INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Group A streptococcus (GAS), specifically Streptococcus pyogenes, is a common cause of pharyngitis, impetigo, and many other human respiratory tract and soft tissue infections. Recently, there has been a dramatic increase in reports of severe invasive GAS infections, including necrotizing fasciitis and toxic shock syndrome (44). Although GAS produces a wide array of virulence factors, those responsible for the rapid bacterial spread and tissue injury seen with invasive GAS infections are unknown.

A hallmark feature of GAS in the clinical laboratory is the zone of beta-hemolysis observed surrounding colonies grown on the surface of blood agar media. The factor responsible for the beta-hemolytic phenotype is streptolysin S (SLS), the oxygen-stable cytolytic toxin of GAS. Despite detailed investigations over several decades, the exact chemical nature of SLS has remained a great mystery of GAS biology. SLS can exist in intracellular, cell-surface-bound, and extracellular forms (16), but attempts at purification are complicated by instability of the cytolytic activity in the absence of high-molecular-weight carriers, such as yeast RNA core or albumin (47).

We have adopted a molecular genetic approach towards the identification of SLS and the study of its contribution to disease pathogenesis. Recently, we reported that Tn916 mutagenesis of two GAS clinical isolates yielded mutants with an SLS-negative phenotype (8). The single transposon insertion in each mutant mapped to a promoter motif just upstream of a unique open reading frame (ORF), encoding a 53-amino-acid (aa) peptide, which we named sagA (SLS-associated gene). In contrast to the parent strains, these SLS-deficient transposon mutants were incapable of inducing necrotic lesions in a mouse model of subcutaneous infection, establishing SLS as a potential GAS virulence factor.

In the present study, we have used Tn917 mutagenesis, chromosomal walking steps, and data from the University of Oklahoma GAS genome sequencing project (www.genome.ou.edu /strep.html) to discover eight additional ORFs (sagB to sagI) situated immediately downstream of sagA in an operon structure. Targeted plasmid integrational mutagenesis was used to probe the requirement of each gene for SLS production and to define the functional boundaries of the sag locus. A recombinant plasmid containing the nine-gene sag operon was sufficient for complementation and heterologous expression of the SLS phenotype in Lactococcus lactis. Sequence analysis of the sag locus gene products suggests that SLS is related to the bacteriocin family of microbial toxins, which are genetically encoded by an operon including the structural (prepropeptide) gene and genes required for the chemical modification, processing, and export of the mature form.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, culture conditions, and transformation. Bacterial strains used in this study are listed in Table 1. GAS strain NZ131 is a T14/M49 isolate from a patient with acute poststreptococcal glomerulonephritis that produces SLS and streptolysin O (SLO), streptokinase, DNase, and pyrogenic exotoxin (SPE) B (42). GAS were grown in Todd-Hewitt broth (THB) or on Trypticase soy agar + 5% sheep's blood (SBA). For antibiotic selection, 2 µg of erythromycin (EM) or chloramphenicol (CM) per ml was added to the media. Escherichia coli strains DH5 and DH10B were grown in Luria-Bertani broth or on Luria-Bertani agar; antibiotic selection utilized 100 µg of ampicillin (AMP) per ml, 130 µg of spectinomycin (SPC) per ml, 300 µg of EM per ml, or 5 µg of CM per ml. L. lactis NZ900 was grown in M17 broth (Difco) supplemented with 1% glucose (GM17) or on SBA plates, with selection at 5 µg of CM per ml. GAS strains were rendered competent for transformation by growth in THB media plus 0.3% glycine as described for Streptococcus agalactiae (13). L. lactis was made competent for transformation by growth in GM17 plus 2.5% glycine (22). Plasmids were introduced by electroporation (Eppendorf 2510, 1,500 V), with recovery in THB plus 0.25 M sucrose (GAS) or GM17 media plus 20 mM MgCl₂ plus 2 mM CaCl₂ (L. lactis) for 1 to 2 h prior to antibiotic selection on agar media.

Tn917 mutagenesis and cloning of transposon insertion site. Mutagenesis of NZ131 with Tn917 was performed as described for Streptococcus mutans, by using the temperature-sensitive delivery vector pTV₁OK (20). The chromosome junction site of the Tn917 insertion in the single nonhemolytic mutant discovered on SBA screening was cloned as a HindIII fragment encompassing the transposon erm gene into pUC19, selecting for Amp^r and Em^r in the rpsL mutant E. coli strain DH10B. The resultant plasmid was named pSLSneg.

Chromosome walking steps. A 484-bp PCR fragment was amplified from within sagB, was digested with EcoRV and HindIII, and was cloned into XmnI/HindIII-digested pACYC184, resulting in pSagB'. Subsequently, the Spc^r determinant from pFW12 was cloned into EcoRI/XmnI-digested pSagB' as a compatible HincII-EcoRI fragment, resulting in pSagB-spc. NZ131 was transformed with the nonreplicative plasmid pSagB-spc, and Spc^r colonies exhibited an SLS-negative phenotype. Chromosomal DNA was isolated from the NZ131:pSagB-spc mutant, digested with EcoRI, self-ligated, and used to transform E. coli strain DH10B to Spc^r. This step resulted in the cloning of ~7,600 bp of additional DNA upstream of sagB (pSLSup). To complete sagH and identify downstream sequence, NZ131:sagA.Erm (see below) chromosomal DNA was partially digested with BclI, was self-ligated, and was used to transform E. coli DH10B to Erm^r. The resulting plasmid purified from an E. coli transformant was termed pSLSdown.

Southern analysis. Total cellular DNA was isolated (31) from NZ131 and 11 additional GAS strains representing a spectrum of emm types (4). The genotyped strains, obtained from recent epidemiologic investigations at the Centers for Disease Control and Prevention were 2109/98(st2974), 4356/99(emm3), 1603/99(emm76), 1143-99(emm11), 159-99(emm28), 134/98(st2967), 4315/99(emm1), 1489/99(emm75), 317/98(emm59), 1630/99(pt4245), and 3466/99(emm5). DNA was digested to completion with SpeI, was separated by 0.6% Tris-acetate-EDTA (TAE)-agarose gel electrophoresis, and was transferred to a nylon filter by the Southern method. A digoxigenin-labeled probe of a PCR amplicon comprising sagD to sagF was prepared by using a Genius Kit (Boehringer Mannheim). The probe was hybridized to the target filter under standard conditions, and the probe-positive bands were visualized by chemiluminescence.

Targeted plasmid integrational mutagenesis. PCR with specific primer pairs (Table 2) was used to amplify intragenic fragments from sag genes, as well as fragments from upstream of the sagA promoter and downstream of the terminator following sagI. PCR products were recovered by T-A cloning in the vector pCR2.1-TOPO (Invitrogen) and then were cloned by HindIII/XbaI digestion in the temperature-sensitive vector pVE6007 (34). The resultant knockout plasmids (Table 1) were introduced into NZ131 by electroporation, and Cm^r transformants were identified at the permissive temperature for plasmid replication (30°C). Single-crossover Campbell-type chromosomal insertions were selected by shifting to the nonpermissive temperature (37°C) while maintaining CM selection (52). SLS phenotype was determined on SBA plus CM at 37°C. The fidelity of site-directed recombination events was confirmed by PCR. Confirmed intragenic plasmid integrational mutants were designated NZ131:sagA.KO, NZ131:sagB.KO, etc. (Fig. 1). The targeted integrations upstream and downstream of the sag locus were designated NZ131:sagUp.KO and NZ131:sagDown.KO, respectively. An SLS-negative allelic replacement mutant of sagA in NZ131 was also constructed. In this mutant, NZ131.sagA::Erm, the sagA gene was replaced with a sagA derivative containing the nonreplicative plasmid bearing Em^r, pFW15 (U50977), inserted in the SpeI site.

View larger version (14K): [in this window] [in a new window]   FIG. 1.   Map of the sag genetic locus required for SLS production by GAS. Loci targeted for mutagenesis are also shown.

Transcriptional analysis. For Northern blots, a 30-ml culture of GAS was grown for 2 to 4 h post-mid-log phase (optical density at 600 nm, 0.8 to 1.2). Bacteria were pelleted and resuspended in Trizol reagent (Gibco) and were transferred to Fast RNA tubes with matrix (BIO101), and RNA was isolated by using the FASTPREP machine (BIO101) for 23 s at a speed setting of 6.0. RNA (3 or 15 µg) was electrophoresed on a 1.2% formaldehyde denaturing gel and was transferred to a nylon membrane (Hybond N⁺; Amersham) by standard methods. Membranes were incubated with digoxigenin-labeled (Boehringer Mannheim) sagA or sagB probe overnight at 50°C. Chemiluminescent detection was performed by using an anti-digoxigenin-alkaline phosphatase conjugate and autoradiography according to the manufacturer's instructions.

Complementation analysis and heterologous expression of SLS in L. lactis. A 9,440-bp amplicon comprising the entire nine-gene sag locus was amplified by using the high fidelity Expand Long Template PCR System (Boehringer Mannheim) according to the manufacturer's instructions. The amplicon sequence, from ~600 bp upstream of the sagA promoter motif to ~300 bp downstream of the terminator motif behind sagI, was captured by T-A cloning in the vector pCR2.1-TOPO (Invitrogen). The complete sag locus DNA was then directionally cloned as a BamHI/ApaI fragment into the shuttle vector pDC123 (9) cut with the same enzymes. The resultant plasmid, pSagLocus, was used to transform the nonhemolytic sagA allelic exchange mutant NZ131.sagA::Erm, with selection for Em^r and Cm^r. For the heterologous expression study, pSagLocus was used to transform nonhemolytic L. lactis NZ9000 with selection for Cm^r. In each case, control transformations were performed with vector pDC123 alone.

Nucleotide sequence accession number. All new sequence data from the present study have been submitted to the DDBJ/EMBL/GenBank databases under the accession numbers AF067723, AF067649, and AF163682.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Tn917 mutagenesis and discovery of downstream sag genes. One of 1,200 Tn917 mutant colonies of GAS strain NZ131 exhibited an SLS-negative phenotype when screened on SBA. The Tn917 insertion site in this mutant was cloned in pUC19 to produce pSLSneg. Sequence analysis of pSLSneg with primers annealing to Tn917 and pUC19 revealed a 660-bp fragment of the NZ131 chromosome with 100% sequence identity to an ORF with a contiguous sequence (contig) from the ongoing M1 GAS genome project and served as a starting point for the discovery of the apparent sag operon.

Chromosome walking and DNA sequence linkages yield extended sag locus. The complete sag locus was deduced from a compilation of sources: (i) sequencing of the previously reported 3.8-kb HindIII fragment from the M1 GAS strain MGAS166 containing sagA (8); (ii) sequencing of pSLSneg containing the chromosomal insertion site of Tn917 (within sagC) in the SLS-negative NZ131 mutant described above; (iii) chromosomal walking steps in which plasmid recovery from SLS-negative Campbell insertion mutants captured additional sequence information; and, importantly, (iv) discovery of homology with two contigs in the ongoing M1 GAS sequencing project at the University of Oklahoma. All sag ORFs identified from the M1 sequencing project data were confirmed to be present in M49 GAS strain NZ131 by PCR amplification and the targeted mutagenesis experiments described below.

Sequence analysis of the sag locus genes. Figure 1 shows the genetic structure of the nine-gene sag locus, including promoter and terminator motifs, while Table 3 lists the ORF size, predicted protein mass, PSORT prediction, and GenBank homologies of each identified gene. The sagA ORF encodes a 53-aa product with sequence features suggestive of a bacteriocin precursor (prepropeptide). These features include a potential glycine-glycine cleavage site, which would yield leader peptide (23 aa) and propeptide (30 aa) of characteristic size and an abundance of target residues (cysteine, serine, and threonine) known to undergo posttranslational modification. Figure 2 shows similarity between the leader sequences and the C-terminal propeptide regions of the predicted sagA gene product with the microcin B17 precursor (McbA) from E. coli. Microcin B17, a bacteriocin, is produced by an operon that includes modifying enzymes (McbB to -D) for posttranslational modification of propeptide amino acids and an ATP transporter secretion-immunity apparatus (McbE to -F) (53). The predicted protein product of sagB is a 36.0-kDa species likely localized to the cytoplasm that shares weak homology with McbB. A sequence possessing an inverted repeat stem of 17 bp (or 15 bp to allow a loop of 4 bp) followed by 9 nucleotides, 6 of which are Ts, is located between the sagA and sagB ORFs, consistent with a rho-independent terminator (GTAA TTAGCAGGTACTA...GA...TAGTACCTGCTAATTAC ...TATATGTTT).

View larger version (12K): [in this window] [in a new window]   FIG. 2.   Amino acid sequence similarity between SagA and McbA of E. coli.

The sag locus is conserved across GAS strains of differing serotypes. Recognition sites for restriction enzyme SpeI are found within sagA, twice within sagE, and downstream beyond the sagI terminator sequence (Fig. 1). Southern blot analysis was performed on SpeI-digested chromosomal DNA from NZ131 and 11 other GAS strains of differing emm (M protein gene) types (Fig. 3). When probed with a sagD to -F PCR amplicon, two hybridizing bands corresponding to the expected 4,602-bp (sagA to -E) and 4,182-bp (sagF to -I) fragments were invariably observed, showing that the sag locus is conserved among GAS.

View larger version (51K): [in this window] [in a new window]   FIG. 3.   Southern blot analysis showing the conservation of the sag locus among GAS isolates from a variety of emm (M protein) genotypes. Blot was probed with a digoxigenin-labeled PCR amplicon encompassing sagD to sagF.

Targeted mutagenesis of sag locus genes defines the functional boundaries of the sag locus. In order to (i) confirm the requirement of individual sag locus genes for SLS expression and (ii) define the functional boundaries of the sag locus, a series of targeted plasmid integrational mutants were constructed by using homologous recombination cassettes cloned into a temperature-sensitive suicide vector. Our earlier studies had shown that a random Tn916 integration that mapped within the sagA promoter resulted in an SLS-negative phenotype (8). In the present study, we found that a random Tn917 integration that mapped to the sagC ORF was also associated with loss of SLS expression. Figure 4 shows that targeted plasmid integrational mutagenesis of sagA, sagB, sagD, sagE, sagF, sagG, sagH, and sagI resulted in NZ131 mutants that invariably failed to produce detectable SLS activity. An SLS-negative double-crossover allelic exchange replacement of sagA (NZ131.sagA::Erm) was also constructed. Plasmid integrations 300 bp upstream of the sagA promoter (sagUp.KO) and 50 bp downstream of the apparent rho-independent terminator following sagI (sagDown.KO) retained the wild-type SLS activity of parent strain NZ131.

View larger version (159K): [in this window] [in a new window]   FIG. 4.   SLS phenotype following targeted knockouts of sag locus genes.

Transcription analysis reveals operon structure of GAS sag locus. Northern blot analysis (data not shown) revealed that parent strain NZ131 produced an abundant transcript of approximately 450 bp in size which hybridized with sagA. This transcript, corresponding to the size predicted from the sag promoter to the terminator motif between sagA and sagB, was absent in the SLS-deficient mutants NZ131:sagA.KO and NZ131.sagA::Erm. These findings were identical to our earlier observations with M1 GAS parent strains and SLS-negative Tn916 mutants of the sag promoter (8). Northern blot analysis of SLS-negative targeted mutants in downstream genes of the locus (sagB.KO through sagI.KO) all revealed wild-type levels of the 450-bp sagA transcript. Targeted plasmid integration upstream of the sagA promoter (sagUp.KO) also failed to abolish production of the sagA transcript. These findings confirmed the start of transcription at the sagA promoter and demonstrated that sagA transcript alone is insufficient to produce the SLS phenotype. Multiple attempts at detection of sagB transcript by Northern analysis of NZ131 and M1 GAS strains were unsuccessful, even when five times the amount of RNA preparation was analyzed. We have performed dot blot analysis on total RNA from NZ131, which yields very weak but detectable signals of equivalent intensity when probed with downstream genes (sagB, sagC, and sagG), compared to a very strong signal when probed with sagA (data not shown). These data suggested that mRNA for downstream genes in the sag locus was unstable or was produced in very low abundance.

View larger version (61K): [in this window] [in a new window]   FIG. 5.   (a) RT-PCR analysis of total RNA from NZ131. Transcript of predicted size is identified using a sagA forward primer and a sagB reverse primer, sagB forward primer and sagC reverse primer, etc. (b) RT-PCR analysis of total RNA from SLS-negative mutants NZ131:sagA.KO and NZ131:sagB.KO, demonstrating presence of transcript extending from the 3' end of the integrative vector (T7) to the next gene in the operon.

Reintroduction of the sag locus in trans restores hemolytic activity to SLS-negative mutant. A 9,440-bp PCR amplicon comprising the entire nine-gene sag locus was amplified and cloned in vector pDC123 to yield plasmid pSagLocus. On repeated experiments in which this construct was used to transform the nonhemolytic sagA allelic exchange mutant NZ131.sagA::Erm, wild-type levels of hemolytic activity were restored (Fig. 6). The pSagLocus-complemented mutants grew more slowly than mutants containing the vector pDC123 alone, suggesting some toxicity with overexpression of the sag locus gene products, many of which are predicted integral membrane proteins.

View larger version (82K): [in this window] [in a new window]   FIG. 6.   Restoration of SLS-activity to the nonhemolytic sagA allelic exchange mutant by reintroduction of the nine-gene sag operon on a plasmid vector.

Heterologous expression of the SLS phenotype in L. lactis. Plasmid pSagLocus did not confer a beta-hemolytic phenotype to the gram-negative E. coli used as an intermediary for cloning. However, when this plasmid was used to transform the nonpathogenic, nonhemolytic gram-positive coccus L. lactis NZ9000 (29), a robust and stable beta-hemolytic phenotype was observed in transformants (Fig. 7). The transformed L. lactis did not exhibit any other differences from the parent strain on the API-20 system (BioMerieux) for identification of streptococci and related lactic acid bacteria. The presence of the intact pSagLocus in the beta-hemolytic transformants was confirmed by plasmid purification, restriction analysis, and individual PCR amplification of all nine genes. Therefore, the sag locus is not only necessary for SLS activity in GAS, it is sufficient to confer the SLS phenotype to a heterologous, nonhemolytic bacterial species. These findings strongly suggest that the sag locus encodes the structural determinant for the SLS toxin and any additional genes required for its proper processing and export.

View larger version (162K): [in this window] [in a new window]   FIG. 7.   Plasmid pSagLocus confers beta-hemolytic phenotype to L. lactis.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

SLS is the much studied, yet poorly understood, oxygen-stable and nonimmunogenic beta-hemolysin elaborated by GAS. In addition to the classic lytic effect on sheep erythrocytes, SLS can damage other cell membranes, including those of lymphocytes (23), neutrophils, and platelets (15), certain tissue culture and tumor cells (46), and subcellular organelles such as lysosomes (7) and mitochondria (26). By weight, SLS is one of the most potent cytotoxins known (28). The SLS peptide has never been fully purified nor has the genetic basis of its production been previously elucidated.

In this paper, we present the identification and analysis of a genetic locus (sagA to sagI) required for SLS activity in GAS. These studies extend and confirm our earlier discovery of the sagA gene and its importance for the SLS phenotype (8). Targeted mutagenesis and complementation analysis were used to define the functional boundaries of the sag locus, and transcriptional analysis demonstrated its operon structure. Our heterologous expression of a beta-hemolytic phenotype in L. lactis demonstrated that genes of the sag operon are also sufficient for production of functional SLS. Beta-hemolysis was not observed in E. coli transformed with the identical sag operon construct. We hypothesize that use of a closely related bacterial species was needed for heterologous expression of SLS because of fundamental differences between the gram-positive and -negative cell walls and the prediction that a number of the sag locus genes including the ATP transporter apparatus are membrane-associated. There are no homologies between GAS sagA and -F genes and known L. lactis genes. However, L. lactis, like all prokaryotes and eukaryotes studied, possesses ABC-type transporters involved in translocation of substrates (e.g., ions, sugars, amino acids, peptides, and proteins) across membranes. Recent precedent for heterologous expression of a multigene streptococcal operon in L. lactis exists: introduction of the 13-gene exopolysaccharide synthesis cluster from Streptococcus thermophilus into L. lactis resulted in production of an altered polysaccharide capsule (45).

Our Southern analysis demonstrates that the sag locus is conserved among GAS strains regardless of serotype. Failure of targeted mutagenesis upstream of the sagA promoter and downstream of the sagI terminator to affect SLS production defines the functional boundaries of the locus. Inactivation of each and every gene in the locus results in an SLS-negative phenotype. The unique requirement of each gene in SLS biosynthesis cannot yet be absolutely confirmed, however, because of the possibility of polar effects. The integrative plasmid pVE6007 used for targeted mutagenesis does not contain translational or transcriptional stop signals (52). Moreover, each mutant was constructed with the cat gene of pVE6007 in the same orientation as the sag locus genes, such that transcription initiated from the cat promoter may be sufficient to transcribe downstream genes and diminish polar effects. Our RT-PCR results confirm some level of transcription of downstream genes in both the pVE6007 and allelic exchange knockouts of sagA.

Sequence homologies suggest that SLS is related to the bacteriocin class of antimicrobial peptides. The bacteriocins, including the colicins, microcins, lantibiotics, and nonlantibiotic bacteriocins, typically possess antimicrobial activity against closely related bacterial species and can sometimes exhibit broader hemolytic and cytolytic properties (10, 36). Examples include nisin of L. lactis (18), subtilin from B. subtilis (3), and the Enterococcus faecalis plasmid-encoded hemolysin-bacteriocin (14).

Bacteriocins are synthesized ribosomally as a prepropeptide, comprising an N-terminal leader sequence of 23 to 36 aa and a propeptide of 22 to 60 aa. The propeptide portion then undergoes specific posttranslational modification (e.g., dehydration) of amino acid residues, particularly serine, threonine, and glycine (25). These modified amino acids may then react with a neighboring cysteine residue to produce thioether bridges and unique cyclic structures (40). Following its modification, the prepropeptide is translocated to the cell surface in a process that requires an ABC export complex, at which time the leader peptide is proteolytically cleaved from the mature toxin at a specific sequence site (10).

Examination of sagA demonstrates features consistent with a structural gene encoding a bacteriocin-like prepropeptide. In particular, SagA possesses a Gly-Gly sequence motif that is known to immediately precede the cleavage site in several bacteriocins (48). Cleavage at this site would remove a 23-aa leader sequence and leave a 30-aa propeptide with a predicted molecular weight of 2.9 kDa. Koyama (28) calculated the molar ratio of SLS polypeptide to oligonucleotide in highly purified SLS-RNA core preparations to be 0.3. Gel filtration analysis of such preparations yielded a molecular mass of 12.0 kDa for the SLS-oligonucleotide complex, suggesting the polypeptide moiety is approximately 0.3 × 12,000 to 1.3 × 12,000 kDa or 2.8 kDa in size (6), which is consistent with the predicted size of the SagA propeptide.

The sequence of the 30-aa SagA propeptide region is unusually rich in threonine (four), serine (five), glycine (six), and cysteine (seven) residues, the precursors for posttranslational modification in several bacteriocin toxins. The presence of such unusual residues may help explain why amino acid analyses of partially purified SLS preparations (2, 28) have differed from the sequence predicted for the SagA propeptide (e.g., absence of cysteine). Previous attempts at sequencing SLS peptide by Edman degradations were unsuccessful (2), which may reflect cyclical thioether bridge structures or N-terminal blockage by a 2-oxobutyryl group as seen in other bacteriocins (35).

Despite the fact that all bacteriocins possess thioether rings and dehydrated residues, there is surprisingly little overall homology pattern among their propeptide sequences (41). SagA propeptide possesses the strongest homology (37% identity, 57% similarity) to McbA, the precursor of microcin B17 of E. coli. Microcin B17 is a unique bacteriocin peptide containing posttranslationally modified cysteine, serine, and glycine residues that form directly connected heteroaromatic ring structures (33). The microcin B17 genetic locus includes genes required for chemical modification of prepropeptide amino acids (mcbB, mcbC, and mcbD) and an ABC transporter secretion/immunity apparatus (mcbE and mcbF). Sequence homology of SagB with McbB, together with its predicted cytoplasmic localization, suggests SagB may be involved in posttranslational modification of amino acid residues in the SLS propeptide.

Bacteriocin peptides are typically exported across the bacterial cytoplasmic membrane by a dedicated ABC transporter (48). The ATP-binding cassette (ABC) transporter complex may sometimes carry out proteolytic processing and removal of the leader peptide concomitant with export of the toxin. Sequence homology indicates that sagG to sagI encodes the ABC transporter complex of the SLS production pathway, with SagG containing the ABC. It is interesting to speculate that SagG to SagI may also carry out a maturation protease function. Production of functional SLS by GAS has been shown to be inhibited by treatment with protease inhibitors such as tolylsulfonyl phenylalanyl chloromethyl ketone or phenylmethylsulfonyl fluoride, even in the presence of RNA core (1). However, due to relatively small size and lack of a conserved N-terminal segment, SagG itself (and McbF of E. coli to which it shares homology) do not appear to belong to the proteolytic GG leader cleaving group of ABC transporters (48).

The homology of SagA bacteriocin-like toxins suggests that SLS may possess antimicrobial activity. Preparations of stabilized SLS are toxic to bacterial protoplasts and spheroplasts (5) but have not to date been shown to inhibit cell-wall-competent bacteria. Certain GAS are known to produce bacteriocins, for example, the lantibiotic streptococcin A from strain FF-22 (24). The GAS M1 genome contains a close homologue of genetic apparatus for the lantibiotic salivaracin A (39). We have performed limited screening of targeted SLS-deficient mutants versus wild-type to probe for antimicrobial activity of the toxin, but we have yet to identify disparities, perhaps due to residual bacteriocin activity attributable to the above-mentioned bacteriocins (data not shown). Identification of potential antimicrobial properties of SLS may require future purification of the mature toxin.

The mechanism(s) whereby bacteriocin-producing bacteria themselves are immune to the antimicrobial effects of the mature toxin are not well understood (41). Genes involved in self-protection have been described in the gene clusters for nisin (29) and subtilin (27) biosynthesis. ABC transporter systems may also participate in bacteriocin immunity, including MbcE to -F of E. coli (27). SagE (25.4 kDa) is a possible membrane protein with weak homology to PlnP, a putative L. plantarum bacteriocin immunity protein (11). The best sequence match to residues 103 to 179 of SagE in the GenBank database is, in fact, PlnP.

Our discovery of a rho-independent terminator sequence downstream of sagA corresponds well with our previously reported 450-bp sagA transcript seen by Northern analysis (8). RT-PCR documents the continued presence of sagA transcript in downstream SLS-negative mutants (sagB.KO to sagI.KO), supporting the hypothesis that SagA requires the action of downstream gene products to become the mature SLS toxin. RT-PCR analysis showed that sagB to -I is transcribed as a polycistronic message along with sagA, but at a level too low for detection by Northern blotting. Thus, the terminator sequence following sagA allows some level of read-through. In several bacteriocin gene clusters examined to date, similar gene sequences have been observed downstream of the genes encoding the prebacteriocin, along with mRNA transcripts corresponding to termination at this site (29, 43). This regulatory mechanism, resulting in a differential abundance of mRNAs through leaky termination plus or minus the differential stability of resultant short and long transcripts, appears to be a common feature of bacteriocins (43). GAS may thus generate an excess of the SagA prepropeptide transcript without expending the biochemical resources required to produce equimolar amounts of each gene product in the biosynthesis and export pathway.

Recently, mutagenesis of a GAS response regulator identified as covR increased transcription of sagA, along with hasA (the first gene of the operon for capsule synthesis), ska (streptokinase), and speMF (mitogenic factor) (12). Mutagenesis of another two-component GAS regulator, CsR-CsrS, resulted in enhanced transcription of sagA and speB, the gene encoding pyrogenic exotoxin B (21). Thus, it appears that multiple gene repressor systems may exist for SLS and other potential GAS virulence factors. It is also possible that SLS itself is involved in GAS regulatory networks, as increased sagA expression from an inducible plasmid resulted in increased message levels for M protein (emm) in an M49 parent strain (32).

In summary, we have discovered and characterized by targeted mutagenesis, complementation, and heterologous expression a nine-gene operon required for SLS production by GAS. The sag operon is headed by a candidate gene for a bacteriocin prepropeptide, SagA, the likely SLS precursor. The sag operon also contains candidate genes for chemical modification of the bacteriocin propeptide and self protection, an ABC transporter for export and maturation proteolysis of the leader peptide, and an internal terminator motif for differential transcription of structural gene and accessory gene mRNAs. We have previously demonstrated the significance of SLS as a virulence determinant in an animal model of necrotic skin infection (8). The targeted isogenic SLS-negative mutants we have created will allow further study of cytotoxic properties of this molecule and its role in disease pathogenesis. In addition, the expression of SLS in a hemolysin-negative heterologous host will facilitate attempts at purification of the recombinant toxin, determination of its chemical structure, and elucidation of its mechanism of action.