mM Tris-HCl, pH 8.0, and substrates (Pep1 or P+1f) were used at a concentration of 150 μM. When present, EDTA was added to 1 mM final concentration. For assays using washed whole cells, a 400 μl aliquot of freshly washed cells (see below) was pelleted in a microcentrifuge, and the cells were suspended in the reaction mixture. Following incubation, the cells were removed by centrifugation and reaction products were analysed by agarose gel electrophoresis. Reactions using SpsA or SpsB protein contained 25% glycerol to improve activity of the signal peptidase (data not shown). In other assays, glycerol was added to 25% to facilitate sample loading. All assays were conducted at 37°C with a typical incubation time of 24 h, and in all cases, activity was observed after this amount of incubation time. To improve band visualization, particularly with membrane samples, incubation times of 48–60 h were typically used. Control peptides were stable in the incubations with no apparent degradation (data not shown). Reaction products were resolved by electrophoresis on 4% or 4.5% TAE agarose gels. Molecular biology grade agarose was used for all assays (Fisher Scientific). Typically 0.75–3 ng of control peptide or reaction product was loaded in each well and gels were run at 125 V. Peptide cleavage was visualized using a Bio-Rad Gel Doc 2000 (Hercules, CA) in DNA mode.

Protease inhibitor profiling

Inhibitor profiling studies were performed with a Sigma Protease Inhibitor Panel (product code INHIB1) supplemented with 2,2′-dithiodipyridine and phenylmethyl sulphonyl fluoride (PMSF). All inhibitors were prepared and stored according to manufacturers' directions. The inhibitors were added to reaction mixtures at the high end of their respective typical working concentration ranges. The following protease inhibitors and concentrations were tested: 4-(2-Aminoethyl) benzenesulphonyl fluoride hydrochloride (ABESF), 1 mM; 6-aminohexanoic acid, 38 mM; antipain hydrochloride, 100 μM; aprotinin, 800 nM; benzamidine hydrochloride, 4 mM; bestatin hydrochloride, 40 μM; chymostatin, 100 μM; 2,2′-dithiodipyridine, 1 mM; trans-Epoxysuccinyl-l-leucylamido-(4-guanidino)butane (E-64), 10 mM; EDTA, 1 mM; N-ethylmaleimide, 1 mM; leupeptin hemisulphate, 100 μM; pepstatin A,1.45 mM; phosphoramidon, 10 μM; PMSF, 1 μM; trypsin inhibitor type I-S, 150 μM.

Preparation of washed cells and sterile conditioned media

Overnight cultures of SH1000 and SH1001 were grown at 37°C with shaking in 50 ml of TSB to OD600 values of approximately 1.5. Cells were harvested by centrifugation at 10 000 g for 12 min. Approximately half of the cell suspensions were saved for preparing cytoplasmic and membrane fractions. The remaining cells were washed twice with 20 mM Tris-HCl, pH 8.0, and re-suspended in the same buffer. Conditioned medium was prepared by removing cells with centrifugation, sterilizing the medium with 0.22 μm filters, and concentrating the medium 10-fold using Amicon YM-10 centriplus filter devices (Millipore, Bedford, MA).

Preparation of protoplasts, membrane and cytoplasmic fractions

SH1000 and SH1001 cells were washed twice with 1× SMM buffer consisting of 0.5 M sucrose, 20 mM maleic acid, pH 6.5, and 20 mM MgCl2 (Novick, 1991). Washed cells were suspended in 20 ml of SMM buffer containing 10 μg ml−1 lysostaphin (AMBI Products, Lawrence, NY) and incubated for 1.5 h at 37°C on a nutating shaker. Protoplasts were harvested by centrifugation at 12 000 g for 15 min at 4°C, washed twice with 1× SMM buffer and suspended in 10 ml of ice-cold 20 mM Tris-HCl, pH 8.0. Protoplasts were lysed by three rounds of sonication on ice, for 5 min each, at 50% duty on a Sonifier 450 (Branson, Danbury, CT). Insoluble material was removed by centrifugation at 15 000 g for 20 min at 4°C prior to precipitating membranes by centrifugation at 100 000 g for 90 min at 4°C. Pelleted SH1000 and SH1001 membranes were washed twice with 5 ml of 20 mM Tris pH 8.0 and dissolved in 400 μl of 20 mM Tris pH 8.0. The SH1000 and SH1001 cytoplasmic fractions were centrifuged a second time at 100 000 g for 30 min at 4°C in order to remove any trace amounts of membranes, and concentrated approximately twofold using Amicon YM-10 centriplus filter devices (Millipore).

Purification of aureolysin

Strain RN6911 was used to purify the secreted metalloprotease, later identified as aureolysin. An overnight culture of RN6911 was inoculated into 3 l of LB medium, containing Tet, and grown at 37°C with shaking until stationary phase. The medium was cleared of cells by centrifugation at 12 000 g for 20 min and ammonium sulphate was added slowly to 60% (w/v) while stirring at 4°C. The mixture was stirred for an additional 2 h, and precipitated proteins were pelleted by centrifugation at 15 000 g for 1.5 h at 4°C. The proteins were suspended in 25 ml of 20 mM Tris-HCl, pH 7.8, with 2 mM CaCl2, and 4 volumes of chilled acetone was added. The mixture was held at −20°C for 1 h, and the precipitated proteins were pelleted again by centrifugation at 15 000 g for 10 min at −10°C. The resulting pellet was dissolved in 10 ml of column buffer consisting of 20 mM Tris-HCl, pH 7.8, and 2 mM CaCl2, and then dialysed against this same buffer. The sample was centrifuged at 600 g for 10 min to remove insoluble material and was loaded at 4°C onto a Q-sepharose FF (Sigma) column (1 cm × 15 cm) equilibrated with column buffer. After loading, the resin was washed with several column volumes of the same buffer and proteins were eluted with a 500 ml linear gradient of 0–1 M NaCl at a flow rate of ∼1 ml min−1. Elution fractions (a total of 96, 5 ml fractions) were assayed for activity by incubating 10 μl of each fraction with Pep1 (6.8 mM) for 2 h at 37°C and resolving the reactions on a 4% TAE agarose gel. The desired activity was found in elution fractions 52 through 63, corresponding to NaCl concentrations of ∼0.55 M and ∼0.65 M respectively. These fractions were pooled, concentrated in an Amicon YM-3 centriplus concentration device (Millipore), and dialysed extensively against column buffer. The sample was further purified on Toyopearl DEAE-650 (Tosoh Bioscience, Tokyo, Japan) resin equilibrated with column buffer and eluted with a 300 ml linear gradient of 0–0.3 M NaCl. Five-millilitre fractions were collected and the correct activity was found in fractions 25 through 31, corresponding to NaCl concentrations of ∼125 mM and ∼155 mM. Fractions 26 through 29 we pooled, concentrated and dialysed against the same column buffer. SDS-PAGE indicated the activity was purified to near homogeneity and ran as an approximately ∼40 kDa protein. Zymography analysis of protease activity was performed as described elsewhere (Heussen and Dowdle, 1980).

Purification of SpsA and SpsB

Escherichia coli strains AH486 and AH487 were used for the overexpression and purification of SpsA and SpsB. For purifying either protein, the strain was grown overnight and a 1:200 dilution was inoculated into 4 l of LB supplemented with Kan. After reaching a density of 0.7 at OD600, expression was induced with 1 mM IPTG. Following 6 h of induction, cells were harvested by centrifugation at 12 000 g for 15 min at 4°C and suspended in 60 ml of ice-cold equilibration buffer containing 0.1 M sodium phosphate, pH 6.0, with 6 M urea. Cells were lysed with 6 ml of 10× BugBuster (Novagen, Madison, WI) on a nutating mixer at 4°C. Insoluble debris was removed by centrifugation at 20 000 g for 45 min at 4°C, and the cleared cell lysate was loaded onto a 5 ml column containing Ni-NTA His-Bind Resin (Novagen) equilibrated at 4°C with equilibration buffer (pH 6.0). The resin was washed with 20 column volumes of equilibration buffer, and SpsA or SpsB was eluted with buffer consisting of 0.1 M sodium phosphate, pH 4.6, and 6 M urea. Fractions containing protein were monitored by SDS-PAGE. Proteins were refolded by a series of three dilutions followed by extensive dialysis against 20 mM Tris-HCl, pH 8.0 (dialysis buffer). First, the urea concentration of the pooled elution fractions was decreased from 6 M to ∼1.5 M by dilution with dialysis buffer. Next, the volume was reduced to ∼50 ml in an Amicon 8200 concentrator (Millipore) using a PM10 membrane (62 mm), and the urea was decreased to ∼0.5 M by a second dilution with dialysis buffer. Finally, another round of concentration and dilution lowered the urea to ∼125 mM, and the protein was dialysed extensively against dialysis buffer containing 1 mM EDTA. Protein was concentrated using an Amicon YM-10 centriplus device (Millipore) to a concentration of 0.4 mg ml−1 as determined by the Bradford method (Bio-Rad). The purity of the refolded SpsA and SpsB proteins was assessed by SDS-PAGE, and proteins were diluted 1:1 with 50% glycerol and stored at −20°C.

AIP detection by bioassay

A bioassay was developed to detect the presence of AIP in extracts. As part of the quorum-sensing response, wild-type S. aureus strains with plasmid pDB59 will induce fluorescence in late log to early stationary phase, an effect that is reproducible in both Agr type I and type II strains. Any interference with the Agr system, such as an agr mutation or competing AIP signal, will block GFP induction, creating a convenient assay to monitor AIP levels. To check AIP production with strain AH320, cells were grown to late stationary phase in TSB with Cam. Cells were removed by centrifugation and the conditioned medium was sterilized using 0.22 μm spin filters. Agr type II reporter strain AH430 was grown to an OD600 of 0.1 in TSB with Cam, and 180 μl of fresh cells were dispensed into wells in a microtitre plate (Corning 3603). Twenty microlitres of sterile-conditioned media or 20 μl of sterile conditioned media diluted 1:10 with TSB were added to the wells. The 200 μl cultures, four for each dilution of conditioned media, were grown at 37°C with vigorous shaking using a PMS-1000 Microplate Shaker (Boekel, Feasterville, PA). Eight 200 μl control cultures containing 180 μl of AH430 culture diluted with 20 μl of TSB were also prepared. A Tecan GENios microtitre plate reader (Research Triangle Park, NC) was used to monitor cell density and GFP expression by measuring optical density (595 nm) and fluorescence (excitation at 485 nm, emission at 535 nm).

Monitoring the effects of peptide inhibitors on growth and the Agr system

Strains AH462 and AH430 were grown to an OD600 of 0.1 in TSB with Cam. Aliquots of culture (190 μl) were dispensed into microtitre plates (Corning 3595 for AH462 and Corning 3606 for AH430) and 10 μl of P+1, APST or NIF in sterile water were added to give the final desired peptide concentrations. A minimum of four wells were prepared at each peptide concentration and eight ‘no peptide’ control wells were made by diluting the AH462 or AH430 cultures with 10 μl of sterile water. The microtitre plates were incubated at 37°C and monitored for cell growth at an optical density of 595 nm in the plate reader. The peptide effects on the Agr system were monitored by following GFP fluorescence.

β-Lactamase secretion assays

An overnight culture of strain AH488 was inoculated into 25 ml of TSB containing Cam at a dilution of 1:500. The culture was grown at 37°C in a 250 ml flask until it reached an OD600 of 0.25. Cell culture was dispensed into a 96-well microtitre plate (Corning 3595 plate) and P+1 or APST peptides were added to appropriate wells at final concentrations of 1 mM, 5 mM or 10 mM. The final volume for each well was 200 μl. The microtitre plate was incubated at 37°C with vigorous shaking using a PMS-1000 Microplate Shaker. Cell growth was monitored by measuring absorption at 595 nm in a Tecan GENios plate reader. After 6 h and 10 h, the amount of secreted β-lactamase activity was measured using a modified version of the nitrocefin assay (Novick, 1991). Briefly, 10 μl of culture was removed from each well and pelleted by centrifugation, supernatant was loaded onto a new microtitre plate, and 200 μl of 0.1 M nitrocefin (Calbiochem) in 50 mM sodium phosphate buffer, pH 7.0, was added to start the reaction. Hydrolysis of nitrocefin was monitored by following the absorption change at 490 nm. Initial velocities were converted into units of μmoles of nitrocefin hydrolysed per minute per millilitre (μmole min−1 ml−1) using a molar extinction coefficient for hydrolysed nitrocefin of 20 500 M−1 cm−1 at 486 nm.

NIF inhibition of AIP biosynthesis

An overnight culture of SH1000 was diluted 250-fold into test tubes (13 × 100 mm) containing 800 μl of TSB with the desired final concentrations of P+1, APST or NIF peptides. Cultures were grown shaking at 37°C until reaching an OD600 of 1.2. The supernatants were sterilized by filtration using 0.22 μm Spin-X® centrifuge tubes (Corning), and the filtered supernatants were kept on ice until use. The AH430 reporter strain was grow in TSB with Cam for 12 h, diluted 500-fold into fresh media and grown to a final OD600 of approximately 0.015–0.05. One hundred and eighty microlitres of the reporter culture was dispensed into a microtitre plate (Corning 3606 plate), and 20 μl of the filtered supernatants were added to each well. A minimum of four wells were used for each sample. As a control, TSB was added in place of the peptides. The microplate was incubated with shaking at 37°C on the PMS-1000 Shaker and growth and fluorescence were monitored using a Tecan GENios plate reader.

Quantitative assay for SpsB

The fluorescence quenched synthetic peptide, Y(NO2)–F–S–A–S–A–L–A–K–I–K(Abz), was obtained from California Peptide Research (Napa, CA). The fluorescence assay for signal peptidase activity was performed essentially as described (Zhong and Benkovic, 1998), except that the assay was performed in a 96-well microplate and monitored in a Tecan GENios plate reader. The concentrations of APST control peptide and the P+1 and NIF inhibitors were varied from 0 to 15 mM.

Acknowledgements

We thank Dr George O'Toole, Dr Ambrose Cheung, Dr Simon Foster, Dr Jerrold Weiss, Dr Simon Silver and Dr Doug Bartels for providing plasmids and strains. We also thank Dr Bartels for experimental advice on the AIP bioassay. This work was supported by a Cystic Fibrosis Foundation pilot grant and a Roy J. Carver Charitable Trust medical research initiative grant.

Ancillary