AMethodforStructure4

We also used this mutant (SMdC) as a host to generate two types of lacZ transcriptional reporter fusion strains to assay the promoter activity of QS-controlled target genes, comDE and nlmAB, in response to addition of CSP. The lacZ reporter strain, SMdC-pYH2 (PcomDE::lacZ, comC ^?, Spec^r, Kan^r), along with its background control strain SMdC-pSL (comC ^?, Spec^r, Kan^r) were then assayed for β-galactosidase activity. The results confirmed that upon addition of CSP strain SMdC-pYH2 was induced to express predicted β-gal activity at similar levels to those of the previously constructed lacZ reporter strain in UA159 (10). In contrast, the background control strain SMdC-pSL did not show significant difference in the expression of β-gal activity. To assay QS activation of the downstream genes controlled by the ComCDE signaling transduction pathway, we also constructed two nlmAB–lacZ fusion strains by transforming a previously generated fusion construct PnlmAB::lacZ (6) into SMdC-GS5 and parent GS5. These strains were genetically confirmed and named SMdC-PnlmAB (PnlmAB::lacZ, comC ^?, Kan^r, Spec^r) and SMGS5-PnlmAB (PnlmAB::lacZ, Kan^r), respectively. The β-gal activity assay showed that prior to addition of CSP SMdC-PnlmAB was defective in expressing a normal level of β-gal activity (Fig. 4 a). Upon addition of CSP, however, SMdC-PnlmAB was rapidly induced to express an increasing level of β-gal activity and reached to the highest level around 90 min after addition of CSP (Fig. 4 b). Interestingly, SMdC-PnlmAB still maintained such a high level of the activity for at least 2 h. Clearly, such expression of β-gal activity appeared to be a major difference between this reporter strain and SMdC-pYH2, in which the expression of β-gal activity declined around 40 min after the addition of CSP (10). However, the expression of β-gal activity by SMdC-PnlmAB was highly inconsistent with those of other reports (6, 7), suggesting that the expression of nlmAB gene was likely enhanced by an unknown factor following initial activation by CSP. Thus, both lacZ reporter strains representing two levels of gene regulation provided us with a sensitive detecting system to assay and identify signaling peptide agonists.

Application of the Methods to Identify Signaling Peptide Agonists

Based on the three-dimensional structures of S. mutans wild-type signaling peptide UA159sp (CSP) and C-terminally truncated peptide TPC3 from mutant JH1005 defective in genetic competence, we designed and synthesized a series of truncated peptides and peptides with amino acid substitutions (10). By functional analysis of these peptides, we found that CSP from S. mutans displayed two functional domains. The C-terminal structural motif consisting of a sequence of polar hydrophobic charged residues is crucial for activating the signal transduction pathway, while the core α-helical structure extending from residue 5 through the end of the CSP is required for receptor binding. The AC₅₀ of CSP (UA159sp) was determined to be about 25 nM (10). With these data, we developed a rationale to design and synthesize several new peptide agonists (Table 1 ). By assaying their activity in quorum-sensing activation using the methods described here, we found that hydrophobic residues, leucine (L) or phenylalanine (F), could replace the hydrophilic residues at positions 9, 12, or 13 of the core α-helix of CSP without significant change of its activity (Table 1 ). In contrast, peptides, F7Q, F11Q, and F15Q that had hydrophobic phenylalanines at positions 7, 11, or 15 replaced by a hydrophilic residue, glutamine (G), failed to activate quorum sensing (10). In addition, we found that alanine (A) at position 18 appeared to be important for the activity, since a deletion or replacement of this residue abolished the peptide activity. The evidence strongly suggests that the maintenance of the hydrophobic face of the signaling peptide is crucial for receptor recognition. These results are consistent with the results of Havarstein and colleagues (9), who have found that the hydrophobic patch of CSP1 and CSP2 of S. pneumoniae mainly contributes to the receptor recognition and the specificity. Interestingly, the hydrophobic patch in the S. pneumoniae CSP1 also consists of three highly hydrophobic phenylalanines at positions 7, 8, and 11 and an isoleucine at position 12. In contrast to S. pneumoniae, however, CSP receptor recognition in S. mutans appears to be less specific among strains. For example, the peptides with a truncation of N-terminal residues, such as TPN1 or TPN2, or C-terminal residue TPC1, also functioned as agonists with AC₅₀ values similar to CSP (10). These appear to be consistent with the report of Allan et al. (28), who have found no strict correlation between the CSP genotype and the ability to induce quorum sensing among S. mutans strains. Together, these findings suggest that the S. mutans CSP and its variants may have less recognition specificity of interacting with its receptor protein. Unlike S. pneumoniae that requires two independent signaling systems, the ComCDE and BlpRH, to regulate genetic competence and bacteriocin production, S. mutans only uses one signaling system with more flexible peptide–receptor interaction to regulate these phenotypes. We speculate that more peptide variants may be found in S. mutans isolates, and they likely function in the same fashion as CSP (UA159sp) to activate quorum sensing for bacteriocin production and genetic competence. To support this speculation, we designed a shorter peptide agonist, SPA-18, to assay its activity. We found that this shorter peptide could completely replace CSP (UA159sp) in activating quorum sensing (Fig. 4 ), genetic competence (data not shown), and bacteriocin production (Fig. 5 ).

Table 1 Peptide analogs and their agonist activity in quorum sensing

^aModifications of amino acid residues are in bold and underlined

^bThe AC₅₀ values were calculated from a dose–response curve using Graphpad Prism 4

Fig. 5 QS-controlled bacteriocin production in S. mutans strains. Only peptides CSP, SPA-18 (5), and R13F (4) induce the same level of bacteriocins as GS5 (1).