Bacterial Interference Caused by Autoinducing Peptide Variants

See allHide authors and affiliations

Science  27 Jun 1997:
Vol. 276, Issue 5321, pp. 2027-2030
DOI: 10.1126/science.276.5321.2027


The synthesis of virulence factors and other extracellular proteins by Staphylococcus aureus is globally controlled by theagr locus, which encodes a two-component signaling pathway whose activating ligand is an agr-encoded autoinducing peptide. The cognate peptides produced by some strains inhibit the expression of agr in other strains, and the amino acid sequences of peptide and receptor are markedly different between such strains, suggesting a hypervariability-generating mechanism. Cross-inhibition of gene expression represents a type of bacterial interference that could be correlated with the ability of one strain to exclude others from infection or colonization sites, or both.

“Bacterial interference” refers to the ability of one organism to interfere with the biological functioning of another. Although interference has been assumed to involve growth inhibition, this has been demonstrated in only a few instances (1, 2). We now describe a type of bacterial interference in staphylococci that does not involve growth inhibition, but rather is mediated by inhibition of the synthesis of virulence factors and other extracellular proteins. Expression of the genes encoding these proteins is coordinately controlled by theagr locus (3-5) (Fig.1), which consists of two divergent transcription units driven by promoters P2 and P3. The P3 transcript RNAIII, rather than any protein, is the effector of the agr response, which involves the up-regulation of genes encoding secreted proteins and down-regulation of genes encoding surface proteins (5,6). The P2 operon contains four genes—agrB,D, C, and A—all required for transcriptional activation of the two agr promoters (4). AgrC corresponds to the signal receptor and AgrA to the response regulator of a standard two-component signal transduction pathway (4). AgrB and D generate an autoinducing peptide that is secreted by the bacteria, can be isolated from culture supernatants, and is the activating ligand for AgrC (7). Addition of the autoinducing peptide to an early exponential phase culture of the producing strain causes the immediate activation of transcription from the two agr promoters (7).

Figure 1

The agr locus of S.aureus. Schematic map of the agr locus showing the major transcripts RNAII and RNAIII (arrows) and the genes indicated by boxes (3-7).

The existence of a form of bacterial interference involving this peptide was suggested by a test of naturally occurring staphylococcal strains for the production of signal molecules activating agr transcription in a standard strain, RN6390B. Surprisingly, although all of the strains showed autologousagr activation by their own culture supernatants, in many instances these supernatants inhibited rather than activatedagr expression by RN6390B (8). We then tested a set of seven strains for the effects of their culture supernatants on the same and on different strains, with respect to the agrresponse (Fig. 2, A and B). Because the time course ofagr activation varied from strain to strain (8), the tests used either early (EEP) or mid (MEP)-exponential phase cells. On the basis of the cross-activation or -inhibition revealed by these tests, Staphylococcus aureus strains could be divided into three groups, a group being defined as a set of strains showing mutual cross-activation of the agr response (Fig. 2C). In all instances, members of one group inhibited agr expression by members of the other two. Non–S. aureus strains generally inhibited the agr response of S. aureus strains from each of the three groups (8).

Figure 2

(A to C) Effects of conditioned media on the transcription of RNAIII in different S. aureusstrains. Conditioned media from cultures of various S. aureus strains (9) were prepared (7). (A) RN6390B RNAIII transcription was measured as described (Table 1). (B) Conditioned medium (10%) (or CYGP medium as control) was added to each culture of RN7843. After 30-min (for EEP) or 60-min (for MEP) incubation at 37°C, whole cell lysates were prepared and used for Northern blot (RNA) hybridization with a 32P-labeled RN6390B RNAIII-specific DNA probe (7). The blots were exposed to x-ray film (Kodak). For strains RN6596, RN7690, SA502A, RN8462, and RN8463, similar experiments were performed (8). (C) Summary of the data in (A) and (B). (D) Effects of purified peptides on the RNAIII transcription in different strains. The P3 promoter regions of SA502A and RN8463 were sequenced (12). The DNA sequences of these two promoters are identical to that of RN6390B (4), so the RN6390Bagr P3-blaZ construct was used. The peptides of RN6390B, SA502A, and RN8463 were purified from cells containing the cloned agrBD genes (9) as described (19). RNAIII transcription was measured as described (Table 1) with 20 mM tris-HCl (pH 7.5) as control.

These results led us to clone and sequence the agrBDCregions from three group II and two group III strains plus a second group I strain, using primers flanking agrBDC(9). A compilation of the predicted AgrBDC sequences from one strain of each S. aureus group plus that fromS. lugdunensis (10) is shown in Fig.3. The corresponding sequences for the two or three strains within each group were identical (11), whereas those for strains from different groups were highly divergent between amino acid positions 34 (AgrB) and 205 (AgrC).

Figure 3

Comparisons of the predicted (A) AgrB, (B) AgrD, and (C) AgrC amino acid sequences from different S. aureus groups and S. lugdunensis. The sequences were aligned by MegAlign software (GeneStar). The sequences are as follows: S. aureus group I (RN6390B and RN7690) (4), II (SA502A, RN6923, and RN6925) (GeneBank accession number AF001782), and III (RN8462 and RN8463) (GeneBank accession number AF001783); and S.l (S. lugdunensis) (10). Within the same group, the AgrBDC sequences are identical. Residues identical in all four or in all three groups of S. aureus are shaded. Dashes indicate gaps generated by the align program. In (B), the sequences of the peptides are boxed and the conserved aspartate-glutamate are in bold. Asterisks indicate stop codon. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

To determine whether an agrD-encoded peptide was responsible for both the activation and inhibition activities exhibited by culture supernatants, we used a group II and a group IIIagrBDC clone plus the original RN6390B (group I)agr clone (7) and a polymerase chain reaction (PCR) product containing S. lugdunensis agrBD genes to prepare agrBD clones that were introduced into anagr-null host strain, RN6911 (5, 9). Supernatants from each of these clones showed the same activity as that of the supernatant from the corresponding parental strain (8), confirming that the agrBD complex was responsible for the agr-inhibiting as well as theagr-activating activity of these strains.

We then purified and sequenced the active material from the supernatants of RN6911 derivatives containing these agrBDclones and isolated from each a single active peptide, which showed the same autologous activation and heterologous inhibition as the crude supernatant from which it was purified (Fig. 2D). The four peptides varied in length from seven to nine amino acyl residues and had highly divergent sequences with the exception of a conserved cysteine five residues from the COOH-terminus (Fig. 3B). Sequences of the corresponding agrD loci (Fig. 3B) confirmed that each of the active peptides, like that of RN6390B (7), was processed from within the (predicted) AgrD peptide. Synthetic peptides corresponding to the sequences of the RN6390B and S. lugdunensis autoinducers, however, had no detectable activity. Mass spectroscopy showed that the synthetic peptides were dimeric, whereas the native peptide molecules were monomeric and had molecular masses that were 18 ± 1 atomic mass units less than those predicted by their respective amino acid sequences (7,8).

Taken together, these results suggested that the cysteines in the synthetic peptides had spontaneously formed intermolecular disulfides, whereas those in the native peptides were involved in an intramolecular bond, most likely a cyclic thioester introduced posttranslationally and including the COOH-terminal carboxyl, because there is no other conserved carboxyl group in the molecule. Consistent with this possibility were the results of treatment of the native peptides with iodoacetic acid and hydroxylamine. Iodoacetic acid, expected to react with free –SH groups, had no effect, whereas hydroxylamine, expected to react with thioesters, abolished activity (12). We recently synthesized a small quantity of a cyclic thioester derivative of the RN8463 (group III) octapeptide and showed that the synthetic material, still impure, inhibits agr expression by RN63909B (group I) (13). We have not demonstratedagr activation with this material and are currently preparing a larger batch to enable better purification.

The cyclic thioester bond is probably introduced during processing, and we suspect that AgrB is responsible for this step and possibly for secretion as well. The predicted AgrB sequences (Fig. 3A) suggest that AgrB is a transmembrane protein, and we have confirmed this by means of PhoA fusions (12). Except for the first 34 amino acyl residues, AgrB is divergent, and we have shown by analysis of separateagrB and agrD subclones that AgrB determines the specificity of AgrD processing. An agr-null strain containing the cloned group I agrB and group IIIagrD produced group III autoinducer activity, and vice versa (Table 1); no other heterologous combinations were active, although each of the parental agrBD pairs generated the expected activity. We conclude that despite the sequence divergence of agrB and D, the two gene products have retained the specific interactions that are required for maturation of the peptide derivatives of AgrD. Genetic and in vitro biochemical analyses to determine the precise role of AgrB and the different roles of the conserved and divergent regions of the protein are in progress. A single transmembrane protein is responsible for processing and secretion of peptidic bacteriocins by lactococci (14); however, this protein is unrelated to AgrB and is much larger.

Table 1

Complementation analysis of agrB andagrD genes on the production of peptide activity among three groups of S. aureus and S. lugdunensis.Staphylococcus aureus RN6390B (pRN6683, containing RN6390Bagr P3-blaZ fusion) (4) cells were grown in CYGP medium at 37°C to EEP or MEP. To each culture, 10% conditioned medium prepared (7) from S.aureus cells containing cloned agrB, oragrD or various combinations from representative strains of three groups of S. aureus (I, RN6390B; II, SA502A; and III, RN8463) and S. lugdunensis (9), or CYGP medium as control, was added. The cultures were incubated at 37°C with shaking. After 55-min (for EEP) or 80-min (for MEP) incubation, β-lactamase activity was measured as described (7). +, activation; +*, inhibition; −, no effect.

View this table:

Similarly, AgrC has diverged in parallel with AgrD so as to retain the specificity of the receptor-ligand interaction that leads to activation. Here, the divergent region is the NH2-terminal half, which was predicted by sequence analysis and confirmed by PhoA fusions (15) to span the cytoplasmic membrane and which must contain the site of interaction with the different peptides. The COOH-terminal half, which contains the conserved histidine and transmits the autophosphorylation signal (16), is highly conserved.

The sequence divergence is confined to the region of the agrlocus that is responsible for the specificity of processing and of the ligand-receptor interactions, and the junctions between the conserved and divergent regions are very sharp, at the nucleotide as well as at the amino acid level. This type of sequence organization is similar to that of the exchangeable cassettes that are responsible for antigenic switching and other types of phase variation in microorganisms (17). It would not be surprising if some sort of cassette-switching mechanism were responsible for exchanging these divergent gene segments between strains or species, or both. The evolutionary mechanism for divergence in concert, however, remains unknown and may involve a hypervariability-generating system.

Studies are presently directed toward the biological role or roles of these groupings. Group-specific differences in the expression of subsets of virulence factors or other extracellular proteins could be related to differences in disease patterns. Group-specific differences in the expression of colonization factors could be related to interstrain interference with colonization (1) and to differences in colonization site preference. Indeed, we have observed that the vast majority of menstrual toxic shock strains belong toagr group III and are characterized by a coherent overall biotype (12). This finding could reflect a tissue tropism for the human vaginal mucosa; however, it is inconsistent with the fact that many surface proteins, presumably including colonization factors, are down-regulated by agr (18). Finally, ligand-based inhibition of virulence or colonization factor expression, or both, could represent the basis of a therapeutic or prophylactic initiative that may not be limited to staphylococci.


View Abstract

Navigate This Article