Genetic Analysis of a Unique Bacteriocin, Smb, Produced by Streptococcus mutans GS5

Abstract

A dipeptide lantibiotic, named Smb, in Streptococcus mutans GS5 was characterized by molecular genetic approaches. The Smb biosynthesis gene locus is encoded by a 9.5-kb region of chromosomal DNA and consists of seven genes in the order smbM1, -T, -F, -M2, -G, -A, -B. This operon is not present in some other strains of S. mutans, including strain UA159. The genes encoding Smb were identified as smbA and smbB. Inactivation of smbM1, smbA, or smbB attenuated the inhibition of the growth of the indicator strain RP66, confirming an essential role for these genes in Smb expression. Mature Smb likely consists of the 30-amino-acid SmbA together with the 32-amino-acid SmbB. SmbA exhibited similarity with the mature lantibiotic lacticinA2 from Lactococcus lactis, while SmbB was similar to the mersacidin-like peptides from Bacillus halodurans and L. lactis. We also demonstrated that Smb expression is induced by the competence-stimulating peptide (CSP) and that a com box-like sequence is located in the smb promoter region. These results suggest that Smb belongs to the class I bacteriocin family, and its expression is dependent on CSP-induced quorum sensing.

Dental caries has plagued humans since the dawn of civilization and still constitutes one of the most common human infectious diseases. Multiple species of bacteria inhabit the human oral cavity, and the species most commonly associated with human caries is Streptococcus mutans (22). Among the attributes thought to contribute to the virulence of S. mutans is its ability to elaborate antimicrobial or bacteriocin-like substances, which may provide a selective advantage for initial or sustained colonization in a milieu of densely packed competing organisms found in dental plaque (38, 48).

Bacteriocins are a family of ribosomally synthesized peptide antibiotics that are produced by bacteria (11, 14, 18, 37 ). They are subdivided into four different classes based on biochemical and genetic characteristics (14, 16, 17). Class I and class II bacteriocins are by far the most extensively studied because they are the most abundant and most prominent in industrial applications (26). Class I bacteriocins, named lantibiotics, contain two modified amino acid residues, lanthionine and/or methyllanthionins, which are formed posttranslationally (7). The primary product of the lantibiotic structural gene is a precursor with an N-terminal leader sequence followed by a C-terminal propeptide which undergoes modification. Once modified within the cell, the bacteriocin is secreted by a dedicated transporter and the N-terminal leader sequence is cleaved by a protease (12, 49).

Some strains of S. mutans produce antimicrobial substances called mutacins (3, 4, 30, 34, 35, 36). Mutacins have been classified into two families: the lantibiotics and the nonlantibiotics. Classification of mutacin-producing strains based on their bactericidal activities, their sensitivities to other or self-produced mutacins, and the presence of plasmids divides the mutacins into four types, I, II, III, and IV (28, 34, 35, 36). Mutacins I, II, and III belong to the lantibiotic family, while mutacin IV is a dipeptide nonlantibiotic bacteriocin. The structural genes for the prepropeptides of mutacins I, II, III, and IV have been sequenced, and their biosynthetic loci are composed of multiple genes, including those involved in regulation, cleavage, transport, and immunity to the produced mutacins (28, 34, 35, 36). However, a recent report suggests that the bacteriocin previously demonstrated to be synthesized by S. mutans GS5 (31) is not a member of the mutacin I, II, or III family (23).

Quorum sensing in gram-positive bacteria has been found to regulate a number of physiological activities, including competence development in Streptococcus pneumoniae (19) and S. mutans (21). A quorum-sensing system essential for genetic competence in S. mutans was recently identified (21). This cell-cell signaling system involves at least five gene products encoded by comAB (33) and comCDE (21). The comC genes encode a competence-stimulating peptide (CSP) precursor. Recently, several competence-specific genes which are likely involved in the DNA uptake process and in recombination, such as cilA (ssb2, a gene for single-stranded breaks), cilB (similar to dprA in Haemophilus influenzae) (15), cilC (ccl, similar to comC in Bacillus subtilis), cilD (cglABCDE), cilE (celAB), and coi (2, 32), were identified. These operons contain a conserved consensus sequence, TACGAATA (com box), at position −10 from the transcription start site and a T-rich region at −25 (2).

In the present study, we have used Tn916 mutagenesis and the single-specific-primer PCR (SSP-PCR) (29, 43, 44) technique to characterize a novel smb (S. mutans bacteriocin) operon in S. mutans GS5. We demonstrate that the smb genes are present in an operon structure using transcriptional analysis. Targeted gene integration mutagenesis was also used to probe the essentiality of the operon genes for Smb production. In addition, the results of sequence analysis and homology searches demonstrated that Smb is a class I two-component bacteriocin and is regulated by CSP.

MATERIALS AND METHODS

Bacterial strains and plasmids.

S. mutans GS5 was used in this study for the production and characterization of the Smb bacteriocin. RP66 (group C streptococcus) was used as an indicator strain for Smb activity assays (31). S. mutans GS5 and RP66 were grown in Todd-Hewitt (TH) medium (Becton Dickinson and Co., Cockeysville, Md.) in an anaerobic atmosphere of 85% N₂, 10% CO₂, and 5% H₂. Transformants of S. mutans were selected following their growth on TH broth agar plates supplemented with 10 μg of erythromycin per ml or 3 μg of tetracycline per ml.

Agar plate assays.

Loopfuls of stationary-phase cultures of S. mutans strains were stabbed into a TH broth agar plate. The plate was incubated at 37°C for 24 h. RP66, to be assayed for sensitivity, was grown to an optical density of 0.2 at 550 nm. Each culture was then diluted 1:100, and 0.2 ml of this dilution was pipetted into a tube containing 4 ml of molten Trypticase soy broth containing 1% agar (Becton Dickinson and Co.). This solution was mixed and poured evenly onto the surfaces of the plates and incubated at 37°C for an additional 24 to 48 h, and the diameters of the zones of inhibition were measured.

Characterization of the Tn916 insertion region.

The broad-host-range conjugative transposon Tn916, originally identified on the chromosome of Enterococcus faecalis (9), has been used as a mutagen in streptococci. SSP-PCR (29, 43, 44) was performed for the characterization of the Tn916 insertion sites. Briefly, chromosomal DNA from a transposon-containing GS5 mutant was isolated and digested with the restriction endonuclease EcoRI and then ligated into EcoRI-digested pUC19. The ligation mixture served as the template for amplification with transposon-specific primers (6) and an M13 primer. Subsequently, the PCR product was used as a template for sequencing. Amino acid homology searches and comparisons were carried out with the FASTA and BLAST network services of GenBank.

Construction of the smb mutants.

The mutants of the smbM1, smbA, smbB, and smbA-smbB genes were created by double-crossover homologous recombination via insertion of an erythromycin resistance determinant into each gene. The plasmids used for disruption of the smbM1, smbA, and smbB genes were prepared as follows. The PCR fragments of the upstream and downstream regions of each gene were amplified with pairs of primers and chromosomal DNA from GS5 as a template. Initially, PCR products of the downstream region were ligated into the pResEmMCS10 plasmid (42) containing the Erm cassette. Next, PCR products of the upstream regions were ligated into the other flanking site of the Erm cassette in the plasmids. The resulting plasmids were linearized following BamHI digestion, and the linearized plasmids were used to transform S. mutans GS5. Confirmation of plasmid insertions causing gene disruption was performed either by Southern blotting or by PCR (data not shown).

Extraction of RNA.

Total RNA was isolated from 15 ml of log-phase cell cultures. After centrifugation, the cells were suspended in 0.3 ml of diethylpyrocarbonate-treated water. The samples were transferred to FastRNA tubes with blue caps (Qbiogene, Inc., Carlsbad, Calif.), and 0.9 ml of TRIzol reagent (Invitrogen) was then added. Cells were broken by a FastPREP FP120 homogenizer (Qbiogene) at a speed setting of 6.0 for 30 s. After samples were placed on ice for 2 min, 0.2 ml of chloroform was added and the tubes were vortexed for 1 min. The mixtures were then placed at room temperature for 2 min and centrifuged at 12,000 × g for 5 min at 4°C; 0.5 ml of chloroform was then added to the supernatant fluids, and the mixtures were vortexed and centrifuged again as described above. The RNA was finally precipitated from the aqueous phase with isopropanol, and the resulting pellets were dried and resuspended in 20 μl of diethylpyrocarbonate-treated water.

Transcription analysis.

For Northern blot analysis, a quantity (4.5 μl) of RNA (15 μg) was mixed with 15.5 μl of sample buffer (2.0 μl of 10× MOPS [morpholinepropanesulfonic acid], 3.5 μl of 37% [vol/vol] formaldehyde, 10 μl of formamide) and denatured at 65°C for 10 min. After dye solution was added, the RNA fragments were separated by electrophoresis in 1% agarose gels containing 3% formaldehyde at 4°C. The gel was washed twice with 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) for 15 min each time to remove formaldehyde, and blotting was carried out with 20× SSC overnight. The blotted membrane was washed with distilled H₂O twice for 5 min and fixed by UV cross-linking. Hybridization was then carried out with DIG Easy Hyb (Boehringer Mannheim Corp., Indianapolis, Ind.) with a digoxigenin (DIG)-labeled probe at 50°C according to the directions of the supplier.

For reverse transcription-PCR (RT-PCR) analysis, RNA samples were treated for 15 min at 37°C with 1.0 U of RNase-free DNase (Amersham Biosciences Corp., Piscataway, N.J.) per ml to remove contaminating DNA. Reverse transcription was carried out with SuperScript III (Invitrogen) according to the directions of the supplier.

Nucleotide sequence accession number.

The nucleotide sequences of the smb genes have been submitted to the DDBJ, EMBL, and GenBank nucleotide sequence databases under the accession number AB179778.

RESULTS

Identification of the bacteriocin gene locus following transposon mutagenesis.

The broad-host-range conjugative transposon Tn916, originally identified on the chromosome of E. faecalis (9), was used to mutagenize S. mutans GS5. One of 8,000 Tn916 transformants of strain GS5, mutant H1, exhibited a pronounced bacteriocin-negative phenotype when it was screened on agar plates with indicator strain RP66 (Fig. 1). Southern blotting indicated the presence of a single copy of the Tn916 transposon in the H1 mutant chromosomal DNA (data not shown). Transformation of strain GS5 with the DNA isolated from mutant H1 resulted in mutants with reduced Smb production (data not shown).

FIG. 1.

Results of bacteriocin plate assays with wild-type strain GS5 (a) and mutant H1 (b).

In order to characterize the Tn916 insertion site, SSP-PCR was performed. A 2.5-kb fragment was generated from the TnLO-2 and M13-Fw primer set using ligation mixtures of EcoRI-digested H1 mutant chromosomal DNA and plasmid pUC19. The other flanking region of the Tn916 insertion was also amplified by using the same technique with the TnR-O and F13-Rev primer set. Sequence analysis of these regions indicated that the GS5 bacteriocin genes, designated smb, were located downstream of the Tn916 insertion site (Fig. 2). Additional primers were then designed according to the newly derived DNA sequences. The original smb locus was amplified by PCR from the wild-type strain GS5, and the PCR products were used as templates for further sequence analysis. Comparison of the sequence adjacent to the Tn916 insertion region of H1 with the wild-type GS5 sequence showed that Tn916 was inserted into the promoter region of the smb locus. A total of about 11 kb of DNA was sequenced in this region. The smb locus is not present in the S. mutans UA159 database (http://www.genome.ou.edu.smutans.html). Sequence analysis revealed that there are 12 putative open reading frames in the order of cyl-smbM1-smbF-smbT-smbM2-smbG-smbA-smbB (Fig. 2), which were followed by three predicted transposases and an ABC transporter gene. The cyl and ABC transporter genes are found adjacent to one another in the UA159 database. The cyl gene encodes a leucyl-tRNA synthetase and therefore was presumed to define the upstream border of the smb gene locus in strain GS5.

FIG. 2.

Smb biosynthesis genes. The orientation of the genes and their relative distances are shown. The cyl gene encodes leucyl tRNA synthetase and is therefore presumed not to be part of the smb biosynthesis gene operon. The arrows represent promoters (P), and a potential terminator (black oval) is also depicted.

Sequence analysis of the smb locus and homology of the Smb gene products with other proteins.

Inspection of the upstream region of smbM1 revealed a potential ribosomal binding site with the sequence AAGGG, which was 6 bp upstream of the predicted initiating codon, GTG. A GenBank search for similar peptides revealed that the first gene, smbM1, encoded a protein of 958 aa, which showed significant similarity to the lantibiotic-mersacidin-modifying enzyme from Bacillus halodurans (46) and scnM from Streptococcus pyogenes (13). The second gene, smbF, began 11 bp after the stop codon for smbM1. This gene would encode a protein of 274 aa which shows homology with the SpaF protein from Bacillus subtilis (45) and MutF from another bacteriocin-producing strain of S. mutans, UA787 (36). The SpaF and MutF proteins are immunity proteins for lantibiotic bacteriocins. The third gene, smbT, encoded 243 aa and bore similarity to an ABC transporter (27). The reading frames of smbF and smbT overlapped in 5 bp. The fourth gene in the operon, smbM2, encoded a protein of 876 aa, which resembled the salivaricin A modification protein from S. pyogenes (1) and the lantibiotic-modifying enzyme from Staphylococcus aureus (51). The reading frame of smbM2 overlapped with that of smbT by 16 bp and is followed closely by smbG. The latter gene encoded a putative protein of 698 aa and exhibits similarity with the plnG immunity gene from Lactobacillus plantarum (8). A GenBank search for similar peptides revealed that SmbA was homologous to lacticinA2 (39) and displayed 46.9% identity and 75% similarity with the lacticin at the amino acid level. Furthermore, SmbB was similar to a mersacidin-like peptide (24, 46), with the putative mature peptide sharing 52.9% identity and 70.6% similarity with the marsacidin-like peptide (Fig. 3). LacticinA2 produced by Lactococcus lactis, the mersacidin-like peptide produced by B. halodurans, and lacticinA1 secreted by L. lactis are components of lantibiotic bacteriocins (24, 46). These results suggested that the two peptides, SmbA and SmbB, belong to the dipeptide lantibiotic bacteriocin family (10, 24, 25, 39, 40).

FIG. 3.

Similarity of prepropeptides of Smb with other lantibiotic peptides. (top) Sequence alignment of SmbA and lacticinA2 prepeptide; (bottom) sequence alignment of SmbB and mersacidin-like peptide secreted by B. halodurans. Dark-gray boxes represent identical amino acids, and light boxes denote conserved changes. Arrows indicate putative cleavage sites for the prepeptides.

Inactivation of the smb genes and their effects on Smb production.

To determine whether the two putative structural genes smbA and smbB are required for Smb activity, we disrupted the two genes individually or together by inserting an erythromycin resistance gene cassette within the genes (Fig. 4a). The three resulting mutants, GS5SmbAEm (H2A), GS5SmbBEm (H2B), and GS5SmbABEm (H2AB), were assayed for Smb production by plate assays against RP66 (Fig. 4b). Although RP66 displays some sensitivity to most S. mutans strains, including strain UA159 (data not shown), which lacks the smb genes, each of the mutants was markedly attenuated in its ability to inhibit the growth of indicator strain RP66 compared with that of parental strain GS5. This result suggests that both the smbA and smbB genes are necessary for Smb activity. We also disrupted the smbM1 gene by insertional inactivation and also constructed a mutant with the Erm cassette inserted into the same region disrupted by Tn916 in mutant H1. The fact that we detected transcription of the genes downstream from the Erm cassette in mutants H2A and H2M1 by RT-PCR (data not shown) demonstrated that the H2A and H2M1 mutations did not produce polar effects. The resulting mutants, GS5SmbM1Em (H2M1) and GS5SmbPEm (H2P), were also markedly attenuated in bacteriocin production (Fig. 4b).

FIG. 4.

Effects of smbM1 (H2M1), smbA (H2A), smbB (H2B), smbA, and smbB (H2AB) as well as smbP mutations (H2P) on Smb production. (a) Loci targeted for mutagenesis are indicated by arrows. (b) Results of bacteriocin plate assays.

Transcriptional analysis of the smb operon.

To determine how many transcription units comprise the Smb biosynthetic locus, Northern blotting and RT-PCR were performed with wild-type GS5. We performed Northern blot analysis with DIG-labeled RNA probes (200 to 300 bp) specific to each of the seven smb genes. Hybridization with the smbA or smbB probe detected the same transcript of about 500 bp in size (Fig. 5). This transcript size was approximately equal to that predicted for cotranscription of the two genes. This result suggested that transcripts for both smbA and smbB corresponded to an initiation site upstream of smbA. Multiple attempts at detection of smb operon transcripts by Northern blot analysis of GS5 using the other probes were unsuccessful, even when 10 times the amount of RNA was analyzed (data not shown). Therefore, we hypothesized that the mRNA encoding smbM1 to smbG or smbB is transcribed together and is too large to be detected following Northern blot analysis. Furthermore, it is unlikely that these negative results were due to weak transcription from the promoter upstream of smbM1. We further tested this hypothesis using the RT-PCR approach. The results of RT-PCR analysis indicated that mRNA encoding SmbM1 was carried on the same transcript as that encoding SmbF, since a product of the expected size was amplified with specific smbM1 forward and smbF reverse primers from wild-type GS5 (data not shown). A similar RT-PCR analysis revealed that the smbF transcript was cotranscribed with smbT, smbT was cotranscribed with smbM2, and smbT was cotranscribed with smbA, indicating that the smb locus represents a seven-gene operon. Taken together, these data suggest that there is a single smb operon with two promoters, one upstream of smbM1 and the other flanking smbA with a terminator sequence downstream of smbB.

FIG. 5.

Northern blot analysis of transcripts of the smbA and smbB loci. RNA isolated from wild-type GS5 was hybridized with a DIG-labeled probe specific to smbA (a) or smbB (b). The arrow indicates the smbA-smbB transcript. The left lane contains the RNA molecular size markers.

The CSP is required for the transcription of smbA and smbB.

Recent results have indicated that a GS5 comC null mutant (CC1301) (52) was attenuated in bacteriocin production (49a). However, addition of synthetic CSP to the culture of this mutant restored the production of Smb. With cocultured samples, it was demonstrated that mutants H2AB, H2M1, and H2P also complemented the comC mutant for bacteriocin production (data not shown). These data indicated that these smb mutants produce CSP and that their inability to secrete bacteriocin was not due to a defect in CSP secretion. In order to determine which smb genes are affected by the competence of strain GS5, Northern blot analysis and RT-PCR were used to examine expression of these genes in the presence and absence of CSP. Transcription of the smbA and smbB genes appeared to be weaker in the comC mutant than in the parental strain GS5 following Northern blot analysis (Fig. 6). Furthermore, even weaker expression of smbA and smbB was apparent in the mutant whose promoter region upstream of smbM1 was altered. This result suggests that this promoter may be the major promoter for regulating the expression of these two genes. RT-PCR analysis also confirmed these results (data not shown). These results are consistent with the agar plate assay results demonstrating that the comC mutant inhibition zone against RP66 is much smaller than that of GS5 but larger than that of H2P (data not shown). Using RT-PCR, the other transcripts of the smb operon (M1 to G) were detected at the same levels in all of strains (data not shown). These results show that the reduction of Smb production in the comC and H2P mutants resulted from decreased transcription of the smb operon.

FIG. 6.

Northern blot analysis of wild-type S. mutans (lane 1), the comC mutant (lane 2), and the H2P mutant (lane 3) with the DIG-labeled smbAB probe. The arrow indicates the smbA-smbB transcript. The RNA molecular size marker is present in the leftmost lane (lane M).

Most of the CSP-induced genes in S. pneumoniae and S. mutans have a com box (TACGAATA) sequence located in their promoter regions (9, 20, 50). We also identified candidate sites which shared sequence elements with the com box and which were located in the apparent extragenic regions. We identified a com box-like sequence upstream of smbA (Fig. 7a), suggesting that the promoter upstream of smbA may be regulated by the competence state of the cells. We also observed that the smb operon promoter region (upstream of smbM1), containing the transposon insertion site in mutant H1, was very highly homologous to the promoter region of the comC gene. The two sequences are very closely related, with over 93% identity (Fig. 7b). Based on the results of H2P Northern blot analysis and the reduction of the levels of transcription of smbA and smbB, these com box sequences likely are involved in the regulation of the two Smb structural genes.

FIG. 7.

Analysis of the promoter regions of Smb. (a) A putative com box is also present in the putative smbA promoter region. The consensus sequence is depicted in the bottom line. Consensus −10 and −25 elements are in boldface and underlined. The initial start codon of the downstream open reading frame (smbA) is underlined. (b) DNA sequence of the promoter regions of the smb and comC genes. The top sequence is that of comC and the comC promoter region (sequences are the complement of the sense strands). The bottom sequence is that of smbM1 and its promoter region. The boxes indicate identical sequences in both of the promoter regions. Underlined is the com box-like structure in the smbM1 promoter region. The transposon insertion site in mutant H1 is denoted by the arrow.

DISCUSSION

In the present study, we identified an S. mutans GS5 unique bacteriocin locus, smb, following Tn916 mutagenesis. One of the resulting transformants, H1, exhibited a defective phenotype for the production of Smb and was isolated following screening of approximately 8,000 Tn916 transformants. Sequence analysis of wild-type GS5 and mutant H1 demonstrated that Tn916 inserted into the upstream promoter region of the smb operon. The smb operon consists of seven putative open reading frames: smbM1, smbF, smbT, smbM2, smbG, smbA, and smbB. A GenBank search for similar proteins revealed that all of the proteins encoded within the smb operon are similar to several lantibiotic bacteriocin components. Northern blots and RT-PCR analyses confirmed its operon structure (Fig. 5 and data not shown). The operon arrangement of the genes for modification, transport, and immunity is a common feature of lantibiotic gene clusters. These results suggested that Smb belongs to the lantibiotic bacteriocin family (class I bacteriocin).

A novel finding of this study was that Smb is apparently a dipeptide antimicrobial complex. The two peptides SmbA and SmbB are encoded by two genes in a single operon. Sequence examination of smbA and smbB demonstrated features consistent with structural genes encoding bacteriocin prepropeptides. In particular, SmbA possesses a Gly-Gly sequence motif that is known to immediately precede the cleavage site in several bacteriocins (47). In SmbB, the presence of Gly-Ala may play a similar role (7). A GenBank search for similar peptides revealed that these two peptides have homologies with lantibiotic peptides; SmbA was similar to lacticinA2 encoded by L. lactis, and SmbB was homologous to mersacidin-like peptides expressed by B. halodurans and L. lactis. Inactivation of the smbA or smbB gene resulted in marked attenuation of the inhibitory effects on indicator strain RP66. That the products of both genes are required for bacteriocin activity was further suggested by the observation that inactivation of the smbA gene did not interfere with transcription of the smbB gene, ruling out possible polar effects in the smbA mutant (data not shown). However, we did not obtain direct evidence that Smb is a dipeptide. Nevertheless, these data show that both SmbA and SmbB are required for bacteriocin activity and that either the active bacteriocin is a dipeptide or each peptide acts in a synergistic manner to produce inhibition against the indicator strain RP66. Purification and chemical characterization of the active bacteriocin will be required to resolve this issue.

Recently, several dipeptide lantibiotics, including cytolysin produced by E. faecalis (10), staphylococcin C55 produced by S. aureus C55 (25), and lacticin 3147 produced by L. lactis subsp. lactis DPC3147 (39, 40), have been identified. It has been observed for these bacteriocins that equivalent amounts of both peptides are required for an interaction with target cells. This observation suggests that both SmbA and SmbB may be required at equivalent levels for bacteriocin activity. The Smb operon also appears to contain two modification genes. The biosynthesis of lantibiotics involves several posttranslational modification steps (5, 7, 41). Following translation of the smb transcript into prepropeptides, these products must be modified. The observation of the presence of two putative modification genes (smbM1 and smbM2) and two structural genes (smbA and smbB) within the smb gene cluster suggests several possibilities: (i) both propeptides may be modified sequentially by both modification enzymes, (ii) either one of the modification enzymes may alter both propeptides (with the second modification protein being redundant), and (iii) each propeptide may be modified by one specific modification enzyme. In the smb operon, a gene coding for a potential transporter (smbT) is also present in the smb operon. Typically, the bacteriocin-encoding genes are processed and secreted out of the cell via a dedicated transporter, which is a typical feature of most class I and class II bacteriocins (6). In most of the bacteriocin operons described to date, a gene encoding an immunity protein which protects the producing bacteria against autotoxicity is usually located downstream of the bacteriocin structural genes (26). The bacteriocin- and immunity protein-encoding genes are generally cotranscribed to ensure that the producer strain is not killed by its own bacteriocin (26). For the strain GS5 bacteriocin, smbF and smbG are good candidates to encode Smb immunity proteins. Significant homology among immunity proteins in the bacterial databases has been observed. For example, SmbF showed homology with the immunity proteins SpaF and MutF (36, 45), while SmbG exhibited similarity with the PlnG immunity protein (8).

Recently, it was observed that a GS5 comC mutant, CC1301 (52), was attenuated in Smb expression (49a). Addition of synthetic CSP was able to restore Smb production to the comC mutant. S. mutans uses a typical gram-positive CSP secretion and detection system for quorum sensing which affects several physiological properties (21). Interestingly, the promoter regions of both comC and the smb operon share similar sequences. The H1 and H2P mutants are disrupted by insertion of transposon Tn916 or the Erm cassette, respectively, resulting in a reduction in Smb expression. This decrease in Smb production results from a reduction in the transcription of the smb operon, including the smbA and smbB genes. However the mechanism by which these identical promoter sequences are involved in the transcription of smbA and smbB is still unknown. A recent report suggests that ComE regulated comC expression directly by interacting with the major RNA polymerase and the direct repeats in the comC promoter region. Based on these reports, Smb production may be directly regulated by ComE. The smb promoter region, upstream of smbM1, has a com box-like structure present at the end of the common sequence of both the comC and smb promoter regions. The smb locus may have originally been inserted into the strain GS5 chromosome on an insertion element (direct repeats flank the smb locus, and sequences similar to those of transposase genes are found directly downstream of this locus [data not shown]). Therefore, it is possible that the insertion sequence element containing the smb locus may have been inserted near the com locus originally and may have excised with the promoter region of this regulatory locus. In addition, the fact that the G+C content of the S. mutans genome is 37% while the putative inserted region containing the Smb exhibits a G+C content of 32% suggests that this region in the chromosome might have been imported from another bacterium. Several lantibiotic operons are also known to be regulated by quorum-sensing systems (3, 34). However, to our knowledge, this is the first demonstration that CSP regulates the expression of a bacteriocin in S. mutans.

In summary, we have identified the genes for a putative dipeptide lantibiotic-type bacteriocin produced by S. mutans strain GS5. Other strains of this organism, including strain BM71, also appear to produce the same bacteriocin (unpublished results). Smb is unique and distinct from the other previously characterized bacteriocins, mutacins I to IV, produced by other S. mutans strains and characterized by Caufield's group (3, 34, 35, 36). In addition, we have determined that strain GS5 also produces a nonlantibiotic mutacin IV bacteriocin (data not shown). Thus, bacteriocin production may be used by S. mutans as a means to compete with other oral bacteria present in dental plaque. However, the in vivo role of the S. mutans bacteriocins in dental caries formation still remains to be determined.