|Home | About | Journals | Submit | Contact Us | Français|
Lantibiotics are antimicrobial peptides that have been the focus of much attention in recent years with a view to clinical, veterinary, and food applications. Although many lantibiotics are produced by food-grade bacteria or bacteria generally regarded as safe, some lantibiotics are produced by pathogens and, rather than contributing to food safety and/or health, add to the virulence potential of the producing strains. Indeed, genome sequencing has revealed the presence of genes apparently encoding a lantibiotic, designated Bsa (bacteriocin of Staphylococcus aureus), among clinical isolates of S. aureus and those associated with community-acquired methicillin-resistant S. aureus (MRSA) infections in particular. Here, we establish for the first time, through a combination of reverse genetics, mass spectrometry, and mutagenesis, that these genes encode a functional lantibiotic. We also reveal that Bsa is identical to the previously identified bacteriocin staphylococcin Au-26, produced by an S. aureus strain of vaginal origin. Our examination of MRSA isolates that produce the Panton-Valentine leukocidin demonstrates that many community-acquired S. aureus strains, and representatives of ST8 and ST80 in particular, are producers of Bsa. While possession of Bsa immunity genes does not significantly enhance resistance to the related lantibiotic gallidermin, the broad antimicrobial spectrum of Bsa strongly indicates that production of this bacteriocin confers a competitive ecological advantage on community-acquired S. aureus.
Staphylococcus aureus can be a human commensal bacterium, colonizing the skin and mucosal surfaces such as the nares, pharynx, and vagina in approximately 25 to 40% of the population. However, it is also a human pathogen that can cause epidemics of invasive disease. Genome sequencing of S. aureus strains has highlighted that the species is highly clonal, with approximately 78% of the genes being conserved and representing the core genome. The remaining 22% of the genes, which are variable and include those present on genomic islands, pathogenicity islands, prophages, integrated plasmids, and transposons, can in turn be regarded as an accessory genome (for a review, see reference 19) that provides a means via which S. aureus can evolve to adapt to particular niches and environmental pressures. The environmental pressure that has most strongly influenced S. aureus evolution in the past century has been the development and application of different antibiotics. These advancements have dictated that the strains that have flourished in hospitals, most notably hospital-acquired methicillin-resistant S. aureus (HA-MRSA) strains, tend to be multidrug resistant but suffer from a concomitant reduction in fitness relative to isolates from the community, due to being encumbered with staphylococcal cassette chromosome mec (SCCmec) types I to III and additional antibiotic resistance genes (48, 55). The negative consequences of this reduction in fitness are, however, mitigated by the reduction in competition from the human commensal microbiota by antibiotic exposure.
Since the late 1990s, MRSA infections have been detected among the general population and among healthy individuals (typically children and young adults) who lack traditional risk factors (26). It was apparent that the S. aureus strains responsible for these community-acquired MRSA (CA-MRSA) infections were genetically distinct from their HA counterparts, possessing the more simple type IV (and to a lesser extent, type V and VII) allelic versions of SCCmec (13, 55) and fewer antibiotic resistance genes (20). While this fact indicated that these strains might represent less of a health care challenge than the HA strains, it quickly became apparent that the enhanced competitiveness of these strains, resulting in rapid growth (CA-MRSA strains grow much faster than HA-MRSA strains) (4) and increased virulence (67) of CA-MRSA, meant that any delay in switching from the β-lactam antibiotics normally used to treat infections of unknown etiology could have very serious medical implications, including death. Indeed, paradoxically, CA-MRSA strains have since spread to hospitals and have been responsible for a number of infections.
In contrast to HA-MRSA strains, which by virtue of their multidrug-resistant nature, coupled with exposure to antibiotics, have a selective advantage over other microorganisms in the hospital environment, CA-MRSA strains, like commensal S. aureus strains, often face stiff competition from the natural flora of healthy individuals. It has been speculated that the production of an antimicrobial compound may provide CA-MRSA isolates with a competitive advantage in such environments (4, 14). The theory was first suggested when sequencing of strain FPR3757 (part of the virulent USA300 clonal group) revealed the presence of bsa (bacteriocin of S. aureus) genes, which resembled those associated with production of the epidermin subgroup of lantibiotics (2, 60). Lantibiotics are ribosomally produced, posttranslationally modified peptide antibiotics that are generally active against bacterial species which are closely related to the producing organism, and these antimicrobials are thought to have a role in niche competition in many natural environments (41). Lantibiotics have been the focus of much attention in recent years with a view to clinical, veterinary, and food applications (10, 72). Although many lantibiotics are produced by food-grade bacteria or bacteria generally regarded as safe, there have also been a few examples of lantibiotic production by pathogens (11, 46, 69). In this instance, despite the identification of the bsa genes, the production of a lantibiotic by CA-MRSA isolates has remained speculative. Indeed, to date, there has been only one confirmed example of a lantibiotic, i.e., staphylococcin C55 (46), produced by S. aureus and no definitive evidence that CA- (or HA)-MRSA strains produce such compounds. There is, however, some evidence to suggest that staphylococcin Au-26, which is produced by a vaginal isolate of S. aureus and has an inhibitory spectrum encompassing lactobacilli isolated from the endocervix and representative strains of Staphylococcus hominis, Staphylococcus warneri, Streptococcus pyogenes, Streptococcus salivarius, Streptococcus mutans, Lactococcus spp., and oral Neisseria spp., may also be a lantibiotic (63). Here, 17 years after its initial characterization, we have carried out a closer inspection of staphylococcin Au-26 and the associated producer and have established that the staphylococcin Au-26 and Bsa genetic loci are almost identical. Prompted by this finding, we employed a combination of mutagenesis and mass spectrometry (MS) to reveal that these genes are functional in a number of other staphylococci, including a large percentage of CA-MRSA isolates. We suggest that, as a consequence of eliminating competing human microbiota, this lantibiotic contributes strongly to the fitness of these community-associated isolates.
Bacterial strains, plasmids, and primers used in this study are listed in Tables Tables11 and and2.2. S. aureus strains were grown at 37°C in Mueller-Hinton broth/agar (Oxoid) or a blood agar-calcium carbonate mixture (BACa) comprising Columbia blood agar base (Difco, Sparks, MD) supplemented with 5% human blood and 0.1% calcium carbonate (Oxoid). Escherichia coli strains were grown in Luria-Bertani (LB) broth/agar at 37°C with aeration. Micrococcus luteus strains were grown in tryptic soy broth/agar (Merck) or BACa at 37°C with aeration unless otherwise stated. Lactobacillus strains were grown anaerobically on BACa at 37°C. Microbacterium oxydans DPC 6277 was grown aerobically in tryptic soy broth/agar at 30°C. Corynebacterium testudinoris, Staphylococcus epidermidis DPC 6293, and M. luteus DPC 6275 were grown aerobically in Mueller-Hinton broth/agar at 30°C. Psychrobacter sp. strain DPC 6277 was grown aerobically in brain heart infusion broth/agar (Oxoid) at 30°C. Antibiotics were used, where indicated, at the following concentrations: chloramphenicol, 10 μg ml−1 for E. coli and 10 μg ml−1 for S. aureus, and erythromycin, 250 μg ml−1 for E. coli and 5 μg ml−1 for S. aureus. X-Gal (5-bromo-4-chloro-3-indoyl-β-d-galactopyranoside) was used at a concentration of 40 μg ml−1.
Staphylococcin Au-26 was purified from cultures of S. aureus strain 26, grown in Columbia agar base-liquid medium (Difco) supplemented with 0.1% Tween 80, by using cation exchange on XAD-2 resin followed by reverse-phase high-performance liquid chromatography (HPLC) as described previously (63). Fractions separated during chromatography were assayed for biological activity by being spotted in 20-μl aliquots onto a freshly seeded lawn of M. luteus cells on Columbia blood agar. Plates were incubated for 18 h at 37°C and examined for zones of inhibition of M. luteus growth. Fractions displaying activity against M. luteus were examined for activity against a number of other genera, including strains of lactobacilli, and were analyzed using matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) tandem MS as described previously (74). Collision-induced dissociation (CID) fragment spectra were manually interpreted to gain sequence tag information for peptide identification. The amino acid derivatives dehydroalanine (Dha) and dehydrobutyrine (Dhb) were considered to represent sites of thioether bridge cleavages under CID conditions. For colony MS (CMS), bacteria were collected with sterile plastic loops and mixed with 50 μl of 70% isopropanol adjusted to pH 2 with HCl. The suspension was subjected to a vortex, the cells were spun down in a benchtop centrifuge at 14,000 rpm for 2 min, and the supernatant was removed for analysis. MS was performed with an Axima CFR plus MALDI-TOF mass spectrometer (Shimadzu Biotech, Manchester, United Kingdom). A 0.5-μl aliquot of matrix solution (α-cyano-4-hydroxycinnamic acid [CHCA] at 10 mg ml−1 in 50% acetonitrile-0.1% [vol/vol] trifluoroacetic acid) was placed onto the target and left for 1 to 2 min before being removed. The residue from the solution was then air dried, and the resulting sample solution was positioned onto a precoated sample spot. Matrix solution (0.5 μl) was added to the sample, and the sample was allowed to air dry and subsequently analyzed in positive-ion reflectron mode.
Bioactivity was assessed by well diffusion assays unless otherwise stated. These assays were carried out as described previously (70) or as follows. Molten agar was cooled to 48°C and seeded with the indicator of choice (~106 cells from a fresh overnight culture per ml). The inoculated medium was dispensed into sterile petri plates in 20-ml volumes, allowed to solidify, and dried. Wells (4.6 mm in diameter) were bored into the seeded agar plates, and 50 μl of an antimicrobial-containing sample was placed into each well before overnight incubation. Deferred antagonism assays were carried out as described previously (65) or as follows. Aliquots of 10 μl (2 × 108 cells per ml) of overnight cultures of Staphylococcus epidermis and Staphylococcus gallinarum were spotted onto petri dishes containing 20 ml of solidified LB agar, and the dishes were incubated overnight at 37°C. The resultant growth was killed by UV irradiation, and the plates were flooded with 20 ml molten agar containing 106 cells of the various CA-MRSA strains per ml. Plates were incubated at 37°C overnight.
PCR with degenerate primers was carried out using Hotmaster Taq polymerase (Eppendorf, Hamburg, Germany) in 50-μl reaction volumes containing 41.5 μl PCR-quality water, 5 μl of 10× buffer, 10 mM deoxynucleoside triphosphate mix (Roche Diagnostics Ltd., Lewes, England), 1 μl of each forward and reverse primer (primer stocks at 0.1 ng μl−1), and 0.5 μl Taq (5 U μl−1). Amplification was carried out with reaction conditions as follows: initial denaturation at 95°C for 2 min, followed by 30 cycles of 95°C for 30 s, annealing at 40°C for 30 s, and elongation at 65°C for 1 min, with final extension at 72°C for 5 min. PCR products were gel extracted with a Qiagen gel extraction kit, cloned into pGEM-T per the instructions of the manufacturer (Promega), and sequenced. Amplification of the entire staphylococcin Au-26 locus was carried out using a long-range PCR kit per the instructions of the manufacturer (Roche Diagnostics, Mannheim, Germany). Plasmid DNA was isolated from E. coli strains by using the High Pure plasmid isolation kit as recommended by the manufacturer (Roche Diagnostics, Mannhein, Germany). Total cell DNA was isolated using the High Pure PCR template preparation kit according to the recommendations of the manufacturer (Roche Diagnostics, Mannheim, Germany). The presence of bsa genes was assessed using the following primer pairs: BsaA1seqF and BsaBseqR, BsaEFsoeC and BsaEFsoeD, and BsaBsoeA and BsaBsoeB. E. coli EC10B was used as an intermediate cloning host for the plasmid pORI280. S. aureus cells were made electrocompetent by using the protocol outlined by Schenk and Laddaga (58). PCR was performed according to standard procedures using BioTaq DNA (Bioline), Vent polymerase (New England Biolabs), and KOD Hot Start DNA polymerase (Novagen) in a PTC-200 DNA engine (MJ Research). Colony PCR was implemented following lysis of cells in 10% IGEPAL 630 (Sigma-Aldrich) at 94°C for 10 min. Restriction digestions and DNA ligations were carried out according to established procedures using restriction enzymes PstI and EcoRI and T4 DNA ligase supplied by Roche Diagnostics. DNA sequencing was performed by MWG Biotech AG.
Multilocus sequence typing (MLST) of strain 26 was carried out using previously published protocols (18). Sequencing reactions were carried out using CEQ Dye terminator cycle sequencing with a Quick Start kit (Beckman Coulter, Buckinghamshire, United Kingdom). Cycle sequencing reactions were carried out in 10-μl (1/4-strength) reaction volumes containing 0.5 μl of a purified PCR product, 5 pmol of each primer, 1 μl of halfCEQ buffer (Genetix, Hampshire, United Kingdom), 2 μl of Dye terminator cycle sequencing Quick Start master mix (Beckman Coulter), and 6 μl of sterile water. Reaction conditions were as follows: 40 cycles of 96°C for 20 s and 60°C for 4 min. Reaction products were analyzed using a CEQ 8000 genetic analysis system, and raw data were analyzed using Sequencher software (Gene Codes Corporation, MI).
Two PCR fragments from the DNA flanking the bsaB gene were generated by using Vent polymerase (New England Biolabs) with the oligonucleotide pairs BsaBsoeA/BsaBsoeB and BsaBsoeC/BsaBsoeD. The products were mixed in a 1:1 ratio and combined by splicing by overlap extension (SOE)-PCR using the oligonucleotide pair BsaBsoeA/BsaBsoeD. The resultant product was digested with PstI and EcoRI and introduced into similarly digested pORI280 (RepA− shuttle vector), after which competent EC10B cells containing the RepA+ temperature-sensitive helper plasmid pVE6007 were transformed with the construct, designated pORI280bsaB, and transformants were selected on LB agar containing 250 μg ml−1 erythromycin and 40 μg ml−1 X-Gal. Plasmids were reisolated and introduced by electroporation into an S. aureus NCTC8325-4 derivative, RN4220, containing pVE6007. RN4220::pVE6007/pORI280bsaB was grown overnight at 30°C, subcultured twice at the nonpermissive temperature of 42°C, plated onto tryptic soy agar containing 5 μg/ml erythromycin at 42°C, and then subjected to selection for integration of pORI280 into the RN4220 genome. Plasmid integration was confirmed by PCR using primers pOri280R and deltabsaoutFor.
MICs were determined as described by Wiedemann et al. (71). CA-MRSA strains were grown in Mueller-Hinton broth (Oxoid). Serial twofold dilutions of the gallidermin peptide in Mueller-Hinton broth were prepared. Bacteria were added to give a final inoculum density of 105 CFU per ml in a volume of 0.2 ml. After incubation for 16 h at 37°C, the MIC was read as the lowest peptide concentration causing inhibition of visible growth.
Previous attempts to elucidate the amino acid sequence of staphylococcin Au-26 were hampered by the presence of a modified residue at position 2 of the peptide, which blocked N-terminal Edman degradation assays. It was established, however, that the first residue in this peptide (estimated to have a mass of 2,700 Da) is an isoleucine and that the peptide contains two or three lanthionine residues and thus is likely to be a lantibiotic (63). Given that isoleucine is also the first residue of the S. epidermidis-produced lantibiotic epidermin and of two putative epidermin-like peptides potentially encoded within the genome of a number of S. aureus isolates (corresponding to the theoretical Bsa lantibiotic), we speculated that staphylococcin Au-26 may also be an epidermin-like antimicrobial. To assess this possibility, we designed two degenerate primers, one (StaphLanF) designed on the basis of nucleotides conserved within epidermin-encoding genes and a second (StaphLanBR) designed to amplify a region conserved across genes encoding lantibiotic-modifying proteins, i.e., lanB homologues, in a manner analogous to that described previously (31, 73). A PCR product of approximately 450 bp was amplified, and DNA sequencing and BlastX searches against the nonredundant protein sequence database established that this amplicon was almost identical to corresponding regions from the type II υSaβ virulence islands of a number of sequenced genomes of S. aureus strains associated with the putative Bsa lantibiotic (Fig. (Fig.1A).1A). Further PCR amplification and partial sequencing were employed to better assess the extent to which this region of the νSaβ island in strain 26 corresponded to the type II versions of the islands from sequenced strains. Amplification with primers LeuDF (designed to amplify a sequence from the leukocidin-encoding gene lukD, located upstream of the Bsa locus in type II νSaβ islands) and StaphLanBR yielded an amplicon 2.043 kb in length, and sequencing revealed that this region displayed 99% identity to bsa gene sequences over the entire length (E value, 0.0) and potentially represented genetic determinants encoding two structural lantibiotic peptides and a partial LanB-encoding gene sequence, which were designated bsaA1Au-26, bsaA2Au-26, and bsaBAu-26, respectively (Fig. (Fig.1A).1A). PCR carried out with the primers LeuDF and Au26LocR, designed to amplify a region downstream of the bsa locus (Table (Table1),1), yielded a product of >10 kb in length. The length of this product, combined with sequencing of the 3′ ~700 bp of this fragment, confirmed that this was a portion of the S. aureus strain 26 νSaβ island. Given the high level of identity shared between the loci in this strain and other bsa gene-containing strains, sequencing of the remainder of the υSaβ island in S. aureus 26 was deemed unnecessary and a detailed in silico analysis of the sequence information available to date was undertaken.
As noted above, the nature of the genes present within the various Bsa-encoding loci indicates that the corresponding strains may produce lantibiotics that are related to the previously characterized lantibiotics epidermin (produced by S. epidermidis DSMZ 3095) (62) and epidermin′ and gallidermin (produced by S. gallinarum DSMZ 4616), which are active against other staphylococci, streptococci, and Propionibacterium acnes (the causative agent of acne) (34), as well as other skin bacteria (59), and have been the subject of a number of studies investigating their potential for clinical application (45) (Fig. (Fig.1).1). In fact, the bsa genes so closely resemble those associated with epidermin production that they have on occasion been incorrectly annotated or referred to as epi (epidermin) genes. bsa genes are located on the type II νSaβ genomic islands in strains MW2 (USA400; multilocus sequence type 1 [ST1]) and FPR3757 and USA300_TCH1516 (USA300; ST8), the laboratory strains NCTC8325-4 and Newman (ST8), and MSSA476 (a CA methicillin-sensitive strain that shares common ancestry with USA400 strains; ST1) (28). While the bsa genes are identical in these strains, related but not identical genes are also present in ET3-1, formerly known as RF122, a representative of a hypervirulent bovine mastitis clone (ST151) (27). In contrast, these genes are absent from the type I forms of the island found in N315 (an ST5 HA-MRSA strain) and Mu50 (an ST5 vancomycin-intermediate S. aureus [VISA] strain) (4, 6, 28) and in JH1 (a VISA strain; GenBank accession no. AAPK01000000) and JH9 (a VISA strain; GenBank accession no. AAPL00000000), both single-locus variants of ST5 (64), and are also absent from the type III island found in MRSA252 (an ST36 HA-MRSA strain). Thus, among genome-sequenced S. aureus strains, ST1, ST8, and ST151 strains possess the island and all ST5 and ST36 strains lack it. Notably, a fragment of one of the Bsa-associated genes, bsaG, corresponding to the 92 C-terminally located residues of the intact type II protein, is present in type I νSaβ islands, establishing that the bsa genes were lost from type I islands rather than acquired by type II islands (3). The chromosomal location of the bsa genes is not atypical in the context of epidermin-like determinants in that the gallidermin determinants are also situated on the chromosome of S. gallinarum (61); however, the location does contrast with that of the plasmid-associated epidermin genes (62).
The predicted prepropeptide and propeptide domains of staphylococcin Au-26 elements BsaA1Au-26 and BsaA2Au-26 share 83% identity (39 of 47 amino acids are identical) and 77% identity, respectively. The most notable difference between the two prepropeptides arises from a leucine start codon (TTG) in the gene encoding BsaA1Au-26. This is also the only difference between the BsaA1Au-26 and BsaA1 prepropeptides. UUG start codons have been reported to lead to reduced levels of expression in Gram-positive bacteria (1, 75). The predicted BsaA1Au-26 propeptide shares 86, 64, and 68% identity with the ET3-1 BsaA1 (BsaA1ET3-1), epidermin, and gallidermin propeptides, respectively, while the deduced 22-amino-acid sequence of the BsaA2Au-26 propeptide is identical to that of the BsaA2 propeptide and shares 81, 77, and 77% identity with those of the BsaA1ET3-1, gallidermin, and epidermin propeptides, respectively. Alignment of the propeptide forms of these, and other, epidermin-like peptides (Fig. (Fig.1B)1B) reveals that while 9 of the 10 C-terminally located amino acids are conserved in all such peptides (excluding mutacin I), the same is true for only 4 of the 12 N-terminally located residues.
Interestingly, BsaA1, BsaA1Au-26, and BsaA1ET3-1 do not possess the conserved CTPGC stretch of residues in the unmodified peptide, thought to be essential for binding of peptides of this nature to their target, i.e., the precursor of cell wall peptidoglycan, lipid II (29). All three contain an alternative stretch corresponding to CSFGC in the unmodified peptide. It should also be noted that the presence of two lantibiotic structural gene homologues is unusual in that epidermin and gallidermin producers possess only one such gene. Should these peptides be produced and undergo ring formation in a manner equivalent to that observed for epidermin and gallidermin, we predict the masses to be 2,281 Da for BsaA1 and 2,091 Da for BsaA2 (assuming all hydroxy-amino acids not involved in bridge formation are completely dehydrated). It should also be noted that while the annotation of the genome of FPR3757 (a USA300 CA-MRSA strain) (14) did not include a bsaA1 gene, we can confirm that it is indeed present (Fig. (Fig.1).1). Similarly, although annotation of the S. aureus USA300_TCH1516 genome suggests the existence of a hypothetical protein (USA300HOU_1816) encoded upstream of bsaA2, our inspection of this region has revealed the existence of a bsaA1 gene. Interestingly, in strain COL (a HA-MRSA strain which is related to NCTC8325-4 and is classified as ST250, i.e., an ST8-like sequence type [6, 23]), there is a divergently transcribed open reading frame (ORF; SACOL1879) located upstream of bsaA2, which is predicted to encode an IS1181 transposase (12). This finding indicates the ongoing evolution of the locus mediated by the transposition of mobile DNA. In this strain, bsaA1 is interrupted but the lukDE genes are present (Fig. (Fig.1A1A).
Of the other proteins potentially encoded within the genomic island, BsaB, BsaC, and BsaD correspond to those required for posttranslational modification of epidermin-like peptides, BsaP is predicted to be involved in leader cleavage/transport (2, 22, 36, 60), and BsaEFG are likely to serve as immunity proteins to protect the producing strain from that which it produces (49, 51). Our analysis reveals that the bsa genes in MW2, FPR3757, USA300_TCH1516, NCTC8325-4, Newman, and MSSA476 (which are 100% identical across these strains) are 88 to 97% identical to their ET3-1 equivalents (Fig. (Fig.1A).1A). The level of homology to the corresponding epidermin genes is lower (40 to 72% identity) (Fig. (Fig.1A),1A), and there are no bsa equivalents of epiH, epiT, and epiQ. During sequencing of ET3-1, an additional ORF located in this region, sab1684, was annotated (27). This ORF, which we have designated bsaX, is located between bsaA1ET3-1 and bsaA2ET3-1 but has the opposite orientation. Our analysis has revealed that, although not always annotated, such an ORF exists between bsaA1 and bsaA2 in the other genome-sequenced Bsa+ strains and that the predicted 63-amino-acid products BsaXET3-1 and BsaX differ by only one residue (Fig. (Fig.1C).1C). bsaX and bsaXET3-1 do not display a significant degree of homology to any other ORFs and are not present in epidermin- or gallidermin-producing strains, and the absence of an obvious ribosomal binding site means that it is not yet clear if these ORFs correspond to genes.
A number of approaches were taken to definitively confirm that the bsa and bsaAu-26 loci encode an active lantibiotic (Bsa/staphylococcin Au-26). The first strategy involved purification and investigation of the active staphylococcin Au-26 peptide. Cation-exchange chromatography analysis of ammonium sulfate-precipitated fractions from the broth supernatant of S. aureus strain 26 revealed a titer of 64 activity units/ml against M. luteus, a saprophyte often isolated from the skin of animals. Following HPLC separation using an acetonitrile gradient, an active fraction was eluted at approximately 60% acetonitrile. The spectrum of activity of this semipurified preparation corresponded to that of strain 26, thereby suggesting that, under these conditions, this strain produces a single inhibitory agent (M. Upton and J. R. Tagg, unpublished data). MS analysis identified three dominant peptide species in the range from 1,000 to 2,100 Da. One of the peptide species observed upon MS had a mass of 2,089 Da, which is very close to the predicted mass of 2,091 Da for BsaA2 (presuming a structure corresponding to that of epidermin and the dehydration of all remaining hydroxyl residues) (Fig. (Fig.1D).1D). Tandem MS analysis identified two sequence tags (Dhb-PGCAK/Q-Dhb-G and I/L-Dhb-Dha-H [due to the use of CID, Dha/Dhb represents either a dehydrated amino acid or a site of thioether bridge cleavage]) that matched the deduced amino acid sequence of BsaA2. The first of these sequence tags provided further evidence that the peptide detected corresponded to BsaA2 rather than BsaA1 (Fig. (Fig.1B).1B). Rigorous analysis of the MS spectra suggested that a peptide species corresponding to BsaA1 was not present in the HPLC-purified sample displaying inhibitory activity.
To determine if the bsa genes are active in other strains, the laboratory strain NCTC8325-4 was selected as a representative for further investigation. Deferred antagonism assays using a selection of skin-associated microorganisms, i.e., C. testudinoris DPC 6273, S. epidermidis DPC 6293, Psychrobacter sp. strain DPC 6277, M. luteus DPC 6275, M. oxydans DPC 6277 (44), and M. luteus NCIMB9278, as indicator strains established that NCTC8325-4 is also the producer of an antimicrobial agent with a broad spectrum of activity against competing skin microbiota (Fig. (Fig.2A).2A). M. luteus NCIMB9278 was again employed as an indicator to establish that strain ET3-1 also exhibited antimicrobial activity (Fig. (Fig.2B).2B). CMS analysis was carried out to determine if NCTC8325-4 produced peptides with masses corresponding to those predicted for BsaA1 and BsaA2. Although a peak corresponding closely to that predicted for BsaA2 was observed, no peaks corresponding to the predicted BsaA1 mass were apparent, indicating that, as in strain 26, only one of these peptides is produced. To provide a further link between antimicrobial production and the presence of bsa genes, we sought to disrupt the Bsa operon in strain RN4220, a readily transformable restriction-modification mutant derivative of NCTC8325-4 (35). This process involved the use of the pVE6007/pOri280 plasmid pair that was originally developed to generate mutants of Lactococcus lactis (9, 38) and has more recently also been applied to mutagenesis in other Gram-positive bacteria (43). The polar disruption of bsaB was targeted, as plasmid integration at this point would be expected to prevent lantibiotic biosynthesis. Thus, a 600-bp bsaB fragment was amplified and cloned into the RepA− vector pOri280, and the resultant construct was introduced into an RN4220 strain containing the temperature-sensitive RepA+ vector pVE6007. Following temperature upshift and loss of pVE6007, pORI280 integrants were identified by antibiotic selection and checked by PCR. The resultant mutant was designated RN4220::pOri280bsaB. The antimicrobial activities of the parental and mutant strains were assessed by deferred antagonism agar diffusion assays using M. luteus NCIMB9278. From these assays, it was apparent that disruption of the Bsa operon resulted in a strain that was unable to produce an antibiotic (Fig. (Fig.2C2C).
The CA-MRSA strains that have spread most rapidly belong to one of five clonal groups and are associated with specific STs in each case, i.e., ST1 (corresponding to the USA400 clonal group) (48, 66), ST8 (USA300 and USA500) (7, 15, 21, 25, 33), ST59 (USA1000) (30, 68), ST80 (Europe) (66), and ST30 (Pacific) (15, 48). Although not absolute (57), an association between the clinical spectrum of infections caused by CA-MRSA and the presence of Panton-Valentine leukocidin (PVL) genes, which code for the production of cytotoxins that cause tissue necrosis and leukocyte destruction, has been noted (24). When bsa genes were first identified, it was suggested that possession of these genes might represent an alternative distinctive feature of CA-MRSA isolates. However, as is evident from analyses of genome-sequenced strains (as described above) and comparative genomics studies, an exact correlation does not exist (8). Nonetheless, the production of the Bsa lantibiotic by CA-MRSA strains may be significant given the virulent nature of many of these isolates, in particular the PVL-positive (PVL+) forms due to their need to be more competitive than their HA-MRSA counterparts with respect to flourishing on the skin surface. We therefore set about investigating the distribution of bsa genes among a selection of 21 HA, health care-associated (HCA), and CA PVL+ isolates. PCR-based analysis established that eight of the isolates, M02/0088 (ST80), M04/0101 (ST80), M04/0266 (ST8), M05/0028 (ST8), M05/0060 (ST8), M05/0199 (ST8), M05/0259 (ST8), and M05/0267 (ST8), possessed the bsa genes (Table (Table3).3). Thus, both ST80 strains and all except one of seven ST8 strains tested possessed the genes (ML224 being the exceptional strain), whereas all strains of the ST5, ST22, ST30, and ST154 genotypes lacked the genes. Notably, during the present study it was determined that strain 26 is also a representative of the ST8 lineage and has a methicillin-sensitive phenotype. With respect to the MRSA isolates, it was also evident that only 1 (M02/0088) of 5 HA/HCA strains (20%) but 7 of 16 CA strains (44%) possessed the Bsa operon (Table (Table33).
Deferred antagonism antimicrobial assays, using NCIMB9278 as an indicator, were conducted with the 21 PVL+ strains. The results demonstrated a close correlation between the presence of the Bsa operon and antimicrobial activity, in that all except one of the eight bsa+ strains inhibited the indicator whereas none of the bsa-negative strains exhibited antimicrobial activity (Table (Table3).3). Perhaps significantly, the only bsa+ strain that did not exhibit antimicrobial activity was M02/0088, i.e., the sole HCA bsa+ strain. Thus, from an antimicrobial activity perspective, none of the HA/HCA strains exhibited activity whereas 44% of CA strains did. Finally, CMS analysis of all 21 strains was carried out. Here, a perfect correlation between antimicrobial activity and the presence of a peak corresponding to BsaA2, but not BsaA1, was apparent (Table (Table3;3; Fig. Fig.3).3). This peak was absent from the spectra for all strains that lacked antimicrobial activity.
In addition to the inhibition of competing microorganisms, another potential advantage associated with the presence of the Bsa operon may theoretically be the provision of immunity against other epidermin-like lantibiotics either produced by members of the human (or animal) microbiota or introduced in the form of a clinical therapeutic. All lantibiotic producers possess self-protective immunity mechanisms. One of the advantages associated with the application of lantibiotics for clinical use is that these immunity mechanisms are usually quite specific, which makes resistance, as a consequence of possessing homologous immunity genes, rare (for a review, see reference 17). Initially, deferred antagonism assays were carried out to investigate if Bsa-producing MRSA strains were more resistant than strains without Bsa to the activity of the related lantibiotics epidermin and gallidermin. Although the epidermin producer failed to inhibit the target strains, with the exception of strain 26, regardless of their bsa status (data not shown), the gallidermin producer effectively inhibited growth to various extents (Fig. (Fig.4).4). To more accurately assess gallidermin sensitivity, MIC determination assays were carried out with purified gallidermin (Table (Table3).3). Here again, sensitivity levels did vary within both the bsa+ and bsa-negative groups. However, when the bsa+ strains were examined in combination, it was apparent that they did not possess statistically enhanced resistance to gallidermin (F1,19 = 0.727; P > 0.05 by one-way analysis of variance). It would thus seem that BsaEFG does not provide protection against gallidermin and that this lantibiotic could be employed as an anti-CA-MRSA chemotherapeutic option. By extension, the lack of immunity against gallidermin suggests that no protection would be provided against other, more distantly related lantibiotics either.
Lantibiotics are antimicrobial peptides, many of which have potent and broad-ranging antimicrobial activities. In response to the ever-increasing emergence of antibiotic-resistant bacteria, many researchers have investigated the application of these compounds in clinical settings, with some very positive results (for a review, see Piper et al. ). One of the areas that has received most attention is the application of lantibiotics as anti-S. aureus agents, with nisin, lacticin 3147, Pep5, mersacidin, gallidermin, epidermin, and others having potential in this regard. However, the production of lantibiotics is not a feature associated solely with nonpathogenic Gram-positive bacteria. Although the focus of much less attention, a number of pathogens have also been shown to produce these compounds. It may be that production of various mutacins by S. mutans, streptin and streptococcins by S. pyogenes, and staphylococcin C55 by S. aureus, etc., could provide these microorganisms with a competitive advantage when colonizing/infecting a human host. In the most extreme case, that of cytolysin produced by enterococci, the lantibiotic is itself a cytolytic virulence factor (11).
From the findings of the present study, it is evident that the bsa genes initially identified during the course of S. aureus genome-sequencing projects do indeed encode a novel epidermin-like lantibiotic. The individual genes closely resemble those within the corresponding Epi and Gdm operons, but there are key differences in that no BsaH or BsaT homologues exist and, thus, export of the Bsa lantibiotic must progress in a manner that differs from that for epidermin and gallidermin. The presence of two structural peptide-encoding genes is particularly noteworthy. Although a number of two-peptide lantibiotics which are active by virtue of the combined activities of two quite different lanthionine-containing peptides exist (37), the benefits of possessing multiple highly homologous structural peptide-encoding genes is not evident. Despite this, a number of lantibiotic producers, especially among strains producing the type AII (lacticin-481-like) lantibiotics (e.g., ruminococcin A , mutacin K8 , streptococcin A-M49 , streptococcin A-FF22 , and macedocin ), carry multiple copies of homologous structural genes. This phenomenon has previously also been associated with the other epidermin-like peptides mutacin I (53), mutacin III (54), and mutacin B-Ny266 (5). In the case of mutacin B-Ny266, there is only 57.4% amino acid identity between the two putative structural gene products and there is no evidence for the transcription of the second structural gene. For mutacin I and mutacin III, insertional inactivation of the first structural gene abolished antimicrobial activity while inactivation of the second did not and, thus, the role of the latter gene has remained undetermined. Similarly, the significance of the presence of bsaA1 or bsaA1Au-26 remains unclear. Although our failure to detect a peptide with a mass corresponding to that of BsaA1 indicates that in this instance BsaA2 may be the sole significant peptide, further analysis will be required to confirm this conclusion unequivocally. The fact that BsaA1 was not detected in these analyses indicates that the peptide is not produced in a fully modified form, is present in very low abundance, or is not produced at all. Although the UUG start codon may have an impact on the production of BsaA1Au-26, it does not explain the apparent lack of production of BsaA1 by strains other than strain 26.
The production of a lantibiotic by clinical MRSA strains is potentially of great significance. Our investigations focused on pvl+ MRSA strains responsible for HA, HCA, and CA infections. From these studies, it was apparent that there was a particular association between strains with an ST8 or ST80 genotype and the presence of the bsa genes; it was also evident that although 44% of strains associated with CA infections possessed these genes, exhibited antimicrobial activity, and produced a BsaA2 peptide, none of the HA/HCA strains produced the antimicrobial. While a larger collection of strains will need to be assessed to determine definitively whether this finding represents a general association between Bsa production and CA strains in general or reflects a more specific association between ST8/ST80 strains and the lantibiotic-encoding genes, the high proportion of ST8 strains that produce the lantibiotic is particularly interesting in light of the extreme success of the USA300 (ST8) CA strain. It was also noteworthy that the sole bsa+ strain that did not produce the Bsa lantibiotic, M02/0088, was, despite being an ST80 strain, isolated from a patient with an HCA infection. While the basis for nonproduction remains undetermined, we have established that the disruption of bsaA1 through transposase insertion, akin to that observed in COL (ST8), has not occurred (K. M. Daly, unpublished data). It will be interesting to investigate M02/0088 in greater depth to determine if the absence of Bsa production is a reflection of its adaptation to the health care environment where, due to the presence of antibiotics, competition from other bacteria is greatly reduced. Although Bsa production by S. aureus COL has yet to be assessed, it may be that the presence of the IS1181 transposase also has an impact on the ability of this HA-MRSA isolate to produce a lantibiotic.
As the genes required for production of staphylococcin Au-26/Bsa are relatively widespread in staphylococci and it has now been established that these genes do indeed encode a functioning lantibiotic, it seems logical to assume that the producing strains must derive some associated benefit. Given that strain 26 was originally recovered from an endocervical environment, one could hypothesize that Au-26 plays a role in staphylococcal toxic shock syndrome. However, strain 26 was not associated with this disease and does not express toxic shock syndrome toxin 1 (63). In addition, an in silico survey of the prevalence of the tst gene reveals that only strain ET3-1 carries both the lantibiotic-encoding genes and tst. Thus, Bsa is unlikely to play a significant role in invasive S. aureus disease in the vagina. Instead, we propose that Bsa is important for the survival of staphylococci in the vagina, where antagonistic activity against competing commensal bacteria, including lactobacilli, would be beneficial (indeed, although perhaps a coincidence, the HCA bsa-positive strain, M02/0088, was originally isolated from an episiotomy wound specimen). Similarly, skin-associated CA-MRSA strains face competition from other members of the skin microbiota, including coagulase-negative staphylococci, Micrococcus spp., Corynebacterium and Brevibacterium, Propionibacterium spp. (P. acnes, P. avidum, and P. granulosum), Acinetobacter, and Streptococcus spp., as well as common transient species such as E. coli, Bacillus species, and Pseudomonas aeruginosa, while bovine mastitis-causing S. aureus strains, such as ET3-1, encounter an equally complex microbial challenge in order to establish an infection in the community setting, i.e., environments where exposure to antibiotics does not eliminate/reduce the competing microbiota. While production of an antimicrobial is of obvious advantage to a strain competing against a complex microbial consortium, the presence of the associated immunity proteins provides a less obvious, but potentially important, advantage. These immunity proteins may be present in lantibiotic producers themselves, although more recently, the phenomenon of immune mimicry (i.e., the provision of immunity as a consequence of possessing immunity gene homologues not associated with lantibiotic production) has also been reported (16). Given that the bsaE, bsaF, and bsaG genes are most closely related to the genes responsible for epidermin and gallidermin immunity, it was deemed most likely that cross-protection, if observed, would be against these lantibiotics. As the epidermin producer did not possess sufficient antimicrobial activity to inhibit any of the targets other than S. aureus strain 26, the possibility of cross-protection from gallidermin became the focus of greatest attention. Gallidermin has been shown in a number of studies to possess activity against skin-associated microorganisms, including P. acne (34), and is viewed as an antimicrobial with commercial potential. From our investigations, it was notable that despite homology between the BsaEFG and GdmEFG proteins, the presence of the Bsa operon did not confer a significantly higher level of resistance on the associated strains. Thus, while additional investigations using isogenic mutants containing nonpolar deletions of immunity genes or using indicators carrying heterologously expressed immunity determinants are required, it would seem that the presence of these genes does not preclude the use of gallidermin (or, presumably, more distantly related lantibiotics) as an antimicrobial to target/prevent CA-MRSA infection.
In conclusion, we have established that many S. aureus isolates from the community, including disease-causing antibiotic-resistant forms, produce an antimicrobial agent that enhances their ability to inhibit competing bacteria while potentially reducing their sensitivity to similar such compounds. However, now that the significance of the Bsa operon has been uncovered, attempts can be made to reduce the competitiveness of bsa+ strains by developing means by which production can be switched off or using other lantibiotic-producing members of the skin microbiota to control their expansion.
This work was supported in part by the Irish government under the National Development Plan, through a Science Foundation Ireland Investigator award to C.H., R.P.R., and P.D.C. (no. 06/IN.1/B98). S.K.S. was supported as a visiting researcher at the University of Otago by a laboratory fellowship grant from the Society for Applied Microbiology.
We dedicate this work to the memory of the late Julie Scott, who pioneered work with staphylococcin Au-26. We thank staff at the Otago Centre for Protein Research, Department of Biochemistry, University of Otago, for analysis of samples purified from S. aureus strain 26 and Alan Marsh for assisting in the optimization of peptide purification at University College Cork.
Published ahead of print on 18 December 2009.