Temperature shock, injury and transient sensitivity to nisin in Gram negatives

Authors


Adams School of Biomedical and Life Sciences, University of Surrey, Guildford, Surrey GU2 7XH, UK (e-mail: m.adams@surrey.ac.uk).

Abstract

Aims: The effect of thermal stresses on survival, injury and nisin sensitivity was investigated in Salmonella Enteritidis PT4, PT7 and Pseudomonas aeruginosa.

Methods and Results: Heating at 55°C, rapid chilling to 0·5°C or freezing at –20°C produced transient sensitivity to nisin. Cells were only sensitive if nisin was present during stress. Resistance recovered rapidly afterwards, though some cells displayed residual injury. Injury was assessed by SDS sensitivity, hydrophobicity changes, lipopolysaccharide release and NPN uptake. LPS release and hydrophobicity were not always associated with transient nisin sensitivity. Uptake of NPN correlated better but persisted longer after treatment.

Conclusions: Thermal shocks produce transient injury to the outer membrane, allowing nisin access. After treatment, the permeability barrier is rapidly restored by a process apparently involving reorganization rather than biosynthetic repair.

Significance and Impact of the Study: Inclusion of nisin during food treatments that impose sub-lethal stress on Gram negatives could increase process lethality, enhancing microbiological safety and stability.

INTRODUCTION

The outer membrane (OM) of Gram-negative bacteria confers resistance to the food preservative, nisin, by acting as a permeability barrier, preventing access of nisin to its site of action, the cytoplasmic membrane. If the OM permeability is altered by chelators or physical treatments, however, Gram negatives can show nisin sensitivity (Stevens et al. 1991; Kalchayanand et al. 1992, 1994; Boziaris et al. 1998).

The cell envelope is a common site of injury by chilling, freezing and heating, giving rise to leakage of cations and u.v. absorbing material (Russell and Harries 1967; Strange and Ness 1963; Calcott and MacLeod 1975). In particular, a variety of changes can occur in the outer membrane, including morphological and structural changes, involving blebs and vesiculation, and damage or release of lipopolysaccharides. These changes can alter the permeability barrier, causing efflux of periplasmic enzymes, and sensitivity to hydrophobic compounds, dyes and surfactants (Ray et al. 1976; Kempler and Ray 1978; Katsui et al. 1982; Tsuchido et al. 1985). Such injury can be temporary, with bacteria having the capacity to recover or repair subsequently. In some case this process can be very rapid. Transient sensitivity to nisin has previously been reported in several Gram negatives subjected to rapid chilling, where cells re-established nisin resistance as soon as they returned to room temperature (Boziaris and Adams 2000). In this paper, the relationship is explored between different temperature stresses (heating, rapid chilling and freezing/thawing), transient nisin sensitivity and measures of sub-lethal injury such as sensitivity to hydrophobic compounds, lipopolysaccharide loss, changes to cell hydrophobicity, and permeability to fluorescent probes in the Gram negatives, Salmonella Enteritidis PT4 and PT7, and Pseudomonas aeruginosa.

MATERIALS AND METHODS

Organisms, media and chemicals

Salmonella Enteritidis phage type 4 P167807 and phage type 7 P469815 were supplied by the Division of Enteric Pathogens, Central Public Health Laboratory, London, UK. Pseudomonas aeruginosa USCC 2186 was from the University of Surrey Culture Collection. All bacteria were stored on beads (Protect; Technical Service Consultants Ltd, Heywood, UK) at –70°C. For resuscitation, one bead was added to 10 ml Nutrient Broth (NB) and incubated at 37°C for 24 h.

Nisin was supplied in a purified form (5 × 107 IU g–1) by Aplin and Barrett Ltd, Dorset, Beaminster, UK. Nisin solutions in 0·02 N HCl (pH 2) were sterilized by filtration through 0·45 μm filters (Minisart, NML, Sartorius, Göttingen, Germany) and stored at 4°C.

Microbiological media were supplied by Oxoid unless otherwise stated. Nutrient agar (NA) supplemented with 0·15% SDS was prepared by adding SDS solution, sterilized by filtration through 0·45 μm filters, to molten autoclaved NA.

All chemicals were supplied by Sigma.

Experimental design

To eliminate variation between batches of culture, one culture of micro-organisms was grown for each series of replicates, centrifuged, washed and resuspended to give a concentrated suspension. Specific volumes were then added to the treatment medium and the property (population, hydrophobicity or lipopolysaccharide level) was measured. This permitted paired observations between the control and each of the treated samples. All trials were carried out at least in triplicate.

Temperature shock treatment

Chilling, freezing and heating of bacterial populations were conducted in NB or phosphate-buffered saline (PBS) in glass MacCartney bottles. Resuscitated cultures were diluted 10-fold in Maximum Recovery Diluent (MRD) for the inoculation of 50 ml pre-warmed NB at 37 or 25°C to give an initial suspension of approximately 1–10 cfu ml–1. All broths were incubated statically at 37 or 25°C for 18–24 h and centrifuged (1500 g for 15 min at 20°C). Previous determination of a growth curve under these conditions had shown the cultures to be into the stationary phase at this point. The pellets were washed in saline, resuspended in NB to a cell concentration of about 109–1010 cfu ml–1 and 100 μl of the suspension added to 9·9 ml NB, with or without nisin, to give a population of about 107–108 cfu ml–1.

For heating, the bottles containing the bacterial suspensions were transferred to a water-bath (Haake DC-1 circulator heater, Fisons Scientific Equipment, Loughborough, UK) operating at 55 ± 0·1°C and left for 10 min. The contents reached > 54°C within 5 min. On removal, they were cooled immediately in cold water at 4°C and placed in a water-bath at 37°C for 5 min to reach 37°C.

Freeze-thaw shock was performed by placing the cell suspension in a freezer at –20°C and leaving for 2 h. The suspensions were then transferred to a water-bath at 37°C for 10 min where they were defrosted and the temperature raised to 37°C.

For chilling shock, bottles containing the treatment menstruum were pre-cooled to 0·5°C in an iced-water-bath. The cell suspensions were then added and left for 10 min, before placing in a 37°C water-bath for 5 min to reach 37°C. The temperature was recorded with a NAMAS certified probe and digital indicator (Pt 100 probe and Series 268 indicator, Anville Instruments, Camberley, UK).

Treatments with nisin

For treatments in the presence of nisin, 0·1 ml nisin solution or HCl control was added to 9·8 ml NB (or PBS) in MacCartney bottles, before 0·1 ml of cell suspension was added. When nisin was added after treatments, 0·1 ml nisin solution or HCl was added after the temperature stress had been carried out and the suspension was back at 37°C.

Enumeration and evaluation of the injury

After the treatments, 5 ml samples of the test suspension were transferred to 10 ml centrifuge tubes and centrifuged at 1500 g for 15 min. The pellets were washed in 5 ml saline to remove remaining nisin and re-centrifuged. The bacterial pellet was resuspended in 5 ml MRD, serially-diluted, and plated as 0·1 ml spread plates on Nutrient Agar (NA) and NA supplemented with 0·15% SDS. The plates were incubated at 37 or 25°C for 24–48 h.

Treatments with NPN

1-N-phenylnaphthylamine (NPN), as a 1 mmol l–1 solution in acetone, was added to 9·9 ml of cell suspension in PBS to give a final concentration of 10 μmol l–1. The fluorescent intensity of the samples was measured at 420 nm, following excitation at 350 nm, using a Perkin Elmer LS-5 Luminescence Spectrometer (Perkin Elmer, Beaconsfield, UK).

Determination of cell surface hydrophobicity

The methods used were a modification of microbial adhesion to hydrocarbons (MATH), using hexadecane (Mozes and Rouxhet 1987) and hydrophobic interaction chromatography (HIC) (Smyth et al. 1978).

For the MATH method, stationary phase cell suspensions grown in NB at optimum temperatures were centrifuged at 1000 g for 15 min and the pellets washed and re-suspended in Phosphate, Magnesium, Urea (PUM) buffer (22·2 g l–1 K2HPO4.2H2O, 7·26 g l–1 KH2PO4, 1·8 g l–1 urea, 0·2 g l–1 MgSO4.7H2O, pH 7·1). The absorbance and the counts of the resulting suspension were determined at 400 nm and as 0·1 ml spread plates on NA, respectively. A 2 ml sample of the suspension was mixed with 0·5 ml n-hexadecane and left at room temperature for 10 min before vortexing for 60 s. The suspension was left for a further 30 min before the aqueous phase was removed and A400 measured. The percentage of hydrophobicity was estimated using the formula: (1 – A2/A1) × 100%, where the subscripts 1 and 2 are the measurements taken before and after mixing with hexadecane, respectively.

For HIC, the cell suspensions were centrifuged at 1000 g for 15 min and the pellets washed and resuspended in high ionic strength buffer (10 mmol l–1 potassium phosphate + 1 mol l–1 (NH4)2SO4) to give a population of approximately 1010 cells ml–1. Columns comprising short-ended glass Pasteur pipettes were plugged with glass wool and filled to a height of 20 mm with octyl sepharose CL-4B or sepharose CL-4B as a control column. Columns were equilibrated with 2 ml buffer before the addition of 100 μl test suspension. Cells were eluted with 2 ml buffer. A400 readings and spread plates of the initial cell suspension and the column eluate were conducted, and the percentage of cells retained within the gel matrix, and the percentage eluted, were calculated. The percentage of retained cells in the octyl sepharose column gives a relative measure of the hydrophobicity of the cell population, while the sepharose CL-4B control demonstrates the extent of non-specific, non-hydrophobic interaction in the columns. The hydrophobicity index (HI) of the cell population was calculated as:

inline image

(Van der Mei et al. 1987).

Determination of lipopolysaccharides

Lipopolysaccharides were assayed by measuring levels of β-hydroxy fatty acids, characteristic constituents of lipid A, using gas liquid chromatography (GLC) according to Lambert (1991). The amount of β-hydroxy fatty acids remaining on the whole cells after the treatments was measured and expressed per milligram of dry cell mass. Dry weights were determined by drying 1 ml of washed cell suspensions at 105°C.

Suspensions of test organisms in 10 ml NB containing approximately 109 cfu ml–1 were treated by heating, rapid chilling or freezing. After treatment, the cells were centrifuged at 1500 g for 15 min, washed in saline (0·85% NaCl) and resuspended in 1 ml distilled water prior to analysis.

The assay was carried out in 15 ml glass hydrolysis tubes fitted with Teflon-lined screw caps (Fisher Scientific, Loughborough, UK). To each series of tubes, 40 μl of a 0·5 mg ml–1 solution of nonadecanoic acid in hexane were added as an internal standard and the solvent allowed to evaporate at room temperature. To each tube, 1 ml of the cell suspension in distilled water was added, followed by 4 ml 6 mol l–1 HCl. The tubes were then sealed and heated at 100°C for 4 h in a heating dry block (Techne DB3A, Scientific Laboratory Supplies Ltd, Nottingham, UK). The tubes were then cooled, 4 ml of boron trifluoride/methanol complex 14% w/v added and the tubes heated for 10 min at 80°C to produce the methyl esters.

The mix was then extracted twice with 2 ml chloroform:hexane (1:4) mixture, the upper phase removed and reduced to dryness on a rotary evaporator (Buchi, rotavapor-R, Flawl, Switzerland). The residue was re-dissolved in 50 μl hexane and transferred to small vials for storage at –20°C.

For the GLC analysis of the fatty acid methyl esters, a Chrompack CP-SIL 88, WCOT fused silica 30 m (length) × 0·25 mm (internal diameter) column was used on a Varian 3400, with FID detector and helium at 0·12 MPa as the carrier gas. The split ratio was 1/30. The column temperature was programmed for an initial temperature of 135°C for 10 min, 1°C min–1 to 150°C, and then to a final temperature of 200°C at 5°C min–1. The run duration was 35 min. The sample size was 1 μl. The characteristic β-hydroxy fatty acids for the Enterobacteriaceae and Pseudomonas, β-hydroxy tetradecanoic and dodecanoic acids, respectively (Meadow 1975), were identified by spiking samples with the authentic compounds.

Levels of acids were determined from calibration curves of analyte/internal standard peak area ratios against concentration, using standard solutions of the authentic acids put through the same extraction and esterification procedure.

RESULTS

The effect of treatments on injury and inactivation

Cells were subject to three types of thermal shock: heating at 55°C, rapid chilling to 0·5°C, and freezing at –20°C for 2 h followed by thawing. Survivors were enumerated on a non-selective medium (NA), and the extent of outer membrane injury in the surviving population was evaluated by comparing the total counts with counts on selective media (NA containing SDS). A concentration of 0·15% SDS in NA was used following preliminary trials which showed this concentration to be the highest that had no effect on healthy cells. It gave almost identical results to XLDA when used to detect injury in Salmonella, but was applicable to both Pseudomonas and Salmonella.

Figure 1(a-c) shows the effect of heating, chilling and freezing, respectively, on the bacterial populations. All three physical treatments reduced the culturable population. Heating had the greatest effect, giving a 1–2 log cycle reduction in culturable numbers. The low temperature treatments generally gave reductions of less than 0·5 log cycles, with cold shock having the lowest bactericidal effect.

Figure 1.

 Effect of (a) heating, (b) cold shock and (c) freezing and the presence of nisin on injury and inactivation of Gram negatives. (▮), Reduction in total count; (□), reduction in uninjured cells. Error bars show the standard deviation of the mean of at least three replicates

Sub-lethal injury, measured by comparing the count reduction on selective medium with the non-selective medium, was also greatest in survivors of the heating process, where counts on the two media differed by 0·5–2 log cycles depending on the organism studied. Chilling and freezing gave far lower differences between NA and NA/SDS counts, indicating that less than 50% of the survivors were injured.

The effect of nisin on temperature shocked bacteria

Unstressed cells of Ps. aeruginosa and the two salmonellas were not sensitive to nisin. When nisin was present in the medium during the different treatments, however, a dose-dependent increase in lethality was observed in most cases (Fig. 1). Pseudomonas aeruginosa was very sensitive to chilling and freezing in the presence of nisin, where there was a 2–3 log reduction in the number of survivors, but not to heating. Salmonella Enteritidis Phage Type 7 was far more sensitive to freezing in the presence of nisin than Salmonella Enteritidis Phage Type 4.

The presence of nisin during physical treatment reduced counts on both NA and NA/SDS, but to different extents. Heating with nisin reduced the total population of survivors (NA counts) more than the population of uninjured survivors (NA/SDS counts), showing that nisin inactivated primarily the injured cells. Addition of nisin 5 min after heating had little or no effect on the surviving population, indicating that the injury carried by the survivors at that stage was not sufficient to make them sensitive to nisin (Fig. 2a).

Figure 2.

 Effect of (a) heating, (b) cold shock and (c) freezing and subsequent addition of nisin on injury and inactivation of Gram negatives. (▮), Reduction in total count; (□), reduction in uninjured cells. Error bars show the standard deviation of the mean of at least three replicates

In freezing and chilling, where the proportion of injured survivors was far less, the presence of nisin during the treatment reduced the counts on NA and NA/SDS to a similar extent. Again, the survivors were largely insensitive to nisin added minutes after the treatment. (Fig. 2b,c). Pseudomonas aeruginosa was slightly sensitive to high nisin concentrations (2500 IU ml–1) added after chilling and freezing, but the salmonellas were completely unaffected. Thus, in freezing and chilling, the activity of nisin during the treatment was against cells which did not exhibit injury when cultured after treatment.

Similar results were seen when all the treatments were carried out in phosphate buffer rather than nutrient broth; cells showed sensitivity to nisin when present during treatments but not when it was added after (data not shown).

Effect of stress on NPN fluorescence

NPN was added to unstressed cells and to physically-stressed cells and its fluorescence measured (Table 1). The fluorescence intensity increased during the physical treatments, confirming OM damage. When NPN was added after treatment, the fluorescence intensity remained higher than the untreated control, indicating that molecules such as NPN could still gain access to the cytoplasmic membrane.

Table 1.   Effect of stress on NPN fluorescence Thumbnail image of

Cell surface hydrophobicity

The hydrophobicity of the bacterial cell surfaces was measured using two different methods: hydrophobic interaction chromatography (HIC) and microbial adhesion to hydrocarbons (MATH). Both methods showed that Salmonella Enteritidis PT7 was the most hydrophobic organism and that the other two had similar hydrophobicities (Fig. 3).

Figure 3.

 Percentage hydrophobicity determined by MATH (▮) and HIC procedure (□), using absorbance at 400 nm. The error bars show the standard deviation of the mean from at least three replicates

Changes in hydrophobicity occurring during heating, rapid chilling and freezing/thawing were also recorded. Paired observation t-tests between the controls and the treated samples were used to evaluate the significance of observed differences in hydrophobicity. For the strains tested, the hydrophobicity increased during heating and freezing (< 0·05) but did not change significantly as a result of chilling (Table 2). The increase was greater for heating than for freezing. For Salm. Enteritidis PT4, PT7 and Ps. aeruginosa, the hydrophobicity increase was 65, 32 and 57%, respectively, for heating, and 26, 12 and 25%, respectively, for freezing.

Table 2.   Effect of heating, chilling and freeze/thaw on cell surface hydrophobicity Thumbnail image of

The results did not correlate with the nisin sensitivities shown by the stressed bacteria during the different treatments (Fig. 1a–c). Salmonella Enteritidis PT7 was the most hydrophobic strain, and its hydrophobicity increased to the highest levels during heating, but it showed intermediate nisin sensitivity. In the freezing treatment, Salm. Enteritidis PT4 was the least sensitive to nisin while Ps. aeruginosa was the most sensitive, despite their similar hydrophobicities. No changes in hydrophobicity were apparent as a result of chilling, yet chilling did induce transient nisin sensitivity.

Loss of lipopolysaccharides

The amount of LPS hydroxy fatty acid markers remaining on the whole cell after treatment was determined by GLC using nonadecanoic acid as an internal standard.

The salmonellas and Pseudomonas contain β-hydroxy tetradecanoic acid and β-hydroxy dodecanoic acid, respectively (Meadow 1975). This was confirmed for the strains used here and the respective peaks identified by spiking samples with the authentic acid esters. The amount of β-hydroxy fatty acids in the two salmonellas, 5·8 μg mg–1 dry wt (PT4) and 5·1 μg mg–1 dry wt (PT7) (Table 3) did not differ significantly (P=0·46), indicating that the rough strain (PT7) contains similar amounts of LPS to the smooth strain (PT4).

Table 3.   Lipopolysaccaharide loss, as loss of β-hydroxy fatty acid Thumbnail image of

Heating caused release of LPS, although only salmonellas showed significant release of their lipopolysaccharides after 10 min heating at 55°C (Table 3), with losses of 41 and 42% of the total lipopolysaccharides for Salmonella Enteritidis PT4 and PT7, respectively. No significant release was detected for Ps. aeruginosa (Table 3). Pseudomonas aeruginosa was the most resistant to the presence of nisin during heating, and Salm. Enteritidis PT4 the most sensitive. The results show a positive correlation between OM damage, as measured by release of lipopolysaccharide, and the ability of nisin to inactivate heat-injured cells.

Chilling did not cause significant release of LPS from any of the tested organisms (Table 3), although the stressed cells did display nisin sensitivity during chill stress.

Freezing/thawing gave similar losses of LPS in Salm. Enteritidis PT4 and PT7, which were higher than those seen in Ps. aeruginosa (43, 46 and 34%, respectively). However, freeze-stressed Ps. aeruginosa showed a much greater reduction in viability in the presence of nisin, compared with Salm. Enteritidis PT4 and PT7.

DISCUSSION

The stress treatments used affected both the viability and sub-lethal injury of survivors to different extents, although heating produced greater inactivation and injury than chilling or freezing. Cells which survived the various treatments were not sensitive to nisin when it was added after the stress, regardless of whether or not they displayed residual injury.

Cells were, however, sensitive to nisin when it was present during treatments, and reduced counts were seen on both NA and NA/SDS. Heating with nisin reduced the NA counts (total population of survivors) more than the NA/SDS counts (uninjured population), indicating that many of the organisms, which would normally survive heating in an injured state, had been inactivated by nisin during heating. In freezing and chilling, the proportion of injured survivors was far less, and the presence of nisin during treatment reduced the counts on NA and NA/SDS to a similar extent. This, and the fact that reduction in viable numbers was far greater than the extent of injury in survivors, suggests that transient injury during processing had made the cells nisin-sensitive, and that all cells injured in this way had either been inactivated by nisin or had recovered rapidly after treatment. The inability of the cells to form colonies on NA with SDS reflects structural injury, especially associated with the permeability barrier (Ray and Speck 1973; Ray 1979). Even though large numbers of heated survivors still showed enough OM injury to allow SDS access, recovery had been sufficient to abolish nisin sensitivity. Similar transient susceptibility of Gram negatives to nisin has also been observed in cells injured by high pressure (Hauben et al. 1996).

Injury during physical treatments was also measured by changes in cell surface hydrophobicity, loss of LPS and NPN absorption. The MATH test to assess surface hydrophobicity is considered unreliable unless supported by alternative techniques and in this case, was supplemented with hydrophobic interaction chromatography (Mozes and Rouxhet 1987). The higher hydrophobicity found for Salm. Enteritidis PT7 compared with PT4 can be ascribed to shorter LPS seen in rough mutants of Salmonella (Magnusson and Johansson 1977). Increases in cell surface hydrophobicity were found only in the heating and freezing treatments. Similar observations have been made with Escherichia coli subjected to heating and freeze/thaw stress (Mackey 1983; Tsuchido et al. 1985), in which treated cells were also found to be sensitive to hydrophobic compounds. Kobayashi et al. (1991) correlated the higher sensitivity of Bacteroides fragilis to hydrophobic antimicrobials (lincomycin and clinamycin) with cell surface hydrophobicity. In the present study, the stress-induced hydrophobicity changes did not correlate with the observed transient sensitivity to nisin across all treatments and organisms. Nisin sensitivity was seen where there was no change in hydrophobicity (chilling), and similar changes in hydrophobicity produced markedly different nisin sensitivities (freezing). Though nisin sensitivity did correlate with hydrophobicity changes in the two salmonellas during heating, this seemed only to apply for very closely related organisms, as the Pseudomonas showed a large increase in hydrophobicity on heating but the lowest nisin sensitivity. Clearly, changes in hydrophobicity detectable after stress are not always responsible for transient nisin sensitivity.

Increased hydrophobicity appears to be related to LPS release, since freezing and heating also produced the greatest losses of LPS. This is presumably due to the temporary appearance of phospholipid in the OM filling the void space (Nikaido and Vaara 1985). The release of LPS by heating has been reported previously (Katsui et al. 1982; Tsuchido et al. 1985) but not as a result of freeze/thawing, though conformational alteration and structural damage of LPS in the heptose region have been described (Ray et al. 1976; Kempler and Ray 1978).

NPN is a fluorescent probe for OM permeability and fluoresces when in a hydrophobic environment, e.g. the cytoplasmic membrane. When the permeability of the OM is altered, NPN can gain access to the cell membrane and fluoresce (Tsuchido et al. 1989). The fluorescence of NPN present during treatments indicated that there is an increase in permeability during all treatments. For the two salmonellas, the fluorescence increase correlated with the transient nisin sensitivity seen during chilling and freezing treatments. Greater NPN uptake was seen in PT7 than in PT4 during heating, although PT4 showed the greater transient sensitivity to nisin. This can be ascribed to the greater heat sensitivity of PT7 which resulted in larger numbers of dead cells that would also take up NPN. Unlike nisin sensitivity, uptake of NPN was similar when the NPN was added immediately after the treatment. This indicates that although recovery of the OM permeability barrier was rapid enough to exclude nisin, it took longer to recover sufficiently to exclude smaller molecules such as NPN. This is supported by the results of Tsuchido et al. (1989), who also found that NPN could still permeate E. coli cells after heat treatment but at a rate which decreased with time. When NPN was added 70 min after the heat treatment, the rate of permeation was similar to that of untreated cells.

Results with the Pseudomonas do not correlate well with those for the two salmonella strains, indicating that differences between organisms in their intrinsic sensitivity to the physical stresses and to nisin are important factors. Nisin is only likely to inactivate Gram negatives which have suffered sufficient damage to their OM to allow the nisin molecule to pass, and this will vary between treatments and with differences in OM composition. There may also be differences in the intrinsic resistance of the damaged cell to nisin due to differences in the composition of the cytoplasmic membrane. The fatty acid and phospholipid composition affect the fluidity of the cytoplasmic membrane and this has been reported to affect nisin sensitivity (Ming and Daeschel 1993, 1995; Crandall and Montville 1998).

Salmonella Enteritidis PT4 and PT7 differ in their LPS chain length, with PT7 possessing shorter chains (Chart et al. 1989). During heating and chilling treatments, they were almost equally sensitive to the presence of nisin, though this was not the case with freezing in the presence of nisin. It could be that the projection of the longer LPS chains from the OM surface protected cells of the PT4 strain from mechanical damage by ice crystals. This is supported by an earlier observation that variations in the LPS chain length affect the sensitivity of freeze/thaw-injured Salm. typhimurium to novobiocin, polymyxin B and SDS (Bennet et al. 1981).

The results show that a variety of stresses can produce transient injury to the OM permeability barrier, allowing nisin temporary access to the cytoplasmic membrane. The transient nature of the nisin sensitivity indicates a relatively rapid reorganization or recovery of the OM after the treatments. Nisin resistance was restored within 5 min of treatment in the case of chilling and heating, and within 10 min of freezing. Biosynthetic repair seems unlikely to play a role in the short-term recovery of nisin resistance, since resistance was apparent when nisin was added immediately after the treatments and recovery was also apparent in cells suspended in buffer. More probably, physical reorganization of the existing molecules in the OM (LPS or possibly OM proteins) is able to restore barrier functions.

Of the various measures of injury examined, uptake of NPN correlated best with transient nisin sensitivity. The more severe treatments, heating and freezing, resulted in loss of LPS and increased hydrophobicity, but this did not correlate well with observed nisin sensitivity. Since NPN was present during the treatments, however, it gave a better indication of transient changes in OM architecture. It may be possible to characterize these transient changes further by employing membrane probes of different molecular size or a range of defined outer membrane mutants.

The results demonstrate that inclusion of nisin during food processing treatments that impose sub-lethal stress on Gram negatives could increase the lethality of the process, enhancing both microbiological safety and stability.

Acknowledgements

The authors would like to thank the State Scholarship Foundation (I.K.Y.) of the Hellenic Republic for the financial support of I.S.B.

Ancillary