We identified a genetic context encoding a transcriptional regulator of the Rgg family and a small hydrophobic peptide (SHP) in nearly all streptococci and suggested that it may be involved in a new quorum-sensing mechanism, with SHP playing the role of a pheromone. Here, we provide further support for this hypothesis by constructing a phylogenetic tree of the Rgg and Rgg-like proteins from Gram-positive bacteria and by studying the shp/rgg1358 locus of Streptococcus thermophilus LMD-9. We identified the shp1358 gene as a target of Rgg1358, and used it to confirm the existence of the steps of a quorum-sensing mechanism including secretion, maturation and reimportation of the pheromone into the cell. We used surface plasmon resonance to demonstrate interaction between the pheromone and the regulatory protein and performed electrophoretic mobility shift assays to assess binding of the transcriptional regulator to the promoter regions of its target genes. The active form of the pheromone was identified by mass spectrometry. Our findings demonstrate that the shp/rgg1358 locus encodes two components of a novel quorum-sensing mechanism involving a transcriptional regulator of the Rgg family and a SHP pheromone that is detected and reimported into the cell by the Ami oligopeptide transporter.
Quorum-sensing (QS) is a cell–cell communication mechanism in bacteria that controls gene expression via secreted signalling molecules, also called autoinducers or pheromones. Despite differences in the chemical nature of the signalling molecules between Gram-positive and Gram-negative bacteria (Bassler, 2002; Waters and Bassler, 2005; Antunes et al., 2010), QS is triggered by a similar circuit in the two groups. First, the signalling molecule is released into the extracellular environment by active or passive transport; it accumulates in the extracellular environment and once at a threshold concentration is detected by a sensor protein. This sensing leads cells to modulate gene expression in a co-ordinated manner in the bacterial population.
There are two general activation pathways for Gram-positive QS (Williams et al., 2007). The signal molecule, which is a peptide, can be sensed outside the cell by the histidine kinase of a two component system. Detection of the peptide leads to phosphorylation or dephosphorylation of the histidine kinase and then of a specific cytoplasmic transcriptional regulator. This phosphorylation state modifies the ability of the response regulator to bind DNA, modulating its ability to activate or repress the transcription of its target gene(s). This mechanism has been described in detail for many systems, including triggering of competence for natural transformation in Streptococcus pneumoniae and Bacillus subtilis (for reviews see Claverys and Håvarstein, 2002; Claverys et al., 2006) and control of the agr locus in Staphylococcus aureus inducing the regulation of accessory virulence genes (Novick and Geisinger, 2008). The second activation pathway involves sensing the signalling molecule inside the cell after its internalization by an oligopeptide permease transport system called Opp or Ami, member of the ubiquitous ATP-binding cassette superfamily (ABC transporters) (Linton and Higgins, 2007). Once internalized, the pheromone interacts with a transcriptional regulator or a Rap protein, both belonging to the RNPP family (for Rap, NprR, PlcR and PrgX), thereby modifying their activity and consequently the expression of their target gene(s) (Declerck et al., 2007; Rocha-Estrada et al., 2010). Three groups of relevant peptides have been described in detail: (i) Rap-associated Phr peptides in B. subtilis involved in the control of sporulation, competence, and production of degradative enzymes and antibiotics (Pottathil and Lazazzera, 2003), (ii) peptides involved in the control of plasmid transfer in Enterococcus faecalis that interact with PrgX or PrgX-like proteins (Dunny, 2007) and (iii) PlcR-associated PapR peptides involved in triggering virulence of cereus group bacteria (Slamti and Lereclus, 2002; Bouillaut et al., 2008).
We previously identified, by in silico analyses, a subcluster of Rgg regulators defined by the association of pairs of genes transcribed divergently. These pairs of genes, widespread only in the Streptococcus genus, combine a gene encoding a transcriptional regulator of the Rgg family with another gene encoding a small hydrophobic peptide called shp. One or two copies of such genetic contexts have been found in nearly all streptococci genomes, including S. pneumoniae, S. pyogenes and S. agalactiae, and up to seven copies have been found in S. thermophilus (Ibrahim et al., 2007a). We have studied one of these shp/rgg loci, named shp/rgg1358, in S. thermophilus LMD-9 and identified one target gene of the encoded Rgg regulator. This target gene, called ster_1357, encodes a secreted cyclic peptide and its expression is positively controlled by Rgg1358 and indirectly by SHP1358 and the Ami oligopeptide transporter (Ibrahim et al., 2007b). These findings led us to investigate whether the Rgg protein and a small hydrophobic peptide (SHP) reimported pheromone are components of a novel QS regulatory system. Recently, another group of signalling peptides, associated with Rgg-like proteins named ComR, has been identified. These peptides are named ComS or XIP and are involved in triggering competence for transformation in S. thermophilus, S. salivarius and S. mutans and probably in other streptococci belonging to the pyogenic and bovis group (Gardan et al., 2009; Fontaine et al., 2010; Mashburn-Warren et al., 2010).
In this study, we started by constructing a phylogenetic tree of the Rgg and Rgg-like proteins found in Gram-positive bacteria to compare the respective distributions of SHP-associated Rgg with the XIP-associated ComR regulators. Then, we tested the function of the shp/rgg subcluster as a QS mechanism by studying the shp/rgg1358 locus of S. thermophilus LMD-9. We identified another Rgg1358 target gene (shp1358) and used it to confirm the existence of the first steps of a QS mechanism: (i) secretion of the putative SHP pheromone, (ii) maturation of the pheromone, (iii) detection of the pheromone at a cell density threshold and (iv) reimportation of the pheromone into the cell by the Ami oligopeptide transporter. Surface plasmon resonance (SPR) analyses demonstrated that the pheromone interacts with the transcriptional regulator Rgg1358. Electrophoretic mobility shift assays (EMSA) revealed that the Rgg1358 binds to the promoter regions of target genes. Thus, this comprehensive survey of SHP-associated regulators combined with the study of the shp/rgg1358 locus of S. thermophilus LMD-9 provides evidence of a novel QS mechanism: it involves a Rgg transcriptional regulator associated with a SHP pheromone detected and reimported into the cell by the Ami oligopeptide transporter.
SHP-associated Rgg are phylogenetically distinct from the XIP-associated ComR regulators
Using genome sequences for 19 streptococci, we previously identified the SHP-associated Rgg as a subfamily of Rgg regulators (Ibrahim et al., 2007a). Since then, numerous genomes have been deposited in the databases. Moreover, another subfamily of Rgg-like regulators, ComR, also working with a peptide, ComS or XIP, imported by an oligopeptide transporter has been discovered (Fontaine et al., 2010; Mashburn-Warren et al., 2010). We therefore updated our list of SHP-associated Rgg and investigated their place in the repertoire of the Rgg and Rgg-like proteins. In particular we compared them with the XIP-associated ComR proteins.
Rgg proteins are only found in the order of Lactobacillales and the family Listeriaceae for which 90 complete genome sequences are available. We identified 484 sequences corresponding to Rgg proteins sensu largo, including all XIP-associated ComR proteins previously identified (Fontaine et al., 2010; Mashburn-Warren et al., 2010). Rgg regulators were found in all the genomes considered except the two Lactobacillus salivarius isolates. To map SHP and XIP peptides, the GenBank CDS annotation was complemented by predicting short CDSs with our software SHOW as described by Ibrahim et al. (2007a). Putative CDSs with length between 10 and 50 aa and adjacent to the 484 rgg genes were retrieved and screened manually to identify exhaustively all SHP and XIP peptides. The sequence of the 12 SHP (Ibrahim et al., 2007a) and 10 XIP (Mashburn-Warren et al., 2010) identified in complete genome sequences and already published were used to define the selection criteria as described in the Experimental procedures section. We recovered 61 SHP and 27 XIP. As a result of some redundancy between strains, only 22 and 12 unique amino acid sequences of SHP and XIP, respectively, were identified; they included 10 new SHP sequences and two new XIP.
The 484 sequences were further used to construct a phylogenetic tree in which SHP-associated Rgg and XIP-associated ComR were highlighted (Fig. 1). The complete list of Rgg and Rgg-like proteins and the sequences and positions of the SHP and XIP peptides is provided in Table S1. This analysis confirmed that both SHP-associated Rgg and XIP-associated ComR are specific to the streptococci family. Although they share similarities, they form two different branches of the phylogenetic tree indicating that they correspond to two distinct subfamilies of Rgg regulators. Furthermore, the tree distinguished between the two groups of SHP-associated Rgg previously identified, i.e. groups I and II, which again correlated well with the presence of a conserved aspartate or glutamate respectively (Ibrahim et al., 2007a). We found no exception in group I and only one SHP (in six strains of Streptococcus suis) with a conserved aspartate associated with a Rgg belonging to group II. In this new phylogenetic tree, the two SHP previously described as unclassified because of the absence of conserved aspartate or glutamate residue clustered with group I Rgg proteins. Finally, detailed examination of the Rgg proteins in close vicinity to the SHP-associated Rgg revealed another group (group III) of seven peptides with amino acid sequences that fulfil the SHP definition but that are encoded by genes overlapping the end of the rgg genes in a convergent orientation. One is in S. thermophilus strain LMD-9 and the others are from six (of nine studied) strains of S. pneumoniae.
To confirm the function of the shp/rgg subcluster as a component of a QS mechanism, we investigated the shp/rgg1358 locus of S. thermophilus LMD-9.
The expression of the shp1358 gene is controlled by Rgg1358, SHP1358 and, AmiCDEF in S. thermophilus LMD-9
The shp/rgg1358 locus of S. thermophilus LMD-9 is composed of shp1358 and rgg1358, two genes divergently transcribed with overlapping putative promoters. We studied the expression of the shp1358 gene suspected to encode a pheromone (Ibrahim et al., 2007b). The production of most such pheromones is auto-regulated, so we constructed a Pshp1358-luxAB transcriptional fusion and used it to transform the LMD-9 strain and several mutants deleted for the rgg1358, shp1358 or ami genes. The luciferase activity of the Pshp1358-luxAB fusion was 80 RLU/OD600 in the LMD-9 strain but only 6 and 3 RLU/OD600 in the Δrgg1358 and Δshp1358 mutants respectively; no activity was detected in the amiCDE::spec mutant (Fig. 2A). Therefore, the shp1358 gene is a target of the Rgg1358 regulator, and the shp1358 gene product is implicated in its own transcription. The Ami transporter is also involved in the positive control of the transcription of the shp1358 gene, most probably by an indirect mechanism. The Pshp1358-luxAB fusion was then used to investigate all steps of the QS mechanism involving SHP1358 and Rgg1358.
The small hydrophobic peptide SHP1358 is the precursor of a secreted pheromone of nine amino acids
In our QS model, SHP1358 is proposed to be the precursor of the pheromone. We tested whether, like other pheromones in these systems, SHP1358 is secreted. Supernatants from cultures of strains producing or not producing the SHP1358 peptide (strain LMD-9 and its Δshp1358 mutant, respectively) were mixed with pellets of cells of a strain containing a Pshp1358-luxAB fusion and deleted for the shp1358 gene (TIL 1200, named the ‘reporter strain’ hereafter). The expression of the shp1358 promoter was then tested (Fig. 2B). The expression of the shp1358 promoter in the reporter strain was not activated by the supernatant from the Δshp1358 mutant, whereas the supernatant from strain LMD-9 induced its activity (55 RLU/OD600). This demonstrates functional complementation of the cells of the reporter strain by the supernatant of the LMD-9 strain possibly through secretion of the product of the shp1358 gene.
To determine the amino acid sequence of the active form of the secreted pheromone, the supernatant from strain LMD-9 was analysed by mass spectrometry (LC-MS/MS) and compared with that from strain Δshp1358 (data not shown). Only one mass, not detected in the supernatant of strain Δshp1358, was identical to the mass of a fragment of SHP1358. This mass corresponds to a highly hydrophobic nonapeptide with the amino acid sequence EGIIVIVVG. This form is the putative product of a C-terminal cleavage of the 23 amino acid peptide precursor. This sequence was validated by fragmentation followed by an analysis on the LTQ orbitrap (Fig. 3).
A synthetic peptide corresponding to this nine amino acid sequence was produced and added at different concentrations to cultures of the reporter strain. Functional complementation by the synthetic peptide was observed and was correlated with its concentration in the medium (Fig. 2C). These results demonstrate and identify one active secreted peptide, hereafter called SHP1358(15–23), derived from the SHP1358 precursor.
A critical biomass is required to activate the expression of the shp1358 gene
Quorum-sensing mechanisms involve signalling molecules reaching a threshold concentration to trigger the expression of their target gene(s) (Winzer et al., 2002; Podbielski and Kreikemeyer, 2004; Siehnel et al., 2010), and signalling molecule concentration correlates with cell density. To check whether this applies to the SHP/Rgg1358 system, the expression of the shp1358 promoter was measured in cultures of strain LMD-9 containing the Pshp1358-luxAB fusion inoculated at different concentrations (Fig. 2D). In all cultures, irrespective of initial cell density, a similar maximum of luciferase activity was observed (approximately 80 RLU/OD600). However, the higher the initial cell density, the earlier the expression of the luciferase activity was triggered and its maximum reached.
To check that this effect was the result of the presence of the SHP1358(15–23) pheromone at a sufficient concentration, and not of the transition from exponential growth to stationary phase, we inoculated media with strain LMD-9 containing the Pshp1358-luxAB fusion at a low cell density and added or did not add synthetic SHP1358(15–23) at the beginning of the exponential phase. The addition of SHP1358(15–23) was sufficient to trigger the expression of the shp1358 gene, independent of the growth phase of the culture (Fig. S1). These findings are consistent with a QS mechanism.
The lipoprotein AmiA3 detects the SHP1358(15–23) pheromone before its reimportation by the oligopeptide transporter, AmiCDEF
Oligopeptide permease transporters can be divided in two functional entities: a detection entity composed of oligopeptide binding proteins allowing the capture of the target substrate (AmiA1 and AmiA3 in the strain LMD-9) and a transport entity allowing the import of the substrate into the cell (AmiCDEF in the strain LMD-9). The expression of the shp1358 gene is abolished in an amiCDE::spec mutant and also in a ΔamiA1 amiA3::erm mutant (data not shown). Therefore, it is plausible that the Ami transporter is involved in the reimportation of the SHP1358(15–23) pheromone into the cell. To confirm this, we introduced the Pshp1358-luxAB transcriptional fusion into various Δami mutants, also deleted for the shp1358 gene: ΔamiA1 amiA3::erm Δshp1358 blp::Pshp1358-luxAB to confirm the role of the Ami transporter in the reimportation of the pheromone, ΔamiA1 Δshp1358 blp::Pshp1358-luxAB and amiA3::erm Δshp1358 blp::Pshp1358-luxAB to study, respectively, the activity of the AmiA3 and AmiA1 oligopeptide binding proteins. The reporter strain, TIL 1200, was also used as a positive control of luciferase activity. These four strains were then cocultivated with strain LMD-9, providing SHP1358(15–23) to the medium, and luciferase activity was followed (Fig. 2E). The luciferase activity of the positive coculture control increased to 28 RLU/OD600 whereas there was no detectable activity in the LMD-9/ΔamiA1 amiA3::erm Δshp1358 blp::Pshp1358-luxAB coculture. Therefore, the SHP1358(15–23) pheromone provided by the LMD-9 strain was not sensed or imported by the cells lacking the oligopeptide binding proteins of the Ami transport system, but was by the ami wild type. Two different levels of luciferase activity were obtained for the cocultures of LMD-9 with single oligopeptide binding protein mutants. The luciferase activity of the LMD-9/ΔamiA1 Δshp1358 blp::Pshp1358-luxAB coculture was similar to that of the LMD-9/reporter strain coculture. The activity of the LMD-9/amiA3::erm Δshp1358 blp::Pshp1358-luxAB coculture was half that of the positive control coculture. Therefore, detection of the SHP1358(15–23) pheromone mainly requires the AmiA3 oligopeptide binding protein.
The membrane peptidase Eep is involved in the SHP1358 maturation process
The Eep protease plays a key role in the maturation of various pheromones and anti-pheromones of E. faecalis (An et al., 1999) and S. gordonii (Vickerman et al., 2010). The presence of an eep-like gene in the S. thermophilus LMD-9 genome and some similarities between the anti-pheromone of E. faecalis and the SHP1358 pheromone (sequences of approximately 23 amino acids, presence of lysine in the N-terminal part and high hydrophobicity) led us to investigate the involvement of this protease in the SHP1358 maturation process. Eep proteases also contribute to the cleavage of the peptide signal of some lipoproteins (An and Clewell, 2002; Denham et al., 2008). As the oligopeptide binding proteins of the Ami transporter are lipoproteins, we hypothesized that they may be matured by the Eep protease and therefore tested their functionality in a Δeep mutant. The toxicity of the aminopterin peptide for the Δeep strain was compared with that for the wild-type strain and the ΔamiCDE strain. The toxicity of the aminopterin peptide for the Δeep strain was intermediate between that for strain LMD-9 (sensitive) and the ami mutant (resistant) (data not shown), suggesting that the functionality of the Ami transporter is probably partially impaired in the Δeep mutant. Consequently, a ΔamiCDE mutant, rather than the wild-type strain, was used as a positive control for the following experiment. A replicative plasmid containing the shp1358 gene under the control of a constitutive promoter was introduced into the Δeep and ΔamiCDE strains, leading to strains TIL 1206 and TIL 1213 respectively. The effects of culture supernatants from both these strains were then tested on reporter strain (TIL 1200) cell pellets (Fig. 2F). The luciferase activity of the reporter strain was tenfold lower with supernatant from TIL 1206 than with supernatant from TIL 1213, indicating that the extracellular concentration of the pheromone is lower in TIL 1206 cultures than in TIL 1213 cultures. The endopeptidase Eep is therefore involved in the SHP1358 maturation process.
The SHP1358(15–23) pheromone interacts with the Rgg1358 regulatory protein
We used SPR analyses to investigate whether the SHP1358(15–23) pheromone binds the regulatory protein Rgg1358. The interaction between chip-immobilized SHP1358(15–23) (898 Da) and purified His-tagged Rgg1358 protein (at 5 µM) gave a high SPR signal response corresponding to approximately 3700 RU (Fig. 4, signal 1), under the flow condition tested. Moreover, very little dissociation was observed, suggesting a high affinity between these two partners. In contrast, when purified Rgg1358 protein was injected over another immobilized small hydrophobic peptide (IAILPYFAGCL, ComS of S. thermophilus LMD-9) in similar conditions, no signal response was obtained (Fig. 4, signal 2). The specificity of the binding was confirmed by replacing the regulatory protein Rgg1358 by BSA (Fig. 4, signal 3). These SPR analyses demonstrate that SHP1358(15–23) and Rgg1358 specifically interact with each other.
Rgg1358 binds the promoter regions of shp1358 and ster_1357 genes
Electrophoretic mobility shift assays were carried out to test for interactions between the Rgg1358 regulatory protein and DNA fragments of promoter regions of the two known target genes, shp1358 and ster_1357. To design relevant DNA probes, we mapped the transcriptional start site of both genes and identified putative −10 promoter sequences (Fig. 5A). A conserved motif – GCAWATATGGGAATA – of 15 nt was found 25 nt upstream from the −10 motif of the shp1358 and ster_1357 genes. This motif differs from that described for PlcR, PrgX and ComR regulators (Fig. 5B). DNA probes centred on the conserved motif were amplified by PCR. Nine pg of each labelled probe was mixed with purified recombinant His-tagged Rgg1358 protein and analysed by EMSA. The recombinant Rgg1358 protein bound specifically to the shp1358 and ster_1357 probes and the intensity of the shift increased with increasing concentration of the Rgg1358 (data not shown). In parallel, the binding to these two probes of the Rgg1358 protein mixed with the SHP1358(15–23) pheromone was also tested. In this case, to prevent peptide-dependent binding being missed because of a possible excess of Rgg1358 protein, an intermediate Rgg1358 protein concentration (at 70 nM) was used; this concentration did not allow a complete shift of the shp1358 and ster_1357 probes (Fig. 6A, lanes 1 and 2). The experiments were performed as described above in the presence of 27.8 nM, 278 nM and 2.78 µM, of the synthetic SHP1358(15–23) pheromone (corresponding to 25, 250 and 2500 ng ml−1, respectively). DNA shifts were observed for the shp1358 and ster_1357 probes (Fig. 6A, lanes 3 to 5) but the presence of the pheromone did not appear to influence the binding of the Rgg1358 protein to DNA. For all experiments, the ldh probe was used as negative control (Fig. 6A, lanes 6 and 7). These experiments show that Rgg1358 binds sequences found in the shp1358 and ster_1357 gene promoters and that the presence of the pheromone does not seem to be required for binding. To confirm this novel result, a Prgg1358-luxAB fusion was constructed and used to transform strain LMD-9 and mutants deleted for the rgg1358 or shp1358 genes. The maximal expression of the rgg1358 promoter was the same in the wild-type strain background and the Δshp1358 mutant (70 RLU/OD600) but was twofold higher in the Δrgg1358 mutant (130 RLU/OD600) (Fig. 6B). This indicates that the Rgg1358 protein is a repressor for the transcription of its own gene with or without the SHP1358(15–23) pheromone. This repressor effect is most probably direct. Indeed, the DNA-binding motif recognized by the Rgg1358 protein in the promoter region of the shp1358 gene is located 16 nt upstream from the ATG of the rgg1358 gene and consequently in its promoter region. These findings are in good agreement with the EMSA results and suggest that the Rgg1358 protein does not require the presence of pheromone to bind DNA.
Here, we demonstrate the existence of, and describe, a QS mechanism involving a Rgg transcriptional regulator encoded by the rgg1358 gene and a small hydrophobic peptide called SHP1358 (summarized in Fig. 7). We show that the triggering of the mechanism is biomass-dependent and relies on the presence, in the extracellular medium, of the mature form of the shp1358 gene product, SHP1358(15–23), which is released by a C-terminal cleavage of the SHP1358 precursor. SHP1358(15–23) is most probably the pheromone of this QS mechanism. However, we can not completely exclude the possibility that it is an intermediate precursor to a more active species, or that there are other minor derivatives involved. The maturation of the SHP1358 precursor involves the transmembrane endopeptidase Eep. The capture of the pheromone from the extracellular medium appears to be performed mainly by one of the two oligopeptide binding proteins of the AmiCDEF oligopeptide transporter, AmiA3. Following the reimportation of SHP1358(15–23), it interacts with the regulatory protein Rgg1358. This interaction probably changes the Rgg1358 protein to an activated state, resulting in, at least, upregulation of the expression of the genes shp1358 and ster_1357. The effect of the interaction remains unclear, but we suggest that it may involve a change of either conformation or oligomerization. At a low cell density, the Rgg1358 regulator may already be bound to its recognition sequences upstream from the promoters of the genes shp1358 and ster_1357 but in an inactive form.
Further study of two steps of this mechanism would be particularly interesting. First, the secretion mechanism responsible for the export of SHP pheromones is unknown. As the SHP sequences are similar to signal sequences, it is difficult to predict whether or not they use the Sec machinery, need specific transporter(s) or insert into the cytoplasmic membrane to be exported. Second, the details of the roles of the two oligopeptide binding proteins are unclear. The oligopeptide binding protein AmiA3 is also the major actor for the capture and the sensing of the signalling peptide ComS (Gardan et al., 2009; Fontaine et al., 2010). The amiA1 gene is the first gene of the ami operon whereas the amiA3 gene is transcribed independently, and is separated from the ami operon by an insertion sequence. It would be of interest to determine if AmiA1 is more dedicated to nutritional function and if AmiA3 is largely responsible for the transport of pheromones into the cell as a result of a higher affinity for these hydrophobic peptides; alternatively, the presence of larger amounts of AmiA3 may be responsible for its dominant role.
This QS mechanism, involving a signalling peptide that is detected in the intracellular medium and imported by an oligopeptide transporter, has similarities with those described for the peptide-associated transcriptional regulators of the RNPP family. This family includes the PapR/PlcR and the NprX/NprR activators and the cCF10-iCF10/PrgX repressor (Declerck et al., 2007). Only the QS mechanisms involving PlcR and PrgX have been studied in detail: the role of the peptide in the control of the activation state of their cognate regulator has been revealed by structural studies (Shi et al., 2005; Declerck et al., 2007). PSI-BLAST analyses indicate amino acid sequence similarities between various Rgg proteins (MutR of S. mutans and RggD of S. gordonii) and the transcriptional regulators PlcR and PrgX, throughout the protein (Declerck et al., 2007). Like PlcR and PrgX, Rgg proteins have a Helix–Turn–Helix domain of the xenobiotic regulatory element family, which allows their binding to DNA (Neely et al., 2003; Vickerman et al., 2003; Rawlinson et al., 2005; Samen et al., 2006). We demonstrate such binding in this study. Rgg-associated peptides, like the anti-pheromone peptides from E. faecalis, resemble signal sequences and, like the pheromone and anti-pheromones, SHP1358 is matured by the endopeptidase Eep (An et al., 1999; An and Clewell, 2002). Rgg-associated peptides are not similar to the PapR peptide, which is longer and in which the presence of a pro-peptide leader upstream from the mature form of the peptide indicates that it is probably exported and matured by the Sec machinery. A common feature of the RNPP regulators is the presence of tetratricopeptide repeat (TPRs) domains in their C-terminal part that are involved in protein/protein or protein/peptide interactions (Goebl and Yanagida, 1991). TPRpred software (Karpenahalli et al., 2007) failed to identify such domains in the Rgg proteins. However, the TPR domains of the PrgX repressor were only discovered after structural analysis (Shi et al., 2005). Consequently, structural analyses, involving for example crystallography experiments, are required to determine whether or not the Rgg regulatory proteins contain TPR domains. We therefore propose that the SHP-associated Rgg are new members of the RNPP family. Indeed, SHP/Rgg systems have also some specific features. By searching the complete genome sequences of streptococci, we retrieved 68 SHP representing 28 different amino acid sequences. Among the various peptides associated with regulators of the RNPP family, SHPs constitute the largest group. They have one conserved aspartate or glutamate residue in their C-terminal part, possibly important for the maturation step and nearly always a conserved glycine residue at the end of their sequence. The DNA recognition sequence also seems to be specific with no similarities to the inverted repeat sequences recognized by PlcR or PrgX (Fig. 5B). Similar sequences have been found upstream from 80 % of shp genes belonging in S. thermophilus and other streptococci (e.g. B. Fleuchot, unpublished) and has the particularity of not being palindromic. Finally, our in vitro and genetic experiments demonstrate that SHP1358(15–23) interacts with the transcriptional regulator Rgg1358 but that it is probably not necessary for the binding of the Rgg1358 protein to DNA. In B. cereus, PapR pheromone binding to the PlcR activator leads to its multimerization, which is essential for its binding to the PlcR box and therefore for the regulation of the PlcR regulon (Slamti and Lereclus, 2002; Declerck et al., 2007). In E. faecalis, the situation is more complex. Indeed, although PrgX is able to bind DNA without the absence of its associated pheromones (Bae et al., 2002), their interaction is essential to regulate the expression of the target genes. The binding of the cCF10 sex pheromone to the PrgX repressor modifies its tetrameric conformation to become two dimers with less affinity for DNA, leading to the opening of a DNA loop and allowing access of the RNA polymerase to the target prgQ promoter. In contrast, the binding of iCF10 neutralizes the cCF10 pheromone activity and favours the return to the tetrameric form of PrgX (Shi et al., 2005; Dunny, 2007). Further experiments are needed to determine if the SHP1358(15–23) pheromone influences the oligomerization state of the Rgg1358 protein, and if this is the cause of the activation of the Rgg1358 regulator, independently of its binding capacity. However, these results indicate that this QS mechanism involving Rgg regulators differs from those described for PlcR and PrgX.
The regulation of competence in S. thermophilus LMD-9 is controlled by a Rgg-like protein (ComR) associated with a hydrophobic pheromone (ComS) reimported into the cell by the Ami oligopeptide transporter (Gardan et al., 2009; Fontaine et al., 2010). Mashburn-Warren et al. (2010) showed that comS/comR orthologues are also present in the genomes of S. mutans and in pyogenic and bovis streptococci. Moreover, they found results in S. mutans similar to those in S. thermophilus: (i) a deletion of comR, comS or of subunits of the oligopeptide transporter abolish transformability, (ii) ComR controls the transcription of the gene encoding the alternative sigma factor, SigX, itself involved in the control of the transcription of the genes encoding the DNA uptake and recombination machinery and (iii) extracellular addition of a synthetic peptide derived from the ComS precursor induces development of competence. The comS orthologues encode peptides, named XIP, which are 17 to 32 amino acids long and contain a conserved double-tryptophan in their C-terminal part (except in the case of S. thermophilus). The ComR regulators have similarity with both Rgg and PlcR proteins, and their predicted structures have similarities with the crystal structure of PlcR. These various observations led the authors to propose that XIP-associated ComR regulators are new members of the RNPP family. Although we agree with this, we think that SHP/Rgg and XIP/ComR pairs are distinct branches of the RNPP family for the following reasons: first, our phylogenetic analysis of the Rgg and Rgg-like proteins clearly shows that SHP-associated Rgg and XIP-associated ComR cluster in different parts of the tree and that the two families of SHP and XIP peptides differ in terms of their amino acid sequences and their conserved amino acids; second, their predicted DNA recognition sequences are not similar; and finally, the shp and rgg genes are in all cases transcribed in opposite directions whereas the comS genes all map downstream from the comR genes and in the same orientation. This last point has an incidence on the SHP/Rgg mechanism. Where the shp gene is located upstream from the rgg gene and divergently transcribed, Rgg can simultaneously be the repressor of its own gene and the activator of the shp gene. This potentially prevents the system from becoming over-active or overexpressed. For the new SHP/Rgg group described in this study (group III, in which the shp gene is downstream from the rgg gene and convergently transcribed), the 3′ ends of the two genes always overlap. This genetic configuration may also have consequences for the regulation of the system.
To facilitate discrimination between these small ComS/XIP hydrophobic peptides and the SHP we describe herein, we propose to restrict the use of the acronym SHP to small hydrophobic peptides with a conserved aspartate or glutamate residue about nine amino acids before the C-ter end of the sequence, encoded by a gene that is not transcribed in the same orientation as the rgg gene.
Another important issue is the range of physiological functions controlled by these systems. The discovery of 68 SHP/Rgg systems suggests a QS mechanism that is widespread in the streptococci family. Unlike XIP/ComR systems that all seem to control the transcription of at least the sigX gene and consequently the triggering of the competence state in different species of streptococci, the targets controlled by the SHP/Rgg systems are not obvious. For the SHP/Rgg1358 QS mechanism described here, one target, ster_1357, encoding a cyclic secreted peptide called Pep1357C was identified. This gene, mapping downstream from the rgg1358 gene, is followed by two genes encoding a protein belonging to the S-adenosylmethionine radical enzyme family and an efflux transporter; it has been suggested that the proteins encoded by these two genes are responsible for the maturation and the secretion of Pep1357C outside the cell (Ibrahim et al., 2007b). A nearly identical locus including the shp/rgg genes has been found in a recently sequenced Streptococcus mitis strain (SK564). We found a small number of homologous genes i.e. encoding a peptide, a S-adenosylmethionine radical enzyme and an exporter downstream from SHP-associated Rgg, in various streptococci species. Four similar genetic structures in B. subtilis and more recently in Bacillus thuringiensis have been studied. None of them are close to genes encoding Rgg proteins (Zheng et al., 2000; Brede et al., 2004; Butcher et al., 2007; Rea et al., 2010). All these loci encode modified peptides with antimicrobial activities. We therefore tested Pep1357C for bactericidal and bacteriostatic activity against several strains and species but no bacterial growth inhibition or killing effects were detected in the conditions tested.
Finally, it would also be interesting to explore other pathways linked to this QS mechanism. For example, shp/rgg loci are present in multicopy in streptococci, so it would be interesting to test for cross-activation within cells, or between strains or species. Any such interspecies interaction would be highly relevant to the study of various natural ecosystems, particularly for example the human oral microbiome where several species of streptococci live together (Dewhirst et al., 2010). Another level of complexity is brought by the presence of other RNPP systems in the same strain. Elucidating how the bacteria manage this plethoric repertoire of peptides and how it could be manipulated would be potentially valuable.
Phylogenetic analysis of the Rgg family
A preliminary PSI-BLAST search, against the non-redundant NCBI database, detected protein sequences with statistically significant similarity to Rgg regulators only in the order of Lactobacillales and the family Listeriaceae. We thus conducted a comparative analysis of the Rgg and Rgg-like repertoire in the genomes belonging to these two taxonomic groups. The analysis included a total of 90 complete genome sequences corresponding to all the sequenced isolates available in Genome Reviews release 125 (05-OCT-2010). This list includes representatives of eight genera: Streptococcus (48 genome sequences), Lactobacillus (22), Listeria (9), Lactococcus (5), Leuconostoc (3), Enterococcus (1), Oenococcus (1) and Pediococcus (1). Three prey representative of the three types of SHP/Rgg systems described in Ibrahim et al. (2007a) (stu.0182 in S. thermophilus LMG18311, smu.1509 in S. mutans UA159 and gbs1555 in S. agalactiae NEM316) were used to retrieve an exhaustive list of Rgg and Rgg-like proteins. Hits accounting for a least 80% of the length of one prey were collected after three iterations with PSI-BLAST (E-value ≤ 0.001). The sequences were aligned with muscle (default parameters) and a maximum-likelihood phylogenetic tree was constructed with phyml (JTT model of protein evolution, gamma distributed rate of evolution, 4 discrete categories of sites). The statistical support for each branch was quantified by analysing 200 bootstrap replicates. The tree was drawn in R using the library ‘ape’ (Paradis et al., 2004). Rgg-associated SHP or XIP peptides were recovered as follows. First, all hydrophobic peptides with basic residues (lysine and more rarely arginine) at the N-terminus were selected. Then, peptides with a glutamate or an aspartate at their C-terminus and encoded by a gene upstream from the rgg gene and divergently transcribed were considered to be SHP. Peptides with a double tryptophan motif and encoded by a gene downstream from the rgg gene and in the same orientation were considered to be XIP.
Bacterial strains and growth conditions
The bacterial strains used in this study are listed in Table 1. S. thermophilus strains were grown at 28°C or 42°C in M17 medium (Difco) supplemented with 10 g l−1 lactose (M17lac) or in a chemically defined medium (CDM) without shaking, under atmospheric air and with a ratio of air space to liquid of approximately 90% (Letort and Juillard, 2001). Escherichia coli strains were grown at 30°C or 37°C in Luria–Bertani (LB) broth with shaking (Sambrook & Russell, 2001). Agar (1.5%) was added to the media as appropriate. When required, antibiotics were added to the media at the following final concentrations: erythromycin, 200 µg ml−1 for E. coli or 5 µg ml−1 for S. thermophilus, kanamycin, 30 µg ml−1 for E. coli or 1 mg ml−1 for S. thermophilus and spectinomycin, 100 µg ml−1 for S. thermophilus. The optical density at 600 nm of the cultures was measured with a Uvikon 931 spectrophotometer (Kontron).
Erm, Spec, Km, Cm indicate resistance to erythromycin, spectinomycin, kanamycin and chloramphenicol respectively.
Arrows indicate construction by natural transformation or electroportation with chromosomal DNA or plasmid; ‘updown’ followed by the name of a gene indicates that the fragments upstream and downstream from the gene were inserted into pG+host9 to construct in frame deletions of the gene by double cross-over event (Garault et al., 2000).
Standard methods were used for DNA purification, restriction digestion, PCR, ligation and sequencing. E. coli TG1 or TG1 repA+ strains were used as hosts for cloning experiments. The oligonucleotides used for PCR are listed in Table 2. S. thermophilus was transformed using natural competent cells or electrocompetent cells (Ibrahim et al., 2007b). The 5′ tag-RACE method described by Fouquier d'Hérouel et al. (2011) was used to determine the transcriptional start site of the genes ster_1357 and shp1358 with oligonucleotides shp1358-tag and ster_1357-tag.
Table 2. Primers used in this study.
Plasmid and strain constructions
The plasmids used are listed in Table 3. Plasmid pGICB004, which allows the integration of transcriptional fusions to the luxAB reporter genes at the blp locus in S. thermophilus, was constructed in two steps. First, a 0.95 kb fragment including the blpR promoter sequence, the blpR ORF and the first 96 bp of blpH was amplified by PCR with primers UPINSLUX1 and UPINSLUX2, digested with PshAI and SpeI, and ligated upstream from luxAB into similarly digested pJIM4900. Next, a 1.05 kb fragment located downstream from blpX was amplified by PCR with primers DNINSLUX1 and DNINSLUX2, digested with SalI and PvuII, and inserted downstream from luxAB between the SalI and PvuII sites of the plasmid obtained in the first step. Plasmid pGICB004::Pshp1358 and pGICB004::Prgg1358 were constructed as follows. The shp1358 and rgg1358 promoter were amplified by PCR with oligonucleotides Pshp1358-EcoRI/Pshp1358-SpeI and Prgg1358-EcoRI/Prgg1358-SpeI, respectively, digested with restriction enzymes SpeI/EcoRI and ligated between the same restriction sites of pGICB004. Integration and excision of these plasmids in strains LMD-9, TIL 778 (Δshp1358), TIL 775 (Δrgg1358) and TIL 1198 (ΔamiA1) (Garault et al., 2000 for experimental procedure) led to the construction of strains TIL 1165 (blp::Pshp1358-luxAB) and TIL 1164 (blp::Prgg1358-luxAB), TIL 1200 (Δshp1358 blp::Pshp1358-luxAB) and TIL 1208 (Δshp1358 blp::Prgg1358-luxAB), TIL 1169 (Δrgg1358 blp::Pshp1358-luxAB) and TIL 1168 (Δrgg1358 blp::Prgg1358-luxAB) and TIL 1204 (ΔamiA1 blp::Pshp1358-luxAB). Competent cells of strain TIL 1204 were transformed with pG+host9::updown.shp1358. Integration and excision of the plasmid gave strain TIL 1303 (ΔamiA1 Δshp1358 blp::Pshp1358-luxAB). The SHP1358 peptide was overproduced by inserting the shp1358 gene into pBV5030 downstream from the P32 promoter. The shp1358 gene was amplified by PCR with oligonucleotides shp1358-NcoI and shp1358-PstI. The resulting fragment was digested with restriction enzymes NcoI/PstI and inserted into pBV5030::P32-ster_1357 digested with the same enzymes. The resulting plasmid pBV5030::P32-shp1358 allowed the overproduction of a modified SHP1358 peptide containing a glycine residue between the methionine at position 1 and lysine at position 2 of the wild-type sequence. Strain TIL 1202 (amiCDE::spec) was constructed using the overlapping PCR method. Briefly, the spectinomycin cassette (spec) was amplified by PCR with oligonucleotides Spec-R and Spec-F2 and the pAT28 as the template (Trieu-Cuot et al., 1990); it was fused by PCR to the beginning of the amiC gene (upstream fragment) and the end of the amiE gene and the amiF gene (downstream fragment). Oligonucleotides used for the amplification of the upstream fragment were amiCDE_up-F and amiCDE_up-R and for the downstream fragment amiCDE_down-F and amiCDE_down-R. The resulting 3 kb fragment was used to transform LMD-9 cells. Strains TIL 1304 (amiA3::erm Δshp1358 blp::Pshp1358-luxAB) and TIL 1305 (ΔamiA1 amiA3::erm Δshp1358 blp::Pshp1358-luxAB) were constructed as follows. Cells of strain TIL 1200 were transformed with chromosomal DNA from strains TIL 1197 (amiA3::erm) and TIL 1199 (ΔamiA1 amiA3::erm) respectively. A similar procedure was used to construct strain TIL 1203 (amiCDE::spec blp::Pshp1358-luxAB): TIL 1165 was transformed with chromosomal DNA from strain TIL 1202. Strain TIL 773 (Δeep) was constructed by deleting an internal fragment of the gene by a double cross-over event using pG+host9. Briefly, oligonucleotides eep-XhoI with eep-EcoRIA and eep-EcoRIB with eep-SpeI were used to amplify upstream and downstream fragments of the eep gene. These two fragments were double digested with restriction enzymes XhoI with EcoRI and EcoRI with SpeI, respectively, and ligated between the XhoI and SpeI restriction sites of pG+host9. The resulting plasmid was used to transform electrocompetent cells of strain LMD-9. Integration and excision of the plasmid led to the in-frame deletion of the eep gene. Strains TIL 1206 (Δeep pBV5030::P32-shp1358) and TIL 1213 (ΔamiCDE pBV5030::P32-shp1358) were constructed by electrotransformation of strains TIL 773 and TIL 883, respectively, with pBV5030::P32-shp1358. Plasmid pET28a::Rgg1358 was constructed to allow the expression and the purification of the N-terminus His-tagged Rgg1358 protein in E. coli as follows. The rgg1358 gene was amplified by PCR with oligonucleotides Rgg1358-NdeI and Rgg1358-XhoI. The resulting 882 bp fragment was digested with restriction enzymes NdeI/XhoI and ligated into in the expression vector pET28a digested with the same restriction enzymes. The resulting plasmid was used to transform E. coli strain Rosetta to obtain TIL 1318 (E. coli Rosetta pET28a::Rgg1358). All constructions were verified by PCR and validated by sequencing.
Erm, pJIM4900 with a 0.95 kb insert containing two fragments upstream and downstream respectively from the luxAB gene and allowing the integration of the plasmid by double cross-over event at the blp locus of S. thermophilus.
Erm, Derivative of pGICB004 used to introduce a Pshp1358-luxAB transcriptional fusion at the blp locus.
Erm, Derivative of pGICB004 used to introduce a Prgg1358-luxAB transcriptional fusion at the blp locus.
Erm, Replicative plasmid in E. coli and Gram positive bacteria.
Erm, Derivative of pBV5030 allowing the overexpression of the ster_1357 gene of S. thermophilus LMD-9.
Gardan (e.g. unpublished)
Erm, Derivative of pBV5030::P32-ster_1357 allowing the overexpression of the shp1358 gene of S. thermophilus LMD-9.
Km, vector for production of His tagged proteins
A derivative of pET28a used to overproduce Rgg1358
Cells were grown overnight at 42°C in CDM. These cultures were then diluted in 50 ml of CDM to a final OD600 of 0.05 and incubated at 42°C, except for the experiments reported in Fig. 2D where cultures were diluted to a final OD600 of 0.012, 0.025, 0.05 or 0.1. Aliquots of 1 ml of the culture were sampled at regular intervals until the culture reached stationary phase and analysed as follows: OD600 was measured, then 10 µl of a 0.1% nonyl-aldehyde solution was added and the luminescence was measured with a Luminoskan TL (Labsystems). Results are reported in Relative Luminescent Units divided by the OD600 (RLU/OD600). For cocultures, each culture was diluted to an OD600 of 0.025 and results are expressed as luminescence divided by the total OD600. For experiments involving mixing supernatants of one strain with cell pellets of another strain containing a luxAB reporter fusion, both supernatants and pellets were harvested by centrifugation (5000 g for 10 min at RT) of equal volumes of cultures at OD600 0.6. Finally, the synthetic peptide EGIIVIVVG, stored in lyophilized form and freshly prepared in DMSO, was added to cultures with an OD600 of 0.5.
Supernatants from cultures of strains LMD-9 and TIL 778 were recovered by centrifugation. Aliquots of 10 µl of ultra-filtered supernatant (0.22 µM, Millipore) were loaded on a Pepmap C18 column (length 150 mm, 75 µm ID, 100 Å; Dionex, Voisin-le-Bretonneux) and analysed on-line by mass spectrometry on a LTQ-Orbitrap Discovery apparatus (Thermo Fischer, San Jose). First, only ion current of m/z corresponding to all possible C-terminal amino acid sequences produced by the cleavage of the SHP1358 sequence were sought in the supernatant of strain LMD-9. Next, we checked that these masses, detected in the LMD-9 supernatant, were not present in the supernatant of the Δshp1358 mutant and that they were eluted from the Pepmap C18 column with a retention time compatible with the hydrophobicity of the putative amino acid sequence. Then, the ion with mass that fulfilled these conditions was fragmented and analysed on the Orbitrap mass analyser, to identify accurately the sequence of the corresponding peptide. Finally, the absence of the mature form of this peptide from the Δshp1358 supernatant was confirmed after fragmentation on LTQ by monitoring the current of the b7 ions daughter (724.46 m/z) in the Orbitrap mass analyser.
Production and purification of Rgg1358
A preculture of TIL 1318 was diluted in 500 ml of LB with kanamycin to an OD600 of 0.05 and incubated at 37°C with shaking. At OD600 0.5, recombinant protein production was induced by adding IPTG at a final concentration of 1 mM and the incubation continued for 4 h at 30°C. Cells were harvested by centrifugation (5000 g for 10 min at 4°C), rinsed and resuspended in disrupter buffer (20 mM Tris HCl pH 8.0, 250 mM NaCl, 0.1% Tween 20, 10% v/v glycerol and 10 mM Imidazole) to obtain a OD600 of 100 ml−1. Cells were disrupted at 2.0 kbars with a Basic Z Cell Disruption System (Constant Systems). The soluble fraction was collected by centrifugation (5000 g for 20 min at 4°C) and production of the recombinant His-tagged Rgg1358 protein checked by SDS-PAGE. The product was then purified with the QIAexpressionist kit (Qiagen), according to the manufacturer's instructions with the following modification: elution was carried out with an imidazole concentration gradient from 100 mM to 250 mM.
Real-time binding kinetics experiments were conducted on a BIAcore 3000 apparatus (GE Healthcare Europe). The synthetic peptide EGIIVIVVG was immobilized on a CM5 sensor chip using amine-coupling chemistry. For this, the surface of the chip was activated for 7 min with a mixture of 0.05 M NHS and 0.2 M EDC (Amine coupling kit, Biacore, GE Healthcare). The synthetic pheromone (37 µM, in sodium acetate buffer, pH 5.0) was then covalently linked to the surface giving up to 2000 resonance units (RU). Ethanolamine (1 M, pH 8.5) was injected for 7 min to block the remaining activated groups. The purified Rgg1358 protein (5 µM) was injected over 2 min, at flow of 25 µl min−1 in 20 mM HBS (pH 7.4) to allow binding to the immobilized SHP1358(15–23) pheromone; dissociation was recorded for 12 min after the end of injections. The surface was regenerated by injecting glycine buffer (pH 2.0) for 2 min. The sensorgram observed in a cell immobilizing pheromone was corrected by subtracting the response observed in a cell immobilizing only ethanolamine (reference surface). Two sets of control experiments were performed to check the specificity of binding: (i) BSA was injected over the immobilized SHP1358(15–23), and (ii) Rgg1358 was injected over a chip surface with the control peptide, IAILPYFAGCL, covalently immobilized (purchased from eurogentec). All measurements were performed at 25°C. Sensorgrams were analysed using BIAevaluation Software.
DNA probes of approximately 110 bp of the shp1358 and ster_1357 promoter regions were amplified by PCR with oligonucleotides EMSA-shpF with EMSA-shpR and EMSA-ster_1357F with EMSA-ster_1357R respectively. The ldh promoter region was amplified with oligonucleotides EMSA-ldhF and EMSA-ldhR as a control. DNA probes were 3′ end labelled and gel shift reactions carried out by using the DIG GEL Shift Kit, 2nd Generation (Roche), according to the manufacturer's instructions. The DNA binding reactions, coupling probes to the Rgg1358 protein with or without the synthetic peptide EGIIVIVVG (purchased from eurogentec), involved incubation for 30 min at 42°C. Samples were then loaded on a 4–16% native polyacrylamide gels (nativePAGE 4–16% Bis-Tris Gel, Invitrogen) and subjected to 70 V for approximately 2 h in 1 × TBE buffer. The labelled probe/protein complexes were transferred to a positively charged nylon membrane (GE Healthcare Amersham Hybond – N+) by electro-blotting for 30 min in a Mini Trans-Blot Cell (Biorad) in 0.5 × TBE buffer. DNA complexes were detected by chemiluminescence on X-ray film (Amersham Hyperfilm ECL) according to the manufacturer's instructions.
This study was supported by the Institut National de la Recherche Agronomique (INRA) and the Ministère de l'Education Nationale de la Recherche et de la Technologie (MENRT). The ‘Plateforme d’Analyse Protéomique Paris Sud-Ouest' (PAPPSO, Alain Guillot) received the financial support from the Ile de France regional council and from CEMAGREF. Work in the group of P. Hols was financially supported by FNRS. L. Fontaine is postdoctoral researcher at FNRS. P. Hols is research associate at FNRS. We thank L. Topisirovic for plasmid pBV5030. We thank F. Rul, V. Juillard and P. Serror for their critical reading of the manuscript, E. Chambellon and M. Cote for their technical assistance, and E. Bruneau for her technical advice. Finally, we are grateful to F. Repoila for helpful advice concerning his 5′ tag-RACE method.