INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteriocins are ribosomally synthesized antimicrobial polypeptides that are usually inhibitory only to strains closely related to the producing bacteria. These antimicrobial compounds are thought to provide the producer strain with a selective advantage over other strains. Bacteriocins produced by gram-positive bacteria are often membrane-permeabilizing cationic peptides with fewer than 60 amino acid residues (25, 29). In recent decades, the major advances in this field have been made in the lactic acid bacterium (LAB) family, due to the eminent economic importance of these microorganisms. Hence, the great structural diversity of LAB bacteriocins in combination with the fact that many bacteriocin producing LAB are present in a variety of naturally fermented food and feed products has led to a great interest in the potential of these bacteria as biopreservatives that could, at least partially, replace chemical preservatives (50). The bacteriocins of LAB have been divided into four distinct classes by biochemical and genetic means (28, 29). Bacteriocins of class I and II are by far the most studied because they are both the most abundant ones and the most prominent for industrial application (41). Class I bacteriocins called lantibiotics contain modified amino acid residues, lanthionine and methyllanthionine, which are formed posttranslationally (10). Class II consists of bacteriocins that lack modified residues. The pediocin-like bacteriocins constitute a large subgroup within class II: they are all small, heat-stable, membrane-active peptides that have a YGNGVXC consensus motif and are also characterized by their strong inhibitory effect on Listeria (29, 41). Representatives are widespread among LAB, including pediocins AcH and PA-1 produced by Pediococcus acidilactici (18); mesentericin Y105 produced by Leuconostoc mesenteroides (21); enterococin A produced by Enterococcus faecium (1); carnobacteriocins BM1, B2, and piscicocin V1a produced by Carnobacterium piscicola (2, 45); and divercin V41 produced by Carnobacterium divergens (35).

Antimicrobial compounds are also produced by other gram-positive bacteria (25), including the following Bacillus species: B. subtilis (26, 54), B. thuringiensis (8, 43), B. stearothermophilus (49), B. licheniformis (5), B. megaterium (24), B. thermoleovorans (42), and the food spoilage microorganisms B. cereus (39) and B. coagulans (23). Nevertheless, these reports usually suffer from limited biochemical characterization, often carried out on culture supernatant and not on the purified peptide, and a lack of genetic information. The only bacteriocins from Bacillus to be characterized at the amino acid and DNA sequence levels are subtilin (26, 30) and subtilosin (55) produced by B. subtilis. Subtilin is a member of linear lantibiotics (class I). Subtilosin is a cyclic bacteriocin, and the precursor peptide is reported to undergo several unique and unusual chemical posttranslational modifications, unlike those occurring during the synthesis of class I or some class II bacteriocins (55).

According to some authors (41), nonlantibiotic bacteriocins, which are widespread in LAB, are probably produced by other gram-positive bacteria. Nevertheless, no class II bacteriocin has so far been characterized in the genus Bacillus. In this paper, we report on the detailed study of the antilisterial peptide produced by B. coagulans strain I₄, isolated in the course of a survey of spore-forming LAB for novel antimicrobial compounds (23). Coagulin is described by the N-terminal sequencing of the purified peptide and sequencing of the dedicated operon. Analysis demonstrates that coagulin is a new member in the pediocin-like family of bacteriocins. A putative Mob-Rs_A module was identified in the vicinity of the coa operon. The implications of our data for the intergeneric transfer of the bacteriocin operon are discussed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and media. The bacteriocin producer B. coagulans I₄ was previously isolated from cattle feces (23). The strain was propagated aerobically in MRS (de Man, Rogosa, and Sharpe) broth (Difco Laboratories, Detroit, Mich.) at 37°C and maintained as a frozen stock at 20°C in the same medium supplemented with 20% (vol/vol) glycerol. Escherichia coli NM522 [supE thi (lac-proAB) (hsdMS-mcrB)5 F'(proAB lacI^q lacZM15)] (33) was used for all genetic manipulations and propagated in Luria-Bertani broth or agar (15 g/liter; Difco Laboratories) at 37°C (46).

Determination of bacteriocin activity. Bacteriocin activity was determined by the well diffusion assay (deferred method) described by Tagg and McGiven (51) in tryptose agar (Difco) seeded with Listeria innocua. The overnight culture of the indicator strain was grown in tryptose broth. To quantitate inhibitory activity, the diameter of the inhibition zone (in millimeters) was measured. For strong activities, the titer (expressed in activity units [AU] per milliliter), corresponding to the reciprocal of the highest dilution showing definite inhibition of the indicator lawn, was determined.

DNA manipulations. Plasmids from E. coli and B. coagulans were extracted and purified as previously described (23, 46). Plasmid DNA was digested with restriction enzymes (Eurogentec, Seraing, Belgium) according to the supplier's recommendations. For cloning purposes, EcoRI or BamHI fragments of pI₄ DNA were ligated into EcoRI- or BamHI-digested pUC19 using T4 DNA ligase (Eurogentec). The ligated DNAs were transformed into competent E. coli NM522 cells as described previously (31). Analytical and preparative agarose gel electrophoresis in Tris-borate-EDTA (pH 8.3) was performed as described (46). DNA fragments were isolated and purified from 1% (wt/vol) agarose gels with the Nucleotrap kit (Clontech, Palo Alto, Calif.). Southern blot hybridizations (46) were performed using Hybond-N⁺ nucleic transfer membranes (Amersham, Little Chalfont, Buckinghamshire, United Kingdom). DNA probes were labeled and used for hybridization experiments with the enhanced chemiluminescence kit according to the high-stringency conditions specified by the supplier (hybridization and washing steps at 42°C and in 0.1× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate]; Amersham). Nucleotide sequencing, based on the chain termination method (47), was done using an Auto-read sequencing kit (Pharmacia Biotech, Uppsala, Sweden) with either standard or specific primers in an automated DNA sequencer (ALF; Pharmacia).

Sequence analyses and alignments. DNA and protein homology searches (Genbank, EMBL and SWISS-PROT) and sequence analysis were performed with the programs of the Genetics Computer Group sequence analysis software package (University of Wisconsin). Multiple sequence alignment was carried out with CLUSTAL W 1.60 (52).

Bacteriocin purification. (i) Concentration. The bacteriocin was purified from 1-liter cultures of B. coagulans I₄ grown in MRS broth at 37°C, to late logarithmic phase (A₆₀₀, approximately 10⁸ CFU/ml). The cells were removed by centrifugation at 4,000 × g for 15 min at 4°C, after which the supernatant was adjusted to pH 6.5 and filtered through a 0.45-µm-pore size HVLP membrane (Millipore SA, Molsheim, France). The filtrate was concentrated 10-fold using a spiral cartridge concentrator (model CH 2; Amicon Division, W. R. Grace and Co., Beverly, Mass.) with a molecular size cutoff of the membrane of 10,000 Da.

(ii) Solid-phase extraction. The lyophilized retentate (concentrated crude extract) was solubilized in 5 mM NH₄HCO₃ buffer, pH 8.0, and loaded on a Sep-Pak Plus C₁₈ cartridge (Waters, Division of Millipore Corp., Bedford, Mass.) for solid-phase extraction. Nonactive material was discarded after washing the Sep-Pak cartridge with 5 ml of NH₄HCO₃-2-propanol (80:20, vol/vol). Active fractions (i.e., those displaying anti listeria activity) were eluted using 5 ml of 5 mM NH₄HCO₃-2-propanol (30:70, vol/vol) and then dried under vacuum centrifugation.

(iii)RP chromatography. The dry residue (fraction I) was dissolved in 5 mM NH₄HCO₃-2-propanol (30:70, vol/vol) and centrifuged at 10,000 × g for 5 min. The supernatant was further used for bacteriocin purification by reverse phase (RP) chromatography performed with an analytical high-performance liquid chromatography (HPLC) system (Kontron Instruments) with an automated gradient controller, 332 UV detector, and 425 integrator. The supernatant (0.5 ml) was loaded on a Delta-Pack C₁₈ (Waters) semipreparative RP-HPLC column (8 by 100 mm; diameter, 15 µm; 100 Å), equilibrated at a flow rate of 2 ml/min with H₂O-acetonitrile (90:10, vol/vol) containing 0.1% trifluoroacetic acid (solvent A). Activity was eluted with a linear gradient ranging from 15 to 60% acetonitrile containing 0.07% trifluoroacetic acid (solvent B). Active fractions were pooled, dried under vacuum (fraction II), dissolved in 5 mM NH₄HCO₃-2-propanol (30:70, vol/vol), and then rechromatographed on a Licrospher C₁₈ (Merck, Darmstadt, Germany) analytical RP-HPLC column (4 by 125 mm; diameter, 5 µm; 100 Å). The column was equilibrated at a flow rate of 1 ml/min with solvent A. The bacteriocin was eluted with a gradient of solvent B. Elution conditions were as follows: 0 to 5 min, 100% A to 90% A-10% B (vol/vol); from 5 to 30 min, linear gradient to 75% A-25% B (vol/vol). Active fractions (1 ml) containing the purified bacteriocin were collected, dried under vacuum and stored at 20°C (fraction III).

Determination of protein. Protein content was determined by the Bradford method (4) using the Bio-Rad assay reagent (Bio-Rad, Munich, Germany) with bovine serum albumin as the standard.

Amino acid sequencing, enzymatic cleavage, and MS of coagulin. Amino acid sequencing was performed by Edman degradation with an automatic liquid phase sequence analyzer (model 473; Applied Biosystems, Foster City, Calif.). Presence of disulfide bonds was pointed out by hydrolysis of the purified bacteriocin with the Lysobacter enzymogenes Lys-C endoproteinase (Sigma Chemical Chimie, St Quentin Fallavier, France).

Analysis of coagulin C-terminal sequence hydrophobicity and hydrophobic moment. The average sequence hydrophobicity and average sequence hydrophobic moments of coagulin sequence regions were calculated by using the normalized amino acid hydrophobicity scale and hydrophobic moment calculation method as described by Eisenberg et al. (12).

Nucleotide sequence accession number. The nucleotide sequence corresponding to the coa operon and mob-Rs_A module has been deposited in the GenBank database under accession no. AF 300457.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Purification of antimicrobial peptide produced by B. coagulans I₄. The antimicrobial peptide produced by B. coagulans I₄ was purified to homogeneity from a late-logarithmic-phase culture, grown at 37°C in MRS broth. Results are summarized in Table 1. Ultrafiltration was used successfully to concentrate the activity from the growth medium. Approximately a two-fold increase in the specific activity and a 65% recovery of the peptide were obtained (Table 1). Ultrafiltration was followed by solid-phase extraction on a silica-based bonded phase with strong hydrophobicity. The recovery, 52%, remained high after binding the peptide contained in the concentrated crude extract and eluting with 5 mM NH₄HCO₃-2-propanol (Table 1, fraction I). The largest increase in specific activity, about 109 times, was obtained after the final analytical RP-HPLC steps (Table 1, fraction III). A single absorbance peak, corresponding with the activity peak, was obtained (Fig. 1). The final specific activity of the purified peptide was evaluated to be about 140,000 AU/mg.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Reverse-phase chromatography of the antimicrobial peptide produced by B. coagulans I4. The amount applied to the column was obtained from a 1-liter culture.

The antimicrobial peptide produced by B. coagulans I₄ is a pediocin-like bacteriocin. The purified peptide was analyzed by mass spectrometry, which confirmed the purity of the sample and showed a molecular mass of 4,612 Da. Amino acid sequencing enabled the partial determination of the first 39 amino acid (aa) residues. Seven positions (9, 14, 19, 24, 29, 33, and 34) could not be identified. Comparisons with protein sequences contained in the data banks demonstrated a total identity of the 32 identified amino acid residues with the sequence of pediocin PA-1 (22). This 44-aa bacteriocin, with a molecular mass of 4,624 Da, contains four cysteine residues forming two disulfide bridges, between C-9 and C-14 and between C-24 and C-44, which are required for the antibacterial activity (6, 13, 22). Digestion of the newly purified bacteriocin by Lys-C endoprotease produced only two RP-HPLC peaks (results not shown) as has been previously observed for pediocin when submitted to the same treatment (15, 22). These results demonstrate the presence of disulfide bridge(s) in the structure of the peptide. In spite of the observed similarities between the partial sequence of the two bacteriocins, the difference between the molecular masses might be suggestive of a slight variation in the primary structure.

Cloning and nucleotide sequence analysis of the bacteriocin locus. Evidence for plasmid-encoded bacteriocin was obtained previously by plasmid curing (23), suggesting that production and immunity are associated with the 14-kb plasmid, named pI₄, present in B. coagulans I₄. In an attempt to express the bacteriocin production gene in E. coli, restriction fragments from plasmid pI₄ DNA were cloned in pUCI9 (see Materials and Methods), and the bacteriocin production of transformants was tested using the agar well diffusion method. L. innocua was used as the indicator strain. Cells carrying the recombinant clone pBC8 showed a zone of inhibition in the indicator lawn, while the control strain E. coli NM522 containing pUC19 did not. This observation indicated that the bacteriocin was produced and, most likely, secreted into the medium. Southern blot hybridization analysis showed that this clone contained a 5-kb EcoRI fragment of plasmid pI₄ from B. coagulans I₄. To localize the bacteriocin production gene, the 5-kb EcoRI_1-2 fragment was mapped and the identified restriction sites were used to obtain deletion derivatives. An overview is given in Fig. 2. The SacI deletion derivative retained activity while neither of the other derivatives obtained displayed zones of inhibition against the indicator strain. Since the SacI site was observed to be close to the EcoRI₁ cloning site (Fig. 2), the sequence of the entire 5-kb EcoRI fragment from pI₄ was determined.

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Schematic representation of the cloned region of pI4 that mediates expression of coagulin in construction pBC8 and its deletion derivatives. The shaded area spanning coordinates 1046 to 4577 represents the section homologous to the ped-pap operon, present on the P. acidilactici pSMB 74 plasmid, and required for pediocin PA-1-AcH expression. The coding region containing coaA, coaB, coaC, and coaD spans position 1089 to 4474. The structures resembling rho-independent terminators are indicated. The vertical arrow represents the putative promoter for coagulin expression in B. coagulans. Abbreviations for restriction enzymes: C, ClaI; EI, EcoRI; EV, EcoRV; H, HindIII; P, PvuII; Sa, SacI; Sm, SmaI; X, XbaI. +, coagulase expressed; , coagulase not expressed.

The bacteriocin operons from P. acidilactici and B. coagulans differ by six mismatches and display different promoter sequences. As mentioned above, six mismatches were identified between the Pediococcus and Bacillus 3.532-kb homologous sequences. Five mapped in the transport genes ORF III and IV. They resulted in three substitutions in ORF IV, encoding the ATP-binding cassette transporter (V488F, F670L, and Y712H), and two in ORF III, encoding the accessory protein (R54K and V151I). The last of these mismatches (V151I) resulted in the presence of an EcoRV restriction site, demonstrated during the mapping of the 5-kb EcoRI fragment of pI₄, and absent in pSMB74 (38). Hence, these five amino acids are not essential for transport activity since active coagulin was recovered in the supernatent of B. coagulans I₄ cultures. The sixth mismatch was identified in ORF I, the bacteriocin production gene, and resulted in a single substitution at the C terminus of the peptide. Hence, a threonine residue was present at position 41 (T41) in coagulin while pediocins AcH and PA-1 contain an asparagine residue (N41) (22, 38).

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Translational and transcriptional signals upstream of coaA in pI4 from B. coagulans and pedA in pSMB 74 from P. acidilactici. Numbering for the pSMB74 DNA sequence is that reported for the complete sequence of the plasmid (38). The boxed area denotes the homologous sequences, starting at position 1046 in the EcoRI fragment from pI4. 35 and 10 boxes and the ribosome binding sequence (rbs) sequence are underlined.

A putative mobilization module is present immediately downstream of the coa operon. The 3.532-kb homologous region containing the coa locus is located between two sequences of 1.045-kb and 0.384-kb, with G+C contents of 36 and 44%, respectively. Both sequences lack homology with pSMB 74 DNA and sequences found in the nucleotide sequence databases (EMBL and GenBank). When translated, each nonhomologous region revealed a truncated ORF (ORF X and ORF Y, respectively) (Fig. 2). ORF X, is followed by a region of dyad symmetry with a calculated free energy (G₀) of 32.6 kcal/mol, representing a putative rho-independent terminator. Complete elucidation of the sequences of ORF X and ORF Y was achieved by sequencing the restriction fragments contiguous to the 5-kb EcoRI fragment containing coa, which were positioned and cloned during the mapping of pI₄ (C. Le Marrec, B. Hyronimus, P. Bressollier, and M. C. Urdaci, unpublished data). The additional 1.5-kb sequence containing the 5' part of ORF X has a G+C content of 46%. Analysis did not reveal any homology to pSMB 74 DNA or to other known sequences (data not shown). Data revealed that ORF X specified a 181-aa protein with no homology between it and other protein sequences contained in the databases.

View larger version (94K): [in this window] [in a new window]   FIG. 4.   Comparison of mobilization modules (Mob-RsA) from plasmids isolated from gram-positive bacteria. EMBL-GenBank accession numbers, given between square brackets, for the following plasmids are indicated pLAB1000 (Lactobacillus hilgardii [P35856]), pLB4 (L. plantarum [P20046]), pK214 (L. lactis [WX92946]), pMV158 (Streptococcus agalactiae [P13925]), pS86 (E. faecalis [AJ223161]), pIP1714 (Staphylococcus cohnii [AF015628]), pCB16 (B. cereus [U32369]), pUB110 (S. aureus [P22490]), and pTB19 (thermophilic Bacillus, [JQ1212]). (A) Partial alignments of Mob proteins (generated by using the CLUSTAL W program [52]). Conserved residues are marked with an asterisk. Closely related residues are indicated by colons, while less related residues are indicated by single dots. The two conserved regions I and II (20, 34) are indicated with filled bars. (B) Comparisons of RSA sites found upstream of Mob family members. The inverted repeats are indicated by arrows. Conserved nucleotides are printed in boldface type.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In order to gain insight into the production of bacteriocins among Bacillus species, we characterized the antimicrobial compound produced by B. coagulans I₄, using biochemical and genetic analyses. Purification of the bacteriocin was achieved using a four-step procedure that included ultrafiltration (UF) and RP chromatography. Good recovery of activity (65%) after the UF step is promising, because this technique is relevant for large-scale production. The purification level (109-fold) appears to be lower than those obtained for other structurally related bacteriocins (14, 15, 22). This was mainly the result of the greater sensitivity of the Bradford method, which allows a more acute determination of protein concentration in crude extract compared to A₂₈₀ or A₂₅₄ measurements.

Surprisingly, analysis of the structure of the purified bacteriocin produced by B. coagulans I₄ and the sequence deduced from the gene show a strong similarity with that of pediocins AcH and PA-1. The only difference is a threonine residue at position 41 instead of asparagine (N41T). This single modification in the structure is confirmed by the difference observed between the molecular masses of purified coagulin and AcH, estimated by mass spectrometry to be 4,612 and 4,624 Da, respectively. Furthermore, this difference in measured (4,612 Da) and estimated (4,616 Da) molecular masses may suggest the existence of two disulfide bonds linking the four cysteine residues on the sequence of coagulin. The N41T modification occurs at the C-terminal part of the peptide, which is less conserved than the N-terminal region within the family of pediocin-like bacteriocins (22). Recently, pediocin AcH substitution mutants with altered bactericidal activity have been described (36). Such single-residue mutants obtained in the C-terminal I25-H42 region (I26T, M31T, A34D, G37E, G37R, and H42L) displayed a modified hydrophobic moment. The authors suggested that this domain, when folded into a putative  helix, becomes amphiphatic and would mediate the adsorption to the bacterial membrane. Modification of its hydrophobic moment would change the angle of contact at which pediocin adsorbs to the membrane interface. This modification was observed for all mutants, except for one case (N41K), which showed a 10-fold-reduced activity, although the substitution did not change the hydrophobic moment of the sequence. Interestingly, the discovery of coagulin shows that an N41T variant exists naturally and that this modification does not result in a significant change in average hydrophobicity and hydrophobic moment of the I25-H42 C-terminal part (Table 2). As suggested (36), the formation of the C-24-C-44 disulphide bond probably interferes with the formation of an  helix. A theoretical loss of specific activity of coagulin compared to pediocin AcH and the N41K variant could not be calculated. Such studies would be interesting to elucidate the influence of such a mutation on activity and host spectrum. Preliminary studies indicate that coagulin is not active on Lactobacillus plantarum and Staphylococcus aureus, contrary to pediocin AcH. Such results seem to be contradictory to those obtained by Miller et al. (36), who obtained a single mutant, A34D, whose mutation may have a species-specific effect.

Pediocins have been isolated from different P. acidilactici strainsPAC1.0 (32), LB 42-923 (3, 37), SJ-1 (48), M (14)and from Pediococcus pentosaceus strains FBB61 (44) and L7230 (7). These strains were isolated from fermented sausages or cucumbers. Recently, a pediocin-like antilisterial bacteriocin was purified from an L. plantarum strain named WHE 92, which was isolated from munster cheese (15). To our knowledge, the present study provides the first evidence for pediocin production by a non-LAB bacterium. Interestingly, available data concerning the host spectra of these bacteriocins reveal that pediocins PA-1-AcH (3, 22), SJ-1 (48), AcM (14), and A (44) have been found to be active against L. plantarum and that pediocin AcM is active against B. coagulans (14).

Production of pediocin by L. plantarum WHE 92 has been reported to involve genes that are located on an 11-kb plasmid. This size is different from those of the plasmids involved in pediocin production in P. acidilactici (37). These authors envisaged an intergeneric conjugal transfer of a plasmid harboring the pediocin determinants, followed by a modification event (15). Analysis of the 5-kb fragment harboring the coa locus revealed extremely high conservation of these determinants, with only six mismatches out of 3.532-kb, which may support the idea of a similar genetic transfer by conjugation between Pediococcus and Bacillus spp. Nevertheless, the following observations remain intriguing: (i) the operon is located between two regions (2.545 and 1.885-kb) with no homology to plasmid pSMB 74 DNA, (ii) the G+C content of the coa operon and the upstream 2.545-kb sequence (39%) is lower than that reported for the chromosome of B. coagulans (45 to 47%), and (iii) the promoter identified in P. acidilactici is absent in pI₄. It is tempting to speculate that these elements may be relevant to an alternative situation, involving the extremely precise excision of the Pediococcus operon sequence devoided of its promoter, and its rearrangement downstream of a promoter on a different resident plasmid. Although the mobilizing capacity of pI₄ remains untested, the presence of a Mob-RS_A module downstream of coa could explain the following step, i.e., plasmid transmission between bacteria. The interaction of the oriT-RS_A site and Mob protein are well known to contribute to the dissemination of small antibiotic plasmids among gram-positive bacteria (53). Hence, the suggested precise recombination, then combined to a conjugative mobilization and/or interplasmidic recombination event, would have enabled the transfer of the pediocin determinants to other gram-positive bacteria, such as B. coagulans.

Complete nucleotide sequence of pI₄ and detailed functional analysis will be necessary to understand the organization of this plasmid and to enable some conclusions about its evolution. In particular, the availability of this sequence should help to determine whether the bacteriocin determinants are contemporary additions or not and to understand the mechanism that led to their inheritance by B. coagulans I₄. Recent comparative studies of the sequences of the S. aureus antimicrobial multiresistance plasmid pSK41 (16) and the Lactococcus lactis bacteriocin production plasmid pMRC01 (11) suggested that insertion sequence-mediated cointegrate formation may play an important role in the evolution of these plasmids, allowing bacteria to collect and subsequently disseminate advantageous traits. To date, little information is available on plasmids replicating in B. coagulans, except for the report of a cryptic 1.6-kb plasmid called pBC1 (9). This rolling-circle type plasmid (RCR) was classified in the pC194/pUB110 family (19, 27). RCR plasmids are characterized by the modular organization of functional regions and known to be highly recombinogenic, a feature that may favor their intra- and intergeneric dissemination (27).