High Frequency and Diversity of Antimicrobial Activities Produced by Nasal Staphylococcus Strains against Bacterial Competitors

The human nasal microbiota is highly variable and dynamic often enclosing major pathogens such as Staphylococcus aureus. The potential roles of bacteriocins or other mechanisms allowing certain bacterial clones to prevail in this nutrient-poor habitat have hardly been studied. Of 89 nasal Staphylococcus isolates, unexpectedly, the vast majority (84%) was found to produce antimicrobial substances in particular under habitat-specific stress conditions, such as iron limitation or exposure to hydrogen peroxide. Activity spectra were generally narrow but highly variable with activities against certain nasal members of the Actinobacteria, Proteobacteria, Firmicutes, or several groups of bacteria. Staphylococcus species and many other Firmicutes were insusceptible to most of the compounds. A representative bacteriocin was identified as a nukacin-related peptide whose inactivation reduced the capacity of the producer Staphylococcus epidermidis IVK45 to limit growth of other nasal bacteria. Of note, the bacteriocin genes were found on mobile genetic elements exhibiting signs of extensive horizontal gene transfer and rearrangements. Thus, continuously evolving bacteriocins appear to govern bacterial competition in the human nose and specific bacteriocins may become important agents for eradication of notorious opportunistic pathogens from human microbiota.


Introduction
The microbiomes of human body surfaces contribute to health and wellbeing in multiple ways [1,2]. At the same time, they represent major reservoirs for many human bacterial pathogens such as Staphylococcus aureus and members of the Enterococcaceae, Streptococcaceae, and Enterobacteriaceae. The composition of microbiota is highly variable between different human individuals and it changes over time [3]. Environmentally exposed microbiomes such as those of the human skin and airways are much more variable than those of the gut [4]. Changing environmental stressors, climate, personal hygiene, and contact to other persons contribute to the dynamic alteration of skin and airway microbiota [3,[5][6][7].
The success of individual bacterial strains in competition with other microbes is governed by multiple types of interaction [8][9][10][11]. The underlying mechanisms are complex and have been hardly explored. They may include different capabilities to bind to a limited number of host attachment sites, to take up and metabolize scant nutrients, or to produce antibacterial substances, referred to as bacteriocins, that inhibit competitors. Recent studies on the availability of nutrients in the human nose indicated that this habitat is extremely poor [12] compared e.g. with the gastrointestinal tract [13], which is constantly exposed to ingested food. Accordingly, the nasal microbiome is less dense and less diverse than the intestinal microbiome [5,14], and it is tempting to assume that bacterial strains that colonize the nose have efficient means to compete with other bacteria. [3,5].
Most of the Staphylococcus species found in the nose are either exclusive commensals (Staphylococcus hominis) or cause infections in hospitalized patients with indwelling devices (Staphylococcus epidermidis, Staphylococcus warneri, Staphylococcus lugdunensis Staphylococcus capitis, and others). In contrast, Staphylococcus aureus is not only an opportunistic but a highly aggressive pathogen that is able to infect also healthy humans. S. aureus is one of the most frequent causes of skin and soft-tissue infections as well as blood stream infections often resulting in endocarditis or sepsis [15,16]. Only one third of the human population is permanently colonized by S. aureus and carriers are at a higher risk of invasive S. aureus infections [17,18]. The preferred S. aureus niche is the anterior nares and nasal eradication of S. aureus with the antibiotic mupirocin is a valuable strategy to reduce the risk of endogenous infections in immunocompromised patients [19,20]. Why only certain persons are colonized and why this trait is stable over time has remained unknown. While host genetic factors have been thought to play a role in the carrier status [21], a recent twin cohort study suggests that other than genetic factors may be more important [22]. These may include mechanisms of microbial interference in the nose, but the identity and frequency of such processes have not been elucidated.
Bacteria found in the human nose or in other habitats have been occasionally reported to produce ribosomally synthesized antimicrobial peptides referred to as bacteriocins [23][24][25]. Bacteriocin prepeptides are often modified e.g. by formation of thioether rings in the case of lantibiotics, and producing strains have immunity genes providing resistance specifically to the cognate bacteriocin but not to other antimicrobial compounds [26]. Most of the known bacteriocins have rather narrow activity spectra, often against closely related bacteria, which may reflect their specific ecological roles but remains an obstacle for their use as antimicrobial agents [27]. Bacteriocin genes are usually encoded on mobile genetic elements (MGEs) such as plasmids and only rarely found in bacterial core genomes [23,24,28]. All these findings point to diverse and important roles of bacteriocins in bacterial interference and microbiota composition, but systematic analyses of bacteriocin abundance, diversity, and activity spectra in a given microbial niche have not been conducted so far.
In this paper, to examine the ability of nasal Staphylococcus strains to produce antimicrobial substances and substances' activity spectra, we tested 89 human nasal Staphylococcus isolates against a selection of nasal commensal bacteria. We further characterized the structure and capacity of a newly identified S. epidermidis bacteriocin with a specific activity spectrum, and evaluated its impact on other nasal commensals.

Results
A large number of nasal Staphylococcus strains produce antimicrobial substances, which differ largely in activity spectra To determine the frequency and diversity of bacteriocin production by nasal Staphylococcus species, and if their bacteriocins preferentially inhibit other nasal bacteria, we analyzed 89 Staphylococcus isolates (IVK strains) recovered from the noses of 37 healthy human volunteers. These strains belonged to six different species (S. epidermidis, S. aureus, S. hominis, S. lugdunensis, S. warneri and S. capitis). Whereas these species have been reported in the human nose in several independent studies [29][30][31], the collected strains may represent a random and not necessarily representative panel of Staphylococcus strains form the nasal microbiota. Eleven bacterial species, frequently found in human nasal microbiomes [5], were used as test strains for the identification of inhibition zones around IVK colonies. They comprised representatives from the Actinobacteria families Micrococcaceae (Micrococcus luteus, Rothia mucilaginosa), Corynebacteriaceae (Corynebacterium pseudodiphteriticum, Corynebacterium accolens), and Propionibacteriaceae (Propionibacterium acnes), Firmicutes families Staphylococcaceae (S. aureus, S. epidermidis), Carnobacteriaceae (Dolosigranulum pigrum), Streptococcaceae (Streptococcus pyogenes), and Proteobacteria families Moraxellaceae (Moraxella catarrhalis) and Pasteurellaceae (Haemophilus influenzae). Notably, production of antimicrobial compounds was a common and highly variable phenotype among the Staphylococcus strains tested (Figs 1 and S1) with 84% of them inhibiting at least one of the test strains. In total, 77 strains exhibited antimicrobial activity, but the various Staphylococcus isolates differed largely in the strains targeted by their antimicrobial substances. While the majority of S. epidermidis inhibited D. pigrum and M. catarrhalis, only some S. aureus were found to target these species. Conversely, the majority of S. aureus but only some S. epidermidis inhibited M. luteus. Whereas none of the S. aureus strains was able to inhibit C. pseudodiphteriticum, almost half of all S. epidermidis isolates showed this capacity. S. epidermidis was by far the most frequent bacteriocin-producing Staphylococcus species (96%), whereas only 69% of other CoNS and 52% of S. aureus isolates showed inhibitory activity ( Table 1).
The inhibition patterns of antimicrobial compounds from our bacteriocin-producing IVK strain panel were highly diverse and most had a rather narrow spectrum that covered one or two (49.4%) or three to four (46.8%) of the test strains. According to the panel of inhibited strains 24 different inhibition patterns were defined. Even though it cannot be excluded that some strains may differ in the amounts of the produced compounds or may produce more than one antimicrobial, these findings suggest that there is a remarkable variety of antimicrobials produced by nasal Staphylococcus strains. In order to elucidate if specific inhibitory patterns are associated with particular clonal lineages, the clonal identities of 19 S. epidermidis isolates with inhibitory patterns no. 4 to 7 and of all 19 S. aureus isolates were analyzed by multi-locus sequence typing (MLST) and spa typing, respectively. However, with only a few exceptions, the inhibitory patterns and the clonal identities did not correlate, suggesting that bacteriocin production is a very variable trait that may often depend on acquisition of MGEs (Fig 1, S1 and S2 Tables).
Staphylococcus strains encounter a wide range of different bacterial species in the nasal microbiota, and it is largely unclear, which bacteria may be major competitors and whose inhibition by a particular bacteriocin may provide a significant fitness benefit for the producer. Among the 11 test strains M. luteus, C. pseudodiphteriticum, D. pigrum, and M. catarrhalis were most frequently inhibited by bacteriocins from the majority of S. epidermidis and some of the S. aureus or other CoNS strains (Fig 1), suggesting that interfering with growth of these species may provide particular fitness benefits. Only sporadic activities were found against C. accolens, P. acnes, R. mucilaginosa, S. pyogens, or other Staphylococcus strains from the species S. epidermidis and S. aureus (Fig 1), but none of the IVK strains was able to inhibit H. influenzae. Several of the detected bacteriocins inhibited preferentially Proteobacteria (M. catharrhalis) or Actinobacteria species (e.g. M. luteus or C. pseudodiphteriticum) but many combined activities against certain Proteobacteria and Actinobacteria species. Inhibition of Firmicutes was generally rare except for D. pigrum, which was susceptible to many of the detected bacteriocins from S. epidermidis and other CoNS. Only a small minority of the IVKs (three strains) exhibited broad-spectrum antimicrobial activity against the majority of test strains. Such molecules may represent interesting candidates for new antibiotics.
While only one of the IVK strains inhibited wild-type S. aureus under standard conditions, this number increased to fourteen when an isogenic S. aureus mutant, lacking the D-alanine modification of teichoic acids, was used (ΔdltA; Fig 1, column L). Teichoic acid alanylation modifies the net charge of the cell envelope and leads to resistance to many cationic antimicrobials [32]. Thus, many bacteriocins can act against S. aureus, but these bacteria appear to be well protected by their intrinsic antimicrobial peptide resistance mechanisms.

Many bacteriocins are induced in Staphylococcus isolates by colonization-related stress conditions
Staphylococcus species have quite different life styles and habitats [15,18,33] raising the possibility that they may regulate bacteriocin production and express bacteriocin genes only under colonization-related conditions. In order to simulate stress conditions encountered in the human nose, we analyzed antimicrobial activities of the 89 IVK strains under conditions of iron limitation and in the presence of H 2 O 2 . Bacteria have been found to strongly up-regulate genes of iron acquisition [12,34] and to be exposed to H 2 O 2 released by Streptococcus strains or by phagocytes in the human nose [35][36][37]. M. luteus and S. aureus were chosen as representative test strains that were inhibited only by a minority or by almost no IVK strains, respectively (Fig 1), and because they grew well on test agar plates with iron limitation or H 2 O 2 . Addition of the iron chelator 2,2'-bipyridine at a final concentration of 200 μM to test agar plates increased IVK strains' antimicrobial activity against M. luteus by varying degrees; however, some IVK strains did not inhibit M. luteus, irrespective of bipyridine addition (Figs 2 and S2). In contrast, the addition of bipyridine did not substantially change IVK strains' antimicrobial activity against S. aureus, except for strain IVK28. The vast majority (82.5%) of the activities against M. luteus or S. aureus, were only detectable (32.5%) or substantially increased (52.5%) under conditions of iron limitation (Fig 2).
H 2 O 2 did not elicit bacteriocin production against M. luteus or S. aureus in any of the nonproducers but 18% of the activities active against M. luteus were increased in the presence of H 2 O 2 (Figs 2 and S2). Taken together, these data show that many nasal Staphylococcus strains regulate bacteriocin production and release antimicrobial substances only under conditions resembling their nasal habitat.

S. epidermidis strain IVK45 produces a nukacin-related lantibiotic
The highly diverse antimicrobial activities among nasal Staphylococcus isolates and the fact that the type of inhibition pattern did not correlate with clonal identity suggests that different bacteriocin genes may have been repeatedly introduced and exchanged by Staphylococcus strains. S. epidermidis strain IVK45 was chosen to elucidate the molecular basis of its bacteriocin production, because this strain showed strong antimicrobial activity against several unrelated bacteria (M. luteus, C. accolens, S. pyogenes, D. pigrum, M. catarrhalis; Fig 1) that was inducible by H 2 O 2 and iron limitation (Fig 2). The genes responsible for bacteriocin production were identified by sequencing of a transposon mutant lacking antimicrobial activity and found on a plasmid, subsequently named pIVK45, of 21.840 nucleotides (accession number KP702950).
The bacteriocin gene cluster exhibited similarity to genes required for biosynthesis of a ribosomally synthesized lanthionine-containing bacteriocin (lantibiotic). The highest similarity was found with genes previously described in S. warneri ISK-1 [38] and S. hominis KQU-131 [39] to be required for biosynthesis of the lantibiotic nukacin ( Fig 3A). The mature bacteriocin of S. epidermidis IVK45 was predicted to consist of 27 amino acids including a dehydrobutyrine, a 3-methyllanthionine, and two lanthionine residues, originating from threonine, cysteine, and serine, respectively (Figs 3B and S3). It differs from nukacin ISK-1 and KQU-131 by five and six amino acids, respectively, in the mature peptide and another five amino acids in the leader peptide (S3 Fig).
To further confirm its identity, the compound named nukacin IVK45 was purified from culture filtrates by a combination of adsorption, cation exchange, and size exclusion chromatography. The resulting active fraction contained only one notable peak in RP-HPLC chromatograms that was absent from the corresponding preparation of the nukacin-deficient mutant (S4A and S4B Fig). Mass spectrum (MS) analysis confirmed that the peak corresponds Nukacin IVK45 is encoded on a composite plasmid whose sequence suggests a history of extensive horizontal gene transfer and rearrangements The precursor sequence of nukacin IVK45 does not only align with those of nukacin ISK-1 and KQU-131 from S. warneri and S. hominis, respectively, but also with several hypothetical peptides from other bacterial genomes (S3 Fig). A peptide from Streptococcus agalactiae was particularly similar to nukacin IVK45, whereas peptides from other Firmicutes from the genera Streptococcus, Lactococcus, or the Actinobacteria species Kocuria were less related. Thus, lantibiotic genes appear to be exchanged even between bacteria from different orders and phyla. The nukacin IVK45 genes were located on an MGE with obvious composite structure. The operon was most similar to that of S. warneri ISK-1, but the location and orientation of the precursor peptide gene nukA was different (Fig 3A). The other parts of pIVK45 including the origin of replication, open reading frames encoding replication and hypothetical proteins have highly similar counterparts in plasmids or other MGEs from S. aureus, S. epidermidis, and S. lugdunensis (Fig 3C) underscoring a common DNA exchange between several Staphylococcus strains and species.
Nukacin IVK45 increases fitness of S. epidermidis in competition with microbial competitors In order to study how the production or absence of a bacteriocin may impact on the competitive capacities of nasal bacteria, a defined S. epidermidis IVK45 mutant, lacking the nukA gene, was constructed, which lacked antimicrobial activity as expected (Fig 4A). The plasmid pRB474-nukA was used for complementing the mutant with a wild-type copy of nukA expressed from a constitutive promoter. The complemented strain resumed production of antimicrobial activity thereby confirming that the nuk genes and their product are responsible for the bacteriocin activity of IVK45.
To further investigate how nukacin IVK45 may affect the growth of other nasal bacteria, M. luteus, C. accolens, and M. catarrhalis were cultivated in medium supplemented with culture filtrate of IVK45 wild type or the nukacin-deficient mutant (Fig 4B to 4D). The CFU per ml of M. luteus and C. accolens decreased almost 10,000-fold within 24 hours in cultures containing culture filtrates of IVK 45 wild type, while culture filtrates of the isogenic nukA mutant did not inhibit growth of the test strains. Increasing amounts of culture filtrate of IVK45 wild type showed dose-dependent inhibition of the test strains. Culture filtrate of IVK45 but not of the nukA mutant prevented also growth of M. catarrhalis but did not lead to a major reduction of viable cells even if nukacin IVK45 was strongly concentrated. Growth of nukacin-insensitive S. aureus Newman, which was included as a negative control, was not affected by IVK45 or IVK45 ΔnukA culture filtrate (Fig 4E). Co-cultivation of IVK45 wild type or ΔnukA with M. catarrhalis on agar plates demonstrated that nukacin IVK45 allows S. epidermidis to outcompete M. catarrhalis ( Fig 5). Thus, nukacin IVK45 is a bactericidal or bacteriostatic bacteriocin, depending on the target strain, with a strong capacity to interfere with growth of susceptible nasal bacteria.

Discussion
Bacteriocins have been described in many bacterial species, but their ecological significance has remained unclear [27,40]. Specifically, we are lacking insights into the frequency, diversity, regulation, and activity spectra of bacterial antimicrobial molecules in a given microbial niche. It is obvious that bacteriocins can augment bacterial fitness when producing strains reside in a complex microbiota such as those of human body surfaces. The poor availability of nutrients in the nose [12], compared to other human body surfaces, such as those of the gastrointestinal tract [13], may cause a particularly high competitive pressure and may be a reason why we found an unexpectedly high frequency and diversity of bacteriocins in nasal bacteria.
The effectiveness of a particular bacteriocin should depend on its capacity to restrict growth of most critical competitors of the producing strain. It is interesting to note that the investigated bacteriocins from Staphylococcus strains varied largely in activity spectra suggesting that this group of bacteria does not have a preferred competitor but may need to combat quite different bacteria in different hosts. The differences in activity spectra may reflect the enormous variability in composition of individual nasal microbial communities, which can be dominated by different bacterial genera such as Moraxella, Corynebacterium, Propionibacterium, or Staphylococcus [5,29,[41][42][43] depending on the individual host. On the other hand, the activity spectra of bacteriocins produced in a particular nose may as well contribute to the capacity of a given bacterial species to dominate the microbiota. If this hypothesis holds true, then there should be a chance for developing 'probiotic' strains that would support a favorable microbiota composition and could eradicate facultative pathogens such as S. aureus. To this end it will be crucial to associate microbiota composition and pathogen colonization with bacteriocin diversity in large cohorts of human volunteers in the future.
Individual nasal bacterial strains have been found to affect each other in specific ways. For instance, H 2 O 2 produced by Streptococcus strains, living in close proximity to Staphylococcus strains on human body surfaces [43], can induce prophages in S. aureus, which destroy Staphylococcus cells during the propagation process [37]. Our finding that H 2 O 2 also induces certain bacteriocins from Staphylococcus strains suggests that these bacteria can fight back and produce anti-Streptococcus compounds. There is evidence that an inverse correlation exists between Staphylococcaceae and Corynebacteriaceae, and it has been shown that high numbers of Corynebacterium species but also of of S. epidermidis can limit nasal colonization by S. aureus [43][44][45][46]. Our study suggests that this competition is governed at least in part by bacteriocins. Of note, only two of 77 inhibitory activities were found to be active against S. aureus, supporting the notion that this pathogen has particularly effective resistance mechanisms against antimicrobial peptides encoded by the dltABCD and mprF genes [47]. In contrast to the wild type, the ΔdltA mutant of S. aureus (Fig 1, panel L) was sensitive to compounds from 14 of the investigated strains including nukacin IVK45.Therefore, it is very likely that such compounds resemble cationic antimicrobial peptides (like nukacin IVK45), against which the ΔdltA mutant is most sensitive [32].
S. aureus nasal colonization depends on expression of several molecules, such as host ligand-binding surface proteins, teichoic acids, capsule, and proteases [18]. S. aureus is eradicated from the nose of risk patients, but increasing resistance rates against mupirocin, the antibiotic currently used for eradication, demand the development of new decolonization strategies [48]. Identifying and harnessing bacteriocins that act against S. aureus will be crucial for potential future strategies for selective eradication of S. aureus. It is interesting to note that nasal bacteria do not only inhibit each other but, in certain instances, appear also to depend on each other and support each other's growth [45]. Such growth-promoting capacities may be based on release of specific nutrients or on production of siderophores subducted by other bacteria that cannot produce but take up such iron-scavenging molecules. Thus, there are probably several other processes governing the fitness of bacteria in the nasal microbiome in addition to production of or resistance to bacteriocins. Such mechanisms may explain why 16% of the nasal Staphylococcus isolates from this study can colonize the nose without producing bacteriocins.
The observed diversity of inhibition patterns of nasal bacteriocins implies that there is a substantial pool of bacteriocin-encoding MGEs that are frequently rearranged and horizontally transferred or exchanged between different nasal strains. In line with this assumption most bacteriocins are encoded on plasmids [23,24,28] that can be easily exchanged either by transducing phages or by conjugative elements [49][50][51]. Such horizontal gene transfer events can be observed across the boundaries of species and genera and are directed e.g. by the capacity of a transducing phage to bind to wall teichoic acids in donor and recipient strains [52]. Transposases and recombinases found on pIVK45 and many other MGEs [53,54] may contribute to recombination events that continuously create new bacteriocin variants with altered activity spectra. Nukacin IVK45 may be regarded as a paradigm for bacteriocin evolution: It is encoded together with several gene fragments of transposases, recombinases, IS-and IS-like elements on a mosaic plasmid composed of DNA segments probably originating from many different bacterial species in combination with other MGEs and DNA fragments. It differs from the first discovered nukacin ISK-1 from S. warneri in five amino acid positions in the mature peptide, which may have altered its activity spectrum. Along this line, nukacin ISK-1 has been described to be bacteriostatic for S. aureus and other Gram-positive and bactericidal for M. luteus and Staphylococcus simulans but inactive against Gram-negative species [55]. In contrast, nukacin IVK45 was bacteriostatic for the Gram-negative M. catarrhalis, bactericidal for the Gram-positive M. luteus, but inactive against S. aureus. Nukacin is proposed to target the peptidoglycan precursor lipid II but may have additional modes of action as shown for other lantibiotics [56].
Our study should stimulate more extensive research on the ecological roles of bacteriocins and other fitness traits in human microbiomes. It will be particularly interesting to see how frequent and diverse bacteriocin production may be in more complex microbiota such as those of the human ileum and colon. Among oral Streptococcus species only 16% have been found to produce bacteriocins [57], mainly active against closely related Streptococcus pneumoniae, suggesting that bacteriocins may have a less important ecological role in upper gastrointestinal compared to upper respiratory microbiomes. Bacteriocins, acting against S. aureus or other endogenous facultative pathogens, and the bacteriocin-producing strains may become attractive agents for future selective decolonization strategies. The nukacin-related lantibiotic mersacidin has been successfully used to completely eradicate methicillin-resistant S. aureus (MRSA) from the nasal epithelium in a mouse model of rhinitis [58] suggesting that sophisticated bacteriocin-producing 'probiotics' could become valuable new agents for prevention of opportunistic infections in immune-compromised at-risk patients.

Materials and Methods
Bacterial strains and growth conditions C. accolens, C. pseudodiphtheriticum, D. pigrum and R. mucilaginosa were a kind gift from the Medical Microbiology Department of Münster. M. catarrhalis, P. acnes, and H. influenzae were provided by the microbial diagnostics facility of University Hospital Tübingen. S. pyogenes BK2192, S. aureus Newman, S. aureus Newman ΔdltA and S. epidermidis 1457 are frequently used laboratory strains. Additionally, a nasal isolate of M. luteus was used. Tryptic soy broth (TSB) and tryptic soy agar (TSA) were used to cultivate S. aureus, S. epidermidis, M. luteus, S. pyogenes, R. mucilaginosa, M. catarrhalis, P. acnes, and C. pseudodiphtheriticum. Brain heart infusion (BHI) agar and broth were used for C. accolens. Staphylococcus isolates from nasal swabs from 37 healthy volunteers (IVK strains) have been described in a recent study [12]. In short, the nasal swabs were resuspended in 1 ml PBS. 100 μl of 10 −2 and 10 −3 dilutions were plated on blood agar plates, and single colonies were analyzed by MALDI-TOF (Matrix Assisted Laser Desorption/Ionisation-Time of Flight) for species determination.

Ethics statement
The nasal sample collection procedure was approved by the clinical ethics committee of the University of Tuebingen (No. 109/2009 BO2) and informed written consent was obtained from all volunteers. Nasal swabs were taken exclusively from healthy adults.

Analysis of antimicrobial activities
The nasal Staphylococcus isolates were stamped on TSA agar plates inoculated with the relevant test strain and incubated at 37°C. In order to prepare S. aureus and S. epidermidis test plates, autoclaved tryptic soy agar (TSA) was cooled to 45°C and inoculated with the test strains to an OD 600 = 0.001. After mixing the medium with the cells, plates of 20 ml volume were poured. For M. luteus plates bacteria were inoculated at an OD 600 of 0.01. For all other strains, OD 600 = 0.1 was used.

DNA manipulation and sequencing
For DNA manipulation standard procedures were used [59]. All restriction enzymes and High-fidelity polymerase (used for all PCRs) were obtained from Fermentas GmbH, Germany. Primers used for PCR or DNA sequencing were obtained from Eurofins MWG Operon, Germany (Table 2). DNA sequencing was done by GATC Biotech AG, Germany and Lasergene software (DNAStar, Inc., USA) was used for sequence analysis. Preparation of electro-competent Staphylococcus cells and electroporation was performed as described elsewhere [60].

Spa-typing and MLST analysis
The spa-types of all S. aureus IVK strains except two have been published recently [61]. Since the two S. aureus isolates characterized here were non spa-typeable by the common method, we amplified the whole spa gene with primers spa3-F and spa3-R (Table 2) [62] The PCR products were subsequently sequenced and the repeat region was analyzed using Ridom StaphType 1.5.21 software [63].
MLST of the seven house-keeping genes for S. epidermidis was performed with the primers listed on the MLST web site (http://pubmlst.org/sepidermidis/) and sequences were compared with the MLST database.

Transposon mutagenesis
The transposon mutagenesis procedure was performed as described earlier [64]. Briefly, S. epidermidis IVK45 harboring the transposon vector pTV1ts [65] was grown in TSB containing 5 μg/ml erythromycin (Em) and 10 μg/ml chloramphenicol at 30°C overnight. The culture was then diluted in TSB containing 2.5 μg/ml Em and cultivated overnight at 42°C. After two further cycles of cultivation at 42°C with Em 2.5 μg/ml the cells were spread on TSA plates containing Em 2.5 μg/ml. Erythromycin-resistant and chloramphenicol-sensitive mutants were screened for their ability to inhibit M. luteus growth. Among 100 mutants, 10 did not inhibit the test strain M. luteus. To determine the exact insertion site, the isolated plasmid DNA was sequenced with primers Tn917 up and Ptn2 down, which read upstream and downstream of the transposon.

Construction and complementation of a nukacin IVK45-deficient mutant
The gene for the nukacin IVK45 precursor peptide nukA was exchanged for an erythromycin resistance cassette. The flanking regions of nukA were amplified with primers Nukacin KO up1, Nukacin KO up2 (F1) and Nukacin KO down1, Nukacin down2 (F2). After digestion of F1 with EcoRI/Acc65I and F2 with BamHI/SalI, the two restricted flanking region PCR products F1 and F2 were subsequently ligated into the EcoRI/Acc65I and BglII/SalI digested vector pBase6-erm/lox2. The resulting plasmid pBASE6-ΔnukA was used to transform E. coli DC10B [66]. The plasmid was subsequently isolated and transferred to S. epidermidis IVK45, where the homologous recombination process resulted in nukA-deficient mutants [67], which were confirmed by PCR analysis. For complementation, the shuttle vector pRB474 [68] was used. With primers compl.-promotor and compl. down, nukA was amplified. The resulting PCR product was digested with SalI/PstI and ligated into SalI/PstI-restricted pRB474. The resulting complementation vector pRB474-nukA was used to transform E. coli DC10B. After plasmid isolation, the vector was directly transferred to S. epidermidis IVK45 ΔnukA via electroporation.

Antimicrobial activity assay for nukacin IVK45
Bioactivity was assessed by agar diffusion assays against the sensitive indicator strain M. luteus. To this end, TSA was inoculated with an overnight culture of M. luteus to a final OD 600 of 0.01. After solidification of the agar, 10 μl of an overnight culture were put on the agar plate and incubated at 37°C. After incubation overnight, inhibitions zones were measured.

Nukacin activity in liquid medium
Fresh TSB Medium was supplemented with sterile filtered supernatant of IVK45 wild type or the nukacin-deficient mutant, or spent TSB medium of the cultures was used directly for monitoring growth of the test strains in a photometer. The cultures were inoculated with M. luteus or M. catarrhalis (to OD 600 = 0.01) and colony-forming units per ml were determined for several time points. C. accolens only grew in liquid BHI medium in the presence of 0.2% Tween 80. The amount of conditioned culture filtrates required for optimal growth and inhibition of potential competitors were different for the test strains corresponding to 20%-40% (M. luteus), 50%-100% (C. accolens), and 100% and enriched activity (M. catarrhalis). For the 4-fold enrichment of nukacin IVK45 activity the 100% ethanol XAD-1180 eluate, described in the nukacin IVK45 purification section, was dried and resuspended in TSB. Subsequently, the dilution factor was determined to obtain the same inhibitory activity as with the corresponding 100% culture supernatant. Experiments were performed with four-fold amount of the determined concentration.

Co-cultivation experiments
S. epidermidis IVK45 wild type, S. epidermidis IVK45 ΔnukA and M. catarrhalis were grown in TSB overnight at 37°C under continuous shaking. These strains were then adjusted to 1 x 10 7 c.f.u./ml in 1 x PBS. For the starting conditions each of the S. epidermidis strains was mixed with M. catarrhalis at a ratio of 3:1. 20 μl of these mixtures were spotted on TSB agar and incubated at 37°C. Samples were taken at 0 and 48 hours by scraping cells from the agar plates, which then were suspended in 1 x PBS. Serial dilutions of these samples were plated on TSB agar. After overnight incubation at 37°C colony counts were determined, and the bacterial ratios of S. epidermidis and M. catarrhalis were calculated. The individual strains could be distinguished by size and color of their colonies.
Nukacin IVK45 purification XAD-1180 resin, cation exchange (SP Sepharose FF column), and size exclusion (Sephadex LH-20 column) chromatography were used to purify nukacin IVK45 from liquid cultures. Cell-free culture supernatant was incubated with 1/10 volume of XAD-1180 adsorber resin (Acros Organics) and incubated for 2 h under constant agitation. Subsequently XAD-1180 was washed twice with 2 x bed volumes (BV) 60% ethanol. For elution 2 x BV of 100% ethanol were applied. Subsequently, ethanol was removed using a rotary evaporator. The resulting extracts were dissolved in 25 mM sodium phosphate buffer pH 7.2 (binding buffer). The sample was injected into a fast protein liquid chromatography (FPLC) system (ÄKTA Purifier) using a 10-ml SP Sepharose FF (strong) cation exchanger column (GE Healthcare) equilibrated with the same buffer at a flow rate of 1 ml/min. Then the column was washed with a 10-fold column volume of the binding buffer before nukacin IVK45 was eluted using the following step gradient: 1.5 M NaCl in 25 mM sodium phosphate buffer at pH 7.2 was used as elution buffer (Buffer B); 25 min 10% buffer B, 35 min 13% buffer B, 20 ml 100% buffer B; nukacin IVK45 eluted immediately after the gradient reaches more than 13% of buffer B. All fractions were tested for inhibitory activity using M. luteus as the indicator strain.
The active SP Sepharose fraction of the wild type and the corresponding fraction of the deletion mutant were freeze dried, dissolved in 100% methanol, and subsequently separated via a 60-ml Sephadex LH-20 size exclusion column (GE healthcare) with methanol as running solvent. In the active fractions nukacin IVK45 was detected by High Pressure Liquid chromatography-Mass Spectrometry (HPLC-MS).

Nukacin analysis by HPLC electrospray Ionization mass spectrometry (ESI-MS)
The methanolic fractions obtained from the size exclusion chromatography were analyzed by a HPLC-LC/MSD Ultra Trap System XCT 6330 (Agilent Technologies). For the HPLC measurements, a C18 Nucleosil 100 column (100 x 2 mm ID, 3 μm) and a pre-column (100 x 2 mm ID) (Dr. Maisch) were used. Solvent A (0.1% formic acid) and solvent B (0.06% formic acid in acetonitrile) were used as mobile phase applying the following gradient (t 0 = 10% B, t 15