Lacticin 3147 is a two-peptide lantibiotic produced by Lactococcus lactis in which both peptides, LtnA1 and LtnA2, interact synergistically to produce antibiotic activities in the nanomolar concentration range; the individual peptides possess marginal (LtnA1) or no activity (LtnA2). We analysed the molecular basis for the synergism and found the cell wall precursor lipid II to play a crucial role as a target molecule. Tryptophan fluorescence measurements identified LtnA1, which is structurally similar to the lantibiotic mersacidin, as the lipid II binding component. However, LtnA1 on its own was not able to substantially inhibit cell wall biosynthesis in vitro; for full inhibition, LtnA2 was necessary. Both peptides together caused rapid K+ leakage from intact cells; in model membranes supplemented with lipid II, the formation of defined pores with a diameter of 0.6 nm was observed. We propose a mode of action model in which LtnA1 first interacts specifically with lipid II in the outer leaflet of the bacterial cytoplasmic membrane. The resulting lipid II:LtnA1 complex is then able to recruit LtnA2 which leads to a high-affinity, three-component complex and subsequently inhibition of cell wall biosynthesis combined with pore formation.
Lacticin 3147 is a two-peptide bacteriocin produced by Lactococcus lactis ssp. lactis DPC3147 which was first isolated from an Irish kefir grain and which has broad spectrum activity against Gram positive bacteria (Ryan et al., 1996). Among the bacteriocins, lacticin 3147 belongs to the Class 1 lantibiotics, for which intramolecular rings formed by the thioether amino acids lanthionine and β-methyllanthionine are characteristic features (Gross and Morell, 1971). As with several other bacteriocins, both modified and unmodified, lacticin 3147 is unusual in that two peptides, LtnA1 and LtnA2, are required for full antimicrobial activity. When both peptides are present, lacticin 3147 was found to elicit a range of effects in target cells through formation of pores in the cell membrane. Rapid efflux of K+ and phosphate ions was observed causing immediate dissipation of the membrane potential, hydrolysis of intracellular ATP and ultimately cell death (McAuliffe et al., 1998). Membrane depolarization has been described for many other small bacteriocins, including two-peptide non-lantibiotic systems, and is generally found to be an intrinsic part of the antibiotic activity of cationic amphiphilic peptides which are important factors of innate immune responses.
Such peptides have been shown to impair microbial cytoplasmic membranes when applied in μM concentrations and membrane disruption models, e.g. the toroidal pore or carpet model, may adequately describe such activities (for a review see Zasloff, 2002). However, many bacteriocins of Gram positive bacteria are active in nanomolar concentrations, suggesting specific target-mediated modes of action. Such a target has been recently identified for some lantibiotics which act through binding of the membrane-bound cell wall precursor lipid II (for a recent review see Hechard and Sahl, 2002). Lipid II is the substrate of the cell wall biosynthesis machinery which mediates the transport of disaccharide-pentapeptide units from the cytoplasm to the outside of the cell, where it is incorporated into the growing peptidoglycan network. Its central function makes lipid II aprominent target for antibiotics such as the glycopeptides or the globular type-B lantibiotic mersacidin (Brötz et al., 1997) which mainly inhibit cell wall biosynthesis by sequestration of the precursor. Nisin and the closely related epidermin are both typical type-A lantibiotics with an elongated and flexible structure which had been shown to primarily permeabilize bacterial membranes (Ruhr and Sahl, 1985). In their case it became clear that lipid II is used as a docking molecule for the formation of defined and stable pores (Brötz et al., 1998), thus providing these lantibiotics with dual modes of action, i.e. inhibition of cell wall biosynthesis and target-mediated pore formation (Breukink et al., 1999; Wiedemann et al., 2001).
When the primary structures of the individual lacticin peptides were recently elucidated by means of multidimensional NMR spectroscopy, it became obvious that lacticin A1 has a lanthionine bridging pattern which resembles that of the globular type-B lantibiotic mersacidin, whereas the A2 peptide is a member of the elongated type-A lantibiotic class. These structural similarities (Fig. 1) led to the hypothesis that lacticin A1 may interact with the cell wall precursor lipid II and inhibit cell wall biosynthesis, while lacticin A2 may form a membrane-spanning pore (Martin et al., 2004). Here, we provide evidence that the activity of lacticin 3147 indeed involves binding of LtnA1 to lipid II. However, both activities, pore formation and inhibition of cell wall biosynthesis, require the presence of both peptides whose intermolecular interactions appear to be stabilized by lipid II.
Pore formation by lacticin 3147 in whole cells
After treatment of a lactococcal cell suspension with lacticin 3147 for up to 30 min, increased levels of K+ had been found in the supernatant, indicating that lacticin may impair the integrity of the cytoplasmic membrane and induce K+ leakage (Morgan et al., 2005). For kinetic studies, we used a potassium electrode-based method that allows online monitoring of the lacticin 3147-induced K+ release from Lactococcus lactis HP cells after peptide addition (Fig. 2A). Lacticin pore formation was concentration dependent and potassium release started already at low concentrations, i.e. with 0.05 μM or 0.1 μM. Application of higher concentrations, i.e. 0.5 μM, resulted in a more rapid potassium efflux that reached a plateau at 90–100% after only 1–2 min. When the individual peptides, i.e. lacticin A1 and lacticin A2 were added alone at a concentration as high as 1 μM, no leakage of potassium occurred.
When lacticin A1 and A2 were either added together or when cells were pre-incubated with A2 for 3 min followed by addition of lacticin A1, potassium release only started after a lag period of 30–50 s; the efflux rates and the plateau reached were comparable (Fig. 2B). In contrast, when cells were pre-incubated with A1 followed by addition of A2 potassium release was induced immediately and was slightly faster, indicating that when the A1 peptide is bound first, subsequent pore formation by A2 is accelerated. These results demonstrate that the specific interaction of both peptides is necessary for the maximal activity leading to the formation of membrane pores. When both lacticin peptides were first incubated together in the presence of purified lipid II prior addition to the cells, the pore-forming activity was almost completely inhibited, strongly indicating that lipid II is a specific target for lacticin 3147.
Pore formation by lacticin 3147 in liposomes
The requirements for the pore formation process induced by lacticin 3147 were further analysed by using carboxyfluorescein (CF)-loaded liposomes. The highest marker release of about 80% were obtained from 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) liposomes containing 0.1 mol% purified lipid II after addition of 0.5 and 1 μM lacticin A1/A2 (Fig. 3). In contrast, liposomes made of pure DOPC or of DOPC with 0.1 mol% C55-P, the bactoprenol carrier lacking the disaccharide-pentapeptide moiety of lipid II, were significantly less affected, i.e. the addition of lacticin at 0.5 μM induced release levels of about 30% and at 1 μM 50% respectively. When 50% negatively charged phospholipids were incorporated [DOPC:DOPG (1,2-dioleoyl-sn-glycero-3-phosphoglycerol); 1:1 molar ratio], liposomes were less susceptible than those doped with lipid II at peptide concentrations up to 0.1 μM. However, this difference disappeared at higher concentrations of 0.5 μM and 1 μM. The individual lacticin peptides, as observed in the potassium release assay with whole cells (Fig. 2), did not induce significant marker release from all kinds of liposomes tested (Fig. 3).
These results indicate that the pore formation of lacticin is significantly enhanced in the presence of lipid II, especially at concentrations of 0.1 μM, and lower. As this effect was not achieved with C55-P, the carbohydrate moiety of lipid II appears to be involved in the interaction with lacticin 3147. At higher peptide concentrations, i.e. above 0.1 μM, similar release rates as compared with the lipid II containing liposomes were achieved by incorporation of 50% negatively charged phospholipids in the target membrane. However, the lipid II effect cannot be simply attributed to a charge effect, because incorporation of 0.1 mol% C55-P in the membrane was not sufficient to enhance pore formation.
Pore formation by lacticin 3147 in asymmetric membranes
In order to analyse the role of lipid II in the lacticin-induced pore formation process in more detail, we performed electrical measurements on planar bilayer membranes with both leaflets being composed of an equimolar mixture of DOPG and DOPC. One leaflet of the bilayer, the PL+ side, was supplemented with 1 mol% (with respect to the total amount of phospholipids) of the purified cell wall precursor lipid II. The opposite side, lacking lipid II, is labelled as PL–.
When lacticin A1/A2 (2.5 μg of each peptide corresponding to 0.5 μM A1 and 0.58 μM A2) was added to the PL– side of PL–/PL+ membranes, no increase in membrane current was observed even after 20 min of incubation and at voltages up to ±100 mV (Fig. 4A, a). Subsequent addition of the same amount of lacticin A1/A2 to the PL+ side of the membrane led to an increase in membrane current already at a clamp voltage of −10 mV (Fig. 4A, b) confirming that lacticin uses lipid II for targeted pore formation. The same effect was observed, when lacticin A1/A2 was added only to the PL+ side (data not shown). Pore formation was only observed, when the polarity of the applied voltage was trans-negative, corresponding to the situation in living bacterial cells.
The addition of 5 μg of the individual lacticin A2 peptide alone to the PL+ side of a PL+/PL– membrane had no effect on membrane conductance (Fig. 4B, a). After subsequent addition of 5 μg lacticin A1 to the same side of the membrane, an increase in membrane current was observed, indicating that the A2 peptide is inactive in the absence of the A1 peptide (Fig. 4B, b). In contrast, the addition of 5 μg lacticin A1 alone to the PL+ side also caused some increase in membrane current (data not shown) as observed with the liposome system.
To determine the size of the pores induced by lacticin A1/A2 the early period of pore formation was recorded as depicted in Fig. 5A. The inset shows that lacticin A1/A2-induced pores have a lifetime in the range of milliseconds. An amplitude histogram was derived from this current trace (Fig. 5B). The peak positions (grey lines) in the histogram were obtained from Gauss fits, and the single channel current was found to be 8.2 ± 0.4 pA. Assuming a cylindrical pore geometry and a membrane thickness of 6 nm, the diameter of a pore was calculated to be ∼0.6 nm.
Furthermore, the macroscopic I/U-characteristics of lacticin A1/A2-induced pores were determined. A typical I/U-trace for two subsequent cycles of positive and negative clamp voltages (depicted in black and grey) is shown in Fig. S1. At positive polarity of the clamp voltage, the I/U-characteristic exhibits an ohmic behaviour up to a voltage of ∼70 mV. When applying voltages >70 mV the current increases super proportionally, indicating that more pores were formed. With decreasing positive voltages, the formation of additional pores stops, and the I/U-characteristic becomes ohmic again. At negative clamp voltages, the I/U-characteristics behave ohmic, indicating that no further pores were formed.
From these results we conclude that (i) lipid II serves as a specific target molecule for lacticin 3147 pore formation; (ii) in the absence of lipid II in the phospholipid layer lacticin 3147 cannot translocate through the membrane to reach the lipid II molecules incorporated in the layer facing the opposite compartment; (iii) lipid II cannot move spontaneously from one leaflet into the other (no ‘flip-flop’).
Inhibition of in vitro lipid II synthesis by lacticin 3147
The interaction of lacticin A1 and/or A2 with lipid II implies that, in addition to the targeted pore formation, the ongoing cell wall biosynthesis should be blocked by lacticin (as had been demonstrated for other lipid II targeting lantibiotics such as nisin and epidermin (Wiedemann et al., 2001; Bonelli et al., 2006). We analysed the inhibitory effect of lacticin 3147 on the membrane-bound steps of peptidoglycan formation (Fig. 6), using isolated micrococcal membranes, supplemented with C55-P to which the addition of MurNAc-pentapeptide results in the formation of lipid I, and the addition of the second sugar GlcNAc yields lipid II. The individual lacticin peptides, lacticin A1 and lacticin A2 inhibited the conversion of the substrate C55-P to lipid II to a similar extent, i.e. lipid II synthesis was reduced from 100% in the control assay to about 70% and 75% respectively. As it had been shown recently that A2 does not bind to whole cells (Morgan et al., 2005), the inhibitory effect of A2 in this system was rather unexpected; however, the highly amphiphilic A2 may cause a detergent-like effect at the concentrations applied (67 μM) in this test. In contrast to the individual peptides, when lacticin A1 and A2 were combined, the amount of synthesized lipid II was clearly reduced to only 15%. The extent of inhibition induced by lacticin was comparable to the effect observed for nisin when applied in the same concentration indicating that lacticin 3147 (A1 and A2) obviously binds tightly to lipid I.
Tryptophan fluorescence spectroscopy
The lacticin A1 peptide contains three intrinsic Trp residues, which enabled us to use fluorescence spectroscopy to study the interaction of the A1 peptide with different types of liposomes. The Trp residues in the A1 peptide are located in ring A at position 1 and 18 and in ring C at position 28.
The fluorescence emission spectra of lacticin A1 in buffer is shown in Fig. 7A. The maximum fluorescence emission was detected at 354 nm, which is typical for Trp in a polar environment (Surewicz and Epand, 1984). In general, a blue shift of the emission maximum and an increase of the fluorescence intensity are indicative of a more hydrophobic environment of the Trp side-chains.
The maximum emission of the A1 peptide in buffer was unaltered when it was combined with the lacticin A2 peptide, which lacks Trp residues, indicating that the peptides do not interact in aqueous solution in a manner that affects the environment of the Trp residues (Fig. 7A). In the presence of DOPC or DOPC/DOPG-liposomes the emission maxima of lacticin A1 changed only marginally (blue shift of 4 nm) (Fig. 7B and C). However, after subsequent addition of lacticin A2 a significant blue shift of 12 nm and an increase in fluorescence intensity was observed for the DOPC/DOPG liposomes indicating that the A2 peptide causes a deeper insertion of the A1 peptide into the membrane (Fig. 7C).
A different picture emerged when the DOPC liposomes were supplemented with 1 mol% lipid II (Fig. 7D). Already the spectrum of individual A1 peptide significantly changed, i.e. a net blue shift of the emission maximum of 6 nm accompanied by an increase in fluorescence intensity. The fluorescence intensity increased even more when both lacticin peptides were present and here we also observed a larger blue shift of 14 nm. These changes were not observed in the presence of DOPC liposomes supplemented with C55-P indicating that the lacticin A1 peptide specifically interacts with the carbohydrate-pyrophosphate moiety of lipid II. Obviously, binding of A1 causes a conformational change allowing the tryptophan residues, in particular ring A and/or C, to move into a more hydrophobic environment. Moreover the conformation of the A1 peptide in complex with lipid II appears to promote binding of the A2 peptide, which may pull the entire complex even deeper into the hydrophobic core.
Lacticin 3147 (Ryan et al., 1996) is one of only four lantibiotic two-peptide systems identified to date, including the structurally closely related plantaricin W (Holo et al., 2001) and staphylococcin C55 (Navaratna et al., 1998) and the completely unrelated streptococcal cytolysin which combines bacteriocin and cytolytic activity against blood cells (Gilmore et al., 1994). Currently, molecular concepts for the synergistic effect of two bacteriocin peptides are not available. Generally, the two-peptide systems mentioned above operate best at equimolar concentrations (1:1 stoichiometry). The data presented here allow the first detailed analysis of the synergistic activity of a two-peptide bacteriocin at the molecular level (Fig. 8). Given the structural relatedness of plantaricin W and staphylococcin C55 the proposed model is likely to apply to these systems as well.
We propose a three-step model for the antibiotic activity of lacticin 3147 (i) the A1 component of lacticin associates with the membrane and with lipid II. (ii) Binding to lipid II induces or stabilizes a conformation of LtnA1 which facilitates the interaction with the LtnA2 component and enables the formation of a two peptide:lipid II complex. (iii) When bound to the LtnA1:lipid II complex the LtnA2 peptide is able to adopt a transmembrane conformation which allows it to form a defined pore. Such a model accommodates the data reported here and is consistent with structural information which is currently available on the interaction of the model lantibiotics nisin and mersacidin with lipid II (Fig. 8).
Several lines of evidence suggest that the initial contact between lacticin 3147 and the bacterial cell involves the A1 peptide. In particular, the lag in K+ release which is observed when A1 and A2 are added simultaneously or when A2 is added first, and the absence of the lag when cells are pre-incubated with A1 (Fig. 2B), indicate that binding of A1 takes place first and is rate limiting for pore formation. This agrees well with the observation that A2 do not have a specific binding site on cells in the absence of A1 (Morgan et al., 2005). Like mersacidin, but unlike most other lantibiotics and antimicrobial peptides in general, the A1 component is not positively charged. As a result, its accumulation within the anionic cell wall polymer and on the negatively charged membrane surface, generally considered a most important step for the activity of antimicrobial peptides, may not occur in contrast to the positively charged A2 component.
The interaction between LtnA1 and lipid II most likely involves the mersacidin-like binding motif which is conserved in a large group of lantibiotics, some of which have been shown to target lipid II (Wiedemann et al., 2006). In a recent NMR study, Hsu et al. (2003) reported on the structure of mersacidin. In light of the structural similarity between mersacidin and the A1 component, it appears appropriate to discuss this structural information in context with the above model. In contrast to nisin, for which the same authors (Hsu et al., 2004) were able to characterize the structure of the lipid II:nisin complex, no intermolecular nuclear Overhauser effects (NOEs) could be detected between mersacidin and lipid II which could help to identify groups involved in binding. However, it was possible to resolve defined structures in MeOH/water, and in dodecylphosphocholine (DPC) micelles in the presence and absence of lipid II. In DPC micelles, as compared with the aqueous solution, mersacidin assumed a ‘closed’ conformation, in which the N-terminus and the Glu residue in the conserved motif (Fig. 1) were in close proximity. Upon addition of lipid II the molecule appeared to ‘open’ and large chemical shift perturbations were observed. Two distinct conformations of mersacidin in the presence of lipid II were distinguished which differed in the positioning of ring A (Fig. 1) and a minute flexible hinge was identified (Ala12-Abu13) which allowed for the observed conformational changes. The flexible hinge (Ala19-Abu20 in case of the A1 peptide) and the general architecture of the adjacent rings A and B are conserved in lacticin A1, in particular ring B with the essential Glu residue (Szekat et al., 2003), and it is conceivable that the structural similarity is reflected in a similar course of events as found for mersacidin.
When lipid II was incorporated in DOPC liposomes, the fluorescence emission spectrum of A1 changed even in the absence of A2, confirming that, in analogy to mersacidin, LtnA1 is the binding partner for the cell wall precursor. Addition of the A2 peptide to this complex leads to a strong blue shift and deeper insertion into the membrane. The results obtained with the asymmetric planar membranes are fully in line with the liposome results, except that, in the absence of lipid II, pore formation was not detectable in this model membrane system. In both systems, however, it was also observed that, like with intact cells, the A2 component is completely inactive on its own, while the A1 component has some lipid II-dependent pore formation activity at high concentrations.
The interaction with lipid II appears to induce a change in the conformation of A1 in which a binding site for the A2 peptide becomes accessible. Binding of the A2 peptide to this complex (lipid II-A1) could stabilize the resulting three-component complex and allow it (i) to insert deeper into the membrane and (ii) to assume a transbilayer orientation. It is tempting to speculate that the C-terminal segment of A2, which contains three positively charged residues, may move across the membrane in response to the trans-negative polarization of the energized bacterial membrane, while the N-terminal helical segment, which is largely hydrophobic (Martin et al., 2004) may be located in the core of the membrane and keep contact with the A1:lipid II complex.
A scenario in which the A2 peptide stabilizes the interaction of A1 with lipid II is strongly supported by the data obtained from the lipid II synthesis assay (Fig. 6). Inhibition of lipid II formation occurs, as shown for nisin (Reisinger et al., 1980), via complexation of the intermediate product lipid I which lacks the second sugar N-Acetylglucosamine (GlcNAc). The affinity of the A1 peptide for lipid I is obviously low and strongly increases after addition of the A2 peptide, such that the inhibitory effect of nisin, the gold standard so far, is easily matched. The data suggest that the synergistic action of LtnA1 and LtnA2 is needed for stabilizing the interaction with the target and is therefore necessary for both activities, pore formation and efficient inhibition of cell wall biosynthesis; with this respect, a working hypothesis (Martin et al., 2004) in which binding of lipid II and inhibition of murein synthesis was attributed to the A1 peptide and pore formation to the A2 component, needs some amendment.
The data reported here also give some hints as to which parts of the molecules may interact in the three-component complex. As lipid I and lipid II are both suitable targets, the GlcNAc-moiety, like in the nisin:lipidI/II complex (Hsu et al., 2004), does not appear to be involved. Similarly, we found no indication that C55-P is acting as a target or is supporting the synergistic activity; also the pentapeptide appears not involved (Bonev et al., 2004). This restricts the relevant part in lipid II to the MurNAc-PP moiety which is likely to interact with the conserved ring B in the A1 peptide. The similarity of A1 to mersacidin prompted us to test whether A2 could synergize with this lantibiotic; however, no synergism was observed (L. Deegan and I. Wiedemann, unpubl. results) which strongly argues for specific interactions between the A1 and A2 peptide. In the mersacidin structure, the N-terminus was found to fold back onto the conserved Glu in ring B in the absence of lipid II and to move away upon binding of the cell wall precursor. If this holds true for the A1 peptide, in which the N-terminal part is extended, a similar conformational change might expose a binding site for A2. In a recent study (Cotter et al., 2005), the importance of the conversion of Ser residues in the N-terminal segments of the A1 and A2 peptide into D-Ala residues was shown. Neither the dehydration of Ser to dehydroalanine, nor substitution of Ser with L-Ala restored full activity, demonstrating that the chirality of the D-Ala residues may be essential for stereospecific interactions. It will be interesting to test such a structure-activity model with A1 and A2 fragments and site-specific mutant peptides.
The pore size of the lacticin 3147 pore, 0.6 nm, is small as compared with nisin; for the latter, an average diameter of 2 nm was found (Wiedemann et al., 2004). The use of different model membranes, solvent containing black lipid membranes for nisin and solvent free Montal–Mueller membranes in the present study, makes it difficult to directly compare these values. Comparison is even more difficult with liposome model membranes. The CF leakage observed in this system suggests, because of the molecular radius of CF, a somewhat larger pore. However, for nisin it was observed that metabolites and amino acids leak out of treated cells (Ruhr and Sahl, 1985) while with lacticin 3147 only K+ leakage was reported; thus the pore dimensions determined with planar bilayer systems seem to correlate better with the in vivo situation. To build a pore of 2 nm in diameter, it is necessary that several nisin:lipid II complexes assemble into a functional pore structure and it has been suggested that such a pore complex may consist of eight nisin and four lipid II molecules (Hasper et al., 2004). For a pore of 0.6 nm in diameter, such an intricate architecture may not be necessary and it seems conceivable that the lacticin 3147 pore represents just a monomeric or dimeric complex with a 1:1:1 stoichiometry (lipid II:A1:A2).
Lacticin 3147 is an excellent example for the exceptional antibiotic efficiency that can be achieved when two killing mechanisms combine. However, it needs to be addressed that such high activities in the nanomolar concentration range are to some extent strain or species specific (Table 1), a feature which is characteristic for bacteriocins rather than for antibiotics in general. Currently, it remains enigmatic why the synergism between A1 and A2 is so strong against Lactococcus lactis, but much less pronounced with Micrococcus flavus. Also, there is no satisfactory explanation for the fact that the A1 peptide and mersacidin are almost equally effective against the lactococcal strain, but differ by a factor of 30 against the Micrococcus. The low susceptibility of M. flavus could arise from an inaccessibility of lipid II in this species for lacticin; a lipid II-independent killing mechanism of this strain is suggested by the high minimal inhibitory concentration (MIC), which is in the same range as lipid II-independent poration of liposomes (Fig. 3) and by the observation that lacticin-induced K+ leakage from this strain was still observed after the lipid II had been blocked with a non-pore-forming gallidermin variant (data not shown). In a recent study with gallidermin (Bonelli et al., 2006) which contains the lipid II binding motif of nisin, but is considerably shorter (30 Å versus 50 Å for the overall length of the peptides in solution), we found that it was able to induce pore formation only in some strains (e.g. Micrococcus flavus and Staphylococcus simulans) but not in Lactococcus. Membrane thickness may contribute to this phenomenon because in lactococci the average fatty acid acyl chains contain more than 17 carbon atoms whereas just about 15 were reported in staphylococci and micrococci (Bonelli et al., 2006). In spite of the missing pore formation, gallidermin was as active against the Lactococcus strain as lacticin 3147 in the present study (Table 1), i.e. by a factor of eight more active than nisin. Obviously, lantibiotics, and bacteriocins in general have been optimized in the course of evolution towards a narrow spectrum and limited number of competitors in defined ecological niches. While this could make it difficult to correlate structural features with functions for each individual pair of peptide and target species, it provides a unique opportunity to develop highly active and highly targeted antibiotic agents for defined biomedical applications.
Table 1. Antimicrobial activity of lantibiotics determined as MIC.
L. lactis ssp. cremoris HP (μM)
M. flavus DSM 1790 (μM)
Lacticin A1 and A2 were used at a 1:1 ratio (w/w); calculation of the MIC is based on the MW of the A1 peptide.
All chemicals were of analytical grade or better. Radiolabelled [14C]-UDP-GlcNAc (7.4 GBq mmol−1) was purchased from Amersham Bioscience. The phospholipids DOPC and DOPG were purchased from Avanti Polar Lipids (Alabaster, AL, USA) and used without further purification. CF was purchased from Sigma (Deisenhofen, Germany). Protein concentration of membrane preparations was determined using the bicinchonic acid protein assay reagent (Pierce, Rockford, IL, USA) with bovine serum albumin as standard.
Bacterial strains and culture conditions
The lacticin 3147 overproducing strain Lactococcus lactis ssp. cremoris MG1363 (pMRC01, pOM02) (Cotter, P.D., Draper, L., Lawton, E., Hill C. and Ross, R.P., 2006. Applied and Environmental Microbiology, in press) and the indicator strain L. lactis ssp. cremoris HP were grown in M17 medium (Merck, Darmstadt, Germany) supplemented with 0.5% glucose (GM17) at 30°C without aeration. Micrococcus flavus DSM 1790 was grown at 30°C and Staphylococcus simulans 22 at 37°C in TSB (Merck, Darmstadt, Germany) with aeration.
Preparative purification of the lacticin 3147 peptides
Lacticin A1 and A2 were isolated as described (Martin et al., 2004) with some modifications. Briefly, the over-producer L. lactis subspec. cremoris MG1363 (pMRC01, pOM02) was grown in the production medium for 20 h at 30°C without aeration. After addition of ammonium-sulphate [50 g l−1] and DTT [75 mg l−1] cells from 4 l of culture were collected by centrifugation (20 min, 10 000 g, 4°C). Pellets were resuspended in 250 ml of 70% isopropanol 0.1 M DTT. After stirring for 3 h at 4°C 3.5 h cell debris were removed by centrifugation (10 000 g, 20 min, 4°C), the volume was reduced by rotary evaporation and the concentrated solution was lyophilized. The crude extract was dissolved in 20 ml of 24% isopropanol, 0.1% trifluoroacetic acid (TFA) and the insoluble material was removed by centrifugation (2500 g, 3 min). The sample was applied to a preparative high-performance liquid chromatography column (Nucleosil 100-C18-10 μm 225 × 20 mm ID). The column was equilibrated with buffer A (H2O, 0.1% [v/v] TFA) and peptides were eluted using a linear gradient of buffer B (isopropanol, 0.1% [v/v] TFA). Gradient: 24% buffer B for 5 min, then climbing to 50% in 45 min and returning to 24% in 1 min at a flow rate of 8 ml min−1. Peptides were detected at 220 nm. Electro spray mass spectrometry was used to confirm the correct mass and the purity of the isolated lacticin A1 and A2 peptides. Purified samples were lyophilized and stored at −20°C; stock solutions were prepared in 30% isopropanol and kept at −20°C.
Minimal inhibitory concentration determinations
Minimal inhibitory concentration determinations were carried out in microtitre plates. M. flavus DSM 1790 and S. simulans 22 were grown in half-concentrated Mueller–Hinton (MH) broth (Oxoid). L. lactis ssp. cremoris HP was grown in M17 broth plus 0.5% glucose (Oxoid). Serial twofold dilutions of the peptides were made in the growth medium of the respective indicator strain. Bacteria were added to give a final inoculum of 105 cfu ml−1 in a volume of 0.2 ml. After incubation for 16 h at 37°C for S. simulans 22; at 30°C for L. lactis ssp. cremoris HP or 24 h at 30°C for M. flavus DSM 1790, the MIC was read as the lowest peptide concentration causing inhibition of visible growth. Results given are mean values of three independent determinations.
Potassium release from whole cells
Cells were harvested at an OD600 of 1.0–1.5 (3300 g, 5°C, 3 min), washed with 50 ml of cold choline-buffer (300 mM choline chloride, 30 mM Mes, 20 mM Tris, pH 6.5) and resuspended in the same buffer to an OD600 of 30. The concentrated cell suspension was kept on ice and used within about 90 min. For each measurement the cells were diluted in 2 ml of choline-buffer (25°C) to an OD600 of about 3. Peptide-induced potassium efflux was monitored using a pH-meter pH 213 (Hanna Instruments, Kehl am Rhein, Germany) with a MI-442 potassium electrode and MI-409F reference electrode. Before each experiment, the electrodes were calibrated with standard solutions containing 0.01, 0.1 or 1 mM KCl in buffer and calculations of potassium-efflux in per cent were performed as described (Orlov et al., 2002). Peptide-induced leakage was expressed relative to the total amount of potassium release induced by addition of 1 μM nisin (data not shown). The amount of K+ released by nisin correlated to an intracellular K+ concentration of approximately 300 mM.
Preparation of unilamellar vesicles
Large unilamellar vesicles (LUVs) were prepared for CF and tryptophan quench experiments by the extrusion technique (Mayer et al., 1986). When indicated, vesicles (400 nm) were supplemented with 0.1 mol% or 1 mol% lipid II or undecaprenylphosphate (C55-P) (referring to the total amount of phospholipids).
Carboxyfluorescein efflux experiments
Carboxyfluorescein-loaded vesicles were prepared with 50 mM CF and then diluted in 1.5 ml of K+ buffer [50 mM Mes-KOH, pH 6.0, 100 mM K2SO4] in a final concentration of 25 μM phospholipid on a phosphorous base. After addition of the peptide, the increase of fluorescence intensity was measured at 520 nm (excitation at 492 nm) at room temperature. Peptide-induced leakage was documented relative to the total amount of marker release after solubilization of the vesicles by addition of 10 μl of 20% Triton X-100.
Tryptophan fluorescence measurements
Tryptophan fluorescence measurements were performed in 1.5 ml of K+ buffer at room temperature. The peptides were added at a concentration of 0.25 μM in the absence and presence of different liposomes at a concentration of 100 μM lipid Pi. Emission spectra were recorded from 300 to 400 nm (bandwidth 10 nm) with excitation at 280 nm (bandwidth 5 nm) and corrected for the vesicle blank. All fluorescence measurements were recorded using a RF-5301 spectrophotometer (Shimadzu, Duisburg, Germany).
Preparation of planar bilayer membranes and electrical measurements
Planar bilayer membranes were prepared according to the Montal–Mueller technique (Montal and Mueller, 1972) as described earlier (Wiese and Seydel, 1999). Briefly, symmetric and asymmetric bilayers were formed by opposing two identically or differently composed lipid monolayers prepared on aqueous subphases from chloroformic solutions of the lipids at a small aperture (Ø∼150 μm) in a Teflon septum. For electrical measurements, the membranes were voltage-clamped via a pair of Ag/AgCl-electrodes (type IVM E255, Advanced Laboratory Research, Franklin, MA, USA) connected to the head-stage of an L/M-PCA patch-clamp amplifier (List-Medical, Darmstadt, Germany). If not mentioned otherwise, lacticin was added to the side of the bilayer named first, the opposite compartment was grounded. All measurements were performed at a temperature of 37°C with subphases consisting of 1 M KCl buffered with 5 mM Hepes (specific conductance: 18.6 S m−1) and adjusted to pH 6.
At the beginning of each experiment, membrane formation was checked by measuring membrane current and capacitance. Only membranes with a basic current of less than ±2.5 pA at a clamp voltage of ±100 mV and a capacitance >90 pF were used for the experiments. To determine the macroscopic current/voltage-characteristics (I/U-characteristics) triangular voltage ramps (1.85 mV s−1) were applied to the membrane. The current voltage plot shown in Fig. S1 is a representative of at least three independent experiments. For the characterization of single pores, the peptide-induced membrane current was filtered with a frequency of 2 kHz and recorded with 5 kHz. Statistical analysis of pore diameter was based on evaluation of equidistant current steps.
Inhibition of the in vitro lipid II synthesis
Inhibition of the in vitro lipid II formation was analysed using the lipid II synthesis assay (Schneider et al., 2004) with the addition of radiolabelled [14C]-UDP-GlcNAc. Reaction mixtures were carried out in a final volume of 150 μl containing 400–800 μg of membrane protein of M. flavus DSM 1790, 10 nmol C55-P, 100 nmol UDP-N-acetylmuramyl pentapeptide (UDP-MurNAc-PP), 100 nmol [14C]-UDP-GlcNAc in 60 mM Tris-HCl, 5 mM MgCl2, pH 8 and 0,5% (wt/vol) Triton X-100. Peptides were added to the reaction mixture in molar ratios of 0.5:1, 1:1 and 1.5:1 referring to the total amount of C55-P (10 nmol). After 1 h at 30°C, the lipids were extracted with 1 vol. of n-butanol/6 M pyridine-acetate (2:1, v/v), pH 4.2. The reaction products were separated by TLC (silica plates, 60F254; Merck) using chloroform-methanol-water-ammonia (88:48:10:1) as the solvent (Rick et al., 1998). Radiolabelled spots were visualized by iodine vapour, excised and quantified by β-scintillation counting (1900 CA Tri-Carb scintillation counter, Packard).
The German Research Foundation (DFG, Sa 292/9-1-9-4), the BonFor programme of the Medical Faculty, University of Bonn and the financial assistance of the Irish Government under the National Development Plan 2000−06 are gratefully acknowledged. We thank A. Tossi, Trieste and B. Lindner, Borstel for determination of the masses.