A {sigma}54-dependent PTS permease of the mannose family is responsible for sensitivity of Listeria monocytogenes to mesentericin Y105

Bacteriocins are antibacterial proteinaceous compounds produced by bacteria. In Gram-positive bacteria, they are divided into four classes (Klaenhammer, 1993 ). Within the class II, constituted by non-modified peptides produced mainly by lactic acid bacteria, bacteriocins of the subclass IIa, such as mesentericin Y105 (Héchard et al., 1992 ), are of particular interest (for a review see Ennahar et al., 2000 ). They are active against the foodborne pathogen Listeria monocytogenes and share a similar primary structure, with a conserved N-terminal motif (YGNGV). Subclass IIa bacteriocins induce membrane permeabilization (Abee, 1995 ; Maftah et al., 1993 ) of sensitive strains, but their target specificity and their molecular mode of action remain elusive. Whether a receptor or a docking molecule is necessary for activity of subclass IIa bacteriocins is still debated and, to date, such molecules have not been reported. We previously hypothesized that the transcriptional factor σ⁵⁴ (encoded by rpoN) could direct expression of a receptor since rpoN mutants of L. monocytogenes (Robichon et al., 1997 ) and Enterococcus faecalis (Dalet et al., 2000 ) are resistant to mesentericin Y105 and related subclass IIa bacteriocins. σ factors are subunits of the RNA polymerase holoenzyme involved in the initial step of transcription. Among them, σ⁵⁴ is unique since it targets conserved -24/-12 promoter sequences and requires an activator protein for transcription initiation, referred to as σ⁵⁴-associated activator (Morett & Segovia, 1993 ; Shingler, 1996 ). σ⁵⁴-associated activators are classically composed of three distinct domains respectively involved in signal reception, transcriptional activation and DNA binding. The central domain, which directs the transcriptional activation via σ⁵⁴ interaction, is the most highly conserved and clearly identifies these activators. To date, neither σ⁵⁴-regulated operons nor activator genes have been described in L. monocytogenes and, consequently, a direct link could not be established between σ⁵⁴ and sensitivity of L. monocytogenes to subclass IIa bacteriocins. Recently, two bacteriocin-resistant spontaneous mutants of L. monocytogenes have been linked to phosphotransferase systems (PTSs). In the first mutant, resistant to leucocin A, a two-dimensional SDS-PAGE protein analysis revealed absence of a IIAB subunit of a PTS permease (Ramnath et al., 2000 ). In the second mutant, resistant to pediocin PA-1, a ß-glucoside PTS permease was described to be overexpressed (Gravesen et al., 2000 ). PTSs are involved in both phosphorylation and transport of sugars (Postma et al., 1993 ; Saier & Reizer, 1994 ). A phosphate moiety is transferred from phosphoenolpyruvate (PEP) to the transported sugar via the PTS proteins EI, HPr and EII. The latter is a complex permease, made of three or four subunits which can be fused: IIA and IIB are cytoplasmic subunits responsible for phosphorylation, whereas IIC is an integral membrane subunit involved in sugar transport. A IID membrane subunit, associated with IIC, is found specifically in permeases of the mannose family. Interestingly, σ⁵⁴ has been described to be involved in PTS expression. Finally, we recently described that interruption of genes encoding a σ⁵⁴-associated activator (MptR) and a σ⁵⁴-dependent PTS permease of the mannose family (termed ) led to resistance of E. faecalis to mesentericin Y105.

To find a link between σ⁵⁴ and sensitivity of L. monocytogenes to mesentericin Y105, we searched for σ⁵⁴-associated activators and σ⁵⁴-dependent genes in the L. monocytogenes EGDe genome. We found an activator and a PTS permease of the mannose family that are required for sensitivity of L. monocytogenes to mesentericin Y105.

Bacterial strains and growth conditions.
L. monocytogenes EGDe and its derivatives were grown at 37 °C in brain heart infusion (BHI) or in LuriaBertani (LB) medium supplemented or not with various sugars that support L. monocytogenes growth (i.e. glucose, mannose, fructose and cellobiose). Escherichia coli XL-1 Blue, used for molecular cloning, was grown at 37 °C in LB medium with vigorous shaking. Erythromycin (5 µg ml^-1) or ampicillin (100 µg ml^-1) was added, as needed. Leuconostoc mesenteroides Y105, which produces the bacteriocin mesentericin Y105, was grown in MRS medium at 30 °C.

DNA manipulations and gene interruption.
Molecular cloning and DNA manipulations were performed as described by Sambrook et al. (1989) . Restriction and modification enzymes purchased from Life Technologies were used as recommended by the manufacturer. DNA fragments, used for gene interruption experiments, were amplified by PCR using Taq polymerase and specific primers bearing a HindIII site, as follows: manR primers, 5'-CTGCCAAGCTTGGAAGAACG-3' and 5'-CATCATCTTCCAAAGCTTGATCC-3'; mptA primers, 5'-GCTGAAGCTTTTTTGCAGTCCG-3' and 5'-GATGAGAAAGCTTCAATCAACATTGG-3'; mptD primers, 5'-CATCTCCAAAGCTTGGGGTAAC-3' and 5'-GGCGCAAGCTTGATATTTACCC-3'. A 1559 bp HindIII fragment, corresponding to the Tn917lac/rpoN junction of pRT758 (Robichon et al., 1997 ), was used for rpoN interruption. The latter DNA fragment and the PCR products were digested with HindIII and ligated at the same site in the erythromycin-resistant pHV1248ΔTn10 plasmid (Petit et al., 1990 ), giving rise to plasmids pLMK50 (rpoN), pLMK51 (manR), pLMK54 (mptA) and pLMK55 (mptD). These plasmids were used to create independent knockouts in rpoN, manR, mptA and mptD by homologous recombination with the L. monocytogenes EGDe chromosome, as previously described (Kocks et al., 1992 ). Several mutants from each transformation were analysed by Southern blotting of chromosomal DNA digested by HindIII and hybridized with probe labelled by random priming from the PCR products described above (Sambrook et al., 1989 ).

Allelic exchange of mptD.
Allelic exchange of mptD was achieved by integration-excision of a plasmid bearing a deleted fragment of mptD. This fragment was obtained by PCR as follows. One gene fragment from each side of the additional domain we wanted to delete was amplified with the following primers: Del1 (5'-ATGAAGCTTTTCAAGGGGTTAAAGT TGG-3') and Del2 (5'-GCAAGGTTGTTACTTTAATTTCCGCACCTTCATCAAGTTTAACTTTCG-3') together and Del3 (5'-CGAAAGTTAAACTTGATGAAGGTGCGGAAATTAAAGTAACAACCTTGC-3') and Del4 (5'-ACGTAAGCTTTAAGTCCAGTATACGC-3') together. The resulting PCR products, overlapping by 25 bp, were then used in a PCR reaction, leading to a 84 bp deleted fragment of mptD (Δ₆₅₅₇₃₈mptD). This fragment was first cloned in pGEM-T (Promega), confirmed by sequencing and then subcloned as a HindIII fragment in pHV1248ΔTn10. The resulting plasmid, pLMY2, was used to achieve integration in mptD by homologous recombination. Excision events were then screened by their erythromycin sensitivity. Finally, erythromycin-sensitive clones were tested for the presence of the required deletion by PCR with primers Del1 and Del4 and sequencing. This gave rise to the strain EGY2 (EGDe-Δ₆₅₅₇₃₈mptD).

Bacteriocin purification and assays.
Mesentericin Y105 was purified as reported (Guyonnet et al., 2000 ). Nisin, a class I bacteriocin, was purchased from Sigma. L. monocytogenes sensitivity was assayed by spot-on-lawn or microtitre plate tests. The former was achieved by overlaying a BHI agar (1·5%) plate with a BHI agar lawn (0·7%) previously inoculated with 1% L. monocytogenes. Purified bacteriocin was then spotted on the lawn, the plate was incubated overnight at 37 °C and zones of inhibition were recorded. The microtitre plate tests were conducted as follows. Bacteria were grown overnight in LB medium and inoculated in 1 ml fresh LB medium to a final OD₆₂₀ between 0·01 and 0·03. The culture was supplemented or not with either fructose, glucose, mannose or cellobiose at 2 g l^-1. Four aliquots of 200 µl from each sample were then distributed in a microtitre plate. Plates were incubated at 37 °C with agitation at 120 r.p.m. and bacterial growth was monitored by measurement of the OD₆₂₀. Purified mesentericin Y105 (100 ng) was added after 2 h, when the culture had reached an OD₆₂₀ between 0·05 and 0·1.

Transcription analysis.
L. monocytogenes EGDe and its derivatives were grown in 3 ml LB medium supplemented or not with glucose, mannose, fructose or cellobiose (2 g l^-1) to an OD₆₀₀ of 0·6. The bacterial pellets collected by centrifugation were resuspended in 100 µl lysis buffer (10 mg lysozyme ml^-1, 10 mM Tris, 1 mM EDTA, pH 7·5) and incubated for 1 h at 37 °C. Total RNA was then extracted with the RNAwiz reagent (Ambion) and treated as recommended. The RNA pellets were finally resuspended in 50 µl diethylpyrocarbonate-treated water containing 20 U RNase-free DNase I. Slot-blot hybridization was performed as described by Sambrook et al. (1989) . A ³²P-labelled mptD probe was prepared by random priming from the PCR product obtained with the mptD primers described above. Radioactivity was measured with an Instant Imager apparatus (Packard).

DNA sequencing.
Cycle sequencing was achieved with the ABI Prism BigDye terminator cycle sequencing ready reaction kit (Perkin-Elmer) and analysed with the ABI Prism 310 genetic analyser.

σ⁵⁴-associated activators and σ⁵⁴-dependent genes in the EGDe genome
We used the central domain of σ⁵⁴-associated activators to screen the L. monocytogenes EGDe genome. Three ORFs, encoding putative σ⁵⁴-associated activators, were found in BLAST searches. One ORF has similarity (35% identity) with activators of the NifA/NtrC family whereas the two others display similarity (31 and 38% identity) with activators of the LevR family. Inactivation of each activator gene was performed; only one, named manR, was further studied since, in contrast to the others (data not shown), its inactivation led to resistance of L. monocytogenes EGDe to mesentericin Y105 (see below). The manR gene encodes a putative 938 aa protein (GenBank accession number AF397144), which displays highest similarity with LevR of Bacillus subtilis (38% identity) (Débarbouillé et al., 1991 ) and MptR of E. faecalis (40% identity) (Héchard et al., 2001 ). According to Prosite, ManR has classical motifs found in other σ⁵⁴-associated activators: an ATP/GTP-binding site motif A (position 144151) and a σ⁵⁴-interaction domain (position 209224). Interestingly, ManR also bears a 10 aa sequence (position 217226, see Fig. 1) integrally matching the classical DEAH motif, usually found in helicases (Luking et al., 1998 ). Because helicase activity could explain the energy coupling of DNA unwinding in σ⁵⁴-dependent transcription, such similarities between σ⁵⁴-associated activators and helicases were searched for but not yet found (Buck et al., 2000 ). In addition, we found a DEAH motif, differing in only one or two positions (see Fig. 1) in other σ⁵⁴-associated activators. These observations suggest that ManR and other σ⁵⁴-associated activators could harbour both an ATPase and a helicase activity, allowing initiation of transcription. Finally, ManR has two PTS regulation domains (PRD-I and PRD-II in positions 501566 and 863929, respectively), described in LevR of B. subtilis to be involved in regulation of its activity by PTS components.

(42K): Fig. 1. Alignment of helicase DEAH domains found in several σ54-associated activators. Only the residues displayed in lower-case characters deviate from the consensus pattern.

LevR controls the expression of the lev operon encoding a PTS permease of the mannose family (i.e. with a IID subunit). Consequently, we searched, in the L. monocytogenes EGDe genome, for similar operons harbouring a consensus -24/-12 promoter sequence. Two operons, mpo (mannose permease one) and mpt (mannose permease two), encoding putative mannose PTS permeases were found and subsequently interrupted. The mpo operon is located just downstream of manR whilst mpt was found elsewhere on the chromosome. In contrast to interruption of the mpo operon (data not shown), knockout of the mpt operon led to full resistance of L. monocytogenes to mesentericin Y105 (see below). The mpt operon (GenBank accession number AF397145) is delineated by two hairpin loops, T1 and T2, with a calculated free energy of -18 and -16 kcal mol^-1 (75·3 and 66·9 kJ mol^-1), respectively. The mpt operon is composed of three ORFs, mptA, mptC and mptD (Fig. 2). They putatively encode the proteins MptA (321 aa), MptC (268 aa) and MptD (303 aa), which share high similarities with subunits of PTS permeases of the mannose family (i.e. IIAB^Man, IIC^Man, IID^Man, respectively). MptA, MptC and MptD thus compose a putative PTS permease, named .

(4K): Fig. 2. Genetic organization of the mpt operon. The -24/-12 denotes a σ54 promoter. The black rectangle represents the DNA fragment deleted in Δ655738mptD. T1 and T2 indicate putative transcription terminators.

ManR and are involved in sensitivity to mesentericin Y105
rpoN, manR, mptA and mptD were interrupted by recombination between plasmids pLMK50, pLMK51, pLMK54 and pLMK55, respectively, and the chromosome of L. monocytogenes EGDe. The corresponding derivative strains were named EGK50, EGK51, EGK54 and EGK55. Southern blots with specific homologous probes verified gene interruption in these derivative strains (data not shown). Sensitivity of L. monocytogenes EGDe and its derivatives to purified mesentericin Y105 was then assessed by a spot-on-lawn assay. As expected, L. monocytogenes EGDe was sensitive down to 0·5 ng mesentericin Y105. By comparison, the mutant strains EGK50, EGK51, EGK54 and EGK55 (interrupted in rpoN, manR, mptA, mptD, respectively) were fully resistant, even to 250 ng mesentericin Y105. Accordingly, the mutations increased the resistance at least 500-fold. All these mutants remained sensitive to nisin, a class I bacteriocin, underlying a specificity in resistance to mesentericin Y105. These results indicate that σ⁵⁴, ManR and are involved in sensitivity of L. monocytogenes EGDe to mesentericin Y105, as previously shown for σ⁵⁴ of L. monocytogenes L028 (Robichon et al., 1997 ).

MptD plays a particular role in sensitivity
The distal gene of the mpt operon, mptD, seems to play a particular role in sensitivity since its interruption led to resistance. Interestingly, MptD bears an additional domain compared to three other IID^Man subunits found in the L. monocytogenes EGDe genome and to most of IID^Man subunits described in the literature except E. faecalis (Héchard et al., 2001 ) and Streptococcus salivarius (Lortie et al., 2000 ) or found in GenBank (Fig. 3). Among 22 bacterial sequenced genomes (finished or unfinished) found to possess at least one orthologue of IID^Man, only several Gram-positive bacteria (i.e. E. faecalis, Streptococcus spp. (5 examples), Lactococcus lactis and Clostridium acetobutylicum) have a IID^Man with an additional domain. Interestingly, these Gram-positive bacteria have sometimes been described to be sensitive to subclass IIa bacteriocins. The mutant strain EGY2 has an 84 bp in-frame deletion of mptD (Δ₆₅₅₇₃₈mptD), created by allelic exchange (see Methods). It likely encodes the Δ₂₁₉₂₄₆MptD protein with a 28 aa deletion in the additional domain (bold characters in Fig. 3). Strain EGY2 was tested for sensitivity to mesentericin Y105, as described above. It was fully resistant, indicating that the presence of the additional domain is of primary importance for sensitivity.

(19K): Fig. 3. Partial alignment of IID subunits of PTS permeases. Listeria monocytogenes (Lmo), Enterococcus faecalis (Efa), Steptococcus salivarius (Ssa), Lactobacillus casei (Lca), Bacillus subtilis (Bsu) or Escherichia coli (Eco). Residues in bold characters are those of the additional domain deleted in mutant Δ219246IID2Man.

Expression of the mpt operon is controlled by ManR and glucose or mannose
Slot-blot analysis was first performed with total RNA of L. monocytogenes EGDe grown in LB medium supplemented or not with either cellobiose, fructose, glucose or mannose at 2 g l^-1. Expression of the mpt operon was followed with the mpt probe. Fig. 4 shows that expression of the mpt operon is highly induced in the presence of glucose or mannose. This indicates that the permease, putatively encoded by the mpt operon, is responsible for glucose and mannose transport.

(13K): Fig. 4. Expression of the mpt operon assayed by slot-blot analysis. L. monocytogenes EGDe was grown in LB medium supplemented or not with cellobiose (Cel), fructose (Fru), glucose (Glu) or mannose (Man). Radioactivity is expressed in c.p.m. per mm2.

Slot-blot analysis, using the same mptD probe, was performed with total RNA of L. monocytogenes EGK50 and EGK51 grown in LB medium supplemented with mannose at 2 g l^-1. The mpt operon was not expressed in these derivative strains (data not shown), showing that σ⁵⁴ and ManR are responsible for its expression.

Mannose and glucose influence sensitivity of L. monocytogenes to mesentericin Y105
L. monocytogenes EGDe was grown in LB medium supplemented with glucose, mannose, fructose or cellobiose at 2 g l^-1. Fig. 5(a) shows that, in the absence of mesentericin Y105, no significant difference in L. monocytogenes EGDe growth curves could be observed, whereas the presence of mesentericin Y105 affected L. monocytogenes growth in a medium supplemented with mannose or glucose but not with cellobiose or fructose. These results show that glucose and mannose specifically induce sensitivity of L. monocytogenes to mesentericin Y105.

(16K): Fig. 5. Growth curves of L. monocytogenes EGDe in LB medium supplemented with various sugars: (a) cellobiose (•), fructose (), glucose () or mannose () at 2 g l-1; (b) mannose 0·125 g l-1 (), 0·5 g l-1 (), 2 g l-1 (•). Bacterial growth without (unbroken line) or with mesentericin Y105 (dotted line) was followed by measurement of optical density at 620 nm. The arrow indicates the addition of 0·5 ng mesentericin Y105 µl-1.

To see whether mannose or glucose influences the sensitivity in a dose-dependent manner, L. monocytogenes EGDe was grown in LB medium supplemented with mannose at various concentrations (2, 0·5 and 0·125 g l^-1, respectively). Fig. 5(b) shows that, in the absence of mesentericin Y105, growth of L. monocytogenes increases along with mannose concentration whereas, in the presence of mesentericin Y105, growth inhibition increases along with mannose concentration. These results show that mannose (and glucose, data not shown) has a dose-dependent effect on L. monocytogenes sensitivity to mesentericin Y105. It underlines that the level of sensitivity is tightly linked to sugar availability, suggesting that mannose or glucose directly causes expression of a molecule responsible for sensitivity to mesentericin Y105. Mesentericin Y105 and other subclass IIa bacteriocins have been described to permeabilize the membrane of target strains. However, the main unanswered question is whether or not these bacteriocins need a docking molecule or receptor, as described for nisin, a pore-forming bacteriocin (Breukink et al., 1999 ). Since σ⁵⁴ has been described to direct sensitivity of L. monocytogenes and E. faecalis to subclass IIa bacteriocins, we thus hypothesized that it could be responsible for the expression of such a receptor.

We demonstrate here that the σ⁵⁴ regulon of L. monocytogenes is clearly involved in sensitivity to mesentericin Y105, bringing experimental support to assessments arising from our earlier work on E. faecalis (Héchard et al., 2001 ). Interruption of either rpoN, manR, mptA or mptD (encoding, respectively, σ⁵⁴, a σ⁵⁴-associated activator and two subunits of the permease) led to resistance of L. monocytogenes EGDe. , a PTS permease of the mannose family, is encoded by the mpt operon. This operon, which bears a -24/-12 promoter, was not expressed in the rpoN and manR mutants, showing that its expression is positively controlled by σ⁵⁴ and ManR. In E. faecalis, the mpt operon also bears a putative -24/-12 promoter although it was not experimentally demonstrated to be regulated by MptR or σ⁵⁴ (Héchard et al., 2001 ). The localization of the mpo operon immediately downstream from manR suggests that it could also be controlled by ManR and preliminary results indicate a possible cross-regulation between the mpo and mpt operons. Accordingly, ManR likely controls these two operons whereas MptR of E. faecalis presumably controls expression of mpt only. Relationships between these operons would not be surprising according to the known regulation of carbohydrate metabolism.

The presence of glucose or mannose induced sensitivity of L. monocytogenes and E. faecalis to mesentericin Y105. These sugars also induced expression of the mpt operon of L. monocytogenes (not shown in E. faecalis), indicating that transports glucose and mannose in accordance with previous observations showing a specific inducible effect of the transported sugar on PTS permease expression (Postma et al., 1993 ). These correlated results suggest that the level of expression is directly linked to sensitivity of L. monocytogenes to mesentericin Y105.

Since mesentericin Y105 and related subclass IIa bacteriocins have a narrow spectrum of inhibition, we are wondering about the specificity of the target strains. The IID^Man membrane subunit of , MptD, contains an additional domain compared to most other IID^Man proteins. The mutant EGY2, which putatively encodes a truncated MptD protein, i.e. lacking 28 aa in this additional domain, became resistant to mesentericin Y105. Assuming that the truncated MptD protein is expressed and remains functional, it strongly suggests a primary role of this domain in the sensitivity of L. monocytogenes. The introduction of point mutations in the additional domain constitutes the next step of our work to confirm the role of this domain and to identify the implicated amino acids as well as their possible interaction.

The involvement of in sensitivity to subclass IIa bacteriocins is emphasized by the report of a spontaneous mutant resistant to leucocin A, a subclass IIa bacteriocin (Ramnath et al., 2000 ). The authors clearly showed the absence of expression of a IIAB^Man PTS component in this mutant. Moreover, the N-terminal sequence of the protein shares high identity (17 of 20 residues) with the MptA protein described here. In addition, Gravesen et al. (2000) recently reported the overexpression of a ß-glucoside PTS in a spontaneous L. monocytogenes mutant resistant to pediocin PA-1, another subclass IIa bacteriocin. This pediocin PA-1 resistant mutant was then shown to be defective in expression (Y. Héchard, unpublished results) and the ß-glucoside PTS was shown to be overexpressed in our L. monocytogenes LUT758 mutant lacking σ⁵⁴ (A. L. Gravesen, personal communication). In S. salivarius, mutants that lack the synthesis of IIAB_L^Man (similar to MptA) are derepressed for several genes, such as the ß-galactosidase gene (Gauthier et al., 1990 ). We thus speculate that overexpression of the ß-glucoside PTS is a consequence of the lack of expression. Taken together, these results are evidence that is a key component for sensitivity of L. monocytogenes to different subclass IIa bacteriocins.

In conclusion, we propose that could either influence the expression of an unknown molecule involved in sensitivity or, via its IIC^ManIID^Man membrane complex, be a docking molecule or a receptor for mesentericin Y105 and other subclass IIa bacteriocins. Since deletion of the additional domain of MptD led to resistance, we propose that this domain could directly interact with bacteriocins or that its deletion could change the structure of the permease, leading to a lower affinity for the bacteriocins. Finally, a IIC^ManIID^Man complex has already been described to facilitate penetration of phage lambda DNA across the inner membrane of Escherichia coli (Esquinas-Rychen & Erni, 2001 ). It would be interesting to see whether both phage and bacteriocin could interact with cells via a similar mechanism.

The authors thank Philippe Glaser for his precious help. This work was partly supported by a partnership with the Rhodia Food company. Karine Dalet is supported by a fellowship from the Région Poitou-Charentes.