Use of the Cell Wall Precursor Lipid II by a Pore-Forming Peptide Antibiotic

See allHide authors and affiliations

Science  17 Dec 1999:
Vol. 286, Issue 5448, pp. 2361-2364
DOI: 10.1126/science.286.5448.2361


Resistance to antibiotics is increasing in some groups of clinically important pathogens. For instance, high vancomycin resistance has emerged in enterococci. Promising alternative antibiotics are the peptide antibiotics, abundant in host defense systems, which kill their targets by permeabilizing the plasma membrane. These peptides generally do not act via specific receptors and are active in the micromolar range. Here it is shown that vancomycin and the antibacterial peptide nisin Z use the same target: the membrane-anchored cell wall precursor Lipid II. Nisin combines high affinity for Lipid II with its pore-forming ability, thus causing the peptide to be highly active (in the nanomolar range).

Nisin Z is a member of the lantibiotic family (lanthionine-containing antibiotics) and is produced by certain strains of Lactococcus lactis. Because of its nontoxicity for humans and its high bactericidal activity, it is used as a food preservative. Nisin is posttranslationally modified (Fig. 1A). Characteristic features of nisin are the ring systems formed by thioether bonds and the dehydrated amino acids. Nisin shares some properties with other pore-forming antibacterial peptides, such as an overall positive charge and amphipathicity. However, the high bactericidal activity of nisin against Gram-positive bacteria, relative to the lower activity of the antibacterial peptides of animal origin (such as magainin), has not been accounted for in the literature. Elucidation of the cause of the activity difference may lead to the development of new classes of antibiotics with high activities.

Figure 1

Activity of nisin Z and magainin toward intactM. flavus cells and model membrane vesicles. (A) Primary structure of nisin Z. Dha, dehydroalanine; Dhb, dehydrobutyrine; Ala-S-Ala, lanthionine; Abu-S-Ala, β-methyllanthionine; S, the sulfur atom of the thioether bond. (B) Activity of nisin and magainin in a cell viability assay. Nisin (closed squares) and magainin (open squares) were added at the specific concentrations to M. flavus cells in complex growth medium (18), and viability was determined by measuring the absorbance at 600 nm as described (19). (C) Activity of nisin and magainin toward intact M. flavus cells. The peptide activity was measured by monitoring the effect on the membrane potential with the fluorescent membrane potential–sensitive probe 3,3′-diethylthiodicarbocyanine iodide [DiS-C2(5)] (20). Cells were grown until mid-log phase, harvested, and washed once with a buffer solution of 250 mM sucrose, 5 mM MgSO4, and 10 mM potassium phosphate (pH 7.0), then resuspended in the same buffer. Cells were added to the fluorescence cuvette at an optical density at 600 nm of 0.075 together with DiS-C2(5) at 1 μM. (D) Activity of nisin and magainin toward model membranes composed of a lipid extract from M. flavus. The peptide activity was measured by monitoring the leakage of carboxyfluorescein from vesicles made from a lipid extract of M. flavus by measuring the increase in fluorescence due to dilution of the dye from self-quenching concentrations as described (20). The concentration of both nisin and magainin was 2.5 μM. The arrows in (C) and (D) mark the time point of peptide addition.

The difference in bactericidal activity between nisin and magainin is illustrated for the Gram-positive bacterium Micrococcus flavus in Fig. 1B. The M. flavus cells were killed by nanomolar concentrations of nisin [minimal inhibitory concentration (MIC) = 3.3 nM] (1). In contrast, magainin (2) was at least two orders of magnitude less active than nisin, because even at the highest peptide concentration tested (0.4 μM) not all M. flavus cells were killed.

The results of the viability assay correlate with the ability of the peptides to permeabilize the membrane of M. flavus cells (Fig. 1C). This permeabilization caused the dissipation of vital ion gradients such as potassium, resulting in dissipation of the proton-motive force and eventually cell death. Nisin efficiently permeabilizes M. flavus membranes: At 5 nM nisin a permeabilization effect was detected, whereas magainin was active only at 3 μM. However, when the two peptides were tested with membranes composed of a lipid extract of M. flavus, the membrane-permeabilizing activity of nisin dropped markedly (now active in the micromolar range), and magainin was the more active peptide (Fig. 1D). Similar results were obtained when potassium leakage from vesicles was measured (3), indicating that these large differences are not a reflection of the nature of the indicator molecule. In pure phospholipid systems both peptides form short-lived transmembrane pores, and it has been shown using synthetic phospholipids that negatively charged lipids play an active role in this process (4–7). Thus, a membrane composed of an isolate from a chloroform/methanol extract (containing mostly lipids and hardly any protein), or pure synthetic phospholipids, may lack one or more specific molecules needed for high nisin activity.

Next we examined the effect of nisin on M. flavuscells in the presence of vancomycin (Fig. 2). Vancomycin inhibited the membrane leakage activity of nisin against intact cells but did not affect the activity of magainin. Vancomycin kills bacteria by blocking the cell wall biosynthesis by binding to the Lys-d-Ala-d-Ala motif of the pentapeptide of the membrane-anchored cell wall precursor Lipid II (8, 9) (inset, Fig. 2). This suggests that Lipid II is used by nisin to permeabilize the membrane (10). If this is the case, then the cells should become more sensitive to nisin upon increasing the Lipid II content of their membranes. This is true for isolated cytoplasmic membranes of both M. flavus andEscherichia coli (Fig. 3A). Fusing Lipid II–containing model membranes with the bacterial membranes increased the sensitivity of both membrane types for nisin but not for magainin. The effect was the highest for the E. coli membranes. One explanation for this effect lies in the intrinsic low Lipid II content of E. coli, which contains about 2000 molecules per cell (11), whereas M. flavus contains about 105 Lipid II molecules (12).

Figure 2

Vancomycin blocks the activity of nisin against M. flavus cells. The activity of the peptides was monitored as described for Fig. 1. When present, vancomycin was added 2 min before the addition of nisin or magainin. Washing the cells after treatment with vancomycin gave similar results. The arrow marks the time point of peptide addition. Inset: Schematic structure of Lipid II with vancomycin bound to the COOH-terminal Lys-d-Ala-d-Ala sequence. The structure consists of a membrane-incorporated undecaprenyl moiety to which the amino sugar MurNAc is attached via a pyrophosphate. To the MurNAc (M) a pentapeptide is attached, of which the composition may slightly differ within different bacterial genera. The final subunit of Lipid II is GlcNAc (G). Pi, phosphate.

Figure 3

Increase in the Lipid II content of isolated cytoplasmic membranes increases their nisin sensitivity. (A) Lipid II incorporation via vesicle fusion (solid bars, fused; open bars, control). Small unilamellar vesicles composed of 1,2-dioleoyl-sn-glycero-3-phosphoglycerol and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine in an equimolar mixture with or without 1 mol % purified Lipid II were fused with isolated membrane vesicles of E. coli or M. flavus by rapidly freezing the mixture into liquid nitrogen and thawing on ice (21). Lipid II was purified fromMicrococcus luteus as described (22). A membrane potential was generated in the membranes with the artificial electron donor/acceptor system ascorbate-phenazine methosulfate (23). The activity of 30 nM nisin or 400 nM magainin was monitored using the fluorescent dye DiS-C2(5) as described (20). Maximum depolarization was determined by addition of a mixture of valinomycin and carbonyl cyanidep-trifluoromethoxyphenylhydrazone, which completely dissipates the membrane potential. (B) Regeneration of the Lipid II pool in M. flavus membranes increases the sensitivity toward nisin. The M. flavus membranes were subjected to eight freeze-thaw cycles in the presence of 0.1 mM UDP-MurNAc-pentapeptide and 0.1 mM UDP-GlcNAc (solid bars), or only 0.1 mM UDP-MurNAc-pentapeptide (gray bars), or with no addition (open bars). After freeze-thawing, the membranes were incubated for 1 hour at 20°C, which was essential for the enzyme reactions to take place. A membrane potential was generated, and the activity of nisin (30 nM) and magainin (400 nM) was determined as described for (A).

We observed that the M. flavus membranes were less sensitive toward nisin (by a factor 3 to 6) relative to the intact cells (13). This can be explained by a loss of part of the Lipid II pool during the isolation of the membrane vesicles, which is due to the high turnover rate of Lipid II. Because the Lipid II biosynthesis machinery is still active, this allowed us to regenerate Lipid II in the presence of biosynthetic precursors. Upon regeneration of the Lipid II pool by supplying the complete set of precursors, the sensitivity of the isolated M. flavus membranes to nisin increased, whereas the sensitivity for magainin was unaffected (Fig. 3B). When only one of the uridine 5′-diphosphate (UDP)–activated sugars was present, the vesicles did not become sensitive to nisin. This can be explained from current knowledge of the Lipid II biosynthesis pathway. Lipid II is synthesized at the cytosolic side of the plasma membrane. First, the UDP-activated amino sugarN-acetyl muramic acid (MurNAc), containing a pentapeptide, is attached to the undecaprenol carrier; then, Lipid II synthesis is completed upon attachment of the second UDP-activated amino sugarN-acetyl-d-glucosamine (GlcNAc). Lipid II is transported to the exterior side of the membrane and becomes available for binding by nisin only when the synthesis is completed. An additional finding was that the freeze-thawed membranes had to be incubated at 20°C for 1 hour to obtain the effect of the UDP-activated sugars (3). Apparently, this is one way to control the onset of the Lipid II synthesis. Thus, this experimental setup could be used in the investigation of the transport process of Lipid II, an important step in the cell wall biosynthesis.

These results show not only that Lipid II functions in the activity of nisin, but also that it appears to be the sole target of nisin (Fig. 4A). Incorporation of purified Lipid II in small amounts (one Lipid II molecule per 1500 phospholipid molecules) to model membrane systems composed of pure lipids markedly increased the nisin activity, whereas the magainin activity remained unchanged. In the absence of Lipid II, nisin induced leakage from this system only at concentrations above 1 μM. In the presence of Lipid II, leakage was detected at 1 nM nisin, which suggests that nisin has high affinity for Lipid II (14). The same effects of Lipid II were present when membranes containing negatively charged lipids were used (3). Because the activity of nisin in these model systems is in the range of the activity against intact cells, nisin appears to use Lipid II as its sole target. Nisin is dependent on Lipid II concentrations in the range of 0.001 to 0.1 mol % (Fig. 4B). This suggests that the diverse sensitivities to nisin displayed by different bacteria are caused by different concentrations of Lipid II in the membrane, although, in the case of intact cells, differences in the accessibility of Lipid II for nisin should also be considered.

Figure 4

(A) The presence of Lipid II in model membrane systems markedly increases the nisin activity. The activity of nisin (squares) and magainin (circles) was measured by monitoring the leakage of carboxyfluorescein from model membranes composed of 1,2-dioleoyl-sn-glycero-3-phosphocholine in the presence (filled symbols) or absence (open symbols) of 0.065 mol % purified Lipid II. (B) Nisin activity is dependent on the concentration of Lipid II in the membrane. The activity of nisin (100 nM) was measured as described for (A).

The effect of Lipid II in all tested systems was specific for nisin, which suggests a specific interaction of Lipid II with one or more of the structural elements of nisin. Moreover, mutagenesis experiments with nisin showed that relatively subtle variations in the three NH2-terminal rings had strong influences on the bactericidal activity of these mutants as well as on the interaction with Lipid II–containing membranes. For instance, a Ser3 → Thr (S3T) mutation (actually, changing the first lanthionine residue into a β-methyl lanthionine) made the peptide less active by an order of magnitude (MIC for S3T = 39 nM, versus 3.3 nM for the wild type). Changing the thioether bond of the third ring into a disulfide bond had even larger effects (MIC > 60 nM). These findings suggest that at least the NH2-terminal rings of nisin are involved in the interaction with Lipid II (15).

The high activity of nisin (in the nanomolar range) compared to magainin is the result of a combination of high-affinity binding to Lipid II and permeabilization of the plasma membrane, resulting in cell death. In contrast to recent reports on resistance or tolerance to vancomycin (16), no resistance to nisin has been reported, despite its prolonged use as a preservative for almost 50 years. This combination of factors makes nisin an attractive antimicrobial agent. The insights into its mode of action uncovered in this study can now be used as a blueprint for the development of a new class of highly efficient antibiotics.

  • * To whom correspondence should be addressed. E-mail: e.j.breukink{at}


View Abstract

Stay Connected to Science

Navigate This Article