INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Listeria monocytogenes, the causative agent of listeriosis, has resulted in numerous major food-borne outbreaks worldwide (16). The ability of L. monocytogenes to grow at temperatures ranging from 1 to 45°C (19, 41, 42), its high tolerance for salt (11), and its ability to initiate growth at a relatively low pH (19) make this pathogen particularly difficult to control in food. A novel approach to controlling L. monocytogenes in food is the use of antimicrobial bacteriocins from lactic acid bacteria (36). Nisin, a lanthionine-containing peptide produced by certain strains of Lactococcus lactis (26) with antimicrobial activity against L. monocytogenes (2-4, 25), is a bacteriocin with many "generally recognized as safe" applications in the United States (12-14).

Nisin's action stems from the disruption of the cell's cytoplasmic membrane, as evidenced by the rapid efflux of small molecules from both whole cells and liposomes (1, 3, 18, 39, 44). As a result, nisin depletes the proton motive force (PMF) of sensitive cells and artificial liposomes (3, 17, 37, 39). Nisin acts through a multistep process which includes binding of nisin to the cell, insertion into the membrane, and pore formation (10, 18, 35, 40, 44). Anionic phospholipids play an important role in nisin's interaction with membranes (9, 10, 30). On binding to anionic phospholipids, nisin causes a local perturbation of the lipid bilayer (10), followed by electrical potential ()- or pH gradient (pH)-enhanced insertion into the membrane to form a wedge-like pore (35).

Nisin's efficacy as an antilisterial agent would be compromised by the emergence of nisin-resistant L. monocytogenes in a food-processing situation. The generation of nisin-resistant (Nis^r) L. monocytogenes mutants in the laboratory is easily achieved by exposure to high concentrations of nisin (21, 32, 33). Given the role of the cytoplasmic membrane in nisin's mechanism of action, several researchers have examined membrane compositional changes to explain nisin resistance in L. monocytogenes. Resistance has been correlated with both an altered fatty acid composition (32, 33) and an altered phospholipid composition (34). When grown in the presence of nisin, the Nis^r strain (32) investigated in this study contains a lower ratio of C₁₅ to C₁₇ fatty acids than the wild-type or Nis^r strain grown in the absence of nisin. In addition to membrane compositional changes, the cell wall may also be involved in nisin resistance in L. monocytogenes (8).

The objectives of this research were to investigate the roles of both the cytoplasmic membrane and the cell wall in a single nisin-resistant strain of L. monocytogenes, to investigate whether nisin resistance results in intrinsic resistance to other antimicrobials, and to develop a model of nisin's interaction with nisin-resistant L. monocytogenes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial cultures, growth media, and nisin stock preparations. All bacterial stock cultures were maintained in their appropriate broth containing 20% glycerol at 80°C. L. monocytogenes Scott A was cultured in Trypticase soy broth without dextrose (BBL Microbiology Systems, Cockeysville, Md.), supplemented with 0.6% yeast extract (Difco Laboratories, Detroit, Mich.) and 0.5% glucose (Fisher Scientific Co., Fair Lawn, N.J.) (TSBYEG). The nisin-resistant strain of L. monocytogenes Scott A was isolated at 30°C after the wild-type strain was plated on TSBYEG agar plates containing 1,000 IU of nisin preparation per ml (32). This Nis^r strain was recently deposited in the American Type Culture Collection with accession no. ATCC 700302. The Nis^r strain was maintained and cultured in TSBYEG containing 1,000 IU of nisin preparation per ml. L. lactis ATCC 11454, L. lactis ATCC 21053, Leuconostoc paramesenteroides OX (28), and Pediococcus acidilactici PAC1.0 (provided by J. L. Swezey, USDA Agricultural Research Service Midwest Area, Peoria, Ill.) were cultured in Lactobacillus MRS (Difco) broth. Working cultures were maintained at 4°C on agar slants solidified with 1.5% Bacto-Agar (Difco). All cultures were grown at 30°C. TSBYEG diffusion agar contained 0.1% Tween 20 and was solidified with 1.5% Noble agar (Difco).

Identification and quantification of membrane phospholipids. Overnight cultures of wild-type cells and of Nis^r cells grown in the presence of 1,000 IU of nisin preparation per ml were centrifuged; the cell pellets were resuspended in TSBYEG containing 20% (wt/vol) sucrose, 0.2% (wt/vol) MgSO₄, 8 mg of lysozyme (Sigma) per ml, and 1,000 U of penicillin G (Sigma) per ml and incubated for another 2 h at 30°C. The lysozyme-penicillin treatment disrupted the cell wall, a step necessary for the complete extraction of cardiolipin from stationary-phase gram-positive bacteria (15). The cells were then pelleted again and the lipids were extracted as previously described (45).

Lysozyme sensitivity. Overnight cultures of wild-type cells, Nis^r cells, and Nis^r cells grown with 1,000 IU of nisin preparation per ml were inoculated at 1% (vol/vol) into fresh TSBYEG containing 10 mM MgCl₂ and 4 mg of lysozyme (Sigma) per ml. The cultures were incubated at 30°C, and absorbances at 620 nm were monitored spectrophotometrically.

MIC determinations. The MICs of antimicrobials for the wild-type and Nis^r strains were determined by a spiral gradient agar dilution technique as previously described (32). This method is based on the establishment of a spiral gradient of antimicrobial in an agar plate by using a Spiral Plater (model D; Spiral Biotech, Inc.). Antimicrobial stock solutions were spiral plated over 20 ml of TSBYEG agar, and the plates were held at room temperature for 4 h. Overnight cultures of wild-type cells, Nis^r cells, or Nis^r cells grown in the presence of 1,000 IU of nisin preparation per ml were then radially swabbed across the antimicrobial gradient, from low to high concentrations, with a sterile swab. The plates were incubated at 30°C for 48 h. The growth endpoint was measured and the MIC was calculated. Stock solutions consisted of gramicidin S, chloramphenicol, and tetracycline in methanol; nigericin and erythromycin in acetone; benzylpenicillin, ampicillin, gentamicin, cycloserine, streptomycin, and vancomycin in water; and nisin in nisin diluent. Control plates were prepared with either methanol, acetone, distilled water, or nisin diluent. All antimicrobials other than nisin were obtained from Sigma.

Transmission EM. Electron microscopy (EM) was utilized to visualize the cell wall of wild-type and Nis^r L. monocytogenes. Mid-log-phase wild-type cells, Nis^r cells, and Nis^r cells grown with 500 IU of nisin per ml were prepared for EM by resuspending cell pellets in TSBYEG broth containing 2% (vol/vol) glutaraldehyde (Sigma). The cells were then again pelleted and shipped under the glutaraldehyde broth supernatant. The EM work was kindly conducted by Peter Cooke at the USDA Eastern Regional Research Center in Philadelphia, Pa.

Determination of the PMF, pH gradient, and . The PMF, an electrochemical gradient of protons across the cytoplasmic membrane, is required to drive a number of bacterial energy-requiring processes, such as ATP synthesis, protein phosphorylation, flagellar rotation, and active transport. The PMF is composed of a and a pH. These two components of the PMF were determined as previously described (4). The pH potential is calculated as ZpH. Factor Z, used to convert pH units to millivolts, represents 2.3 RT/F, where R is the universal gas constant, T is temperature in degrees kelvin, and F represents Faraday's constant. The value of Z is 59 mV at 25°C. The results are calculated as the means of three independent experiments. Variances were analyzed and compared by a general linear model procedure using SAS (SAS Institute Inc., Cary, N.C.).

Divalent-cation requirement of Nis^r L. monocytogenes. Overnight cultures of wild-type or Nis^r cells were harvested by centrifugation, washed once with phosphate-buffered saline, and resuspended in 50 mM MES (morpholineethanesulfonic acid) buffer (pH 6.5) containing either divalent cations alone, divalent cations plus EDTA, EDTA alone, or none of these. The metal ions tested, all at a final concentration of 10 mM, were MgSO₄, MgCl₂, CaCl₂, MnSO₄, BaCl₂, NaCl, and KCl. The final concentration of EDTA was 20 mM. The cells were then treated with a final concentration of 60 IU of nisin per ml for 20 min. Control samples were not treated with nisin. The cells were then diluted in 0.1% peptone water and spiral plated on TSBYEG agar plates. The plates were incubated at 30°C, and colonies were counted with a laser colony counter after 48 h. The results were reported as log reductions in cell viability relative to an untreated control.

Determination of bacteriocin cross-resistance. The sensitivities of wild-type and Nis^r L. monocytogenes to bacteriocins produced by different lactic acid bacteria were determined. MRS agar containing 0.5% glucose was prepared in MES buffer at pH 6.5 to neutralize acid production. Aliquots (20 µl) of overnight lactic acid bacteria cultures were dispensed in 6.8-mm-diameter wells cut into a 15-ml MRS agar layer. After a 48-h incubation at 30°C, the plates were overlaid with 10 ml of TSBYEG agar with 0.1% Tween 20 containing either wild-type or Nis^r L. monocytogenes. The L. monocytogenes inocula were obtained by adjusting the absorbance at 660 nm of overnight cultures to 0.4 and adding 1 µl/ml to the TSBYEG agar to yield 10⁵ CFU/ml. The plates were incubated at 30°C, and the inhibition zones were measured from the edge of the well containing the lactic acid bacteria to the edge of the inhibition zone.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Phospholipid composition. The Nis^r cells, when grown in the presence of 1,000 IU of nisin preparation per ml, contained significantly (P < 0.05) more phosphatidylethanolamine than either wild-type cells or Nis^r cells grown in the absence of nisin (Table 1). The Nis^r cells also contained less cardiolipin and phosphatidylglycerol. These decreases were not statistically significant. However, when cardiolipin and phosphatidylglycerol were combined to give the total anionic phospholipids, this value was significantly decreased for Nis^r cells grown with nisin. No significant differences between wild-type cells and Nis^r cells grown in the absence of nisin for any of the phospholipids or between any of the strains for an unidentified amine-containing phospholipid and two other minor phospholipids were found.

Lysozyme sensitivity. In the absence of lysozyme, the wild-type and Nis^r strains exhibited similar growth characteristics (Fig. 1), attaining a final optical density at 660 nm of approximately 0.5. In the presence of 4 mg of lysozyme per ml, the Nis^r strain reached a higher final optical density of approximately 0.4, compared to 0.25 for the wild-type strain.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Growth of wild-type (circles) and Nisr (triangles) strains of L. monocytogenes in TSBYEG at 30°C. The Nisr culture inoculum was grown either in the absence ( and ) or in the presence ( and ) of 1,000 IU of nisin preparation per ml. The cells were treated with 4 mg of lysozyme at time zero (closed symbols) or left untreated (open symbols).

MIC determinations. The MICs of all the antibiotics tested for Nis^r cells grown in the absence of nisin (data not shown) were not significantly different (P < 0.05) from those for the Nis^r strain grown in the presence of 1,000 IU of nisin preparation per ml (Table 2). The MIC of nisin for the Nis^r strain was 1,423 ± 310 IU/ml, compared to 240 ± 31 IU/ml for the wild-type strain (means ± standard deviations). The MICs of benzylpenicillin and ampicillin for the Nis^r strain were approximately 10-fold lower than those for the wild-type strain (significant at P < 0.001). The Nis^r strain was also slightly more sensitive to cycloserine (significant at P < 0.001). In addition, small but significant (P < 0.001) increases in the MICs of gramicidin S and gentamicin for the Nis^r strain were observed. No significant differences in the MICs of any of the other antibiotics tested were found.

Transmission EM. No significant morphological differences between wild-type and Nis^r cells were observed by transmission EM at ×70,000 magnification. Wild-type and Nis^r cells contained cell walls of comparable thicknesses (approximately 360 nm).

PMF measurements. The PMFs of wild-type and Nis^r L. monocytogenes were measured both in the absence and in the presence of nisin. Basal PMFs of 122 ± 11 and 117 ± 13 mV (means ± standard deviations) were measured for the wild-type and Nis^r strains, respectively (Fig. 2). These values were not significantly different (P < 0.05). The addition of 60 IU of nisin per ml to the wild-type strain caused a significant (P < 0.05) dissipation of the PMF, to 15.0 ± 5.3 mV. Treatment of the Nis^r strain with the same concentration of nisin caused the PMF to be lowered by only 21%, to 100.3 ± 9.8 mV. The partially depleted PMF of the Nis^r strain was significantly lower (P < 0.05) than that strain's resting PMF and significantly higher (P < 0.05) than the wild-type strain's depleted PMF. Under the conditions of the PMF assay, nisin resulted in a 3.2-log reduction in wild-type cell numbers and a 0.5-log reduction in Nis^r cells.

View larger version (32K): [in this window] [in a new window]   FIG. 2.   Total PMF values (means and standard deviations), represented as the sum of and ZpH, for wild-type and Nisr L. monocytogenes in the absence and presence of 60 IU of nisin per ml. Mean values for and ZpH appear to the right of each bar. Different letters indicate statistically significant differences (P < 0.05) for the total PMF.

Divalent-cation requirement of Nis^r L. monocytogenes. In the course of this study, we realized that while the Nis^r strain was able to resist the effects of nisin either in broth or on agar plates, it was unable to do so when suspended in buffer. A major component of growth media which could be required by the Nis^r strain to resist nisin is divalent cations. Therefore, the nisin sensitivity of the Nis^r strain in the absence and presence of different divalent cations was assessed. The viability of Nis^r cells suspended in MES buffer and treated with 60 IU of nisin per ml was reduced by 2.5 log cycles (Fig. 3). Inclusion of either 10 mM MgSO₄, MgCl₂, CaCl₂, MnSO₄, or BaCl₂ reduced the lethality caused by nisin to an approximately 1-log reduction or less. In the absence of nisin, the divalent cations had no effect on cell viability. This effect was attributed to the divalent cation, as opposed to the anion, on the basis of the similar results obtained with MgSO₄ and MgCl₂ and the lack of effect of KCl and NaCl. The effect was confirmed to be due to the divalent cation by experiments involving EDTA, a chelator of divalent cations. Inclusion of 20 mM EDTA in any of the systems containing divalent cations increased the lethality caused by nisin to a 3- to 4-log reduction. In a parallel experiment with wild-type cells, nisin was equally lethal regardless of the presence or absence of divalent cations (data not shown). The metal salts and EDTA, either alone or in combination, had no effect on the viability of either strain.

View larger version (29K): [in this window] [in a new window]   FIG. 3.   Log reduction in cell viability of Nisr L. monocytogenes by 60 IU of nisin per ml in the absence () and presence (+) of 20 mM EDTA and 10 mM metal salts.

View larger version (13K): [in this window] [in a new window]   FIG. 4.   Concentration dependence of MgSO4 on the log reduction in cell viability caused by 1,000 IU of nisin per ml against wild-type () and Nisr () L. monocytogenes.

Determination of bacteriocin cross-resistance. The sensitivity of the Nis^r strain to other lactic acid bacteria bacteriocins was assessed. Neither the wild-type nor the Nis^r strain was inhibited by L. lactis ATCC 21053, the control strain which does not produce bacteriocin (Table 3). As expected, the Nis^r strain was not inhibited by L. lactis ATCC 11454, the nisin-producing strain. The Nis^r strain was less sensitive to pediocin PA-1 and leuconocin S than the wild-type strain.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Nisin resistance in L. monocytogenes has been correlated with altered cytoplasmic membrane fatty acids (32, 33) and phospholipid compositions (34) and with alterations in the cell wall (8). This study identifies all of these changes, in addition to a requirement for divalent cations, in a single nisin-resistant strain, demonstrating that resistance to nisin in L. monocytogenes occurs via a complex mechanism.

Alterations in the cytoplasmic membrane composition of nisin-resistant L. monocytogenes were first reported by Ming and Daeschel (33, 34). Their nisin-resistant mutant contained more straight-chain and fewer branched-chain fatty acids (33) and less phosphatidylglycerol and cardiolipin (34) than the wild type. Similar changes in membrane fatty acid composition, including increased long-chain fatty acids, decreased short-chain acids, and a lower C₁₅/C₁₇ ratio, were found for Nis^r L. monocytogenes ATCC 700302 (32). Both laboratories concluded that the observed fatty acid composition changes were consistent with a less fluid cytoplasmic membrane and that this increase in rigidity might prevent nisin from inserting into the membrane. In addition, we report here that the Nis^r strain ATCC 700302 contained more of the zwitterionic phospholipid phosphatidylethanolamine and less of the anionic phospholipids phosphatidylglycerol and cardiolipin than the wild-type cells. A major difference between the Nis^r strain ATCC 700302 and the nisin-resistant strain studied by Ming and Daeschel (34) is that phosphatidylethanolamine was a major component of the Nis^r ATCC 700302 strain. While this study confirmed previous reports that the major phospholipids in L. monocytogenes are phosphatidylglycerol and cardiolipin (27, 31), it is the first to identify and quantify the phosphatidylethanolamine component of the cytoplasmic membrane of L. monocytogenes. Given the role that anionic phospholipids play in nisin's interaction with membranes (10, 18, 30), a decrease in the net negative charge of the lipid bilayer might hinder nisin's ability to bind and interact with the membrane.

Both the fatty acid composition change (32) and the alteration in membrane phospholipids were observed only when Nis^r L. monocytogenes ATCC 700302 was grown in the presence of nisin, indicating that nisin induced these changes. This suggested that the cytoplasmic membrane alterations might not be the cell's primary defense against nisin. We therefore looked for changes in the cell wall of the Nis^r strain by evaluating the strain's sensitivity to cell wall-acting compounds. The Nis^r strain was more resistant than the wild type to lysozyme, which catalyzes the hydrolysis of the -1,4 glycosidic bond between N-acetylmuramicglucosamine and N-acetylglucosamine of cell wall peptidoglycan, and more sensitive to the cell wall-acting antibiotics benzylpenicillin and ampicillin, which block the cross-linking reaction of peptidoglycan synthesis. These altered sensitivities suggest compositional changes in the cell wall of the Nis^r strain. The nature of these compositional changes remains to be determined. In addition, small but significant (P < 0.001) increases in the MICs of gramicidin S and gentamicin for the Nis^r strain were observed. The sites of action of gramicidin S and gentamicin are the cytoplasmic membrane and protein synthesis, respectively. Given that no other changes in sensitivities to other membrane-acting or protein synthesis-disrupting compounds were observed, the increases in resistance to gramicidin S and gentamicin might result from an alteration in the cell wall which prevents these compounds from reaching their targets. The altered sensitivities to lysozyme and the penicillins were observed even when the Nis^r strain was cultured in the absence of nisin. Therefore, unlike the fatty acid composition change, the cell wall alterations are constitutive. No gross morphological changes in the cell wall of the Nis^r strain were apparent by EM. The cell wall has also been implicated in nisin-resistant Listeria innocua, in which altered sensitivities to cell wall-acting antibiotics and enzymes and a thickened cell wall were observed (29). In addition, removal of the cell wall from nisin-resistant L. monocytogenes F6861 resulted in the loss of nisin resistance, suggesting that differences in the cell wall were responsible for resistance in this strain (7).

Next, to determine whether nisin causes significant membrane disruption in the Nis^r strain, the cell's PMF was measured. In the absence of nisin, the wild-type and Nis^r strains of L. monocytogenes maintained similar basal PMFs of around 120 mV, indicating that the mutation to nisin resistance did not affect the Nis^r strain's ability to generate and maintain a PMF. Addition of nisin at 60 IU/ml to wild-type cells caused a 3-log reduction in cell viability and depleted the PMF by 88%. In the Nis^r strain, nisin at 60 IU/ml reduced the PMF by only 21%. While this PMF reduction was statistically significant, the remaining PMF was still within the range required to support bacterial growth (22). The reduction of PMF in the Nis^r strain indicates that some membrane disruption occurred in this strain, but this decrease was not sufficient to cause significant lethality.

In addition to changes in the cytoplasmic membrane and cell wall, the Nis^r strain of L. monocytogenes required divalent cations to resist the inhibitory effect of nisin. The effect was dependent on the concentration of divalent cations. Abee et al. (1) found that di- and trivalent cations (Mg²⁺, Ca²⁺, and Gd³⁺) decreased the nisin Z-induced rate of K⁺ efflux from whole cells of L. monocytogenes Scott A. They suggested that di- and trivalent cations might inhibit the electrostatic interactions between the positive charges on the nisin molecule and negatively charged phospholipid head groups. Alternatively or additionally, the neutralization of the negative head group charges may induce a condensation of these phospholipids, resulting in a more rigid membrane (1). In our study, divalent cations did not prevent nisin from killing wild-type cells. As for the role of divalent cations in protecting Nis^r L. monocytogenes from nisin, it seems unlikely that the cations are interfering only with the electrostatic binding of nisin to the anionic phospholipids. If this were simply the case, divalent cations would also be expected to prevent binding to and subsequent killing of wild-type cells. However, Nis^r cells also have reduced amounts of the anionic phospholipids phosphatidylglycerol and cardiolipin in their membrane, as well as an altered fatty acid composition consistent with a more rigid membrane. These two membrane composition changes alone were still insufficient to prevent nisin from interacting with the Nis^r cells in the absence of divalent cations. Perhaps divalent cations are required to sufficiently stabilize the altered Nis^r cell's cytoplasmic membrane against disruption by nisin. This stabilization might involve interfering with nisin's binding or a specific interaction of the cations with membrane components or some combination of the two. Alternatively, the divalent cations may also play a role in the altered cell wall of the Nis^r strain. Divalent cations interact extensively with cell wall components, especially the negatively charged teichoic acids. Perhaps the change in the cell wall of the Nis^r strain requires additional divalent cations to either stabilize extra negative charge or prevent nisin from binding to anionic sites.

A model illustrating nisin's interaction with wild-type L. monocytogenes and how this interaction might be disrupted in the Nis^r strain is presented in Fig. 5. With wild-type cells, nisin passes through the cell wall; binds to the cytoplasmic membrane, probably via electrostatic interactions with the anionic phospholipids phosphatidylglycerol and cardiolipin; and disrupts the membrane through the formation of pores. In the Nis^r strain, cell wall alterations may prevent nisin from interacting with the cytoplasmic membrane. Nisin which does reach the membrane interacts with it to induce the changes in fatty acid and phospholipid composition. The bound nisin probably forms a limited number of pores, since the PMF of the Nis^r strain was partially depleted by nisin. These pores, however, are not sufficient to cause significant lethality. Nisin's ability to bind to the cytoplasmic membrane of the Nis^r strain may be hindered by the decrease in net negative charge of the membrane surface, and its ability to insert may be hampered by a decrease in membrane fluidity. Divalent cations, which are required by the Nis^r strain to resist the effects of nisin, may play a role in stabilizing the cell wall and/or the cytoplasmic membrane or may prevent nisin from binding to the cell wall and/or the membrane. This model is consistent with data from a variety of sources and provides a conceptual framework for more-detailed studies of various aspects of the Nis^r phenotype.

View larger version (52K): [in this window] [in a new window]   FIG. 5.   Model, as described in Discussion, for the action of nisin against wild-type (A) and Nisr (B) L. monocytogenes.

Finally, the Nis^r strain of L. monocytogenes was also cross-resistant to the class IIa bacteriocin pediocin PA-1 and the class IV leuconocin S. Pediocin PA-1 is a 44-amino-acid protein whose sequence has been determined (24) and modeled into a three-dimensional structure (5) which predicts that initial binding to membranes is through electrostatic interactions. Leuconocin S is a small (molecular weight, <10,000) glycoprotein (28). Both leuconocin S and pediocin PA-1, like nisin, act against L. monocytogenes by depleting the PMF (4). L. monocytogenes mutants resistant to mesenterocin 52, curvaticin 13, and plantaricin were each also cross-resistant to the other bacteriocins (38). In addition, piscicolin 126-resistant mutants of L. monocytogenes which emerged in cheese made from milk containing the bacteriocin were also resistant to pediocin P02 (43). These reports of cross-resistance indicate that the use of multiple bacteriocins to achieve greater antibacterial efficacy (20) might not be feasible. The development of resistance to one of the bacteriocins in the combination might render the organism resistant to the others.