Technical Comments

Activation and Inhibition of the Staphylococcal AGR System

Science  21 Jan 2000:
Vol. 287, Issue 5452, pp. 391
DOI: 10.1126/science.287.5452.391a

In their report, Balaban et al. (1) proposed that the Staphlococcus aureus group I agractivator is a thermostable 38-kD protein (RAP) that is produced by an agr-null strain and is therefore neither encoded within the agr locus nor related to the AgrD autoinducing peptide. This protein was reported to be immunoprotective in a murine subcutaneous abscess model. Additionally, a linear heptapeptide, RIP, thought to be derived from RAP, was reported to inhibit agractivation in vitro and to interfere with an experimental staphylococcal infection in vivo (1). These results represent the first indication that bacterial extracellular factors other than the agr-encoded autoinducing peptides (AIPs) may be involved in the regulation of staphylococcal exoprotein synthesis and the control of staphylococcal virulence, and the first report that any linear peptide is active in the agr system. The idea that the agr activator is a protein was based on a preliminary purification of the activator by gel filtration chromatography of boiled and concentrated culture supernatants (2). In this experiment, the activity was stable to boiling (100°C for 10 min), was associated with a RAP that eluted anomalously, was in the same fraction as a gel filtration standard with a molecular weight of 1000, and was largely retained by a dialysis membrane. The same fraction would also have contained the AIP (see below), and Balaban et al. (1) noted the possibility that the activator was actually a small peptide and that RAP was not the activator but was rather an inactive coeluting protein. Because mutations affecting AgrB or AgrD totally eliminated the production of agr-activating substances (3), we later concluded that this was the case (3). We have also synthesized and tested several AIP-related linear peptides and found them to be totally inactive (3, 4). This result does not, of course, preclude the possibility of linear peptides that could act as agr inhibitors. We thus reexamined the possibility that an agr-activating protein, encoded outside of theagr locus, might have been overlooked in our previous studies.

Balaban et al. (1, 2) purified the RAP protein from concentrated and boiled S. aureus RN6390B (agr +) or RN6911 (agr-null) supernatants by gel filtration chromatography on a Bio-Rad HPLC SEC250 7 × 300 column, using a 1-ml sample for each run. Supernatants of both strains had activity, as demonstrated by Northern blot hybridization with an RNAIII probe, but, surprisingly, all of the activity was in a single 0.5-ml fraction containing only the anomalously eluting 38-kD protein, which was present only in this fraction (2). The activation of agr by RAP in culture supernatants of theagr-null mutant (1) meant that the activating factor was not agr-encoded.

It had previously been observed that a heavily mutagenized derivative of an S. aureus strain (Foggi—our RN833) produced a small peptide that strongly inhibited agr activation in a standard group I S. aureus strain (2). This peptide was found to have the sequence YSPXTNF (1), where X could be cysteine or tryptophan, and a synthetic version was prepared with the sequence YSPWTNF (1). This synthetic linear heptapeptide (Pep) was reported to have inhibitory activity against S. aureus, both in vitro and in vivo, although the in vivo dose used was probably about 1000-fold higher than that of the native staphylococcal peptide isolated from RN833 culture supernatants (1).

We obtained culture supernatants from three strains: RN6390B (standard group I agr +), RN6911 (agr-null [5]) and RN6734 (a derivative of RN6390B lysogenic for phage φ13). The first two were the same strains from which Balaban et al. (1, 2) isolated the RAP protein. The third was included to confirm the generality of RAP production. The supernatants were concentrated 10-fold by lyophilization, one 3-ml sample was dialyzed, and a second 3-ml sample was boiled, centrifuged, and fractionated on a Bio-Rad SEC250 gel filtration column (7.5 × 300 mm) 1 ml at a time. The fractions from the three separate column runs were combined in order to have larger amounts of protein for the subsequent analyses. The starting material, the dialysate, the retained material, and the gel filtration fractions were then analyzed for agr activating activity, using anagr-P3-blaZ fusion, and separated by PAGE (Fig. 1).

Figure 1

Activity and protein contents of gel filtration column fractions. (A) RN6911 (agr-null) was grown in CY broth for 7 hours starting at 3 × 108 cells/ml. The supernatant was concentrated 10× by lyophilization, boiled for 10 min, centrifuged (14 Krpm, 15 min, Eppendorf centrifuge), and the resulting supernatant separated by HPLC on a 7.5 × 300 mm Bio-Rad SEC250 column. Three 1-ml samples were run sequentially, and 1-ml fractions were collected and combined. Two-fold serial dilutions of each fraction were assayed for agractivation, using the agr-P3-blaZ fusion as a reporter, as described (3, top panel). The remainder of each fraction was analyzed by SDS-PAGE (13), stained with Coomassie Blue and scanned with a Molecular Dynamics scanner (bottom). (B) As in (A), except that the supernatant was from a culture of RN6734 (agr +). (C) As in (B), except that the column fractions were dialyzed (three cycles, 10 mM pH 6 phosphate buffer) and the retentates analyzed.

Figure 1

Purification of RAP and the octapeptide by gel filtration column chromatography. (A) Wild-type S. aureus postexponential supernatants were fractionated on a gel filtration Bio-Sil SEC-125 300 × 7.8 mm Bio-Rad column in 1 mM phosphate-buffered saline, pH 7.2 (0.1 × PBS), at a flow rate of 0.5 ml/min 1-ml fractions collected, concentrated to one-tenth of their original volume by lyophilization, and tested for activation of RNAIII synthesis by Northern blotting and the membrane autoradiographed (A). The number of eluted fraction is indicated (5 to 21). Positive control, material that was applied to the column. Negative control, 0.1 × PBS. (The amount of RNA applied to the gel was identical in all samples but in the positive control of <3 where one-tenth of the amount was applied.) Total, total supernatant containing RAP and the octapeptide. >8, supernatant that was ammonium sulfate precipitated and dialyzed against PBS in a 8-kD dialysis membrane and contains RAP. <3, Supernatant that was centrifuged through a 3-kD centriprep (Amicon) cutoff membrane and which contained the octapeptide. (B) Gel filtration standard (Bio-Rad) ranging from 670 kD (thyroglobulin, fraction 5) to 1.3 kD (vitamin B-12, fraction 10).

In neither experiments of this type (using different starting cultures) nor in experiments in which the supernatant was concentrated 50-fold, have we been able to detect any activity in fractionated or unfractionated supernatants of the agr-null strain (Figs. 1A and 2A). Nor have we detected any activity in similarly analyzed but unboiled supernatants. The agr + strains had the expected activity (Figs. 1B and 2A), which always eluted as a broad peak centering at an elution time of 10 min, 1 min before the cobalamin standard (MW 1350). A sample of the synthetic thiolactone-containing group I AIP (4) eluted from this column approximately one min later (not shown). All of the detectableagr-activating activity in supernatants of theagr + strains and in the gel filtration fractions passed through a dialysis membrane, although three cycles of dialysis were necessary (Fig. 2C). Each of the active fractions showed multiple protein bands, two of which were in the 30- to 46-kD range and paralleled the activating activity; either of these could correspond to the 38-kD band of Balaban et al. (1,2) (Fig. 1B). In no case was there a fraction containing only a single protein species. The two bands in the 30- to 45-kD range were retained by the dialysis membrane (Fig. 1C). There was no significant protein band in this region in supernatants of the agr-null strain, although other protein bands were clearly visible (Fig. 1A). This analysis has been repeated five times with similar results (not shown). On the chance that our reporter gene assay was an imperfect reflection of the agr activation process, we assayed culture supernatants by Northern blot hybridization. On the chance that RN6911 had undergone a spontaneous mutation, we included RN7206, a φ13 lysogen of RN6911. As shown in Fig. 2B, there was no detectable activity in concentrated culture supernatants of either of the agr-null strains. We conclude, therefore, that supernatants of the agr-null strain contain neitheragr activating factor nor detectable thermostable protein in the 30- to 45-kD range, and that the agrD peptide accounts for all of the agr-activating activity in theagr + strain.

Figure 2

Tests for agr activation. (A) Comparison of agr + and agr-null strains. Dilutions of concentrated boiled culture supernatants were assayed for agr activity, using theagr-P3-blaZ fusion as a reporter. (B) Analysis of agr activation by Northern blot hybridization. Tenfold concentrated boiled culture supernatants, 0.5-ml samples, were added to an early exponential phase culture of RN6734, with concentrated and unconcentrated CY broth as controls. Following 1 hour of incubation at 37°C, a 5-ml sample was analyzed by Northern blot hybridization (11), using an RNAIII-specific probe prepared by PCR in the presence of 32P-dATP (3). The blot was analyzed and quantified with a Molecular Dynamics PhosphorImager. Lanes 1, 3, 5, and 7, CY broth; lane 2, 10× concentrated CY broth; lanes 4, 6, and 8, RN6911, RN7206, and RN6734 supernatants, respectively. (C) Effect of dialysis. A 1-ml sample of the concentrated boiled supernatant used in the analysis shown in Fig. 1B was dialyzed against three changes of 10 mM pH 6 phosphate buffer, 100 ml each. Both dialysates and retentates were assayed for agractivation after each change of buffer. Activation activity in the starting material and in the retentates is represented by the hatched bars; total activity in the dialysates is indicated at right.

To explain the apparent discrepancy between our results and previous results that a significant portion of agr-activating material was retained by a dialysis membrane (2), we note that the dose-response curve with synthetic group I AIP is linear over only a ten-fold concentration range, between 3 and 30 nM (4), and that the AIP concentration in the culture supernatant of our standard agr + strain, RN6390B, is approximately 100 nM. As shown in Fig. 2C, during a single overnight dialysis, ∼20% of the activity was retained by the dialysis membrane, two more cycles of dialysis were required to remove the rest, and the activity was quantitatively recovered in the dialysate. Starting with a 10× concentrated sample (at about 1 mM), retention of 20% is equivalent to a retentate concentration of ∼200 nM—well above the saturation level for the assay of activity with undiluted material. Because the dialysate was not assayed in the original study (2), the conclusion—based on this retained activity—that the activity is largely nondialysable (2), was incorrect. An ultrafilter with a 3-kD cutoff has been successfully used to separate the peptide from higher molecular weight materials in the supernatants (3); Balaban et al. (1) reported that such a filter retains a significant portion of agr activating activity, consistent with a proteinaceous activating factor. We agree that a significant amount of activity is retained by such filters (not shown); however, in our experience, the concentrated crude culture supernatant rapidly clogs the filter, so that retention cannot be taken as evidence of material with molecular weight greater than that corresponding to the filter cutoff.

We examined the possibility that a linear heptapeptide inhibitsagr activation. Because all of the knownagr-encoded activating/inhibiting peptides contain a five-membered thiolactone ring (4, 6), the report of an active linear heptapeptide lacking the conserved cysteine (1) suggested that such a peptide might have an origin independent of agr and might represent a mechanism ofagr inhibition different from that of the known cross-reacting AIPs (4, 6). To obtain a supply of the native agr-inhibiting peptide produced by strain RN833, we prepared a concentrated postexponential phase supernatant from a culture of this strain as previously described, and confirmed its potent agr-inhibiting activity (Fig. 3). We also prepared a synthetic sample of the linear heptapeptide YSPWTNF by standard peptide synthesis, and then determined its sequence for confirmation. This material had no detectable agr-inhibiting activity in vitro (Fig. 3) or in vivo (not shown) when compared to the potent activity of the native material. Identical results have been obtained independently by Wright and Larrick (J. Larrick, personal communication). This result suggested that the native and synthetic RIP peptides may not be the same and that the native peptide might belong to the AgrD family after all. To test for this, we compared its sensitivity to boiling at pH 8 versus pH 2 with that of the native and synthetic AIPs ofagr groups I and II. All of the peptides tested were highly and equally sensitive to boiling at pH 8 but were stable to boiling at pH 2. The data for native RIP and synthetic AIP II are shown in Fig. 4. As determined from these data, the half-life of native RIP at pH 8 and 100°C was about 14 min, and that of AIPII about 23 min. This behavior is expected for a peptide with an essential thiolactone bond, but is not expected for any simple linear peptide. The Pep could not be tested in this manner, because it had no measureable activity.

Figure 3

Agr-inhibiting activities of native and synthetic RIP and of AIP II. Culture supernatants prepared as inFig. 1A, plus a sample of Pep, with the sequence YSPWTNF, were assayed for inhibition of agr activation using theagr-P3-blaZ fusion with β-lactamase activity as the readout (5).

Figure 4

Effect of boiling and pH on AIP activity. Samples of synthetic AIP II (RN6607) (4) and of an RN833 supernatant were adjusted to pH 2 or pH 8 and placed in a boiling water bath. After boiling for the required time, the material was returned to pH 5.5 (the usual pH of a culture at the end of growth) for assay with the agr-P3-blaZ fusion.

We next turned our attention to strain RN833, kindly provided by G. Omenn many years ago, which was described as a mutant defective in the production of staphylococcal nuclease obtained by extensive nitrosoguanidine mutagenesis of a standard strain of S. aureus known as the Foggi strain (7). RN833 has been in our strain collection for over 25 years, maintained at −80°C, and has been used as a recipient for DNA-mediated transformation because of its lack of nuclease activity. It is nonhemolytic on sheep blood agar and does not produce S. aureusvirulence factors such as α-hemolysin, protein A, and coagulase in detectable quantities. We have considered the possibility that it is an S. aureus mutant producing a mutationally altered AgrD peptide that inhibits its own agr activation. Its failure to produce protein A or coagulase make this unlikely because the protein A and coagulase genes are ordinarily down-regulated by agr. Nevertheless, we considered it important to confirm that RNAIII, theagr effector (5), was not produced. Accordingly, we tested for RNAIII production by Northern blot hybridization analysis of whole-cell RNA from RN833 by our standard method. The result, shown in Fig. 5B, lanes 1 and 2, was that RN833 does produce RNAIII, but that the RN833 material has considerably slower mobility than RNAIII produced by agr + S. aureus (Fig. 5B, lanes 3 and 4). Therefore, while it is possible that the phenotype of RN833 is the result of an agr mutation—perhaps one affecting RNAIII structure—it cannot be the result of agr autoinhibition by a mutant peptide, because this would block RNAIII production. We next considered the possibility that RN833 is not S. aureus after all, noting that S. warnerii has been reported to produce an RNAIII variant larger than that of most other staphylococcal strains studied, including several S. aureus strains, S. simulans, S. lugdunensis (8, 9) (which produces a smaller RNAIII), and S. epidermidis(10). Biotyping, kindly performed by the NYU/Tisch Hospital bacteriology laboratory, revealed that RN833 was not S. aureusbut was 85% likely to be S. warnerii, and that therefore our stock culture of the nuclease-negative mutant of strain Foggi is, and has always been, a non–S. aureus strain. Using conserved regions of the agr locus, we obtained and sequenced PCR products corresponding to most of the agrlocus of RN833. The sequence of the 5′ half of the RNAIII region from RN833 closely matched that of S. warnerii(8) (Fig. 5A), and the sequence of the agrDregion was typical of other agrD sequences, revealing the putative AIP sequence as YSPCTNFF (Fig. 5C). This is clearly the same peptide as that described by Balaban et al. (1), except that it contains a cysteine residue at the usual position rather than the arbitrarily inserted tryptophan, and, on the basis of its NH2-terminal amino acid and the position of the conserved COOH-terminal processing site, it would unquestionably be an octapeptide rather than a heptapeptide (Fig. 5C). Our conclusion is that RIP is the native thiolactone-containing AIP of S. warnerii, and that it fits very well into the general picture of AIPs developed in the past few years (see Fig. 6 for a diagram of the agrlocus as we currently understand it.).

Figure 5

RNAIII and AgrD of S. warneriiversus RN833. (A) Comparison of RNAIII sequences. Published RNAIII sequences (6, 8, 9) were aligned by a multiple sequence alignment program (Skirball Bioinformatics) to which was added the sequence of RN833 RNAIII. Only the 5′ half of the sequence alignment is shown. (B) Northern blot analysis. A preparation of total cell RNA was prepared and analyzed by the method of Kornblum, et al. (11), using as probe a32P-labeled PCR product specific for RNAIII (3). (C) Comparison of agrD sequences. PublishedAgrD sequences (6, 12) were aligned starting with the NH2-terminal methionine, with the addition of the newly determined sequences of S. aureus agr group IV and of RN833AgrD (bottom). The AIPs are shown in boldface and the sequence of Pep (1) shown below that of RN833AgrD. Groups I, II, III, and IV = S. aureus agrgroups I to IV; Se = S. epidermidis; SI = S. lugdunensis.

Figure 6

The agr locus and its regulation. (Top) Schematic diagram showing the two divergent agrpromoters (P2 and P3) and their transcripts. The P2 operon contains 4 genes, agrA, B, C andD. Two of these, agrC and agrArepresent a two-component signal transduction pathway and the other two, agrB and D, combine to generate a peptide that is the activation ligand for the signal receptor. The function of the signaling pathway is to activate the two agrpromoters so that there are two levels of positive feedback. The P3 transcript, RNA III, regulates target gene transcription by an unknown mechanism. At bottom right is a blowup of agrD, showing the sequence of the peptide that serves as activating ligand for the pathway by binding to agrC. (Bottom left) Blowup of agrC, showing five transmembrane helices, predicted by sequence analysis and confirmed by phoAfusions. Deletion analysis suggests that the last extracellular loop is the site of peptide binding, which causes the predicted phosphorylation of a conserved histidine in the COOH-terminal cytoplasmic domain.

We are unable to corroborate the finding of soluble agractivation or inhibition activity associated with any staphylococcal product other than the AIPs encoded by agrDand processed to yield the thiolactone derivatives. We have not attempted to determine the possible basis for mouse protection described for the protein(s) tracking on SDS-PAGE with molecular weights between 30 and 46 kD. It is possible that one of these proteins contains an epitope cross reactive with either the AIP or the receptor; alternatively, it is possible that one or another is a protective antigen on its own. It is certain, however, that none is an agractivator. The activity reported for the agr-null strain (1) could conceivably be explained by a mix-up in strains. The inhibitory activity reported for the linear synthetic heptapeptide has been explained by the presence of impurities in the preparation.


Activation and Inhibition of the Staphylococcal AGR System

Response: Novick et al. state that because they were not able to purify the ∼38-kD RAP that RAP does not exist, and they therefore conclude that RNAIII synthesis can only be activated by the octapeptide. Below, we demonstrate that Novicket al. applied purification methods different from ours, which perhaps explains their failure to purify RAP. We also show that the NH2-terminal sequence of RAP highly resembles that of RIP and demonstrate that RAP and RIP use a signal transduction different from the octapeptide to regulate the synthesis of RNAIII.

Novick et al. claim that the synthetic peptide RIP (YSPWTNF) synthesized by them was inactive and suggest that it was inactive because it does not contain a thiolactone structure. Their arguments are based on their conclusion that RIP is an octapeptide YSPCTNFFand that all octapeptides are in need of a thiolactone structure in order to be active. As we demonstrate below, RIP (YSPWTNF) is active. We further demonstrate that derivatives of RIP (designed according to the putative NH2-terminal sequence of RAP) are also active. None of the peptides were synthesized with a thiolactone structure. Both groups agree, however, that inhibition of RNAIII synthesis is of therapeutic potential (1, 2).

The different results between the groups may be explained by different handling of the product. Handling conditions are of primary importance. We do not know how the peptide was synthesized, solubilized, and stored. RIP, as we observed, can degrade (as can any peptide) and can become inactive under certain handling conditions.

Novick et al. also conclude (with 85% certainty) that the bacterial strain RN833 that produces RIP is S. warnerii. They back this by showing that the agr of S. warneriicontains a sequence YSPCTNFF and conclude that RIP is an octapeptide and is consequently made byS. warnerii.

Our results indicate that the bacteria producing RIP (RN833) may beS. xylosus (as identified by our collaborators with 99% certainty) and that a sequence identical to RIP (YSPWTNF) was identified in RN833 (TATTCGCCGTGGACCAATTTTTGA). Discussion of these findings follows.

RAP is sensitive to boiling (3). Therefore, if maximal activity of RAP is desirable, supernatants should not be boiled prior to RAP purification. Separation between the octapeptide and RAP by dialysis can be carried out, but only for a limited amount of time, and must be kept at 4°C. Novick et al. too show (figure 2C) that 20% activity could be retained after an overnight dialysis (at 4°C, we assume). Unfortunately, they could not use the cutoff membranes because their concentrated crude culture supernatant rapidly clogged the filter. If Novick et al. had used Centriprep instead of Centricon (Amicon), their membranes would not have clogged.

Use of β-lactamase activity (using theagr-P3-blaZ fusion as a reporter) as a test for RNAIII is not sensitive enough and can lead to experimental artifacts. We therefore tested for RNAIII exclusively by Northern blotting. β-Lactamase is naturally produced and secreted by the wild-type strain RN6390B [a common source of RAP in our lab (4)]. If the supernatant of this strain is collected in order to purify RAP from it, and each purification step is tested for its ability to activate RNAIII synthesis in the bacterial strain containing the P3:blaZ fusion by the β-lactamase assay, what one actually detects is β-lactamase activity in the supernatant of RN6390B rather than the consequent induction of P3 in the bacterial strain containing the P3:blaZ fusion. β-Lactamase is sensitive to boiling. If the supernatant is boiled prior to RAP purification, one can assume that most of the enzymes' activity in the supernatant is lost. In theory, one could then use the β-lactamase assay that would in fact reflect P3 activity. However, this assay is not sensitive enough for the detection of low levels of P3 activity induced by boiled or purified RAP. If RAP+octapeptide activity is maximal (100%) (RAP activity alone is only 20%), further boiling or purification leads to a further decrease in its activity. If the positive control is still the unpurified material, the activity of purified material becomes negligible and goes undetected by this assay. The inability to detect pure material by the β-lactamase assay is not only related to RAP but also to the octapeptide. It is not possible to detect induction of P3 by the octapeptide (HPLC-purified from wild-type supernatants) with the use of the β-lactamase assay.

RAP contributes to approximately 20% of the total activity in the supernatant. In the process of purification, some activity is lost. Therefore, to properly detect the active fraction eluting from the HPLC column, the positive control of the assay should be the partially purified material that was used to load the column that only contain RAP. If the positive control always remains the total supernatant (containing RAP and the octapeptide), and if the autoradiogram of the Northern blot is developed just to detect the positive control, the faint bands can go undetected.

To purify RAP, the HPLC column should not be overloaded. In figure 1, A, B, and C, of Novick et al., the gel filtration column was overloaded (as observed in the SDS-PAGE) and RNAIII was detected with the use of the β-lactamase assay instead of Northern blotting. In figure 1A of Novick et al., a wild-type autoinduction of RNAIII synthesis by postexponential supernatants contributed 80% by the octapeptide and 20% by RAP. In RN6911 (4) (which does not contain the agr), only RAP contributes to the induction of RNAIII synthesis, the overall activity is relatively low (1) and decreases with further purification steps. Therefore, β-lactamase assay should not be used for the detection of RAP in RN6911. If this assay is nevertheless carried out, the proper positive control (supernatant of RN6911, which was applied to the column) should be used. Fraction 9 (a possible elution time of RAP) does in fact contain activity. In figure 1B of Novick et al., the high results obtained using the β-lactamase assay may reflect purification of RAP (smaller quantities that probably go undetected) together with the octapeptide. If not boiled, the results might reflect purification of RAP, the octapeptide, as well as β-lactamase itself from the supernatant. In figure 1C of Novicket al., no activity is observed because the material applied to the column should not have been boiled or dialyzed for an extended period of time, particularly if no protease inhibitors were added (see comments on figure 2C of Novick et al.).

In figure 2A of Novick et al., RN6390 supernatant contains octapeptide + RAP if boiled. If not boiled, the supernatant also contains β-lactamase. Therefore, high results do not necessarily reflect P3 activity, but rather β-lactamase in the supernatant. On the other hand, supernatants of RN6911 do not contain the octapeptide but do contain RAP, but its quantity is too low to be properly detected by the β-lactamase assay. However, careful inspection of the graph shows that some activity can be detected in the supernatant (about 20% of activity observed in RN6390 supernatants). In figure 2B of Novicket al., the positive control, which was induced by the octapeptide + RAP (lane 8), gives a strong signal. The film should be exposed longer in order to detect possible lower activity in other lanes. In figure 2C of Novick et al., about 20% activity was retained by dialysis (which could not be extended beyond 1 day at 4°C without adding protease inhibitors). Loss of activity after second and third dialysis observed in figure 2C of Novick et al. might be due to the overall length of dialysis time in the absence of protease inhibitors. If activity was lost only due to dialysis, as Novick et al. suggest, after the first dialysis (of 1-ml sample against 100-ml buffer), less than 20% activity would have been retained. We cannot comment on Novicket al.'s observation that 100% activity was recovered in the dialysis buffer. To be tested appropriately, the buffer (100 ml of 10 mM phosphate) would have to be concentrated to the original volume of the sample (1 ml), thus increasing the phosphate concentration from 10 mM to 1 M. No positive control using 1 M phosphate is presented.

Figure 2

TRAP is specifically phosphorylated by RAP: (A) Wild-type early exponential S. aureus cells were incubated for 1 hour in phosphate- free buffer together with32P, with RAP, with RIP, with the octapeptide (<3), or with PBS as a control. Cells were collected, assayed for RNAIII synthesis by Northern blotting (presented as percent of maximum RNAIII observed in the specific experiment), or (B) applied to SDS- PAGE, and the gel autoradiographed.

What follows is a discussion of one of the protocols that we used for purification of analytical amounts of RAP. Wild-type S. aureus RN6390B (4) cells were grown to the postexponential phase of growth. Growth culture was centrifuged at 6000 × g for 10 min at 4°C. The supernatant was collected and filtered through a 0.22-μm filter to remove residual cells. The supernatant was lyophilized and resuspended in water to 1/10 of the original volume (total 10×). Fifteen milliliters of total 10× was applied to a 10-kD cutoff membrane [Centriprep 10 (Amicon)]. This enabled us to concentrate the material further and to remove material smaller than 10 kD. One milliliter concentrated material greater than 10 kD was washed twice in PBS by resuspending it each time in 15-ml PBS and reconcentrating it on the Centriprep 10, and the material greater than 10 kD collected (>10). Alternatively, postexponential supernatants were precipitated using 75% ammonium sulfate. The precipitate was resuspended in water to 66× and extensively dialyzed through an 8-kD dialysis membrane with PBS. Twenty to 30% of RNAIII upregulating activity was retained in the dialyzed high molecular weight fraction. One hundred microliter material greater than 10 kD was applied to an HPLC gel filtration column (Bio-Sil SEC-125 300 × 7.8 mm, Bio-Rad) in 1 mM PBS, pH 7.2 (0.1 × PBS), at a flow rate of 0.5 ml/min, and 1-ml fractions collected. Fractions were concentrated to 1/10 of their original volume by lyophilization and tested for activation of RNAIII synthesis as described (3). RAP can be further purified by anion exchange chromatography. Specifically, active gel filtration fraction (1 ml) was fractionated by anion exchange chromatography (HPLC SynchroPak Q300, Keystone Scientific, Inc.) in water, pH 7.2. Bound material was eluted by a salt gradient of 0 to 1 M NaCl in water. Pure ∼38 kD RAP eluted at 0.75 M NaCl. As demonstrated in Fig. 1, RAP and the octapeptide elute at close but different fractions.

The fraction that activated RNAIII synthesis was collected, separated by SDS- PAGE, and Coommassie-stained, and the protein band of approximately 38 kD (1) was amino acid–sequenced commercially by Edman degradation chemistry. The NH2-terminal sequence of RAP was determined to be IKKYKPITN (6). This sequence was compared to the S. aureus database, and the sequence of the open reading frame suggests that it is a possible 279–amino acid polypeptide that has a high (76%) sequence identity compared to the Bacillus subtilisribosomal protein L2 (5). We have not yet been able to express the protein in Escherichia coli or to inactivate the gene (possibly because of its essentiality and its high sequence similarity among bacterial species), and therefore still consider these results preliminary. However, we were struck by the similarity between the NH2- terminal sequence of RAP and RIP (YKPITN as compared to YSPWTN). Based on our hypothesis (3) that RAP and RIP may bind to the same receptor, one as an agonist and the other as an antagonist, RIP derivatives were synthesized [by Fmoc chemistry (15)] according to the putative NH2-terminal sequence of RAP. These peptides were tested for their ability to inhibit RNAIII synthesis in vitro and for their ability to prevent cellulitis in vivo. The results of these experiments (see below) indicate that the peptides most successful in inhibiting cellulitis were in fact the peptides that most resembled the NH2-terminal of RAP and contained the sequence YKPITN. Our results suggest that RAP and RIP may bind to the same receptor, one as an agonist the other as an antagonist.

The pathway by which RAP activates and RIP inhibits RNAIII synthesis was unknown. However, it seemed reasonable to hypothesize that, like other quorum sensing molecules, RAP would activate a classical bacterial two-component system by phosphorylation (7). Wild-type S. aureus cells were incubated with the octapeptide (<3) or with RAP which was purified from the postexponential supernatants of the same strain (8), with RIP (native or synthetic) or RIP derivatives, or with PBS as a control. Cells were assayed for RNAIII synthesis by Northern blotting (3) and for in vivo phosphorylation (9). Figure 2A shows that RAP and the octapeptide activate RNAIII synthesis while RIP inhibits it. As shown in Fig. 2B, RAP specifically phosphorylates a 21-kD protein while RIP inhibits its phosphorylation. We termed this protein TRAP (for target of RAP). The sequence of TRAP (9) revealed that it is a novel 167–amino acid polypeptide. Figure 2, A and B, also shows that, while RAP activates RNAIII synthesis and activates TRAP phosphorylation, the octapeptide activates RNAIII synthesis but inhibits TRAP phosphorylation. Furthermore, while RAP and RIP regulate TRAP phosphorylation in the agr null mutant RN6911, the octapeptide does not, suggesting that the octapeptide inhibits TRAP phosphorylation indirectly by activating the agr system (9). Our results suggest that RAP and the octapeptide regulate RNAIII synthesis by different signal transduction pathways, RAP (and RIP) by the TRAP system, and the octapeptide by the agrsystem (10).

In bacteria, many important processes are known to be mediated by extracellular signal molecules produced by the bacteria (11). A single bacterial cell can contain many signaling modules that operate in parallel (12) or in series (13), where convergent quorum sensing pathways lead to important cellular functions (14, 18–21). Production of virulence factors is essential for the survival of S. aureusin the host. It is therefore not surprising that more than one pathway would contribute to achieve the important goal of toxin production.

RIP peptides were synthesized by the Fmoc chemistry, as described below. The synthesis was performed with an automatic peptide synthesizer (PS3, Rainin-Protein Technologies, Emeryville, California), and all of the procedures and programming followed the manufacturer's instructions as previously described (15). The peptide was synthesized utilizing 9-fluoroenylmethoxycarbonyl (Fmoc) chemistry and high-capacity (0.7 to 1.1 mmol/g) Knorr resin, Fmoc-2,4-dimethoxy-4′-(carboxymethyloxy)-benzhydrylamine resin with 1% divinylbenzene cross linker (100 to 200 mesh, Advanced Chemtech, Louisville, Kentucky). The first amino acid was allowed to couple for 2 hours and the remainder amino acids (Advanced Chemtech) were coupled for 20 min at room temperature. The peptide was cleaved and deprotected by the addition of 90% trifluoroacetic acid (TFA), 5%, 1,2-ethanediol, and 5% water solution to the resin. The resin was incubated at room temperature for 14 hours and then washed several times with TFA. The peptide was extracted with cold ether. The peptide/TFA solution was reduced to a volume of 1.0 ml with nitrogen gas. After adding 25 ml of ether, the peptide solution was mixed and incubated on dry ice for 5 min. Samples were then centrifuged at 1000g for 5 min, the ether removed, and the extraction with ethyl acetate. Ether (1.5:1 v/v) on dry ice was repeated three times. Finally 1.0 ml of water and 25 ml of ether were added to the peptide followed by another incubation on dry ice and centrifugation. The top layer was removed, the ether evaporated with nitrogen gas, and the peptide resuspended in water and dialyzed. Following dialysis, the peptide was lyophilized and stored at room temperature in a dessicator under vacuum. Proper molar ratios of the amino acids in the peptide were confirmed by amino acid analysis. YSPWTNF was soluble in dimethyl sulfoxide (DMSO).

The peptide RIP (YSPWTNF) was also synthesized by the Fmoc chemistry using FMOC-amino acyl Wang resin followed by reverse phase chromatography (courtesy of M. Booth) at the Molecular Biology Research Facility, William K. Warren Medical Research Institute, University of Oklahoma Health Science Center, Oklahoma City, Oklahoma. The peptide that was synthesized by the Fmoc chemistry, was extensively dialyzed in water and evaporated, and resulted in a yellowish powder. The powder was resuspended in 30% acetic acid, extensively vortexed and sonicated, and was further diluted to a final concentration of 6% acetic acid and 0.083% TFA. Soluble material was applied to a C8 Dynamax 300A HPLC column (Rainin Inc., Woburn, Massachusetts) and eluted by the following gradient: 0 to 21% acetonitrile in 0.083% TFA for 6 min, 21 to 27% acetonitrile in 0.083% for 18 min, followed by 27 to 60% acetonitrile in 0.083% TFA. Peptide was eluted within 21 to 27% acetonitrile gradient. The resulting peptide was white and could be solubilized in water or in DMSO.

The RIP peptides (YSPWTNF) were tested successfully for inhibition ofS. aureus sepsis, septic arthritis, keratitis, osteomyelitis and mastitis (16). Our findings substantiate RIP as an effective suppressor of toxin production, that RIP is not strain specific in its inhibitory activity, and that RIP is an effective inhibitor of bacterial pathology at multiple body sites following diverse routes and doses of administration. These findings are strong evidence for the potential value of RIP as a chemotherapeutic agent.

The RIP peptide was tested following preparation by two methods described above, and peptide in each form was found to be effective in suppressing RNAIII synthesis in vitro and suppressed S. aureusin vivo. Similar positive results were obtained for those RIP samples prepared by the Fmoc system (solubilized in DMSO) and for those that were prepared in a similar fashion, then further purified by HPLC (solubilized in water). The material prepared by the Fmoc chemistry and solubilized in DMSO has been demonstrated to be more stable (that is, longer shelf life) than the more purified peptide that was water-soluble. The water-soluble form of RIP could have a greater direct penetration of biological fluids, while the DMSO preparations could release RIP to aqueous fluids in a sustained fashion. Comparisons of the two preparations are needed to determine if the drug's uptake from each formulation differs.

RIP derivatives were designed according to the putative NH2-terminal sequence of RAP (YKPITN, see above), synthesized by the Fmoc chemistry as described above and tested for their ability to inhibit RNAIII synthesis in vitro and to inhibitS. aureus cellulitis in vivo, as described (1,25). For the in vitro experiments, 1× culture supernatant of RN833 (native RIP) or 150 μg of synthetic peptides were incubated with 2.5 × 106 wild-type S. aureus cells (RN6390B) for 2 hours at 37°C and RNAIII was determined by Northern blotting, as described (3). As shown in Fig. 3, A and B, native RIP and RIP peptides inhibited RNAIII synthesis.

Figure 3

Inhibition of RNAIII synthesis by synthetic RIP and derivatives: Early exponential wild-type S. aureusRN6390B [2.5 × 106 colony- forming units (CFU)] were incubated for 2 hours at 37°C with native RIP, with synthetic RIP peptides and derivatives or with carrier buffer. Cells collected and assayed for RNAIII synthesis by Northern blotting. (A) Autoradiography. (B) The autoradiogram was scanned and the density of bands determined.

RIP peptides were tested for their ability to inhibitS. aureus cellulitis in mice. The number of bacteria injected per mouse (3.5 × 108) is beyond the range of protection of 1× native RIP and causes a high mortality rate (1). 3.5 × 108 S. aureus Smith diffuse (SD) were incubated with buffer, with 1× native RIP (8), or with 200 μg of synthetic RIP peptides for 30 min at room temperature, and the mixtures were injected (17). Mice were followed for mortality and for development of lesion. As shown in Table 1, some of the RIP derivatives protected animals 100% from mortality (B1, B3, Q, T), as well as from cellulitis (B1). Although one animal died (1/8) in an animal group treated with the peptide B6, the remaining animals (7/8) were 100% protected also from cellulitis. Our results indicate that in the cellulitis model, peptides containing the sequence YKPITN may even have a greater therapeutic potential than the original YSPWTNF RIP peptide. We do not yet know if the high therapeutic potential of certain RIP derivatives comes from better binding of the peptide to its receptor or from differences in chemical properties and stability of the peptides in vivo. We do know, however, that differences in stability do exist because, for example, peptides that are synthesized by the Fmoc chemistry are as active but have a longer shelf life than the ones further purified by HPLC. RIP derivatives were also tested successfully for their ability to inhibit TRAP phosphorylation (not shown), suggesting that RIP and its derivatives inhibit RNAIII synthesis in a similar way to the native RIP peptide (9). None of the peptides were synthesized with a thiolactone structure, suggesting that they may be distinct from the octapeptides that require the thiolactone structure for their activity.

Table 11

Suppression of S. aureus SD sepsis/cellulitis. 3.5 × 108 CFU S. aureusstrain SD were incubated with 200 μg synthetic peptides, 1× native RIP or carrier buffer for 30 min at room temperature and the mixture injected subcutaneously into Balb/C mice (17). Mortality and lesion size 7 days post challenge are indicated.

View this table:

Interestingly, the identity of the Staphylococcusstrain RN833 that produces RIP remains elusive. This strain was originally thought to be a mutated form of S. aureus(3), but has recently been identified as a coagulase negative strain. However, while one commercial Clinical Microbiology Laboratory determined it to be S. warnerii (with 85% certainty) another lab determined it to be S. xylosus(with 99% certainty) and yet another lab recently determined it to be S. hominis. The reason for these discrepancies could be that RN833 may have been mutagenized in 1968 (3) and is therefore no longer a classical strain. In an attempt to resolve the difference, we tested postexponential supernatants of various coagulase negative staphylococci for their ability to inhibit RNAIII synthesis ofS. aureus. Figure 4 shows that all supernatants tested inhibited RNAIII synthesis of S. aureus. The actual inhibitory molecules have not yet been purified and therefore their sequences are not known. We also tested other strains of S. xylosus and S. saprophiticus(4) and discovered that not all strains were able to inhibit RNAIII synthesis of S. aureus, suggesting strain variability (not shown). The use of the degenerate sequence of RIP (YSPWTNF), a sequence identical to RIP, was identified in RN833 (TATTCGCCGTGGACCAATTTTTGA), which was not within the agr locus. These results suggest that RIP may be YSPWTNF and not YSPCTNFF, as suggested by Novicket al. Another possibility is that RN833 produces YSPWTNF (encoded by the above locus) as well as YSPCTNFF, encoded by theagr locus. In conclusion, as shown by Balaban et al. (1) and confirmed by Mayville et al. (2), inhibition of RNAIII synthesis by inhibitory peptides can protect animals from infections caused by S. aureus.

Figure 4

Inhibition of S. aureus RNAIII synthesis by postexponential supernatants of coagulase negative bacteria: Early exponential wild-type S. aureus RN6390B (2.5 × 106 CFU) were incubated for 2 hours at 37°C with 2× postexponential supernatants of coagulase negative bacteria (S. warnerii, S. xylosus, and S. saprophiticus). Cells were collected and assayed for RNAIII.


Related Content

Navigate This Article