INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

For production of high-quality "red-smear cheese," complex, undefined consortia composed of many species of bacteria and yeast from different genera are often used (53). Most species of these consortia belong to the "coryneform" bacteria, but many isolates have not yet been classified. Since contamination of red smear cheese with Listeria monocytogenes has caused considerable problems, several studies have been performed on the antilisterial effects of coryneform bacteria (11, 21, 30, 40, 50). Only one bacteriocin produced by Brevibacterium linens has been characterized at the molecular level so far (51, 52). This strain has also been evaluated with respect to its antilisterial potential in situ on model soft cheese under laboratory conditions (15). The bacteriolytic substance linenscin produced by B. linens OC2 was found to inhibit L. monocytogenes, but it was not classified as a bacteriocin (28) and its structure has never been published.

A smaller fraction of the red-smear cheese bacteria belong to genera other than coryneform bacteria, such as Staphylococcus or Micrococcus. By screening the surface ripening flora of German and French smeared cheeses with respect to their antilisterial potential, the highly inhibitory Staphylococcus equorum was isolated from the surface of a traditional French Raclette cheese, where it was found in cell numbers up to 10⁸ per cm² (the present study).

The genus Staphylococcus is known to produce many antibacterial substances (41, 42). According to Klaenhammer (27), these can be classified into four main groups: Lantibiotics (class I), low-molecular-weight antibiotic-like substances (class II), and high-molecular-weight (100,000 to 500,000 Da) bacteriocins (classes III and IV). Staphylococcins are class III bacteriocins in the classical sense that are sensitive to proteolytic enzymes, often being relatively heat stable (15 min, 120°C). In some cases they were found to be lipoprotein-carbohydrate complexes (class IV, e.g., staphylococcin 1580). Lantibiotics, such as gallidermin, epidermin, Pep5, epicidin, or epilancin, are low-molecular-weight polypeptide antibiotics which are characterized by the presence of unusual amino acids (lanthionine, -methyllanthionine, forming intramolecular thioether bridges) and a high content of unsaturated amino acids (dehydroalanine and dehydrobutyrine [23, 24, 26, 44]). Class I, III, and IV molecules are ribosomally synthesized as precursor peptides which are sometimes posttranslationally modified and processed (26). In the genus Staphylococcus, some low-molecular-mass (1,200 to 1,400 Da), heat-stable, but non-lanthionine-containing antimicrobial peptides were found (class II), but their structure and their mode of biosynthesis is thus far unknown (7).

In this study S. equorum WS 2733 was shown to produce micrococcin P₁, which is an acceptor-site-specific inhibitor of ribosomal protein biosynthesis (12, 35) reported for the first time in the genus Staphylococcus. A potential application of S. equorum as a protective starter culture was evaluated in ripening experiments performed with model soft cheese artificially contaminated with L. monocytogenes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, media, and bacteriocin assay. The micrococcin P₁ producer strain S. equorum WS 2733 was isolated from French Raclette cheese using the HGMF (hydrophobic grid membrane filter) method described previously (11). About 20 cm² (2 to 3 g) of the cheese surface was scraped off aseptically, homogenized in 100 ml of sterile Peptone-Tween diluent (containing 1.0 g of Peptone and 10.0 g of Tween 80 per liter) and filtered through HGMF membranes filter (QA Life Sciences, Inc., San Diego, Calif.). HGMF membranes were transferred grid-side-up to plate count agar supplemented with 3% NaCl (PC3+). After incubation at 30°C for 3 days, the membranes were transferred to plates of tryptose soft agar (TB; Merck, Darmstadt, Germany) with 8 g of agar/liter inoculated with 100 µl of a 24-h culture of the indicator strain Listeria ivanovii (WSLC 3061) to detect inhibitory effects. Colonies corresponding to inhibition zones were picked from the membranes and purified.

Isolation and purification of micrococcin P₁. Micrococcin P₁ was purified from 20-liter cultures of S. equorum WS 2733 grown at 30°C in BHI broth (Merck) for 24 h. The cells were pelleted by centrifugation (12,000 × g, 20 min, 4°C). The antibiotic present in the culture supernatant was concentrated by ammonium sulfate precipitation. The pellet was resuspended in 650 ml of sodium phosphate buffer (50 mM, pH 7.0) and applied to a reversed-phase column (Pharmacia XK 16/70; 700 by 16 mm, silica gel 100 C₁₈, 40 to 63 µm; Fluka) equilibrated with buffer A (0.1% [vol/vol] trifluoroacetic acid [TFA] H₂O) at a flow rate of 1 ml/min. After a washing with 240 ml of buffer A the column was eluted at a flow rate of 4 ml/min with increasing amounts of buffer B (80% acetonitrile-20% H₂O-0.1% TFA [vol/vol/vol]). The antibiotic-containing fractions were collected, pooled, and redissolved in 50% ethanol (fraction II). Final purification was achieved by reversed-phase high-performance liquid chromatography (RP-HPLC) on an RP column (Eurospher 100-C₁₈, 7 µm, 250 by 8 mm; Knauer, Berlin, Germany) equilibrated with buffer A. Micrococcin P₁ was eluted by a linear gradient of 0 to 80% buffer B in 30 min at a flow rate of 3.5 ml/min. All purification steps were performed at room temperature. The chromatographic equipment was obtained from Pharmacia-LKB (Gradi-Frac System; pump, P-50; detectors, Uvicord SII and VWM 2141; detection wavelengths, 280 and 220 nm, respectively) and Perkin-Elmer (Liquid Chromatograph Series 400). In order to monitor purity, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed using a discontinuous 8 to 18% gradient gel (Excel Gel; Pharmacia-LKB). One-half of the gel was silver stained according to the method of Blum et al. (5), while the other half was assayed directly for antimicrobial activity by the modified test described by Bhunia et al. (4).

GC-MS analysis. The RP-HPLC-purified sample was hydrolyzed under N₂ atmosphere in 6 N HCl at 110°C for 24 h. The hydrolyzate was derivatized to n-propyl esters with 3 N HCl in n-propanol (110°C, 10 min). After completion of the reaction the reagents were removed with N₂. Reaction with trifluoroacetic anhydride (110°C, 10 min) subsequently gave the trifluoroacetamides. After removal of the reagents with N₂, the sample was dissolved in toluene and analyzed on a Carlo-Erba 2900 gas chromatograph (GC). The separation was performed on a fused-silica capillary (20 by 0.25 mm, inner diameter; d_f  0.13 µm) coated with Chirasil--Dex (19) as stationary chiral phase. The GC was coupled to a MAT 112S mass spectrometer (MS; Finnigan, Bremen, Germany) with an AMD Intectra Data System. For electron-impact ionization, an energy of 70 eV was used.

MS and NMR spectroscopy. MS analysis was performed with positive-ion-mode detection on a Apex II-ESI-FT-ICR MS (Bruker Daltonics, Billerica, Mass.) operating at 4.7 T. One- and two-dimensional nuclear magnetic resonance (NMR) experiments were performed on a AMX2-600 spectrometer (Bruker, Karlsruhe, Germany) operating at a proton frequency of 600.13 MHz and equipped with a 5-mm inverse triple resonance probehead with a self-shielded Z-gradient coil. The following experiments were recorded at 298 K using a solution of 1.4 mg of the purified sample dissolved in 0.5 ml CD₃OH: ¹H NMR, TOCSY (total correlation spectroscopy), NOESY (nuclear overhauser spectroscopy), and HSQC (heteronuclear single quantum coherence). For suppression of the hydroxy resonance of the solvent a Watergate sequence was appended to the homonuclear experiments. An HMBC (heteronuclear multiple bond correlation) spectrum was acquired at 318 K using a solution of 6 mg of the purified sample dissolved in 0.5 ml of CD₃OH. HSQC and HMBC experiments were performed with gradient selection. A ¹³C NMR spectrum of the second sample was recorded on a AC250 spectrometer (Bruker, Karlsruhe, Germany) operating at a carbon frequency of 62.90 MHz and equipped with a 5-mm dual-probe head. Chemical shifts (see Table 2) were referenced to the signals of the solvent at (¹H) = 3.35 ppm and (¹³C) = 49.0 ppm.

Cheese-ripening experiments. To evaluate the antilisterial potential of the S. equorum strain WS 2733 in situ on soft cheese, model ripening with defined single-strain cultures was performed according to the method of Eppert et al. (15) under laboratory conditions in glass desiccators. Micrococcin-deficient (mic) mutants and the nonbacteriocinogenic (Bac) S. equorum type strain DSMZ 20674 were used as negative controls. Unripened soft cheese of the type "Weinkäse" were obtained from a local dairy after salting and being hand-smeared in petri dishes using gloves and 50 ml of a brine solution with 5% NaCl inoculated with the test strain and the control strains, respectively. The first day of smearing was designated day 1 of the ripening period. Smearing was applied five times (days 1, 3, 5, 7, and 10) at intervals of 2 or 3 days. The cheese was artificially contaminated with L. monocytogenes WSLC 1364 (from the Vacherin Mont d'Or outbreak) on day 1 of smearing. The contamination levels were 4.5 × 10⁴ CFU/ml of brine (resulting in 1.5 × 10² CFU/cm² of cheese surface), 1.5 × 10³ CFU/ml of brine (10 CFU/cm²), and 2.0 × 10² CFU/ml of brine, respectively. The bacterial counts of the Staphylococcus ripening culture in the smear brine were 1.3 × 10⁸/ml (DSMZ 20674), 1.5 × 10⁸/ml (mic mutant), and 1.8 × 10/ml (WS 2733), respectively. In order to cope with a potentially nonuniform distribution of listeriae on the cheese surface at low contamination levels, a large part of the surface was investigated for cell counts (see below).

1

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of S. equorum WS 2733 and inhibitory spectrum. An S. equorum strain, inhibiting a variety of L. monocytogenes strains, was isolated from the surface of a French Raclette cheese. This bacterial consortium was composed of approximately 7.8 × 10⁸ CFU/cm² (18.5%) orange, 4.4 × 10⁸ CFU/cm² (10.4%) yellow pigmented, and 2.9 × 10⁹ CFU/cm² (68.7%) white coryneform bacteria partly identified as B. linens and Corynebacterium fascians. The fraction of the S. equorum strain was 2.5% (1.0 × 10⁸ CFU/cm²) of the total aerobic plate count (4.2 × 10⁹ CFU/cm²). The peptide antibiotic present in the culture supernatant showed a wide spectrum of inhibitory activity that was not restricted to closely related species (Table 1). A total of 95 out of 95 Listeria strains and 130 out of 138 other gram-positive bacteria were inhibited. Nearly all gram-positive bacteria tested were sensitive, whereas no inhibition of 37 gram-negative bacteria was observed. The antimicrobial peptide revealed to be highly inhibitory against all tested L. monocytogenes strains, while only 36% of the tested Bacillus cereus strains were sensitive. The mode of action in the tested concentrations (12,800 AU/ml) was bacteriostatic (data not shown).

Purification of the active substance. Purification of micrococcin P₁ was achieved by RP chromatography. The final polishing step (RP-HPLC) resulted in a 1,600-fold increase in specific activity with 25% recovery. The elution profile from the RP-C₁₈ column is presented in Fig. 1. The antimicrobial substance was eluted at about 84% buffer B (67% acetonitrile). The strong binding of the compound to the C₁₈ column at relatively high acetonitrile concentrations was indicative of a highly hydrophobic molecule. The purification procedure was followed by SDS-PAGE analysis. A single band was detected migrating at an M_r of <2,500 (Fig. 2A, lane 3), which is consistent with a pure preparation and indicative of a relatively small compound. The substance was localized in situ by its antibacterial activity. A clear and distinct zone of inhibition (Fig. 2B), corresponding to an M_r of <2,500 was revealed on the gel overlaid with a L. ivanovii WSLC 3061 indicator lawn.

View larger version (31K): [in this window] [in a new window]   FIG. 1.   Elution profile of micrococcin P1 (A) RP-HPLC on a Eurosphor 100-C18 column (250 by 8 mm; sample volume, 10 µl; flow rate, 3.5 ml/min; linear acetonitrile gradient, 0 to 80% in 30 min) of crude concentrate (fraction II). All contaminating compounds were separated at the beginning of the linear gradient (0 to 60% B). (B) Analytical separation of pooled fraction IV. The pure antibacterial compound was eluted as a distinct, symmetrical peak at about 84% of solvent B (67% acetonitrile). The arrow indicates the activity.

View larger version (87K): [in this window] [in a new window]   FIG. 2.   SDS-PAGE of preparations after various stages of purification. (A) Silver-stained Excel Gel (gradient, 8 to 18% T; Pharmacia-LKB). Lane 1, crude concentrate after ammonium sulfate precipitation; lane 2, fraction after preparative RP-LC; lane 3, purified peptide after RP-HPLC; M, molecular weight marker SDS-7L (Sigma); P, peptide length standard (Pharmacia-LKB). (B) Direct detection of antilisterial activity on the gel overlaid with tryptose soft agar seeded with the indicator strain L. ivanovii WSLC 3061. The inhibition zones after incubation overnight at 30°C correspond to the silver-stained bands in Fig. 2A, lanes 2 and 3.

Structure. MS analysis confirmed the purity of the sample and revealed a molecular mass of 1,143 Da for the substance. ¹H and ¹³C NMR spectra (data not shown) comprised signals for 46 hydrogens (hydrogens of the three hydroxy groups exchanged fast with the solvent) and 48 carbon atoms. Considering the molecular mass, the moderate number of carbon and hydrogen atoms in the molecule indicated a high content of heteroatoms. The spin systems of one valine (Fig. 3, V), two threonine (T), and two didehydrobutyrine (Dhb) residues, as well as of a 2-hydroxypropylamide (HPA) moiety, were traced in a TOCSY spectrum (data not shown). Sequential connectivities between HPA and Dhb¹ and between T¹ and Dhb² (Fig. 3), respectively, were determined by a NOESY experiment by which also the Z-configuration of both Dhb residues was confirmed. The HSQC spectrum contained cross-peaks for eight (hetero)aromatic methine units. Heteronuclear one-bond coupling constants of ¹J_CH = 196 Hz which were observed in the HMBC spectrum for six of the respective methine units suggested the presence of six thiazole rings in the molecule. Methods for detection of thiazole containing amino acids were described for the microcin B17 structure elucidation (3).

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Structure of micrococcin P1. Structural units separated by peptide bonds are divided by dashed lines.

Growth reduction of L. monocytogenes on cheese surface. The development of the pH and total aerobic plate counts during the ripening process are shown in Fig. 4A and can be considered typical for the ripening of industrial red-smear cheese using single-strain cultures (15). The total aerobic plates counts of the cheese surface before smearing were about 2.0 × 10⁶ CFU/cm², and the pH values were in the range of 5.0. At the end of the ripening process all cheeses reached total aerobic cell counts of approximately 6.0 × 10⁸ CFU/cm². A growth reduction of L. monocytogenes on cheese ripened with the micrococcin P₁-producing (Mic⁺) S. equorum wild-type strain WS 2733 was observed when compared to control cheese ripened with the bacteriocin-negative strain DSMZ 20674 and the micrococcin-deficient mutant (Mic). The effect was dependent on the contamination level. When challenged with 10⁴ CFU/ml of brine (10² CFU/cm²), a reduction of the viable cell counts by more than 1 log unit could be observed, reaching a final level of 8.0 × 10⁵ CFU/cm², while listeriae on control cheese grew to 9.0 × 10⁶ CFU/cm². When challenged with 10³ CFU/ml of brine, a growth reduction of 1 to 2 log cycles could be demonstrated. Remarkable effects could be achieved with low contamination levels (10² CFU/ml of brine). The Listeria cells grew to approximately 1.2 × 10⁵ CFU/cm² on the control cheese ripened with the micrococcin-negative strain DSMZ 20674 (Bac) and to 1.5 × 10⁶ CFU/cm² on control cheese ripened with the micrococcin-deficient knockout mutant (Mic), whereas on cheese inoculated with the wild-type S. equorum WS 2733 the growth of listeriae was completely inhibited.

View larger version (25K): [in this window] [in a new window]   FIG. 4.   Cheese ripening with defined single-strain cultures of S. equorum WS 2733. (A) Development of pH on the cheese surface and aerobic plate counts (PC3+ agar). (B) Growth reduction of L. monocytogenes WSLC 1364. Listeria cell counts on the cheese surface (Oxford agar) after contamination at day 1 with 104 CFU/ml (102 CFU/cm2) ( and ), 103 CFU/ml (10 CFU/cm2) ( and ), and 102 CFU/ml ( and ), are shown. Control cheeses were ripened with the micrococcin-negative type strain DSMZ 20674 (bac), closed symbols. (C) Listeria cell counts on the cheese surface (Oxford agar) after contamination at day 1 with 103 CFU/ml ( and ) and 102 CFU/ml (10 CFU/cm2) ( and . Control cheeses were ripened with the micrococcin-deficient mutant (mic), closed symbols. The micrococcin-producing wild-type strain (mic+) is represented by open symbols. Arrows indicate that even by using enrichment procedures, listeriae could not be detected (n.d.). The measuring points mark the detection limit (50 to 100 CFU/cm2) of the direct plating method. +, Listeriae could be detected only after enrichment procedures.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Micrococcin P₁ (Fig. 3) is a macrocyclic peptide antibiotic of the thiocillin-thiazolyl class (8, 9, 13, 25, 32, 54). Recently, a complete synthesis was published by Okumura et al. (34). The peptide binds to complexes formed between the L11 protein of the 50S ribosomal subunit and a nucleotide sequence within the 23S rRNA and inhibits the elongation factor EFG-dependent translocation step of the growing peptide chain. Resistant mutants of B. subtilis and B. megaterium either possess a structurally altered 50S subunit or are specifically methylated at a single site (nucleotide A-1067) of the 23S rRNA (39, 45, 46). It has been demonstrated that this macrocyclic thiopeptide antibiotic is synthesized nonribosomally and involves a multifunctional peptide synthetase, which was designated micrococcin synthase (Carnio et al., submitted).

So far, micrococcin P₁ has been purified from a bacterium related to Micrococcus varians (20, 48) and from Bacillus pumilus (1, 16), isolated from sewage and soil, respectively. Breiter et al. (7) have reported antibiotically active substances excreted by seven strains belonging to Staphylococcus and Micrococcus species isolated from animal sources (partridge, pig, and dog). Two basic substances were isolated and designated micrococcin M1 and M3. The physicochemical and spectroscopic data, as well as investigations of hydrolysis products, indicated a close relationship to micrococcin P₁. However, the structure of these compounds has never been reported. The antibacterial substance secreted by S. equorum WS 2733 was identified as micrococcin P₁ by spectroscopic evidence. Based on biosynthetic considerations, the absolute configuration of the sterocenters was assumed to be as reported for micrococcin P₁. In this context it is of interest that other thiazole- and oxazole-containing polypeptides such as the bacteriocin microcin B17 originate from ribosomally synthesized precursors which are posttranslationally modified (3).

S. equorum was revealed to be a potent inhibitor of growth of L. monocytogenes on the cheese surface. The phenomenon of complete inhibition of listeriae occurs only when the inhibitory strain is used as sole starter and when contamination with Listeria sp. is performed in the early stage of ripening at a low concentration level (10² CFU/ml of brine), which is a high titer for contamination occurring in the dairy industry. This is in accordance with other studies reporting on the inhibition of Listeria sp. on cheese surfaces (9, 14, 15, 17, 18, 29, 33, 36, 47, 49, 51, 55). The limited effects of antilisterial bacteria clearly show that there is a need for developing hurdle concepts in the dairy industry in which bacteriocinogenic bacteria play a certain role. However, such protective starters cannot replace a good hygienic practice during food processing.

Although micrococcin P₁ exhibits a broad activity spectrum, inhibiting nearly all gram-positive bacteria when tested on agar plates (spot-on-the-lawn-assay; Table 1), the antibiotic allows the formation of a stable ecological system on the surface of the French Raclette cheese. However, when combined with other surface-ripening cultures in cheese-ripening experiments under laboratory conditions, S. equorum WS 2733 dominated the ripening flora by suppressing the other gram-positive coryneform bacteria (data not shown). This finding may be explained by the different cheese types: on the surface of a soft cheese the development of the ripening microflora may be completely different from the one on the Raclette surface. It appears, however, to be more likely that the bacterial consortium growing on the surface of the French Raclette cheese constitutes a well-balanced, stable ecosystem which has developed over decades of cheese production. Biocontrol of listeriae on soft cheese by protective ripening bacteria will therefore be more complicated than just adding an antilisterial strain to a well-established smear-cheese flora.

S. equorum was originally isolated from the skin of healthy horses and was first described by Schleifer et al. (43), but it has often been associated with food products. For instance, together with Staphylococcus xylosus, it constitutes the predominant organism present during ripening of an Iberian dry cured ham (37), contributing to the characteristic flavor of the final product. It is also part of the ripening flora from traditional French cheeses (22) and was isolated from goat milk and cheese (31). Moreover, 5 to 15% of the total cell counts of the microflora of the German Tilsit cheese consist of staphylococci classified as S. equorum (6). This species belongs to the group of coagulase-negative staphylococci and has never been reported to be involved in diseases. There would, therefore, be good reason to consider S. equorum a GRAS (generally recognized as safe) organism. In the case of our isolate, however, it should be noted that micrococcin P₁ has to be classified as an antibiotic. With respect to its potential pharmaceutical use it may, therefore, be wise to be careful in spreading this strain widely in the human community before its pharmaceutical potential is evaluated.