INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Peptide bacteriocins are a heterogeneous group of ribosomally synthesized antimicrobial compounds (31). In recent years, a large number of bacteriocins produced by lactic acid bacteria have been described, but relatively few have been characterized at the molecular level (12, 14, 28, 31, 38, 41). The lantibiotics (class I) are small (1.8- to 3.5-kDa) polycyclic peptides containing posttranslationally modified amino acids, such as ,-didehydroalanine, ,-didehydrobutyrine, lanthionine, and -methyllanthionine (30, 41). Other bacteriocins, the so-called nonlantibiotics (class II), consist of small (<13-kDa), heat-stable, cationic, and hydrophobic peptides containing only unmodified amino acids (31, 32). Both lantibiotics and nonlantibiotics are ribosomally synthesized as precursor peptides containing an N-terminal leader sequence, which is cleaved off concomitantly with export (23, 41). The leader sequences of most nonlantibiotics and some lantibiotics are of the double-glycine type. They are clearly related and may serve as a recognition signal for the processing and secretion apparatus (22, 52). The removal of double-glycine-type leader sequences and the translocation of bacteriocins across the cytoplasmic membrane are accomplished by dedicated ATP-binding cassette (ABC) transporters and their accessory proteins (23). Some bacteriocins, such as acidocin B (34), divergicin A (58), bacteriocin 31 (50), and enterocin P (7), have recently been shown to be exported by the general secretory pathway (GSP) (40). These GSP-dependent proteins contain a hydrophobic N-terminal extension, called the signal peptide, which consists of a positively charged N terminus, a hydrophobic core, and a cleavage site (18, 27, 54). Signal peptides are removed by a specific signal peptidase during protein translocation across the cytoplasmic membrane (40, 55).

Most bacteriocins secreted into the medium display full antimicrobial activity by themselves (the so-called one-peptide bacteriocins); however, two-peptide bacteriocins require the complementary action of two peptides for full activity. This group includes lactococcin G (39), lactococcin M (51, 53), lactacin F (1), plantaricin S (29), plantaricins EF and JK (3, 13), thermophilin 13 (36), and probably acidocin J1132 (49).

Recently, Enterococcus faecium L50 (previously identified as Pediococcus acidilactici L50) has been reported to produce an antimicrobial substance which inhibits growth of a broad spectrum of spoilage and food-borne pathogenic bacteria, including Listeria monocytogenes, Staphylococcus aureus, Bacillus cereus, Clostridium perfringens, and Clostridium botulinum (5, 6, 8). The data from the initial biochemical studies suggested that the antimicrobial activity arises from a single 5,250-Da thermostable peptide consisting of 41 amino acids, whose NH₂ terminus is blocked for sequencing by Edman degradation (5, 6). However, we show in this work that the antimicrobial activity arises from two highly similar, small, unmodified, cationic, and hydrophobic peptides, both closely related to the bacteriocin previously identified as enterocin L50.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, media, and growth conditions. E. faecium L50, E. faecium T136, Lactobacillus sake 148, P. acidilactici 347, and Lactococcus lactis subsp. cremoris CNRZ117 were cultured in MRS broth (Difco, Detroit, Mich.) at 30°C. Escherichia coli strains were propagated in Luria-Bertani (LB) broth (40) at 37°C with shaking (250 rpm). Transformants were selected on LB agar (1%) and propagated in LB broth, containing ampicillin (50 µg/ml) and/or chloramphenicol (34 µg/ml). All cultures were maintained and stored at 80°C in 20% glycerol.

Antimicrobial activity assays. Cell-free culture supernatants of E. faecium L50 and E. coli BL21 (DE3)(pLysS) (Novagen Inc., Madison, Wis.) were obtained by centrifugation at 12,000 × g for 10 min at 4°C and subsequent filter sterilization. Antimicrobial activity was assayed by spotting 20 µl of supernatants or 2 µl of the in vitro-synthesized peptides EntL50A and/or EntL50B onto MRS agar (1.5%) plates previously seeded with 10⁵ CFU of a fresh overnight culture of P. acidilactici 347 ml¹. Spotted plates were incubated at 30°C for 18 h before analyzing the results. A quantitative estimate of antimicrobial activity was obtained by using a microtiter plate assay (25). Each well of the microtiter plate contained 50 µl of twofold serial dilutions in MRS broth of the samples and 150 µl of a 400-fold-diluted fresh overnight culture of P. acidilactici 347. The plates were incubated at 30°C for 16 h, and growth inhibition of the indicator microorganism was measured spectrophotometrically at 620 nm with a microtiter plate reader (Titertek Multiskan Plus; Flow Laboratories, Helsinki, Finland). One antimicrobial unit was defined as the reciprocal of the highest dilution of the sample causing 50% growth inhibition (50% of the turbidity of the control culture without bacteriocin). To measure synergistic activities, in vitro-synthesized EntL50A and EntL50B were tested separately and combined in a 1:1 ratio against the indicator microorganisms listed in Table 1, by using the microtiter plate assay described above.

Plasmid DNA isolation and sequencing. Plasmid DNA of E. faecium L50 was obtained by the alkaline lysis method (2). The DNA sequence of a region of plasmid pCIZ1, encoding EntL50A and EntL50B, was determined by direct sequencing of several PCR products. Plasmid DNA was cleaved with DraI, EcoRI, EcoRV, HincII, NlaIV, PvuII, RsaI, ScaI, SspI, or StuI and ligated (T4 DNA ligase; Promega Corp. Madison, Wis.) to dephosphorylated pBluescript II SK+ (Stratagene, La Jolla, Calif.) digested with HincII or EcoRI. Restriction enzymes were obtained from New England Biolabs Inc. (Beverly, Mass.) and Promega and were used in accordance with the suppliers' instructions. Samples from these ligation reactions were subsequently used as templates to amplify pCIZ1 DNA segments near entL50A and entL50B. When the biotinylated polylinker primer SK2 (5'-CCGCTCTAGAACTAGTGG ATC-3') was used in combination with the degenerated primer L50-1 [5'-ATTAGGATCC(C/T)TGICCIAT(A/G)AA(C/T)TGCATIAT(C/T)TG-3'], two specific PCR products consisting of approximately 300 and 175 bp were obtained with the DraI ligation reaction as template. The sequence of primer L50-1 was derived from the published amino acid sequence of enterocin L50 (pediocin L50) obtained by Edman degradation (6). Oligonucleotides used for PCR and DNA sequencing were obtained from Kebo-Lab (Spanga, Sweden). The PCRs were carried out by denaturing template DNA at 97°C for 2 min followed by 35 cycles consisting of annealing at 50°C for 30 s, polymerization at 72°C for 2 min, and denaturation at 94°C for 1 min. Amplification reactions were performed by using the Gene Amp PCR kit and a DNA Thermal Cycler according to the supplier's instructions (Perkin-Elmer Cetus, Norwalk, Conn.). PCR products were analyzed by electrophoresis on 0.8 or 2.0% (wt/vol) agarose gels (1× Tris-acetate-EDTA buffer [pH 8.0]). Fragments to be sequenced were eluted from the agarose gel with a QIAEX II agarose gel extraction kit (Qiagen GmbH, Hilden, Germany) and further purified by using a QIAquick PCR purification kit (Qiagen). The strands of the two DNA fragments were separated with Dynabeads M-280 coupled to streptavidin as prescribed by the supplier (Dynal, Oslo, Norway). Sequencing reactions based on the dideoxynucleotide chain termination method of Sanger et al. (43) were performed on the single-stranded templates by using the Sequenase system (U.S. Biochemical, Cleveland, Ohio). Based on the sequence information obtained from these DNA fragments, new sequence-specific primers were synthesized and the procedure described above was repeated until the 3,560-bp pCIZ1 DNA sequence described in this report was determined.

Computer analysis of DNA and protein sequences. Analyses of DNA and protein sequences were performed by using the PC/Gene program package (version 6.8; Intelligenetics, Inc., Mountain View, Calif.) and the Genetics Computer Group package (version 8; University of Wisconsin, Madison).

Purification and amino acid sequencing of EntL50A and EntL50B. The antimicrobial peptides EntL50A and EntL50B were purified from a 1-liter E. faecium L50 culture which was grown in MRS broth at 30°C (optical density at 620 nm of approximately 0.6), in accordance with the chromatographic procedure previously described (6). The amino acid sequences were determined by Edman degradation with an Applied Biosystems (Foster City, Calif.) model 477A automatic sequence analyzer with an on-line 120A phenylthiohydantoin amino acid analyzer, as described by Cornwell et al. (9). Since the N termini of the enterocins were blocked, sequences were obtained only after cleavage with cyanogen bromide (CNBr) as described by Sletten and Husby (44).

Expression of entL50A and entL50B. A 405-bp PCR fragment carrying entL50A and entL50B was amplified from pCIZ1 DNA extracted from E. faecium L50 as outlined above. The following primers were used for amplification: L50AB-1 (5'-ATTCAATACATATGGGAGCAATCGCAAAATTAGTAGC-3') and L50AB-2 (5'-ATATAAGCTTGCGTTAAGCCGAATGTTTACACAAC-3'). The forward primer, L50AB-1, contained an in-frame ATG codon and an NdeI cleavage site (nucleotides underlined in sequence above). The reverse primer, L50AB-2, was located downstream of entL50B and included a HindIII site (nucleotides underlined in sequence above). The PCR product was cleaved with NdeI and HindIII and analyzed by agarose gel electrophoresis. The 390-bp NdeI-HindIII fragment was purified as described above and subsequently ligated into the NdeI and HindIII sites of the prokaryotic expression vector pRSETB (Invitrogen Corp., San Diego, Calif.) by using standard molecular techniques (42). The resulting recombinant plasmid (pRSETB-entL50AB) was chemically transformed into E. coli DH5 (Gibco BRL, Life Technologies, Inc., Grand Island, N.Y.) cells, as described by Inoue et al. (26). Transformants were grown in selective medium for 18 h, and plasmid DNA was prepared by the alkaline lysis method of Sambrook et al. (42). The recombinant plasmid was then chemically transformed into E. coli JM109 (DE3) (Promega) and E. coli BL21 (DE3)(pLysS) cells. Only the E. coli BL21 (DE3)(pLysS) cells gave transformants, indicating that the recombinant plasmid pRSETB-entL50AB cannot be maintained in E. coli JM109 (DE3). A 20-ml culture of transformed E. coli BL21 (DE3)(pLysS) cells, grown in a medium containing suitable antibiotics, was harvested at an optical density of 0.2. The culture was then washed twice to remove antibiotics, and cells were resuspended in fresh broth to 20 ml. IPTG (isopropyl--D-thiogalactopyranoside) was then added to a final concentration of 0.5 mM to induce expression of target EntL50A and EntL50B genes, and growth was continued for 2 h. A noninduced culture was used as a control. Aliquots of cultures were taken at appropriate intervals for measurements of optical density and antimicrobial activity.

In vitro transcription/translation of entL50A and entL50B separately or together. EntL50A and EntL50B were synthesized in vitro by using the E. coli T7 S30 Extract System for Circular DNA (Promega) as described by the manufacturer. The combined in vitro transcription/translation reactions were carried out at 37°C for 2 h. Recombinant plasmid pRSETB-entL50AB was used to express entL50A and entL50B together. A transcription/translation reaction using wild-type pRSETB as DNA template was also set up as a control.

Nucleotide sequence accession number. The nucleotide sequence reported in this publication has been submitted to the EMBL database under accession no. AJ223633.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

DNA sequence of the EntL50 locus. Southern analysis of chromosomal and plasmid DNA isolated from E. faecium L50 revealed that the enterocin L50 structural gene is carried on a 50-kb plasmid, pCIZ1 (5). In addition to pCIZ1, E. faecium L50 harbors a smaller plasmid, pCIZ2 (7.2 kb), which is phenotypically cryptic. Plasmid pCIZ1 was digested with several different restriction endonucleases, and the resulting restriction fragments were ligated into the vector pBluescript II SK+. Samples from these ligation reactions were subsequently used directly as templates to amplify regions of the pCIZ1 plasmid near the gene encoding enterocin L50. One PCR primer was complementary to the polylinker, whereas the other was a degenerated primer deduced from the amino acid sequence of enterocin L50 (6) or a primer based on new sequence information. By using this technique repeatedly, overlapping PCR products were obtained, and the nucleotide sequence of a contiguous 3,560-bp fragment was determined. Analysis of this sequence revealed the presence of six putative open reading frames (ORFs), designated orfA, orfB, orfC, orfD, orfE, and orfF, plus one incompletely sequenced ORF (orfG) (Fig. 1). Each complete ORF is preceded at an appropriate distance by a candidate ribosome binding site (results not shown). orfA and orfB are colinearly arranged and spaced by a 19-bp intergenic region. They encode small, cationic, and hydrophobic peptides with theoretical molecular masses of 5,190 Da (44 amino acids) and 5,178 Da (43 amino acids), respectively (Fig. 2). Comparison of the deduced amino acid sequences of these peptides and that of the bacteriocin previously referred to as enterocin L50 (6, 8) revealed that the sequence determined for enterocin L50 by Edman degradation was a mixture of the OrfA and OrfB sequences. We therefore termed their gene products enterocin L50A (EntL50A) and enterocin L50B (EntL50B), respectively. Alignment of the sequences of EntL50A and EntL50B (Fig. 3) shows that they share 31 amino acid residues, corresponding to 72% identity.

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Genetic organization and partial restriction map of the 3.5-kb pCIZ1 fragment containing the enterocin L50A and enterocin L50B structural genes (orfA and orfB) and surrounding ORFs. Candidate ribosome binding sites are indicated by black rectangles. ORFs within the IS1514 are represented by hatched arrows, and inverted repeats (IRR and IRL) are represented by hatched rectangles. Putative ORFs are represented by white arrows.

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Nucleotide sequence of a 466-bp fragment containing the structural genes of EntL50A (entL50A) and EntL50B (entL50B). The deduced amino acid sequences of EntL50A and EntL50B are shown below the DNA sequence. Putative ribosome binding sites (RBS) are overlined.

View larger version (24K): [in this window] [in a new window]   FIG. 3.   Amino acid sequence alignment of EntL50A, EntL50B, the antigonococcal substances (AGS-1, -2, and -3) secreted by S. haemolyticus (57), and the SLUSH peptides (SLUSH-A, -B, and -C) produced by S. lugdunensis (15). The alignment was performed by using the Genetics Computer Group programs Pileup and Prettybox. Identical residues are shown in a black background, and conservative substitutions are shown in a gray background.

Purification and amino acid sequencing of EntL50A and EntL50B. By using a purification procedure which includes ammonium sulfate precipitation and cation-exchange, hydrophobic-interaction, and reverse-phase liquid chromatographies, two purified fractions containing antimicrobial activity were obtained from a 1-liter culture of E. faecium L50 (results not shown). The procedure was essentially the same as the one originally used by Cintas et al. (6) to purify EntL50, but the last step (reverse-phase liquid chromatography) was rerun several times in an attempt to separate the two peptides. The fractions were subjected to Edman degradation, but since the N termini of both peptides turned out to be blocked, no sequence information was obtained. From the DNA sequence data, it was deduced that the sequences of EntL50A and EntL50B contained two methionine residues. Therefore, samples from the purified fractions were cleaved by CNBr. Subsequent sequencing revealed that one of them was pure and contained enterocin L50A, while the other contained a mixture of enterocin L50A and enterocin L50B. Furthermore, the results of the Edman degradation after CNBr cleavage gave the N-terminal amino acid sequence GAIAKLV, which demonstrates that the deduced gene products of entL50A and entL50B have been secreted without removal of any N-terminal extension (Fig. 2 and 3).

Coexpression of entL50A and entL50B in E. coli. A PCR fragment containing entL50A and entL50B was cloned into the NdeI and HindIII sites of the prokaryotic expression vector pRSETB (Invitrogen Corp.), under control of the strong bacteriophage T7 promoter (48). The recombinant plasmid (pRSETB-entL50AB) could not be established in E. coli JM109 (DE3), probably because small amounts of toxic bacteriocins were produced prior to induction. Even in the absence of IPTG, there is some expression of T7 RNA polymerase from the lacUV5 promoter in DE3 lysogens such as strain JM109 (DE3). For this reason, we selected another bacterial host [E. coli BL21 (DE3)(pLysS)] which contains a compatible plasmid (pLysS) that provides a small amount of T7 lysozyme. This bifunctional protein is a natural inhibitor of T7 RNA polymerase (47). Consequently, background expression can almost be abolished in DE3 lysogens harboring the pLysS plasmid. As expected, it was possible to establish the recombinant plasmid in the new host, indicating that small amounts of EntL50A and/or EntL50B produced in the cytoplasm are toxic to E. coli. Coexpression of entL50A and entL50B was induced by IPTG at 37°C. Within 60 min after induction, the optical density of the culture producing the bacteriocins dropped from 0.36 to 0.05, whereas the noninduced culture continued to grow normally. Cell-free supernatants of induced and noninduced cultures were assayed at regular intervals by a microtiter plate assay. Antimicrobial activity was first detected after 30 min and reached a maximum after approximately 1 h, in accordance with the observed rate of lysis of the E. coli host cells. No activity was detected in the cell-free supernatants of the noninduced culture.

In vitro transcription/translation of entL50A and entL50B. To investigate whether EntL50A and EntL50B have any antimicrobial activity by themselves, we cloned their structural genes into different pRSETB vectors, resulting in the constructs pRSETB-entL50A and pRSETB-entL50B. Due to the difficulties encountered when the two peptides are expressed together in vivo, we decided to prepare the material needed for further analyses by in vitro transcription/translation. By using the constructs pRSETB-entL50A, pRSETB-entL50B, and pRSETB-entL50AB in coupled in vitro transcription/translation reactions, the two bacteriocins were synthesized separately or together. The in vitro transcription/translation product containing both peptides exhibited strong antimicrobial activity against the indicator strain (P. acidilactici 347), as demonstrated by the sharp 14-mm-diameter inhibition zone observed in the spot-on-agar test (Fig. 4). Interestingly, in vitro-synthesized EntL50A and EntL50B also displayed antimicrobial activity by themselves (Fig. 4). The inhibition zone produced by the sample containing EntL50A was 13 mm in diameter, whereas the zone produced by the corresponding EntL50B sample was only 9 mm in diameter. In both cases, the inhibition zones were less sharp than the larger zone produced by both peptides. These results suggest that the two peptides together possess a greater antimicrobial effect than that exerted by the most active peptide acting alone (EntL50A). To confirm this synergistic effect, we employed a microtiter plate assay to determine the number of antimicrobial units in samples containing either EntL50A, EntL50B, or a mixture of both peptides (see Materials and Methods). The results presented in Table 1 conclusively show that when the two peptides are mixed in about a 1:1 ratio, their antimicrobial effect is much greater than the additive effect of the two bacteriocins acting independently. Furthermore, the degree of synergism varies considerably (4 to 73 times) depending upon which species of indicator bacterium is used. Since the same amount of plasmids (pRSETB-entL50A and pRSETB-entL50B) was added to each in vitro transcription/translation reaction mix, and since the translation products are very small, it is likely that the bacteriocins were synthesized at similar rates. If this is the case, EntL50A is 4 to 12 times more effective than EntL50B in killing the indicator bacteria.

View larger version (173K): [in this window] [in a new window]   FIG. 4.   Antimicrobial activity of in vitro-synthesized EntL50A and/or EntL50B against P. acidilactici 347 by a spot-on-agar test. Upper left spot, in vitro-synthesized EntL50A; upper right spot, in vitro-synthesized EntL50B; lower spot, in vitro-cosynthesized EntL50A and EntL50B.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The genetic and biochemical data presented in this report show that the antimicrobial activity previously assigned to enterocin L50 arises from two closely related, small, unmodified, cationic, and hydrophobic peptides, termed enterocin L50A and enterocin L50B (Fig. 3). At first glance, the enterocins seem to belong to a large group of antimicrobial compounds called class II bacteriocins. Class II bacteriocins are defined as small, cationic, hydrophobic, and heat-stable unmodified peptides with antimicrobial activity (31). This class includes bacteriocins consisting of a single polypeptide chain, as well as bacteriocins whose activity depends on the complementary action of two peptides. In some two-peptide systems, e.g., lactococcin G (39) and lactococcin M (53), both peptides are needed for activity. In others, e.g., plantaricin S (29), lactacin F (1), and thermophilin 13 (36), one or both peptides have some activity on their own but the combined action of both peptides enhances the activity much more than the additive effect of the peptides acting independently. EntL50A and EntL50B seem to have more in common with two-peptide bacteriocins of the last category. However, some characteristics of enterocin L50 peptides clearly distinguish them from class II bacteriocins. (i) The enterocins are secreted without a leader sequence or signal peptide. (ii) Apparently, the structural genes of the enterocins are not cotranscribed with a gene encoding an immunity protein. The immunity proteins of class II bacteriocins, whose genes are located on the same operon as the structural genes, specifically protect the producer bacteria against self-toxicity. (iii) The organization of the enterocin L50 locus is different from class II bacteriocin gene clusters, which in addition to the bacteriocin gene(s) include genes encoding proteins involved in immunity, processing, secretion, and regulation. (iv) The enterocins inhibit growth of a wide range of gram-positive bacteria, whereas most class II bacteriocins have a narrow antimicrobial spectrum. (v) Unlike the enterocins, class II bacteriocins are synthesized as inactive precursors (28, 38), presumably to protect the producer cells against the toxin while it is still in the cytoplasm.

All features considered, the enterocin L50 system seems to have more in common with members of a small group of antimicrobial and/or hemolytic peptides secreted by staphylococci than with class II bacteriocins. These peptides, the -lysin, three peptides with hemolytic activity (SLUSH A to C), and three peptides termed antigonococcal substances (AGS 1 to 3), are produced by Staphylococcus aureus (17, 33) Staphylococcus lugdunensis (15), and Staphylococcus haemolyticus (57), respectively. They are ribosomally synthesized, unmodified, secreted without leader sequences or signal peptides, and do not seem to be cotranscribed with an immunity protein gene (15, 17, 33). The antigonococcal substances have not been characterized at the genetic level, but their striking similarity to the SLUSH peptides (Fig. 3) strongly suggests that the characteristics listed above also apply to them. The -lysin, which is lytic to a wide range of eukaryotic cells, is a small peptide of only 26 amino acids which ruptures cells by forming pores selective for cations in their plasma membrane (17, 33). As far as we know, the -lysin has not been shown to possess antimicrobial activity. The SLUSH and AGS systems both consist of three closely related peptides (Fig. 3), which exhibit hemolytic activities on erythrocytes from various species (15, 57). In addition, the AGS peptides have antimicrobial activity, showing a broad antigonococcal spectrum and a narrow antibacterial spectrum (57). Preliminary data suggest that the SLUSH peptides also have the ability to kill bacteria. It has been shown hat crude extracts from hemolytic, but not nonhemolytic, S. lugdunensis strains inhibit growth of other staphylococci (15). However, since bacteria often produce multiple toxins, it is uncertain whether this antimicrobial activity should be ascribed to the SLUSH peptides. Considering the similar features of the enterocins and the staphylococcal toxins, we wondered if the enterocins EntL50A and EntL50B possess hemolytic activity as well. This hypothesis was tested by growing E. faecium L50 on calf, horse, porcine, and human blood agar plates. Furthermore, in vitro-synthesized EntL50A and EntL50B were spotted on corresponding blood agar plates, either separately or in combination. Both assays were negative, indicating that the enterocins have no hemolytic activity (data not shown). Evidence of a relationship between the enterocins and the SLUSH and AGS peptides was obtained by comparing their primary structures. This comparison revealed a low but significant homology between the enterocins and the staphylococcal peptides (Fig. 3). Together, amino acid sequence homology and other unique characteristics shared by the enterocins and the staphylococcal peptides strongly indicate that they constitute a separate, hitherto-unrecognized, family of peptide toxins.

Reminiscent of the SLUSH peptides, the genes encoding the enterocins are situated next to each other, presumably on a single transcription unit. The simplicity of this locus is striking compared to the organization of the gene clusters involved in production of class II bacteriocins. For instance, none of the ORFs neighboring the structural genes of the enterocins seem to be involved in their secretion. The secretion machinery of the staphylococcal hemolysins has not been identified, but it does not seem to be located close to the structural genes. Many larger bacterial proteins are known to be secreted without an N-terminal leader sequence or signal peptide. These include E. coli hemolysin (10, 16, 21), Pasteurella haemolytica leukotoxin (46), Erwinia chrysanthemi and Serratia marcescens proteases (11, 56), Bordetella pertussis cyclolysin (19), and the virulence plasmid-encoded proteins termed Yops (Yersinia outer membrane proteins) (35, 37). It has been established for most of them that secretion is mediated by specific ABC transporters (24), rather than by the secretion apparatus encoded by the GSP genes (4), and that a stretch of about 50 amino acids at their C termini targets them for export (35, 45). Even though no genes encoding a possible transport protein have been identified on the sequenced 3.5-kb fragment, we speculate that EntL50A and EntL50B are secreted by a dedicated ABC transporter, as demonstrated for the larger proteins discussed above.

Ribosomally synthesized prokaryotic proteins initially conain an N-terminal formylmethionine, which is usually removed together with the leader sequence or signal peptide in extracellular proteins. Both EntL50A and EntL50B are blocked for direct N-terminal amino acid sequencing by Edman degradation, indicating that the N-terminal methionine is modified or that the formylmethionine is retained. We favor the last possibility since it has been reported that about 90% of the -lysin molecules contain formylmethionine at their N termini after secretion (17). Microcin C7, another ribosomally synthesized antimicrobial peptide lacking a leader sequence or signal peptide, is also secreted without removal of the N-formyl group (20). In addition, it is known that this kind of modification will block Edman degradation.

For all the four species tested, the levels of antimicrobial units of EntL50A are much higher (4 to 12 times) than the corresponding values of EntL50B. There could be two reasons for this: (i) the in vitro transcription/translation reactions produce more EntL50A than EntL50B or (ii) EntL50A is more active against the bacteria tested than EntL50B. Since in vitro synthesis of the peptides was carried out in parallel with the same amount of plasmids, we consider it unlikely that quantitative differences alone can account for the much higher levels of antimicrobial units determined for EntL50A. If future investigations confirm that EntL50A is more active than EntL50B, domain swapping and site-directed mutagenesis could be used to pinpoint which amino acid residues make EntL50A the most active antimicrobial compound. Most likely, the C-terminal part of the molecule will be found to play a crucial role.

Elucidation of the mechanisms involved in the secretion, immunity, and mode of action of enterocins L50A and L50B, as well as identification of structural motifs required for their broad antimicrobial activity, may be helpful in the development of new antimicrobial compounds to be used in either the pharmaceutical or the food industry.